Progress in Nucleic Acid Research and Molecular Biology, Volume 65

Some Articles Planned for Future Volumes

Proceedings

of the Workshop

on “Base Excision

Repair 2000”

SANKAKMITRA AND R. STEPHENLLOYD Exoribonucleases

and Their Multiple Roles in RNA Metabolism

MURRAYDEUTSCHER A Unique Combination of Thyroid

of Transcription

Factors Controls

Differentiation

Cells

ROBERTODI LAURO, G. DAMANTE,AND L. PELLIZARI ATP Synthase:

The Missing

Link

STANLEYD. DUNN, D. T. MCLACHLIN, AND M. J. RE~INCTON Molecular

Characterization

of Cation-Chloride

Cotransporters

BLISSFORBUSHAND JOHNPAYNE Functional Analysis

of MUC 1, a Carcinoma-Associated

Mucin

SANDRAJ. GENDLER Manipulation

of Aminoacylation

and Combinational

Properties

of tRNAs

by Structured-Based

in Vitro Approaches

RICHARDGIEGE AND JOEMPUETZ The Role of Acid B-Glucosidase

and Saponins

in Glycosphingolipid

Metabolism

and Disorders

GREGORYA. GRABOWSKI Understanding

Nuclear

to the interphase

Receptor Function: From DNA to Chromatin

Nucleus

GORDON HAGER Regulation of Yeast Glycolytic

Gene Expression

MICHAEL HOLLAND AND JOHNJ. KING Branched Chain Aminotransferases

SUSANM. HUTSON,NIMBE TORRES,AND ARMANDOTOVAR Molecular

Mechanisms

for the Interaction of LDL with the LDL Receptor

T~IOMASL. INNERARITYAND JANBOREN Control of Metallothionine

SAMSONT. JACOB

Gene Expression

SOMEARTICLESPLANNEDFORFUTUREVOLUMES

X Specificity

and Diversity

in DNA

Recognition by E. co/i Cyclic AMP

Receptor Protein

JAMES CLEE Molecular

Mechanisms

of Error-Prone

DNA

Repair

ZVI LIVNEH Translation

Initiation Factors in Eukaryotic

Protein Biosynthesis

UMADASMAITRA Regulation and Function of the Cyclic Nucleotide Phosphodiesterases

(PDE)3

Family

VINCENTC.MANG.~NIELLOANDEVADEGERMAN DNA

Polymerase

III Holoenzyme,

a Prototypical

Replicative Complex

CHARLESMCHENRY Multiple

Routes from the Ribosome to and across the Membrane

MATHIASMULLER,H.-G.KOCH,K.BECK,ANDU.SCHAFER Distinct Regulatory and Phosphatase

Properties

of Pyruvate Dehydrogenase

Kinase

lsozymes

THOMASROCHE Complexity Hormone

of Transcriptional

Regulation Associated

Biosynthesis

MICHAELR.WATERMANANDLARRYJ.BISCHOF The Expanding

World

of DNA Triplet Repeats

ROBERTD.WELLSANDRICHARDP.BOWATER

with Steroid

Cyclic

Nucleotide

Phosphodiesterases: Structure

Relating

and Function SHARR~N

H. FRANCIS,

ILLARION

V. TURKO,

AND JACKIE

D. CORBIN

Department of Molecular Physiology and Biophysics Vanderbilt University School of Medicine Nashville, Tennessee 37232 I. II. III. IV.

.

2 3 3 5 7 8 8 9 9 11 14 16 18 20 20 23 24 26 29 33 37 38 39 40 40 41 41

Introduction .................................................. Background .................................................. Classification ................................................. Structural Features of PDEs .................................... .......................................... A. CatalyticDomain B. Regulatory Domains ........................................ Features of Catalysis ........................................... ....................................... A. CatalyticMechanism B. Determinants of Nucleotide Specificity ........................ C. Structural Determinants for PDE Catalytic Activity .............. D. Metal Requirements ........................................ E. Inhibitors .................................................

VI. Mechanisms Utilized for Regulation of PDEs VII. PDEFamilies ................................................. A. PDElFamily .............................................. B. PDE2 Family .............................................. ............................................ C. PDE3Family.. D. PDE4 Family .............................................. E. PDE5 Family .............................................. E PDEGFamily .............................................. G. PDE7 Family .............................................. ............................................ H. PDE8Family.. I. PDE9 Family .............................................. ............................................. J. PDElOFamily K. PDEll Family .............................................

......................

VIII. Concluding Remarks .......................................... References ....................................................

Cyclic nucleotide phosphodiesterases allophosphohydrolases of CAMP and/or

that specifically

cGMP

critical determinants Progress in Nucleic Acid Research and Molecular Biology, Vol. 65

to produce

for modulation

(PDEs) comprise a superfamily of metcleave the 3’,5’-cyclic

the corresponding

phosphate

5’-nucleotide.

moiety

PDEs are

of cellular levels of CAMP and/or cGMP by

1

Copylight 0 2001 by Academic Prrs. All rights ofreproduction m any form reservrd. 0079.6fiwnl $1.500

2

SHARRON many stimuli. Eleven cGMP

families

of PDEs

have been identified in mammalian

with varying

H. FRANCIS

selectivities

ET AL.

for CAMP

or

tissues. Within these families, multiple

isoforms are expressed either as products of different

genes or as products of the

same gene through alternative splicing. Regulation

of PDEs is important for con-

trolling

the visual response,

myriad

physiological

muscle relaxation, and cardiic

platelet

functions, aggregation,

contractility. PDEs

lular CAMP and cGMP by a panoply

including

such as cGMP

immune

are critically involved in feedback phosphorylation

or phosphatidic

association with specific protein partners. a major target for pharmacological tant maladies.

fluid homeostasis,

levels. Activities of the various PDEs

of processes,

small molecules

including

intervention

control of cel-

are highly regulated

events, interaction

acid, subcellular

The PDE

smooth

responses,

superfamily

localization,

with and

continues to be

in a number of medically impor-

0 2000Academic Press.

I. Introduction Cellular levels of cyclic adenosine 3’,5’-monophosphate (CAMP) and guanosine 3’,5’-monophosphate (cGMP) are determined by the relative activities of adenylyl and guanylyl cyclases, which catalyze their synthesis, and cyclic nucleotide phosphodiesterases (PDEs), which hydrolyze them to the respective 5’-nucleoside monophosphates. The sensitivity of physiological processes to cAMP/cGMP signals requires that their levels be precisely maintained within a relatively narrow range in order to provide for optimal responsiveness in a cell. Agents acting through CAMP or cGMP typically produce maximum responses in tissues with only a transient two- to three-fold increase in cyclic nucleotide. Decline of cyclic nucleotide levels often occurs despite the continued presence of hormone. Several mechanisms are likely to contribute to the rapid decline, which involves increased PDE activity. Cyclic nucleotide PDEs provide the major pathway for eliminating the cyclic nucleotide signal from the cell. Cellular levels of CAMP or cGMP rarely achieve the Km values that have been determined for the PDEs in vitro, so that increased cyclic nucleotide synthesis will undoubtedly be accompanied by an increased rate of hydrolysis. Thus, according to the rule of mass action, cyclic nucleotide accumulation is dampened when synthesis increases. In addition, PDEs are highly regulated by inputs from many pathways; the integration of these signals modulates the catalytic activities of the PDEs and plays a major role in determining the intensity and duration of the cellular response to an external stimulus. PDEs are regulated by intracellular CAMP and cGMP concentrations, binding of Ca2+/calmodulin, phosphorylation events, interaction with regulatory proteins, subcellular localization, and alterations in protein level. This article focuses on some of the major advances that have been made in our understanding of the physical and kinetic char-

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

3

acteristics of PDEs, and the roles that PDEs play in modulating intracellular cyclic nucleotide levels. Numerous other reviews provide excellent resource materials for information pertaining to PDEs (1-24).

II. Background Early studies by Sutherland and colleagues (25,26) demonstrated that tissue extracts contained a PDE activity that could ablate the biological actions of CAMP. This was subsequently identified as PDE catalytic activity, which was increased in the presence of divalent cations and inhibited by methylxanthines such as caffeine and theophylline (26). In the ensuing four decades, PDEs have been shown to be a large superfamily of diverse enzymes that are highly regulated by multiple signaling pathways. These enzymes are targets for many drugs that are used to treat cardiovascular diseases, asthma, depression, male impotence, and other maladies. In addition to the vital role of PDEs in hydrolyzing CAMP and cGMP, a group of these enzymes may also have functions as intracellular receptors or even sinks for cGMP; these PDEs (PDEB, PDE5, and PDEG) bind cGMP with high specificity and affinity at homologous allosteric sites that are arranged in tandem in their amino-terminal regulatory domains, but the physiological roles of these binding sites are poorly understood.

III. Classification Members of the PDE superfamily differ substantially in their tissue distributions, physicochemical properties, substrate and inhibitor specificities, and regulatory mechanisms. Based on differences in primary structure of known PDEs, they have been subdivided into two major classes, class I and class II (27). To date, class I contains the largest number of PDEs and includes all known mammalian PDEs, four freshwater sponge (Ephydutiajluuiatilis) PDEs (28), a Drosophila PDE (29), two nematode (Cuenorhubditis) PDEs (28), and the Sacchuromyces ceretksiue PDE2 gene product (30); each class I PDE contains a conserved segment of -250-300 amino acids in the carboxyl-terminal portion of the proteins, and this segment has been demonstrated to include the catalytic site of these enzymes (Fig. 1) (27, 31). All known class I PDEs are contained within cells and vary in subcellular distribution, with some being primarily associated with the particulate fraction or the cytoplasmic fraction of the cell, others being more evenly distributed in both compartments. Certain PDEs appear to be expressed selectively in different mammalian tissues, but the tissue distribution of PDEs across a num-

SHARRON

H. FRANCIS

ET AL.

PDE Family Number Regulatory

Regions +

Conserved Catalytic Domain

Calmodulin binding sites 4

1

I

I

Calmodulin-stimulated

. Id-l

cAMPlcGYP

I

’PDE

cGYP-binding sites

I

2

cGMP-stimulated cAMPlcGMP ’ PDE

I

membrane association region

3

I \I/0

cGYP-inhibited cAMPlcGMP PDE

.

I II

I

1

UCR sites

A

4

4,

::.C..: _. _.

cAMP-specltic, Rollpram-inhibited PDE

]

I

cGMP-binding sites

I

)

I

Photoreceptor cGMP-specific PDE

6

I

7

I CAMP-specific ;&pram-insensitive

m

8

I

9

I

I

I

I

CAMP-specific PDE IBMX-insensitive

I

High affinity cAMPIcGMP PDE

I

cGMP-binding sites?

10

cGMP-binding, cGMP-specific PDE

cGMP-binding, cAMP/cGYP PDE

4

I

I

I

I

cGMP-binding site?

11

4 WA

I

I

CAMPI cGMP PDE

FIG. 1. Schematic depiction of characteristic structural features of known mammalian (class I) PDE families. Arrangements of functional domains are shown for a monomer, although all known mammalian PDEs exist as dimeric enzymes. Known sites of phosphorylation (P) are shown; upstreamed conserved regions (UCR) are characteristic for PDE4 family. The cGMPbinding sites denote the two allosteric cyclic nucleotide-binding sites that have relatively high specificity for cGMP; PDEs 10 and 11 have sequence homologous to the cGMP-binding sites, but have not been shown to bind cGMP. These conserved motifs occur in other signaling proteins and have also been noted as “GAF domains” (31a). The open block in PDE3 catalytic domain indicates a -44-amino acid insertion that characterizes this family.

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

5

ber of species has not been systematically examined. Because properties of the catalytic activities of PDEs commonly overlap, associating specific physiological functions with an apparent selective tissue distribution of PDEs should be approached with caution. Only three PDEs have been classified as class II PDEs; these include a PDE from S. cerevisiae (product of the PDEl gene) (32), a periplasmic PDE from Vibrio fischeri (33), and a PDE from Dictyostelium discoideum (34) that occurs as an extracellular enzyme as well as being associated with the cell membrane. At this time, no mammalian PDE has been included in class II. Class II PDEs share -32% sequence homology. Due to the absence of sequence similarities between class I and class II PDEs, the evolutionary relationship between these classes cannot be ascertained at this time; similar function in the absence of structural homology could have arisen either by a convergent or a divergent evolutionary process. However, it is likely that additional members of both classes of PDEs and perhaps even additional classes of PDEs will be identified. The focus of this article is on mammalian class I PDEs.

IV. Structural Features of PDEs PDEs from mammalian tissues have been subdivided into 11 families that are derived from separate gene families (Fig. 1) (15, 35-39). PDEs within a given family may differ significantly, but the members of each family are functionally related to each other through similarities in amino acid sequences, specificities and affinities for cGMP and CAMP, inhibitor specificities, and regulatory mechanisms. Comparison of amino acid sequences of PDEs suggests that all are chimeric multidomain proteins possessing distinct domains that provide for catalysis and a number of regulatory functions (40). A phylogenetic tree of the class I PDE superfamily has been constructed based on the amino acid sequences of the mammalian PDE families 1-7, the Saccharomyces PDEQ, the two nematode PDEs (PDEl and PDE4), and the four freshwater sponge PDEs (PDEs l-4) (Fig. 2) (28). From this analysis, vertebrate PDEs are proposed to have diverged from a common ancestral gene by gene duplication and domain shuffling early in the evolution of animals. This comparison reveals that the catalytic domain, the calmodulinbinding domain, the upstream conserved regions (UCRs), and the allosteric cGMP-binding domain are ancient in origin. Furthermore, both the amino acid sequences and the arrangement of the functional domains within the respective PDE structures that served to generate the different families of PDEs had been achieved prior to the evolutionary divergence of sponges and eumetazoans (28).

-

L

TWD3.3

R153.1

human PDE4A

human PDE48

human PDMD

human PDE4C

Drosophila dunce

nematode

human PDEIB

human PDESA

human PDElB

human PDElC

human PDElA

nematode

human PDE7A

PDE3

PDEl

PDE4

1

3

11

PDE7

PDEP

1

1

PDE6

m

-

hydrophobic membrane association domain

cGMP binding domaln

PDE catalytic domain

CAAX motif

@

conservec

CaM-interaction domain

upstream region 2 0

ea

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

A. Catalytic Domain The 11 PDE families exhibit the greatest sequence similarities (-2O45% identity) in the carboxyl-terminal segments, and among the members of a respective family, the homology is much higher (70-80%). This region was predicted to contain the PDE catalytic domain (31). This proposal was proved correct by experiments demonstrating that products derived from partial proteolysis of PDEs 1,2, and 3 and expressed truncation mutants containing the respective catalytic domains of PDEs 3,4, and 5 exhibited PDE catalytic activity (41-49) (49u). Deletion of a 99-amino acid region at the carboxyl terminus of the conserved catalytic domain of a PDE4 isoform altered the kinetic properties, but catalytic function was retained. Internal deletion of a 96-amino acid segment from this PDE4 produced an enzyme with greater activity, thus suggesting the presence of an autoinhibitory domain, but expression of the conserved catalytic core of PDE4 produced inactive protein (46). Studies with PDE5 demonstrate that the catalytic properties (Km for cGMP, and IC,, values for inhibitors) of a monomeric PDE5 catalytic domain closely resemble those of native or wild-type PDE5 (49u). These results indicate that the components required for catalyzing hydrolysis of the phosphodiester bond are contained within a single catalytic domain and that interactions between two catalytic domains within a dimer or between the catalytic domain and the regulatory domain are not required for this process. These results are particularly important because all known mammalian PDEs are either dimeric or oligomeric, and the functional importance of this quatemary structure is not known. Although regions outside the catalytic domain can alter catalytic efficiency for many of the PDEs, the catalytic domain alone is sufficient for catalysis. Notably, the catalytic rates of different PDEs vary markedly and are likely to result from subtle differences in the microarchitecture of the respective catalytic sites. If the metals associated with the respective PDEs vary, this might also account for or contribute to the wide variation in catalytic rates observed with different PDEs. Attempts to specifically photoaffinity label the catalytic sites of a number of PDEs using compounds related to cyclic nucleotides have been modestly successful (42, 50- 53). Using a CAMP analog, S-[(4-bromo-2,3-dioxobutyl) 4 FIG. 2. Phylogenetic tree of PDEs. The tree includes seven vertebrate PDE families, four sponge PDE families, two nematode PDE families, one Drosophila PDE, and one PDE derived from Sacchuromyces. This evolutionary tree was inferred from the amino acid sequences of the catalytic domains; the phylogeny was derived using the NJ method with a fungal PDE as the outgroup. 0, Parazoan-eumetazoan split; 0, human-Drosophila (or nematode) split; + , gene duplications that gave rise to different subtypes; 0, gene duplications whose divergence time is unknown; 0, gene duplications in the same subtype. The branch length is proportional to the number of accumulated amino acid substitutions. Domains that were integrated during evolution by domain shuffling are shown. Reproduced from Ref. 28 with permission.

8

SHARRON

H. FRANCIS

ET AL.

thioladenosine 3’,5’-cyclic monophosphate, Colman and co-workers (51) covalently labeled a peptide in the carboxyl-terminal portion of PDE4A; the labeled peptide was isolated and the sequence was determined to be 6g7GPGHPPLPDK706, but this peptide is located carboxyl terminal to the core of the conserved catalytic domain. The catalytic domain of the rod outer segment PDE6 has been affinity labeled using an inhibitory y-subunit derivatized at the carboxyl terminus; this construct covalently linked the labeled y-subunit specifically into the conserved sequence near the carboxyl terminus of the rod PDE6 catalytic domain (54). This suggests that the inhibitory action of the y-subunit results from its direct interaction with this region of the PDE6 catalytic domain.

B. Regulatory Domains The more amino-terminal and extreme carboxyl-terminal sequences in the various PDEs are highly divergent and vary in length. Numerous studies have established that the amino-terminal sequences in PDEs contain distinct domains that provide for quite varied modes of regulation of PDE function, e.g., allosteric cyclic nucleotide-binding sites, phosphorylation sites, Ca2+/ calmodulin binding sites, inhibitory protein binding site(s), and autoinhibitory domains. In some instances, sequences have been identified that provide for specific subcellular localization of a given PDE when overexpressed in a heterologous cell line. The functional features of the various regulatory mechanisms for each of the PDE families will be discussed in the following respective sections.

V. Features of Catalysis The molecular mechanisms that provide for specific binding of cyclic nucleotides and for hydrolysis of the phosphodiester bond are poorly understood. Although development of selective PDE inhibitors for therapeutic uses has been a major interest for many years, information about catalytic domains of these enzymes is surprisingly modest. The novel six-member phosphodiester ring of CAMP or cGMP is critical for interaction with PDEs because other nucleotides, including ATP, ADP, GTP, GDP, and the corresponding 5’-nucleoside monophosphates, do not interact appreciably with PDEs (Fig. 3). Cyclic AMP and cGMP are resistant to other phosphoesterases, such as intestinal phosphatases and ribonucleases. Enthalpies of hydrolysis of phosphodiester bonds of CAMP and cGMP have been estimated at - 14.1 and - 10.5 kcabmol,

respectively,

which approximates

the ener-

of ATP or GTP (55). Cleavage of the cyclic phosphodiester bond of either CAMP or cGMP by PDEs requires a negative charge,

gy in the y-phosphate

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

cGMP

CAMP

FE. 3. Molecular structures of cGMP and CAMP.

an equatorial oxygen atom in the ring, and the presence of divalent cations (56).

A. Catalytic

Mechanism

The phosphohydrolase action of PDEs involves the nucleophilic substitution of a solvent hydroxyl at the phosphate, resulting in the disruption of the P-O bond at the 3’-oxygen of the ribose. When catalysis is conducted in [r80]water, a single atom of IsO is introduced as one of the three equivalent terminal oxygens of the 5’-nucleotide product (5’~AMP or 5’-GMP); this stoichiometry of incorporation has been interpreted to indicate that the reaction proceeds without exchange of oxygen between phosphate and water (57). Other studies that examined the stereochemical course of catalysis using a CAMP analog with a chiral phosphorous [CAMP(S)] demonstrated that, on lysis of the phosphodiester bond, there is an inversion of configuration at the phosphorous group (58,59). Although other mechanisms are possible, the results are consistent with an in-line mechanism of nucleophilic substitution of the hydroxyl of a water molecule at the phosphate.

B. Determinants

of Nucleotide

Specificity

Two structural components of CAMP and cGMP, i.e., the cyclic phosphodiester ring and the purine moiety, provide for the specificity with which these compounds interact with PDEs. Several PDE families are highly selective for either adenine (PDE4, PDE7, and PDES) or guanine (PDE5, PDEG, and PDES), whereas others (PDEl, PDEB, PDE3, PDElO, and PDEll) accommodate both. In solution, CAMP and cGMP are in equilibrium between two conformations, syn and anti, based on the orientation between the purine and the ribose, whereas the spatial arrangement of the cyclic phosphate ring

10

SHARRON H. FRANCIS ET AL.

and ribose are fixed. Cyclic nucleotide analog studies have suggested that the preferred configuration, either syn or anti, varies among the PDE families (60- 62). The selectivity of the respective catalytic domains for either CAMP or cGMP is undoubtedly determined by interaction of select residues within the cyclic nucleotide-binding pocket with particular chemical properties of the purine, such as the relative hydrophobicity or direction and magnitude of the dipole moments of the adenine and guanine bases (Fig. 3) (56, 63, 64). However, for some PDEs, sequences outside this region have also been implicated in producing effects on the relative cAMP/cGMP specificity of the enzyme (65) or on the potency with which PDE inhibitors (compounds that are analogs of cyclic nucleotides) interact with the enzyme (11, 66- 70). Guanine has a higher dipole moment than does adenine and could therefore form stronger stacking interactions with hydrophobic amino acids. A conserved hydrophobic residue (Tyr/Phe) has been shown to be important for substrate interaction in PDE5 (71). Cyclic GMP and CAMP have similar hydrogen bonding potential in the imidazole ring, but quite different hydrogen bonding potential in the N-l, N-6, and C-2 positions. The purine selectivities of the substrate-binding sites of PDEs most likely reflect constraints in the spatial structure within the substrate-binding site as well as positive and negative chemical interactions between specific catalytic site residues and the adenine or guanine. Hydropathy analysis of the catalytic domains of CAMP- and cGMP-specific families of PDEs suggests that PDE substrate selectivity could result from the pattern of hydrophobic/hydrophilic residues in a short segment of sequence surrounding an invariant Glu; this Glu (Glu-775) has been shown to be critical for cGMP binding in the catalytic domain of PDE5 (64). Sitedirected mutagenesis of PDE5 was used to replace within this segment the residues that are conserved in cGMP-specific PDEs with the conserved residues in the corresponding positions of CAMP-specific PDEs (Table I). The results indicate that the substrate-binding site of PDE5 may contain positive elements for accommodating cGMP, as well as negative elements that disTABLE I CHANCES IN SELECTED RESIDUES ALTER SUBSTRATESELECTIVITY OF THE cCMP-BINDING

cCMP-SPECIFIC

Km(44 PDE5 Wild type A769T/L771R W762L/Q765Y W762L/Q765Y/ A769T/L771R

cGMP

CAMP

2 8 36

330 84 77

43

67

PDE (PDE5)

Relative affinity for substrate (Km cGMP& CAMP) 165 11 2 1.6

CYCLIC

NUCLEOTIDE

11

PHOSPHODIESTERASES

criminate against binding of CAMP Furthermore, the cGMP/cAMP selectivity of PDE5 can be shifted -SO-fold by substituting only two residues (Trp762 and Gln-765) of PDE5 (Table I) with the residues in the corresponding positions (Leu and Tyr, respectively) of PDE4, a CAMP-specific PDE. This produces a 4-fold gain in the affinity of PDE5 for CAMP and an l&fold decrease in the affinity of PDE5 for cGMP, so that the affinity of the PDE5 double-mutant for CAMP is only twice that for cGMP These studies provide further evidence that a significant portion of the interaction of PDEs with cyclic nucleotide is provided by this region.

C. Structural

Determinants

for PDE Catalytic

Activity

Catalytic domains of all mammalian PDEs contain three highly conserved regions of sequence (Fig. 4); the first, a histidine-rich region of 80120 amino acids, is located in the amino-terminal region of the catalytic domain and contains two Zn2+-binding motifs (HX,HX24_2,E), each of which mimics the single motif for coordination of a catalytic Zn2+ in metalloendoproteinases such as thermolysin (Fig. 4) (72- 75). The five histidines that are invariant in all class I PDEs are designated as H-l through H-5 for ease of comparison among PDEs. H-l and H-2 denote the two histidines in Zn2+-binding motif A; H-3 and H-4 denote the 2 histidines in Zn”+ -binding motif B. H-5 indicates the invariant histidine represented by His-675 in PDE5 (Fig. 4). Th e second conserved sequence in the catalytic domain of PDEs includes a conserved dyad of residues, threonine-aspartic acid, and then another cluster nearer the carboxyl terminus. It is now clear that each of these conserved regions in the catalytic domain participates in generating a functionally efficient catalytic site (71).

ZnZ*-Binding Motif A

Znz+-Binding Motif B

/

754DLSAITWWPIQQRIAELVATEFFDQGDRE783 FIG. 4. Features of the conserved catalytic domain in PDEs. General arrangement of the catalytic domain and partial amino acid sequence of catalytic domain of bovine PDES. Bold letters indicate residues that are highly conserved among class I PDEs. Zn2+-binding motifs A and B indicate the two Zn2+-binding motifs (HXaHX,,E/D) that include invariant histidines (H-l through H-4) and are crucial to catalysis. Components of motifs A and B provide for interaction with metal(s) required to support catalysis.

12

SHARRON

HI. FRANCIS

ET AL.

Point mutations of a number of invariant residues in PDEs have been made and assessed for their effects on catalytic function. From these studies, it has been concluded that several of the conserved histidine residues as well as a conserved threonine are critical for efficient catalytic function (46, 68, 71, 76). I n dr‘vid ua 1 point mutations of all invariant residues within the catalytic domain of PDE5 have been made in order to construct a comprehensive structure-function map of the conserved elements within the catalytic domains of PDEs (71). Because these residues are conserved in all mammalian PDEs, the conclusions regarding the function of these amino acids in PDE5 are likely to apply generally to other mammalian PDEs, although minor variations will almost certainly occur. Each of two mutations in PDE5, Y602A and E775A, caused a dramatic increase (-35-fold) in the Km of the enzyme for cGMP (Fig. 5), whereas the kcat for each was reduced only 3- to 4-fold (Fig. 6). Substitution of a phenylalanine for Tyr-602 (Y602F) caused minimal perturbation of the kcat of the enzyme, and the Km was actually slightly improved (4-fold) (no t sh own). These results are consistent with the interpretation that an aromatic residue that occurs at this position in all class I PDEs is probably involved in stacking interactions, which improves catalytic function. Significant deterioration (lo- to 15-fold) in the Km for cGMP

FIG. 5. Comparison of K_ for cGMP and IC,, for zaprinast of catalytic domain mutants of PDE5. The values for the Km for cGMP and IC,, for zaprinast of wild-type PDE5 were taken as 1.0, and the corresponding values for each mutant are expressed as a fold increase with respect to the wild-type PDE5. Reproduced from Ref. 71 with permission.

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

13

also occurred with substitutions at two other conserved residues (Glu-672, and Thr-713), thereby establishing a potentially important role for these residues in cyclic nucleotide binding as well. Other studies using site-directed mutagenesis of the catalytic domain of PDE5 have shown that the cGMP/ CAMP selectivity of PDE5, an enzyme highly specific for cGMP, can be shifted 106-fold by substituting four residues of PDE5 with those in the corresponding positions of the CAMP-specific PDE family, PDE4 (64). Several mutations of PDE5 [His-603 (H-l), His-607 (H-2), His-643 (H-3), His-647 (H-4), Asp-714, and Asp-7541 profoundly disrupt catalysis by reducing kcat of PDE5 with only modest changes in Km for cGMP (Figs. 5 and 6). The decrease in free energy of binding for several of these mutants is within the range predicted for loss of a hydrogen bond or an electrostatic bond involving a charged residue. It must be emphasized that these conclusions depend on specific assay conditions, because subsequent studies reveal that when the divalent cation supporting catalysis is changed, much of the catalytic activity can be restored in some mutants (76~). Several of these residues, His-607 (H-2) and His-643 (H-3) have now been shown to be important for the affinity with which a divalent cation binds to the enzyme as well as important for the relative specificity for different divalent cations (76~). Deciphering the precise roles of each of these various residues in PDE catalytic function will require much additional investigation,

FIG 6. Changes in the kot associated with site-directed mutagenesis of the conserved residues in the catalytic domain of PDE5A. Reproduced from Ref. 71 with permission.

14

SHARRON

H. FRANCIS

ET AL.

D. Metal Requirements Divalent cation is required to support PDE catalytic activity, and the more amino-terminal conserved region in the catalytic domain of PDEs has been shown to provide a major portion of the metal binding by these enzymes (Fig. 4) (72, 76~). Traditionally, Mg2+ has been used to support catalysis, but Mn2+ and Co’+ are also quite effective. A potentially prominent role for Zn2+ has emerged because this cation potently promotes catalysis in a number of PDEs (69, 72, 77). However, a clear picture of the metal requirements for each of the PDEs has yet to be definitively established. Based on the results of a number of studies, it is likely that metal requirements of PDEs are complex and may involve more than one cation. Kinetic studies using bovine rod outer segment PDE (PDEGoB) suggest that Mg2+ and cGMP interact with the PDE in a rapid equilibrium random binding order to form a ternary complex between the PDE, Mg2+, and cyclic nucleotide (78). These same studies discount the likelihood of the formation of a free Mg 2+/cGMP complex due to the high K, of this complex. Independent association of cGMP and Mg2+ with the catalytic site is also supported by studies in PDE5 utilizing cyclic nucleotide analogs (3). Other workers have proposed that divalent metal ions may facilitate catalysis by stabilizing the charged transitional intermediate (59). Work by our own laboratory demonstrated that for PDE5 and PDEGc$, Zn2+ is the most potent cation for supporting catalysis. For PDE5, catalytic activity was supported by submicromolar concentrations of Zn2+ (72), and for PDEGaB, Zn2+ was required even in the presence of 10 mM Mg2” (Fig. 7A) (S. Francis, unpublished results). For these two PDEs, order of potency for these cations is Zn2+ >>Mg 2+. Subsequent reports demonstrate that catalytic activity of recombinant PDE4A is supported by submicromolar concentrations of Zn2+ compared to significantly higher concentrations of other divalent cations (69, 77). However, Zn2+ apparently does not activate certain PDEs (51, 79, 80). Using atomic absorption spectroscopy and 65Zn binding, PDE5 has been shown to bind -3 Zn2+/PDE5 monomer (Fig. 7B) (72). PDE4A is reported to bind specifically Zn 2+ (77, 80), and both class I and class II PDEs in yeast bind Zn2+ specifically (81). Th ese combined results strongly support an important role for Zn 2+ in PDE catalytic function. The strict conservation of a histidine-rich region in catalytic domains of class I PDEs compares well with other catalytic Zn 2+-binding sites, for which X-ray crystal and the importance of two histidines (H-2 structures are known (82-84), and H-3) in this region for metal coordination in PDE5 has now been established (76~). Thus, it seems highly likely that this segment of conserved sequence in PDEs provides the salient features for coordinating required metals and participates in producing a nucleophilic hydroxyl ion from water, as suggested by earlier studies using isO-labeled water and cyclic nucleotide analogs (57- 59).

CYCLIC

NUCLEOTIDE

15

PHOSPHODIESTERASES

A 2.1 Metals Added After Preincubation with EDTA 4

Control

Treated Control

190 gM

1 mM

zn2+

Mn2+

>

B 1.6 -

0

0

I 0.6

I

I 1.2

I 1.6

Zn2+ (PM) FIG. 7. Importance of Zn2+ in PDE function. (A) Requirement of the rod PDE for Zn2+ to support catalysis. Following pretreatment of the rod PDE6 with EDTA, PDE catalytic activity was measured in the presence of 10 mM Mg 2+ either in the absence of other metals or in the presence of the indicated metals. PDE6 that was not pretreated with the chelator was used as control. (B) Saturation curve for 65Zn2+ binding to bovine PDE5A. Reproduced from Ref. 72 with permission.

The presence of two catalytic Zn 2+-binding motifs separated by 10 residues in the PDEs is novel, because other enzymes that utilize this motif to bind Zn2+ contain a single copy of this sequence. Studies in PDE5 have now shown that metal binding by PDEs has unique features compared to the proteases (76~). Stoichiometries of Zn 2+ binding to PDE4, PDE5, and PDE6

16

SHARRON

H. FRANCIS

ET AL.

suggest that PDEs may contain a multinuclear metal-binding site. Site-directed mutagenesis of select residues that comprise the two Zn2+-binding motifs causes a marked loss in PDE catalytic activity of each PDE that has been studied (Fig. 6) (46, 53, 68, 71, 76, 76u, 80). However, the relative contribution of these residues in maintaining catalytically active enzymes has varied. In PDE5, each of the residues in the two Zn2+-binding motifs have been individually mutated; five of the six residues proved to be critical to normal catalytic function (Fig. 6) (71). Th e exception was the glutamic acid in Zn2+-binding motif A. In studies of PDE5, other functions, such as allosteric cGMP-binding, were intact, suggesting that the deleterious effects of these mutations on catalysis are specific. These combined results indicate that residues in both Zn2+-binding motifs in PDEs are important for efficient catalysis. The particular importance of a single his&line in each Zn-binding motif (H-2 and H-3, respectively) for interaction with catalytic metal establishes that metal binding in this region of PDEs will not mimic that in metalloendoproteinases such as thermolysin, which contain only a single motif. In this region, the importance of another conserved residue (Tyr-602 in PDE5), which is not part of the metal-binding motif, is also established by these studies (71). Based on metal requirements, metal binding, and results of site-directed mutagenesis of several PDEs, it seems likely that class I PDEs use a multinuclear metal binding site to effect catalysis. At least a portion of the metalbinding sites is provided by the conserved histidine-rich region in these PDEs, including H-2 and H-3 (76~). Metal-binding site in PDEs could resemble complex metal-binding sites found in other phosphohydrolases, such as alkaline phosphatase or phospholipase C; in each of these sites, one metal is considered the “catalytic” metal whereas the other facilitates the catalytic process (75,84,85). Current evidence strongly suggests that Zn2+ plays a key role in catalysis effected by several PDEs. However, whether a particular metal is absolutely required for rupture of the phosphodiester bond by PDEs and whether one or more metals act directly or indirectly as “cocatalytic metals” to facilitate this process are still open questions. Given significant differences in the Vmax of PDE families, it is also possible that the complement of metals that optimally provides for support of catalysis may vary among PDE families. Last, if multiple metals are involved in catalysis, it is possible that cyclic nucleotide substrate binding may be independent of occupation of one of these metal-binding sites, but dependent on occupation of another.

E. Inhibitors The ognized phylline for use

prospect for utilizing PDE inhibitors therapeutically was quickly recwhen the inhibitory effects of agents such as caffeine and theowere discovered. The quest for specific and potent PDE inhibitors in physiological studies and therapeutic settings has continued un-

CYCLIC

NUCLEOTIDE

17

PHOSPHODIESTERASES

abated, and compounds that exhibit high selectivity for the various PDEs are continually reported (Table II) (13, 86-98). In some instances, these inhibitors are highly selective for a particular PDE family, as in the case of the inhibition of PDE4 by rolipram; however, most PDE inhibitors exhibit some inhibitory effect on multiple PDE families. For this reason, it is risky to invoke physiological roles for specific PDEs based exclusively on the effects of “specific” inhibitors. For example, zaprinast, a compound that for years has been referred to in the literature as a PDE5-specific inhibitor, also inhibits the photoreceptor PDEs (PDEG) and the PDElC isozyme quite well. Likewise, sildenafil (marketed as Viagra) is quite selective for PDE5, but it also inhibits the PDE6 family to some extent (93). Kinetic analysis of PDE inhibition by many of these agents commonly shows competitive kinetics with cyclic nucleotide substrate, thus suggesting that the compounds interact within the same binding pocket on the enzyme, but in other instances, a more complex pattern of interaction has been documented. For example, the kinetics of interaction of rolipram with some members of the PDE4 family suggests multiple sites of interaction (99). Therefore, careful molecular characterization of the interactions of these compounds with PDEs is crucial in order to more effectively design pharmacologically active inhibitors that exploit selective active site features of various PDEs. For PDE5, zaprinast and sildenafil (Viagra) are competitive inhibitors with respect to cGMP, which suggests that the two compounds interact with the same site. However, when catalytic domain mutants are

TABLE II COMMONINHIBITORSOF CYCLIC NUCLEOTIDE PHOSPHODIESTERASES Family PDEl PDEZ PDE3 PDE4 PDE5 PDE6 PDE7 PDES PDE9 PDElO PDEll

Inhibitor” Vinpocetine, zaprinast, SCH51866 EHNA Milrinone, cilostamide, amrinone, enoximone Rolipram, Ro 20-1724, denbufylline, zadarverine, RP73401 CDP840, SB207499, RS25344, LAS31025 Sildenafil, zaprinast, DMPPO, dipyridamole, E4021, SCH51866, GF248 Sildenafd, zaprinast, DMPPO, E402 1, GF248 Benzotbieno- and benzotbiadiazine dioxides Dipyridamole Zaprinast ? Dipyridamole, zaprinast

~‘3.Isobutyl-1-methylwanthineis commonly used as a general inhibitor for most known PDEs, but it is ineffective for inhibition of PDE8 and PDES.

18

SHARRON

H. FRANCIS

ET AL.

compared for changes in Km for cGMP and IC,, for zaprinast, the two parameters do not correlate exactly (Fig. 5). For PDE5, the most dramatic changes in IC,, for zaprinast occur in D754A and G780A mutants, whereas the greatest change in Km for cGMP is observed with the Y602A, T713A, and E775A mutants. Interestingly, effects of the mutations on the potency of sildenafil more closely mimic the pattern for cGMP, with the greatest change (25-fold) occurring in the Y602A mutant (100). These results suggest that the site occupied by zaprinast and sildenafil overlaps the substrate cGMP-binding site, but the residues for binding these ligands differ in some instances. These results emphasize the importance of profiling PDE mutants with respect to these parameters because these types of studies may more fully define important interactions that provide for selectivity and potency of interactions between the respective PDEs and inhibitors. The interaction of the various PDE4 isoforms with rolipram is complex; in some binding studies two types of rolipram binding can be observed, i.e., a high-affinity component and a low-affinity component (99). Amino-terminal deletion of the regulatory region of the PDE4 eliminates the higher affinity binding site, whereas the lower affinity interaction and inhibition of PDE activity is retained (49). Studies using the human recombinant PDE4B suggest that the enzyme contains a single binding site for (R)-rolipram and that different conformational states of the enzyme determine the affinity with which rolipram is bound (101). These studies also support the involvement of both the amino terminus and the catalytic domain in high-affinity rolipram binding by this enzyme.

VI. Mechanisms Utilized for Regulation of PDEs The PDEs are highly regulated enzymes. The general schemes by which the rate of cyclic nucleotide breakdown can be regulated include (1) the availability of substrate, (2) regulation by extracellular signals, (3) short-term feedback regulation, and (4) long-term changes in PDE protein levels. Because the Km values for many PDEs are relatively high compared to cellular cyclic nucleotide levels, it is likely that many of these enzymes are functioning at less than maximum rates. Consequently, PDE activity would increase due to mass action on elevation of cyclic nucleotide levels. Furthermore, when one of the cyclic nucleotides is selectively increased, the rate of hydrolysis of both nucleotides may be changed due to competition by the respective cyclic nucleotides for the catalytic sites in the dual-specificity PDEs (such as PDEs 1, 2, 3, 10, and 11). Extracellular signals modify intracellular PDE activities through a panoply of signaling pathways, e.g., phosphorylation events, vtiations in the levels of small molecules such as Ca2+, phosphatidic acid, in-

CYCLIC

NUCLEOTIDE

19

PHOSPHODIESTERASES

ositol phosphates, and protein-protein interactions. Examples of this type of regulation include the stimulation of PDE3 activity by insulin, leptin, or insulin-like growth factor (16, 102-106), stimulation of PDE5 by atrial natriuretic factor (107), stimulation of PDE6 activity by photons through the transducin system, which alters PDE6 interaction with its inhibitory y-subunit (6, 18, IO??), and stimulation of PDEl activity by its interaction with Ca2+/ calmodulin (10). Short-term feedback regulation of PDEs is an important and immediate physiological response that can involve many of the schemes described above; examples include (1) phosphorylation of PDE3 or PDE4 catalyzed by PKA after CAMP elevation and resulting in an increased Vmax for each of these enzymes (109-111) and (2) allosteric cGMP binding to PDEB, which lowers the Km of PDE2 for substrate, thereby promoting breakdown

A MEETTGKV INO..... KDG..... KDG..... KDG.....

f

N-

-C cGMP-blndlng

snea

cGMP-blndlng

sitek

cattdytk

domain

B

FK. 8. (A) Conserved sequences in the allosteric cGMP-binding sites in PDEs 5,2, and 6. (B) Possible interactions of cGMP with the putative NKX,,D motif based on comparison with interactions of GTP in GTF-binding proteins. Reproduced from Ref. 261 with permission.

20

SHARRON H. FRANCIS ET AL.

of CAMP or cGMP after cGMP elevation (Fig. 8) (112-114). Long-term changes in the level of PDE protein are caused by a persistent change in the cellular environment. Examples of this type of regulation include (1) the densensitization that occurs by increased levels of PDE3 or PDE4 following chronic exposure of cells to CAMP-elevating agents (46, U-122) (2) the developmentally related changes in PDE5 protein (123) or enterotoxin-induced increases in PDE5 level (124, 125), and (3) the selective induction of particular PDE families or isoforms such as the induction of PDE7 that occurs in T cell activation (126) and specific expression of PDE4 isoforms (127, 128). Additional considerations that could influence regulation include selective cellular compartmentalization of PDEs effected by covalent modifications such as prenylation or by specific targeting sequences in the PDE primary structure, and perhaps translocation of PDEs between compartments within a cell (129-133).

VII. PDE Families At least 11 gene families are included in the mammalian PDE gene family. Most of these families have been shown to contain two or more genes; in certain instances, alternative mRNA splicing or utilization of different translational initiation sites produces multiple protein products that exhibit distinct kinetic and regulatory properties. Variations produced by alternative splicing will be discussed for each family. Regulation of gene expression of mammalian PDEs has previously been the topic of discussion in this series (134) and will not be discussed further here.

A. PDEl Family CALCIUMKALMODULIN-BINDING

PDEs

Ca2+/calmodulin (CaM) binds to and activates members of the PDEl family, also known as CaM-PDEs. The Ca2+/calmodulin PDEs are derived from three CaM-PDE genes; alternative mRNA splicing of these gene products produces a number of amino-terminal and carboxyl-terminal PDEl splice variants (135-141). Comparison of nucleotide and amino acid sequences has been used to categorize the PDEl as follows: PDElAl (-59 kDa) and PDElA2 (-61 kDa) are products of the same gene, PDElBl is the product of a separate gene, and PDEl Cl - 5 are products of a separate gene, with variants resulting from alternative splicing at both the amino and carboxy1 termini. The predicted molecular masses of the PDElC variants approximate the determined size of a 75-kDa CaM-PDE that has been described (142).

CYCLIC

NUCLEOTIDE

21

PHOSPHODIESTERASES

Each of the CaM-PDEs hydrolyzes both CAMP and cGMP, but PDElAl, PDElA2, and PDElBl exhibit higher affinity for cGMP (Table III); the five splice variants of PDElC h ave high affinity for both cGMP and CAMP and hydrolyze the two nucleotides with equal efficiency, but the potency with which Ca2+/calmodulin stimulates catalysis varies somewhat (139, 143). Both PDElAl and PDElA2 are dimeric (subunit molecular masses of -59 and 61 kDa, respectively), and their amino acid sequences are identical except for a segment near the amino terminus. Catalytic activity in the PDEl family can be regulated by both Ca2+/ calmodulin binding and phosphorylationdephosphorylation. Ca2+/calmodulin allosterically activates these PDEs through interactions with sequences near the amino termini, although the affinities with which the respective PDEl isoforms interact with calmodulin differ (10, 244-146). For example, half-maximal stimulation of catalysis in PDElAl occurs at a lo-fold lower concentration of Ca2+/calmodulin (Scam - 0.3 nM) than is required for activation of PDElA2 (-4 nM). A sequence near the amino terminus of PDElAl was predicted to form an amphipathic helix suitable for binding Ca2+/calmodulin (43). Synthetic peptides based on this sequence block activation of PDElA2 by Ca2+/calmodulin, thus supporting this interpretation. However, only a portion of this sequence is conserved in PDElA2, which is still dependent on Ca 2+/calmodulin; following deletion of this sequence in PDElAl, the PDE activity can still be activated -3-fold by calmodulin -3 nM). Using several experimental approaches, a region of sequence (%&Xl has been identified in PDElAl and PDElA2 that is likely to account for at least one site of Ca2+/calmodulin binding to these enzymes; the putative Ca2+/calmodulin binding sequence includes residues 115-126 in PDElA2 or residues 97-110 in PDElAl (147). PDElA and PDElB can be phosphorylated in this region by PKA and calmodulin-dependent protein kinase

TABLE III KINETIC PROPERTIES OF CALMODULIN-DEPENDENT

CYCLIC

NUCLEOTIDE PDEs (PDEl)”

Km(FM) PDElA2 PDElBl PDElC2

V miLXratio

CAMP

cGMP

112.7 24.3 1.2

5.0 2.7 1.1

(cAMP/cGMP) 2.9 0.9 1.2

aFrom Ref. 143,A. Z. Zhao, C. Yan, \V K. Sonnenburg,and J. A. Beavo; “Advancesin SecondMessengerPhosphoproteinResearch” (Corbin et al., Eds.), Vol. 3 1, pp. 237-251. Lippincoti-RavenPress, 1997.

22

SHARRON

H. FRANCIS

ET AL.

II, resulting in decreased sensitivity to activation by Ca2+/calmodulin (148 152). In PDElA2, phosphorylation of a single serine (Ser-120) accounts for much of the decreased affinity for calmodulin; whether PKA phosphorylates PDEl in viva has not been established. The two calmodulin-binding regions in PDElAl have been dubbed domains A and B, and although the role of the B domain in activating PDEl is assured, the role of the A domain is unclear and may be to somehow enhance the affinity for binding calmodulin. A number of studies have determined a stoichiometry of one calmodulin bound per monomer of PDElA (149, 153, 154); thus, it is possible (1) that one PDElA2 monomer has two sites of interaction with a single calmodulin molecule, (2) that calmodulin interacts with binding sites on two separate monomers, (3) that only one of these domains in PDElA2 is crucial to the activation by calmodulin whereas the other may enhance the binding affinity for Ca2+/calmodulin, or (4) that the stoichiometry of calmodulin binding has been underestimated. Modulation of the catalytic activity of PDEl family members by an autoinhibitory process has been well documented. Limited proteolysis of PDElA2 fully activates the enzyme, but the Ca2+/calmodulin binding function is retained (41). Likewise, using amino-terminal truncation of PDElA2, a region of sequence [residues 89-98 (-VPSEVBDWLA-)] has been identified that fulfills the criteria of an autoinhibitory domain, i.e., deletion of this sequence activates the enzyme, but Ca2+/calmodulin binding is preserved. Furthermore, other parameters (Vm,, oligomeric structure) in this truncated, activated PDElA are apparently unperturbed. Whether other PDEs contain a well-defined autoinhibitory domain or whether in some instances autoinhibition is effected through more general steric constraints on the enzyme structure remains to be determined. Selective tissue distribution of the various CaM-PDEs has been interpreted to implicate the respective enzymes in specific physiological functions, and the distinctive kinetic and regulatory features of the CaM-PDEs may contribute to their roles in these tissues. The PDEl family has a broad tissue distribution, but it is particularly well represented in testis, heart, and neural tissues, e.g., in cerebellar granule cells as well as in olfactory epithelial cells (138, 139, 155). PDEl is selectively expressed in different areas of the brain, and the particular abundance of PDEl in neurons of the central nervous system suggests that the PDEl family may participate importantly in a variety of neural functions. PDElC is particularly abundant in olfactory cilia and has been suggested to play an important role in the Ca2+ modulation of odorant signaling, perhaps in conjunction with a Ca2+-regulated olfactory adenylyl cyclase (138, 143, 155-157). Proliferating vascular smooth muscle cells contain high levels of PDEl, in contrast to the paucity of this PDE family in nonproliferating cells (158).

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

23

Given the broad tissue distribution of the PDEl family members and their diverse kinetic properties, these enzymes must always be considered as potential participants in signaling processes involving cGMP/cAMP and/or Ca’+-mediated events.

B. PDE2 Family The PDE2 family (also known as the cGMP-stimulated PDE, even though CAMP also stimulates the enzyme) occurs in both cytosolic and membranebound forms: it has a broad tissue distribution, but it is particularly abundant in the central nervous system and adrenal cortex (159-162). These enzymes are typically homodimers with a molecular mass of -240 kDa, with two subunits of -105 kDa each. Sequences of cDNAs derived from several species and tissues suggest that there are at least three isoforms of PDE2 that are thought to be derived from one gene. The respective PDE2 isoforms differ in their amino termini; alternative splicing of the PDESA mRNA is likely to account for the differences (138,162-164). Forms of PDE2 that are localized to the particulate fraction of the cell contain more hydrophobic amino-terminal domains that may specify their membrane association. Members of the PDE2 family are multidomain proteins with a catalytic domain and an amino-terminal cGMP-binding domain (Fig. 8). Interactions contributed by structures within the latter domain are thought to provide for dimerization (45, 165). Cyclic AMP and cGMP are hydrolyzed with similar Vmax values (-150 ~mol/min/mg); the affinity for cGMP is -twofold greater. Although FM the reported Km values vary, they generally fall in the range of -15-30 for cGMP and 30-50 pM for CAMP. Catalysis exhibits positive cooperativity (Hill coefficient ranging from 1.2 to 1.6 for cGMP and 1.6 to 1.9 for CAMP); hydrolysis of either CAMP or cGMP can be markedly stimulated (as much as 50-fold) by the other nucleotide, although cGMP is preferred both as substrate and effector (113, 166-169). The stimulatory effects of CAMP or cGMP are due to an increased affinity for substrate, i.e., a Km effect, with no change in the V_, in the enzyme; activation is mediated through interactions of CAMP or cGMP with the allosteric cyclic nucleotide-binding domains in the amino-terminal regulatory portion of PDE2 (Fig. 8). Occupation of the cyclic nucleotide-binding allosteric sites induces a conformational change in PDE2, converting it to a more active form that is also more sensitive to proteolysis. Studies of the molecular requirements for cyclic nucleotide interaction with either the allosteric cyclic nucleotide-binding sites or the catalytic site of PDEs have been performed using PDEB. Low levels of 3-isobutyl-1-methylxanthine (IBMX), a classical PDE inhibitor, can also stimulate catalytic activity by interacting with the allosteric cyclic nucleotide-binding sites on PDE2 (61, 62, 113,166,170-174). Based on predictions from the amino acid sequence, there are two putative allosteric

24

SHARRON

H. FRANCIS

ET AL.

cyclic nucleotide-binding sites per PDE2 monomer, but the stoichiometry of cGMP binding is -2 cGMP bound per dimer (45,175). The apparent activation constant for cGMP stimulation is substantially lower (-0.2-0.4 FM) than that for CAMP, and cGMP at concentrations as low as lo-7 M can stimulate PDE2 hydrolysis of micromolar levels of CAMP. Therefore, at concentrations of CAMP and cGMP that are estimated to occur in most cells (O.l- 1 t.&), it is predicted that PDE2 would preferentially hydrolyze CAMP. Elevations of cGMP in such conditions would lower CAMP by the action of PDEB, which could provide for cross-talk between these two signaling pathways. The PDE2 associated with the particulate fraction of the bovine brain can be phosphorylated in vitro by PKA, but no functional changes in the enzyme’s properties were detected (160); the PDE2 isolated from bovine heart is not phosphorylated by PKA in vitro (86). Among the group of PDE families that bind cGMP at allosteric sites in their regulatory domains, the role of these cyclic nucleotide-binding sites is best understood in PDES. There is significant experimental evidence to suggest that hydrolysis of CAMP by PDE2 is important in a number of physiological processes, including regulation of Ca2+ channels (176, 177), olfactory signaling pathways (155), platelet aggregation (178), and aldosterone secretion (179, 180). PDE2 is particularly abundant in the adrenals. Adrenocorticotropic hormones such as ACTH induce increases in CAMP, thereby promoting increased synthesis and secretion of aldosterone, which promotes an expansion in the blood volume. Atria1 natriuretic peptide, the release of which is sensitive to blood volume, also binds to these cells and activates a guanylyl cyclase that increases cGMP synthesis; subsequent cGMP binding to cGMP-binding sites in PDE2 is thought to stimulate hydrolysis of CAMP (179). As CAMP is lowered, aldosterone production diminishes, which reduces blood volume and blood pressure. This effect on steroidogenesis is one example in which a cGMP-binding PDE may regulate the cellular response to changes in cyclic nucleotide. Given the broad tissue distribution of the PDE2 family and the capacity of this family to hydrolyze both CAMP and cGMP, the PDE2 family has the potential to play vital regulatory roles in many tissues, including cross-talk between the CAMP and cGMP signaling pathways. In addition, because PDE2 degrades both cGMP and CAMP in vitro, it could also mediate negative feedback regulation of cGMP signaling in certain tissues.

C. PDE3 Family The PDE3 family hydrolyzes both CAMP and cGMP (reported Km values for CAMP and cGMP range from 0.1 to 0.8 FM), albeit with a 2- to lo-fold higher Vmax for CAMP (3-9 p,mol/min/mg); the purified enzyme exhibits normal Michaelis-Menten kinetics (181, 182). Because the two nucleotides

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

25

compete for hydrolysis at the same catalytic site, this family has been dubbed with the misleading name of cGMP-inhibited PDE (cGI-PDE). Cyclic GMP can compete with CAMP hydrolysis with a Ki of -0.2 FM; such a competition could play an important role in certain biological effects because elevation of either cGMP or CAMP might result in increased hydrolysis of that nucleotide while sparing the other (114, 183-186). This would effectively increase the intracellular levels of the latter nucleotide and enhance signaling through that pathway. Members of the PDE3 family are widely distributed dimeric enzymes, are highly susceptible to partial proteolysis during purification (187) and occur as both cytosolic and “peripheral” membranebound species [in both the endoplasmic reticulum and the plasma membrane (188, ISS)]. PDE3 s are particularly abundant in adipose tissue, cardiac muscle, vascular smooth muscle, liver, and platelets (17, 35, 190). The particulate PDE3s from adipocytes, hepatocytes, and platelets are regulated by a number of hormones, including insulin, glucagon, catecholamines, and prostaglandins; the activation induced by these agents ranges from 1.5- to 4-fold and is the result of a change in the Vmax (16, 102, 103, 191-193). Leptin has also been reported to activate the PDE3B in pancreatic beta cells (104) and in hepatocytes (104~). PDE3 plays a central role in a short-term negative feedback loop that modulates intracellular levels of CAMP (191, 1944196). Th is action of PDE3 contributes substantially to the hormonal regulation of glycogenolysis, to the antilipolytic effect of insulin, and to the leptin-induced decrease in insulin secretion from the pancreas (16, 104,191,195,197-201). PDE3 s are targets for the action of many drugs [such #as cilostamide, amrinone, fenoximone, and mihinone (Table I)] that exhibit cardiotonic, vasodilatory, thrombolytic, and antiplatelet aggregation properties (17, 86, 90, 95, 190,202). A mong these inhibitors, cilostamide (IC,5,, -40 nM) is widely used as a specific inhibitor of the PDE3 family. Two PDE3 cDNAs (PDE3A and PDE3B) that are predicted to encode proteins of -122,000-125,000 kDa have been cloned from a number of species (47,203-206). The PDE3A gene product encodes the PDE3 that is abundant in the cardiovascular system, including the myocardium and the vascular smooth muscles. It is also expressed in the smooth muscle of the airways and of the genitourinary and gastrointestinal tracts (207). The PDESB gene product encodes the PDE3 that is abundant in adipose tissue, hepatocytes, spermatocytes, and the renal collecting duct epithelium; the human genes for both PDE3s are located in chromosomal region 11~15 (205, 206). These predicted proteins have very similar domain structures and are clearly distinguished from other PDE families in that each contains a novel insertion of 44 amino acids in the amino-terminal conserved sequence in the catalytic domain (Fig. 1) (203). Th is insertion appears to be unique to PDE3s and is positioned within the Zn 2+-binding motif A (Figs. 1 and 4). Deletion

26

SHARRON

H. FRANCIS

ET AL.

of this insertion produces an inactive PDE (65, 80, 208). The more aminoterminal regulatory region in the PDE3s contains hydrophobic regions that may contribute to membrane association, and a number of consensus phosphorylation sites (-RRXSX-) for PKA are located in this region. PDE3s in adipocytes and hepatocytes are activated by phosphorylation of their regulatory domains by PKA (193, 201, 209, 210). In rat adipocytes treated with either isoproterenol and/or insulin, PDE3 is phosphorylated at a single serine (Ser-302) in a PKA consensus sequence (-RRPSLPC-) (211). However, treatment of adipocytes with both insulin and catecholamines causes phosphorylation and activation of the enzyme that are greater than additive (212). Cross-talk between the two signaling pathways involving upstream regulatory components may account for this apparent disparity. An insulin-stimulated PDE3 kinase that is distinct from PKA has been partially purified from platelets and has been shown to phosphorylate and activate PDES. Some evidence suggests that this kinase may be protein kinase B (213-215). The PDE3 family of enzymes clearly plays a major role in modulating cAMP/cGMP levels in many tissues. This family has a broad tissue distribution, occurs in both the soluble and particulate compartments of the cell, is hormonally regulated, and perhaps most importantly, the enzymes have Km values for CAMP and cGMP that approximate the levels of cyclic nucleotides in cells. Thus, even slight changes in either nucleotide would have the potential to markedly increase the catalytic rate in PDE3s. Last, although the PDE3 family has been convincingly demonstrated to participate in a rapid negative feedback control of CAMP levels that is mediated through PKA, a similar pathway has not been demonstrated for the cGMP pathway. However, the possibility that this type of regulation occurs is entirely feasible.

D. PDE4 Family The PDE4 family is characterized by being highly specific for CAMP as substrate, having a low Km for CAMP (l-3 FM), being insensitive to cGMP and Ca2+/calmodulin, and being potently and specifically inhibited by rolipram and RO 20-1724. It is commonly described as the “rolipram-inhibited CAMP-specific PDE.” The PDE4 family has been the focus of vast research efforts over the years because this family is considered to be a prime target for therapeutic intervention in a number of maladies. A comprehensive overview of the PDE4 family is beyond the scope of this article, but several recent reviews cover this topic in depth (11,14,21,23). The PDE4 family has a broad tissue distribution, is typically expressed in small amounts, and contains a highly diverse number of variants, many of which occur as both soluble and membrane-bound species. In human, rat, and mouse cells, the PDE4 family is encoded by four sep-

CYCLICNUCLEOTIDEPHOSPHODIESTERASES

27

arate genes (PDE4A-4D) located on three chromosomes, and there is high sequence conservation among species. Genomic organization of human and rat PDE4 genes is very similar, suggesting a unique ancestor (23, 134, 216, 217). Use of different promoters, multiple transcription start sites, and alternate splicing in these genes generates a multiplicity of mRNA and protein products for this large PDE family, and regulation of members of the PDE4 family is remarkably complex. PDE4A, PDE4B, PDE4C, and PDE4D genes are differentially spliced to generate more than 15 isoenzymes with aminoterminal heterogeneity that can be generally classified into long (85-110 kDa) and short (68-75 kDa) forms. Products of PDE4 genes show greater than 80% sequence identity over their catalytic regions, although there are segments of sequence in these regions that are unique to each PDE4 gene family. Despite differences in these various PDE4s, they share many of the same features, and genomic sequences of rat PDE4A, PDE4B, and PDE4D show high conservation of intronexon boundaries (218 -225). Mice lacking PDE4D exhibit growth and fertility abnormalities (22%). 1. ROLEOFTHEAMINO-TERMINALSEQUENCESINPDE~S The PDE4 gene family in rat has multiple members that utilize altemative splicing to form proteins with distinct amino-terminal regions. These distinct amino termini affect a number of properties of the PDE4s, including catalytic activity, inhibition of catalysis by rolipram, temperature stability, response to ligands such as phosphatidic acid, and localization to specific subcellular compartments (129,132,133,226-231). The membrane localization of rat PDE4A and PDE4B activity expressed in COS cells depends on the presence of a sequence of some 25 amino acids in the amino-terminal region of the protein (232). One of these (RNPDE4Al) is a membrane-bound PDE that can be solubilized by mild detergent treatment; two other PDE4A varants (RNPDE4A and RNPDE4A6) are present in both the cytosolic and particulate fractions of cells. Houslay and colleagues have shown that the aminoterminal 25 amino acids of RNPDE4Al account for its membrane localization; creation of a chimera between this sequence and the classically cytosolic protein, chloramphenicol acetyltransferase, localized the transferase to the particulate component of the cell (228, 233, 234). lH nuclear magnetic resonance (NMR) analysis of a synthetic peptide representing the amino-terminal 25 amino acids of RNPDE4A reveals an amino-terminal amphipathic a-helix connected by a hinge region to a compact, tryptophan-rich helix (235). The latter helix, which is formed by a hydrophobic dodecapeptide, contains a novel heptapeptide (-PWLVGWW-) that is apparently responsible for the membrane-targeting properties. This targeting apparently involves specific hydrophobic interactions because substitution of these residues by alanine disrupts PDE4 association with the membrane. Howev-

28

SHARRONH.FRANCISETAL.

er, the membrane components that interact with this peptide to anchor the PDE4A to specific regions have not been identified. The functional properties of the different particulate and cytosolic forms of human PDE4B and PDE4D have been described in much detail as well (132, 236). The aminoterminal domains of certain PDE4s may also provide for specific interactions with components of other signaling pathways. The long form of rat PDE4A (RNPDE4A5) in t eracts with the SH3 domain of v-Src by its amino-terminal splice region (237,238). A shorter amino-terminally truncated PDE4A splice variant fails to associate with v-!&c SH3. 2. VARIATIONSINTHEREGULATORYDOMAINFUNCTIONSIN PDE4s In addition to the differences in subcellular localization, the long and short forms of PDE4 differ in the presence or absence of two upstream conserved regions (UCRl and UCRB), which are unique to PDE4s and are located in the amino-terminal portion of these enzymes (Fig. 1). UCRl (approximately 55 amino acids) and UCR2 (approximately 75 amino acids) show no homology to any other known protein sequence; however, these sequences are preserved in PDE4s from evolutionarily distant species, which implies their importance in PDE4 function (216). UCRl and UCR2 differ markedly in sequence and are separated by a short linker region (LRl) of -33 amino acids that shows no homology between PDE4s (217); UCR2 is separated from the catalytic domain by a second linker region (LR2) of lo-28 residues. LR2 is hydrophobic and shows some homology between members of the PDE4 family. Full importance of the UCRl/UCRB region for PDE4 functions has not been established, but several lines of evidence support the likelihood that important regulatory features of PDE4s are contributed by this region. It has been proposed that the linker sequences could influence or provide for specific regulation of UCRl/UCR2 interactions in various PDE4 isoforms (238). Long and short forms of PDE4 can be divided into two groups: (1) those that contain UCRl and UCR2 and (2) th ose in which UCRl and/or portions of UCR2 are absent. To investigate functional significance of these domains, deletion mutagenesis has been peformed for PDE4A (239). Results indicate that catalytic activity of a long form of PDE4A is not significantly affected by amino-terminal truncation to remove either or both UCRs from the sequence. The truncated PDE4A s were less sensitive to rolipram inhibition, were not stereoselective for the enantiomers of rolipram, and no longer contained the high-affinity rolipram binding site. These combined results suggest that alternative splicing of PDE4 mRNA, particularly in the region involving the UCRs and the amino terminus, produces PDE4 isoforms with different potentials for regulation. Evidence has been presented that rolipram

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

29

interacts with two different affinities at a single site on PDE4 and this may relate to different conformational states of the enzyme (67, 101, 240). A number of agents produce a rapid activation of PDE4 (-twofold increase in the VmJ (234, 241, 242). Th is activation of PDE4D provides for short-term feedback regulation of CAMP levels because this enzyme is activated by a CAMP-induced phosphorylation by PKA. The amino-terminal regulatory portion of PDE4D contains two PKA phosphorylation consensus sequences (Ser-13 and Ser-54 in PDE4D3), and Ser-54 is located in the UCRl (109, 111). Site-directed mutagenesis studies also support a crucial role for phosphorylation of Ser-54 for the activation of the enzyme by PKA-catalyzed phosphorylation, and the activated enzyme has a lower IC,, for rolipram (243). Replacement of Glu-53 with alanine blocked the activation of the enzyme despite rapid phosphorylation at Ser-54. These results suggest that disruption of hydrogen bonding in this region of the protein is involved in the effects manifested following phosphorylation (243). Persistent long-term treatment with hormones that elevate CAMP promotes an increase in the level of transcription of PDE4Dl and PDE4D2 variants and an increase in PDE4D protein (116, 120, 244). There has been a sustained and widespread interest in PDE4 for many years. Although low in abundance, the PDE4 family is homologous to the products of the dunce gene that accounts for learning defects in Drosophila (245) and is implicated as a major participant in signaling processes in immunocompetent and proinflammatory cells. The relatively low Km of this enzyme for CAMP, its high specificity for CAMP, its molecular complexity, and its abundance in neural and endocrine tissues implicate it as a major determinant in the modulation of CAMP levels in these tissues.

E. PDES Family The PDE5 family belongs to the group of PDEs that contain allosteric cGMP-binding sites in their regulatory domains (Figs. 1 and 4); it is known as the cGMP-binding cGMP-specific PDE. In this family both the allosteric sites and the catalytic site are highly specific for cGMP PDE5 is a homodimer of -99-kDa subunits and is abundant primarily as a cytosolic enzyme in lung, platelets, vascular smooth muscle, and kidney (246, 247, 247a). PDE5AI gene product was first cloned from bovine tissues (248). Subsequently cDNAs (PDE5A1, PDESA2, and PDE5A3) representing splice variants of PDE5A gene have been found. Although the tissue distribution of these differs (249, 250, 25Oa), the catalytic properties and inhibitor profiles do not. Bovine and human PDE5Als are 96% identical after excluding an insertion of a 10 amino-acid glutamine-rich segment near the amino terminus in the human enzyme. PDESAI, PDE5A2, and PDESA3 cDNAs differ only at the 5’ end and encode proteins that are 100 kDa (875 amino acids), 95 kDa (833 amino

30

SHARRONH. FRANCISET AL.

acids), and 95 kDa (823 amino acids), respectively. Human PDE5A maps to chromosome 4q25-27 (250-252).

gene

1. CATALYTICDOMAIN The catalytic site of PDE5 has significantly higher affinity for cGMP (Km - l-5 pv, although CAMP (Km 2 300 FM) is hydrolyzed with a higher V may.PDE5 was the first mammalian PDE for which Zn2+ was shown to potently support catalysis and that specifically binds Zn2+ with a stoichiometry of -3 mol Zn2+ per PDE5 monomer (72). Although Mn2+ and Mg2+ also support catalysis, Zn2+ is significantly more potent. Site-directed mutagenesis has been used to introduce alanine substitutions at 23 conserved residues in the catalytic domain of PDE5 (Fig. 4B). Under the conditions used for assaying these mutant PDE5s, substitution of His-603 (H-l), His-607 (H-2), His-643 (H-3), His-647 (H-4), Glu-672, Asp-714, and Asp-754 produced marked changes in kcat (Fig. 6); therefore, all of these residues appear to be highly important for efficient catalysis. The most dramatic decrease in kcat occurred with substitutions at either His-643 (H-3) or Asp-754, which decreased free energy of binding -4.5 kcal/mol for each; this is within the range predicted for loss of a hydrogen bond involving a charged residue. These PDE5 catalytic domain mutants were initially studied using standard assay conditions that include Mg2+ as supporting cation (71). More recent studies that in many instances, substitution of Mn2+ for Mg2+ dramatically increases the catalytic rate in certain mutants. These studies reveal that the mutations at His-607 (H-2), His-643 (H-3), and Glu-672 impaired metal binding to the PDE5, which then manifested as dramatic changes in kcat (76~). The Km for cGMP was most profoundly altered by alanine substitutions at Tyr-602 and Glu-775. The Km of the Y602F mutant was actually fourfold improved compared to that of the wild-type enzyme. These results suggest that Tyr-602 forms important stacking interactions in the substrate-binding site of PDE5 and is likely to serve a similar role in other PDEs. An acidic residue is important at the Glu-775 position and is suggested to interact with the N-l nitrogen of the guanine ring (71). The two putative Zn 2+-binding motifs that are conserved in mammalian PDEs include His-603, His-607, Glu-632, His-643, His-647, and Glu-6 72 in PDE5 (Fig. 4). Five of these residues were implicated as critical for efficient catalysis (Fig. 6). The only residue in this group that seems to be nonessential for catalysis is Glu-632; this residue is not conserved in the class I yeast PDE, so this finding is not surprising (27). However, the results support an important role for His-603 (H-l), His-607 (H-2), His-643 (H-3), His-647 (H-4), Glu-672, Asp-714, and Asp-754 in catalysis, and one possible function of this combination of residues is to provide metal-binding sites. The results establish that mutations in either Zn 2+-binding motif are profoundly delete-

CYCLIC NUCLEOTIDE PHOSPHODIESTERASES

31

rious to catalysis and that a single Zn 2+-binding motif is insufficient to support normal catalysis in PDEs. With sufficiently high levels of the appropriate metal, significant activity can be largely restored in certain mutants [His607 (H-2), His-643 (H-3), Glu-6721, but not in others (76~). Results of these mutagenesis studies demonstrate that residues located in each of the three conserved subdomains of the catalytic domains of PDEs provide for efficient catalysis and for establishing the Km of PDE5 for cGMP. This distribution suggests that the active site may be formed at the interface between these blocks of conserved residues. Thus, catalytic and substratebinding components are overlapping and utilize several subdomains. 2. INHIBITOR SPECIFICITY AND INTERACTION PDE5 is potently inhibited by sildenafil (IC,, - l-4 nM), DMPPO (I% -3 nA4), UK-122764 (IC,s, - 5 nM), WIN 65579 (IC,, - 2-3 nM), zaprinast (IC,, - 300 nM), and dipyridamole (IC,, - 1 k,M) (3, 93, 96, 253). Inhibition by sildenafil, zaprinast, and IBMX shows competitive kinetics, and the inhibitors interact with the catalytic site of PDE5 in a mutually exclusive manner (100). The IC,, values of sildenafil and IBMX for isolated catalytic domain of PDE5 are essentially the same as for the wild-type enzyme (49u). Among the inhibitors that have been examined using a range of PDE5 catalytic domain mutants, sildenafil exhibits a pattern of changes in the IC,, most similar to that found for the substrate, cGMP. This implies similar interactions of cGMP and the inhibitors with the catalytic domain. However, the potencies of a number of inhibitors of PDE5 (sildenafil, DMPPO, UK122764, and zaprinast) are significantly higher than the affinity for cGMP. Thus, these inhibitors must form additional novel contacts with PDE5 to provide for the tighter binding (100). 3. ALLOSTERIC cGMP-BINDING SITES The two homologous allosteric cGMP-binding sites (a and b) are kinetically distinct, and cGMP occupation of both sites is required for phosphorylation and activation of bovine PDE5Al at Ser.92 by either PKG or PKA (Fig. 1) (247u, 254-255~). Whether these sites have other effects on PDE5 function is unknown. The demonstration that both cGMP-binding sites are required for phosphorylation of Ser-92 provided the first evidence that both cGMP-binding sites in PDE5 are important for a structural change in this enzyme. Furthermore, occupation of the catalytic site by IBMX, cyclic nucleotide analogs, or substrate promotes cGMP-binding to the allosteric sites (3, 60, 247). Th us, the allosteric cGMP-binding sites clearly influence interdomain communication in PDE5. Phosphorylation of Ser-92 in PDE5A by PKG proceeds with at least a lofold higher rate than that by PKA. The primary sequence surrounding Ser-

32

SHARRON

H. FRANCIS

ET AL.

92 provides a phosphorylation site that is relatively selective for PKG over PKA (256). Phosphorylation of the PDE5 by PKG has been shown to cause an increase in the activity of the enzyme using low substrate concentrations with no discernible change in IC,, for the inhibitors sildenafil and zaprinast (25%). Using a partially purified PDE5, Bums and Pyne reported that phosphorylation of PDE5 by addition of exogenous PKA increased the V,,, of the enzyme and weakened the potency of inhibition of the enzyme by zaprinast (257). However, the site phosphorylated in these studies was not identified, and the phosphorylation was not dependent on cGMP binding to the allosteric sites on the enzyme, indicating that the differing results of the respective phosphorylation studies may result from different processes. Wyatt et cd. reported a correlation between phosphorylation of PDE5 and increased catalytic activity in cultured vascular smooth muscle cells treated with atria1 natriuretic peptide (107). Last, Lochhead and co-workers reported that activation of partially purified PDE5 by PKA is modulated by interaction with low-molecular-weight proteins (- 14 and 18 kDa) that resemble the inhibitory y-subunits present in the photoreceptor PDEs (258). However, other workers have been unable to demonstrate an interaction of purified -y-subunit with PDE5 or with a chimeric PDEGo/PDE5 (54). The allosteric cGMP-binding sites (a and b) of PDE5 are highly specific for cGMP, and at least a portion of the cGMP/cAMP selectivity may result from the ionization state of a specific aspartic acid residue in these sites (Fig. 8) (259); the more amino-terminal site a exhibits higher binding affinity compared to site b (t,,, - 0.26 and 1.0 hr-r, respectively) (260). Half-maximal saturation of the allosteric cGMP-binding sites is achieved at -0.2-2 pM cGMP, but the stoichiometry of cGMP binding approaches one cGMP per PDE5 monomer. The reason for this low stoichiometry of binding is unclear given the presence of two homologous sequences that provide for cGMP binding to allosteric sites in each PDE5A monomer and the kinetically distinct binding sites. However, a similarly low stoichiometry of binding has been observed for other families of cGMP-binding PDEs, i.e., PDE2 and PDEG, so it is possible that the actual binding stoichiometry for these cGMPbinding PDEs is one cGMP per PDE5 monomer. The allosteric cGMP-binding sites in cGMP-binding PDEs represent a newly recognized class of cyclic nucleotide-binding site that is distinct from those sites in the catabolite activator protein (CAP) family, i.e., PKA, PKG, and cyclic nucleotide-gated channels (40). The cGMP-binding sites in PDEs have no sequence homology with CAP-related proteins, and the analog specificities of the PDE sites differ markedly from those sites in the CAP-related proteins. Site-directed mutagenesis of residues in the cGMP-binding sites of PDE5 has established that a conserved -N(K/R)XnD- motif is critical for cGMP binding (260,261). Th is motif (Fig. 8) resembles the canonical NKXD

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

33

motif that participates in binding guanine in GTP-binding proteins (262). Because the NKXnD motif is conserved in the allosteric cGMP-binding sites of other cGMP-binding PDEs, it is likely that these residues will prove to be important in cGMP binding in the PDE2 and PDE6 families. Based on changes in free energy of binding of cGMP on mutation of these conserved residues, on studies using cGMP analogs, and on the known functions of the respective residues in the NKXD motif in GTP-binding proteins, a functional role for each residue in the NKXnD motif in PDEs has been proposed (261). In PDE5A, disruption of cGMP binding in either site by site-directed mutagenesis had no apparent effect on the function of the other cGMP-binding site or on catalytic function. This result is further supported by studies of the properties of a truncation mutant containing the PDE5 catalytic domain (49~). This catalytically active PDE5 fragment was shown to be monomeric and to exhibit catalytic properties that are comparable to those of native PDE5. However, a subtle, but physiologically relevant, difference in catalytic activity of this mutant could not be ruled out. The remarkable therapeutic success of sildenafil (Viagra), a highly specific inhibitor of PDE5, in treating male erectile dysfunction has generated much interest in this enzyme, and the potential for other drugs that might target PDE5 is now well appreciated (93, 263). The tissue distribution of the PDE5 family is relatively restricted compared to some other PDEs; this makes the use of agents that selectively target PDE5 and particular physiological processes somewhat more practical. Given the abundance of PDE5 along with other components of the cGMP signaling pathway in smooth muscle, it is likely that regulation of PDE5 activity is a central feature in modulating contractile tone in the vasculature, airways, and gastrointestinal smooth muscle. Although significant progress has been made in defining the molecular features that provide for the properties of PDE5, definitive data regarding the catalytic mechanism and the manner in which substrates and inhibitors interact with the enzyme are still very limited.

F.PDE6 Family Members of the PDE6 family are highly specific for cGMP as substrate and are found in the rod and cone cells of the vertebrate retina, where they function as key participants in the visual response to light. The population of PDEs in the photoreceptor cells is almost exclusively restricted to the PDE6 family. The PDE6 family is characterized by (1) its high catalytic specificity for cGMP, (2) th e p resence of allosteric cGMP-binding sites that are located in the amino-terminal half of the enzyme, (3) a high kCat,and (4) the fact that this is the only PDE family known to be regulated by G proteins. Numerous visual defects have been traced to mutations affecting the various protein subunits of the rod and cone PDEs (1, 9,18,19,108,264,265).

34

SHARRONH.FRANCISETAL.

1. STRUCTURALFEATURES The PDE6 family is a member of the cGMP-binding subgroup of PDEs because enzymes of this family bind cGMP with high specificity at allosteric sites in their amino-terminal regulatory domains. PDE6 family is closely related to the PDE5 family, and the allosteric cGMP-binding sites are homologous with those in PDEs 2, 5, and 10. Photoreceptor PDEs occur as heterotetramers that contain two dimerized homologous catalytic subunits [either 01and B (rod PDEG) or two (Y’(cone PDEG)]; each of these three catalytic subunits is the product of separate genes. The predominant form of rod PDE holoenzyme contains one PDE-ol and one PDE-B subunit (-99 and 98 kDa, respectively), although minor components involving PDE-ecu. and PDE-BB have been detected; cone PDE6 is a homodimer of two catalytic PDE-a’ subunits (-94 kDa) (266-268). In each instance, two inhibitory y-subunits of -9.7 kDa that are specific for either rod PDE or cone PDE are complexed with the respective PDE dimers, and the affinities of these y-subunit interactions with either rod or cone catalytic subunits differ (269, 270). Th e majority of PDE6 activity is associated with the membrane fraction of the cell. This may result from posttranslational modification of the PDE6 subunits in a CAAX sequence at their respective carboxy1 termini; PDE6o. is farnesylated and PDEGB is geranylgeranylated (271,272). The soluble portion of PDE6 contains an additional subunit, i.e., the S-subunit (-15,000 Da) (273-275). 2. ~LOSTERIC

CAMP-BINDING

SITES

PDE6 family members bind cGMP with high selectivity and affinity to the allosteric cGMP-binding sites (276-279). Both allosteric cGMP-binding sites contribute to cGMP binding by the regulatory portion of PDEG, and the two binding sites appear to be coupled (54, 280). Like other cGMP-binding PDEs, the best estimate of cGMP-binding stoichiometry is one cGMP bound per PDE6 monomer. Affinity for cGMP binding varies substantially (Kd 60 nM-1 (IM)depending on the activity state of PDE6 (278, 281), and the dynamics of the protein-protein interactions that determine overall activity of PDE6 appear to be modulated by state of occupation of the cGMP-binding allosteric sites (278, 281, 282). Cyclic GMP binding to the allosteric sites is increased by binding of Py to the enzyme, and the polycationic region of Py (residues 24-45) accounts for this increased affinity (283, 284). Activation of PDE6 by transducin causes an initial loss of cGMP-binding affinity of -lO-fold, and subsequent removal of Py an&or delta subunits from the catalytic subunits decreases affinity further (278, 284~). Although cGMP binding to allosteric cGMP-binding sites in PDE6 does not appear to affect catalytic function directly, occupation of these sites alters binding affinity of PDE-CYB for Py and thereby modulates catalytic function.

CYCLICNUCLEOTIDEPHOSPHODIESTERASES

35

3. PDE6 INVISUALTRANSDUCTION In dark-adapted rods and cones, cellular cGMP (-60 FM) is quite high, but a significant portion is apparently bound to allosteric cGMP-binding sites of the photoreceptor PDEs. Concentration of unbound cGMP (-5 t&I) is still sufficient to provide for cGMP occupation of the cGMP-gated cation channels; under these conditions, the channel exists in an open state, thereby facilitating inward conductance of Naf/Ca2+. The membrane potential is thus relatively depolarized. Absorption of light converts rhodopsin to a photoactivated rhodopsin, which, in turn, activates a heterotrimeric GTP-binding protein, transducin (265). The GTP-bound o-subunit of transducin (CIJ interacts with the inhibitory y-subunits of PDEG, thereby relieving inhibition and causing a rapid increase in PDE catalytic activity. GDP-bound ot can bind to the P-y, but with markedly lower affinity (284). The increased PDE activity rapidly lowers intracellular cGMP, such that cGMP dissociating from the channel is not replaced. Consequently, channel pores then close, thereby abolishing the depolarizing influx of cations into the photoreceptor. However, activity of a Na+ exchanger persists, so that the membrane potential of the cell becomes increasingly hyperpolarized. This provides the signal that is perceived as light in higher integrative centers. Specifics of this pathway continue to be studied intensely. Amplification of a light signal occurs at the levels of the G protein, the PDE, and the cGMP-gated channel. A single photon converts rhodopsin into photoactivated rhodopsin, which promotes GDP/GTP exchange on as many as 500 transducins in 1 sec. Hundreds of o-GTPs dissociate from transducin free ot-GTP concomplexes, reaching a local concentration as high as 80 p,M; centration is estimated to approximate available binding sites for (w,-GTP on the PDE-Py complex. The resulting au,-GTP forms a tightly membranebound complex with PDE, thereby activating the enzyme. Two c-w,-GTPscan be bound per PDE, but whether one or two o,GTPs are required to activate each PDE molecule fully is not clear (285). However, the physiologically activated PDE is most likely to be complexed with two CY,-GTPs. Estimates of affinity with which o,GTP binds to PDE vary widely, and binding is suggested to be positively cooperative; intrinsic fluorescence studies of the interaction between ot and PDEy determined the binding constant to be co.1 nM (286). The o,GTP activation occurs in a membrane-bound PDE complex, and a number of studies have shown that presence of membranes also enhances the process (287, 288). Each activated PDE molecule then hydrolyzes perhaps 1000 cGMP molecules to 5’-GMP. The maximal activity of rod PDE is estimated at -3000-4000 cGMP/sec/PDE molecule, and ‘Y+ GTP-activated PDE is likely to reach -90% of this rate due to the positive cooperativity of CY~-GTPbinding. Lowered cGMP causes closure of the cGMP-gated channel, and the influx of thousands of cations is blocked.

36

SHARRONH.FRANCISETAL.

4. CHARACTERISTICSOFTHEINTERACTIONSTHATCONTRIBUTE TOTHEREGULATIONOFPDEGBYPDE-?/SUBUNIT ANDTHETRANSDUCINIX~-GTP/GDP Interactions that occur between PDE-v and PDE6 catalytic subunits, and transducin o-subunit, are central to regulation of photoreceptor PDE activity. Myriad advances have been made regarding the complex mechanisms that contribute to modulation of the interactions of these and other components both in the dark, in the acute response to light, and in events associated with light adaptation. Furthermore, the number of proteins participating in modulating this pathway has progressively increased (280, 285, 289-293). Extensive experimental evidence suggests that the whole process involves direct contacts between Py and PDE6 catalytic subunits, between transducin o-subunit, ot, and Py, and between ot and PDE catalytic subunits (195, 291, 294,295). More recently, proteins from the RGS family have been shown to interact with o,-GTP to stimulate GTP hydrolysis (293, 296-298). The inhibitory y-subunits interact with PDE-c+ with a very high affinity (Kd < 10 100 PM) to maintain the enzyme in an inactive state in absence of light. The high affinity of interaction between Py and PDE-c@ is provided by at least two subdomains on Py (54,283,299-302). One set of contacts involves the amino-terminal portion of a centrally located polycationic region between Py residues 24-45, which interacts with a region in PDE-o subunit (residues 46 l- 553) that is located between the catalytic domain and the cGMP-binding domain. A synthetic peptide corresponding to residues 517-541 of the PDE-a subunit blocks inhibition of PDE activity by Py and also competes with interaction between Py and qGTP. A second subdomain in Py, the carboxyl-terminal5 -7 amino acids, also interacts with the PDE-of3, and this contact is critical to effect the inhibition of catalysis. Carboxyl-terminal deletion of these residues renders Py ineffective, and peptides corresponding to the carboxy-terminal region of Py inhibit PDE catalysis (299). Studies have shown that the carboxy-terminal 5-7 amino acids of Py interact directly with a site (residues 751-763) in the catalytic domain of one of the catalytic subunits of rod PDE, i.e., PCL Using a fluorescent probe attached to the carboxyl terminus of Py, it was shown that K, for this interaction was -4 nM and a competitive inhibitor, zaprinast, displaced the probe from the interaction on PoyP. Furthermore the carboxyl terminus of Py could be directly cross-linked with a specific region in the catalytic site of PDE6 near an NKXD motif, thought to provide for specific interaction with the guanine of cGMP (303). These studies suggest that the carboxyl terminus of Py inhibits photoreceptor PDE by physically blocking access of cGMP to the catalytic site. Alternatively, Py could cause a conformational change near the catalytic site that would block catalysis.

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

37

Interaction between the Py and transducin or,GTP complex occurs through two major regions of contact. On Py this involves the carboxyl-terminal portion of the Py polycationic region and a segment surrounding Cys-68 in the carboxyl-terminal region (Py 63-76) (54, 284, 290, 299). Studies involving modification of several residues, including two an&nines (Arg-33 and -36) and a threonine (Thr-35) located in the polycationic region of Py, support the importance of this region in forming contacts with o+, Both arginines can be covalently modified in the absence of ot, but when complexed with ot (GTPor GDP-bound forms), they appear to be masked (304). Phosphorylation of Thr-35 in Py by PKA functionally alters Py so that its potency of inhibition of PDE is increased, and its potency of interaction with the ot is also altered (305). Most recently a highly important role has emerged for participation of RGS proteins in photoreceptor signaling. The RGS proteins interact with otGTP to increase the GTPase activity and thereby increase responsiveness of this whole pathway (280, 293, 296-298, 306-308). Evidence has been presented that supports a direct interaction between ot and catalytic subunits of PDE6 as well as with a specific RGS protein (291, 309, 310). The structure of ot contains a GTP-binding domain and an o-helical domain (HD); th ere is now evidence that a 22-amino acid sequence in the HD region of (3~~ interacts with catalytic subunits of PDEG, a site that is distinct from the site on PDE-cxB that interacts with Py (291), and that this interaction facilitates interaction of the at-GTF-binding domain with Py, thereby enhancing activation of PDE6 (291, 309). It has been proposed that this additional interaction of o+ with catalytic subunits of PDE6 may contribute to the high efficiency of PDE activation/inactivation. The investigation of the proteins involved in determining the visual phototransduction pathway has been unrelenting, and despite the monumental amount of effort we still have only a general picture of the molecular events that provide for this highly complex process. However, the major advances that have been made in studying this pathway have yielded important insights into the central role of PDE6 in this system and into the basic properties of the PDE superfamily, as well as the complexity of the controls that can overlay a relatively simple catalytic event.

G. PDE7 Family The PDE7 family is characterized by having a high specificity for CAMP (K,,, = 0.2-l ~_LLM) an d resistance to inhibition by a variety of common PDE inhibitors, including rolipram, zaprinast, Q-(2-hydroxy-3-nonyl)adenine (EHNA), Ro 20-1724, and enoximone. There are no known PDE7-selective inhibitors, but PDE7B is inhibited by IBMX (IC,, -2 t_J4), dipyridamole (IC,, = 1.5 PM) (313~2,b). PDE7 catalytic (I%, -2 FM), and SCH51866 activity is unaffected by either Ca’+/calmodulin of cGMP By comparison of

38

SHARRON

H. FRANCIS

ET AL.

amino acid sequences, PDE7 family is most closely related to PDE4 family, but the similarity is modest. Two cDNA clones (PDE7Al and PDE7A2) representing variants of the PDE7A gene and one cDNA clone representing PDE7B have been isolated (311-313b). Amino acid sequence of PDE7B is - 70% identical to PDE7A, and PDE7A variants differ in their amino termini. These variants arise through alternative mRNA splicing with the divergence occurring just amino terminal to the second conserved -RRGAIS- sequence: PDE7A2 has one of these sequences beginning at Arg-21, whereas PDE7Al has two copies, one beginning at Arg-2 1 and the second at Arg-45. Protein products of PDE’IAl, PDE7A2, and PDE7B are 481-, 456. and 446amino acid proteins (-57, 50, and 52 kDa), respectively. PDE7Al was detected in both cytosolic and particulate fractions, whereas PDE7A2 is primarily associated with the particulate fraction in skeletal and cardiac muscle. Proteins of the predicted sizes are detected in Sf9 cells overexpressing the respective cDNAs, and a PDE activity consistent with that of PDE7Al has been detected in adult skeletal muscle. PDE7Al is particularly abundant in human lymphoid tissue (126, 311); in several cell lines of T lymphocytes the PDE7Al represents -40% of total CAMP PDE activity. PDE7A mRNA is broadly distributed in fetal tissues, but in some instances PDE7 protein levels do not correlate well with PDE7 mRNA content. Although PDE7 mRNA is abundant in adult skeletal muscle, T lymphocytes, and B lymphocytes, PDE7 protein and activity are readily measurable only in the T lymphocytes (311). The basis for this disparity and the physiological relevance of such a difference is unclear, although it has been suggested that PDE7A2 predominates in skeletal muscle and is largely particulate. Increased level of PDE7 in T cell lymphocytes is implicated as a major determinant in T cell activation and proliferation (126). PDE7B transcripts are abundant in the putamen, caudate nucleus, heart, skeletal muscle, and pancreas (311-313b).

H. PDE8 Family The PDE8 family was discovered only recently and has been descriptively dubbed as the high-affinity CAMP-specific and IBMX-insensitive PDE (36, 38). PDE8 has a very low Km for CAMP (0.06-0.15 pJ4 compared to a Km of 124 ~.LLM for cGMP). Cyclic GMP at concentrations as high as 100 p~J4 do not significantly inhibit CAMP catalytic activity. When compared to PDE4A, PDE8 has a 40-fold higher affinity for CAMP, but PDE4A has a lo-fold higher V,,. PDE8 is a soluble PDE with a predicted sequence of 823 amino acids. The catalytic domain of PDE8 contains 22 of the 24 invariant amino acids conserved among the seven earlier recognized PDE families, including the two histidine-rich metal-binding motifs. Although PDES is inhibited potently by dipyridamole (IC,, - 5 k&f), it is notably insensitive

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

39

to a number of other well-known PDE inhibitors, including IBMX, zaprinast, rolipram, vinpocetine, and SKF-94 120. In the catalytic domain, the primary sequence is most similar to those of the PDE4 and PDE7 families (39 and 34% identity, respectively) and least similar to those of PDE5 and PDE6 (23 and 200/o,respectively). The PDE8 mRNA is a single species of 4.5 kb, is expressed widely, and appears to be most abundant in testis, ovary, small intestine, and colon. However, the amount of PDES protein in these tissues is relatively low. Maximum catalytic activity of PDE8 requires 1 mM Mn2+ or Mg”+, whereas Ca2+ (100 kh’l) is significantly less effective (20% maximum).

I. PDE9 Family The PDE9 family is also a recently recognized family and is described as a very high-affinity cGMP-specific PDE. This PDE family has been cloned from both murine and human tissues, and four different PDE9 mRNA transcripts have been identified (PDESAl, PDE9A2, PDE9A3, and PDE9A4); these are produced as a result of alternative splicing of 5’ exons. The predicted protein products from these transcripts would encode proteins of 593, 533, 466, and 465 amino acids (37, 39, 314). The primary structure of the catalytic domains of these PDE9s is most similar to that of PDE8 (34%) and least similar to that of PDE5A (28%). Of the 22 residues considered to be invariant in other mammalian PDE families, 21 residues are present in the PDE9 catalytic domain. The PDESA mRNA (-2 kb) is widely expressed and is particularly abundant in spleen, intestine, kidney, heart, and brain. However, the abundance of PDE9 protein and activity in tissues has not been assessed, so the possible involvement of the PDE9 family in physiological processes is still unclear. In humans the PDESA gene contains 20 exons that extend over I22 kb; it maps to chromosome 21q22.3, a region of the genome that is associated with a number of medical disorders (314). The PDE9 family is highly selective for cGMP and has an affinity for cGMP that is 40-170 times higher than that of the two other families of cGMP-selective PDEs, PDE5 and PDEG. PDE9 has been expressed in insect cells and in COS-7 cells and has a Km of 70-170 nA4 for cGMP versus 230 $!4 for CAMP; the Vmax for cGMP is -e-fold greater than PDE4 for CAMP. Due to the high affinity of PDE9 for cGMP and the broad tissue distribution of the mRNA, this PDE family has been proposed to participate in maintaining basal cGMP levels in the cell (3 7). In the concentration range of l10 mM Mn2+ or Mg 2+, the catalytic rate in the presence of Mn”+ is twice that of comparable concentrations of Mg 2+. PDE9 is inhibited by zaprinast (IC,5, - 35 ~~,n/l) and by PDEs 1 and 5 inhibitor SCH51866 (IC,, - 1.6 $t4), but is insensitive to a wide range of other PDE inhibitors, including sildenafil, dipyridamole, IBMX, rolipram, vinpocetine, and SKF-94 120.

40

SHARRON

H. FRANCIS

ET AL.

J. PDElO Family PDElO is dual substrate specificity PDE family. There are at least two variants (PDElOAl and PDElOA2) that have 766 residues in common (315318). Although Km for CAMP is much lower than that for cGMP (0.05 and 3 @4, respectively), Vm, for cGMP is -5-fold greater than that for CAMP. Both human, mouse, and rat PDElO clones have been isolated. Predicted sequences include a region that is homologous to the conserved catalytic domain found in other PDEs, but there is less than 50% sequence identity with catalytic domains of other PDE families. Predicted sequence of PDElOA also contains a segment that is homologous to the cGMP-binding domain in cGMP-binding PDEs (PDES, PDE5, and PDEG), but cGMP binding to PDElO has not yet been demonstrated. Mouse and rat PDElOA encodes proteins of 779 amino acids and 794, respectively (88,516 Da) (315, 316), and a novel form in rat has an amino terminus that is distinct from the human enzymes (317). The mRNA for mouse PDElOA is most abundant in brain and testis. Human PDElO maps to chromosome 6q26-27 (316). Based on Northem blot analysis, the human enzyme shows a broad tissue distribution.

K. PDEll Family A PDEll family has recently been recognized (Fig. l), and human PDEllAl has been cloned, expressed and characterized (318). The cDNA for PDEllAl encodes an open reading frame for a 56-kD protein (490 amino acids), and the sequence of its catalytic domain is most like that of PDE5 (-50%). The regulatory domain of PDEllAl contains a single sequence that is homologous to the two cGMP-binding motifs located in the regulatory domains of PDEs 2, 5, 6, and 10, although cGMP-binding to PDEll has not been demonstrated. This is the first recognized PDE to have a single such site in the regulatory region, and no effect of either cGMP or CAMP to stimulate hydrolysis of the other nucleotide has been observed. Investigators conjecture that this domain may bind either cGMP or perhaps another unknown ligand to regulate function. PDEll Al hydrolyzes both CAMP and cGMP with similar V,= values and similar affinities, with Km values of 0.5 and 1 k.M, respectively. Based on these kinetic values, PDEll may contribute significantly to modulating both CAMP and cGMP in the cell. PDEll activity is inhibited by IBMX (IC,, - 50 FM), and among other PDE inhibitors tested, dipyridamole is the most potent compound (IC,, - 0.4 I.LM) is not inhibited by either followed by zaprinast (IC,, - 12 yM). PDEllAl milrinone or rolipram, selective inhibitors for PDE3 and PDE4, respectively. There are at least three mRNA transcripts (10.5,8.5, and 6 kb, respectively) for PDEllA, suggesting that there may be additional isoforms in this family, and based on RNA expression patterns, PDEll may have a broad tissue

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

41

distribution. Immunoblots of human tissues detect three protein bands (78, 65, and 56 kD); the smallest form, which co-migrates with recombinant PDEllAl, is abundant in prostate gland, while this form, as well as two additional protein bands, are found in skeletal muscle.

VIII. Concluding Remarks Appreciation of the complexity of the PDE superfamily and its importance in regulating physiological processes continues to increase. The diverse features of the various families of PDEs emphasize the potential for selective regulation and function, and it is clear that PDEs provide a critical component of cyclic nucleotide signaling in all tissues. Exploiting the novel and shared features of these enzymes for developing improved therapies for a number of maladies continues to be a major challenge in this field. An improved understanding of the molecular basis for catalytic function and regulation of these enzymes will undoubtedly provide the framework for success in systematically approaching this goal.

ACKNOWLEDGMENTS Supported by NIH Grants GM41269 (JDC) and DK40029 (JDC), and by the American Heart Association (SHF).

REFERENCES 1. J. B. Hurley, Annu. Rev. Physiol. 49, 793-812 (1987). 2. S. J. Strada, P. A Kithas, M. E. Whalin, and W. J. Thompson, Adv. Exp Med. Biol. 255, 409-423 (1989). 3. S. H. Francis, M. K. Thomas, and J. D. Corbin, In “Cyclic Nucleotide Phosphodiesterases: Structure, Regulation, and Drug Action” (J. Beavo and M. D. Houslay, eds.), pp. 117-140. Wiley, New York, 1990. 4. V. C. Manganiello, C. J. Smith, E. Degerman, and P Belfrage, In “Cyclic Nucleotide Phosphodiesterases: Structure, Regulation and Drug Action” (J. Beavo and M. Houslay, eds.), pp. 87-109. Wiley, New York, 1990. 5. W. J. Thompson, Pharmacol. Z’her. 51,13-33 (1991). 6. L. Stryer,]. Biol. Chem. 266,10711-10714 (1991). 7. M. Conti, S. L. Jin, L. Monaco, D. R. Repaske, and J. V. Swinnen, Endow. Rev. 12,218234 (1991). 8. W. J. Thompson, V. R. Lavis, M. E. Whalin, and S. J. Strada, Adv. Second Messenger Phosphoprotein Research 25, 165-184 (1992). 9. J. B. Hurley,J. Bioenerg. Biomemhr. 24,219-226 (1992).

42

SHARRON

H. FRANCIS

ET AL.

10. Z. Wu, R. K. Sharma, and J. H. Wang, Adv. Second Messenger Phosphoprotein Res. 25, 29-43 (1992). 11. T. J. Torphy, W. E. J. DeWolf, D. W. Green, and G. P. Livi, Agents Actions Suppl. 43,5171 (1993). 12. V. C. ManganieIIo, M. Tair~ E. Degerman, and P. Belfrage, Cell. Signal. 7,445-455 (1995). 13. V. C. Manganiello, T Murata, M. Taira, P. Belfrage, and E. Degerman, Arch. Biochem. Biophys. 322,1-13 (1995). 14. M. Conti, G. Nemoz, C. Sette, and E. Vicini, Endocr. Rev. 16,370-389 (1995). 15. J. A. Beavo, Physiol. Rev. 75, 725-748 (1995). 16. V. C. Manganiello, E. Degerman, M. Taira, T Kono, and P. Belfrage, Cum. Top. Cell. Reg. 34,63-100 (1996). (1996). 1% J. B. Polson and S. J. Strada, Annu. Rev. Phurmucol. Toxicol. 36,403-427 (1996). 18. L. Stryer, Proc. N&Z. Acud. Sci. U.S.A. 93,557-559 19. D. Baylor, Proc. N&Z. Acad. Sci. U.S.A. 93,560-565 (1996). (1997). 20. E. Degermann,P. Belfrage, andV. C. ManganieIIo,J. BioZ. Chem. 272,6823-6826 (1997). 21. M. D. Houslay and G. MiUigan, Trends Biochem. Sci. 22,217-224 22. E. Degerman, T. R. Landstrsm, J. Wijkander, L. S. Holst, F. Ahmad, P. Belfrage, and V. ManganieIIo, Methods 14,43-53 (1998). (1998). 23. M. D. Houslay, M. Sullivan, and G. B. Bolger, Adz;. Pharmucol. (N.X:) 44,225-342 (1999). 24. T. I? Dousa, Kidney Int. 55,29-62 25. T W. Rail and E. W. Sutherland,J. BioZ. Chem. 232,1065-1076 (1958). (1962). 26. R. W. Butcher and E. W. Sutherland,J. BioZ. Chem. 237,1244-1250 27. H. Charbonneau, In “Cyclic Nucleotide Phosphodiesterases: Structure, Regulation and Drug Action” (J. Beavo and M. D. Houslay, eds.), pp. 267-296, Wiley, New York, 1990. 28. M. Koyanagi, H. Suga, D. Hoshiyama, K. Ono, N. Iwabe, K. Kuma, and T. Miyata, FEBS Leti. 436,323-328 (1998). 29. C. N. Chen, S. Denome, and R. L. Davis, Proc. N&l. Acud. Sci. U.S.A. 83, 9313-9717 (1986). 30. P. Sass, J. Field, J. Nikawa, T Toda, and M. Wigler, Proc. NatZ. Acad. Sci. U.S.A. 83,9303 9307 (1986). 31. H. Charbonneau, N. Beier, K. A. Walsh, and J. A. Beavo, Proc. NatZ. Acud. Sci. U.S.A. 83, 9308-9312 (1986). 31~. L. Aravind and C. l? Ponting, Trends Biochem. Sci. 22,458-459 (1997). (1987). 32. J.-I. Nikawa, P. Sass, and M. %‘igler, Mol. Cell. Biol. 7,3629-3636 (1995). 33. S. M. Callahan, N. W. Cornell, and I! V. Dunlap,]. BioZ. Chem. 270,17627-17632 34. M.-L. Lacombe, G. Podgorski, J. Franke, and R. Kessin,]. BioZ. Chem. 261,16811-16817 (1986). (1996). 35. F. Bums, A. Z. Zhao, and J. A. Beavo, Adv. Phurmacol. (N.Y.) 36,29-48 36. D. A. Fisher, J. F. Smith, J. S. Pi&r, S. H. St. Denis, and J. B. Cheng, Biochem. Biophys. Res. Common. 246,570-577 (1998). 37. D. A. Fisher, J. l? Smith, J. S. Pillar, S. H. St. Denis, and J. B. Cheng, _Z.BioZ. Chem. 273, 15559-15564 (1998). 38. S. H. Soderling, S. J. Bayuga, and J. A. Beavo, Proc. N&Z. Acad. Sci. U.S.A. 95,8991-8996 (1998). (1998). 39. S. H. Soderhng, S. J. Bayuga, and J. A. Beavo,]. Biol. Chem. 273,15553-15558 40. H. Charbonneau, R. K. Prusti, H. LeTrong, W. K. Sonnenburg, P. J. MuIIaney, K. A. Walsh, and J. A. Beavo, Proc. N&Z. Acud. Sci. U.S.A. 87,288-292 (1990). 41. R. L. Kin&d, I. E. Stitl-Coleman, and M. Vaughan,]. Biol. Chem. 260,9009-9015 (1985). (1989). 42. S. D. Stroop, H. Charbonneau, and J. A. Beavo,J. BioZ. Chem. 264,13718-13725 43. H. Charbonneau, S. Kumar, J. P. Novack, D. K. Blumenthal, P. R. Griffin, J. Shabanowitz, D. E Hunt, J. A. Beavo, and K. A. Walsh, Biochemistry 30,7931-7940 (1991).

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

43

44. J. P. Novack, H. Charbonneau, J. K. Bentley, K. A. Walsh, and J. A. Beavo, Biochemistry 30,7940-7947 (1991). 45. S. D. Stroop and J. A. Beavo, Adv. Second Messenger Phosphoprotein Res. 25, 55-71 (1992). 46. S. L. Jin, J. V Swinnen, and M. Corm,]. Biol. Chem. 267, 18929-18939 (1992). 47. E. Meacci, M. Tan-a, M. Moos, C. J. Smith, M. A. Mousesian, E. Degerman, P. Belfrage, and V. Manganiello, Proc. N&Z. Acad. Sci. U.S.A. 89,3721-3725 (1992). 48. P. P. Cheung, H. Xu, M. M. McLaughlin, F. A. GhazaIeh, C. P. Livi, and R. W. Colman, Blood S&1321-1329 (1996). 49. S. Jacobitz, M. M. McLaughlin, G. P. Livi, M. Burman, and T. J. Torphy, Mol. Pharmacol. 50,891-899 (1996). 49a. T. H. Fink, S. H. Francis, A. Beasley, K. A. Grimes, and J. D. Corbin, J. BioZ. Chem. 274, 34613-34620 (1999). 50. P. G. Grant, D. L. DeCamp, J. M. Bailey, R. W. Colman, and R. F. Colman, Biochemistry 29,887-894 (1990). 51. G. A. Omburo, T. J. Torphy, G. Scott, S. Jacobitz, R. F. Colman, R. W. Colman, T Brickus, (1997). and F. A. Ghazaleh, Blood 89,1019-1026 52. J. D. Corbin, A. Beasley, I. V. Turko, T. L. Haik, K. A. Mangum, J. N. Wells, S. H. Francis, and K. R. Sekhar, Cell Biochem. Biophys. 29,145-157 (1998). 53. G. A. Omburo, S. Jacobitz, T J. Torphy, and R. W. Colman, Cell. Signal. 10, 491-497 (1998). 54. ,4. E. Granovsky, M. Natochin, R. L. McEntaffer, T. L. Haik, S. H. Francis, J. D. Corbin, and N. 0. Artemyev,J. Biol. Chem. 273,24485-24490 (1998). 55. P. Greengard, S. A. Rudolph, and J. M. Sturtevant,!. BioZ. Chem. 244,4798-4800 (1969). 56. T. Braumann, C. Emeux, G. Petridis, W. D. Stohrer, and B. Jastorff, Biochim. Biophys. Actu 871,199-206 (1986). 57. N. D. Goldberg, T. F. Walseth, J. H. Stephenson, T. P. Krick, and G. Graff, J. Biol. Chem. 255,10344-10347 (1980). 58. P. M. Burgers, F. Eckstein, D. H. Hunneman, J. Baraniak, R. W. Kinas, K. Lesiak, and W. J. Stec,J. BioZ. Chem. 254,9959-9961 (1979). 59. R. L. Jarvest, G. Lowe, J. Baraniak, and W. J. Stec, Biochem./. 203,461-470 (1982). 60. M. K. Thomas, S. H. Francis, and J. D. Corbin,J. BioZ. Chem. 265,14964-14970 (1990). 61. E. Butt, J. Beltman, D. E. Becker, G. S. Jensen, S. D. RybaIkin, B. Jastorff, and J. A. Beavo, Mol. Pharmacol. 47,340-347 (1995). 62. J. Beltman, D. E. Becker, E. Butt, G. S. J ensen, S. D. Rybalkin, B. Jastorff, and J. A. Beavo, Mol. Pharmacol. 47,330-339 (1995). 63. D. Couchie, G. Petridis, B. Jastorff, and C. Erneux, Eur. J. Biochem. 136,571-575 (1983). 64. I. V. Turko, S. H. Francis, and J. D. Corbin, Biochemistry 37,4200-4205 (1998). 65. K. M. Tang, E. K. Jang, and R. J. Haslam, Biochem. J. 323,217-224 (1997). 66. F. Burns and N. J. Pyne, Biochem. Biophys. Res. Comm. 189,1389-1396 (1992). 67. J. E. Souness and S. Rao, Cell. Signal. 9, 227-236 (1997). 68. S. Jacobitz, M. D. Ryan, M. M. McLaughlin, G. P Livi, W. E. J. DeWolf, T. J. Torphy, and M. Burman, Mol. Pharmacol. 5&999-1006 (1997). 69. I. Wilkinson, P. Engels, and M. D. Houslay, Pharmacol. Rev. Commun. 9,215-226 (1997). 70. J. M. Atienza, D. Susanto, C. Huang, A. S. McCarty, and J. Colicelli, J. BioZ. Chem. 274, 4839-4847 (1999). 71. I. V Turko, S. H. Francis, and J. D. Corbin,]. BioZ. Chem. 273, 6460-6466 (1998). 72. S. H. Francis, J. L. Colbran, L. M. McAllister-Lucas, and J. D. Corbin, J. Biol. Chem. 269, 22477-22480 (1994). 73. B. L. Vallee and D. S. Auld, Proc. NatZ.Acad. Sci. U.S.A. 87,220-224 (1990). 74. B. L. Vallee and D. S. Auld, Biochemisty 29, 5647-5659 (1990).

44

SHARRON

H. FRANCIS

ET AL.

75. B. L. Vallee and D. S. Auld, Biochemistry 32,6493-6500 (1993). 76. S. Jacobi& M. M. McLaughlin, G. P Livi, M. D. Ryan, and T J. Torphy, FASEBJ. 8, A371 (1994). 76~. S. H. Francis, I. V. Turko, K. A. Grimes, and J. D. Corbin, Biochemistry 39, in press (2000). 77. M. D. Percival, B. Yeh, and J. P Falgueyret, Biochem. Biophys. Res. Commun. 241, 175180 (1997). 78. D. Srivastava, D. A. Fox, and R. L. Hun&z, Bi0chem.J. 308,653-658 (1995). 79. G. A. Omburo, T Brickus, F. A. Ghazaleh, and R. W. Colman, Arch. Biochem. Biophys. 323, l-5 (1995). 80. P P. Cheung, L. Yu, H. Zhang, and R. W. Colman, Arch. Biochem. Biophys. 360,99-104 (1998). 81. J. Londesborough and K. Suoranta, In “Methods in Enzymology” (J. D. Corbin and R. A. Johnson, eds.) pp. 777-785. Academic Press, New York, 1988. 82. M. A. Holmes and B. W. Matthews,J. Mol. Biol. 160,623-639 (1982). 83. A. Volbeda, A. Lahm, F. Sakiyama, and D. Suck, EMBOJ. 10,1607-1618 (1991). 84. E. Hough, L. K. Hansen, B. Birknes, K. Jynge, S. Hansen, A. Hordvik, C. Little, E. Dodson, and Z. Derewenda, Nature (London) 338,357-360 (1989). 85. E. E. Kim and H. W. Wyckoff,J. Mol. Biol. 218,449-464 (1991). 86. S. A. Harrison, M. L. Chang, and J. .4. Beavo, Circulation 73,111109-III116 (1986). 87. J. Beavo, In “Cyclic Nucleotide Phosphodiesterases: Structure, Regulation and Drug Action” (J. Beavo and M. D. Houslay, eds.), pp. 3-15. Wiley, New York, 1990. 88. T. J. Torphy and B. J. Undem, Thorax 46,512-523 (1991). 89. Y. Takase, T Saeki, N. Watanabe, H. Adachi, S. Souda, and I. Saito,J. Med. Chem. 37,21062111 (1994). (1995). 90. C. Schudt, H. Tenor and A. Hatzelmann, Eur. Respir. J. 8,1179-1183 91. K. R. Sekhar, P. Grondin, S. H. Francis, and J. D. Corbin, In “Phosphodiesterase Inhibitors” (C. Schudt, G. Dent, and K. F. Rabe, eds.), pp. 135-146. Academic Press, New York, 1996. 92. A. H. Cohen, K. Hanson, K. Morris, B. Fouty, I. F. McMurty, W. Clarke, and D. M. Rodman,J. Clin. Znzjest. 97, 172-179 (1996). 93. J. Y. Jeremy, S. A. Ballard, A. M. Naylor, M. A. Miller, and G. D. Angelini, Br. J Ural. 79, 958-963 (1997). 94. P. J. Chiu, S. Vemulapalli, M. Chintala, S. Kurowski, G. G. Tetzloff, A. D. Brown, and E. J. (1997). Sybertz, Naunyn Schmiedebergs Arch. Pharmacol. 355,463-469 95. M. J. Perry and G. A. Higgs, Cum. @in. Chem. Biol. 2,472-481 (1998). 96. P. J. Silver, E. D. Pagani, R. L. Dundore, L. de Garavilla, D. C. Bode, and E. R. Bacon, Eur. J. Pharmucol. 349,263-268 (1998). 97. J. W. Ziegler, D. D. Ivy, J. J. Fox, J. P. Kinsella, W. R. Clarke, and S. H. Abman, Am. J. Respir: Crit. Cure Med. 157,1104-1110 (1998). 98. T. J. Torphy, Am.]. Respir. Crit. Cure Med. 157,351-370 (1998). 99. D. R. Buckle, J, R. Arch, B. J, Connolly, A. E. Fenwick, K. A. Foster, K. J. Murray, S. A. (1994). Readshaw, M. Smallridge, and D. G. Smith,J. Med. Chem. 37,476-485 100. I. V Turko, S. A. Ballard, S. H. Francis, and J. D. Corbin, Mol. Pharmacol. 56, 124-130 (1999). 101. W. J. Rocque, G. Tian, J. S. Wiseman, W. D. Holmes, I. Zajac-Thompson, D. H. Willard, I. R. Patel, G. B. Wisely, W. C. Clay, S. H. Kadwell, C. R. Hoffman, M. A. Luther, R. W. Dougherty, 0. Ittoop, and L. Overton, Biochemistry 36,14250-14261 (1997). 102. E. G. Loten and J. G. Sneyd, Biochem.]. l20,187-193 (1970). 103. S. H. Francis and T. Kono, Mol. Cell. Biochem. 42,109-116 (1982). 104. A. Z. Zhao, H. Zhao, J. Teague, W. Fujimoto, and J. A. Beavo, Proc. N&l. Acad. Sci. U.S.A. 94,3223-3228 (1997).

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

45

IO4a.A. Z. Zhao, M. M. Shinohara, D. Huang, M. Shimizu, H. Eldar-Finkelman, E. G. Krebs, J. A. Beavo,J. Biol. Chem. 275,11348-11354 (2000). 105. A. Z. Zhao, K. E. Bornfeldt, and J. A. Beavo,]. Clin. Znr;est. 102,869-873 (1998). 106. C. B. Andersen, R. A. Roth, and M. Conti,J. Biol. Chem. 273,18705-18708 (1998). 107. T. A. Wyatt, A. J. Naftilan, S. H. Francis, and J. D. Corbin, Am. J. Physiol. Heart Circ. Ph~p iol. 274, H448-H455 (1998). 108. S. Yarfitz and J. B. Hurley,J. Biol. Chem. 269,14329-14332 (1994). 109. C. Sette, S. Iona, and M. Conti,J. Biol. Chem. 269,9245-9252 (1994). 110. C. Sette, E. Vicini, and M. Conti, Mol. Cell. Endocrinol. 100, 75-79 (1994). (1996). 111. C. Sette and M. Conti,J. BioZ. Chem. 271, 16526-16534 112. J. A. Beavo, J. 6. Hardman, and E. W. Sutherland,J. Biol. Chem. 246,3841-3846 (1971). 113. V C. Manganiello, T. Tanaka, and S. Murashima, In “Cyclic Nucleotide Phosphodiesterases: Structure, Regulation, and Drug Action” (J. B eavo and M. D. Houslay, eds.), pp. 6185. Wiley, New York, 1990. 114. D. M. Juilfs, S. H. Soderhng, F. Bums, and J. A. Beavo, Rec. Physiol. Biochem. Pharmncol. 135,67-104 (1999). 115. V. C. Manganiello, C. J. Smith, A. H. Newman, K. Rice, E. Degerman, and P. Belfrage, J. Cyclic Nucleotide Protein Phosphor. Res. 11,497-511 (1986). 116. J. V. Swinnen, D. R. Joseph, and M. Conti, Proc. NatZ. Acad. Sci. U.S.A. 86, 8197-8201 (1989). 117. J. V Swinnen, K. E. Tsikalas, and M. Conti,J. Biol. Chem. 266, 18370-18377 (1991). 118. J. A. Girault, B. Chamak, G. Bertuzzi, H. Tixier, J. K. Wang, D. T. Pang, and P. Greengard, /. Nent-o&em. 58, 518-528 (1992). 119. T Kovala, I. A. Lorimer, A. M. Brickenden, E. H. Ball, and B. D. Sanwal,J. BioZ. Chem. 269, 8680-8685 (1994). 120. C. D. Manning, M. M. McLaughlin, G. P. Livi, L. B. Ciesiinski, T. J. Torphy, and M. S. Barnette,J. PharmucoZ. Exp. T’her. 276, 810-818 (1996). 121. E. Vicini and M. Conti, Mol. Endocrinol. 11, 839-850 (1997). (1999). 122. H. Liu and D. H. Maurice, Br. J. Phamzacol. 125,1501-1510 123. L. S. Sanchez, S. M. de la Monte, G. Filippov, R. C. Jones, W. M. Zapol, and K. D. Bloch, Pediatr. Res. 43, 163-168 (1998). 124. M. M. Bakre and S. S. Visweswariah, FEBS Left. 408,345-349 (1997). 125. K. A. Hanson, F. Bums, S. D. Rybalkin, J. W. Miller, J. Beavo, and W R. Clarke, Am. J. Respir. Crit. Care Med. 158, 279-288 (1998). 126. L. Li, C. Yee, and J. Beavo, Science 283,848-851 (1999). 127. A. R. Morena, C. Boitani, S. de Grossi, M. Stefanini, and M. Conti, Endocrinology 136, 687-695 (1995). 128. F. Naro, R. Zhang, and M. Conti, Endocrinology 137,2464-2472 (1996). 129. M. D. Houslay, G. Scotland, L. Pooley, S. Spence, I. Wilkinson, F. McCallum, P. Julien, N. G. Rena, A. M. Michie, and S. Erdogan, Biochem. Sot. Trans. 23,393-398 (1995). 130. M.D. Houslay, Adu. Em. Reg. 35,303-338 (1995). 131. C. C. Chini, J. P. Grande, E. N. Chini, and T. P. Dousa, /. Biol. Chem. 272, 9854-9859 (1997). 132. G. B. Bolger, S. Erdogan, R. E. Jones, K. Loughney, G. Scotland, R. Hoffmann, I. Wilkinson, C. Farrell, and M. D. Houslay, Biochem. J 328,539-548 (1997). 133. S. L. Jin,T. Bushnik, L. Lan, M. Conti, and J. V. Swinncn,J. Biol. Chem. 273,19672-19678 (1998). 134. M. Conti and S.-L. C. Jin, Nucl. Acid Res. Mol. BioZ. 63, 1-38 (1999). 135. J, K. Bentley, A. Kadlecek, C. H. Sherbert, D. Seger, W. K. Sonnenburg, H. Charbonneau, J. P. Novack, and J. A. Beavo,J. Biol. Chem. 267, 18676-18682 (1992).

46

SHARRON

H. FRANCIS

ET AL.

136. W. K. Sonnenburg, D. Seger, and J. A. Beavo,J. Biol. Chem. 268,645-652 (1993). 137. J. W. Polli and R. L. Kin&d,]. Neurosci. 14,1251-1261 (1994). 138. C. Yan, A. Z. Zhao, J. K. Bentley, K. Loughney, K. Ferguson, and J. A. Beavo, Proc. N&Z. Acad. Sci. U.S.A. 92,9677-9681 (1995). 139. C. Yan, A. Z. Zhao, J. K. Bentley, and J. A. Beavo,]. Biol. Chem. 271,25699-25706 (1996). 140. J. Yu, S. L. Wolda, A. L. Frazier, V A. Florio, T J. Martins, P B. Snyder, E. A. Harris, K. N. McCaw, C. A. Farrell, B. Steiner, J. K. Bentley, J. A. Beavo, K. Ferguson, and R. Gelinas, C&11.Signal. 9,519-529 (1997). 141. T. M. Reed, J. E. Browning, R. I. Blough, C. V. Vorhees, and D. R. Repaske, Mammalian Genome 9,571-576 (1998). 142. S. Shenolikar, W. J. Thompson, and S. J. Strada, Biochemistry 24,672-678 (1985). 143. A. Z. Zhao, C. Yan, W. K. Sonnenburg, and J. A. Beavo, Adv. Second Messenger Phosphoprotein Res. 31,237-251 (1997). 144. R. Geremia, P. Rossi, D. Mocini, R. Pezzotti, and M. Conti, Biochem. J. 217, 693-700 (1984). Reprod. Fertil. 74,317-327 (1985). 145. l? Rossi, R. Pezzotti, M. Conti, and R. Geremia,]. 146. R. K. Sharma, Y. Tan, and R. V. Raju, Arch. Biochem. Biophys. 339,40-46 (1997). 147. W. K. Sonnenburg, D. Seger, K. S. Kwak, J. Huang, H. Charbonneau, and J. A. Beavo, J. Biol. Chem. 270,30989-31000 (1995). 148. R. K. Sharma and J. H. Wang, Proc. Natl. Acud. Sci. U.S.A. 82,2603-2607 (1985). 149. R. K. Sharma and J. H. Wang,]. Biol. Chem. 261,1322-1328 (1986). 150. R. K. Sharma and J. H. Wang,]. Biol. Chem. 261,14160-14166 (1986). 151. R. K. Sharma, Biochemistry 30,5963-5968 (1991). 152. V. A. Florio, W. K. Sonnenburg, R. Johnson, K. S. Kwak, G. S. Jensen, K. A. Walsh, and J. A. Beavo, Biochemisty 33,8948-8954 (1994). 153. D. C. LaPorte, W. A. J. Toscano, and D. R. Storm, Biochemistry 18,2820-2825 (1979). 154. M. M. Tucker, J. B. J. Robinson, and E. Stellwagen,J. Biol. Chem. 256,9051-9058 (1981). 155. D. M. Juilfs, H. J. Fiille, A. Z. Zh ao, M. D. Houslay, D. L. Garbers, and J. A. Beavo, Proc. N&l. Acad. Sci. U.S.A. 94,3388-3395 (1997). 156. D. E. Jaworsky, 0. Matsuzaki, F. E Borisy, and G. V. Ronnett, J, Neurosci. 15, 310-318 (1995). 157. I. Boekhoff, C. Kroner, and H. Breer, Cell. Signal. 8, 167-171 (1996). 158. S. D. RybaIkin, K. E. Bornfeldt, W. K. Sonnenburg, I. G. RybaIkina, K. S. Kwak, K. Hanson, E. G. Krebs, and J. A. Beavo,J. Clin. Invest. 100,2611-2621 (1997). 159. J. A. Beavo, R. S. Hansen, S. A. Harrison, R. L. Hun&z, T J. Martins, and M. C. Mumby, Mol. Cell Endotinol. 28,387-410 (1982). 160. M. E. Whahn, S. J. Strada, and W. J. Thompson, Biochim. Biophys. Acta 972, 79-94 (1988). 161. M. E. Whahn, J. G. ScammeIl, S. J. Strada, and W. J. Thompson, Mol. Phurmacol. 39,711717 (1991). 162. W. K. Sonnenburg, P. J. Mullaney, and J. A. Beavo, J. Biol. Chem. 266, 17655-17661 (1991). 163. H. L. Trong, N. Beier, W. K. Sonnenburg, S. D. Stroop, K. A. Walsh, J. A. Beavo, and H. Charbonneau, Biochemistry 29,10280-10288 (1990). 164. G. J. Rosman, T J. Martins, W. K. Sonnenburg, J. A. Beavo, K. Ferguson, and K. Loughney, Gene 19189-95 (1997). 165. S. D. Stroop and J, A. Beavo,]. Biol. Chem. 266,23802-23809 (1991). 166. C. Erneux, D. Couchie, J. E. Dumont, J. Baraniak, W. J. Stec, E. G. Abbad, G. Petridis, and B. Jastorff, Eur.]. Biochem. 115, 503-510 (1981). 167. T J. Martins, M. C. Mumby, and J. A. Beavo,]. Biol. Chem. 257,1973-1979 (1982).

CYCLIC

NUCLEOTIDE

168. T. Yamamoto, V. C. Manganiello, 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186. 187.

188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198.

47

PHOSPHODIESTERASES

and M. Vaughan, J. Biol. Chem. 258,

12526-12533

(1983). T. Yamamoto, S. Yamamoto, J. C. Osborne, Jr., V. C. Manganiello, M. Vaughan, and H. Hidaka,]. Biol. Chem. 258,14173-14177 (1983). (1977). J. Moss, V. C. Manganiello, and M. Vaughan,]. Biol. Chem. 252,5211-5215 C. Emeux, F. Miot, J. M. Boeynaems, and J. E. Dumont, FEBS Leti. 142,251-254 (1982). C. Erneux, F. Miot, P J. Van Haastert, and B. Jastorff, J. Cyclic Nucleotide Protein Phosphor. Res. 10,463-472 (1985). C. Emeux and F. Miot (1988). Methods Enzymol. 159,520-530 (1988). F. Miot, P. J. Van Haastert, and C. Emeux, Eur. J. Biochem. 149,59-65 (1985). J. P. Novack, H. Charbonneau, D. K. Blumenthal, K. A. Walsh, and J. A. Beavo, Ado. Exp. Med. Biol. 255,387-395 (1989). H. C. Hartzell and R. Fischmeister, Nature (London) 323,273-275 (1986). M. A. Simmons and H. C. Hartzell, Mol. Pharmacol. 33,664-671 (1988). N. T. Dickinson, E. K. Jang, and R. J, Haslam, Bi0chem.J. 323,371-377 (1997). (1991). R. T. MacFarland, B. D. Zelus, and J. A. Beavo,J. Biol. Chem. 266,136-142 W. K. Sonnenburg and J. A. Beavo, In “Cyclic GMP: Synthesis, Metabolism and Function” (F. Murad, ed.), p. 87. Academic Press, San Diego, 1994. E. Degennan, P. Belfrage, A. H. Newman, K. C. Rice, and V. C. Manganiello,J. Biol. Chem. 262,5797-5807 (1987). A. Rascon, S. Lindgren, L. Stavenow, P. Belfrage, K. E. Andersson, V. C. Manganiello, and E. Degerman, Biochim. Biophys. Actu 1134,149-156 (1992). D. H. Maurice and R. J. Haslam, Mol. Phurmucol. 37,671-681 (1990). B. E. VanUffelen, B. M. De Koster, and J, G. Elferink, Biochem. Pharmacol. 56,1061-1063 (1998). A. E. Eckly and C. Lugnier, Br. J. Pharmucol. 113,445-450 (1994). A. Kurtz, K. H. Gotz, M. Hamann, and C. Wagner, Proc. Natl. Acad. Sci. U.S.A. 95,47434747 (1998). E. Degerman, M. Moos, Jr., A. Rascbn, V. Vasta, E. Meacci, C. J. Smith, S. Lindgren, K.-E. Andersson, P. Belfrage, and V. Manganiello, Biochim. Biophys. Acta Protein Struct. Mol. Enzymol. l205,189-198 (1994). T. Kono, F. W. Robinson, and J. A. Sarver,J. Biol. Chem. 250,7826-7835 (1975). C. Correze and H. Thibout, J. Cyclic Nucleotide Protein Phosphor. Res. 10, 167-179 (1985). C. H. Macphee, S. A. Harrison, and J. A. Beavo, Proc. N&l. Acad. Sci. U.S.A. 83,66606663 (1986). J. D. Corbin, S. J. Beebe, and P. F. Blackmore,J. Biol. Chem. 260,8731-8735 (1985). V C. Manganiello, C. J. Smith, E. Degennan, V. Vasta, H. Tomqvist, and P. Belfrage, Adu. Exp. Med. BioZ. 293,239-248 (1991). V. C. Manganiello, E. Degerman, C. J. Smith, V. Vasta, H. Tomqvist, and P Belfrage, Ado. Second Messenger Phosphoprotein Rex 25,147-164 (1992). J. W. Eriksson, C. Wesslau, and U. Smith, Biochim. Biophys. Acta 1189, 163-167 (1994). H. Eriksson, M. Ridderstrale, E. Degerman, D. Ekholm, C. J. Smith, V. C. Manganiello, P. Belfrage, and H. Tomqvist, Biochim. Biophys. Acta l266,101-107 (1995). C. H. Macphee, D. H. Reifsnyder, T A. Moore, K. M. Lerea, and J. A. Beavo,J. Biol. Chem. 263,10353-10358 (1988). L. 6. Pawlson, C. J. Lovell-Smith, V. C. Manganiello, and M. Vaughan, Proc. Natl. Acad. Sci. U.S.A. 71, 1639-1642 (1974). H. Makino, P. M. deBuschiazzo, R. H. Pointer, J. E. Jordan, and T. Kono,J. Biol. Chem. 255, 7845-7849 (1980).

48

SHARRON

H. FRANCIS

ET AL,.

199. H. Makino and T. Kono,J. Biol. Chem. 255,7850-7854 (1980). 200. S. J. Beebe, J. B. Redmon, P. F. Blackmore, and J. D. Corbin, J. Biol. Chem. 260, 1578115788 (1980). 201. C. J. Smith and V C. Manganiello, Mol. PharmacoZ. 35,381-386 (1989). 202. V. C. Manganiello,]. Mol. Cell Curdiol. 19, 1037-1040 (1987). 203. M. Taira, S. C. Hockman, J. C. Calve, P. Belfrage, and V. C. Manganiello, J. Biol. Chem. 268,18573-18579 (1993). 204. M. J. Leroy, E. Degerman, M. Taira, T. Murata, L. H. Wang, M. A. Movsesian, E. Meacci, and V C. Manganiello, Biochemistry 35,10194-10202 (1996). 205. T. M&i, M. Taira, S. Hockman, F. Shimada, J. Lieman, M. Napohtano, D. Ward, H. Makino, and V. C. Manganiello, Genomics 36, 476-485 (1996). 206. R. W. Lobber-t, A. Winterpacht, B. Seipel, and B. U. Zabel, Genomics 37,211-218 (1996). 207. R. R. Reinhardt, E. Chin, J. Zhou, M. Taira, T Murata, V. C. Manganiello, and C. A. Bondy, J. Clin. Invest. 95, 1528-1538 (1995). 208. R. He, N. Komas, D. Ekholm, T. Murata, M. Taira, S. Hockman, E. Degerman, and V. C. Manganiello, Cell Biochem. Biophys. 29, 89-111 (1998). 209. T. W. Gettys, A. J. Vine, M. F. Simonds, and J. D. Corbin,]. Biol. Chem. 263,10359-10363 (1988). 210. E. Degerman, C. J. Smith, H. Tornqvist, V. Vasta, P. Belfrage, and V. C. Manganiello, Proc. (1990). Natl. Acad. Sci. U.S.A. 87,533-537 211. T Rahn, L. Ronnstrand, M. J. Leroy, C. Wernstedt, H. Tornqvist, V. C. Manganiello, P. Belfrage, and E. Degerman,J. Biol. Chem. 271,11575-11580 (1996). 212. C. J. Smith, V. Vasta, E. Degerman, P. Belfrage, and V. C. Manganiello, J. Biol. Chem. 266, 13385-13390 (1991). 213. P. Lopez-Aparicio, A. Rascon, V. C. Manganiello, K. E. Andersson, P. Belfrage, and E. Degennan, Biochem. Biophys. Res. Commun. 186,517-523 (1992). 214. P Lopez-Aparicio, P. Belfrage, V. C. Manganiello, T. Kono, and E. Degerman, Biochem. Biophys. Res. Commun. 193,1137-1144 (1993). 215. J. Wijkander, T. R. Landstrom, V Manganiello, P. Belfrage, and E. Degerman, Endocrinology 139,219-227 (1998). 216. G. Bolger, T. Michaeh, T. Martins, T. St. John, B. Steiner, L. Rodgers, M. Riggs, M. Wigler, (1993). and K. Ferguson, Mol. Cell. Biol. 13,6558-6571 217. G. B. Bolger, Cell. Signal. 6, 851-859 (1994). (1994). 218. L. Monaco, E. Vicini, and M. Conti,J. Biol. Chem. 269,347-357 219. Y. M. Horton, M. Sullivan, and M. D. Houslay, Biochem. J. 308,683-691 (1995). 220. S. Iona, M. Cuomo, T. Bushnik, F. Naro, C. Sette, M. Hess, E. R. Shelton, and M. Conti, Mol. Pharmacol. 53, 23-32 (1998). 221. M. Sullivan, M. Egerton, Y. Shakur, A. Marquardsen, and M. D. Houslay, Cell. Signal. 6, 793-812 (1994). 222. C. Szpirer, J. Szpirer, M. Riviere, J. Swinnen, E. Vicini, and M. Conti, Cytogenet. Cell Genet. 69, 11-14 (1995). (1996). 223. G. Nemoz, R. Zhang, C. Sette, and M. Conti, FEBS Lett. 384,97-102, 224. G. B. Bolger, I. McPhee, and M. D. Houslay,J. Biol. Chem. 271,1065-1071 (1996). 225. R. Obemolte, J. Ratzliff, P. A. Baecker, D. V Daniels, P. Zuppan, K. Jamagin, and E. R. Shelton, Biochim. Biophys. Acta 1353,287-297 (1997). 226. P A. Kithas, M. Artman, W. J. Thompson, and S. J. Strada, Circ. Res. 62,782-789 (1988). 227. Y. Shakur, M. Wilson, L. Pooley, M. Lobban, S. L. Griffiths, A. M. Campbell, J. Beattie, C. Daly, and M. D. Houslay, Biochem.]. 306,801-809 (1995). 228. G. Scotland and M. D. Houslay, Biochem. J. 308,673-681 (1995). 229. G. Nemoz, C. Sette, M. Conti, and R. Zhang, Mol. Phurmacol. 51,242-249 (1997).

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

49

230. T. Kovala, B. D. Sanwal, E. H. Ball, I. A. Lorimer, and A. M. Brickenden, Biochemistry 36, 2968-2976 (1997). 231. M. Grange, M. Picq, A. F. Prigent, M. Lagarde, and G. Nemoz, Cell Biochem. Biophys. 29, 1-17 (1998). 232. Y. Shakur, J. G. Pryde, and M. D. Houslay, Biochem. 1. 292,677-686 (1993). 233. I. McPhee, L. Pooley, M. Lobban, G. Bolger, and M. D. Houslay, Biochem. J. 310, 965974 (1995). 234. G. Scotland, M. Beard, S. Erdogan, E. Huston, F McCallum, S. J. MacKenzie, A. H. Peden, L. Pooley, N. G. Rena, A. H. Ross, S. J. Yarwood, and M. D. Houslay, Methods 14,65-79 (1998). 235. K. J. Smith, G. Scotland, J. Beattie, I. P. Trayer, and M. D. Houslay, J. Biol. Chem. 271, 16703-16711 (1996). 236. E. Huston, S. Lumb, A. Russell, C. Catterall, A. H. Ross, M. R. Steele, G. B. Bolger, M. J. Perry, R. J. Owens, and M. D. Houslay, Bi0chem.J. 328,549-558 (1997). 237. J. C. O’Connell, J. F. McCall urn, I. McPhee, J. Wakefield, E. S. Houslay, W. Wishart, G. Bolger, M. Frame, and M. D. Houslay, Biochem. J. 318,255-261 (1996). 238. M. D. Houslay, G. Scotland, S. Erdogan, E. Huston, S. Mackenzie, J. F. McCallum, I. McPhee, L. Pooley, G. Rena, A. Ross, M. Beard, A. Peder, E Begg, I. M’ilkinson, S. Yarwood, C. Ackerman, E. S. Houslay, R. Hoffman, I? Engels, M. Sullivan, and G. Bolger, Biochem. Sot. Trans. 25,374-381 (1997). 239. R. J. Owens, C. Catterall, D. Batty, J. Jappy, A. Russell, B. Smith, J. O’Connell, and M. J. Peny, Biochem. J. 326,53-60 (1997). 240. \V. J. Rocque, W. D. Holmes, I. R. Patel, R. W. Dougherty, 0. Ittoop, L. Overton, C. R. Hoffman, G. B. Wisely, D. H. Willard, and M. A. Luther, Protein Exp. Pwif 9, 191-202 (1997). 241. J. M. Lenhard, D. B. Kassel, W. J. Rocque, L. Hamacher, W. D. Holmes, I. Patel, C. Hoffman, and M. Luther, Biochem. J. 316,751-758 (1996). 242. S. J. MacKenzie, S. J. Yarwood, A. H. Peden, G. B. Bolger, R. G. Vernon, and M. D. Houslay, Proc. N&Z. Acud. Sci. U.S.A. 95,3549-3554 (1998). 243. R. Hoffmann, I. R. Wilkinson, J. F. McCallum, P. Engels, and M. D. Houslay, Biochem. J. 333,139-149 (1998). 244. M. Conti, S. Iona, M. Cuomo, J. V. Swinnen, J. Odeh, and M. E. Svoboda, Biochemistry 34, 7979-7987 (1995). 245. R. L. Davis and B. Dauwalder, Trends Genet. 7,224-229 (1991). (1978). 246. P. Hamet and J. F. Coquil,J. Cyclic Nucleotide Res. 4, 281-290 247. S. H. Francis, T. M. Lincoln, and J. D. Corbin,J. Biol. Chem. 255,620-626 (1980). 247~. J. D. Corbin and S. H. Francis,J. Biol. Chem. 274, 13729913732 (1999). 248. L. M. McAllister-Lucas, W. K. Sonnenburg, A. Kadlecek, D. Seger, H. LeTrong, J. L. Colbran, M. K. Thomas, K. A. Walsh, S. H. Francis, J. D. Corbin, and J. A. Beavo,J. Biol. Chem. 268,22863-22873 (1993). 249. J. Kotera, K. Fujishige, H. Akatsuka, Y. Imai, N. Yam&a, and K. Omori, J. Biol. Chem. 273, 26982-26990 (1998). 250. K. Loughney, T. R. Hill, V. A. Fl orio, L. Uher, 6. J. Rosman, S. L. Wolda, B. A. Jones, M. L. Howard, L. M. McAllister-Lucas, W. K. Sonnenburg, S. H. Francis, J. D. Corbin, J. A. Beavo, and K. Ferguson, Gene 216,137-147 (1998). 2.50~. C. S. Lin, A. Lau, R. Tu, T. F. Lue, Biochem. Biophys. Res. Comm. 268,628-633 (2000). 251. N. Yanaka, J. Kotera, A. Ohtsuka, H. Akatsuka, Y. Imai, H. Michibata, K. Fujishige, E. Kawai, S. Takebayashi, K. Okumura, and K. Omori, Eur. J. Biochem. 255,391-399 (1998). 252. J. Kotera, K. Fujishige, Y. Imai, E. Kawai, H. Michibata, H. Akatsuka, N. Yanaka, and K. Omori, Eur. 1. Biochem. 262,866-872 (1999).

50 253. 254. 255. 255~. 256. 257. 258. 259. 260. 261. 262. 263. 264.

265. 266. 267.

268. 269. 270. 271. 272. 273. 274. 275. 276. 277. 278. 279. 280.

SHARRON

H. FRANCIS

ET AL.

A. J. Carter, S. A. Ballard, and A. M. Naylor,J. Ural. 160,242-246 (1998). M. K. Thomas, S. H. Francis, and J. D. Corbin,J. Biol. Chem. 265,14971-14978 (1990). I. V Turko, S. H. Francis, and J. D. Corbin, Biochem. J. 329,505-510 (1998). J. D. Corbin, I. V. Turko, A. Beasley, and S. H. Francis, Eur. J Biochem. 267,2760-2767 (2000). J. L. Colbran, S. H. Francis, A. B. Leach, M. K. Thomas, H. Jiang, L. M. McAllister, and J, D. Corbin,J. Biol. Chem. 267,9589-9594 (1992). F. Burns, I. W. Rodger, and N. J. Pyne, BiochemJ. 283,487-491(1992). A. Lochhead, E. Nekrasova, V. Y. Arshavsky, and N. J. Pyne, J. Biol. Chem. 272,1839718403 (1997). I. V. Turko, S. H. Francis, and J. D. Corbin, J. Biol. Chem. 274,29038-29041 (1999). L. M. McAllister-Lucas, T. L. Haik, J. L. Colbran, W. K. Sonnenburg, D. Seger, I. V. Turko, J. A. Beavo, S. H. Francis, and J. D. Corbin, J. Biol. Chm. 270,30671-30679 (1995). I. V Turko, T. L. Haik, L. M. McAllister-Lucas, E Bums, S. H. Francis, and J. D. Corbin, J. Biol. Chem. 271,22240-22244 (1996). E. F. Pai, W. Kabsch, U. Krengel, K. C. Holmes, J. John, and A. Wittinghofer, Nature &on&n) 341,209-214 (1989). S. A. Ballard, C. J. Gingell, K. Tang, L. A. Turner, M. E. Price, and A. M. Naylor, J. Ural. (1998). 159,2164-2171 A. Yamazaki, S. Uchida, P. J. Stein, G. L. Wheeler, and M. W. Bitensky, In “Advances in Cyclic Nucleotide and Protein Phosphorylation Research” (S. J. Strada and W. J. Thompson, eds.), pp. 381-392. Raven Press, New York, 1984. M. Chabre and P. Deterre, Eur. J. Biochem 179,255-266 (1989). Y. A. Ovchinnikov, V. V Gubanov, N. V. Khramtsov, K. A. Ischenko, V E. Zagranichny, K. G. Muradov, T. M. Shuvaeva, and V. M. Lipkin, FEBS Leti. 223,169-173 (1987). V. M. Lipkin, N. V Khramtsov, I. A. Vasilevskaya, N. V. Atabekova, K. G. Muradov, V. V. Gubanov, T. Li, J. P. Johnson, K. J. Volpp, and M. L. Applebury, J. Biol. &em. 265,1295512959 (1990). N. 0. Artemyev, R. Surendran, J. C. Lee, and H. E. Hamm, J. Biol. Chem. 271,2538225388 (1996). S. E. Hamilton, R. K. Prusti, J. K. Bentley, J. A. Beavo, and J. B. Hurley, FEBS Leti. 318, 157-161 (1993). 0. C. Ong, H. K. Yamane, K. B. Phan, H. K. Fong, D. Bok, R. H. Lee, and B. K. Fung, J. Biol. Chem. 270,8495-8500 (1995). S. K. Srinivas, R. V Srinivas, G. M. Anantharamaiah, R. W. Compans, and J. P Segrest, J. Biol. Chem. 268,22895-22899 (1993). N. Qin and W. Baehr, J. Biol. Chem. 269,3265-3271 (1994). (1988). P. G. Gillespie and J. A. Beavo, J. Biol. Chem. 263,8133-8141 P. G. Gillespie R. K. Prusti, E. D. Apel, and J. A. Beavo, J. Biol. Chem. 264,12187-12193 (1989). N. Li, S. K. Florio, M. J. Pettenati, P. N. Rao, J. A. Beavo, and W. Baehr, Genomics 49,7682 (1998). A. Yamazaki, 1. Sen, M. W. Bitensky, J. E. Casnelhe, and P Greengard,J. Biol. Chm. 255, 11619-11624 (1980). P. G. Gillespie and J. A. Beavo, Proc. Natl. Acad. Sci. U.S.A. 86,4311-4315 (1989). R. H. Cote, M. D. Bownds, and V. Y. Arshavsky, Proc. Natl. Acad. Sci. U.S.A. 9l,48454849 (1994). M. C. Hebert, F. Schwede, B. Jastorff, and R. H. Cote, J. Biol. Chem. 273, 5557-5565 (1998). S. H. Tsang, M. E. Bums, l? D. CaIvert, P Gouras, D. A. Baylor, S. P. Goff, and V. Y. Arshavsky, Science 282,117-121 (1998).

CYCLIC

NUCLEOTIDE

PHOSPHODIESTERASES

51

281. R. H. Cote and M. A. Brunnock,J. BioZ. Chem. 268,17190-17198 (1993). 282. A. Yamazaki, V. A. Bondarenko, S. Dua, M. Yamazaki, J. Usukura, and F. Hayashi, J. BioZ. Chem. 271,32495-32498 (1996). 283. N. 0. Artemyev and H. E. Hamm, Bi0chem.J. 283,273-279 (1992). 284. A. E. Granovsky, M. Natochin, and N. 0. Artemyev,J. BioZ.Chem. 272,11686-11689 (1997). 284~. H. Mou, H. J. Grazio, T. A. Cook, J. A. Beavo, R. H. Cote, J. BioZ. Chem. 274, 1881318820 (1999). 285. F. Bruckert, P. Catty, P. Deterre, and C. Pfister, Biochemistry 33, 12625-12634 (1994). 286. A. Otto-Bruc, B. Antonny, T. M. Vuong, P. Chardin, and M. Chabre, Biochemistry 32, (1993). 8636-8645 287. F. Pagks, P. Deterre, and C. Pfister,J. BioZ. Chem. 268,26358-26364 (1993). 288. C. Pfister, N. Bennett, F. Bruckert, P. Catty, A. Clerc, F. Pages, and P. Deterre, Cell. Signal. 5,235-241 (1993). 289. N. 0. Artemyev, J. S. Mills, K. R. Thornburg, D. Ft. Knapp, K. L. Schey, and H. E. Hamm, J. Biol. Chem. 268,23611-23615 (1993). 290. N. P. Skiba, N. 0. Artemyev, and H. E. Hamm,J. BioZ. Chem. 270,13210-13215 (1995). 291. W. Liu, W. A. Clark, P. Sharma, and J. K. Northup, J. BioZ. Chem. 273, 34284-34292 (1998). 292. P. D. Calvert, T. W. Ho, Y. M. LeFebvre, and V. Y. Arshavsky,I. Gen. Physiol. ill, 39-51 (1998). 293. W. He, C. W. Cowan, and T G. Wensel, Neuron 20,95-102 (1998). 294. V. M. Lipkin, I. L. Dumler, K. G. Muradov, N. 0. Artemyev, and R. N. Etingof, FEBS Lett. 234,287-290 (1988). 295. A. E. Granovsky, R. McEntaffer, and N. 0. Artemyev, CeZZBiochem. Biophys. 28, 115133 (1998). 296. E. R. Nekrasova, D. M. Berman, R. R. Rustandi, H. E. Hamm, A. G. Gilman, and V. Y. Ar(1997). shavsky, Biochemistry 36,7638-7643 297. T. Wieland, C. K. Chen, and M. I. Simon,J. BioZ. Chem. 272,8853-8856 (1997). 298. C. W. Cowan, R. N. Fariss, I. SokaI, K. Palczewski, andT. G. Wensel, Proc. NatZ.Acad. Sci. U.S.A. 95,5351-5356 (1998). 299. V. M. Lipkin, I. P. Udovichenko, V. A. Bondarenko, A. A. Yurovskaya, E. V. Telnykh, and N. P. Skiba, Biomed. Sci. 1, 305-308 (1990). 300. N. 0. Artemyev, H. M. Rarick, J. S. Mills, N. I? Sk’ 1b a, andH. E. Hamm,]. BioZ. Chem. 267, 25067-25072 (1992). 301. M. Natochin and N. 0. Artemyev,J. BioZ. Chem. 271,19964-19969 (1996). 302. V A. Bondarenko, M. Desai, S. Dua, M. Yamazaki, R. H. Amin, K. K. Yousif, T. Kinumi, M. Ohashi, N. Komori, H. Matsumoto, K. W. Jackson, F. Hayashi, J. Usukura, V. M. Lipkin, and A. Yamazaki,]. BioZ. Chem. 272,15856-15864 (1997). 303. N. 0. Artemyev, M. Natochin, M. Busman, K. L. Schey, and H. E. Hamm, Proc. NatZ.Acud. Sci. U.S.A. 93,5407-5412 (1996). 304. V A. Bondarenko, M. Yamazaki, F. Hayashi, and A. Yamazaki, Biochemistry 38, 77557763 (1999). 305. L. X. Xu, Y. Takana, V. A. Bonderenko, I. Matsuura, H. Matsumoto, A. Yamazaki, and F. Hayashi, Biochemistry 37,6205-6213 (1998). 306. M. Natochin, A. E. Granovsky, andN. 0. Artemyev,J. BioZ. Chem. 272,17444-17449 (1997). 307. K. Seno, A. Kishigami, S. Ihara, T. Maeda, V. A. Bondarenko, Y. Nishizawa, J. Usukura, A. Yamazaki, and F. Hayashi,]. BioZ. Chem. 273,22169-22172 (1998). 308. E. R. Makino, J. W. Handy, T. Li, and V. Y. Arshavsky, Proc. NatZ. Acad. Sci. U.S.A. 96, 1947-1952 (1999). 309. N. Spickofsky, A. Robichon, W. Danho, D. Fry, D. Greeley, B. Graves, V. Madison, and R. E Margolskee, Nature (Struct. Biol.) $771-781 (1994).

52

SHARRON

H. FRANCIS

ET AL.

310. C. Liu, J. M. Ding, L. E. Faiman, and M. U. Gillette,J. Neurosci. 17,659-666 (1997). 311. T. J. Bloom and J. A. Beavo, Proc. N&l. Acad. Sci. U.S.A. 93,14188-14192 (1996). 312. T. Michaeli, T J. Bloom, T. Martins, K. Loughney, K. Ferguson, M. Riggs, L. Rodgers, J. A. Beavo, and M. Wigler,]. Biol. Chem. 268, 12925-12932 (1993). 313. P. Han, X. Zhu, and T. Michaeli,]. Biol. Chem. 272, 16152-16157 (1997). 313a. T Sasaki, J. Kotera, K. Yuasa, and K. Omori, Biochem. Biophys. Rex Common. 271, 575-583 (2000). 313h. J. Hetman, S. H. Soderling, N. A. Glavas, and J. A. Beavo, Proc. Natl. Acud. Sci. U.S.A. 97,

472-476 (2000). 314. M. Guipponi, H. S. Scott, J. Kudoh,

315. 316. 317. 318. 319.

K. Kawasaki, K. Shibuya, A. Shintani, S. Asakawa, H. Chen, M. D. Lalioti, C. Rossier, S. Minoshima, N. Shim& and S. E. Antonarakis, Human Genet. 103,386-392 (1998). S. H. Soderling, S. J. Bayuga, and J. A. Beavo, Proc. N&Z. Acad. Sci. U.S.A. 96,7071-7076 (1999). K. Loughney, P. B. Snyder, L. Uher, G. J. Rosman, K. Ferguson, and V. A. Florio, Gene 234, 109-117 (1999). (1999). K. Fujishige, J. Kotera, and K. Omori, Eur. J. Biochem. 266,1118-1127 J. Kotera, K. Fujishige, K. Yuasa, and K. Omori, Biochem. Biophys. Res. Comm. 261, 555-557 (1999). L. Fawcett, R. Baxendale, P. Stacey, C. McGrouther, I. Harrow, S. Soderling, J. Hetman, (2000). J. A. Beavo, and S. C. Phillips, Proc. Nutl. Acad. Sci. U.S.A. 97,3702-3707

Thyroid

Hormone Regulation

of Apoptotic Tissue Remodeling: implications from Molecular Analysis Amphibian

of

Metamorphosis YUN-BO SHI* AND ATSUKO ISHIZUYA-OKAY

1

*Laboratory of Molecular Embryology National lnstitute of Child Health und Human Development National Institutes of Health Bethesda, Maryland 20892 fDepartment of Histology and Neurobiology Dokkyo Unioersity School of Medicine Mibu, Tochigi 321-02, Japan

I. Introduction .................................................. II. Thyroid Hormone and Vertebrate Development .................... A. Thyroid Hormone in Mammals ............................... 8. Thyroid Hormone in Anurans ................................ III. Mechanism of Thyroid Hormone Action .......................... A. Thyroid Hormone Receptors ................................. B. Transcriptional Regulation by TRs ............................ IV Thyroid Hormone-Dependent Morphological and Cellular Changes during Intestinal Remodeling ................................... ............................................ A. Larval Intestine ............................... B. Metamorphic Transformations C. Organ Autonomous Response to TH .......................... V TR Expression and Function during Intestinal Remodeling .......... A. Correlation of Receptor mRNA Levels with Intestinal Remodeling . B. TH Activation of TRR Genes ................................. C. Cell Type-Dependent Temporal Regulation of TRB Genes ........ VI. Thyroid Hormone Response Genes in the Intestine ................. A. Early TH Response Genes ................................... B. Late TH Response Genes .................................... VII. Functions of TH Response Genes: Implication from Studies on Matrix ............................................ Metalloproteinases ............. A. ECM Remodeling during Intestinal Metamorphosis .............. B. Regulation of MMP Genes during Metamorphosis ............... C. Extracellular Matrix and Cell Fate Determination VIII. Conclusions and Prospects ..................................... References ...................................................

Progress in Zlucleic Acid Research and hlolecular Biology, Vol. 63

53

54 ,5s 55 56 58 !59 61 70 70 73 74 77 77 77 79 80 80 83 84 85 85 91 91 94

54

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

Organogenesis and tissue remodeling are critical processes during postembryonic animal development. Anuran metamorphosis has for nearly a century served as an excellent model to study these processes in vertebrates. Frogs not only have essentially the same organs with the same functions as higher vertebrates such as humans, but also employ similar organogenic processes involving highly conserved genes. Development of frog organs takes place during metamorphosis, which is free of any maternal influences but absolutely dependent on the presence of thyroid hormone. Furthermore, this process can be easily manipulated both in intact tadpoles and in organ cultures by controlling the availability of thyroid hormone. These interesting properties have led to extensive morphological, cellular, and biochemical studies on amphibian metamorphosis. More recently, the cloning of thyroid hormone receptors and the demonstration that they are transcription factors have encouraged enormous interest in the molecular pathways controlling tissue remodeling induced by thyroid hormone during metamorphosis. This article summarizes some of the recent studies on the mechanisms of gene regulation by thyroid hormone receptors and isolation and functional characterization of genes induced by thyroid hormone during Xenopus metamorphosis. Particular focus is placed on the remodeling of the animal intestine, which involves both apoptosis (programmed cell death) of larval cells and de no00 development of adult tissues, and the roles of thyroid hormone-induced genes that encode matrix metalloproteinases during this process. 8 2000Academic PIZSS.

I. Introduction For nearly a century, biologists have used amphibian metamorphosis, particularly anuran metamorphosis, as a model to study postembryonic development in vertebrates. This process is independent of maternal influences but absolutely requires the presence of thyroid hormone (TH) (see below). It is easy to manipulate this process in vitro and access the developing organs and animals for various analyses. Such properties have made anuran metamorphosis a long-lasting model to study the biochemical and morphological changes associated with organogenesis and tissue remodeling in postembryonic animals. All three classes of amphibians-caecilians, urodeles, and anurans-undergo varying degrees of metamorphosis during their development. Anurans have the most dramatic and complete larva/adult transformation. Their metamorphosis involves changes in essentially all organs and tissues of the tadpole. Three major types of changes occur during this process (1,2). The first is the complete resorption of larva-specific organs such as the tail and gills. Tail resorption is one of the best studied processes during anuran metamorphosis. It involves systematic resorption of different tissues, such as the epidermis, connective tissue, and muscles. At the other extreme, limbs develop de nova from undifferentiated blastema cells to form structures that consist

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

55

of tissues similar to those of the resorbed tail. The vast majority of the organs, such as the intestine and liver, are present in both tadpoles and frogs. They undergo a partial but dramatic remodeling to form organs suitable for the frog environmental habitat. In this article, we will focus on this kind of transformation. In particular, we will review our current understanding of the molecular pathways involved in the remodeling of the intestine. We will focus primarily on our studies in Xenopus laevis. Where appropriate, we will describe related studies in other organs and in Rana catesbeiana, another anuran that has been studied at molecular level in considerable detail, to illustrate the potential generality of the basic conclusions. We will begin by reviewing the roles of TH in vertebrate development and the current understanding of the molecular mechanisms of gene regulation by thyroid hormone receptors.

II. Thyroid Hormone and Vertebrate Development A. Thyroid

Hormone

in Mammals

Thyroid hormone is known to affect diverse biological processes in vertebrates. The earliest observed effect associated with TH is human cretinism due to the lack of TH caused by iodine deficiency (3). Human cretins were observed as early as 2600 BC; cretinism is characterized by short stature, mental retardation, and the presence of a goiter, due to overgrowth of the thyroid gland (3). Such developmental defects can be prevented by supplementing TH during fetal and postnatal development, indicating key roles of TH in human development. Consistently, TH levels in human fetal plasma rise sharply a few months prior to birth and remain high during the first several months postbirth (Fig. 1) (4). On birth, the human neonate has a higher rate of metabolism and an increased demand for oxygen (3). TH is known to stimulate metabolic rate both in vitro and in vivo in humans and animals (5 - 7). In addition to increased metabolism after birth, the period of perinatal development is also associated with increased proliferation of glial and neuronal cells and acquisition of several brain functions, as exemplified by the sensory processes (4). Again, the high levels of TH at this period are critical for these and other neural development processes, although the exact processes affected by TH in the developing human brain are not clear. Animal studies have shown that TH influences many aspects of brain development (8). At anatomical and histological levels, both the forebrain and the cerebellum require TH for normal maturation. For example, hypothyroidism delays the appearance of the external germinal layer and decreases the number and density of synaptic contacts

56

YUN-BO SHI AND ATSUKO ISHIZUYA-OKA

with the already defective Purkinje cells, leading to a permanent impairment of neuronal connectivity. The anatomical alterations are accompanied by extensive biochemical changes in the brain, including changes in oxygen consumption and the metabolism of glucose and polyamines. TH continues to play important roles during human postnatal development well past birth (3). Thyroid h ormone is essential for normal growth and development, and deficiency is associated with severe retardation of growth and maturation processes of almost all organ systems. Body weight does not increase and bone growth is also retarded if TH is deficient. The most dramatic effects are seen in tissues that are rapidly proliferating. The severe consequence of insufficient TH or lack of TH during fetal and postnatal development is cretinism, which affects many aspects of the patient (3, 9-13). In addition to the developmental effects, TH deficiency is known to lead to reduced metabolic rate (5 - 7). Similarly, abnormal TH levels are associated with a number of cardiovascular symptoms (14, 15). Thus, proper levels of TH are critical for mammalian development and organ function. 6.

Thyroid

Hormone

in Anurans

Nearly nine decades ago, Gudematsch (16) found that a substance(s) in the thyroid gland could induce metamorphosis. Shortly after, Kendall (17,18) showed that the active ingredient is thyroid hormone. These studies led to the isolation and structural characterization of two natural thyroid hormones, 3, 5,3’, 5’-tetraiodothyronine (TJ, commonly known as thyroxine, and 3, 5, 3’triiodothyronine (T,) (Fig. 1A). Subsequent investigations revealed three independent lines of evidence that firmly establish the causative role of TH in anuran metamorphosis. First, elevations in circulating plasma concentrations of thyroid hormones T, and T, correlate with metamorphosis (Fig. lB, for X. laeuis) (19, 20). In X. Zuevis, there is little TH before stage 54, when tadpoles grow rapidly but exhibit few morphological changes (21). During prometamorphosis (stages 54-58), synthesis of endogenous TH allows for the accumulation of increasing levels of T, and T, in the plasma (Fig. 1B). Accompanying this, tadpoles undergo both growth and morphological transformation, most noticeably the development of the hind limbs. Finally, at the climax of metamorphosis (stages 58 -66), TH is at peak levels and the tadpoles stop feeding and undergo a rapid metamorphic transition. On the completion of metamorphosis at stage 66, plasma TH levels are also reduced. The second line of evidence supporting the role of thyroid hormone in metamorphosis comes from the ability of TH to induce precocious metamorphosis. The first such experiment was the landmark study by Gudernatsch (16), which has since been reproduced for many different anuran species with pure T, and T, at concentrations comparable to endogenous

THYROID

HORMONE-REGULATED

TISSUE

57

REMODELING

T3: 3,.5,3’-triicdothyronine

T4: 3, 5, 3’, 5’-tetraiodothyronine=thyroxine

B

Human Age I,,I

I w

-6

-4

-2

I

I

I

2 0 (Birth)

I

I

(month)

4

6

6

I

I

I

10

12

14

III

100 -

.z LJ 76 F ;

50-

3 2

25-

1

I,,,

I 35/36

I

I

45

51

1

61

63

66

Xenopus Stage

FIG. 1. (A). Structures of two natural thyroid hormones. (B). Plasma TH levels during Xenopus Zaevis and human development. The Xenopus stages are based on Nieuwkoop and Faber (21). The TH levels are based on Leloup and Buscaglia (19) for X. la&s and Tata (4) for humans.

plasma TH levels (1, 20,22).For example, addition of 5 nM of T, to the rearing water of X. laevis tadpoles at stages 55156 leads to drastic changes within 5 days (Fig. 2). Most notable are the external changes, including hind limb morphogenesis and cranial restructuring. Finally, amphibian metamorphosis can be prevented by blocking the synthesis of endogenous TH. In fact, shortly after the landmark experiment of Gudernatsch (16), Allen (23) sh owed that thyroid gland removal resulted in formation of larger than normal tadpoles that were not capable of metamorphosis. However, such giant tadpoles can resume metamorphosis when exogenous TH is added to their rearing water. In addition to thyroidectomy, chemical inhibitors (goitrogens) can be used to inhibit the synthesis of endogenous TH and block metamorphosis (1).

58

YUN-BO

Control

SHI AND ATSUKO

ISHIZUYA-OKA

TH-treated

FIG. 2. TH induces precocious metamorphosis. Stages 55156 Xenopus tadpoles untreated (left) or treated (right) with 10 nM T3 for 5 days showed TH-dependent limb developmentand morphological changes in the head and body.

Their action can again be reversed

by exogenous

evidence

unambiguously

collectively

demonstrates

TH. Thus, all of the above that TH is the causative

agent of anuran metamorphosis.

III. Mechanism of Thyroid Hormone Action Both of the naturally

occurring

thyroid hormones,

T, and T,, are synthe-

sized in the thyroid gland (1, 24). T, can be either secreted ing plasma

or converted The T,

to T, through and T,

deiodination

thyroid

gland.

quently

carried by the plasma to different

ert their biological effects. T, can also be converted (25, 26). In addition,

secreted

into the circulat-

by 5’-deiodinase

.from the thyroid

in the

gland are subse-

organs and tissues, where they ex-

to T3 in various target tissues by 5’-deiodinase

both T3 and T, can be inactivated

through

the action

of 5-deiodinases, producing T, and reverse T,, respectively. Two different 5’deiodinases have been identified in various animal species (25,26). A 5’-deiodinase and a 5-deiodinase

from

R. catesbeiana (27-29) and a 5-deiodinase

THYROID

HORMONE-REGULATED

59

TISSUE REMODELING

from X. lamis (30) have been cloned. They have distinct developmental regulation patterns in different organs, suggesting that T, and T, levels can vary in different organs and tissues. It is unclear whether T, and T, have different biological functions. Some evidence suggests that, at least in certain cases, T, may be first converted to T, to exert its biological effects (29). The biological effects of TH are mostly, if not entirely, mediated by thyroid hormone receptors (TRs). TRs are high-affinity TH-binding proteins with dissociation constants of less than 1 nM. They are localized in the nucleus both in the presence and the absence of TH (31, 32). Using photoaffinity labeling, Samuels and colleagues (33) showed that TRs are chromatin associated. These and other studies show that TRs can regulate gene expression even in the absence of TH and the binding of TH to TRs alters this regulation.

A. Thyroid

Hormone Receptors

There are four TR genes in Xenopus-two TRo genes and two TRB genes (34, 35). Alternative splicing of the TRB transcriptions gives rise to two different isoforms for each TRB gene (Fig. 3). In higher vertebrates, there is only one TRa and one TRB gene, with the latter producing two TRB isoforms due to alternative splicing (36). TRs belong to the superfamily of nuclear hormone receptors, including receptors for glucocorticoid and retinoic acid (37- 41). Members of this family share many structural features. In general, each can be divided roughly into five domains, A/B, C, D, E, and F, respectively, from the amino terminus to the carboxyl terminus (Fig. 3) (42). The DNA-binding domain (domain C) is located in the amino half of the protein and is the most highly conserved domain among different receptors (Fig. 3). The large ligand-binding domain (domain E) is in the carboxyl half of the protein and is conserved among TRs

DNA

Hormone Binding Domain

Binding Domain AIB

C

E

D

TRaAIB 1 tY?&SY

V/////////////////A

TRPAl

f\\\\\‘y

mu

TRW2

p

TRPB 1 TRPB2

F

0

mu m

m

I

FIG. 3. Domain structures of Xenopus TRs. The A/B domain has been shown to have transcriptional activation properties for some TRs, although its role in frog TRs is uncertain. Domain C encodes the DNA-binding domain and E encodes the hormone-binding domain. A transactivation domain (AF2) is located at the carboxyl terminus encompassing the F and part of the E domain. Different boxes for A/B and F domains indicate divergent sequences among different receptors. The C,D, and E domains and part of the F domain are highly conserved.

60

YUN-BOSHI AND ATSUKOISHIZUYA-OKA

in different species. The rest of domains vary in size and sequences among different nuclear receptors. The N-terminal A/B domain is highly variable in sequence and length, the shortest being the TRB in X. la&s (Fig. 3) (35). At least in some TRs, this domain contains a transactivation function (Al?), although its role in amphibian TRs is unclear. Another transactivation function domain is the AF-2 domain, which is located at the very end of the C terminus (F domain and part of the E domain). Between the DNA- and ligand-binding domains is the D domain, or variable hinge region. This region often contains a nuclear localization signal and influences both DNA binding and transactivation (38,43,44), although the underlying mechanisms are unclear. 1. THYROID HORMONE BINDING BY TRs Specific binding of thyroid hormone by TRs requires the minimum thyroid hormone binding domain, which spans the C-terminal domain of about 250 amino acids (36, 45). Th is d omain shares only a low level of homology among different nuclear receptors. However, TRs from species as diverse as frogs and humans are over 90% identical in this region (35). Such conservation accounts for the essentially identical high affinities of TRs for T, (with K, in the subnanomolarity range) (29, 44, 46, 47). The crystal structure of the domain bound with a ligand reveals that the holo-ligand-binding domain consists of 12 o-helices with the ligand buried inside (48). The last helix (helix 12) encompasses the carboxyl terminus where the AF-2 domain (F domain) is located. Although the sequence of the ligandbinding domain for different nuclear receptors diverges extensively, the domain seems to maintain a similar overall structure. The crystal structure of the holo-TH-binding domain is remarkably similar to that of the holo-retinoic acid @.A)-binding domain of the retinoic acid receptor y (RARy) (49). It also resembles that of steroid hormone receptors (50, 51). Interestingly, the apoRA-binding domain of RXRa (52) h as a slightly different structure. Although it has the 12 a-helices in similar locations as seen in RARy and TRa, its last helix (helix 12) stretches away from the rest of the protein, in contrast to the ligand-bound RARy and TRo, where helix 12 wraps tightly against the rest of the ligand-binding domain, with the ligand buried inside. This difference suggests that on ligand binding, there is a conformational change in the ligand-binding domain that leads to the inward folding of the helix 12. Indeed, biochemical studies have provided direct evidence for conformational changes in TRs induced by TH binding (38,39, 53 - 55). 2. DNA BINDING BY TRs The DNA-binding domain of TRs mediates specific recognition of the thyroid hormone response elements (TREs) present in TH response genes

THYROIDHORMONE-REGULATEDTISSUEREMODELINC

61

(38, 39). The DNA-binding domain consists of two adjacent Zns+ fingers, each of which contains two histidine and two cysteine residues that coordinate a Zn2+ ion in a tetrahedral configuration (56). This coordination of the Zn2+ ions by the two Zn 2+ fingers determines the overall structure of the DNA-binding domain. TRs can bind to DNA as monomers, homodimers, and heterodimers formed with other members of the thyroid-retinoid receptor subfamily (36, 38, 40,45). The most stable complexes are those formed by heterodimers of TRs and RXRs (9-cis-retinoic acid receptors). A number of in vitro and in viva studies support the view that TIQRXR heterodimers are the true mediators of the biological effects of TH (38, 44, 57- 61). Many of the natural occurring TREs consist of two direct repeats with a 4-bp spacing and of sequences highly similar to AGGTCA (62-64). The binding of such a TRE by a TR/RXR heterodimer involves the recognition of the 5’ repeat by the RXR and the 3’ repeat by the TR of the heterodimer (65, 66). Such a complex has been directly visualized in an X-ray structure of a cocrystal of a TRE made of two direct repeats separated by 4 bp and a heterodimer consisting of the DNA-binding domain of a TR and an RXR (56). 6.

Transcriptional

Regulation by TRs

TH can both up- and down-regulate gene expression in target tissues or cells. Thus, depending on the target promoters and/or cell types, TH-bound TRs can either activate or repress transcription. The vast majority of the known TH response genes are up-regulated by the hormone and most studies of receptor function have been on these up-regulated genes. Relatively little is known about how liganded TRs repress transcription. The discussions here focus only on the mechanisms involved in regulating the transcription of TH up-regulated genes. 1. ACTIVATIONVS.REPRESSION Transcriptional activation by TH requires the binding of TRs, most likely as heterodimers with RXRs, to TREs present in the regulatory regions of the TH response genes. The binding of TREs by TR/RXR heterodimers is, however, independent of TH both in solution and in chromatin (33, 60). TH appears to bind to TRs and trigger conformational changes in TRs that activate the receptor function (53 - 55). Various experiments have revealed that in the absence of TH, TRs repress the transcription of TRE-containing promoters. In the presence of TH, TRs enhance the transcription from these same promoters (38, 57, 60, 67). Thus, unliganded TRs seem to function as repressors. The binding of TH leads to not only the relief of the repression by unliganded TRs but often, if not always, additional activation of the target gene. Thus, TR/RXR functions as a

62 dual transcriptional regulator-a activator when bound by TH.

YUN-BOSHIANDATSUKOISHIZUYA-OKA

repressor in the absence of the TH and an

2. COACTIVATORSANDCOREPRESSORS Both transcriptional repression by unliganded TRs and activation by Tabound TRs involve TR-interacting cofactors. Many such factors have been isolated based on their ability to interact with TRs in the presence or absence of T, or under both conditions (68 - 75). Th e corepressors bind preferentially or exclusively to unliganded TR while the coactivators have the opposite preference. a. Corepressors. Several corepressors have been cloned, including SMRT, N-CoR, SunCoR, and Alien (73, 75- 78). All are capable of interacting with unliganded TRs and have transcriptional silencing activity. Among them, SMRT and N-CoR have been studied extensively. These are two large proteins of about 2400 amino acids (SMRT also has a smaller form of 1500 amino acids, which was the original form identified). SMRT and N-CoR appear to be members of a related family, sharing considerable structural and sequence similarity. Both interact with the D domain of TRs. Interestingly, a mutation in the D domain of TR that abolishes the ability of the TR to suppress transcription also fails to interact with SMRT, again pointing out the potential involvement of SMRT in TR-mediated repression (76). In addition to the TR, these corepressors also interact with other members of the nuclear hormone superfamily. Both SMRT and N-CoR appear to form multimeric complexes through their interactions with corepressor Sin3. Sin3 in turn binds to histone deacetylases such as RPD3 (79, 80). Th us, the corepressor complexes are expected to be able to modulate the acetylation levels of histones and/or other proteins such as transcription factors. 6. Coactivators. The transcription coactivator proteins interact with nuclear receptors in the presence of the ligand. Often they can interact with many different nuclear hormone receptors and enhance the activities of these receptors when cotransfected with the receptors into mammalian tissue culture cells (68, 69, 71, 72). Further evidence for a role of coactivators in nuclear hormone receptor function has come from studies with gene knockout in mice. For example, inactivation of the coactivator SRC-1 in mice leads to partial resistance to steroid and thyroid hormones (81, 82). Like the corepressors, coactivators also form multimeric complexes containing other proteins and at least in the case of SRC-1 complex, an RNA (71, 74, 84). There are at least two types of coactivator complexes, those with histone acetyltransferase activity and those without. Many of the TR-interacting

THYROID

HORMONE-REGULATED

TISSUE REMODELING

63

coactivators, such as CBP/p300, SRC-1, and P/CAF, are histone acetyltransferases. Interestingly, at least some coactivator complexes contain different TR-interacting coactivators with histone acetyltransferase activities (71, 83), although the functional significance of the inclusion of multiple histone acetyltransferases in a single complex is unclear. A TR-interaction coactivator complex has been characterized that apparently lacks any histone acetyltransferase activity (84-86). On the other hand, it contains many subunits homologous to proteins of the yeast mediator complex, which associates with RNA polymerase II through the carboxylterminal repeat domain of the polymerase large subunit. Thus, it is very likely that the recruitment of this complex by TH-bound TR/RXR will activate the target promoter directly through the RNA polymerase complex. 3. ROLE OF CHROMATIN Most of the functional studies of hormone receptors were carried out in vitro or by transient transfection experiments in tissue culture cells. However, genomic DNA in eukaryotic cells is associated with histones and other nuclear proteins and is assembled into chromatin. Growing evidence indicates that chromatin structure plays important roles in regulating gene transcription (87- 89). It is known that transcriptionally active chromosome domains have distinct structure and protein compositions compared to repressed chromatin. In addition, transcriptional activation is often accompanied by chromatin reorganization. Thus, to understand the mechanism of TR action, it is important to use properly chromatinized templates. The Xenopus oocyte offers a unique system to study transcription due to its large storage of basal transcription factors as well as many other proteins important for early embryogenesis, including histones for chromatin assembly. Furthermore, it is easy to introduce exogenous genes into oocytes through microinjection. DNA injected into the oocyte nucleus is assembled into chromatin (90). Interestingly, the types of chromatin formed differ depending on the forms of the injected DNA. When double-stranded promoter-containing plasmid DNAs are used, they are chromatinized in 5-6 hr with less welldefined nucleosome arrays such that the transcription from the promoters is at high levels (Fig. 4A). In contrast, when single-stranded plasmid DNAs are used, they are quickly replicated (l-2 h r) an d assembled into chromatin in a replication-coupled chromatin assembly pathway, mimicking the genomic chromatin assembly process in somatic cells (90). The resulting templates produce much lower levels of transcriptional activity (Fig. 4A). Thus, by using different forms of promoter-containing DNA, it is possible to study the transcriptional regulation under different chromatin conditions. Xenopus oocytes have only a very low level of endogenous TR, insufficient to affect the transcription of a TRE-containing promoter (59, 61, 91).

64

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

Slow chmmatm

Active basal transcnption

coupled-with second strand DNA svntbesis. l-2 hr

Oocyte

ssDNA

B

dsDNA

TR/RXR

-

TSA

-

+

+

Repressed basal rranscnprion

ssDNA f

+

-

-

+

+

+

-

-

+

-

+

T3

+---

+

-+

C dsDNA

T-R

+ T3

Expt. b

+

---++

Internal) control

/’

2

3

4

5

6

7

8

9

10 \

1.09

0.15

1.13

1.15

0.12

0.95

0.03

0.93

1.25

FIG. 4. (A) Schematic diagram showing the difference in chromatin assembled from double-stranded (ds) and single-stranded (ss) plasmid DNA injected into a frog oocyte nucleus (NU) and the resulting transcriptional activity from the plasmid. (B) The histone deacetylase inhibitor TSA releases the transcriptional repression instigated by both chromatin and unliganded TR/ RXR. Groups of oocytes were first injected with (+) or without (-) TSA (5 ng/ml) or T, (50 nM) overnight. RNA was then prepared from the injected oocytes and the transcription from TH-dependent Xenopus TRpA promoter (pTRPA) was analyzed by primer extension (Expt.). The internal control represents the primer extension product derived from the endogenous storage pool of histone H4 mRNAs, serving as an RNA isolation and primer extension control. Levels of transcription from pTRPA were quantitated by phosphorimaging and were normalized against the internal control. The level of transcription from control ds pTRPA was designated as 1 (lane 1) and the other lanes were compared with it. (C) Liganded TR/RXR can relieve the repression

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

65

However, when exogenous Xenopus TRs and RXRs are cointroduced into the oocytes, they can repress the transcription from the double-stranded template containing a TRE (Fig. 4B; compare lanes 1 and 3) (TRs function much less efficiently in the absence of injected RXRs) (59, 60). Transcriptional repression of the promoter by the TR occurs even when single-stranded DNA is injected to allow the promoter DNA to be assembled into chromatin in the replication-coupled chromatin assembly pathway (Fig. 4B; compare lanes 6 and 8). Independent of whether double- or single-stranded promoter DNA is used, addition of TH leads to transcriptional activation (59, 60). The final transcriptional activity is essentially identical when either single- or doublestranded promoter DNA is used (Fig. 4B; lanes 5 and lo), indicating that THbound TRs can overcome any repression incurred by chromatin. Therefore, the central question for transcriptional activation is how the TR/RXR heterodimer does so. Recent studies have suggested that two levels of chromatin remodeling contribute to this gene regulation-histone acetylation and chromatin disruption.

a. Regulation of Hi&one Acetylation Levels through H&one AcetyltransHistone acetylation has long been implicated in ferases and Deacetylases. influencing gene expression (92- 96). Histone acetylation occurs at the lysine residues on the amino-terminal tails of the histones, leading to the neutralization of the highly positively charged histone tails and reduced affinity toward DNA (97). These changes alter the nucleosomal conformation and chromatin accessibility, allowing easier excess of transcription factors to chromatin templates (98-100). C onsequently, histone acetylation can increase gene transcription. Many TR-interacting coactivators are histone acetyltransferases (68, 69, 71, 72). Thus, liganded TR/RXR heterodimers may activate gene transcription in part through the recruitment of such coactivator complexes to alter histone acetylation levels. The opposite appears to be true for transcriptional repression by unliganded TR/RXR. This is because the corepressors NCoR and SMRT, which bind to unliganded but not Ta-bound TR/RXR, have been shown to form a complex containing histone deacetylases (79 - 80). This deacetylase complex formation involves the binding of N-CoR or SMRT to

established in the presence of histone deacetylase xRPD3. Groups of oocytes were injected with dsDNA of pTRPA, and with or without increasing amounts of xRPD3 mRNA as indicated (0.5 ng, lanes 2 and 4; 1 ng, lanes 3 and 5). After 14 hr some oocytes were injected with (+) TR/RXR rnRNAs and treated with (+) Ta for 14 hr, before the levels of transcription were analyzed by primer extension. From Ref. (101); J. Wong, D. Patterton, A. Imhof, D. Guschin, Y.-B. Shi, and A. P. Wolfe (1998). EMBOJ. 17,520-534, by permission of Oxford University Press.

66

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

the transcriptional repressor Sin3A, which in turn interacts with histone deacetylases such as RPD3. Studies on the regulation of Xenopus TRRA gene by TR/RXR have provided direct evidence for a role of histone acetylation in promoter activation (101). As described earlier, a double-stranded plasmid containing the TRRA promoter is highly transcribed when injected into the oocyte nucleus, and this transcription can be repressed by unliganded TR/RXR (Fig. 4B; lane 3). The addition of a specific inhibitor of histone deacetylase, trichostatin A (TSA), can reverse this repression, mimicking the addition of TH, while having no effect on the transcription in the absence of TR/RXR (Fig. 4B; lane 2), indicating the involvement of histone deacetylase in the repression by TR/ RXR. In addition, replication-coupled chromatin assembly of the singlestranded promoter injected into the oocyte nucleus also repressed gene transcription to a very low basal level (Fig. 4B; lane 6), which can be further repressed by unliganded TR/RXR (Fig. 4B; lane 8). TSA treatment relieves both types of repression (lanes 7 and 9), again just like the addition of T, (lane 10). In contrast to the deacetylase-blocking experiments, overexpression of the catalytic subunit RPD3 of a frog histone deacetylase complex leads to transcriptional repression of the promoter, and this repression can be reversed by the expression of TR/RXR in the presence of T, or the addition of TSA (Fig. 4C) (101). Th us, these two sets of complementary experiments strongly support a role of histone acetyltransferases/deacetylases in transcriptional regulation by TR/RXR. b. Chromatin Disruption. The frog oocyte system has also allowed us to investigate the structural changes of the plasmid minichromosome in response to gene regulation by TR/RXR. This has been carried out by using two assays (60, 61). The first is the plasmid DNA supercoiling assay. This assay is based on the fact that the wrapping of DNA around a nucleosome generates one negative supercoil on deproteinization. Any loss of nucleosomes or changes in the nucleosome-DNA wrapping will lead to alterations in superhelical density of the circular plasmid, which can be detected on an agarose gel (Fig. 5B). The second assay is based on the ability of micrococcal nuclease to digest preferentially the intemucleosomal spacer region. Thus, a partial digestion of the minichromosome with ordered nucleosome will provide a nucleosomal DNA ladder (Fig. 5C). Using these assays, we have found that T, binding to TRs bound to chromatinized templates causes the disruption of the ordered chromatin formed during replication-coupled chromatin assembly (Fig. 5B and 5C). Furthermore, this chromatin disruption occurs even when transcription is blocked by cx-amanitin (Fig. 5A and 5B). Thus, Ta-bound TR/RXR heterodimers can disrupt chromatin structure through an active process. These and other stud-

-

a-amdtin

-

-

+

+ + T3(50nM) a-amanitin

B -

1

2

+ -

Control + +

3

-

4

+ -

5

TWRXR

6

+ +

T3(5OnM)

C

I

--

2

-

3

4

5

Control

6

+

7 8

+

9 10 II 12 13 14 15 16

TWRXR

FE. 5. The transcriptional activation by liganded TR/RXR is accompanied by extensive chromatin disruption. (A) The transcriptional activation by liganded TR/RXR can be inhibited by a-amanitin. Groups of 20 oocytes were injected with ss pTRRA (Fig. 4) (100 ng/ml, 23 nhoocyte) and TR/RXR mRNAs (100 ng/ml, 27 nboocyte) and treated with (+) TH as indicated; o-amanitin was coinjected with ssDNA at a concentration of 10 ng/ml. The transcription was analyzed by primer extension, and the internal control is the primer extension product from an unknown endogenous mRNA (60). (B) The DNA topology assay indicates that liganded TR/RXR also induces extensive chromatin disruption and that this chromatin disruption is not the by-product of processive transcription. The injections were the same as in A. The DNA was purified from each group and the topological status of the DNA was analyzed using chloroquine agarose gel. The top band in each lane represents the nicked form of the plasmid. (C) Liganded TR/RXR instigates extensive chromatin disruption. The oocytes were injected with ssDNA (100 ng/ml, 23 nhoocyte) and TR/RXR mRNAs (100 ng/ml each, 27 nhoocyte) and treated with (+) hormone and processed for micrococcal n&ease (MNase) assay. The amounts of MNase used are 0, 10, 5, and 2.5 U, respectively. Note the disruption of the orderly nucleosomal ladder in lanes 14-16 when both TH and TR/RXR are present. Adapted from Ref. (61); J. Wong, Y.-B. Shi, and A. P. by permission of Oxford University Press. Wolffe (1997). EMBOJ. 16,3158-3171,

+ -

+ + -

123456

-

--

contmi TTfmxR

nwJdw

A

68

YUN-BO SHI AND ATSUKO ISHIZUYA-OKA

ies suggest that chromatin disruption and transcriptional activation are separable but both require the binding of transcriptionally active TR/RXR heterodimers (61). It is unclear how liganded TR/RXR disrupts chromatin. It does not appear to be due to changes in histone acetylation levels, because TSA treatment or overexpression of histone deacetylase RPD3 in frog oocytes has no detectable effects on chromatin structure as measured by these assays (101). On the other hand, studies ranging from yeast to mammals have suggested the involvement of the SNF/SWI family of proteins in chromatin remodeling (88, 89). Similar protein complexes may be involved in chromatin disruption by liganded TR/RXR. 4. PUTATIVE MODEL

OF TR ACTION

The cumulative information of transcriptional regulation by nuclear receptors has clearly indicated a complex, multistep, multicomponent nature of the underlying mechanism. A potential model for TR/RXR function is outlined in Fig. 6. In the absence of TH, TR/RXR recruits a corepressor and its associated deacetylase complex to the promoter, leading to histone deacetylation and transcriptional repression. On TH binding, the corepressor complex is dissociated and a coactivator complex is recruited to the promoter. This recruitment may lead to histone acetylation (102), chromatin disruption, and transcriptional activation. Although the studies so far are supportive of an important role of histone acetylation in transcriptional activation, other pathways are likely involved. First of all, histone acetyltransferases can also acetylate other proteins, such as general transcription factors (103), and other transcription factors, such as

, FIG. 6. A proposed mechanism for transcriptional regulation by TRs. TR functions as a heterodimer with RXR. In the absence of TH, the heterodimer represses gene transcription, likely through the recruitment of a corepressor complex containing the corepressor N-CoR or SMRT. The corepressor interacts with Sin3A, which in turn recruits a histone deacetylase such as RPD3 to deacetylate histones, thus affecting transcription. On binding by TH, a conformational change takes place in the heterodimer, which may be responsible for the release of the corepressor complex. Liganded TR also recruits a coactivator complex containing coactivators such as SRC-1, CBP/pSOO, and PCAF, and/or the DRIP/TRAP coactivator complex. The DRIP/TRAP complex may contact RNA polymerase directly to activate gene transcription. On the other hand, the SRC-1, CBP/pSOO, and PCAF complexes may function through chromatin modification because they possess histone acetylase activity. In addition, transcriptional activation is associated with chromatin disruption, which may be due to the recruitment of chromatin remodeling machinery by TR/RXR. This chromatin disruption may be necessary for transcriptional activation by TR/RXR. The corepressor and coactivator complexes as well as the chromatin remodeling machinery are multicomponent complexes and only a limited number of subunits are shown for simplicity.

69 THYROID

HORMONE-REGULATED

TISSUE

REMODELING

Basal Transcription

Repressed Transcription

Deacetylation

SRCl.p3OO.P/CAF Co-Activator Complex

Co-Activator Complex

Activated Transcription Histone Acetylation and Chromatin Disruption

70

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

p53 (104). Thus they may also affect transcription independent of histone acetylation. Furthermore, there is evidence that corepressors such as N-CoR can interact with basal transcription factors and inhibit transcription independent of their ability to recruit deacetylases. Likewise, the DRIP/TRAP coactivator complex can interact with the transcriptional machinery directly. In addition, chromatin disruption as detected by micrococcal nuclease digestion or plasmid DNA supercoiling assay appears to be necessary but not sufficient for transcriptional activation by liganded TR/RXR (61). On the other hand, overexpression of RPD3 or TSA treatment has little effect on chromatin structure based on two assays, despite their strong influence on transcription (101). Finally, there are many other TR-interacting proteins of yet unknown function, and they are likely to affect transcription through distinct mechanisms. Thus, further studies on the different cofactors and characterization of the nature of chromatin disruption and histone modification are needed to clarify the exact mechanism governing transcriptional regulation by TR/RXR.

IV. Thyroid Hormone-Dependent Morphological and Cellular Changes during Intestinal Remodeling The presence of a free-living larval form necessitates the remodeling of many existing organs for adult use as an animal changes often from being aquatic and herbivorous to being terrestrial and carnivorous. The intestine is one such organ that has been under extensive investigation (2,105, 106).

A. Larval Intestine Intestinal tissues are derived from the endoderm (the epithelium), the mesoderm (the connective tissue, muscles, and serosa), and the neural crest (the nerve). The epithelium is the tissue responsible for the principal physiological function of the organ, i.e., food digestion and absorption. It initially appears as a solid cell mass containing large amounts of yolk granules, which rapidly decrease after hatching (21, 107). This multilayered endoderm then gradually proliferates and differentiates into a simple columnar cell layer, the larval or primary epithelium (Fig. 7) (108). Unlike higher vertebrates, there are no villi and crypts in anuran tadpoles. There is a single fold, the typhlosole, in the anterior one-third of the small intestine in X. Zuevis (Fig. 8A), which is absent in other larval anurans. In addition, the differentiated larval epithelial cells are capable of dividing independently of their location within the typhlosole (109), in contrast to adult frogs or higher vertebrates, where the proliferating cells are less or not differentiated and are localized in the trough of an epithelial fold (frog) or crypt (higher vertebrates). Finally, there has been

\ Multi-layered endodermal cells

Monolayer columnar epithe cells on top of c

Differentiated larval intestine

Differentiated adult intestine

Xenopus laevis

FIG. 7. Comparison of intestinal development in Xenopus la&s and higher vertebrates. The primordial endodermal cells first form a multilayered cell mass. The endodennal cells are then converted into a monolayer of columnar epithelial (EP) cells tightly associated with the connective tissue (CT), which is derived from the mesoderm, through a basement membrane (basal lamina). Further development in amphibians diverges from that in higher vertebrates. In the latter, the columnar cells develop into multiply folded epithelium surrounded by elaborate connective tissue (stippled area) and muscles (derived from the mesoderm) (hatched area; MU). In Xenqnu.s, the epithelium remains as a simple tubular structure with only a single fold, the typhlosole. The differentiated epithelial cells in both cases have numerous microvilli in the brush border (bb) on the luminal surface for efficient nutrient processing and absorption. The Xenqnx intestine then undergoes a second phase of development that results in the replacement of larval epithelium with adult epithelium as well as extensive development of connective tissues and muscles.

Mesoderm

Primordial endoderm

EP Higher vertebrate

FIG. 8. Metamorphic transformations of the anterior intestine of Xenopus tadpoles, stained with hematoxylin (A, C-F) or methyl green-pyronin Y (B), or labeled by bromodeoxyuridine immunohistochemistry (G). (A) Cross-section of the larval intestine at stage 54. The larval epithelium (le) is a monolayer of columnar cells. The layer of connective tissue (ct) is thin except for the typhlosole (Ty). The muscle (m) is also thin. (B) Appearance of small islets of the adult epithelium between the larval epithelium and the connective tissue (arrows) at stage 60. (C, D) Development of adult epithelial( ae ) 1s1e t s and degeneration of the larval epithelium at stage 61. Mitotic cells are numerous in the islets (arrows in D). (E) F ormation of intestinal fold (F) at stage 62. (F) Many intestinal folds developed at stage 64. The adult epithelium differentiates into a simple columnar epithelium. (G) Adult intestine at the completion of metamorphosis at stage 66. Labeled nuclei of the adult epithelium (arrows) are localized in troughs of the intestinal folds. Bars, 50 km.

THYROID HORMONE-REGULATED

TISSUE REMODELING

73

no evidence for the existence of undifferentiated stem cells in the primary tadpole epithelium similar to those in the mammalian small intestine (110).

B. Metamorphic

Transformations

Around the onset of metamorphic climax the long larval small intestine begins to shorten, and this process continues until the end of metamorphosis (108, 111, 112). Following the beginning of this shortening, the morphogenesis of the intestinal folds occurs. These appear as several circular folds that run longitudinally and are straight along the gut axis, gradually increasing in number and height, and finally being modified into longitudinally zigzagged folds. Accompanying those anatomical changes are the transformations of all major tissues within the intestine. 1. EPITHELIAL

TRANSFORMATION

The epithelial transition from larval to adult form of the amphibian intestine can be divided into two processes, degeneration of the larval (primary) epithelium and development of the adult (secondary) epithelium (Fig. 8). Degenerative cellular changes occur around the onset of metamorphic climax. For example, the microvilli composing the brush border decrease in number and height, whereas lysosomes increase in number and in hydrolytic activity (113,114). At the cellular level, the larval epithelial cells die through apoptosis (programmed cell death). The resulting membrane-bound cellular and nuclear fragments, i.e., apoptotic bodies, are at least partially phagocytosed by macrophages (115). The macrophages are eventually extruded into the lumen while still retaining the apoptotic bodies. Concurrent with larval cell death, the adult epithelial cells are detected at the epithelial-connective tissue interface as small islets consisting of undifferentiated epithelial cells (Fig. SB). It is still unclear whether these adult epithelium cells are derived from a pool of undifferentiated cells in the larval epithelium (113) or transformed from differentiated larval cells (109). In any case, the primordia rapidly grow into the connective tissue through active cell proliferation and differentiate to form the secondary epithelium, replacing the degenerating primary epithelium (112,116), (Fig. 8D). With the progression of fold formation (Fig. SE and 8F), proliferative cells of the adult epithelium become localized in the trough of the folds (Fig. 8G) like those in the mammalian crypts (117), in contrast to the primary epithelium (see above). Thus, during metamorphosis, the larval intestine transforms into a structure with a cell renewal system analogous to that in mammals (Fig. 7) (110). 2. CONNECTIVE

TISSUE REMODELING

Growing evidence suggests a close relationship between the epithelium and the connective tissue during intestinal remodeling. When the epithelial

74

YUN-BO

SHI AND A’ISUKO ISHIZUYA-OKA

transition from the larval to adult form begins, the connective tissue suddenly increases in mitotic activity, cell number, and thickness (Fig. 8C). The connective tissue at this time consists of various types of cells, such as immature mesenchymal cells, fibroblasts, macrophages, and mast cells (106, 115, 118). Furthermore, remarkable changes in the connective tissue occur close to the epithelium. When the larval epithelium begins to degenerate, the basal lamina, which is thin through out the larval period, becomes thick in the entire region beneath the epithelium and remains thick until the larval epithelium disappears. In addition, throughout the thick basal lamina, fibroblasts, that possess well-developed rough endoplasmic reticulum often contact the epithelial cells. These cell contacts are most frequently observed around the primordia of the adult epithelium when the epithelial cells most actively proliferate. These observations suggest that the thickening of the basal lamina and the cell contacts are related to the larval epithelial cell death and the adult epithelial cell proliferation, respectively. At later stages, the basal lamina becomes thin beneath the adult epitheliurn. In addition, the cell contacts and all cell types of the connective tissue, except fibroblasts, decrease in number. By the end of metamorphosis, almost all of the connective tissue cells are ordinary fibroblasts. In the trough of the epithelial folds, these fibroblasts are close to the epithelium and aligned parallel to the curvature of the epithelial basal surface. This structure is similar to the subepithelial fibroblastic sheath reported to be present in the crypt of the mammalian small intestine. This sheath has been thought to play important roles in epithelial cell proliferation and/or differentiation (119). 3. OTHER TISSUES Very little information exists on other intestinal tissues. One such tissue, the muscle, becomes considerably thicker (Fig. 7) during metamorphosis, primarily due to the increase of the inner circular muscle layer. In contrast, after metamorphosis, the thickening is mainly due to that of the outer longitudinal muscle layer. Consequently, in adult Xenopus, the thickness of each layer is almost the same (220). Replacement of the neurons from larval to adult types in the myenteric plexus of the bullfrog intestine has been also observed during metamorphosis (121).

C. Organ Autonomous

Response to TH

Like other organs, the intestine can also be induced to undergo precocious remodeling by treating premetamorphic tadpoles with TH. In addition, this regulation is organ autonomous, for intestinal fragments cultured in vitro can be induced to metamorphose by including TH in the culturing medium. The changes induced by TH treatment mimics those in viva, that is, intestinal length reduction, degeneration of the larval epithelium, and devel-

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

75

opment of the adult epithelium (Fig. 9) (122,123). Interestingly, in the organ cultures, the development of the adult epithelium requires the addition of glucocorticoid and insulin in addition to TH, whereas the larval epithelial degeneration can occur in the presence of TH alone (122). Thus, even though TH acts directly on the intestine, other nonintestinal factors also influence intestinal remodeling in viva. Organ culture studies have also revealed an important role of the connective tissue in adult epithelial development. When X. Zaevis anterior small intestine, which contains the connective tissue-rich typhlosole, is treated with TH in u&-o, it undergoes both larval epithelial cell death and adult epithelial development (Fig. 9A). The adult epithelium then differentiates into a simple columnar epithelium expressing intestinal fatty acid-binding protein (IFABP) (Fig. 9B), wh ic h marks the differentiation of absorptive epithelial cells (124). However, when posterior small intestine, which has little connective tissue, is cultured with TH, only cell death is reproduced (Fig. 9C). On the other hand, coculturing either posterior or anterior epithelium with anterior connective tissue in the presence of TH leads to both larval cell death and adult tissue development (125). Culturing epithelium alone only produces TH-dependent epithelial apoptosis. Thus, cell death appears to be tis-

FIG. 9. Larval small intestine of the Xenopus tadpole undergoes TH-dependent changes in organ cultures; stained with hematoxylin (A, C), or intestinal fatty acid-binding protein immunohistochemistry (B). (A, B) Explants of the anterior intestine cultured with T, insulin, and cortisol. On day 5 of cultivation (A), a typical islet (is) grows by rapid cell proliferation (arrows). The connective tissue (ct) develops. On day 7 (B), th e ep’ rth eliurn (e) differentiates into a simple columnar epithelium possessing intestinal fatty acid-binding protein. (C) Explant of the posterior intestine on day 5 in the presence of TH. No islets are formed. The number of epithelial cells is small. Connective tissue cells are few. Bars, 20 urn.

76

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

sue autonomous whereas adult tissue development requires cell-cell and/or cell-extracellular matrix (ECM) interactions. More recently, by culturing isolated larval intestinal epithelial or fibroblastic cells in vitro, we have shown that these cells respond cell autonomously to TH in vitro. TH stimulates the proliferation of both cell types but causes apoptosis specifically for epithelial cells, leading to an increase in fibroblast cell number and a precipitous drop in epithelial cell number

b

bP 4,361 2322~ 2027'

A Fit?, +T3

564-

1250 0

1

2

3

4

5

Culture Time (Day) FIG. 10. (A) Contrasting effects of thyroid hormone on tadpole intestinal epithelial (Ep) and fibroblastic (Fib) cells. The fibroblasts and epithelial cells were isolated from stages 57158 of tadpole small intestine and then cultured on plastic dishes in the presence of 10% TX-depleted fetal bovine serum in the presence or absence of 100 nM T,. The live cells were counted daily by trypan blue staining. (B) The TH-induced intestinal epithelial cell death produces a typical nucleosomal-sized DNA ladder. Tadpole epithelial cells were cultured as in A for 1 day and total DNA was isolated and analyzed on a 1% agarose gel.

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

77

(Fig. 10A) (126, 127). This TH-induced cell death is typical of mammalian apoptosis both in the requirement of caspases and nucleases and in the generation of a nucleosome-size DNA ladder (Fig. 10B) (126, 127). The cell death in vitro has apoptotic morphology similar to that seen during natural development. Thus, at least for larval epithelial cells, TH functions by directly inducing a cell death program within the target cells. However, it is possible that extracellular events can modulate this process as well (see below).

V. TR Expression and Function during Intestinal Remodeling TRs are the presumed mediators of the causative effects of TH on anuran metamorphosis. Their function in vivo depends on the presence of RXRs. Thus, extensive studies have been carried out to analyze the expression of both TR and RXR genes in various organs in both X. Zaevis and R. catesbeiana. In general, strong correlations exist between TR and RXR expression and organ transformations (based on studies of the mRNA and to a lesser extent, protein levels) (91, 128-133).

A. Correlation

of Receptor mRNA Levels

with Intestinal Remodeling In the Xenopus intestine, the TRB and RXRy genes are up-regulated during intestinal remodeling (Fig. llA), paralleling the rise in plasma TH levels (19). Both TRa and RXRo mRNAs are expressed at similar or higher levels compared to TRB and RXRy, respectively, during intestinal development, although their expression does not change substantially (Fig. 11A) (59). Expression of the third RXR gene, RXRB, has not been studied in the intestine. These results indicate that all receptors are likely to be present during intestinal remodeling to mediate the organ autonomous effects of TH.

B. TH Activation of TRP Genes The Xenopus TRB genes are regulated by TH. This regulation appears to be ubiquitous. Furthermore, the up-regulation occurs within a few hours of TH treatment of premetamorphic tadpoles and is independent of new protein synthesis, arguing that the TRB genes are regulated by TRs (134-136). Analysis of the TRB genes shows that Xenopus TRB and TRBB genes span at least 70 kb of genomic DNA. Each gene has two alternative promoters. One of the promoters is expressed at low but constitutive levels whereas the other is highly expressed only in the presence of TH (134, 135). Analysis of

78

YUN-BO SHI AND ATSUKO ISHIZUYA-OKA

54

62

58

66

Developmental Stages

B connective tissue and muscle cell differentiation

I

connective tissue and muscle cell proliferation

Connective tissue

L__

Stage

54

56

58

60

61

+

62

63 epithelial

64

66

molphogenesis

,y’

FIG. 11. Correlation of receptor expression with intestinal remodeling. (A) Coordinated regulation of TR and RXR genes in the intestine. The mRNA levels were quantified from Northern blots of total intestinal RNA. Note that both TRf3 and RXRy are up-regulated during metamorphosis, whereas TRcx and RXRa expression remain fairly constant. However, the absolute levels of TRa and RXRa mRNAs appear to be higher than those of TRS and RXRy mRNAs, respectively (59). (B) TRS mRNA levels correlate with tissue-specific transformations in the intestine [based on in situ data of Shi and Ishizuya-Oka (141)]. For clarity, the mRNA levels in different tissues are plotted on different scales.

the inducible promoter reveals that both genes contain at least a strong TFtE consisting of two nearly perfect direct repeats of AGGTCA separated by 4 bp (60, 64, 137,138), which is likely responsible for the up-regulation of the TFtp genes by TH during metamorphosis in different organs. Consistent with the above data, up-regulation of the TRP genes in the in-

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

79

testine is not only independent of protein synthesis (139) but also is cell autonomous, because isolated larval epithelial cells are induced to expressed TRR genes when treated with TH in cell culture in vitro (T Amano and Y.-B. Shi, unpublished observation).

C. Cell Type-Dependent Temporal Regulation of TRP Genes Different tissues within the intestine undergo distinct changes at different developmental stages during metamorphosis. How such tissue specificity is controlled remains to be investigated. Several factors could contribute to this (140). First, TH levels within individual cells may be regulated through import and export controls. Second, cellular free TH levels may depend on the levels of cytosolic TH-binding proteins and intracellular TH metabolism. Finally, TR levels may vary in a tissue-dependent manner. By using in situ hybridization, we have carried out a detailed analysis of TRR gene expression during intestinal metamorphosis (141). Little or no TRP mRNA is present in any tissues within the premetamorphic intestine. Around stage 5 7, TRP mRNA becomes detectable in some larval epithelial cells, and by stage 59, just prior to the onset of larval epithelial apoptosis, all larval epithelial cells express high levels of TRR mRNA (Fig. 11B). Subsequently, as apoptosis takes place in the larval epithelium, the TRR mRNA levels are down-regulated (stage 61). Similarly, TRP expression is high in the proliferating adult epithelial cells as soon as they can be identified as cell islets, between the larval epithelium and the connective tissue. The TRR expression in the adult epithelial cells remains high until stage 62, when adult epithelial cell differentiation begins. Likewise, in the connective tissue and muscles, TRR mRNA levels are high when cells proliferate but are down-regulated as these cells differentiate to form the adult connective tissue and muscles, respectively (Fig. 11B). Thus, TRR appears to be involved in promoting both cell proliferation and apoptosis, depending on the cell types in which they are expressed. In particular, high levels of TRR mRNA appear to be incompatible with high degrees of differentiation. In the differentiated larval epithelial cells, high levels of TRR expression are associated with apoptosis. On the other hand, as the cells of the adult epithelium, connective tissue, and muscles begin to differentiate, they need to shut down their TRR expression to prevent likely deleterious (apoptotic) consequences associated with high levels of TRl3 mRNA. Although is it unknown whether TRo and RXR genes have similar cell typespecific regulations, the results suggest that TR levels may control the temporal regulation of metamorphosis and that cell type specificity of tissue transformations is determined by genes other than TRs.

80

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

VI. Thyroid Hormone Response Genes in the Intestine Like other tissue remodeling processes, intestinal metamorphosis involves many genes at various stages of development. The binding of TH to TRs will likely lead to the activation of some genes and repression of others. For simplicity, those genes that are regulated within 24 hr of TH treatment of premetamorphic tadpoles are referred to as early TH response genes, whereas the later ones are the late TH response genes. The direct TH response genes are those that are regulated at the transcriptional level by TRs. They may be either early or late response genes; in the latter case, the synthesis of another protein(s) is required for TR to regulate the target genes, consequently requiring longer TH treatment. To understand the molecular pathways leading to intestinal metamorphosis, a key step is to isolate and functionally characterize these various TH response genes.

A.

Early TH Response Genes

A differential screen is a powerful method to isolate genes whose mRNA levels differ in two samples (142). To isolate low-abundance mRNA species, various polymerase chain reaction (PCR)-based methods have been developed to remove selectively genes that are expressed at similar levels in the two samples being compared, while enriching and amplifying the genes whose mRNA levels differ in the two samples. One such method has been applied to amphibian metamorphosis in several organs (136, 143). This method has been used to identify early TH response genes in the remodeling intestine. This was done with intestinal RNAs isolated from stage 54 premetamorphic tadpoles that had been treated for 18 hr with or without 5 n MT,, close to the endogenous plasma T, concentration at the metamorphic climax (stage 62) (19). A total of 22 up-regulated and one down-regulated genes was isolated (139). Most of the genes respond to T, treatment very quickly (within a few hours) and their regulation by T, appears to be independent of new protein synthesis (Table I), suggesting that they are direct TH response genes and represent the first wave of gene regulation induced by TH. The identities of many of the up-regulated genes have been revealed through sequence analysis (Table I). Several of the direct response genes encode transcription factors and thus are likely involved in directly activating or repressing transcription of intermediate and/or late TH response genes. In addition, several genes encoding proteins varying from a transmembrane amino acid transporter to extracellular enzymes are also found to be among the early response genes in the intestine. These results suggest that TH simultaneously induces many intra- and extracellular events, which in turn cooperate to effect intestinal remodeling.

Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Up-regulated Down-regulated Down-regulated

TH/bZip Zn finger TF (BTEB)

Stromelysin-3 Sonic hedgehog Na+/PO, cotransporter IU12 Nonhepatic arginase

Collagenase3 Collagenase-4 Gelatinase-A

Pleiotrophic factor BMP-4

Ubiquitin-activating Collagens Calbindin Villin IFABP

direct direct direct direct

Late Late Late Late Late

Late Late

Late Late Late

Early, direct Early, direct Early, direct Early Early

Early, Early, Early, Early, Early

Kinetics”

signaling signaling Protein degradation ECM proteins Calcium metabolism Brush border structural protein Fatty acid metabolism

Cell-cell Cell-cell

144 144 217 145 123

144 144

139,177,212 139,177,212 139, 175,212

ECM remodeling ECM remodeling ECM remodeling

139, 211, 212 139, 212 139, 141, 212 139,210,212 139,212 139,175,212 204 139,212,214 139,212,215,216 139,212,213

regulation regulation regulation regulation regulation

Ref.

ECM remodeling Cell-cell signaling PO, transport Amino acid transport Proline biosynthesis

Transcriptional Transcriptional Transcriptional Transcriptional Transcriptional

Possible function

“A directresponseindicates that the regulation by TH is resistant to prokin synthesis inhibition; “early”or “late” refers to regulation by TH in the intestine requiring less or more than 1 day of treatment, respectively.

enzyme El

Response to TH

Protein encoded

Tw NFI bZip (Fra-2)

TABLE I THYROID HOHMONE RESPONSE GENES IN Xenopus INTESTINE

a2

YUN-BO

A

SHI AND ATSUKO

ISHIZUYA-OKA

B Stage

64 56 66 60 62 64 66

54565860

626466

56 54 58 62 66

012357

0

1

2

3

5

FIG. 12. Development of expression of three TH response genes during natural (A) and THinduced (B) metamorphosis. Intestinal RNA was isolated from tadpoles at different stages, or from stages 52/54 tadpoles treated with 5 nMT, for the indicated days, and subjected to North ern blot hybridization. TH/bZip and NFI-B are early response genes (210, 211), and IFABP is a late response gene (123). Adapted from Refs. 123,210, and 211 (Temporal and spatial regulation of a putative transcriptional repressor implicates it as playing a role in thyroid hormone-dependent organ transformation; A. Ishizuya-Oka, S. Ueda, and Y.-B. Shi, Deu. Genet. Copyright 0 1997, Wiley-Liss, Inc., a subsidiary of John Wiley & Sons, Inc.)

The first indication that these genes are likely to play important roles in tissue remodeling comes from their dramatic regulation in the intestine during both natural and TH-induced metamorphosis (e.g., Fig. 12 for TH/bZip and NFI-B). Developmentally, these early response genes fall into three general classes (136, 139). The genes in the first class [for example, TRP, a basic and leucine zipper (TH/bZip) motif-containing transcription factor (Fig. 12), the extracellular matrix-degrading metalloproteinase stromelysin-3 (Table I)] are expressed strongly only at the climax of intestinal remodeling (around

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

83

stages 60 - 62). Much lower levels of expression of these genes are present in pre- or postmetamorphic intestine. The second class of genes [for example, the NFI family of transcription factors (Fig. 12) (Table I)] are activated during metamorphosis and their expression remains high in postmetamorphic frog intestine. Finally, two genes, including the Na+/PO; cotransporter (Table I), are in the third class, They are expressed at high levels immediately before or after the climax or metamorphosis but minimally at the actual climax. Interestingly, most of the direct response genes are also regulated by TH in many other organs, even though different organs undergo vastly different transformations. For example, TRR genes are up-regulated by TH in all of the organs analyzed so far, including tail and limb, even though the tail resorbs and the limb undergoes de novo development. This contrasts with the late TH response genes, which are often intestine specific (see below and Table I). Thus, the ubiquitous early TH response genes are likely to function together with preexisting, intestine-specific factors to activate the downstream intestine-specific metamorphic pathways. For an unknown reason, there appears to be relatively few early response genes that are down-regulated by TH within 24 hr in the intestine (only one was isolated) (139). Th is is similar to the findings from similar screens in other Xenopus organs, with the exception of the brain (136, 143). At present, much less is known about the down-regulated genes. It is difficult to assess their importance during metamorphosis.

B. Late TH Response Genes A PCR-based subtractive screen has also been carried out to isolate intestinal genes regulated by a 4-day TH treatment of stage 57 premetamorphic Xenopus Zaevis tadpoles (144). This led to the identification of over 20 genes that are induced by TH. Most of these genes are distinct from the early TH response genes described above. Although no detailed kinetic study of their TH induction has been carried out, they are likely to be late TH response genes. Like the early response genes, these TH-induced genes also belong to distinct classes (144). They encode transcription factors, collagens, components of the ubiquitin proteasome pathway, morphogenetic and growth factors, etc. (Table I). They can potentially participate in the regulation of larval epithelial cell death and/or adult cell proliferation and differentiation at a step(s) downstream of the direct, early TH response genes. Through various means, a number of other genes have been identified as late TH response genes in the intestine. Many of these genes are specific to intestinal epithelium, such as the genes encoding intestinal fatty acid-binding protein (Fig. 12) and villin (an epithelial brush border structural protein).

84

YUN-BO SHI AND ATSUKO ISHIZUYA-OKA

Interestingly, both IFABP and villin are initially down-regulated by TH as the larval epithelial cells undergo apoptosis. Subsequently, they are up-regulated again as adult epithelium develops (123,124,145,146). Presumably, many other epithelium-specific genes behave similarly. The regulation of these genes is likely reflective of the terminal changes of the intestinal epithelium during metamorphosis.

VII. Functions of TH Response Genes: Implication from Studies on Matrix Metalloproteinases Among the early and late TH response genes are those encoding matrix metalloproteinases (MMPs; Table I). MMPs are extracellular enzymes that are capable of degrading various components of the ECM (147-150). This growing family of enzymes includes collagenases, gelatinases, and stromelysins, each of which has different but often overlapping substrate specificity (151, 152). They are secreted into the ECM as proenzymes with the exceptions of stromelysin-3 (ST3), which appears to be secreted in the active form (153), and membrane-type MMPs, which exist as membrane-bound active enzymes. The proenzymes are enzymatically inactive owing to the presence of a propeptide (149, 154-157). The proenzymes can be activated in the ECM or on the cell surface through the proteolytic removal of the propeptide (155-158). Th e mature enzyme has a catalytic domain at the N-terminal half of the protein, which contains a conserved Zn2+ binding site. Once activated, these MMPs can degrade components of ECM. Thus differential expression and activation of various MMPs can result in specific remodeling of the ECM. Many MMPs have been implicated to play a role in the metastasis of cancerous cells owing to the up-regulation of their expression in this process and the fact that metastasis requires extensive ECM degradation/modification (148, 159-161). In addition, the developmental expression profiles of MMP genes suggest that MMPs are also critical players in a number of developmental processes (162-165). The importance of MMPs in development has been substantiated by gene knockout studies in mice (166-168). The activation of MMP genes by TH during metamorphosis suggests a role for these extracellular enzymes in tissue remodeling. Such an idea was first proposed over 30 years ago, based on the drastic increase in collagen degradation activity in the resorbing tadpole tail (169), which led to the identification of the first MMP, the collagenase. Further studies on the expression profiles of MMP genes during X. laevis metamorphosis suggest that MMPs participate in ECM remodeling to influence tissue transformation.

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

85

A. ECM Remodeling during Intestinal Metamorphosis The intestinal epithelium is separated from the connective tissue by a special ECM, the basal lamina, which is composed of laminin, entactin, collagens, and proteoglycans, etc. (170, 171). In premetamorphic X. Zaevis tadpoles, the intestinal basal lamina is a continuous but thin structure separating the connective tissue and the epithelium. As the larval epithelium undergoes degeneration it becomes much thicker and multiply folded, and remains thick until the larval epithelium finally disappears, i.e., along with the massive epithelial apoptosis (118, 172). Interestingly, the basal lamina appears to be much more permeable at the climax of metamorphosis (stage 60- 63), in spite of the increased thickness. This permeability is reflected by the frequently observed migration of macrophages across the basal lamina into the degenerating epithelium, where they participate in the removal of degenerated epithelial cells. In addition, extensive contacts are present between the proliferating adult epithelial cells and the fibroblasts on the other side of the basal lamina (118, 125). L arval epithelial cell removal is essentially complete around stages 62-63. After stage 63, with the progress of intestinal morphogenesis, i.e., intestinal fold formation, the adult epithelial cells differentiate. Concurrently, the basal lamina become thin and flat again, underlining the differentiating adult epithelium. The mechanism governing the ECM remodeling are unclear at present. Several factors may contribute to it. First, as the larval intestine reduces its length, the ECM may increase in thickness due to contraction. Second, synthesis of new ECM components will lead to changes in the composition and nature of the basal lamina. This is supported by the finding that at least some collagen-encoding genes are activated by TH during intestinal metamorphosis (Table I) (144). Finally, ECM degradation by MMPs is likely to be an important aspect.

B. Regulation of MMP Genes during Metamorphosis Several frog MMP genes have been found to be up-regulated by TH during metamorphosis. Among them, the collagenase-1 (Coil) gene of R. cutesbeiana and stromelysin-3 (ST3) of X. Zuevis are early, direct TH response genes (173 - 175). The Xenopus Co/3 was isolated as an early gene in the tail, but its regulation in the intestine by physiological concentrations of TH requires more than 1 day of TH treatment, similar to Xenopus Co/4 (176,177). Thus, Co/3 and Co/4 are likely late TH response genes. In addition, Northem blot analyses with human gelatinase A (GeZA) and ST1 cDNA probes suggest that Xenopus GeZA is a late TH response gene whereas the expression of Xenops ST1 changes little during intestinal remodeling (175). Spatial and temporal expression ofxenopus MMP genes implicates different function for different MMPs.

86

YUN-BOSHIANDATSUKOISHIZUYA-OKA

1. ST3 EXPRESSIONISASSOCIATED WITHTH-INDUCEDCELLDEATH

Among the known amphibian MMPs, the Xenopus stromelysin-3 gene is of particular interest. This is in part because its human homolog is expressed in most human carcinomas (178, 179). Furthermore, both the human and mouse ST3 genes are expressed during development in tissues where cell death takes place (178,180, 181). These results suggest that ST3 is involved in both apoptosis and cell migration, the processes that also occur during frog intestinal remodeling. The developmental expression of ST3 mRNA correlates strongly with organ-specific metamorphosis (Fig. 13A) (175). Furthermore, ST3 expression is temporally correlated with the stages when cell death occurs in all organs analyzed, and the levels of its mRNA appear to correlate with the extents of cell death in these organs (175). More importantly, the activation of the ST3 gene occurs at the onset of or prior to cell death. Thus, in the intestine, high levels of ST3 mRNA are already present at stage 60, when larval apoptosis is first detected by terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) (Fig. 13A) (182, 183). These results suggest that ST3 is involved in regulation of larval cell death. However, because adult epithelial proliferation in the intestine is rapid around stages 60-62 and differentiation also starts around stage 62, it is possible that ST3 may also be involved in adult epithelial development. In situ hybridization analysis has revealed that ST3 expression is spatially correlated with apoptosis in different organs during Xenopus metamorphosis (182-185). In the intestine, ST3 mRNA is localized in the fibroblastic cells adjacent to the epithelium, but not actually in the apoptotic cells of the intestine (Fig. 13B) (175, 183). Thus, as an MMP, ST3 influences epithelial apoptosis by modifying the ECM, in particular the basal lamina separating the STS-expressing fibroblasts and the dying epithelial cells. Consistent with this, ST3 expression is not only temporally but also spatially correlated with the modification of the basal lamina (the ECM that separates the epithelium and connective tissue) (Fig. 14) (182). In both pre- and postmetamorphic intestine, fibroblasts just beneath the thin and flat basal lamina do not express ST3. However, during metamorphosis, as STS-expressing fibroblasts increase in number near the muscular layers (Fig. 14A), the intestinal basal lamina adjacent to them begins to fold (Fig. 14B and 14C). Then at the highest levels of ST3 mRNA just beneath the degenerating epithelium (Fig. 14D), the basal lamina attains its maximal thickness (Fig. 14E) but becomes more permeable, as described above. It is, therefore, possible that ST3 causes specific degradation/cleavage of certain ECM components, aiding in the folding of the ECM and resulting in increased permeability.

THYROIDHORMONE-RECULATEDTISSUEREMODELING

87

Such modifications may facilitate larval epithelial apoptosis and adult epithelial development. Although such a function for ST3 remains to be proved, it is interesting to note that overexpression of the MMP stromelysin-1 in the mammary gland of transgenic mice leads to similar changes in the basal lamina, alters mammary gland morphogenesis, and induces apoptosis (185a187). Furthermore, homologous deletion of ST3 results in mice with reduced incidence of carcinogen-induced tumors whereas overexpression of ST3 in cultured cells leads to increased tumor formation when the ST3-expressing cells are injected into mice (188,189), a g ain supporting a role of ST3 in ECM remodeling to influence cell behavior. 2. DIFFERENTIALEXPRESSIONOFDIFFERENTMMPGENES IMPLICATESDIFFERENTFUNCTIONS

The multicomponent nature of the ECM suggests the participation of different MMPs in its remodeling and degradation during metamorphosis. Thus, it is not surprising to find multiple MMP genes up-regulated during metamorphosis. Northern blot and in situ hybridization analysis reveal unique but overlapping expression profiles for the five Xenopus MMP genes in the intestine (Fig. 13) (175,177). ST3 is the most up-regulated gene whereas ST1 has relatively constant mRNA levels throughout development (Fig. 13A). The two collagenase genes have only a few fold higher levels of mRNA in the metamorphosing intestine around stage 62 compared with those at other stages (Fig. 13A). The putative Xmopus GelA has an expression profile that is most similar to that of ST3 (Fig. 13A). As described above, ST3 is expressed throughout the connective tissue underlying the remodeling basal lamina and degenerating larval epithelium during intestinal metamorphosis (Figs. 13B and 14) (182, 183). In contrast, Co13 is expressed only in sporadic regions/cells within the connective tissue, and little Co14 mRNA can be detected by in situ hybridization (Fig. 13B). Thus, Co13 and Co14 may be involved in connective tissue remodeling to facilitate adult intestinal morphogenesis and play at most a minor role in larval epithelial degeneration. GelA expression in the intestine resembles temporally that of ST3 (Fig. 13A). However, high levels of GelA mRNA in the intestine are reached later (at stage 62) than th ose of ST3 (at stage 60; Fig. 13A). This difference is also observed when premetamorphic tadpoles are treated with T, to induce precocious intestinal remodeling (175). Although the Xenopus ST3 gene is upregulated very quickly (within a few hours) by the T, treatment, the GelA gene up-regulation is detectable only after a treatment of 3 days or longer. As described above, stage 62 is the time when intestinal epithelial cell death is mostly complete and adult epithelial differentiation is taking place. The activation of GelA, and to a lesser extent Co14 (Fig. 13A), at this stage in the in-

YUN-BO SHI AND ATSUKO ISHIZUYA-OKA

a

54565860626466

co13

co14

GelA

ST3

FIG. 13.

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

89

b

ST3

co14

TUNElL

FIG. 13. (continued) (a) Northern blot analyses demonstrate differential regulation of different MMP genes during Xenqnus intestinal metamorphosis. Each lane had 10 pg total intestinal RNA. The probes used were human stromelysin-1 (STI), Xenopus collagenase-3 (Co13), Xenopus collagenase-4 (Co14), human gelatinase A (GelA), and Xenopus ST3. (b) association of ST3 but not Co13 or Co14 expression with larval epithelial cell death during intestinal remodeiing. Panels A-C: In situ hybridization with ST3, Co13, and Co14 antisense RNA probes, respectively, or anterior typhlosole-containing sections of the small intestine at stage 60. Note that ST3 is highly expressed in essentially all regions of the connective tissue (ct) underlying the larval epithelium (e), which undergoes apoptosis (see D), whereas Co13 is only sporadically expressed in the connective tissue and Co14 has no detectable expression. None of the MMPs are expressed in the epithelium. (D) TUNEL assay for apoptotic cells or anterior small intestine at stage 60. Note that labeled apoptotic cells (arrows) are present throughout but are essentially limited to the epithelium; 1, intestinal lumen; m, muscle. Bar, 300 pm.

testine suggests that these MMPs are involved in the removal of the ECM associated with the degenerated larval epithelium and/or remodeling of the ECM for adult epithelium differentiation. The earlier activation of the ST3 gene in the intestine is likely to be important for the ECM remodeling that facilitates larval epithelial apoptosis. How these MMPs influence tissue remodeling is yet unknown. Interestingly, mammalian ST3 is secreted in its enzymatically active form when overexpressed in tissue culture cells (153). This has allowed the identification of one potential physiological substrate, the ol-proteinase inhibitor, a non-

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

FIG. 14. Remodeling of the basal lamina correlates with ST3 expression in the intestine during metamorphosis. (A and D) In situ hybridization using an antisense ST3 probe on cross-sections of the anterior region of the small intestine. (A) At stage 59, hybridization signals (arrows) are observed in some cells of the connective tissue (ct) near the muscular layer (m), but are weaker in the upper region of the typhlosole (Ty). (D) Small intestine at stage 61. Most of the connective tissue cells just beneath the epithelium (e) are positive, i.e., ST3 expressing (arrows). Bars, 20 (J-m. (B, C and E) Electron micrographs of the epithelial-connective tissue interface of the small intestine. (B) Upper region of the typhlosole at stage 59, when ST3 expression is weak. The basal lamina (Bl) remains thin. (C) Bottom region of the typhlosole at stage 59, when ST3 expression is strong. The basal lamina begins to fold. (E) Th’rck ened basal lamina at stage 6 1, when ST3 is highly expressed. The basal lamina is vigorously folding into accordion-like pleats. Bars, 1 km.

ECM-derived serine proteinase inhibitor (190). Although no in viva ECM substrates of ST3 have been identified, such substrates may exist. One the other hand, the ability of ST3 to cleave a non-ECM substrate raises the possibility that ST3 may affect cell behavior through both ECM- and non-ECMmediated pathways. The other frog MMPs-STl, GelA, Co13, and Col4-are expected to digest specific ECM substrates just like their mammalian coun-

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

91

terparts, thus leading to the remodeling and degradation of the ECM during metamorphosis.

C. Extracellular

Matrix

and Cell Fate Determination

The remodeling of theECM during metamorphosis together with the correlation of the expression of MMPs, especially ST3, with apoptosis suggests that ECM remodeling plays a role in TH-induced apoptosis. Direct support for a role of ECM on cell fate has come from a number of studies with mammalian systems. One of the best-studied systems is the involution of the mammary gland. As the epithelial cells undergo postlactation apoptosis, a number of MMP genes are activated (191, 192). More importantly, by culturing the epithelial cells on different ECM matrices, it has been shown that the ECM can directly influence cell differentiation and survival (193, 194). Similarly, the ECM has been shown to be essential for survival of several other types of cells, and blocking the function of ECM receptor integrins can induce cell death (195-199). As described above, tadpole intestinal epithelial cells can be cultured in vitro and induced to undergo apoptosis by TH just as in viva (126, 127). When the plastic culture dishes are coated with various ECM proteins, the cells become resistant to TH-induced cell death (Fig. 15) (127). Consistent with this apoptosis-inhibiting effect of the ECM coatings, when proliferating/ differentiating adult epithelial cells of the intestine at stage 64 are cultured in vitro on plastic dishes, they too undergo TH-induced apoptosis. In viva, these adult cells proliferate and differentiate instead of undergoing apoptosis in the presence of high levels of circulating plasma TH. Thus, dissociating the adult cells from the ECM alters their response to TH. The mechanism by which the ECM influences cellular function is still unknown. Clearly, one way to transduce the ECM signal into the cells is through cell surface ECM receptors, especially integrins (195,200-203). In the case of mammary gland development, it has been proposed that the interaction of the ECM with its integrin receptors leads to the activation of a focal adhesion (tyrosine) kinase (FAK), which in turn transduces the signal through the MAP kinase pathway to the nucleus (194, 205). This or similar mechanisms may be responsible for ECM-dependent gene transcription and cell fate determination.

VIII. Conclusions and Prospects Anuran metamorphosis bears many similarities to postembryonic organ development in higher vertebrates (4, 206), including maturation of a number of adult tissues/organs such the intestine and brain. Similarly, transition

YUN-BO SHI AND ATSUKO ISHIZUYA-OKA

92

1.2 c

1.0

s ’

=

8

g ._ % 2

. 08

+

Plastic

+

Laminin

+

Flbronectim

+

Collagen IV

-O-

Collagen I

0.6 0.4

T3 OW FIG. 15. Epithelial cells from stages 57/58 intestine culture on matrix-coated dishes are more resistant to Ta-induced apoptosis. The epithelial cells were cultured on va.rious dishes in the presence of different concentrations of T, for 3 days and DNA fragmentation was then determined by an enzyme-linked immunoassay method.

from fetal to adult hemoglobin and the activation of liver albumin genes, etc., occur during postembryonic development of both mammals and anurans. Finally, a number of key processes during this period are dependent on the presence of TH in mammals as well. Clearly, TH plays a more critical role in anuran metamorphosis than in mammalian postembryonic development. The absolute dependence of anuran metamorphosis on TH has made it one of the oldest systems to study vertebrate development. This has led to the accumulation of extensive morphological, cellular, and biochemical information on the process and the establishment of a number of in vioo and in vitro systems to study TH-induced cell death and proliferation/differentiation required for tissue remodeling. The current molecular studies have strengthened the idea that TH induces a gene regulation cascade within each metamorphosing tissue through TRs. Many of the direct TH response genes have been isolated. These genes belong to several diverse groups encoding intracellular, extracellular, and membrane-bound proteins, suggesting that TH simultaneously induces intra- and extracellular signal pathways to effect tissue transformation (Fig. 16). The direct response genes encoding transcription factors are expected to di-

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

----+

Various functions

FIG. 16. A gene regulation cascade model for TH regulation of tissue remodeling. The binding of TH to TR/RXR heterodimers leads to conformational changes that activate the receptor complex. The activated TR/RXR heterodimers then up-regulate several classes of TH response genes, which in turn affect intra- or extracellular events, resulting in ultimate changes in cell fates. Note that TH-bound TRiRXR can also repress gene expression. However, such down-regulated genes in the intestine, if any, have yet to be identified and are therefore not included in the figure.

rectly regulate downstream gene transcription while others, such as the MMPs, will indirectly regulated gene expression through intracellular signaling cascades and/or cell-cell and cell-ECM interactions (Fig. 16). One key question that needs to be addressed is whether the early, direct TH response genes are involved in the regulation of the known late TH response genes. It is likely that other genes may play a mediatory role in transducing the upstream signals to the late response genes. The identification of such genes will also be important to the understanding of the TH-induced gene regulation cascade. Perhaps the most urgent need is to investigate directly the roles of various T, response genes during metamorphosis. Existing organ and cell culture systems will be very useful for this purpose, especially for genes that encode extracellular proteins, because overexpressed proteins or functionblocking antibodies, inhibitors, and dominant negatives can be added to the culture systems to modulate the levels and functions of the endogenous proteins. The ability to manipulate Xenopus embryos through microinjection of DNA or RNA encoding proteins of interest into fertilized eggs serves as another means to study gene function. For example, by overexpression TRs and RXRs in developing embryos, we have shown that both TRs and RXRs are required to efficiently mediate the developmental effects of T, and for specific regulation of several genes that are known to be regulated by TH during metamorphosis in embryos (207). Finally, the more recently developed transgenic Xenopus technology (208) will allow study of gene function directly in metamorphosing tadpoles, as has been done for the TH response gene encoding a type III deiodinase (209).

94

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

ACKNOWLEDGMENT We thank Ms. Kieu Pham for preparing the manuscript.

REFERENCES I. M. H. I. Dodd and J. M. Dodd, The biology of metamorphosis. In “Physiology of the Amphibia” (B. Lofts, ed.), pp. 467-599. Academic Press, New York, 1976. 2. L. I. Gilbert and E. Frieden, “Metamorphosis: A Problem in Developmental Biology,” 2nd Ed. Plenum Press, New York, 1981. 3. B. S. Hetzel, “The Story of Iodine Deficiency: An International Challenge in Nutrition.” Oxford University Press, Oxford, 1989. 4. J. R. Tata, BioEssays 15,239-248 (1993). 5. H. C. Freake and J. H. Oppenheimer, Annu. Rev. Nutr. 15,263-291 (1995). 6. D. L. Guernsey and I. S. Edelman, Regulation of thermogenesis by thyroid hormones. In “Molecular Basis of Thyroid Hormone Action” (J. Oppenheimer and H. Samuels, eds.), pp. 293-324. Academic Press, New York, 1983. 7. J. E. Silva, Thyroid 5,481-492 (1995). 8. J. H. Dussault and J. Ruel, Annu. Rev. Physiol. 49,321-334 (1987). 9. F. Debruyne, M. Vanderschueren-Lodeweyckx, and P. Bastijns, Audiology 22,404-409 (1983). 10. G. R. DeLong, J. B. Stanbury, and R. Ficro-Ben&, Dev. Med. Child Neural. 27,317-324 (1985). 11. V.T. Anand, S. B. S. Mann, R. J. Dash, andY. N. Mehra, Acta Otoluyngol. 10883-87 (1989). (1988). 12. S. A. Rivkees, H. H. Bode, and J. D. Crawford, N. EngZ. J. Med. 318,599-602 13. H. L. Schwartz, Effects of thyroid hormone on growth and development. In “Molecular Basis of Thyroid Hormone Action” (J. Oppenheimer and H. Samuels, eds.), pp. 293-324. Academic Press, New York, 1983. 14. W. H. Dillman, Thyroid 9, l-11, (1996). 15. J. A. Franklyn and M. D. Gammage, Trends Endotinol. Met& 7,50-54 (1996). 16. J. F. Gudernatsch, Arch. Entwickl. Org. 35, 457-483 (1912). 17. E. C. Kendall, Trans. Assoc. Am. Phys. 30,420-449 (1915). 18. E. C. Kendall, Am.J. Physiol. 49, 136-137 (1919). 19. J. Leloup and M. Buscaglia, C. R. Acad. Sci. 284,2261-2263 (1977). 20. B. A. White and C. S. Nicoll, Hormonal control of amphibian metamorphosis. In “Metamorphosis: A problem in developmental biology” (L. I. Gilbert and E. Frieden, ed.), pp. 363-396. Plenum Press, New York, 1977. 21. P. D. Nieuwkoop and J. Faber, “Normal table ofxenopus Zaevis.” North Holland Publ., Amsterdam, 1956. 22. J. R. Tata, Dev. BioZ. l&$415-440 (1968). (1918). 23. B. M. Allen,J. Exp. ZooZ. 24,499-519, 24. H. Fox, “Amphibian Morphogenesis.” Humana Press, Clifton, New Jersey, 1983. 25. D. L. St. Germain, Trends Endocrinol. Met& 5,36-42 (1994). 26. D. L. St. Germain and V. A. Galton, Thyroid 7,655-668 (1997). 27. K. B. Becker, M. J. Schneider, J. C. Davey, and V. A. Galton, Endocrinology 136,44244431 (1995). 28. K. B. Becker, K. C. Stephens, J. C. Davey, M. J. Schneider, and V. A. Galton, Endocrinology 138,2989-2997 (1997).

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

29. J. C. Davey, K. B. Becker, M. J. Schneider,

30. 31.

32. 33. 34. 35 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52.

53. 54. 55.

95

D. L. St. Germain, and V. A. Galton, J. Biol. Chem. 270,26786-26789 (1995). D. L. St. Germain, R. A. Schwartzman, W. Croteau, A. Kanamori, Z. Wang, D. D. Brown, (1994). and V. A. Galton, Proc. Natl. Acad. Sci. U.S.A. 9l, 7767-7771 V. A. Galton, Thyroid hormone action in amphibian metamorphosis. In “Molecular Basis ofThyroid Hormone Action” (J. H. Oppenheimer and H. H. Samuels, eds), pp. 445-483. Academic Press, New York, 1983. J. H. Oppenheimer, Science 203,9 7 1- 9 79 (19 79). .4. J. Perlman, F. Stanley, and H. H. Samuels,J. Biol. Chem. 257,930-938 (1982). A. R. Brooks, G. Sweeney, and R. W. Old, Nucleic Acids Res. 17,9395-9405 (1989). Y. Yaoita, Y.-B. Shi, and D. D. Brown, Proc. Natl. Acad. Sci. U.S.A. 87,7090-7094 (1990). M. A. Lazar, Endocr. Rec. 14,184-193 (1993). R. M. Evans, Science 240,889-895 (1988). (1994). M.-J. Tsai and B. W. O’Malley, Annu. Reu. Biochem. 63,451-486 P. M. Yen and W. W. Chin, Trends Endocrinol. Metab. 5, 65-72 (1994). D. J. Mangelsdorf, C. Thummel, M. Beato, P. Herrlich, G. Schutz, K. Umesono, B. Blumberg, P. Kastner, M. Mark, P. Chambon, and R. M. Evans, Cell 83,835-839 (1995). Y.-B. Shi, J. Wong, and M. Puzianowska-Kuznicka, J. Biomed. Sci. 3,307-318 (1996). A. Krust, S. Green, P. Argos, Kumar, P. H. Walter, J.-M. Bomert, and P. Chambon, EMBO J. 5, 891-897 (1986). R. Uppaluri and H. C. Towle, Mol. CeZZ.Biol. 15, 1499-1512 (1995). M. Puzianowska-Kuznicka, J. Wong, A. Kanamori, and Y.-B. Shi, J, Biol. C&m. 271, 33394-33403 (1996). B. M. Forman and H. H. Samuels, Mol. Endocrinol. 4,1293-1301 (1990). J. Sap, A. Munoz, K. Damm, Y. Goldberg, J. Ghysdael, A. Leutz, H. Berg, and B. Vennstrom, Nature &on&n) 324,635-640 (1986). C. Weinberger, C. C. Thompson, E. S. Ong, R. Lebo, D. J. Gruol, and R. M. Evans, Nature &on&n) 324,641-646 (1986). R. L. Wagner, J. W. Apriletti, M. E. McGrath, B. L. West, J. D. Baxter, and R. J, Fletterick, Nuture (London) 378,690-697 (1995). J.-P. Renaud, N. Rachel, M. Ruff, V Vivat, P. Chambon, H. Gronemeyer, and D. Moras, Nature (London) 378,681-689 (1995). D. M. Tanenbaum, Y. Wang, S. P. Williams, and P. B. Sigler, Proc. NatZ. Acad. Sci. U.S.A. 95,5998-6003 (1998). (1998). D. Moras and H. Gronemeyer, Cum Opin. Cell BioZ. 10,384-391 W. Bourguet, M. Ruff, P. Chambon, H. Gronemeyer, and D. Moras, Nature (London) 375, 377-382 (1995). M. K. Bhat, C. Parkison, P. McPhie, C. M. Liang, and S. Y. Cheng, Biochem. Biophys. Res. Commun. 195 385-392 (1993). J. H. Toney, W. Ling, A. E. Summerfield, G. Sanyal, B. M. Forman, J. Zhu, and H. H. Samuels, Biochemistry 32,2-6 (1993). P. M. Yen, A. Sugawara, M. Forgione, R. A. Spanjaard, E. Macchia, S. Y. Cheng, and W. W. Chin, Mol. CeZl. Endocrinol. 97,93-99 (1993).

56. E Rastinejad, T. Perlmann, R. M. Evans, and P. B. Sigler, Nature (London) 375,203-211 (1995). 57. J. D. Fondell, A. L. Roy, and R. G. Roeder, Genes Der;. 7,1400-1410 (1993). 58. D. J. Mangelsdorf and R. M. Evans, Cell 83,841-850 (1995). 59. J. Wong and Y.-B. Shi,J. BioZ. Chem. 270, 18479-18483 (1995). 60. J. Wong, Y.-B. Shi, and A. P. Wolffe, Genes DeG. 9,2696-2711 (1995). 61. J. Wong, Y.-B. Shi, and A. P. Wolffe, EMBOJ. 16,3158-3171 (1997).

96

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

62. K. Umesono, K. K. Murakami, C. C. Thompson, and R. M. Evans, Cell 65, 1255-1266 (1991). 63. A. M. Naar, J.-M. Boutin, S. M. Lipkin, V. C. Yu, J. M. Holloway, C. K. Glass, and M. G. Rosenfeld, CeZZ65, 126 7- 12 79 (199 1). 64. M. Ranjan, J. Wong, and Y.-B. Shi,J. Biol. Chem. 269,24699-24705 (1994). 65. T. Perlmann, P. N. Rangarajan, K. Umesono, and R. M. Evans, Genes Deu. 7, 1411-1422 (1993). 66. R. Kurokawa, V. C. Yu, A. Naar, S. Kyakumoto, S. Han, S. Silverman, M. G. Rosenfeld, and (1993). C. K. Glass, Genes Dev. 7,1423-1435 67. A. P. Wolffe, J. Wong, Q. Li, B.-Z Levi, and Y.-B. Shi, Biochem. Sot. Trans. 25, 612-615 (1997). 68. J. D. Chen and H. Li, 0%. Rev. Eukay. Gene Express. 8,169-190 (1998). 69. R. J. Koenig, Thyroid 8, 703-713 (1998). 70. C. Rachez, Z. Suldan, J. Ward, C.-I? Betty Chang, D. Burakov, H. Erdjument-Bromage, I? Tempst, and L. P. Freedman, Genes Den lZ,1787-1800 (1998). (1999). 71. N. J. McKenna, R. B. Lanz, and B. W. O’Malley, Endocr. Rev. 20,321-344 72. L. Xu, C. K. Glass, and M. G. Rosenfeld, Cum @in. Genet. Dev. 9,140-147 (1999). 73. U. Dressel, D. Thormeyer, B. Altincicek, A. Paululat, M. Eggert, S. Schneider, S. P. Ten(1999). baum, R. Renkawitz, and A. Baniahmad, Mol. Cell. BioZ. 19,3383-3394 74. R. B. Lanz, N. J. McKenna, S. A. Onate, U. Albrecht, J. Wong, S. Y. Tsai, M.-J. Tsai, and B. W. O’Malley, Cell 97,17-27 (1999). 75. I. Zamir, J. Dawson, R. M. Lavinsky, C. K. Glass, M. G. Rosenfeld, and M. A. Lazar, Proc. Nad. Acad. Sci. U.S.A. 94, 14400-14405 (1997). 76. J. D. Chen and R. M. Evans, Nature (London) 377,454-457 (1995). 77. A. J. Horlein, A. M. Naar, T. Heinzel, J. Torchia, B. Gloss, R. Kurokawa, A. Ryan, Y. Kamei, M. Soderstrom, C. K. Glass, and M. G. Rosenfeld, Nature (London) 377,397-404 (1995). 78. E.-J. Park, D. J. Schroen, M. Yang, H. Li, L. Li, and J. D. Chen. Proc. NatZ. Acad. Sci. U.S.A. 96,3519-3524 (1999). 79. T. Heinzel, R. M. Lavinsky, T M. Mullen, M. Solderstrom, C. D. Laherty, J. Torchia, W. M. Yang, G. Brard, S. D. Ngo, J. R. Davie, E. Seto, R. N. Eisenman, D. W. Rose, C. K. Glass, and M. G. Rosenfeld, Nature (London) 387,43-48 (1997). 80. L. Nagy, H. Y. Kao, D. Charkaravarti, R. J. Lin, C. A. Hassig, D. E. Ayer, S. L. Schreinder, and R. M. Evans, Cell 89,373-380 (1997). 81. J. Xu, Y. Qiu, F. J. DeMayo, S. Y. Tsai, M.-J. Tsai, and B. W. O’Malley, Science 279, 19221925 (1998). 82. R. E. Weiss, J. M. Xu, G. Ning, J. Pohlenz, B. W. O’Malley, and S. Refetoff, EMBO]. 18, 1900-1904 (1999). 83. H. Chen, R. J. Lin, R. L. Schiltz, D. Chakravarti, A. Nash, L. Nagy, M. L. Privalsky, Y. Nakatani, and R. M. Evans, CeZZ90,569-580 (1997). 84. C. Rachez, B. D. Lemon, Z. Suldan, V. Bromleigh, M. Gamble, A. M. Naar, H. ErdjumentBromage, P. Tempst, and L. P. Freedman, Nature (London) 398,824-828 (1999). 85. S. Ryu, S. Zhou, A. G. Ladumer, and R. Tjian, Nature (London) 397,446-450 (1999). 86. M. Ito, C.-X. Yuan, S. Malik, W. Gu, J. D. Fondell, S. Yamamura, Z.-Y. Fu, X. Zhang, J. Qin, and R. G. Roeder, Mol. Cell 3,361-370 (1999). 87. J. Svaren and W. Horz, Cum @in. Genet. Dew. 3,219-225 (1993). 88. A. P. Wolffe, “Chromatin: Structure and Function,” 3rd Ed. Academic Press, London, 1998. 89. J. L. Workman and R. E. Kingston, Annu. Rev. Biochem. 67,545-579 (1998). 90. G. Almouzni, D. J. Clark, M. Mechali, andA. P. Wolffe, Nucleic Acids Res. 18,5767-5774 (1990).

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

97

91. B. P. Eliceiri and D. D. Brown,J. Biol. Chem. 269,24459-24465 (1998). and A. E. Mirsky, Proc. Nd Acud. Sci. U.S.A. 51, 786-794

92. V. Allfrey, R. M. Faulkner,

(1964). 93. B. G. T. Pogo, V. G. AIlfrey, and A. E. Mirsky, Proc. AJutZ.Acud. Sci. U.S.A. 55, 805-812 (1966). 94. A. P. Wolffe, Science 272, 371-372 (1996). 95. A. P. Wolffe and D. Pruss, Cell 84,817-819 (1996). 96. K. Struhl, Genes Da. l2,599-606 (1998). 97. L. Hong, G. P. Schroth, H. R. Matthews, P. Yau, and E. M. Bradbury,J. Biol. Chem. 268, 305-314 (1993). 98. V G Norton, B. S. Imai, P. Yau, and E. M. Bradbury, Cell 57,449-457 (1989). 99. Y. Lee and V. Mahdavi,J. BioZ. Chem. 268,2021-2028 (1993). 100. M. Vettese-Dadey, P. A. Grant, T. R. Hebbes, C. Grane-Robinson, C. D. Allis, and J. L. Workman, EMBOJ. 15,2508-2518 (1996). 101. J. Wong, D. Patterton, D. Imhof, D. Guschin, Y.-B. Shi, and A. P. Wolffe, EMBOJ. 17,520534 (1998). 102. R. T. U&y, K. Ikeda, P. A. Grant, J. Cote, D. J. Steger, A. Eberharter, S. John, and J. L. Workman, Nature (London) 394,498-502 (1998). 103. A. Imhof, X. J. Yang, V. V. Ogryzko, Y. Nakatani, A. P. Wolffe, and H. Ge, Curr BioZ. 7,689692 (1997). 104. W. Gu and R. G. Roeder, Cell 90,593-606 (1997). 105. M. Dauca and J. Hourdry, Transformations in the intestinal epithelium during anuran metamorphosis. In “Metamorphosis” (M. Balls and M. Bownes, eds.), pp. 36-58. Clarendon Press, Oxford, 1985. 106. K. Yoshizato, Int. Rev. Cytol. 119, 97-149 (1989). 107. S. Heusser, S. Colin, A. Figiel, C. Huet, J. M. Keller, P. Pornet, S. Robine, J. Vandamme, J. Vandekerckhove, and M. Dauca,J. Cell Sci. 103,699-708 (1992). 108. Y.-B. Shi and A. Ishizuya-Oka, Cum. Top. Dev. BioZ. 32,205-235 (1996). 109. J. A. Marshall and K. E. Dixon,J. Exp. ZooZ. 203,31-40 (1978). 110. M. Bjerknes and H. Cheng, Am. J. Anat. 160,105-115 (1981). 111. J. A. Marshall and K. E. Dixon,]. Anat. l26,133-144 (1978). 112. A. Ishizuya-Oka and A. Shimozawa, Anut. Anz. 164,81-93 (1987). (1963). 113. M. A. Bonneville,J. Cell BioZ. 18, 579-597 114. J. Hourdry and M. Dauca, Znt. Rev. Cytol. [Suppl.] 5,337-385 (1977). 115. A. Ishizuya-Oka and A. Shimozawa, Morphol. 213 185-195 (1992). 116. J. W. McAvoy and K. E. Dixon,J. Exp. ZooZ. 202,129-138 (1977). 117. H. Cheng and C. P. Leblond, Am.J. Anat. 141,503-520 (1974). 118. A. Ishizuya-Oka and A. Shimozawa, J. Morphol. 193,13-22 (1987). 119. M. N. Marsh and J. S. Trier, Gastroenterology 67,622-635 (1974). 120. L. Kordylewski, Z. Mikrosk. Anut. Ford. 97, 719-734 (1983). 121. S. Torihashi,]. Camp. Neural. 320,54-65 (1990). 122. A. Ishizuya-Oka and A. Sh’Imozawa, In Vitro Cell. Dev. BioZ. 27A, 853-857 (1991). 123. Y.-B. Shi and W. l? Hayes, Dev. BioZ. 16l,48-58 (1994). 124. A. Ishizuya-Oka, S. Ueda, S. Damjanovski, Q. Li, V.C.-T. Liang, and Y.-B. Shi, Dev. BioZ. 192,149-161 (1997). 125. A. Ishizuya-Oka and A. Shimozawa, Roux’s Arch. Dev. BioZ. 201,322-329 (1992). 126. Y. Su, Y. Shi, and Y.-B. Shi, FASEBJ. 11,559-565 (1997). 127. Y. Su, Y. Shi, M. Stolow, and Y-B. Shi, J. Cell BioZ. 139, 1533-1543 (1997). 128. Y. Yaoita and D. D. Brown, Gene Dev. 4,1917-1924 (1990). 129. A. Kawahara, B. S. Baker, and J. R. Tata, Development l&933-943 (1991).

98

YUN-BO

130. 131. 132. 133. 134. 135. 136.

M. J. Schneider and V A. GaIton, Mol. Endocrinol. 5,201-208 (1991). C. C. Helbing, G. Gergely, and B. G. Atkinson, Dev. Genet. 13,289-301 (1992). (1994). Y.-B. Shi, V.C.-T. Liang, C. Parkison, and S.-Y. Cheng, FEBS Lett. 355,61-64 L. Fairclough and J. R. Tata, Dev. Growth Diffm. 39,273-283 (1997). A. Kanamori and D. D. Brown,]. Biol. C&m. 267,739-745 (1992). Y.-B. Shi, Y. Yaoita, and D. D. Brown,J. Biol. Chem 267, 733-788 (1992). Y.-B. Shi, Thyroid hormone-regulated early and late genes during amphibian metamorphosis. In “Metamorphosis: Post-Embryonic Reprogramming of Gene Expression in Amphibian and Insect Cells” (L. I. Gilbert, J. R. Tata, and B. G. Atkinson, eds.), pp. 505-538. Academic Press, New York, 1996. J, Wong,V.C.-T. Liang, L. M. Sachs, andY.-B. Shi,J. Biol. Chem. 273,14186-14193 (1998). I. Machuca, G. Esslemont, L. Fairclough, and J. R. Tata, Mol. Endocrinol. 9,96-107 (1995). Y.-B. Shi and D. D. Brown,J. Biol. Chem. 268,20312-20317 (1993). Y.-B. Shi, J. Wong, M. Puzianowska-Kuznicka, and M. A. Stolow, BioEssays 18,391-399 (1996). Y. B. Shi and A. Ishizuya-Oka,J. Biomed. Sci. 4,9-18 (1997). T D. Sargent, Methods Enzymol. 152,423-432 (1987). R. J. Denver, S. Pavgi, and Y.-B. Shi,]. Biol. Chem. 272,8179-8188 (1997). (1998). T. Amano and K. Yoshizato, Wound Repair Regen. 6,302-313 A. Figiel, J.-M. Keller, J. Schilt, and M. Dauca, Differentiation, 40, 166-175 (1989). A. Ishizuya-Oka, A. Shimozawa, H. Takeda, and Y.-B. Shi, Dev. Biol. 204,150-155 (1994). C. M. AIexande and Z. Werb, Extracellular matrix degradation. In “Cell Biology of Extracellular Matrix,” 2nd Ed. (E. D. Hay, ed.), pp. 255-302. Plenum Press, New York, 1991. L. M. Matrisian, BioEssuys 14,455-463 (1992). H. Birkedal-Hansen, W. G. I. Moore, M. K. Bodden, L. J. Windsor, B. Birkedal-Hansen, (1993). A. DeCarlo, and J. A. Engler, Crit. Rev. Oral Biol. Med. 4, 197-250 W. C. Parks and R. P. Mecham, “Matrix Metalloproteinases.” Academic Press, New York, 1998. (1996). Q. A. Sang and D. A. Douglas,J. Protein Chem. 15,137-160 J. A. Uria and Z. Werb, Cell Res. 8,187-194 (1998). D. Pei and S. J. Weiss. Nature (London) 375 244-247 (1995). H. E. van Wart and H. Birkedal-Hansen, Proc. Natl. Acad. Sci. U.S.A. 87, 5578-5582 (1990). G. Murphy, R. Wihenbrock, T. Crabbe, M. O’Sh ea, R. Ward, S. Atkinson, J. O’Connell, and A. Docherty, Ann. N.Y. Acad. Sci. 732,31-41 (1994). D. E. KIeiner, Jr. and W. G. StetIer-Stevenson, Cum @in. Cell Biol. 5, 891-897 (1993). J. Nagase, K. Suzuki, T. Morodomi, J. J. Englhild, and G. Salvesen, Matrix Suppl. 1,237244 (1992). H. Nagase, Cell Res. 8,179-186 (1998). W. G. Stetler-Stevenson, S. Aznavoorian, and L. A. Liotta, Annu. Rev. Cell Biol. 9, 541573 (1993). K. Tiyggvason, M. Hoyhtya, and T. SaIo, Biochim. Biophys, Acta 907, 191-217 (1987). H. Sato and M. Se&i,]. Biochem. 119,209-215 (1996). L. M. Matrisian and B. L. M. Hogan, Curr. Top. Dev. Biol. 24,219-259 (1990). L. A. Salamonsen, Reprod. Med. Rev. 5,185-203 (1996). W. C. 0. Powell and L. M. Matrisian, Mechanisms by which matrix metalloproteinases may influence apoptosis. In “Programmed Cell Death” (Y. G. Shi, Y. Shi, Y. Xu, and D. W. Scott, eds.), pp. 27-34. Plenum Press, New York, 1997. Q. X. A. Sang, Cell Res. 8, 171-177 (1998). S. D. Shapiro, Curr. Opin. Cell Biol. 10,602-608 (1998).

137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149.

150. 151. 152.

153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163.

164.

165. 166.

SHI AND ATSUKO

ISHIZUYA-OKA

THYROID

HORMONE-REGULATED

TISSUE

REMODELING

99

167. T. H. Vu, M. J. Shipley, G. Bergers, J. E. Berger, J. A. Helms, D. Hanahan, S. D. Shapiro, R. M. Senior, and Z. Werb. Cell 93,411-422 (1998). 168. K. Holmbeck, P. Bianco, J. Caterina, S. Yamada, M. Kromer, S. A. Kuznetsov, M. Mankani, P. G. Robey, A. R. Poole, I. Pidoux, J. M. Ward, and H. Birkedal-Hansen. Cell 99, 81-92 (1999). 169. J. Gross,J. Invest. Dermatol. 47,274-277 (1966). 170. E. D. Hay, “Cell Biology of Extracellular Matrix,” 2nd Ed. Plenum Press, New York, 1991. 171. R. Timpl and J. C. Brown, BioEssays l&123-132 (1996). 172. E. Murata and H. J. Merker, Actu Anat. 140,60 - 69 (199 1). (1994). 173. K. Oofusa, S. Yomori, and K. Yoshizato, ht. J. Dev BioZ. 38,345-350 174. K. Oofusa and K. Yoshizato, Roux’s Arch. Dev. Biol. 205,241-251 (1996). 175. D. Patterton, W. P. Hayes, and Y.-B. Shi, Dev. Biol. 167,252-262 (1995). 176. Z. Wang and D. D. Brown,J. BioZ. Chem. 268,16270-16278 (1993). 177. M. A. Stolow, D. D. Bauzon, J. Li, T. Segwick, V.C.-T. Liang, Q. A. Sang, and Y.-B. Shi, Mol. BioZ. Gel2 7, 1471-1483 (1996). 178. P. Basset, J. P. Bellocq, C. Wolf, I. Stoll, P. Hutin, J. M. Limacher, 0. L. Podhajcer, M. P. Chenard, M. C. Rio, and P. Chambon, Nature &on&n) 348,699-704 (1990). 179. D. Muller, C. Wolf, J. Abecassis, R. Millon, A. Engelmann, G. Bronner, N. Rouyer, M. C. Rio, M. Eber, G. Methlin, P. Chambon, and P. Basset, Cancer Res. 53, 165-169 (1993). 180. 0. Lefebvre, C. Wolf, J.-M. Limacher, P. Hutin, C. Wendling, M. LeMeur, P. Basset, and M.-C. Rio,J. Cell. BioZ. 119, 997-1002 (1992). 181. 0. Lefebvre, C. Regnier, M.-P. Chenard, C. Wendling, P. Chambon, P. Basset, and M.-C. Rio, Develqment 121,947-955 (1995). 182. A. Ishizuya-Oka, S. Ueda, and Y.-B. Shi, Cell Tissue Res. 283 325-329 (1996). 183. S. Damjanovski, A. Ishizuya-Oka, and Y.-B. Shi, Cell Res. 9, 91-105 (1999). 184. D. L. Berry, R. A. Schwartzman, and D. D. Brown, Deu. Biol. 203,12-23 (1998). 185. D. L. Berry, C. S. Rose, B. E Remo, and D. D. Brown, Deo. Biol. 203,24-35 (1998b). 18%. C. M. Alexander, E. W. Howard, M. J. Bissell, and Z. Werb,J. CeZZBioZ. 135,1669-1677 (1996). 18% J. P. Witty, J. H. Wright, and L. M. Matrisian, Mol. BioZ. Cell 6, 1287-1303 (1995). 186. J. P. Witty, T Lempka, R. J. Coffey, J r., and L. M. Matrisian, Cancer Res. 55, 1401-1406 (1995). 187. C. J. Sympson, R. S. TaIhouk, C. M. Alexander, J. R. Chin, S. M. Clift, M. J. Bissell, and Z. Werb,J. Cell BioZ. 125, 681-693 (1994). 188. A. C. Noel, 0. Lefebvre, E. Maquoi, L. VanHoorde, M. P. Chenard, M. Mareel, J.-M. Foidart, P. Basset, and M.-C. Rio,]. CZin. Znuest. 97, 1924-2930 (1997). 189. R. Masson, 0. Lefebvre, A. Noel, M. El Fahime, M.-P. Chenard, C. Wendling, F. Kebers, M. LeMeur, A. Dierich, J.-M. Foidart, P. Basset, and M.-C. Rio, J. Cell BioZ. 140, 15351541 (1998). 190. D. Pei, G. Majmudar, and S. J. Weiss,J. Biol. Chem. 269,25849-25855 (1994). 191. L. R. Lund, J. Romer, N. Thomasset, H. Solberg, C. Pyke, M. J. Bissell, K. Dono, and Z. Werb, Development 122,181-193 (1996). 192. R. S. Talhouk, M. J. Bissell, and Z. Werb,J. Cell Biol. 118, 1271-1282 (1992). 193. N. Boudreau, C. J. Sympson, Z. Werb, and M. J. Bissell, Science 267,891-893 (1995). 194. C. D. Roskelley, A. Srebrow, and M. J. Bissell, Curr Biol. 7, 736-747 (1995). 195. E. Ruoslahti and J. C. Reed, Cell 77,477-478 (1994). 196. Y.-B. Shi, Y. Shi, Y. Xu, and D. W. Scott, “Programmed Cell Death.” Plenum Press, New York, 1997. 197. S. M. Frisch and E. Ruoslahti, Cum @in. Cell BioZ. 9,701-706 (1996). 198. J. E. Meredith, Jr. and M. A. Schwartz, Trends CeZZBioZ. 7, 146-150 (1997).

100

YUN-BO

SHI AND ATSUKO

ISHIZUYA-OKA

199. Y.-B. Shi, Q. Li, S. Damjanovski, T. Ammo and A. Ishizuya-Oka, lnt. J. Mol. Med. 2,273282 (1998). 200. C. H. Damsky and Z. Werb, Curt-. Biol. 4,772-781 (1992). 201. K. E. Brown and K. M. Yamada, Dev. Biol. 6,69-77 (1995). 202. J. W. Schmidt, P. A. Piepenhagen, and W. J. Nelson, Semin. Cell Biol. 4, 161-173 (1993). 203. S. Stromblad, P. C. Brooks, J. Becker, M. Rosenfeld, and D. A. Cheresch, The role of integrin cxvB3 in ceII survival and angiogenesis. In “Programmed Cell Death” (Y.-B. Shi, Y. Shi, Y. Xu, and D. W. Scott, eds.), pp. 35-42. Plenum Press, New York, 1997. 204. M. A. Stolow and Y.-B. Shi, Nucleic Acids Res 23,2555-2562 (1995). 205. D. D. Schlaepfer and T. Hunter, Trends Cell Biol. 8,151-157 (1998). (1994). 206. B. G. Atkinson, Dev. Genet. 15,313-319 207. M. Puzianowska-Kuznicka, S. Damjanovski, and Y.-B. Shi, Mol. Cell Biol. 17,4738-4749 (1997). 208. K. L. KroII and E. Amaya, Development l22,3173-3183 (1996). 209. H. Huang, N. Marsh-Armstrong, and D. D. Brown, Proc. Natl. Acad. Sci. U.S.A. 96,962967 (1999). 2IO. M. Puzianowska-Kuznicka and Y.-B. Shi,J. Biol. Chem. 271, 6273-6282 (1996). 211. A. Ishizuya-Oka, S. Ueda, and Y.-B. Shi, Dev. Genet. 20,329-337 (1997). 212. D. D. Brown, Z. Wang, J. D. Furlow, A. Kanamori, R. A. Schwartzman, B. F. Rmo, and A. Pinder, Proc. Natl. Acad. Sci. U.S.A. 93,1924-1929 (1996). 213. D. Patter-ton and Y.-B. Shi,J. Biol. Chem. 269,25328-25334 (1994). 214. A. Ishizuya-Oka, M. A. Stolow, S. Ueda, and Y.-B. Shi, Dev. Genet. 20,53-66 (1997). 215. V.C.-T. Liang, T. Sedgwick, and Y.-B. Shi, Cell Res. 7, 179-193 (1997). 216. D. Torrents, R. Estevez, M. Pineda, E. Femandez, J. Lloberas, Y.-B. Shi, A. Zorzano, and M. Palacin, J. Biol. Chem. 273,32437-32445 (1998). 217. T. Amano, N. Nero, H. Kawabata, Y. Kobayashi, and K. Yoshizato, Dev. Growth Dif_i 40, 177-188 (1998).

Role of S6 Phosphorylation and S6 Kinase in Cell Growth SINISA VOLAREVIC GEORGE

AND

THOMASI

Fried&h Miescher Institute CH-4058 Basel, Switzerland I. 40s Ribosomal Protein S6 ........................ A. S6 Phosphorylation ........................... B. Location .................................... C. Function .................................... D. Drosophila S6 and Extraribosomal Functions of S6 II. S6Kinase ...................................... A. IdentificationofSGKl......................... B. Structure of S6Kl ............................ C. Mechanisms of S6Kl Activation ................ D. Upstream Effecters ........................... III. Downstream Effecters of S6K ..................... A. S6 Phosphorylation and 5’TOP mRNA Translation B. S6Kl and Cell Cycle Progression ............... C. InsulinProduction ............................ IV. Physiological Importance ......................... A. S6Kl Deletion in Mice and Discovery of S6K2 .... B. Drosophila S6K Mutant ....................... . Future Perspectives .............................. References .....................................

This article reviews our current knowledge S6 phosphorylation

of cell growth and proliferation.

...

.

of the role of ribosomal

protein

Although 40s ribosomal protein S6 phosphory-

25 years ago, it only recently has been implicated in the

up-regulation

thetic apparatus.

103 103 10s 105 107 108 108 108 111 113 117 117 118 119 119 119 120 122 123

and the S6 kinase (S6K) signaling pathway in the regulation

lation was first described translational

. .

of mRNAs coding for the components

These mRNAs contain an oligopyrimidine

scriptional start site, termed

of protein syn-

tract at their 5’ tran-

a 5’TOP, which has been shown to be essential for

their regulation at the translational level. In parallel, a great deal of information has accumulated regulatory

concerning

phosphorylation

this knowledge

the identification

of the signaling pathway and the

sites involved in controlling

we are only beginning

involved in growth factor-induced

S6K activation. Despite

to identify the direct upstream

elements

kinase activation. Use of the immunosupres-

1 To whom correspondence should be addressed. Progress in Nucleic Acid Research and Molecular Biology, Vol. 65

101

Copyright 8 2001 by Academic P~PII All rights of reproduction in any form reserved. 0079-6603/01 $35.00

102

SINISA

sant rapamycin,

a bacterial

macrolide,

VOLAREVIC

in conjunction

AND GEORGE

THOMAS

with dominant interfering

and activated forms of S6Kl has helped to establish the role of this signaling cascade in the regulation employing

of growth and proliferation.

the mouse as well as Drosophila

sights into physiological

paramount

S6K2, whereas importance

have provided

function of S6K in the animal. Deletion

in mouse cells led to an animal of reduced homolog,

In addition, current studies

melanogaster

loss of dS6K

in development

size and the identification

function in Drosophila and growth control.

new in-

of the S6Kl gene of the S6KI

demonstrated

its

o 2000AcademicPWS.

Mitogens stimulate cells to exit the G, stage of the cell cycle, progress through the G, phase of the cell cycle, synthesize DNA during S phase, and ultimately pass through mitosis and cell division (1). Earlier studies in yeast indicated that passage through G, requires a coordinated increase in protein synthesis and cell mass. Hartwell and colleagues in their seminal studies demonstrated that conditions that prevented cell growth, such as nutrient deprivation, arrested cell proliferation (2). In contrast, mutations that blocked cell cycle progression did not affect cell growth (3). These findings demonstrated the dominance of cell growth over cell proliferation. However, the molecular mechanisms by which the processes controlling cell growth and proliferation are integrated to bring about the development of the organism are largely unknown. Although the signaling pathways that control cell cycle progression have been studied intensively (I), we know less concerning the signaling pathways that control cell growth, such as those associated with the regulation of protein synthesis. For some time, it has been evident that for the cell to grow it must synthesize new proteins and that this process relies on the activation and maintenance of high rates of protein synthesis to meet this demand (4). However, less recognized is the fact that the greatest demand is for those proteins that make up the translational apparatus, especially ribosomes (5- 7). The highly regulated relationship between ribosome biosynthesis and cell growth was first demonstrated in bacteria over 40 years ago (8). In this way, biosynthetic energy is conserved when cells are not growing, whereas the rate of protein synthesis increases as a function of physiological or pathological growth state, such as cancer (9). The control of ribosome biogenesis is closely monitored, as the energy investment in production of these large multiprotein-rRNA complexes has been calculated to be as much as 80% of the expended energy in a proliferating mammalian cell. The energy investment in ribosome biogenesis is further underscored by the approximate 5 million ribosomes present per mammalian cell, with each ribosome having an overall molecular mass of 4.2 X lo6 Da. These values mean that ribosomes make up greater than 80% of the cellular RNA and 5-100/o of cellular proteins (IO).

s6

PHOSPHORYLATION

AND

s6

KINASE

IN CELL

GROWTH

103

Although, a number of phosphorylation events have been implicated in the regulation of global translation, mostly at the level of initiation of protein synthesis (11, 12), only recently was it demonstrated that the S6K is involved in the translational up-regulation of a specific family of mRNAs that contain an oligopyrimidine tract at their 5’ transcriptional start site (5’TOP mRNAs). This step is presumably regulated by the multiple phosphorylation of 40s ribosomal protein S6 (13). H ere, we will attempt to review the current status of the role of S6 phosphorylation in the regulation of 5’TOP mRNA translation, the molecular mechanisms of S6K activation, and the characterization of the upstream components in this signaling cascade. Recent insight into the physiological relevance of this pathway obtained from the manipulation of S6K in both mouse and Drosophila models also will be addressed.

I. 40s Ribosomal Protein S6 A. S6 Phosphorylation The S6 gene encodes for one of the 33 proteins that, in a complex with one molecule of 18s rRNA, comprise the mature 40s ribosomal subunit (14). Phosphorylation of the 40s ribosomal protein subunit was first observed in the livers of rats injected with 32P0, (15). Gressner and Wool (16) subsequently demonstrated that the only significantly phosphorylated 40s ribosomal protein, following partial hepatectomy in rats, was S6. Because partial hepatectomy induces the remaining hepatocytes to grow and proliferate, this finding led to a number of studies that demonstrated that growth factor stimulation of different cell types also induced hyperphosphorylation of S6 (1719). In addition to the induction of S6 phosphorylation following partial hepatectomy in rats, this response was also induced in vivo during refeeding after fasting, differentiation, fertilization, and infection with viruses (19-22). In general it was also observed that increased S6 phosphorylation correlated with increased rates of protein synthesis (23). Despite this correlation between protein synthesis and S6 phosphorylation, it was initially difficult to rationalize this response with the observation that protein synthesis inhibitors, such as cycloheximide and puromycin, also were found to elicit a potent S6 phosphorylation response (16,23,24). Nevertheless, two main observations prompted further research into the role of S6 phosphorylation in the regulation of the protein synthesis. First, the kinetics of S6 phosphorylation closely parallel changes in translational activity (23), and second, the observation that ribosomes with the highest proportion of phosphorylated S6 are selectively found in large polysomes (25, 26). To identify the sites of S6 phosphorylation, advantage was taken of the fact that

104

SINISA VOLAREVIC

AND GEORGE

Species:

Seauence:

Homo sapiens

RRRLSSLRASTSKSESSCIK’~~

Mus musculus

RRRLSSLRASTSKSESSQK’@

Ratus norvegicus

RRRLSSLRASTSKSESSQK’@

Xenopus laevis

RRRLSSLRASTSKSESSQK“@

Drosophila

RRRSASIRESKSSVSSDKK’”

melanogaster

Drosophila alternative isoform

RGRY’JTIRKPKSSVFSGKK’~~

Saccharomyces

KRRA.SSLKA’=

cerevisiae

THOMAS

FIG. 1. Comparison of the carboxy-terminal domain of ribosomal protein S6 from different species. The sequence alignment begins at the S6K RXFt recognition motif. Phosphorylation sites are shown in bold.

intraperitoneal injection of cycloheximide had been shown to induce S6 phosphorylation in all tissues examined (23). In this way it was found that cycloheximide induces the phosphorylation of S6 on five car-boxy-terminal serine residues, residing within a 32amino acid carboxyl-terminal fragment (24). The five phosphorylation sites were shown to be Ser-235, Ser 236, Ser240, Ser-244, and Ser-247 (Fig. 1) (24). In later studies it was shown that serum stimulation of mouse fibroblasts induced S6 phosphorylation on the same sites as had been shown in rats treated with cyclohexamide (27). Earlier studies had shown that phosphorylation of S6 appeared to proceed in an ordered fashion (28). Subsequent in vivo and in vitro studies showed that phosphorylation proceeds in an ordered manner: Ser-236 > Ser-235 > Ser240 > Ser-244 > Ser-247 (27, 29). With the identification of the S6 kinase (see below) it was possible to demonstrate that its recognition sequence resided immediately upstream of the principal site of phosphorylation, Ser236, in the sequence FtXRXXS, with arginines present in the -5 and -3 position (30). The mouse, rat, and human S6 protein sequences are identical to each other, whereas the identity of the Drosophila and yeast S6 protein with the mammalian sequence is 74 and 60%, respectively (31-34). The yeast homolog of mammalian S6, termed SlO, is present in two copies in the yeast genome, as are most ribosomal proteins. Compared to mammalian S6, SlO lacks the last 10 carboxy-terminal amino acids. Therefore SlO contains only two of the five potential mammalian phosphorylation sites, corresponding to S235 and S236 (Fig. 1). A yeast strain having a single copy of the S6 gene is viable, therefore Johnson and Warner (35) replaced this allele with a mutant S6 gene having the two putative phosphorylation sites changed to alanines.

s6

PHOSPHORYLATION

AND

s6

KINASE

IN CELL

GROWTH

105

Extensive analysis of this strain revealed no obvious alteration in the growth phenotype (35). Neverth e 1ess, it is important to emphasize that yeast do not regulate ribosomal protein expression at the translational level and that yeast ribosomal protein mRNAs do not contain 5’TOP sequences (36, 37). Furthermore, it appears to be the later sites of S6 phosphorylation that correlate with increased translation, whereas the early sites, S235 and S236, appear to regulate phosphorylation at the downstream sites (13).

B. Location Chemical cross-linking and protection studies with specific components of the translational apparatus have localized S6 to the small head region of the 40s subunit at the mRNA-tRNA binding site. This area of the 40s subunit resides at the interface with the larger 60s ribosomal subunit, where S6 apparently comes into direct contact with 28s rRNA (11,38). The location of S6 in the 40s subunit places it in a position where it could potentially have a functional impact on translation. This led to the hypothesis that S6 in the phosphorylated state, through steric effects or conformational changes (11, 19,38), may be involved in regulating specific steps in translation. Other studies have led to the hypothesis that differentially phosphorylated S6 could directly interact with a subset of mRNAs or associated proteins and lead to selective changes in the pattern of translation (see below).

C. Function Early studies correlated increased S6 phosphorylation with the up-regulation of global translation. However, in some instances the kinetics of these two events did not coincide. For example, in HeLa cells, S6 phosphorylation was shown to decline 6 hr after serum stimulation, whereas maximum levels of protein synthesis were achieved at later time points (23). An important observation that led to the suggestion that S6 phosphorylation may be involved in selective translational up-regulation of specific mRNAs was the finding that the levels of at least 16 proteins coincided with increased S6 phosphorylation in serum-stimulated Swiss mouse 3T3 cells (39). The increased synthesis of seven of these proteins was not blocked by actinomycin D, a potent inhibitor of transcription, suggesting that the up-regulation of the seven unaffected proteins was controlled at the posttranscriptional level. The protein with the most increased value was subsequently identified as eukaryotic elongation factor-lo, eEF-lo (40). Consistent with the actinomycin D experiments described above, the absolute levels of eEF-lo mRNA do not change within the first 3 hr of serum stimulation (41). In quiescent cells eEF-lo mRNA is translated inefficiently, with the bulk of the mRNA largely residing in mRNP particles, with a small amount of eEF-la transcript present on either disomes or monosomes (41).

106

SINIL?A VOLAREVIC

AND

GEORGE

THOMAS

Following serum stimulation, both populations of eEF-la mRNA are selectively recruited to very large polysomes, of approximately 11- 12 ribosomes per transcript (41). Under these conditions the mean polysome size does not change and most mRNAs remain in the same position within the polysome profile. Thus, the selective changes in eEF-la mRNA usage correlated with the increased phosphorylation of S6. To obtain better insight into the relationship between selective translational up-regulation of eEF-lo and S6 phosphorylation, advantage was taken of rapamycin, a bacterial macrolide known to inhibit S6 phosphorylation due to its ability to inhibit S6K (42; and see below). Rapamycin selectively repressed recruitment of eEF-lo into large polysomes, without affecting global translation. Interestingly, eEF-lo belongs to a family of mRNAs that contain at their transcriptional start site S’TOP, an element known to act as a translational regulator (43). All mammalian ribosomal protein mRNAs in which the 5’ transcription start site has been mapped, as well as elongation factors, contain this motif (37). Th e motif starts with a cytidine and is followed by a stretch of 5 to 15 pyrimidine residues. However, for efficient translational up-regulation of S’TOP mRNAs, integrity of the region downstream of the S’TOP is also required (37,44). Although the S’TOP family of mRNAs contains only 100 to 200 genes, their transcripts can represent up to up 20% of total cellular RNAs (37, 44). More recent studies employing chimeric mRNAs, in which either the wild-type S’TOP tract or a disrupted S’TOP tract had been fused to a reporter transcript, demonstrated that an intact S’TOP tract is required for rapamycin to elicit an inhibitory effect on the translation of these mRNAs (45). Consistent with these findings, a dominant interfering S6Kl mutant repressed the mitogen-induced translational up-regulation of S’TOP mRNAs to the same extent as rapamycin, whereas expression of a rapamycin-resistant S6Kl mutant negates the inhibitory effects of rapamycin on S’TOP mRNA translation (45). These same constructs either block or promote S6 phosphorylation, respectively (46). Taken together, these results demonstrate that the rapamycin inhibitory block of S’TOP mRNA translation is mediated through S6Kl inactivation and strongly suggest that S6 phosphorylation mediates this response. Direct evidence for a role for S6 phosphorylation in the regulation of 5’TOP mRNAs is lacking. If phosphorylated S6 is functioning to regulate translation of S’TOP mRNAs, it will be of interest to determine the underlying molecular mechanism. It is possible that phosphorylated S6 directly interacts with the S’TOP, that it induces a conformational change in the 40s subunit that favors S’TOP mRNA translation, or that it mediates this interaction through other factors. In support of the latter possibility it has been found that the S’TOP region interacts in vitro with the La protein (47). In addition the cellular nucleic binding protein (CNBP) interacts with the region

s6

PHOSPHORYLATION

AND

s6

KINASE

IN CELL

107

GROWTH

downstream of the polypyrimidine tract, a sequence required for efficient 5’TOP mRNA translation (48). The Ro60 autoantigen mediates the binding of the two proteins to their respective sequences, with the binding of either of the two proteins excluding the other (49). The fact that La and CNBP interact with two sequences required for translational up-regulation of 5’TOP mRNAs suggests that they may exert a modulating activity on the translational control of this family of mRNAs. Although it is unclear how these proteins may affect 5’TOP translation, it is tempting to speculate that they modulate interaction between S6 and the 5’TOP sequence.

D. Drosophilu

S6 and Extraribosomal

Functions

of $5

Drosophila S6 (DS6) is a single-copy gene. The genomic sequence of the DS6 gene has revealed that it is made up of three exons; however, for exon 3A, two alternative tandem repeats exist, 3B and 3C (50). If utilized, as has been claimed (32) exons 3B and 3C would produce a transcript that encodes a 189-amino acid protein that is considerably shorter than the 248-amino acid protein produced by exon 3A. In addition, the shorter protein would have a distinct carboxy terminus. Preliminary experiments indicate that only the longer form of DS6 is found in Drosophila 40s ribosomes and that it is also phosphorylated at five sites (51). This form of DS6 is almost equivalent in length to its mammalian ortholog, sharing 74% identity and 84% similarity with the mammalian protein (51). The correlation between increased DS6 phosphorylation and protein synthesis in Drosophila has not yet been examined carefully. It is clear that the translation of ribosomal proteins in Drosophila, as in mammals, is regulated at the translational level (52). However, the regulation of mRNAs coding for components of the translation apparatus may be different from that of mammals because not all ribosomal protein mRNAs appear to contain 5’TOPs (37). It should also be noted that P-element insertions in the 5’ untranslated (5’UTR) of the DS6 gene, which is X linked, leads to a decrease in DS6 gene expression, with hemizygous mutant males failing to emerge as adults after a prolonged third larval instar (31, 32). During larval development the mutants develop melanotic tumors in the hematopoietic system, with significantly overgrown lymph glands, implying an additional role for DS6 in tumor suppression of larval hemocytes, which mediate the immune response in insects (31, 32). Th e increased number of hemocytes and the presence of giant cells within the hematopoietic organs are not completely consistent with a role for DS6 in cell growth. However, the expression level of DS6 in these tumor cells has not been determined, nor whether DS6 phosphorylation is involved. Obviously the powerful genetics offered by the Drosophila system should prove advantageous in elucidating the role of S6 phosphorylation in translation. An additional indication that S6 may have an extraribosomal function

108

SINISA VOLAREVIC:

AND GEORGE

THOMAS

comes from the finding of a free form of S6 in the nucleus of mammalian cells (53). This pool of S6 is also phosphorylated in response to growth factor stimulation, presumably by a nuclear form of S6Kl (54). However, there is no direct evidence for an extraribosomal function of S6 in mammals. It is interesting to note that the extraribosomal biochemical functions have been ascribed to a number of other ribosomal proteins (55). However, the biological significance of these biochemical functions has been difficult to demonstrate in vivo, although a patient with a mutation in S19 has been described. The mutation is genetically linked to Diamond-Blackfan anemia (56), a clinical feature that suggests an extraribosomal function.

II. S6 Kinase A.

Identification

of S6Kl

Although a number of kinases were initially implicated in S6 phosphorylation, based on their in vitro activities (57,58), it generally has been accepted that the S6K family of kinases mediates this response. However, studies in Xenopus oocytes suggest that in this system p90mk may function as an in vivo S6 kinase (59). S6Kl was first purified from extracts of mitogen-stimulated cells utilizing 40s ribosomes as substrate (60). Subsequent sequencing of the enzyme, followed by cloning and expression studies, led to the identification of two isoforms of S6Kl that are produced from the same transcript (61- 64) by alternative initiation translational start sites (Y. Chen, C. D. Hoemann, G. Thomas, and S. C. Kozma, unpublished data). The shorter form of S6K1, which is largely localized in the cytoplasm, was termed ~70~~~; a longer form, which exclusively localizes in nucleus, is termed ~85 S6K (Fig. 2). The larger isoform of S6Kl is distinct from the shorter form in that it contains an amino-terminal 23-amino acid extension, containing a nuclear localization signal (54). The functional significance of the differential subcellular localization of the two isoforms of S6Kl has not been established, although it is tempting to speculate that the nuclear form is involved in the phosphorylation of the nuclear pool of free S6 described above (53). It was found that deletion of the S6Kl gene in mice impaired neither S6 phosphorylation nor 5’TOP mRNA translation (65). This observation led to the identification of a second S6 kinase (S6K2) gene that is highly homologous to the first gene (65- 67) (Fig. 2).

B. Structure

of S6Kl

The mechanism by which kinase activation is presumed to take place has been acceded to by the dissection of the primary structure of S6Kl into do-

525

FIG. 2. Schematic representation of S6Kl and S6K2. Two forms of S6K1, p70 and ~85, are shown. The positions of S6K domains are indicated. The phosphorylation sites are numbered. Nuclear localization signals in ~85 and S6K2 are depicted. A proline-rich sequence, representing a putative SH3 recognition motif, is indicated in S6K2.

S6K2

356

110

SINISA VOLAREVIC

Kinase:

Seauence:

S6Kl

lVTH ~ TzaFCGT IEYMWE ~-

PKBa

ATM&“FCGTPEyL&&

PKCa

VTRRT@‘FCGTPDY_lE

CaMKlV

VLMKT’=VCGlPGYCAPE

AMPK

EFLRT’ =SCGSPNYAAPE _____

AND GEORGE

THOMAS

AGC family

CaMK family

FIG. 3. Comparison of the sequence surrounding the activation loop T229 site of S6Kl with those of related kinases. The S6Kl T229 phosphorylation site and equivalent sites in other kinases are shown in bold. The identical amino acids in S6Kl and other kinases are underlined.

mains and to the identification of specific regulatory phosphorylation sites (68). Such an understanding was instrumental in the identification of one of the upstream S6Kl kinases as the phosphoinositide-dependent protein kinase, PDKl (69, 70). P rior studies had revealed five distinct domains that cooperate to bring about S6 kinase activation (Fig. 2). The first is an acidic domain that extends from the amino-terminus to the beginning of the catalytic domain and confers rapamycin sensitivity on the kinase (71- 73). The second is the catalytic domain, which contains the mitogen-induced phosphorylation site T229, residing within the activation loop (74, 75). The catalytic domain as well as the sequence immediately amino terminal to the T229 phosphorylation site are highly conserved in the protein A, protein G, and protein C (AGC) family of protein kinases as well as the CaMK-dependent family (Fig. 3) (76). The third domain is the linker domain, which connects the catalytic domain to the autoinhibitory domain. The linker domain is also highly conserved in the AGC family of serine threonine kinases and contains two phosphorylation sites, S371 and T389, which are essential for S6Kl activation (77, 78). The motif surrounding T389 is flanked by hydrophobic amino acids in the + 1 and - 1 positions and is also conserved in a large number of kinases that belong to this family (Fig. 4) (77). In contrast to most members of the AGC family, which end with the linker domain, S6Kl has two additional carboxy-terminal domains. The first contains a motif that is homologous to the amino acid motif surrounding phosphorylation sites in S6, and was initially suggested to serve as an autoinhibitory domain (61, 79). I n support of this model synthetic peptides covering this sequence inhibit S6 kinase activity in vitro (30) and removal of the carboxy terminus raises basal kinase activity (73). This domain contains five

s6

PHOSPHORYLATION

AND

s6

KINASE IN CELL

GROWTH

Kinase:

Sequence:

S6Kl

VELGFT=YVAP

PKCa

DEEGFS=‘YVNP

RSKl

LERGF?FVAT

PKBcx

HEPQFSAS

111

FE. 4. Conservation of sequence motif surrounding S6Kl T389 in related kinases. Alignment of S6Kl sequence with those of PKCq RSKl, and PKBa. The conserved residues are underlined and the residues homologous to T389 are shown in bold.

sites: S404, S411, S418, T421, and S424. It is interesting to note that S404 is surrounded by large bulky hydrophobic amino acids in the - 1 and + 1 positions, similar to T229 and T389, whereas the remaining phsophorylation sites are followed by proline at the +1 position, similar to S371. Finally, the extreme carboxy terminus of S6Kl contains a sequence that has been shown to interact with the PDZ domain of neurabin, a neuralspecific F actin binding protein (80, 81). Presumably, S6Kl is localized to nerve terminals through this interaction. However, the physiological meaning of this interaction remains unclear at this point. Structural differences between S6Kl and S6K2 are discussed below.

phosphorylation

C. Mechanisms

of S6Kl

Activation

The mechanism of S6Kl activation is complex, involving the interplay between four different domains and at least seven specific regulatory phosphorylation sites, implying the existence of multiple upstream regulators (68). Current models suggest that the first step in S6Kl activation is the phosphorylation of the ST-P sites in the autoinhibitory domain, which functions in conjunction with the amino terminus to allow phosphorylation of T389. These two events presumably disrupt the interaction between the carboxy and amino termini, allowing phosphorylation of T229 and resulting in S6Kl activation (Fig. 5). This model is supported by the observation that T229 is phosphorylated in a mitogen-independent fashion when both T389 and the S/T-P sites in the autoinhibitory domain are replaced with acidic residues, generating the S6Kl variant T389EDaE (82). The kinase activity responsible for mediating T229 phosphorylation appeared to be constitutive, initially suggested that PDKl may serve as the in vivo T229 kinase. In support of this hypothesis the protein sequence motif surrounding T229 is homologous to the corresponding PDKl-dependent phosphorylation site in protein kinase B (PKB), T308 (83, 84). That PDKl is

112

SINISA

Mitogen

VOLAREVIti

AND GEORGE

THOMAS

Mitogen 1 T389 Kinase

L

T389-

s371-P

/

inactlve

I

T229-P

-

active

FIG. 5. Mechanism of S6Kl activation. The S6Kl domains and phosphorylation denoted as in Fig. 1. The model represents sequential steps of S6Kl phosphorylation vation as outlined in the text (68).

sites are and acti-

the physiological T229 S6Kl kinase was based on the fact that PDKl could activate the S6Kl acidic variant in vitro and in vivo, whereas the catalytically inactive PDKl blocked insulin-induced activation of S6Kl in vivo (69). Although the activity of the S6Kl acidic variant is up-regulated following phosphorylation by PDKl, the activity of S6Kl can be increased a further twofold by mitogens (72). The above results suggested the existence of additional sites of phosphorylation involved in S6Kl activation. One such candidate is S3 71 in the linker domain (78). The equivalent site has been found to be a major site of autophosphorylation in members of the PKC family (85-87). However, a transiently expressed kinase dead variant of S6Kl is phosphorylated at S3 71 in the presence of rapamycin (78). This result would strongly imply that S6Kl S3 71 phosphorylation is not regulated by autophosphorylation in either a cis or trans manner (78). The importance of this site for kinase activity is derived

s6 PHOSPHORYLATIONAND s6 KINASEIN CELL GROWTH

113

from the fact that substitution of S371 with either an alanine or aspartic acid blocks kinase activation. In parallel, these substitutions also block mitogeninduced T389 phosphorylation, an effect that cannot be rescued by substituting an acidic residue for T389. These findings would imply that S371 phosphorylation has a dual role-to regulate T389 phosphorylation as well as to contribute to kinase activity (78). It will be of interest to determine whether S3 7 1 is the last link required to bring about kinase activation. In the final analysis it will be necessary to reconstitute S6Kl activation in o&-o. This will require the identification of the kinases that phosphorylate the remaining sites, and complimentary in ciao and genetic studies to confirm their identities.

D. Upstream

Effecters

1. PI3K Because the S/T-P sites in the autoinhibitory domain have motifs similar to the consensus sequence for the mitogen-activated protein kinase (MAPK), and because MAPK can phosphorylate S6Kl in vitro, it was initially speculated that the RASMAPK pathway controlled S6Kl activation (88). However, Ballou et al. (89) d emonstrated that insulin treatment of Swiss 3T3 cells had no effect on MAPK activation, but potently activated S6Kl. Further experiments utilizing both dominant interfering mutants of BASMAPK signaling pathway as well as specific docking site mutants of the platelet-derived growth factor (PDGF) receptor demonstrated that MAPK is neither necessary nor sufficient to elicit S6Kl activation (90). As importantly, such studies demonstrated that S6Kl was a downstream effector of the phosphatidylinositide-30H kinase (PI3K) signal transduction pathway, rather than the RASMAPK signal transduction pathway (90, 91). Support for this conclusion came from studies employing PI3K inhibitors, such as wortmannin and LY294002, which blocked the activation of S6Kl (91), as well as membranetargeted, constitutively active forms of PI3K that activate S6Kl. Nevertheless, the molecular events leading from PISK activation to that of S6Kl have yet to be elucidated. Indeed, not all the studies employing PDGF receptor mutants are wholly consistent with PI3K mediating S6Kl activation (90). Furthermore, studies have challenged the specificity of wortmannin and LY294002 for PI3Ks (92), and membrane-targeted, constitutively active al leles PI3K may not accurately reflect wild-type PI3K function (see below). 2. EFFECTOR KINASES OF PISK Protein kinase B (PKB) has been established as a downstream effector of PI3K (93, 94). Furth ermore, constitutively activated membrane-targeted alleles of PKB induce reporter S6Kl activation in a wortmannin-resistant manner (94). However, subsequent studies showed that PKB does not directly

114

SINI??iAVOLAREVIdANDGEORGETHOMAS

phosphorylate S6Kl. In addition, Corms et al. (1998) have demonstrated that the depletion of intracellular stores of Ca 2+ has no effect on PKB activation, but abolishes S6Kl activation (95). In contrast, they observed that an increase in intracellular Ca2+ induces S6Kl activation in a wortmannin-sensitive manner without stimulating PKB activation. Consistent with these findings recent experiments have shown that dominant interfering alleles of PKB, which block insulin-induced reporter PKB activation, glycogen synthase 3kinase (GSK3) inactivation, and initiation factor 4E binding protein (4EBPl) phosphorylation, have no effect on S6Kl activation (96). Additionally, only activated alleles of PKB that are membrane targeted induce S6Kl activation, whereas the constitutively active cytoplasmic form of PKB, harboring acidic residues at T308D and S473D (97), though sufficient to induce GSK3 and 4E-BP1 phosphorylation, have no effect on S6Kl activation (96). These results indicate that membrane-targeted PKB activation does not reflect wildtype PKB function, as has been suggested by others (98, 99), and that PKB and S6Kl represent distinct branches of the PISK signaling pathway. The demonstration that PKB is not involved in physiological activation of S6Kl has initiated the search for additional kinases that are known downstream effecters of PI3K. Earlier studies had revealed that PKCC is regulated in vitro by phosphatidylinositol3,4,5triphosphate ptdIns(3,4,5)P3] (IOO), and that PKCh as well as PKC< are activated in viva through a pathway involving PI3K (101-103). Studies have shown that both PKCh and PKCL associate with S6Kl in a mitogen-independent manner that kinase-dead mutants of both PKCX and PKC[ partially inhibit S6Kl activation in cotransfection studies (104, 105). In support of these findings membrane-targeted alleles of PKC< increase S6Kl activity (105). These results suggest that the atypical PKCs may regulate S6Kl activation through direct or indirect phosphorylation of S6Kl. It should also be noted that as PDKl has been identified as the activation loop kinase for the atypical PKCs (106), implying that PDKl can regulate S6Kl activation directly by phosphorylating T229 and indirectly via the atypical PKCs (Fig. 6). In addition to the atypical PKCs it has been reported that the Rho family of small G proteins, cdc42 and Racl, associate with the hypophosphorylated forms of S6Kl in a GTP-dependent manner. However, this interaction is not sufficient to activate S6Kl (107). Cotransfection of an activated form of Rat can stimulate reporter S6K1, whereas a dominant negative form can block S6Kl activation in viva However, the mechanisms by which small GTP binding proteins contribute to S6K activation remain unclear. 3. MAMMALIANTARGETOFRAPAMYCIN The observation that the immunosuppressant rapamycin inhibits S6Kl activation and S6 phosphorylation has implicated a mammalian target of ra-

s6

PHOSPHORYLATION

AND

s6

KINASE

IN CELL

GROWTH

115

I

\ J Pl3K

PKB

amino acids autophagy

atypical

PKCs

E2Fcell cycle

insulin transcription

5’TOP mRNAs translation

general translation

FIG. 6. Schematic representation of mitogen-induced S6K signaling cascade. The mitogeninduced activation of S6K is mediated by PI3K. The involvement of other molecules in S6K activation is presented as described in the text. Downstream effects of S6Kl activation are also shown.

pamycin (mTOR) as an upstream effector of S6Kl (108-U). The effects of rapamycin on mTOR are exerted through a gain-of-function inhibitory complex between rapamycin and the immunophilin, FKBP-12 (6, 7). Sequence comparisons revealed that mTOR shares homology with the phosphatidylinositide kinases, although no lipid kinase activity has yet been detected. In contrast, mTOR autophosphorylates (112) and has been shown to phosphorylate a bacterially expressed S6Kl on T389 (113), the principal site of rapamycin-induced S6Kl inactivation and dephosphorylation (77). However, mTOR was also shown to phosphorylate 4E-BP1 (Fig. 6) (114), although the extent to which it phosphorylates all of the initially identified sites has been questioned (115). Rapamycin-resistant and kinase-dead alleles of mTOR either protect S6Kl from the inhibitory effects of rapamycin or block the effects of mitogens on S6Kl activation (113). That mTOR directly regulates S6Kl T389 phosphorylation, however, is in conflict with a number of observations. First, a rapamycin-resistant form of S6K1, lacking the amino and carboxyltermini, is unimpaired in insulin-induced T389 phosphorylation in cells that are pretreated with rapamycin. Nevertheless, T389 phosphorylation is still blocked by wortmannin pre-

116

SINISA VOLAREVIC

AND GEORGE

THOMAS

treatment (72). This latter finding also argues against a model that places PI3K upstream of mTOR (92). Second, the T389 phosphorylation site is flanked by large, bulky aromatic amino acids, whereas the phosphorylation sites in 4E-BP1 are followed by a proline at the + 1 position. Finally, only the bacterially expressed form of S6Kl but not a mammalian expressed form serves as a substrate for mTOR in vitro (113). In contrast to the model above, independent studies had led to the hypothesis that mTOR may regulate S6Kl activation by inhibiting a phosphatase rather than directly phosphorylating S6Kl (72). In addition, earlier experiments had shown that S6Kl is selectively dephosphorylated in vitro by protein phosphatase 2A (PPZA) (116), consistent with the finding that treatment of cells with insulin leads to a general inhibition of PPSA activity, which can be blocked by either rapamycin or wortmannin treatment (217). More recent studies have demonstrated that PPBA directly associates with S6Kl (118) and that th is association requires the presence of an intact S6Kl amino terminus (119). It is this domain that is known to impart rapamycin sensitivity to S6Kl (71- 73). The existence of an mTOR-regulated phosphatase activity in mammals has been supported by studies in yeast whereby TOR was observed to mediate the interaction of the SIT4 phosphatase, the yeast homolog of mammalian PP6, with Tap42, an essential yeast protein (120). Genetic evidence indicates that Tap42 positively regulates both PPBA and PP6 and mutations in Tap42 confer rapamycin resistance (120). The closest relative to Tap42 in mammals is a4 (120). Consistent with the results in yeast, rapamycin prevented association of PPBA and cx4 in mammalian cells (121). Additionally, PPBA activity was increased in vitro toward myelin basic protein and phosphorylase A when complexed with a4. However, with regard to a potential in vivo substrate, 4E-BPl, the data are less straightforward. In one case treatment with rapamycin increased PPBA activity toward 4E-BP1 (129), whereas in a second study PPBA activity toward 4E-BP1 in vivo was inhibited by (~4 in a rapamycin-independent manner (122). One possibility that may explain some of these seemingly disparate results is that mTOR may function as a scaffold, tightly associating with both the kinases and the phosphatases responsible for regulating S6Kl activity (6). At this juncture more information will be required to evaluate these possibilities. In addition to mitogenic stimulation, it has been demonstrated that essential amino acids also modulate S6Kl activation, in a PISK-independent manner (123). The first suggestion for a role for essential amino acids in S6Kl activation came from studies on autophagy. Autophagy is distinct from ubiquitin-mediated proteolysis in that it utilizes an invaginated membrane structure, termed an autophagosome, which fuses with the lysosomes targeting proteins for degradation (124). D uring fasting, autophagic degradation of

s6

PHOSPHORYLATION

AND

s6

KINASE IN CELL

GROWTH

117

proteins is stimulated in order to produce amino acids and nucleotides for gluconeogenesis and other essential metabolic pathways (124). It had been shown that amino acid deprivation of isolated liver cells induced autophagy and blocked S6 phosphorylation, leading to the suggestion that these events may be linked (125). A possible connection between S6Kl activation and amino acids levels was shown in experiments in which amino acid removal blocked S6Kl activation by mitogens, whereas readdition of amino acids reversed this response (123,126-129). That this response is mediated through mTOR was suggested from the finding that rapamycin prevented activation by amino acids of wild-type S6K1, but not a rapamycin-resistant form of S6Kl (123). The above results suggest that mTOR is directly or indirectly involved in regulating S6Kl activation by amino acids. In support of this hypothesis, rapamycin induces autophagy in yeast, even in the presence of nutrients (130). Indeed, a TORl-deficient yeast strain bearing a TOR2 temperature-sensitive allele, when grown at the permissive but not the nonpermissive temperature, induces autophagy, consistent with TOR regulating autophagy (130). As pointed out, amino acid activation of S6Kl does not require PI3K, because it is not affected by wortmannin (128). Moreover, Patti et al. found that amino acids inhibit the early steps of insulin signaling, such as increased tyrosine phosphorylation of IRS-l and IRS-2, increases in ~85 and Grb-2 binding to IRS-l and IRS-2, and stimulation of PISK (127). Consistent with these results, amino acids in the same cells blocked insulin-stimulated proliferation (127). It is tempting to speculate that amino acids may have some role in the stimulation of cell growth in the absence of proliferation-for example, in liver cells during feeding following fasting. Further studies will be necessary to determine the role of mTOR and S6Kl activation in the regulation of balance of protein synthesis and protein degradation by autophagy.

III. Downstream Effecters of S6K A. S6 Phosphorylation

and 5’TOP

mRNA Translation

As previously stated, the mitogen-stimulated increase in the biogenesis of translational components is a critical requirement for cells to increase protein synthetic capacity, grow, and ultimately proliferate. S6K1, presumably through S6 phosphorylation, is involved in translational up-regulation of 5’TOP mRNAs (Fig. 6) (45), wh ic h encode for components of the translational apparatus. In support of this conclusion, rapamycin and dominant interfering mutants of S6K1, both of which prevent S6 phosphorylation (46), block the selective translational up-regulation of 5’TOP mRNAs, whereas ra-

118

SINISA

VOLAREVIC

AND GEORGE

THOMAS

pamycin-resistant forms of S6Kl protect mitogen-induced 5’TOP mRNA translation from the inhibitory effects of the macrolide (45). To determine the involvement of S6 phosphorylation in this response will require both an in vitro and a genetic approach. Studies employing purified protein synthesis initiation factors and defined mRNA transcripts have made it possible to measure the formation of 40s and 80s preinitiation complexes in a process termed “toe printing” (131). Employing such strategies should eventually allow the identification of those components required to form competent initiation complexes with 5’TOP mRNAs. To resolve this issue a genetic approach in theory also could be applied in both the mouse and in Drosophila. In both cases it should be possible to generate mutants in which the wild-type S6 is replaced with an S6 mutant harboring specific changes in identified phosphorylation sites. In the mouse this can be approached by generating an inducible-targeted deletion of the S6 gene in specific tissues, and introducing S6 mutant alleles by either a similar strategy or with viral vectors. A parallel approach could be used in Drosophila, by taking advantage of the described P-element mutations in the S6 gene. One could initially attempt to compliment these mutants by generating transgenic flies harboring a wild-type allele. If such a strategy were effective the next step would be to employ S6 mutants. The advantage of the mouse system would be the potential to carry out follow-up biochemical studies, whereas with Drosophila the genetic manipulations could be more readily carried out. An additional advantage of Drosophila would be the possibility to employ genetic tools, such as genetic interacting screens with other potential regulatory components. It should also be noted that rapamycin and dominant interfering mutants of S6Kl do not completely suppress the translational up-regulation of 5’TOP mRNAs, suggesting the existence of an additional regulatory mechanism involved in controlling the expression of this family of mRNAs (42, 45). Th is mechanism by definition is independent of rapamycin, S6K1, and S6 phosphorylation.

B. S6Kl

and Cell Cycle Progression

Earlier studies have shown that microinjection of inhibitory antibodies against S6Kl significantly represses the progression of cells from G, to S phase of cell cycle (132). Similar results have been obtained in some cell types with rapamycin (108). The most logical explanation for this finding would be that by inhibiting S6Kl activation, and consequently S6 phosphorylation, synthesis of nascent translational components also would be repressed. Work in T lymphocytes suggests that S6Kl activation could also regulate the activity of a critical cell cycle components. In T lymphocytes interleukin-2 (IL2) uses PI3K to regulate E2F transcriptional activity, which is a critical step in cell cycle progression (133). IL-2 stimulation of E2F transcriptional acti-

s6

PHOSPHORYLATION

AND

s6

KINASE

IN CELL

GROWTH

119

vation is made up of both rapamycin-resistant and -sensitive components (134). Expression of a rapamycin-resistant S6Kl could restore rapamycin-inhibited E2F responses (Fig. 6) (134). H owever, the same S6Kl mutant could not rescue an inhibitory effect of rapamycin on cell cycle progression, suggesting additional rapamycin-sensitive effecters in the control of IL-2-mediated T cell cycle progression. Whether the effects of S6Kl on E2F transcriptional activity are mediated through a translational-independent or -dependent step has not been established. However, given the fact that rapamycin suppresses retinoblastoma protein phosphorylation in this system it should be possible to dissect the mechanism by analyzing known upstream cell cycle regulators.

C. Insulin

Production

S6Kl has also been implicated in the regulation of insulin-induced gene transcription in pancreatic R cells (Fig. 6). Leibiger et al. have shown that following glucose-induced release of insulin, insulin acts via its R cell receptor to up-regulate its own transcription in primary pancreatic R cells (135). The analysis of this pathway implicated S6Kl as one of the critical components regulating transcription of the insulin gene. This conclusion was based on the observation that rapamycin and a specific PI3K inhibitor, LY294002, blocked insulin transcription, However, more convincingly, rapamycin-resistant alleles of S6Kl rescued the rapamycin-suppressed transcription of the insulin gene. Interestingly, proliferation of rhabdomyosarcomas is driven by an intracellular growth factor (IGF-2) autocrine-loop, which is blocked by rapamycin (136). Furthermore, the IGF-2 transcript that has been implicated contains an oligopyrimidine tract in its 5’UTR and rapamycin suppresses its translation (137). It will be of interest to determine whether these effects on IGF-2 are mediated through S6Kl. The observation that S6Kl is involved in autocrine regulation of insulin secretion suggests that the mutation in the S6Kl gene may be responsible for at least some cases of metabolic dysfunction in noninsulin-dependent diabetes mellitus (NIDDM) (138). Analysis of S6Kl- and S6K2-deficient mice will aid in determining the impact of these signaling components on this complex disease.

IV. Physiological Importance A. S6Kl

Deletion

in Mice

and Discovery

of S6K2

Studies addressing the function of S6Kl in mammalian cells have been largely performed in tissue culture model systems. To obtain corroborative in T&O data in the animal to support a role for S6Kl in cell growth, the S6Kl

120

SINISA

VOLAREVIC

AND GEORGE

THOMAS

catalytic domain was targeted in embryonic stem (ES) cells and the mutation transferred through the germ line to generate heterozygous and homozygous mutant mice lacking S6Kl protein (65). Unexpectedly, homozygous mice deficient for the S6Kl protein were viable and fertile, although they had a reduced body size. The size reduction was most pronounced during the embryonic stages, with heterozygous mice displaying an intermediate phenotype, suggesting that the absence of S6Kl gene product effected an early stage of embryogenesis. The mouse embryonic fibroblasts (MEFs) and splenic T lymphocytes from either mutant or wild-type mice showed similar proliferative responses when stimulated with serum or anti-T cell receptor antibodies, respectively. Rapamycin treatment inhibited the proliferative response equally in both wild-type and S6Kl mutant mice. These findings were supported by the observation that full stochiometric S6 phosphorylation was still observed in the livers of fasted S6Kl deficient mice after refeeding, a treatment known to induce S6Kl activation and S6 phosphorylation. Serum stimulation of SGKl-deficient MEFs also induced S6 phosphorylation and the translational up-regulation of 5’TOP mRNAs in a rapamycin-sensitive manner. The above results suggested the existence of an additional rapamycinsensitive S6 kinase, which was subsequently confirmed by the identification of S6K2 (65). In all tissues examined, compensatory up-regulation of the S6K2 transcript was found in SGKl-deficient mice, consistent with its putative role as an S6Kl homolog (65). S e q uence comparison of S6K2 revealed a kinase with almost identical domain organization and all the same regulatory phosphorylation sites found in S6Kl. The overall identity between S6Kl and S6K2 is 670/o,with the catalytic domains sharing 82% identity (Fig. 2) (65- 67). The major significant difference between the two kinases resides in the carboxy terminus of the molecules, where a proline-rich domain and a potential nuclear localization sequence are found in S6K2, but not in S6Kl. The proline-rich domain could, through its interaction with SH3-containing proteins, affect cellular localization and function of S6K2, allowing nonoverlapping functions with S6Kl. The importance of S6K2 and S6Kl as in viva S6 kinases will be tested following the deletion of the S6K2 gene in mice. Additionally, comparison of the phenotypes of S6Kl- and S6K2-deficient mice could aid in the elucidation of potential differential functions in vivo for the two highly homologous kinases.

B. Drosophila

S6K Mutant

The mammalian genome is thought to have been duplicated two times (139), which in some cases makes phenotypic characterization of a specific gene deficiency difficult, as exemplified by SGKl-deficient mice. Therefore, in some instances it is more attractive to study gene function in less complex

s6 PHOSPHORYLATION

AND

s6 KINASE IN CELL GROWTH

121

organisms, such as Drosophila melanogastw, in which individual genes are usually represented in a single copy. Additionally, the mammalian signaling pathways are largely conserved in Drosophila and Drosophila offers a powerful genetic system to uncover novel components in signaling cascades. As a first step in the use of Drosophila to study S6Kl signaling and function, a Drosophda S6Kl homolog, dS6K was cloned (140, 141). The dS6K protein has an overall identity of 57% with either S6Kl or S6K2 and 78% identity in the catalytic domain. The organization of functional domains and key phosphorylation sites is also conserved between the two enzymes. The observation that dS6K phosphorylated rat S6 in both a rapamycin- and wortmanninsensitive manner further suggested that upstream elements and downstream effecters are conserved in mammals and flies (140, 141). In order to study the functional role of dS6K, Montagne and colleagues characterized a female sterile mutation (142), which they showed contained a P-element insertion in the dS6K 5’UTR (143). Only 25% of the expected number of homozygous flies emerged as adults, with a 3-day delay in development and reduced body size. That mutant phenotype observed was due to insertion of the P-element into the dS6K gene, demonstrated by showing that all aspects of the mutant phenotype were rescued by either excision of the Pelement or by expression of a dS6K transgene or transgenes expressing either of the two mammalian S6Ks. Because the P-element insertion into the dS6K gene resulted only in decreased expression of S6K, a more severe allele was obtained by generating imprecise excisions that removed part of the gene. Only a few of those homozygous flies survived, emerging after a 5-day delay in development, living a maximum of 2 weeks and showing an almost 50% reduction in body size. Morphologically, all body parts were reduced in size to the same extent (143). Given that growth is dominant over proliferation (2), it was rationalized that the reduction in body size of homozygous dS6K mutant flies was the result of decreased cell number. This possibility was tested by counting the number of cells in the wings and in the eyes. Surprisingly, the results revealed that the cell size was reduced, but not the cell number. Furthermore, analysis of the developing wing disc revealed that that the effect on cell size was displayed throughout development. Further experiments demonstrated that mutant cells were proliferating at half the rate of wild-type cells and that that the decrease in the rate of proliferation was exhibited equally throughout all stages of the cell cycle. Because S6Kl has been implicated in the regulation of insulin and IGF-2 (136, 137), it raised the possibility that the effect on cell size was a humoral response. However, the generation of genetically marked homozygous mutant dS6K somatic clones in a heterozygous background revealed that the effect was cell autonomous. Consistent with this conclusion, expression of an extra copy of the dS6K gene in the dorsal compartment of

122

SINISA VOLAREVIti

AND GEORGE THOMAS

the wing caused only those cells and not the cells in the ventral compartment to grow larger. Thus &6K regulates cell size, growth, and proliferation in a cell autonomous manner without impinging on cell number. Because S6K is implicated in the regulation of translational components that make up the protein synthetic apparatus, the dS6K mutant was speculated to mimic ribosomal protein mutants, termed Minutes, a name derived from the strong reduction in bristle size (144). Surprisingly, the phenotype of the BS6K mutant was quite distinct from that of Minutes. Although both were delayed in development, consistent with a reduced ribosome content, dS6K mutants did not exhibit a strong reduction in bristle size and Minutes displayed a wild-type cell and body size (143). Taken together, these data imply that the defect in the number of ribosomes is not the only reason for the observed phenotype in dS6K mutants. It is possible that dS6K is involved in the modulation of a putative cell size regulator. Alternatively, the difference between dS6K mutant cells and Minute cells could be that the &6K-deficient cells, in addition to a decreased number of ribosomes, have a decreased level of phosphorylated S6, which could affect cell size through favoring the translation of mRNAs lacking a 5’TOP. If such mRNAs encoded cell cycle regulators, such regulators could reach a critical concentration at a smaller cell size. Evidence for such a model has been presented for CDC25A in Xenopus mitotic cell cycles (59).

V. Future Perspectives A great deal of information about the mechanism of S6Kl activation has been obtained by dissecting the primary structure of the enzyme into domains and identifying critical regulatory phosphorylation sites. However, less progress has been accomplished in identifying upstream regulators, which are responsible for phosphorylation of critical regulatory sites of S6Kl. Only PDKl kinase has been identified as a direct in vivo and in vitro modulator of S6Kl activity. The genetics of Drosophila will certainly be an important tool in elucidating new components of this signaling cascade. Of particular interest is the molecular mechanism by which mTOR functions to bring about S6Kl activation and 4E-BP1 phosphorylation. Another important field of endeavor will be the identification of signaling components involved in amino acid activation of S6K and determination of the relationship of this signaling pathway to insulin signaling and the regulation of autophagy. Mice deficient in S6Kl and S6K2 will certainly aid in the determination of differential functions of these two isoforms in vivo. Finally, a major task will be to determine the role of S6 phosphorylation in mediating S6K response, both in mammals and in other model organisms such as Drosophila melanogaster.

s6 PHOSPHORYLATION

AND

s6 KINASE IN CELL GROWTH

123

Inhibition of the mTOR/SGK signaling pathway by rapamycin has shown great promise as an immunosuppressant for use in organ transplantation. Selective pharmacological manipulation of this signaling cascade, which is intimately involved in cell growth, may be useful in modulating the proliferative response associated with such diseases as cancer.

ACKNOWLEDGMENTS We thank P. B. Dennis for a critical reading of the manuscript. We are also grateful to the European Economic Community (Grant Number BI04-CT97-2071) and Novartis Research Foundation for their financial support.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9, 10. 11. 12. 13.

14.

15. 16. 17. 18. 19. 20. 21. 22.

C. J. Sherr and J. M. Roberts, Genes Dev. 13,1501-1512 (1999). L. H. Hartwell and M. W. Unger,J. Cell Biol. 75,422-435 (1977). G. C. Johnston, J. R. Pringle, and L. H. Hartwell, Exp. Cell Res. 105, 79-98 (1977). A. B. Pardee, Science 246,603-608 (1989). K. Nasmyth, Nature (London) 382,28-29 (1996). P. B. Dennis, S. FumagaIIi, and G. Thomas, Cum. Opin. Genet. Dec. 9,49-54 (1999). G. Thomas and M. N. Hall, Cum Opin. Cell Biol. 9,782-787 (1997). R. L. Gourse, T Gaal, M. S. Bartlett, J. A. Appleman, and W. Ross, Annu. Rev. Microbial. 50,645-677 (1996). T. Gaal, M. S. Bartlett, W. Ross, C. L. Tumbough, Jr., and R. L. Gourse, Science 278,20922097 (1997). E. V. Schmidt, Oncogene l&2988-2996 (1999). (1991). S. J. Morley and G. Thomas, Pharmacol. Ther. 50,291-319 N. K. Gray and M. Wickens, Annu. Rer;. Cell Den Biol. 14,399-458 (1998). H. B. J. Jefferies and G. Thomas, In “Translational Control” (J. W. B. Hershey, M. B. Mathews, and N. Sonenberg, eds.), pp. 389-409. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. I. G. Wool, Y.-L. Chan, and A. G&k, In “Translational Control” (J. W. B. Hershey, M. B. Mathews, and N. Sonenberg, eds.), pp. 685-732. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. J. E. Loeb and C. Blat, Fed. Eur. Biochem. Sac. Lett. 10,105-108 (1970). A. M. Gressner and I. G. Wool,J. Biol. Chem. 249,6917-6925 (1974). G. Thomas, M. Siegmann, A. Kubler, J. Gordon, and L. Jimenez de Asua, Cell 19, 10151023 (1980). P. J. Nielsen, R. Duncan, and E. H. McConkey, Eur. J. Biochem. 120, 523-527 (1981). M. J. Stewart and G. Thomas, BioEssays 16, 1-7 (1994). S. C. Kozma, H. A. Lane, S. Ferrari, H. Luther, M. Siegmann, and G. Thomas, EMBO J. 8, 4125-4132 (1989). A. Greco, A. M. Laurent, and J. J. Madjar, Mol. Gen. Genet. 256,320-327 (1997). B. Buendia, A. Person-Femandez, G. Beaud, and J. Madjar, Eur. J. Biochem. 162,95-103 (1987).

124

SINISA VOLAREVIC

AND GEORGE

THOMAS

23. I’. J. Nielsen, K. L. Manchester, H. Towbin, J. Gordon, and G. Thomas,]. BioZ. Chem. 257, 12316-12321 (1982). 24. J. Krieg, J. Hofsteenge, and G. Thomas,]. Biol. Chem. 263,11473-11477 (1988). (1982). 25. R. Duncan and E. H. McConkey, Eur.J. Biochem. I23,535-538 26. G. Thomas, J. Martin-Perez, M. Siegmann, and A. M. Otto, Cell 30,235-242 (1982). 27. H. R. Bandi, S. Ferrari, J. Krieg, H. E. Meyer, and G. Thomas,J. Biol. Chem. 268,45304533 (1993). 28. J. Martin-Perez and G. Thomas, Proc. N&Z. Acad. Sci. U.S.A. 80,926-930 (1983). 29. S. Ferrari, H. R. Bandi, B. M. Bussian, and G. Thomas,Z. BioZ. Chem. 266,22770-22775 (1991). 30. H. Flotow and G. Thomas,J. BioZ. Chem. 267,3074-3078 (1992). 31. M. J. Stewart and R. Denell, Mol. Cell. BioZ. 13,2524-2535 (1993). 32. K. L. Watson, K. D. Konrad, D. F Woods, and P. J. Bryant, Proc. NatZ. Acad. Sci. U.S.A. 89, 11302-11306 (1992). 33. T A. Spencer and G. A. Mackie, Biochim. Biophys. Acta 1172,332-334 (1993). 34. R. J. Leer, M. M. van Raamsdonk-Duin, C. M. Molenaar, L. H. Cohen, W. H. Mager, and R. J. Planta, Nucleic Acids Res. 10, 5869-5878 (1982). 35. S. P. Johnson and J. R. Warner, Mol. Cell. Biol. 7, 1338-1345 (1987). 36. R. J. Planta and H. A. Raue, Trends Genet. 4,64-68 (1988). 37. 0. Meyuhas, D. Avni, and S. Shama, In “Translational Control” (J. W. B. Hershey, M. B. Mathews, and N. Sonenberg, eds.), pp. 363-388. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. (1990). 38. 0. Nygard and L. Nilsson, Eur. J. Biochem. 19I,l-17 39. G. Thomas, G. Thomas, and H. Luther, Proc. Natl. Acud. Sci. U.S.A. 78, 5712-5716 (1981). 40. G. Thomas and G. Thomas, J. Cell Biol. 103,2137-2144 (1986). 41. H. B. J. Jefferies, G. Thomas, and G. Thomas,J. BioZ. Chem. 269,4367-4372 (1994). 42. H. B. J. Jefferies, C. Reinhard, S. C. Kozma, and G. Thomas, Proc. NatZ. Acud. Sci. U.S.A. 91,4441-4445 (1994). 43. S. Levy, D. Avni, N. Hariharan, R. P Perry, and 0. Meyuhas, Proc. N&l. Acad. Sci. U.S.A. 88,3319-3323 (1991). 44. F. Amaldi and P. Pierandrei-Amaidi, “TOP Genes: A Translationally Controlled Class of Genes Including Those Coding for Ribosomal Proteins” (P. Jeanteur, ed.). Progress in Molecular and Subcellular Biology. Springer-Verlag, Berlin and Heidelberg, 1997. 45. H. B. J. Jefferies, S. Fumagalli, l? B. Dennis, C. Reinhard, R. B. Pearson, and G. Thomas, EMBOY. l2,3693-3704 (1997). 46. S. R. Von Manteuffel et al., Mol. Cell. BioZ. 17, 5426-5436 (1997). 47. L. Pellizzoni, B. Cardinah, N. Lin-Marq, D. Mercanti, and P Pierandrei-Amaldi,]. Mol. BioZ. 259,904-915 (1996). 48. L. Pellizzoni, F. Lotti, B. Maras, and P. Pierandrei-Amaldi, J. Mol. BioZ. 267, 264-275 (1997). 49. L. Pellizzoni, F. Lotti, S. A. Rutjes, and P. Pierandrei-AmaIdi, J. Mol. BioZ. 281, 593-608 (1998). 50. M. J. Stewart and R. Denell, Mol. BioZ. Evol. 10, 1041-1047 (1993). 51. T. Radimerski, T. Mini, U. Schneider, R. E. H. Wettenhall, G. Thomas, and P Jeno, Biochemistry 39,5766-5774 (2000). 52. R. C. Pate1 and M. Jacobs-Lorena,J. Biol. Chem. 267,1159-1164 (1992). 53. R. Franc0 and M. 6. Rosenfeld,J. BioZ. Chem. 265,4321-4325 (1990). 54. C. Reinhard, A. Femandez, N. J. C. Lamb, and G. Thomas, EMBOJ. 13,1557-1565 (1994). 55. I. G. Wool, Trends Biochem. Sci. 21,164-165 (1996).

s6 PHOSPHORYLATION

AND

s6 KINASE

IN CELL

GROWTH

125

56. N. Draptchinskaia, P. Gustavsson, B. Andersson, M. Pettersson, T. N. Willig, I. Dianzani, S. Ball, G. Tchemia, J. Klar, H. Matsson, D. Tentler, N. Mohandas, B. Car&son, and N. Dahl, Nature Genet. 21, 1699175 (1999). 57. S. C. Kozma, S. Ferrari, and G. Thomas, Cell. Signal. I, 219-225 (1989). 58. R. L. Erikson,]. Biol. Chem 266,6007-6010 (1991). 59. M. S. Schwab, S. H. Kim, N. Terada, C. Edfjall, S. C. Kozma, G. Thomas, and J. L. Maller, Mol. Cell. Biol. 19, 2485-2494 (1999). 60. P. Jeno, L. M. Ballou, I. Novak-Hofer, and G. Thomas, Proc. Natl. Acad. Sci. U.S.A. 85, 406-410 (1988). 61. P. Banejee, M. F. Ahamad, J. R. Grove, C. Kozlosky, D. J. Price, and J. Avruch, Proc. N&l. Acad. Sci. U.S.A. 87,8550-8554 (1990). 62. S. C. Kozma, S. Ferrari, P. Bassand, M. Siegmann, N. Totty, and G. Thomas, Proc. Natl. Acad. Sci. U.S.A. 87,7365-7369 (1990). 63. J. R. Grove, P. Banerjee, A. Balasubramanyam, P. J. Coffer, D. J. Price, J. Avruch, and J. R. Woodgett, Mol. Cell. Biol. l&5541-5550 (1991). 64. C. Reinhard, G. Thomas, and S. C. Kozma, Proc. Natl. Acud. Sci. U.S.A. 89, 4052-4056 (1992). 65. H. Shima, M. Pende, Y. Chen, S. Fumagalli, G. Thomas, and S. C. Kozma, EMBO J. 17, 6649-6659 (1998). 66. I. Gout, T. Minami, K. Hara, Y. Tsujishita, V. Filonenko, M. D. Waterfield, and K. Yonezawa, J. Biol. Chem. 273,30061-30064 (1998). 67. M. Saitoh, P. ten Dijke, K. Miyazono, and H. Ichijo, Biochem. Biophys. Res. Commun. 253, 470-476 (1998). (1997). 68. N. Pullen and G. Thomas, Fed. Eur. Biochem Sot. Leti. 410,78-82 69. N. Pullen, P. 8. Dennis, M. Andjelkotic, A. Dufner, S. Kozma, B. A. Hemmings, and G. Thomas, Science 279,707-710 (1998). 70. D. R. Alessi, M. T. Kozlowski, Q. P Weng, N. Morrice, and J. Avruch, Curr. BioZ. 8,69-81 (1998). 71. L. Cheatham, M. Monfar, M. M. Chou, and J. Blenis, Proc. Natl. Acad. Sci. U.S.A. 92, 11696-11700 (1995). 72. P. B. Dennis, N. Pullen, S. C. Kozma, and G. Thomas, Mol. CelZ. Biol. 16,6242-6251(1996). 73. Q.-P. Weng, K. Andrabi, M. T Kozlowski, J. R. Grove, and J. Avruch, Mol. Cell. Biol. 15, 2333-2340 (1995). 74. Q.-P. Weng, K. Andrabi, A. Klippel, M. T. Kozlowski, L. T. Williams, and J. Avruch, Proc. Natl. Acad. Sci. U.S.A. 92,5744-5748 (1995). 75. R. B. Pearson and G. Thomas, In “Progress in Cell Cycle Research, Vol. 1” (L. Meijer, S. Guidet, and H. Y. L. Tung, eds.), pp. 21-32. Plenum Press, New York, 1995. 76. S. K. Hanks, A. M. Quinn, and T. Hunter, Science 241,42-52 (1988). 77. R. B. Pearson, P. B. Dennis, J. W. Han, N. A. Williamson, S. C. Kozma, R. E. H. Wettenhall, (1995). and G. Thomas, EMBO J. 21,5279-5287 78. B. A. Moser, P. B. Dennis, N. Pullen, R. B. Pearson, N. A. Williamson, E. H. Wettenhall, S. C. Kozma, and G. Thomas, Mol. Cell. Biol. 17,5648-5655 (1997). 79. S. Ferrari, W. Bannwarth, S. J. Morley, N. F. Totty, and G. Thomas, Proc. ,Natl. Acad. Sci. U.S.A. 89,7282-7285 (1992). 80. P E. Burnett, S. Blackshaw, M. M. Lai, I. A. Qureshi, A. F. Burnett, D. M. Sabatini, and S. H. Snyder, Proc. Nutl. Acud. Sci. U.S.A. 95,8351-8356 (1998). 81. H. Nakanishi, H. Obaishi, A. Satoh, M. Wada, K. Mandai, K. Satoh, H. Nishioka, Y. Matsuura, A. Mizoguchi, and Y. Takai, J. Cell Biol. 139,951-961 (1997). 82. P. B. Dennis, N. Pullen, R. B. Pearson, S. C. Kozma, and G. Thomas, J. Biol. Chem. 273, 14845-14852 (1998).

126

SINISA VOLAREVIC

AND GEORGE THOMAS

83. D. R. Alessi, S. R. James, C. P Dowries, A. B. Holmes, I? R. Gaffney, C. B. Reese, and P. Cohen, Cum. F&Z. 7,261-269 (1997). 84. D. Stokoe, L. R. Stephens, T. Copeland, P. R. J. Gaffney, C. B. Reese, G. F. Painter, A. B. Holmes, F. McCormick, and P. T Hawkins, Science 277,567-570 (1997). 85. A. J. Flint, R. D. Paladini, and D. E. J. Koshland, Science 249,408-411 (1990). 86. S. E. Tsutakawa, K. I? Medzihradszky, A. J. Flint, A. L. Burlingame, and D. E. J. Koshland, J. Biol. Chem. 270,26807-26812 (1995). 87. L. M. Keranen, E. M. Dutil, and A. C. Newton, Cum Biol. 5,1394-1403 (1995). 88. N. K. Mukhopadhayay, D. J. Price, J. M. Kyriakis, S. L Pelech, J. Sanghera, and J. Avruch, J. Biol. Chem. 267,3325-3335 (1992). 89. L. M. Ballou, H. Luther, and G. Thomas, Nature (Londonj 349,348-350 (1991). 90. X. F. Ming, B. M. Burgering, S. Wennestrom, L. Claesson-Welsh, C. H. Heldin, J. L. Boss, S. C. Kozma, and G. Thomas, Nature (London) 371,426-429 (1994). 91. J. Chung, T. C. Grammer, K. P. Lemon, A. Kazlauskas, and J. Blenis, Nature (London) 370, 71-75 (1994). 9.2. G. J. Brunn, J. Williams, C. Sabers, G. Wiederrecht, J. C. Lawrence, Jr,, and R. T. Abraham, EMBOJ. 15,5256-5267 (1996). 93. T. F. Franke, S. I Yang, T. 0. Chan, K. Datta, A. Kazlauskas, D. K. Morrison, D. R. Kaplan, and P. N. Tsichlis, Cell 81, 727-736 (1995). 94. B. M. T. Burgering and P. J. Coffer, Nature (London) 376,599-602 (1995). 95. N. M. Conus, B. A. Hemmings, and R. B. Pearson,J. Biol. Chem. 273,4776-4782 (1998). 96. A. Dufner, M. Andjelkovic, B. M. T. Burgering, B. A. Hemmings, and G. Thomas, Mol. Cell. Biol. 19,4525-4534 (1999). 97. D. Alessi, M. Andjelkovic, B. Caudwell, P. Cron, N. Morrice, P Cohen, and B. A. Hemmings, EMBOJ. 23,6541-6551 (1996). 98. D. R. Alessi and C. P. Downes, Biochim. Biophys. Actu 1436,151-164 (1998). 99. M. Aoki, 0. Batista, A. Bellacosa, P Tsichlis, and P. K. Vogt, Proc. N&l. Acad Sci. U.S.A. 95,14950-14955 (1998). 100. H. Nakanishi, K. A. Brewer, and J. H. Exton,J. Biol. Chem. 268,13-16 (1993). 101. K. Akimoto, R. Takahashi, S. Moriya, N. Nishioka, J. Takayanagi, K. Kimura, Y. Fukui, S. Osada, K. Mizuno, S. Hirai, A. Kazlauskas, and S. Ohno, EMBOJ. 15,788-798 (1996). 10.2. M. L. Standaert, L. Galloway, P. Kamam, G. Bandyopadhyay, J. Moscat, and R. V. Farese, (1997). J. Biol. Chem. 272,30075-30082 103. A. Toker and L. C. Cantley, Nature (London) 387,673-676 (1997). 104. K. Akimoto, M. Nakaya, T. Yamanaka, J. Tanaka, S. Matsuda, Q. P. Weng, J. Avruch, and S. Ohno, Biochem. J. 335,417-424 (1998). 105. A. Romanelli, K. A. Martin, A. Toker, and J. Blenis, Mol. Cell. Biol. 19,2921-2928 (1999). 106. J. A. Le Good, W. H. Ziegler, D. B. Parekh, D. R. Alessi, P. Cohen, and P. J. Parker, Science 281,2042-2045 (1998). 107. M. M. Chou and J. Blenis, Cell 85,573-583 (1996). 108. N. H. Sigal, C. S. Lin, and J. J. Siekierka, Transplant. Proc. 23, l-5 (1991). 109. J. Chung, C. J. Kuo, G. R. Crabtree, and J. Blenis, Cell 69, 1227-1236 (1992). 110. C. J. Kuo, J. Chung, D. F. Fiorentino, W. M. Flanagan, J. Blenis, and G. R. Crabtree, Nuture (Londonj 358, 70-73 (1992). 111. D. J. Price, J. R. Grove, V. Calvo, J. Avruch, and B. E. Bierer, Science 257,973-977 (1992). 112. E. J. Brown, I? A. Beal, C. T Keith, J. Chen, T. B. Sh’m, and S. L. Schreiber, Nature (London) 377,441-446 (1995). 113. P. E. Burnett, R. K. Barrow, N. A. Cohen, S. H. Snyder, and D. M. Sabatini, i’roc. Natl. Acud. Sci. U.S.A. 95, 1432-1437 (1998). 114. G. J, Brunn, C. C. Hudson, A. Sekulic, J. M. Williams, H. Hosoi, P. J. Houghton, J. C. Lawrence, Jr., and R. T. Abraham, Science 277,999101 (1997).

s6 PHOSPHORYLATION

AND

s6 KINASE IN CELL GROWTH

127

115. A. C. Gingras, S. P Gygi, B. Raught, R. D. Polakiewicz, R. T. Abraham, M. F. Hoekstra, R. (1999). Aebersold, and N. Sonenberg, Genes Deo. 13,1422-1437 116. L. M. Ballou, P. Jeno, and G. Thomas,Z. Biol. Chem. 263,1188-1194 (1988). 117. N. Begum and L. Ragolia,J. Biol. Chem. 271,31166-31171 (1996). 118. R. S. Westphal, R. L. Coffee, Jr., A. Marotta, S. L. Pelech, and B. E. Wadzinski,J. BioZ. Chem. 274,687-692 (1999). 119. R. T. Peterson, B. N. Desai, J. S. Hardwick, and S. L. Schreiber, Proc. NutZ. Acad. Sci. U.S.A. 96,4438-4442 (1999). 120. C. J. Di Como and K. T. Arndt, Genes Dw. 10,1904-1916 (1996). 121. J. Chen, R. T Peterson, and S. L. Schreiber, Biochem. Biophys. Res. Commun. 247,827832 (1998). 122. M. Nanahoshi, T. Nishiuma, Y. Tsujishita, K. Hara, S. Inui, N. Sakaguchi, and K. Yonezawa, Biochem. Biophys. Res. Commun. 251520-526 (1998). 123. K. Hara, K. Yonezawa, Q. P. Weng, M. T. Kozlowski, C. Belham, and J. Avruch, J. BioZ. Chem. 273,14484-14494 (1998). 124. VI! A. J. Dunn, Trends Cell Biol. 4, 139-143 (1994). 125. E. F. Blommaart, J. J. Luiken, P. J. Blommaart, G. M. van Woerkom, and A. J. Meijer,J. Biol. Chem. 270,2320-2326 (1995). 126. Y. Iiboshi, P. J. Papst, H. Kawasome, H. Hosoi, R. T. Abraham, P. J. Houghton, and N. Terada,Z. BioZ. Qrem. 274, 1092-1099 (1999). 127. M. E. Patti, E. Brambilla, L. Luzi, E. J. Landaker, and C. R. Kahn,]. CZin. Invest. 101,1519 1529 (1998). 128. K. Shigemitsu, Y. Tsujishita, K. Hara, M. Nanahoshi, J. Avruch, and K. Yonezawa, J. BioZ. Chem. 274,1058-1065 (1999). 129. G. Xu, G. Kwon, C. A. Marshall, T. A. Lin, J. C. Lawrence, Jr., and M. L. McDaniel, J. BioZ. Chem. 273,28178-28184 (1998). 130. ‘I Noda and Y. Ohsumi,J. Biol. Chem. 273,3963-3966 (1998). 131. T. V. Pestova, S. I. Borukhov, and C. U. Hellen, Nature &on&n) 394,854-859 (1998). 132. H. A. Lane, A. Fernandez, N. J. C. Lamb, and G. Thomas, Nature (London) 363,170-172 (1993). 133. P. Brennan, J. W. Babbage, B. M. Burgering, B. Groner, K. Reif, and D. A. Cantrell, Zmmunity 7, 679-689 (1997). 134. P. Brennan, J. W. Babbage, G. Thomas, and D. Cantrell, Mol. Cell. Biol. 19, 4729-4738 (1999). 135. I. B. Leibiger, B. Leibiger, T. Moede, and P 0. Berggren, Mol. Cell 1,933-938 (1998). 136. H. Hosoi, M. B. DiIIing, T. Shikata, L. N. Liu, L. Shu, R. A. Ashmun, G. S. Germain, R. T. Abraham, and P. J. Houghton, Cancer Res. 59,886-894 (1999). 137. F. C. Nielsen, L. Ostergaard, J. Nielsen, and J. Christiansen, Nature (London) 377, 358362 (1995). 138. S. I. Taylor, Cell 97, 9-12 (1999). 139. G. L. G. Miklos and G. M. Rubin, Cell 86,521-529 (1996). 140. M. J. Stewart, C. 0. A. Berry, F. Zilberman, G. Thomas, and S. C. Kozma, Proc. N&Z. Acad. 5%. U.S.A. 93,10791-10796 (1996). 141. K. L. Watson, M. M. Chou, J. Blenis, W. M. Gelbart, and R. L. Erickson, Proc. Natl. Acad. Sci. U.S.A. 93,13694-13698 (1996). 142. G. H. Karpen and A. C. Spradling, Genetics 132, 737 (1992). 143. J. Montagne, M. J. Stewart, H. Stocker, E. Hafen, S. C Kozma, and G. Thomas, Science 285, 2126-2129 (1999). 144. A. Lambertsson, Adrj. Genet. 38,69-134 (1998).

Molecular Characterization of Monoamine Oxidases A and B CREED

W. ABELL*~~ AND

SAU-WAH

KWAN”

*LX&ion

of Medicinal Chemistry College of Pharmacy *The lnstitute for Neuroscience ?he Institute fw Cellular and Molecular Biology University of Texas Austin, Texas 78712

I

I. Gene Cloning and Structure ........................ A. Cloning and Sequencing of MAO A and B cDNAs ... B. Chromosomal Location and Gene Organization ..... II. FunctionalRegionsofMAOB ....................... A. Strategy for Identifying Crucial Amino Acid Residues B. Dinucleotide-Binding Site ....................... C. Second FAD-Binding Site ........................ D. Fingerprint Site ................................ E. Covalent-Binding Site ........................... E Active Site ..................................... G. Other Potential Targets .......................... III. Tissue and Cell Distribution ........................ . Physiological Functions ............................ V FuturePerspectives ................................ References .......................................

Monoamine degrading

A and B) are the major neurotransmitter-

enzymes in the central nervous system and in peripheral

A and B cDNAs deduced

oxidase A and B (MAO

131 131 132 135 135 136 141 144 14-i 148 149 149 150 152 152

..

tissues. MAO

from human, rat, and bovine species have been cloned and their

amino acid sequences

enzyme shows approximately

compared.

Comparison

70% sequence

of A and B forms of the

identity, whereas comparison

A or B forms across species reveals a higher

sequence

of the

identity of 87%. Within

these sequences, several functional regions have been identified that contain crucial amino acid residues participating

in flavin adenine dinucleotide

strate binding. These include a dinucleotide-binding

(FAD) or sub-

site, a second FAD-binding

site, a fingerprint

site, the FAD covalent-binding

brane-anchoring

site. The specific residues that play a role in FAD or substrate

binding were identified by comparing with those in soluble flavoproteins

Progress in Nucleic Acid Research and Molecular Biology, Vol. fi5

site, an active site, and the mem-

sequences in wild-type and variants of MAO

of known structures. The genes that encode

129

Copyright 0 2001 IJ~ Academic Prraa. All rights ofrqxoduction in my form resewed. 0079-6603~01 $35.00

130

CREED

W. ABELL

AND SAU-WAH

KWAN

MAO A and B are closely aligned on the X chromosome (Xp11.23), and have identical exon-intron organization. Immunocytochemical localization studies of MAO A and B in primate brain showed distribution in distinct neurons with diverse physiological functions. A defective MAO A gene has been reported to associate with abnormal aggressive behavior. A deleterious role played by MAO B is the activation of I-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), a proneurotoxin that can cause a parkinsonian syndrome in mammals. Deprenyl, an inhibitor of MAO B, has been used for the treatment of early-stage Parkinson’s disease and provides protection of neurons from age-related decay. 8 zoooAcademic press.

Monoamine oxidase A and B (MAO A and B) are the major intracellular enzymes in the brain and in peripheral tissues that degrade neuroactive and vasoactive amines (1). They can also metabolize potentially harmful substances found in the environment (xenobiotics), including protoxins that enter the central nervous system (2). MAO A and B are thought to be linked to psychiatric and neurological disorders such as depression and Parkinson’s disease (PD), respectively. For example, MAO A and B are the targets of drugs that are used to treat depression (moclobemide, an MAO A inhibitor) (3) and PD (deprenyl, an MAO B inhibitor) (4 - 6). C urrent interest in the A form of the enzyme has been intensified by the discovery of a defective MAO A gene, which encodes a tnmcated enzyme in some members of a Dutch family that exhibit abnormal aggressive behavior (7). Several strains of knockout mice, each harboring a different inactivated gene (encoding MAO A, 5-HTlB receptor, o-calmodulin-dependent protein kinase II, or nitric oxide synthase), exhibit increased aggressive behavior (8-9). The existence of defective genes other than that for MAO A illustrates the complex nature of the determinants associated with aggressive behavior. Furthermore, MAO A deficiencies in the general population and in “putative” high-risk groups (e.g., males with adult attention deficit hyperactivity disorder) appear to be highly uncommon (IO). An alternate form (allele) of the MAO B gene has been identified in patients with PD (II), suggesting that an inherited variant form of MAO B may be associated with a predisposition for this disorder. However, other allelic association studies of PD for both MAO A and B failed to find polymorphisms (12). Other recent genetic studies have identified a defective gene on chromosome 4 that is responsible for a relatively rare form of hereditary PD (13). This gene encodes c-w-synuclein,a presynaptic protein that is thought to be involved in neuronal plasticity. Interestingly, ol-synuclein was identified as the precursor protein for the non+-amyloid component of plaques in Alzheimer’s disease. An inverse relationship between cigarette smoking and risk of developing PD has been observed (14). A possible explanation for neuroprotection

MAO STRUCTURE

131

AND FUNCTION

is that one or more components in cigarette smoke inhibit MAO B activity. In fact, cigarette smokers have been found to have 28 and 40% lower MAO A and B activities, respectively, than either nonsmokers or former smokers (15, 16). Because dopamine is degraded by MAO B, elevated levels of dopamine could augment the addictive effect of nicotine in smokers. A recent study has shown that a genetic polymorphism of MAO B eliminated this protection (17). This finding supports the concept that MAO B is linked to PD, but the definitive relationship remains unknown. The purpose of this review is to focus on the structural and functional relationships of these two important neurotransmitter-degrading enzymes. Because it is not possible to reference the many publications on different aspects of this topic, additional review articles on MAO A and B are included (18-22).

I. Gene Cloning and Structure A. Cloning

and Sequencing

of MAO

A and B cDNAs

MAO A and B are integral proteins of the outer mitochondrial membrane (22) and can be distinguished by differences in substrate preference (23), inhibitory specificity (24), tissue and cell distribution (25, 26), and immunological properties (27, 28). Despite these differences in the characteristics of MAO A and B, the true identity of their fundamental nature was unresolved until we and our collaborators (29) isolated and characterized the human liver cDNA clones that encode these enzymes and determined their nucleotide and deduced amino acid sequences. Comparison of sequences of MAO A and B shows that these enzymes have a very high degree of identity (approximately 700/o),but different amino acid residues occur at the same place throughout the polypeptide chains. This indicates that MAO A and B are derived from at least two separate genes rather than by a splicing mechanism. Cloning and sequencing studies (see Fig. 1) in other laboratories and by us have yielded the deduced amino acid sequences for human liver MAO A and B (29), human placental MAO A (30), bovine adrenal MAO A (31), rat liver MAO A and B (32-34), and bovine liver MAO B (Kwan and Abell, Fig. 2). Direct amino acid sequencing of about 30% of bovine liver MAO B has been reported (31). We have cloned and sequenced the entire MAO B cDNA from a bovine liver cDNA library (in the Uni-Zap XR vector, from Stratagene). All oligonucleotide primers were custom synthesized by National Biosciences. The nucleotide and deduced amino acid sequences are shown in Fig. 2. The latter sequence is also included in Fig. 1 to provide easy comparison with MAO B from rat and human tissues.

132

CREED

W. ABELL AND SAU-WAH KWAN

MAO A or B from bovine, rat, and human species (Fig. 1) will show strikingly high similarity in their amino acid sequences, and they contain motifs that are found in the vast majority of flavin-requiring enzymes (e.g., a dinucleotide-binding site) (35-37). Comparison of MAO A or B across species (human, rat, and bovine) shows 86-880/o identity, with the exception of bovine versus human MAO B, which is slightly higher (910/o).Sequence identity between MAO A and MAO B is approximately 70% across all species. Human platelet and frontal cortex MAO B were also compared and were found to have amino acid sequences identical to those of MAO B from human liver (38).

B. Chromosomal

Location and Gene Organization

Earlier work using somatic cell mouse/human hybrid lines indicated that the MAO A and B genes are located on the X chromosome (39, 40). Use of cDNA probes and in situ hybridization studies has confirmed that MAO A and B are located in close proximity within Xp11.23-11.4 (41, 42). Linkage analysis in reference pedigrees placed the MAO A locus between DXS14 and OTC (43). Furth ermore, the MAO A and B genes were found to be proximal to each other and deleted in patients with Norrie disease, a neurological disorder that is a result of a microdeletion on the short arm of the X chromosome (41, 42). Altered MAO genes that are associated with specific diseases have not been identified in human subjects, with the exception of the Dutch pedigree having defective MAO A associated with aggressive behavior (7). However, the pentameric distribution of platelet MAO B activity in healthy men (44) supports the notion that genetically determined variations in MAO occur in the human population. Identification and characterization of these variants await future genetic studies. For studies on gene organization, the human MAO A and B genes were isolated from X chromosome genomic libraries (45-47). Both genes were found to contain 15 exons, and they exhibited identical exon-intron organization. Examination of the promoter sequences in these genes revealed that fragments responsible for transcriptional activation contained GC-rich sequences and potential Spl binding sites, and had 60% sequence identity (48). The promoter region of the MAO A gene has extensive repetitive structures, including two 90-bp tandem repeats, but no TATA or CCAAT box sequences

FIG. 1. Comparison of deduced amino acid sequences of bovine (b), rat (r), and human (h) MAO A and B. Identical amino acids at the same site are indicated by an overline. The C residue that binds FAD covalently is indicated by an underline. The different groups of amino acids are color coded: red, charged; blue, polar: green, hydrophobic; and black, neutral. Sources: bMA0 A (3I); rMA0 A (32, 34); hMA0 A (29, 30); hMA0 B (29); rMA0 B (33); bMA0 B (Fig. 2).

s~IPPT~TAKI~F~~ELPS~~~QLTQXLPY SAIPPILTAKIBPRPPLPPERNQLIQRLPN NAIPPTLTAKIBPRPBLPAPRNQLIQRLPM SAIPPTLQYKIEFNPPLPMMRKQMITRVPL SAIPPVLQYKIBESPPLPILRNQLITRVPL SAVPPVLQMKIBFNPPLPYYRNQLITRVPL

300 300 300 291 291 291

-GAVIKCYYYYKPAFWKKKDYCQCMIIBD~E

-

-

-

-

330 330 330 321 321 321

QAVIKCYVYYKBABWKKKDYCQCMIIEDKK QAVIKCYYYYKEAFWKKKDYCQCMIIBDBD GSVIKCIVYYKPPPWRKKDYCQT~IIDGEB GSVIKCYVYYKBPFWRKKDPCGT~VIl?GE~ GSVIKSIVYYKPPFWRNYDYCQSMIIBGBE WROA -A -A

bUAOB -B ki43OB

-A MAOA -A

-

~ZISITLDDTKPDQSLP~MQFILARKADR APIAITLDDTKPDQSLPAIYQFILARKADR APISITLDDTKPDGSLPAIYGFILARKADR APVAYTLDDTKPBQNYAAIMQFILABKARK APIAYTLDDTKPDAGCAAIYGFILABKARK APVAYALDDTKPDGSYPAIIQFILABKARK -

-

-

360 360 360 351 351 351

---

LAKVRKDIRKR~ICBRYAKVLG~QEALH~

390 390 390 361 361 361

mB IHAOB bYK)B

LAKLBKDIRKRKICELYAKVLGSQZALYPV LAKLBKBIRKKKICELYAKVLGSQEALHPV LARLTKPERLKKLCELYAKVLGSLEALEPV LVRLTKPERLRKLCELYAKVLNSQEALQPV LARLTKEBRLKKLCDLYAKVLQSQXALHPV

bURGA WA bMAOA MJAQB IZNAOB kWAOB

BYPEKNWCQPQYSGQ~YTAYFPPDIYTPYe BYPBKNWCBBQYSGQ~YTAYRPPGIMTQYG RYPEKNNCPBQYSQG~YTAYFPPQIKTQYQ HYEEKNWCEBQYSQG~YTTYFPPGILTQYQ RY~BKNWCPBQYSQQ~YTAYFPPQILTQYG BYEEKNWCEEQYSQQ~YTSYFPPQIMTQYQ

420 420 420

-A -A -A

RVIRQPVQR~YFAQTBTATQ~SQY~KQA~R RVIRQPVGRIYFAGTETATQWSQYMEGAVB RVIRQPVQRIFFAGTBTATKWSGYMEQAVE RVLRQPVDRIYFAQTETATHNSGYMEQAVA RVLRQPVQKIFFAQTETASEWSQY~EQAVE RVLRQPVQRIYFAQTETATHWSGYMEGAVE

450 450 450 441 441 441

hM4OB HAOB bMKIB BUAOA -A blRoA

h?AOB dRL)B kUAOB

AQERAARBVLNALQKLSARDIWIQEPEAE~

411 411

411

--

480

AGPRAAREVLNALGKVAKKDIWVRAPBSKD 460 471 471 471

AGERAARPVLNQLQKVTBKDIWVQEPESKD AQERAAREILBANGKIPEDEIWQSBPESVD AQBRAARPILHAIGKIPEDEIWQPEPESVD AQBRAARPILBAYQKIPBDEIWLPEPl3SVD --VPAVBITPSFWERNLPSVSQLLKIVQ~STVPAIBITRTFLPRNLPSVPQLLKITGVSTVPAVZITBTFIPRNLPSVSQLLKIIQPSTVBAQPITTTBLERELPSVPQLLRLIGLTTI VPARPITNTBLBRRLPSVPQLLKLLGLTTI VPAKPITTTFLQRBLPSVPQLLKLIQLTTI

-

-

509 509 509 501 501 501

-SITALWFVYYRFRLLSRS -SVALLCFVLYKIKKLPC -SVTALQFVLYKYKLLPRS FSATALQFLAAKRQLLVRV LSATALQFLAEKKQLFVRB FSATALQYLAAKRQLLVRI

527 521 527 520 520 320 FIG.

1

continued

WAOA WOA BNAOA hlAOB MAOB M4AOB

30 30 30 21 21 21

YESLQKTSDAQQ~FDVV;*GGQISGL~**K

MTDLEKPNLAQHYFDVVVIQQQISQLAAAK MENQPKASIAQHYFDVVVIQQQISQLSAAK MSNKCDVVVVQQQISGMAAAK MSNKCDVIVVQQQISGYAAAK HSSKCDVVVVQGQISGHAAAK -

-LLAEHEVNVLVLEARERVGQRTYTVRNES~ LLSRYKINVLVLEARDRVGGRTYTVRNERV LLTPYQVSVLVLEARDRVGQRTYTIRNEHV LLHDSQLNVVVLEARDRVGGRTYTLRNQKV LLHDCGLSVVVLEARDCVQGRTYTIRNKNV LLHDSGLNVIVLEARDRVGQRTYTLRNQKV

60 60 60 51 51 51

k44AOA rMAOA hMAoA lMAOB z?dAOB bMAOB

DYVDV=AYVQPTQNRILRLS;Q~LETYK KWVDVGGAYVQPTQNRILRLSKRLQIETYK DYVDVQQAYVGPTQNRILRLSKELQIETYK KYVDLGQSYVQPTQNRILRLAKRLGLETYK KYVDLQGSYVGPTQNRILRLAKELQLETYK KYVDLGQSYVQPTQNHILRLSKELGLETYK

-A -A mAoA hMAOB ZM?LOB bUAOB

VNVNERLVHYVKQKTIPFRGAGPVWNPIA VNVNBRLVQYVKGKTYPFRGAFPPVWNPLA VNVSERLVQYVKQKTYPFRQAFPPVWNPIA "NEVERLIHHVKQKSYPFRGPFPPVWNPIT VNEVERLIHFVKGKSYAFRQPFPPVWNPIT VNEVBRLIHHTKQKSYPFRGSFPSVWNPIT

MAOA HAOA hNFaA hMAOB SMAOB LMAOB

YLDY~LWRTM;NM(IKEIPADAPWR~PH;V YLDYNNLWRTMDEMGKEIPVDAPWQARHAQ YLDYNNLWRTIDNHQKEIPTDAPWEAQHAD YLDHNNFWRTMDDMQREIPSDAPWKAPLAE YLDYNNLWRTHDEHGQEIPSDAPWKAPLAE YLDHNNLWRTHDDYGREIPSDAPWKAPLAE

150 150 150 141 141 141

bNAoA -A maoA hMAOB r-B bMAOB

-EWDKHTMKD;IEXICWTKTARQF;SLFVNI EWDKHTNKDLIDKICWTKTAREFAYLFVNI KWDKHTMKELIDKICWTKTARRFAYLFVNI EWDNYTMKELLDKLCWTESAKQLATLFVNL EWDYYTMKELLDKICWTNSTKQIATLFVNL QWDLHTYKELLDKICWTESSKQLAILFVNL

180 180 180 171 171 171

N~S;PHEVSALWFLWYVKPCQQTTflF~I NVTSEPHEVSALWFLWYVRQCQGTARIFSV NVTSEPHEVSALWFLWYVKQCQGTTRIFSV CVTABTHEVSALWFLWYVKQCQGTTRIIST CVTAETHEVSALWFLWYVKQCQQTTRIIST CVTAEIHEVSALWFLWYVKQCGQTTRIFST

210 210 210 201 201 201

90 90 90 81 81 81 120 120 120 111 111 111 --

H6AOA HADA MAOA

WAOB WAOB bMAOB

-mmoa MAOA hNAoA hWAOB MAOB UAOB

TNGQQERKFVGQSQQVSER;MQLLQDRVKL TNQQQERKFVQQSQQVSPQIMGLLGDKVKL TNQGQERKFVGGSGQVSERIMDLLGDQVKL TNGQQERKFVQQSGQVSERIMDLLGDRVKL

MAOA -A IINAOA hWAOB -B MAGB

R~PUTY~DQSSRNITVETLNREL~ECRYVI

240 240 240 231 231 231

TNGGQBRKFIGGSGQVSERIKDILGDRVKL

SNGGQERKFVGQSGQVSERIMDLLGDRVKL

270 270 270 261 261 261

SSPVTYIDQTDDNIIVETLNKEHYRCKYVI NHPVTHVDPSSDNIIIETLNHEHYECKYVI ERPVIYIDQTRENVLVETLNHEMYEAKYVI ERPVIHIDQTQENVVVKTLNHEIYEAKYVI

ERPVIHIDQTGENVLVETLNHELYRAKYVI

FIG.

1

CTGCAGGCGG

GGGCCGAGAT

CCAGACACCG

AAGCAGCTGG

CACCGGGTAG

CCCGGAGAGG

GGCGAGCAAC

ATG AGC AGC AT&IiTGC GAC GTG GTC GTG GTG GGG GGC GGC ATC TCA GGT ATG GCA GCA GCC MSSKCDVVVVGGGISGMAAA

-70 60 20

AAA CTT CTA CAT GAC TCT GGC TTG AAT GTG ATT GTT CTG GAA GCC CGG GAC CGC GTG GGA KLLHDSGLNVIVLEARDRVG

120 40

GGC AGG ACT TAC ACC CTT AGG AAC CAA AAA GTT AAA TAT GTG GAC CTT GGA GGA TCT TAT GRTYTLRNQKVKYVDLGGSY

180 60

GTT GGG CCA ACT CAG AAT CAT ATC TTA AGA TTA TCC AAG GAG CTA GGA TTA GAA ACC TAC VGPTQNHILRLSKELGLETY

240 80

AAG GTG AAT GAA GTA GAG CGT CTG ATT CAC CAT ACA AAG GGC AAA TCC TAC CCC TTC AGG KVNEVERLIHHTKGKSYPFR

300 100

GGC TCA TTC CCG TCT GTG TGG AAT CCT ATC ACC TAC CTA GAT CAT AAC AAC CTC TGG AGG G S F P S V W N P I T Y L D H N N L W R

360 120

ACG ATG GAT GAC ATG GGA CGA GAG ATT CCC AGT GAT GCC CCG TGG AAG GCA CCC CTT GCA TMDDMGREIPSDAPWKAPLA

420 140

GAA CAG TGG GAC CTG ATG ACA ATG AAG GAG TTG CTG GAC AAG ATC TGC TGG ACA GAA TCT EQWDLMTMKELLDKICWTES

480 160

TCA AAG CAG CTT GCT ATT CTC TTT GTG AAC CTT TGC GTC ACT GCA GAG ATC CAT GAG GTC S K Q L A I L F V N L C V T A E I H E V

540 180

TCC GCT CTC TGG TTC CTG TGG TAT GTG AAG CAG TGT GGG GGC ACG ACC AGG ATC TTC TCA SALWFLWYVKQCGGTTRIFS

600 200

ACA TCC AAT GGA GGG CAG GAG AGG AAA TTT GTG GGT GGA TCT GGT CAA GTG AGT GAG CGG T S N G G Q E R K F V G G S G Q V S E R

660 220

ATA ATG GAC CTC CTG GGG GAT CGA GTG F&G CTG GAG AGG CCT GTG ATC CAC ATT GAC CAG I M D L L G D R V K L E R P V I H I D Q

720 240

ACA GGA GAA AAT GTC CTT GTG GAG ACC CTA AAC CAT GAA TTG TAC GAG GCT AAG TAC GTG K Y V T G E N V L V E T L N H E L Y E A

780 260

ATT AGC GCT OTT CCT CCT GTT CTG GGC ATG AAG ATT CAC TTC AAT CCC CCT CTG CCA ATG I S A V P P V L G M K I H F N P P L P M

840 280

ATG AGA AAT CAG CTG ATC ACT CGT GTG CCT TTG GGT TCA GTC ATC AAG AGT ATA GTT TAT MRNQLITRVPLGSVIKSIVY

900 300

TAT AAA GAG CCC TTC TGG AGA AAT ATG GAT TAC TGT GGA AGC ATG ATT ATT GA?+ GGA GAG YKEPFWRNMDYCGSMIIEGE

960 320

GAA GCT CCA GTT GCC TAT GCA TTG GAT GAT ACT AAA CCT GAT GGC AGC TAT CCT GCC ATA E A P V A Y A L D D T K P D G S Y P A I

1020 340

ATA GGA TTT ATC CTT GCC CAC AAA GCC AGA AAG CTG GCT CGT CTT ACC AAG GAG GAA AGG IGFILAHKARKLARLTKEER

1080 360

FIG. 2. The nucleotide and deduced amino acid sequences of bovine liver MAO B. The amino acid numbering starts at the amino-terminal residue, and the nucleotide sequence is shown above. The cysteine residue to which flavin adenine dinucleotide is covalently bound and the polyadenylation signal sequences are underlined. The stop codon is indicated by an asterisk.

134 TTG AA0

AAA

CREED W. ABELL AND SAU-WAH KWAN CTC

TGT GAC CTC TAT GCA AAG GTT CTG GGC TCA CAA

GAA GCT TTG CAC CCA

LKKLCDLYAKVLGSQEALHP

1140 380

GTG CAC TAT GAA GAG AA0 AAC TOG TGT GAG GAG CAG TAC TCC GGA GGC TGC TAC ACT TCC VHYEEKNWCEEQYSGGCYTS

1200 400

TAC TTC CCT CCT GGG ATC ATG ACT CAA TAT GGA AGG GTT CTA CGC CAG CCA GTG GGC AGG YFPPGIMTQYGRVLRQPVGR

1260 420

ATT TAC TTT GCA GGC ACA GAG ACT GCC ACA CAC TGG AGT GGC TAC ATG GAG GGG GCT GTG

1320 440

IYFAGTETATHWSGYMEGAV GAG GCT GGC GAG AGA GCG GCC CGA GAG ATC CTG CAT GCC ATG GGC AAG ATC CCA GAG GAT EAGERAAREILHAMGKIPED

1380 460

GAA ATC TGG CTG CCT GAA CCA GAG TCT GTG GAT GTC CCT GCG AAG CCC ATC ACC ACC ACC E I W L P E P E S V D V P A K P I T T T

1440 480

TTC TTG CAA AGG CAT TTG CCC TCC GTG CCA GGC CTG CTG AAG CTG ATT GGA TTG ACC ACC F L Q R H L P S V P G L L K L I G L T T

1500 500

ATC TTT TCA GCA ACT GCT CTC GGC TAC CTG GCC CAC AAG AGG GGG CTC CTG GTG CGA ATC IFSATALGYLAHKRGLLVRI

1560 520

TAA

1563

??

AGAGAGGGCA

TCTGTAACCA

CACCCTGGTG

TGTGGGTTTG

GGGGAAGGCA

TTGT-

GTTCCACAAA

1633

GATGCAAAGA

ATGTAGAGTG

AGGCGGCGAG

CATGATGATC

AGTCAGACTT

TCTGACCACA

GGATACACAG

1703

TCTCTTTCTC

CATTTTGACA

CCTGTGTATT

GTCTAGTACC

TAGCTTAGCA

CTGTCTCACC

CACTTCCAAG

1773

TTCACTGGCT

CCAGAATCTT

TACAGTAGTT

AAATTGGCTT

GTGAAAGGTC

CTTGCTATCC

TACTATACAT

1843

TGCCCAGGCA

CACACACACA

CACACACACA

CACACACACC

CCCACACACA

CACTACTTTT

TTCTTACCTC

1913

TATGGCTTTG

TGCTTGTCCT

TCCTCTTTCC

TGTAATGTCC

ACAACCTTCC

AGGTTCTCTG

CATTTGTCCT

1983

TAGAATCCCA

TATTGTTACA

GCTGGAAGAA

CTTAGACACC

ATCCAACCCT

TACTTTCTAT

TTTAGAGTTG

2053

AGCAAACTGA

AGCGGAGAGA

GGAGGAACTT

AATG‘XTCAG

TGTCCACAAT

AAGCCACTGA

TATTTTGGTG

2123

ACTAGGACAC

AGGTCCATTG

CTTTATCCCA

TCTCTCTGGA

TGGATTGCCC

AATCACCCTT

CTCTACTCCC

2193

TGCCAAGGTC

GCTGTGTTCC

CTTGGGTAGG

TTTACTCTGT

ACTAAGCTGT

TTTGTGCTGC

TCAGATGCTA

2263

CTACTCAGTA

TATATCCTTA

AGTCTTACCG

TCTTGCGCAG

TGTGCCTTCA

GCTCATTTTA

CTTTTTTTTC

2333

ATGGTAAGAG

TTCTTGTCTT

TTCTTCCTTT

TGTATCCTCC

ACTGAATCTG

GATACAAAGG

TTGGTGCACA

2403

TTTGGGTAAT

TCAA&TAU

AG-

AAAAAAAAAA

AAACTCGAGG

GGG

2466

G TTGATTGACC

FIG. 2. (continued).

MAO STRUCTURE

AND FUNCTION

135

(48, 49). The promoter activity was inhibited by upstream sequences (48). However, this negative cis regulation was not observed, but a putative initiator (Znr) element was detected (49). Furth er study showed that the presence or absence of Znr-like sequence in promoter constructs did not appear to affect significantly the negative cis regulation in the MAO A gene (50). For the MAO B gene, the promoter fragment consisted of two clusters of overlapping Spl sites [see Shih et al. (21) for a discussion of transcriptional regulation of MAO A and B]. Further study of these elements should increase our understanding of how MAO A and B are regulated and differentially expressed in catecholaminergic and serotonergic neurons, respectively. Knowledge of gene expression at the molecular level may ultimately lead to identification of new determinants (e.g., hormones, vitamins, calcium) that can alter MAO A and B activities in humans.

II. Functional Regions of MAO B A. Strategy for Identifying Crucial Amino Acid Residues The long-term goal in many laboratories is to determine the three-dimensional structure of MAO A and B in order to facilitate an understanding of the molecular and catalytic properties of these enzymes. Once the structures of MAO A and B are known, computer-assisted approaches can be applied to design novel drugs for the treatment of selected psychiatric and neurological disorders. Unfortunately, because MAO A and B are integral proteins of the outer mitochondrial membrane (22), they have resisted crystallization. An alternative approach, which is less definitive but more accessible, is to use site-directed mutagenesis (SDM) to help identify amino acid residues that reside in putative functional regions of these enzymes or to prepare and test hybrids from chimeric constructs of MAO A/MAO B cDNAs. Earlier work had shown that both MAO A and B are flavoproteins. They contain a pentapeptide with a cysteine residue that is covalently coupled to the required cofactor, FAD, through a thio ether linkage (51). Initially, it was thought that one FAD was covalently bound to an enzyme molecule (llO120 kDa) that is composed of two identical subunits (52). However, reexamination of the ratio showed one coupled FAD per subunit of 60 - 65 kDa (53). Moreover, bovine MAO B can exist as a dimer, but the enzyme preferentially functions as larger oligomeric complexes (54). A useful starting point is to examine data bases of soluble flavoproteins of known structures. Using this approach, a few short sequences (motifs) from 5 to 34 amino acid residues long were identified in several flavopro-

136

CREED

W. ABELL

AND SAU-WAH KWAN

teins, including MAO. Furthermore, structural studies of soluble flavoproteins have identified specific amino acid residues that likely interact with its riboflavin cofactor or substrate. To assess whether analogous interactions occur in MAO, mutants that encode variants of MAO proteins were prepared and transiently expressed in COS-7 cells, a mammalian cell line that permits expression and comparison of variants with the wild type (55). Type-specific antisera and monoclonal antibodies that recognize either MAO A or MAO B (27, 55) were used in Western blots to monitor the level of variant expression and showed that it was approximately the same level as the native enzyme. FAD covalent- and noncovalent-binding studies and catalytic activity assays were also performed in parallel on crude extracts of the enzyme. Because the expression of MAO B protein could be quantitated by enzyme-linked immunoassays (ELISAs) with a type-specific monoclonal antibody (e.g., MAO B-lC2) (27), it is possible to express catalytic activity as enzymatic activity (kmol/min/mg MAO B). Furthermore, results derived from one form of the enzyme (e.g., MAO B) can be used to gain insight into the other form (MAO A) because identical amino acid sequences are found in both enzymes in all the functional motifs identified thus far. By using site-directed mutagenesis (56, 57) to examine selected amino acids that are thought to play a role in function, several FAD-binding sites and a putative substrate-binding site were identified in MAO B. These include a dinucleotide-binding site (DBS), a second FAD-binding site (SBS), a fingerprint site (FPS), the FAD covalent-binding site (CBS), and an active site (ACT). The membrane-anchoring site (MEM) has also been identified (58). A schematic representation of our current knowledge of these functional regions in MAO B is shown in Fig. 3.

B. Dinucleotide-Binding

Site

The crystal structures of glutathione reductase, p-hydroxybenzoate hydroxylase, lipoamide dehydrogenase, and other flavoproteins contain a dinucleotide-binding motif (Br-o-BJ that interacts with FAD (36,59). This motif has a consensus sequence of Asp-Val-Val-Val-Ile-Gly-X-Gly-X-X-Gly-LeuX-X-Ala-X-X-Leu-X-X-X-X-X-X-Val-X-Val-Leu-Glu (60) (see Fig. 4). Highly conserved hydrophobic residues are located at positions 7-10, 17,20,23, 30,32, and 33 in MAO B. In contrast, a highly conserved hydrophilic residue is present at position 6 (Asp), and the motif ends with a glutamate residue at position 34. Within this motif, the Gly11-X-Gly13-X-X-Gly16 sequence in MAO B is identical to a sequence found in many flavoproteins (37), and it is likely to constitute a turn between the first B-sheet and the beginning of the a-helix. The y-carboxylate group of Glu-34 is thought to bind to the 2’-hydroxyribose of the AMP moiety of FAD to align this cofactor for participa-

MAO STRUCTURE

137

AND FUNCTION

H273

a

N I

0

b

d

C

I 40

T42a

I

I 60

120

160

200

240

280

320

e

f

I

I

I

360

400

440

I 460

c

I 620

Amino Acid Residue FIG. 3. Putative functional regions and active residues in human MAO B. (a) Dinucleotidebinding site; (b) second FAD-binding site; (c) fingerprint site; (d) covalent-binding site; (e) active site; (f) anchor to outer mitochondrial membrane.

tion in the oxidation-reduction cycle during the oxidative deamination of amines to their corresponding aldehydes. To test the postulated role of Glu-34 in dinucleotide binding, several mutant cDNAs of human MAO B were prepared and transiently expressed in COS-7 cells (55). Variants E34Q and E34A were devoid of activity, and the E34D variant was only 7% of wild-type activity (see Table I for a comparison of the enzymatic activities of all variants with their corresponding wild type). It was not unexpected that the E34A variant was inactive, because Ala has a more hydrophobic character than Glu and does not carry a negative charge. The Gln-34 residue in E34Q is closest to the wild-type Glu-34 in size, but as expected, this variant had no activity because it also lacks a negative charge. However, the dramatic decrease in activity in the variant E34D was not anticipated because Asp contains a negative charge and is similar in size to Glu. The low activity presumably is due to the shorter side chain of the Asp residue, resulting in a decrease of contact between its B-carboxylate side chain and the 2’-hydroxy group of the ribose in FAD. An additional highly conservative mutant (WOI) was also constructed and expressed (Table I). This mutant encoded a variant protein containing Ile-10 in place of Val-10, which is positioned at the end of the first B-sheet immediately before the Gly11-X-Gly13-X-X-Gly16 turn. Analysis of VlOI showed that it had only moderate activity (approximately 6500) compared to the wild type. These results demonstrate that even the most conservative substitutions of essential residues of the Br-c~-Ba motif had a pronounced effect on enzyme activity. This suggests that the pi-o-B2 motif has a precise conformation that accommodates FAD recognition and/or binding to this functional region (Fig. 4).

138

CREED

S

W. ABELL

AND

SAU-WAH

KWAN

ci”D

G

H

L

L L K

N A

A

V

A <% V

M G V

S I

L I

;i,

FIG. 4. Model of FAD interaction with three distinct sites in MAO B. Cys-397 forms a covalent linkage to the FAD via the 8a-methyl group of the isoalloxazine ring. Glu-34 is part of the 8r-o-f3a motif (residues 6-34) and forms hydrogen bonds to the 2’-hydroxyl of ribose in the AMP moiety of FAD. Arg-42, Tyr-44, and Thr-45 are part of a third binding site of MAO B. Arg42 forms hydrogen bonds to the diphosphate moiety of FAD. Tyr-44 forms an aromatic-aromatic interaction with the isoalloxazine ring of FAD. Thr-45 forms hydrogen bonds to the O-4 and/or N-5 of the isoalloxazine ring of FAD. Reprinted with permission from Ref. (73). Copyright 1998 American Chemical Society.

L46 (WT) L46V

G226 (WT) G226A

D227 (WT) D227E D227N D227A

-CH-CH,-OH

T45 (WT) T45A T45S

-CH,

-H

-CH,-COOH -CH,-CH,-COOH -CH,-CONH, -CH,

-CH, -CH,-OH

-(CH,),-NH=C(NH,), -CH, -(CH),-NH,

R42 (WT) R42A R42K

100 33

100 91 70 35

100 ND 12

100 ND 14

100 83

-CH,-CH(CH,), -CH(CH,),

100 93 2 1

VlO W) VlOI -CH,-phenyl-OH -CH,-phenyl -CH,-OH -CH,

100 65

-CH(CH,)CH, -CH(CH,)-CH,-CH,

Y44 (wq Y44F Y44S Y44A

100 7 ND ND

-CH,-CH,-COO-CH,--COOP -CH,-&-CONH, -CH,

Side chain

E34 PW E34D E34Q E34A

SDM

100 89

100 92 89 88

100 28

100 91 67 29

_ _

_ _ -

0

5

-

108 _

-

100

2

_

5

_ 88 -

12 -

100

100

15 -

Covalent FAD binding (O/o of wild type)

100

Total Enzymatic activity FAD binding (O/o of wild type) (O/o of wild type)

64

64, 72, 73

55, 64

Ref.

URadiometic assayswere prepared on equal amounts of wild-type (WIJ and MAO B variants.Tbr results are expressed as percentages of enzymatic activities(~mol/min/mg MAO B) and [‘%]FAD incorporation.SDM, Site-directed mutagenesis; ND, not detectable.

Fingerprintsite

SecondFAD-binding site

Dinucleotide-bindingsite

Functionalregions

TABLE I

COMPARISON OF ENZYMATIC AND FAD-BINDING ACTIVITIES OF VARIANTS WITH WILD TYPES

140

CREED

W. ABELL

AND SAU-WAH

KWAN

Two hybrids of MAO A and B encoded by chimeric cDNAs were prepared and tested for substrate and inhibitor specificity (61). The MAO A hybrid contained the N-terminus 36 amino acids of MAO B, and the MAO B hybrid contained the N-terminus 45 amino acids of MAO A. No significant differences in catalytic properties or sensitivity to inhibitors were found in these hybrids compared to the wild-type MAO A and B enzymes. Similar results were obtained with hybrids at the N terminus, but substitutions at the C terminus of MAO B with MAO A gave inactive enzyme (62). The comparable switch with a MAO B terminus had no effect on MAO A activity. These studies indicate that functional sequences (e.g., the B-o-B motif) in the N-terminal region that contribute to catalytic activity are essentially the same in both the A and the B forms of the enzyme. Flavinylation of MAO B (covalent binding of FAD) has been difficult to study because FAD that is bound to Cys-39 7 cannot be removed without denaturing the apoenzyme. The usual approach has been to study f lavinylation in animals that are fed riboflavin-free diets to deplete endogenous levels of this cofactor, followed by addition of FAD to regenerate activity. We have developed a convenient and rapid method (63) to study flavinylation by growing COS-7 cells in a reduced riboflavin medium (Rib- COS-7 cells). The levels of FAD in these cells are too low to accommodate MAO B modification, but are sufficient for cell growth. Thus, the role played by Glu-34 in the binding of FAD to enzyme can be assessed by transfecting wild-type or mutant MAO B cDNAs and [r4C]FAD into Rib- COS-7 cells. Fluorography results showed that [14C]FAD was incorporated into wild-type MAO B, but the absence of bands in variants E34A and E34Q indicated little or no [14C]FAD coupling to the enzyme. A faint band was observed with variant E34D, showing 10% of [i4C]FAD incorporation into the enzyme compared to the wild type. All variants of MAO B at residue 34 exhibited either a dramatic decrease or total loss of [14C]FAD flavinylation and a corresponding loss of enzymatic activity. In order to assess the noncovalent binding of FAD to the enzyme, dotblot assays were performed (64). In this procedure, immunoprecipitated wild-type and variant MAO B enzymes were loaded onto nitrocellulose membranes and washed with buffer to remove nonspecific bound [i4C]FAD. The total amount of [i4C]FAD incorporated into MAO B included both covalently linked FAD and noncovalently bound FAD. Because bound FAD could be measured independently by fluorography, the level of noncovalently bound FAD could be obtained by difference. Dot-blot analysis of E34D showed that this variant lost its ability to bind FAD noncovalently and correlated closely with low levels of flavinylation and enzymatic activity. Thus, Glu-34 appears to be a critical residue that is required for the initial binding of FAD,

MAO STRUCTURE

AND FUNCTION

141

followed by additional steps (discussed below) that lead to the activation of MAO B through f lavinylation.

C. Second

FAD-Binding

Site

Another putative FAD-binding region in MAO B, located just 8 through 11 residues downstream from Glu-34, was identified by sequence comparisons with other flavoproteins, including spinach ferredoxin-NADP+ reductase (FNR) (65), corn NADH-nitrate reductase (66), Pseudomonas cepacia phthalate dioxygenase reductase (PDR) (67), NADH-cytochrome b5 reductase (68), NADPH-cytochrome P450 reductase (69), and NADPHsulfite reductase (70). A highly conserved flavin-binding sequence consisting of four residues, RXY(T,S), was found in these flavoproteins. This sequence, found in eight proteins, contained two strictly conserved residues (Arg-42 and Tyr-44 in MAO B) and a third residue at position 45 that was serine in half of the proteins and threonine in the other half. The crystal structures of three of these enzymes, FNR, PDR, and nitrate reductase, have been solved and their flavin-binding domains characterized (65- 67, 71). Two of the flavin-dependent enzymes, FNR and nitrate reductase, require noncovalently bound FAD, whereas PDR uses noncovalently bound FMN. These three proteins have similar domain structures consisting of a C-terminal pyridine nucleotide-binding domain and an N-terminal flavin-binding region composed of a six-stranded antiparallel p-barrel. The isoalloxazine ring of the noncovalently bound riboflavin is oriented between the two domains, and the conserved Arg, Tyr, and SerThr residues (corresponding to human MAO B residues Arg-42, Tyr-44, and Thr-45) are contained in the fourth strand of the R-barrel. The conserved Arg-42 hydrogen bonds to the phosphate moiety in the riboflavin, and the conserved Tyr-44 makes van der Waals contacts with the isoalloxazine ring of the riboflavin through its aromatic ring. The hydroxyl group of tyrosine forms a hydrogen bond with the ribityl4’-hydroxyl group of the riboflavin. The conserved Ser/ Thr forms hydrogen bonds to the redox-active N-5 hydrogen and/or O-4 of the flavin (6% 67, 71). Based on these observations, the roles played by Arg-42, Tyr-44, and Thr45 in MAO B were examined to determine their effect on the levels of flavinylation and catalytic activity (see Fig. 4). The role of Tyr-44 will be discussed first (72). Tyr-44 was proposed to interact with FAD in a conformation analogous to that observed for the corresponding Tyr-95 of crystallized FNR. Mutants that encoded serine or alanine in place of Tyr-44 produced variant proteins essentially devoid of catalytic activity, whereas replacement of tyrosine with phenylalanine gave a variant with 93% of the wild-type activity (Table I). Further characterization of these variants with respect to their ability to

142

CREED

W. ABELL

AND SAU-WAH

KWAN

covalently bind FAD showed that losses of catalytic activity in the variants correlate closely with decreased flavinylation. This suggests that the noncovalent interactions of Tyr-44 with FAD are a prerequisite for subsequent covalent coupling of this flavin. Also, these results support the interpretation that the hydroxyl group of Tjr-44 in MAO B does not play a role in noncovalent FAD binding. This observation is in contrast to studies of FNR, PDR, and nitrate reductase, in which the hydroxyl moiety of the corresponding Tyr is found to form a hydrogen bond with the 2’-hydroxyl of the ribityl moiety (65-67, 71). Using SDM, the arginine at position 42 was substituted with alanine or lysine, and the threonine at position 45 was substituted with alanine or serine (see Fig. 4). Substitution of an alanine at either position resulted in proteins devoid of activity, whereas substitution of Arg-42 to Lys and Thr-45 to Ser gave variants that retained partial activity (Table I) (73). The ability of these variants to bind FAD covalently was also examined in COS-7 cells depleted of riboflavin. Although wild-type MAO B expressed in Rib- COS-7 cells is devoid of activity due to the absence of covalently bound FAD (72), the low levels of riboflavin in these cells do not affect the level of expression or insertion of MAO B into mitochondria. Covalent FAD binding and generation of enzymatic activity can be achieved by introducing FAD along with MAO B cDNA during electroporation. The FAD precursors, riboflavin and FMN, are equally as effective as FAD in recovering MAO B activity, presumably because endogenous FAD synthetase can rapidly convert these forms to FAD. These studies also suggest that the conversion of riboflavin to FAD occurs prior to FAD binding to MAO B. Thus, FAD is the form of flavin that initially binds to MAO B. As discussed above, mutant and wild-type MAO B cDNAs can be electroporated into riboflavin-depleted COS-7 cells along with [14C]riboflavin. MAO B expressed in such cells can be immunoprecipitated with MAO B-specific polyclonal antibody, run on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and subjected to fluorography. Because these proteins are submitted to denaturing conditions prior to fluorography, only proteins with covalently bound [14C]FAD will produce a 58-kDa band on the fluorogram. The results of these fluorography studies on the R42A and R42K variants show that the amount of covalent coupling was substantially less than that in the wild-type and correlated closely with the level of enzymatic activity (73). The results of these studies are consistent with the proposed role for Arg42 in which the arginine side chain forms hydrogen bonds with the diphosphate of FAD. Th e arginine side chain has three amino groups, designated E, ql, and q2, which can potentially act as hydrogen bond donors. A complete

MAO STRUCTURE

AND FUNCTION

143

loss of hydrogen bond donors of these side chain amino groups should result in a total loss of FAD incorporation and MAO B enzymatic activity, as is observed for the R42A variant. Alternatively, an amino acid side chain containing only one hydrogen bond donor might result in partial retention of FAD incorporation and MAO B activity, an effect seen by the arginine-to-lysine substitution in the R42K variant (73). From the X-ray crystallographic structures of FNR, PDR, and nitrate reductase (65-67, 71), the arginine residue contained in the consensus sequence RxY(T,S) is thought to form hydrogen bonds from the E and/or q side chain amino groups to the phosphate(s) of the flavin. Based on sequence comparisons with the structures of the crystallized proteins, it seems likely that Arg-42 of MAO B also forms one or more hydrogen bonds to the diphosphate moiety of FAD (Fig. 4). Thus, formation of these noncovalent bonds to FAD appears to be required for subsequent covalent FAD coupling to MAO B and generation of enzymatic activity. The role of Thr-45 in MAO B was also examined in greater detail (73). The T45A variant had no catalytic activity, whereas a more conservative amino acid substitution (Thr to Ser) had 12% of wild-type activity (Table I). The results of the covalent FAD incorporation assays indicate that the loss in catalytic activity is due to an inability of these variants to incorporate FAD covalently. The threonine side chain appears to form noncovalent interactions with FAD that are required for subsequent covalent FAD incorporation into MAO B. The a-hydroxyl of threonine can potentially act as a hydrogen bond donor in such noncovalent interactions. Removal of the hydroxyl group could result in a reduction of FAD incorporation and enzymatic activity. Retention of the y-hydroxyl group, however, should yield at least partial enzymatic activity and FAD binding, as was observed with the T45S variant. Crystallographic studies of FNR and PDR suggest that the conserved Ser is involved in protonation of the N-5 nitrogen of the isoalloxazine ring in the reduction of the flavin. Comparisons of the oxidized and reduced structures of FNR and PDR have suggested that the side chains of Ser-96 of FNR and Ser-58 of PDR rotate into position to hydrogen bond to the N-5 nitrogen of the reduced flavin. Thus, one role proposed for the conserved Ser is to protonate the N-5 nitrogen of the f lavin (71, 74). Mutagenesis studies support the interpretation that Thr-45 of MAO B can protonate the redox-active N-5 nitrogen during the reduction of FAD. A Thr45 + Ala substitution would be expected to cause a decrease in enzymatic activity. This substitution results in the complete loss of activity (Table I) and covalent FAD binding. Although these results indicate that Thr-45 is critical for covalent FAD incorporation, they do not exclude the possibility that this residue also plays some role in the reduction of FAD during oxidation of the

144

CREED

W. ABELL

AND SAU-WAH KWAh’

substrate. Thus, one rational hypothesis is that the hydroxyl of Thr-45 can form hydrogen bonds to the O-4 and/or the N-5 of the isoalloxazine ring of FAD (Fig. 4).

D. Fingerprint Site MAO B contains another FAD-binding site that was identified by sequence comparisons to other flavoproteins. This fingerprint site was first found in the oxidoreductase family (75), and has been studied in flavoproteins with known structures (76). This fingerprint region consists of a Bstrand followed by an invariant glycine residue that folds into a sharp turn to extend the next invariant aspartate residue into close proximity to the ribityl chain of FAD. The B-car-bony1 group of this highly conserved aspartate forms a hydrogen bond with the 3’-hydroxyl group of the ribityl chain of FAD (77). To test the hypothesis that these two residues are involved in FAD binding in human MAO B, Gly-226 was changed to alanine (G226A), and Asp227 to glutamate (D227E), asparagine (D227N), or alanine (D227A) by SDM, and the cDNAs were transiently expressed in COS-7 cells (Table I) (64). The effects of these substitutions were assessed by enzymatic activity assays, quantitation of MAO B expression, [14C]FAD covalent coupling, and [i4C]FAD noncovalent binding. The D22 7E variant retained almost full enzymatic activity (9 lo/a),probably because a carboxyl group is retained in the glutamate residue. The length of the side chain does not appear to be critical. However, a moderate decrease in enzymatic activity was observed for D227N (70% of wild type). This was expected because the replacement of aspartate by asparagine would lead to weaker hydrogen bonding between the B-carbonyl group of asparagine and the 3’-hydroxyl group of the ribityl moiety of FAD. Replacement of the side chain possessing a B-carboxyl with a methyl group (D227A) resulted in a substantial reduction (35% of wild type), but not total loss, in enzymatic activity (Table I). A G226A variant exhibited a comparable loss of activity (33% of wild type). One possible explanation for the partial reduction in activity is that the bulkier side chain of alanine may sterically hinder conformational changes that are required for a sharp turn at position 226. Nevertheless, these results can be interpreted to indicate that both the sharp turn facilitated by Gly-226 and the hydrogen bonding between the B-carbonyl group of Asp22 7 and the 3’-hydroxyl group of the ribityl chain are necessary for generating maximal enzymatic activity of MAO B. Studies on covalent flavinylation of these variant enzymes using quantitative f luorography demonstrated that variant D22 7E incorporated covalently bound [14C]FAD to approximately the same extent (91%) as wild-type MAO B (Table I). The D227N variant retained a moderate amount of covalently bound cofactor (67% of wild type), but D227A and G226A incorpo-

MAO STRUCTURE

AND FUNCTION

145

rated much less covalently bound [i4C]FAD (29 and 28% of wild-type MAO B, respectively). These results are in agreement with the values for enzymatic activities found for all the variants studied, and support the conclusion that both Gly-226 and Asp-227 play essential roles in the flavinylation and activation of MAO B. The total FAD incorporation as measured by the intensity of staining in dot-blot assays corresponding to wild-type and variant enzymes (G226A, D22 7E, D227N, and D227A) were on average approximately 90% of wildtype MAO B, whereas nontransfected COS-7 cells show no dot, indicating the absence of [14C]FAD. These studies indicate that mutations at positions 226 and 227 do not significantly alter the ability of the apoenzyme to bind [14C]FAD noncovalently. Together, the results of the fluorographic and dotblot analyses indicate that the variant enzymes at positions 226 and 227 retain their ability to recruit FAD into the apoenzyme, but they exhibit reduced covalent linkage of FAD to the enzyme compared to the wild type. Presumably, achievement of noncovalent binding of FAD does not assure that the flavin will be delivered in the correct orientation for the next step in forming the holoenzyme. In all variants studied thus far, enzymatic activities correlate closely with covalently bound [14C]FAD, but not to the total amount of bound [i4C]FAD. This suggests that only MAO B variants with covalently bound FAD are catalytically active. These results are in agreement with expression studies of mutants by others (61, 78) who found that substitution of the FAD covalent-binding residue (Cys-39 7) in MAO B to alanine, serine, or histidine abolished catalytic activity. In contrast, Ito and co-workers (79, 80) found that when the FAD covalent-binding residues (i.e., Cys-406 in MAO A and Cys-3 9 7 in MAO B) were substituted with alanine, these variants temporarily retained some activity when expressed in yeast. One possible explanation is that FAD formed a transient complex with variants C406A and C39 7A at the covalent-binding site without becoming covalently coupled to the enzyme. As discussed above, Glu-34 and Tyr-44 are thought to bind FAD initially through noncovalent interactions (55, 72). Direct evidence for this interpretation was obtained using dot-blot analysis (64). Quantitation of the fluorograms of SDS-PAGE and dot-blot autoradiograms showed that the intensity of the dots correlated well with the intensity of the bands on the fluorogram. Furthermore, the total FAD binding (indicated by the intensity of the dot) and covalent binding of FAD (indicated by the band intensity of these variant enzymes) also correlated well with their enzymatic activity (Table I). Unlike the variants of Gly-226 and Asp-227, discussed above, the E34A and Y44A variants not only lost their ability to bind FAD covalently, but they also were unable to recruit FAD noncovalently. In contrast, the two conservative variants, Y44F and E34D (to a much lesser extent), retained partial ability to

146

CREED

W. ABELL

AND SAU-WAH

KWAN

bind FAD covalently, and the extent of covalent FAD binding correlated with their total FAD binding. These studies indicate that Glu-34 and Tyr-44 in the N-terminus of the molecule facilitate the initial noncovalent binding of FAD to position it in the correct orientation for subsequent steps that lead to covalent binding. Alteration of either one of these sites of interaction affects MAO B enzymatic activity. The expression of wild-type and variant MAO B enzymes at the fingerprint site was also analyzed by Western blot analysis using MAO B antisera (64). Wild type and G226A had bands of approximately equal intensity at a molecular mass of 59 kDa. However, none of the variants at position 22 7 was recognized by the MAO B-specific monoclonal antibody, MAO B-lC2. Reduced synthesis of these variants was ruled out, because the enzymatic activity assays, ELISA quantitation, and fluorography all showed the expected level of expression of these variant proteins. The inability of MAO B-lC2 to recognize the enzymes with substitutions at position 22 7 suggests that Asp227 may constitute part of the antigenic determinant. If this interpretation is correct, the recognition site appears to be uniquely specific because mutagenesis of the adjacent residue (G226A) had no effect on recognition. Modification of the apoenzyme by flavinylation is obligatory for the generation of catalytic activity in many flavoproteins. Some examples are succinate dehydrogenase (81), dimethylglycine dehydrogenase (82), 6-hydroxyD-nicotine oxidase (83), trimethylamine dehydrogenase (84), p-cresol methylhydroxylase (85), and MAO A and B (63). Flavinylation of some of these proteins is thought to be achieved by an autocatalytic process rather than catalysis by an as-yet unknown enzyme (72, 83-87). However, the precise steps involved are presently unknown. Furthermore, whether flavinylation of apoflavoproteins is a cotranslational or posttranslational process remains to be definitively established. It was proposed (87) that the covalent coupling of FAD to 6-hydroxy-Dnicotine oxidase occurs only at certain conformational states along the folding pathway. Succinate dehydrogenase, with its C terminus truncated (minus 70 or 90 amino acid residues), cannot bind FAD covalently, although the FAD covalent linkage site is close to the N-terminal end and not the C-terminal end of this enzyme (81). One interpretation of this observation is that FAD linkage occurs only after aposuccinate dehydrogenase has folded into certain conformations that can bind FAD. Studies on flavinylation of p-cresol methylhydroxylase (85) a1so suggest a sequential process for flavinylation of this tetrameric enzyme, which is composed of two (Yand two B subunits. In this folding pathway, FAD binds first to the apoflavoprotein subunit noncovalently, followed by interaction between the flavoprotein and the cytochrome subunits. The participation of the cytochrome subunit is required for eventual covalent attachment of FAD to the flavoprotein subunit.

MAO STRUCTURE

AND FUNCTION

147

Sequential steps in flavinylation have also been proposed for flavoproteins that do not covalently bind FAD. For example, the reconstitution of catalytically active D-amino acid oxidase is a two-stage process (88). Mixing FAD with the apo-D-amino acid oxidase led to the reconstitution of the holoenzyme, including a rapid binding of FAD by the apoenzyme followed by a slower change of the holoenzyme conformation. The authors concluded that the slow conformational change is required for the generation of enzyme activity. These studies support the notion that flavin binding depends on the existence of conformational stages in the folding pathway of catalytically active flavoproteins. In analogous reactions, FAD appears to couple to apo-MAO B in a stepwise process during the translation and folding of the polypeptide. The first step involves recruitment of FAD. The y-carboxyl group of Glu-34 (at the di nucleotide-binding site) in the N terminus provides the initial topological dock for FAD binding (55, 64, 72, 73). The Arg, Tyr, and Thr binding site, which is located immediately adjacent to the Glu site, then participates in binding. This interpretation can explain why mutations at either one of these functional sites result in a loss of both covalent and noncovalent linkage of FAD. FAD is then delivered to the fingerprint site (residues 222-227), which provides another topological dock to further secure the FAD molecule. The incoming FAD, which is held at multiple sites through interaction with several amino acid residues, can then be delivered to Cys-39 7 in a position that places the 8cr-methyl group of FAD in exact and close proximity to the thiol group of Cys39 7 to facilitate covalent f lavinylation. Variants at Gly-226 and Asp-227 in the fingerprint site do not interfere with noncovalent binding of FAD, possibly because they are not involved in FAD recruitment, but they have reduced capability to couple FAD covalently, possibly because the conformations of these variant proteins do not support proper positioning of Cys397 for flavin linkage (64).

E. Covalent-Binding

Site

The role of the nine Cys residues in both MAO A and B (see Fig. 1) was investigated (78). Each Cys residue was individually converted to a Ser residue, and the activities of the variants compared to the appropriate wild type. Single substitutions of two Cys residues in MAO A (Cys374 and Cys406) and three in MAO B (Cys-156, Cys-365, and Cys-397) resulted in total loss of catalytic activity. Because FAD is covalently linked to Cys-406 in MAO A and Cys-397 in MAO B, substitutions at these positions would be expected to abolish activity. The other Cys residues (374 in MAO A, and 156 and 365 in MAO B) may be constituents of the active site. Cys-365 was identified as the recipient residue in MAO inactivation by cyclopropylamines (89). These compounds are mechanism-based inhibitors of MAO that do not

148

CREED

W. ABELL

AND SAU-WAH

KWAN

react with the covalently bound FAD but inactivate the enzyme by binding to Cys-365. Furthermore, recent evidence indicated the presence of a redoxactive disulfide at the catalytically active center (90). Collectively, these studies strongly imply that four Cys residues participate in cofactor or substrate binding. Recent studies on the covalent-binding site have examined the possible participation of the two tyrosines (Tyr-393 and Tyr-398) that flank the covalent FAD-binding residue (Cys-397) in MAO B (B. Wu, B. P Zhou, S.-W. Kwan, and C. W. Abel1 unpublished observations). This region is analogous to the one in flavoproteins that contain a covalent histidyl (N3)-So-FAD linkage (91). Both Tyr-393 and Tyr-398 in MAO B were found to be required for flavinylation and generation of catalytic activity. Examination of a three-dimensional model of MAO A (92) predicts that two antiparallel B strands, which contain the tyrosine residues, flank the covalent cysteinyl S-~CX-FAD linkage. Confirmation of this hypothesized model awaits structural information. F.

Active Site

Comparisons of motifs that occur in a wide variety of f lavoproteins with those in MAO A and B have been useful for identifying amino acid residues that interact with FAD. Examination of motifs in MAO with amino acid sequences and function similar to those in P450 enzymes has also been productive. The deduced amino acid sequence of bovine MAO A was determined and a short segment (eight residues) was identified that exhibited high sequence identity to a region in cytochrome P450 in rat tissue (31). Comparison of this segment in MAO B with that in P450 (93) has led to the identification of three amino acids (Phe-423, Glu-427, and Thr-428) in MAO B that may be part of a central active site helix found in many cytochrome P450 proteins (D. A. Lewis and C. W. Abell, unpublished observations). Other regions of functional significance were identified by chimeric constructs that encode MAO A/B. This approach has the advantage of finding regions that play a role in determining substrate and/or inhibitor specificity. However, loss of activity or failure to change specificity in the hybrid enzymes cannot be readily interpreted, because some combinations are not conformationally compatible. Such hybrids may represent false negatives. Nevertheless, 18 MAO A/B hybrids were prepared and examined by progressively moving the junction of the N terminus of one form with the C terminus of the other (94). Two sequences (residues 62-103 and 146-220) were identified that represent functional regions in MAO B. Using a similar approach (95), replacement of residues 16 l-3 75 of MAO A with the corresponding MAO B segment was found to shift the substrate and inhibitor selectivity of MAO A to that of MAO B.

MAO STRUCTURE

AND FUNCTION

149

Other SDM studies identified a single amino acid residue that is responsible for MAO A or B substrate specificity (96). Substitution of Phe-208 in rat MAO A with Ile at the same site in MAO B converted the A-type substrate specificity to that of MAO B. Likewise, substitution of Ile with Phe in MAO B switched substrate specificity to that of MAO A. Moreover, Phe could be replaced with Tyr for A-type substrate specificity, whereas Ile could be replaced with Val and Ala for B-type specificity. This result suggests that aromatic or hydrophobic residues at this unique site on these enzymes can determine substrate selectivity.

G. Other Potential Targets Earlier work suggested that a His residue in MAO may be involved in generating catalytic activity (97,98). MAO A and B both contain 12 His residues. However, only 3 of these (His-178, His-273, and His-392 for MAO B) are found at the same site in both MAO A and B in the human, bovine, and rat species (see Figs. 1 and 3). It is likely that one of these highly conserved residues could contribute to catalyzing the oxidative deamination of amines. In fact, substitution of His-382 in MAO B with Arg greatly reduced catalytic activity (98). Other studies (90) have uncovered another redox-active group in addition to FAD at the catalytic center. Reductive titration studies of both MAO A and B showed that four electron equivalents were required to reduce the enzyme, indicating the presence of a redox-active disulfide in the active site. However, the physical location of this disulfide in these enzymes remains to be determined.

III. Tissue and Cell Distribution The availability of monoclonal antibodies that can distinguish MAO A from B (27, 28, 99, 100) permit localization of these enzymes to be determined in peripheral tissues and in the central nervous system. Immunocytochemical studies performed in human platelets, lymphocytes, placenta, and liver revealed a pattern of distribution that followed catalytic activity (101). MAO A was found in the syncytiotrophoblast layer of term placenta and in liver, but was not observed in platelets or lymphocytes. In contrast, MAO B was observed in platelets, lymphocytes, and liver, but was found to be very low or absent in placenta (102). Th ese results illustrate the differential expression of the two forms of MAO and suggest that they have independent roles in neurotransmitter and amine metabolism. Immunocytochemical localization of MAO A and B in primate brain (monkey and human) showed that they were distributed in distinct neurons

150

CREED

W. ABELL

AND SAU-WAH

KWAN

(discussed below) with diverse physiological functions (25,26). These aminecontaining cell groups of neurons were previously established in studies of the mouse (103). Some staining for enzyme was also observed in glial cells, particularly MAO B, and in terminal elements of neurons. In both species, the distribution of MAO A-containing neurons correlated closely with the sites of localization of catecholamines, and the distribution of MAO B-containing neurons was essentially identical to that in cells that synthesize and store serotonin. The most densely stained neuronal group for MAO A was found to be the nucleus locus coeruleus, a region in brain that contains noradrenergic neurons that synthesize norepinephrine. However, moderately or lightly stained MAO A-positive cells were also seen in all other regions that contain catecholaminergic neurons. Surprisingly, the extent of immunoreactivity for MAO A in dopaminergic neurons in the substantia nigra was less than expected with respect to the number of cells stained. Ultrastructural studies of MAO A revealed that most staining was associated with the outer membrane of mitochondria within cell bodies, and in axonal terminals (104). Some staining for MAO A was also observed in the rough endoplasmic reticulum in the cell bodies. This distribution was expected because MAO A is encoded by a nuclear, not a mitochondrial, gene. MAO B-containing neurons were found primarily in regions that are known to synthesize and store serotonin. The most prominent location of serotonergic neurons was known to be the nucleus raphe dorsalis. Also, neurons in the posterior hypothalamus stained for MAO B, but stained to a lesser extent for MAO A. Most glial staining was observed in regions known to contain neuronal terminals, but MAO B-containing astrocytes were often found around blood vessels (25, 26). In rat studies, MAO A and MAO B mRNAs were detected in aminergic and nonaminergic cells of the central nervous system by in situ hybridization (105). The highest level of MAO A expression was seen in noradrenergic neurons, whereas moderate and low levels were found in serotonergic neurons and dopaminergic neurons, respectively. MAO B expression was most abundant in serotonergic and histaminergic neurons, as well as in glial cells. In nonaminergic cell populations, MAO A gave a much stronger transcript signal compared to MAO B.

IV. Physiological Functions The distinct patterns of distribution observed for MAO A and B in cells and tissues provide a rational prediction of the function of these enzymes in

MAO STRUCTURE

AND FUNCTION

151

peripheral tissues and in brain. For example, MAO A in placenta may lower the transfer of biogenic or bioactive amines across the placenta, which would provide preferential selection or protection against harmful amines to the fetus. Likewise, MAO B could provide tissue- or cell-specific protection against xenobiotics that are derived from dietary or environmental sources. It is well established from studies in the central nervous system of humans, including healthy controls and those deficient in MAO A and/or MAO B, that these enzymes control, in part, neurotransmitter levels in brain. However, the distribution of the A and B forms in catecholaminergic and serotonergic neurons, respectively, was unexpected. MAO B in serotonergic neurons does not oxidatively deaminate serotonin, unless the concentration of this amine is elevated to very high levels. Under normal conditions, serotonin would be taken up into synaptic vesicles to be released during neurotransmission. Norepinephrine is metabolized by MAO A (K, = 3 7 1 tJ4) for human brain (106) at moderate rates, but uptake of norepinephrine into synaptic vesicles is preferentially favored over degradation by MAO A because the Km values for these systems operate at micromolar levels. Other studies (107) indicated that dopamine can be metabolized by MAO B, which is expressed in serotonergic, but not dopaminergic neurons. Collectively, these findings strongly suggest that one major role of MAO A and B is to reduce or eliminate amines that gain entry into neurons, where they have no legitimate physiological function. A likely role for MAO B, which prefers hydrophobic substrates such as phenylethylamine and benzylamine, is to provide protection by catalyzing the oxidation of xenobiotics that are derived from dietary or environmental sources. However, one such foreign substance, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), has proved to be a neurotoxin that can cause a parkinsonian-like disorder in humans and nonhuman primates (108) by destroying dopaminergic neurons in the substantia nigra (109). MPTP is oxidized by MAO B in glial cells and possibly serotonergic neurons to 1-methyl-4-phenyl-2,3_dihydropyridinium ion (MPDP+), which is unstable and undergoes further oxidation (probably spontaneously) to a highly reactive toxic species, 1-methyl-4-phenylpyridiniurn ion (MPP+) (2, 110). MPP + is actively taken up into brain synaptosomes (111) by th e d op amine uptake transporter and possibly exerts its cytotoxicity by inhibiting NADH dehydrogenase in mitochondria (112). MAO B, and not MAO A, is clearly linked to the generation of MPTP toxicity in uiuo. Pretreatment of animals with MAO B inhibitors (e.g., deprenyl, pargyline) (113, 114) protects against neurotoxicity, whereas pretreatment with an MAO A inhibitor (clorgyline) (112) has no beneficial effect. MPTP administration to mice deficient in MAO B showed no reduction in dopamine levels in the striatum of mice (115). Collectively, these studies led to treatment of early-stage PD with deprenyl(5, 116).

152

CREED

W. ABELL

AND SAU-WAH KWAN

V. Future Perspectives Site-directed mutagenesis studies have greatly enlightened our understanding of cofactor and substrate binding to specific amino acid residues and have provided targets for the development of new drugs (inhibitors of MAO A and B) for treatment of depression and PD. However, without the crystal structures of these mitochondrial membrane proteins, drug design targeted to specific functional domains in these molecules is limited. Thus, the next major goal is to determine the three-dimensional structures of MAO A and B. With regard to the physiological functions of MAO A and B, the study of knockout mice that are deficient in MAO A or B should provide important models for determining the role of specific neurotransmitters that interact to produce specific behaviors. The elevated levels of serotonin in MAO A knockout pups can be modified pharmacologically to achieve a better understanding of the role of serotonin in aggressive behavior (21). Finally, it is of interest to note that one of the specific mechanism-based inhibitors of MAO B, deprenyl, exhibits remarkable effects on selected physiological activities in rodents and/or humans, including protection of neurons from age-related decay and increase in life-span by about 25% (117-119). How deprenyl, through its interaction with MAO B, elucidates responses in these diverse biological pathways is currently unknown. However, attempts to detect determinants of these physiological parameters can be pursued using strategies (e.g., two-hybrid screens or computational methods) for identification of interacting proteins (120). Understanding how these pathways can be modulated represents the potential to improve significantly the quality and quantity of life.

ACKNOWLEDGMENTS This work was supported in part by the Foundation for Research (Carson City, Nevada) and by a Granville Wrather fellowship (to CWA) from the I@ Institute (Austin, Texas). We thank Grace Kubin and Oi-Yin Lee for excellent technical assistance on the bovine MAO B sequencing. We also thank Tommy Tjiptadjaja and Olivia Widjaja for excellent assistance in preparing this manuscript.

REFERENCES I. R. W. Von Korff, Monoamine oxidase: Unanswered questions, In “Monoamine Oxidase: Structure, Function, and Altered Functions” (T. P. Singer, R. W. Von Korff, and D. L. Murphy, eds.), pp. l-7. Academic Press, London, 1979. 2. K. Chiba, A. Trevor, and N. Castagnoli, Jr., Biochem. Biophys. Res. Commun. 120, 574578 (1984).

MAO STRUCTURE

AND FUNCTION

3. M. Da Prada, R. Kettler, H. H. Keller, W. P. Burkard, D. Muggli-Maniglio, 4. 5. 6. 7. 8. 9. 10. II. 12. 13.

14. 15.

16.

17. 18. 19. 20. 21. 22. 23.

24. 25. 26. 27. 28. 29. 30.

153 and W. E. Hae-

fely,]. Phurmacol. Exp. Ther. 248,400-414 (1989). W. Birkmayer, P Riederer, M. B. H. Youdim, and W. Linauer, J. Neural Trunsm. 36,303 326 (1975). J. W. Tetrud and J. W. Langston, Science 245,519-522 (1989). P. A. LeWitt, Actu Neural. Scnnd. 136, 79-86 (1991). H. G. Brunner, M. Nelen, X. 0. Breakefield, H. H. Ropers, and B. A. van Oost, Science 262, 578-580 (1993). 0. Cases, T Vitalis, I. Seif, E. De Maeyer, C. Sotelo, and P Gaspar, Neuron 16,297-307 (1996). R. Hen, Neuron 16,17-21 (1996). D. L. Murphy, K. Sims, G. Eisenhofer, B. D. Greenberg, T George, F. Berlin, A. Zametkin, M. Ernst, and X. 0. Breakefield,Z. Neural Transm. 52, 29-38 (1998). J. H. Kurth, M. C. Kurth, S. E. Poduslo, and J. D. Schwa&ham, Ann. Neural. 33,368-372 (1993). S. Nanko, A. Ueki, and M. Hattori, Neurosci. L&t. 204,125-127 (1996). M. H. Polymeropoulos, C. Lavedan, E. Leroy, S. E. Ide, A. Dehejia, A. Dutra, B. Pike, H. Root, J, Rubenstein, R. Boyer, E. S. Stenroos, S. Chandrasekharappa, A. Athanassiadou, T. Papapetropoulos, W. G. Johnson, A. M. Lazzarini, R. C. Duvoisin, G. D. Iorio, L. I. Golbe, and R. L. Nussbaum, Science 276,2045-2047 (1997). D. M. Morens, A. Grandinetti, D. Reed, L. R. White, and G. W. Ross, Neurology 45,10411051 (1995). J. S. Fowler, N. D. Volkow, G. J. Wang, N. Pappas, J. Logan, R. MacGregor, D. Alexoff, C. Shea, D. Schlyer, A. P. Wolf, D. Warner, I. Zezulkova, and R. Cilento, N&we (London) 379, 733-736 (1996). J. S. Fowler, N. D. Volkow, G. J. Wang, N. Pappas, J. Logan, C. Shea, D. Alexoff, R. R. MacGregor, D. J. Schlyer, I. Zezulkova, and A. P. Wolf, Proc. NutZ. Acad. Sci. U.S.A. 93,1406514069 (1996). H. Checkoway, G. M. Franklin, P Costa-Mallen, T. Smith-Weller, J. Dilley, P. D. Swanson, and L. G. Costa, Neurology 50, 1458-1461 (1998). W. Weyler, Y.-P. Hsu, and X. 0. Breakefield, PharmucoZ. Thx 47,391-417 (1990). l? L. Dostert, M. S. Benedetti, and K. F. Tipton, Med. Res. Reo. 9, 45-89 (1989). P Dostert,J. NeumZ Trunsm. 41,269-279 (1994). J. C. Shih, K. Chen, and M. J. Ridd, Annu. Rev. Neurosci. 22, 197-217 (1999). J. W. Greenawalt and C. Schnaitman,J. Cell BioZ. 46, 173-179 (1970). H. L. White and R. L. Tansik, Characterization of multiple substrate binding sites of MAO. In “Monoamine Oxidase: Structure, Function, and Altered Functions” (T. P. Singer, R. W, Von Korff, and D. L. Murphy, eds.), pp. 129-144. Academic Press, London, 1979. J. P. Johnston, Some observations upon a new inhibitor of monoamine oxidase in brain tissue. Biochem. PhanucoZ. 17, 1285-1297 (1968). K. N. Westlund, R. M. Denney, L. M. Kochersperger, R. M. Rose, and C. W. Abell, Science 230,181-183 (1985). K. N. Westlund, R. M. Denney, R. M. Rose, and C. W. Abell, Neuroscience 25,439-456 (1988). R. M. Denney, R. R. Fritz, N. T. Patel, and C. W. Abell, Science 215,1400-1403 (1982). L. M. Kochersperger, A. Waguespack, J. C. Patterson, C.-C. W. Hsieh, W. Weyler, J. I. Salach, and R. M. Denney,J. Neurosci. 5,2874-2881 (1985). A. W. Bach, N. C. Lan, D. L. Johnson, C. W. Abell, M. E. Bembenek, S. W. Kwan, P H. Seeburg, and J. C. Shih, Proc. N&l. Acad. Sci. U.S.A. 85,4934-4938 (1988). Y.-P. P Hsu, W. Weyler, S. Chen, K. B. Sims, W. B. Rinehart, M. C. Utterback, J. F. Powell, and X. 0. Breakefield,J. Neurochem. 51,1321-1324 (1988).

154

CREED

W. ARELL

AND SAU-WAH

KWAN

31. J. E Powell, Y.-P P. Hsu, W. Weyler, S. A. Chen, J. SaIach, K. Andrikopoulos, J. Mallet, and X. 0. Breakefield, Biochem. J. 259,407-413 (1989). 32. T. Kuwahara, S. Takamoto, and A. Ito, Agric. Biol. Chem. 54, 253-257 (1990). 33. A. Ito, T. Kuwahara, S. Inadome, and Y. Sagara, Biochem. Biophys. Res. Commun. 157, 970-976 (1988). 34. S.-W. Kwan and C. W. Abell, Camp. Biochem. Physiol. 102B, 143-147 (1992). 35. M. G. Rossman, A. Liljas, C.-I. Branden, and L. J. Banaszak, Evolutionary and structural relationships among dehydrogenases. In “The Enzymes” (P. D. Boyer, ed.), pp. 61-102. Academic Press, New York, 1975. 36. R. K. Wierenga, P. Terpstra, and W. G. J. Hol,J. Mol. BioZ. 187, 101-107 (1986). 37. C. W. AbelI, R. M. Stewart, P J. Andrews, and S.-W. Kwan, Heterocycles 39, 933-955 (1994). 38. K. Chen, H.-F. Wu, and J. C. Shih,]. Neurochem. Sl, 187-190 (1993). 39. J. E. Pintar, J. Barbosa, U. Francke, C. M. Castiglione, Jr., M. Hawkins, and X. 0. Breakefield,/. Neurosci. 1, 166-175 (1981). 40. L. M. Kochersperger, E. L. Parker, M. Siciliano, G. J. Darlington, and R. M. Denney,]. Neurosci. Res. 16,601-616 (1986). 41. N. C. Lan, C. Heinzmann, A. Gal, I. KIisak, U. Orth, E. Lai, J. Grimsby, R. S. Sparkes, T. Mohandas, and J. C. Shih, Genomics 4,552-559 (1989). 42. E. R. Levy, J. F. Powell, V. J. Buckle, Y.-P. P. Hsu, X. 0. Breakefteld, and I. W. Craig, Genomics 5,368-370 (1989). 43. L. Ozelius, Y.-P. P. Hsu, G. Bruns, J. F Powell, S. Chen, W. Weyler, M. Utterback, D. Zucker, J. Haines, J. A. Trofatter, P. M. ConneaIly, J. F GuseIIa, and X. 0. Breakefield, Genomics 3,53-58 (1988). 44. C. R. Cloninger, L. von Knorring, and L. Oreland, Psychiatr. Res. 15,133-143 (1985). 45. Z.-Y. Chen, G. S. Hotamisligil, J.-K. Huang, L. Wen, D. Ezzeddine, N. Aydin-Muderrisoglu, J. F. Powell, R. H. Huang, X. 0. Breakefield, I. Craig, and Y.-l? P. Hsu, Nucleic Acids Res. 19,4537-4541 (1991). 46. J. Grimsby, K. Chen, L. J. Wang, N. C. Lan, and J. C. Shih, Proc. N&Z. Acad. Sci. U.S.A. 88, 3647-3641 (1991). 47. S.-W. Kwan, J. M. Bergeron, and C. W. AbeU, Psychophurnaucol. Suppl. 106, Sl-S5 (1992). 48. Q. S. Zhu, J, Grimsby, K. Chen, and J. C. Shih,J. Neurosci. b&4437-4446 (1992). j. Neurochem. 63,843-856 49. R. M. Denney, A. Sharma, S. K. Dave, and A. Waguespack, (1994). 50. Q. Zhu and J. C. Shih,J. Neurochem. 69,1368-1373 (1997). 51. W. H. Walker, E. B. Keamey, R. Seng, and T. P. Singer, Eur. J. Biochem. 24, 328-331 (1971). 52. N. Minamiura and K. T. Yasunobu, Arch. Biochem. Biophys. 189,481-489 (1978). 53. W. Weyler, Biochem. J. 260, 725-729 (1989). 54. B. A. Shiloff, P Q. Behrens, S.-W. Kwan, J. H. Lee, and C. W. AbelI, Eur. J. Biochem. 242, 41-50 (1996). 55. S.-W. Kwan, D. A. Lewis, B. P. Zhou, and C. W. Abell, Arch. Biochem. Biophys. 316,385391 (1995). 56. M. Smith, Annu. Rev. Genet. 19,423-462 (1985). 57. W. P. Deng and J. A. Nickoloff, Anal. Biochem. 200, 81-88 (1992). 58. J. Mitoma and A. Ito,J. Biochem. 111,20-24 (1992). 59. C. Branden and J. Tooze, In “Introduction to Protein Structure.” Garland, New York, 1991. 60. L. D. Arscott, C. H. Williams, Jr., and G. E. Schulz, Zn “Flavins and Flavoproteins” (v. Massey and C. H. Williams, eds.), pp. 44-48. Elsevier North-Holland, Amsterdam, 1982.

MAO STRUCTURE

AND FUNCTION

155

61. J. Gottowik, A. M. Cesura, P. Malherbe, G. Lang, and M. Da Prada, FEBS Leti. 317,152 156 (1993). 62. K. Chen, H.-F. Wu, and J. C. Shih,J. Neurochem. 66,797-803 (1996). 63. B. P. Zhou, D. A. Lewis, S.-W, Kwan, and C. W. Abell,]. BioZ. Chem. 270,23653-23660 (1995). 64. B. P. Zhou, B. Wu, S.-W. Kwan, and C. W. Abell,J. BioZ. Chem. 273,14862-14868 (1998). 65. P. A. Karplus, M. J. Daniels, and J. R. Herriott, Science 251, 60-66 (1991). 66. G. Lu, W. H. Campbell, G. Schneider, and Y. Lindqvist, Structure 2,809-821 (1994). 67. C. C. Correll, C. J. Batie, D. P. BaIlou, andM. L. Ludwig, Science 258,1604-1610 (1992). 68. T. Yubisui, T. Miyata, S. Iwanaga, M. Tamura, S. Yoshida, M. Takeshita, and H. Nakajima, J. Biochem. 96,579-582 (1984). 69. T. D. Porter and C. B. Kasper, Proc. N&Z. Acad. Sci. U.S.A. 82,973-977 (1985). 70. J. Ostrowski, M. J. Barber, D. C. Rueger, B. E. Miller, L. M. Siegel, and N. M. Kredich, J. BioZ. Chem. 264,15796-15808 (1989). 71. C. M. Bruns and P. A. Karplus,J. Mol. Biol. 247, 125-145 (1995). 72. B. P. Zhou, D. A. Lewis, S.-W. Kwan, T. J. Kirksey, and C. W. Abell, Biochemistry 34,95269531 (1995). 73. T. J. Kirksey, S.-W. Kwan, and C. W. Abell, Biochemistry 37, 12360-12366 (1998). 74. C. C. Correll, M. L. Ludwig, C. M. Bruns, and P. A. Karplus, Protein Sci. 2, 2112-2133 (1993). 75. G. Eggink, H. Engel, G. Vriend, P. Terpstra, and B. Witholt, J. Mol. Biol. 212, 135-142 (1990). 76. G. Van Driessche, M. Koh, Z. W. Ch en, F. S. Mathews, T. E. Meyer, R. 6. Bartsch, M. A. Cusanovich, and J. J. Van Beeumen, Protein Sci. 5, 1753-1764 (1996). 77. G. E. Schulz, Curr. Opin. Struct. BioZ. 2,61-67 (1992). 78. H.-F. Wu, K. Chen, and J. C. Shih, Mol. Pharmacol. 43,888-893 (1993). 79. F Ogata, Y. Tsugeno, J. Mitoma, and A. Ito, In “Flavins and Flavoproteins 1993” (K. Yapi, ed.), pp. 795-798. de Gruyter, Berlin, 1994. 80. I. Hiro, Y. Tsugeno, I. Hirashiki, F. Ogata, and A. Ito, Z. Biochem. 120, 759-765 (1996). 81. K. M. Robinson and B. D. Lemire,]. BioZ. Chem. 271,4055-4060 (1996). 82. A. Otto, M. Stoltz, H. P. Sailer, and R. Brandsch,J. Bid. Chem. 271, 9823-9829 (1996). 83. M. Stolh, J. Rassow, A. F. Buckmann, andR. Brandsch,J. BioZ. Chem. 271,25208-25212. (1996) 84. L. C. Packman, M. Mewies, and N. S. Scrutton,J. BioZ. Chem. 270,13186-13191 (1995). 85. J. Kim, J. H. Fuller, V. Kuusk, L. C unane, Z. W. Chen, F. S. Mathews, and W. S. M&tire, J. BioZ. Chem. 270,31202-31209 (1995). 86. R. Brandsch and V. Bichler,J. BioZ. Chem. 266,19056-19062 (1991). 87. R. Brandsch, V. Bichler, M. Schmidt, and J. Buchner, J. BioZ. C/rem. 267, 20844-20849 (1992). 88. V. Massey and B. Curti,J. Biol. Chem. 241,3417-3423 (1966). 89. B. Zhong and R. B. Silverman,]. Am. Chem. Sot. 119,6690-6691 (1997). 90. S. 0. Sablin and R. R. Ramsay,J. BioZ. Chem. 273,14074-14076 (1998). 91. L. Mauch, V. Bichler, and R. Brandsch,J. BioZ. Chem. 265, 12761-12762 (1990). 92. J. Wouters and G. Baudoux, Proteins 32,97-110 (1998). 93. T. L. Poulos, B. C. Finzel, and A. J. Howard,]. Mol. BioZ. 195,687-700 (1987). 94. J. Gottowik, P. MaIherbe, G. Lang, M. Da Prada, and A. M. Cesura, Eur. J Biochem. 230, 934-942 (1995). 95. J. Grimsby, M. Zentner, and J, C. Shih, Lifi. Sci. 58, 777-787 (1996). 96. Y. Tsugeno and A. Ito,J. BioZ. Chem. 272, 14033-14036 (1997).

156

CREED

W. ABELL AND SAU-WAH KWAN

97. A. Hiram&u, S. Tsurushiin, and K. T Yasunobu, Eur.]. Biochem. 57,587-593 (1975). 98. A. M. Cesura, J. Gottowik, H. W. Lah m, G. Lang, R. Imhof, P. Malherbe, U. Rothlisberger, and M. Da Prada, Eur. J. Biochem. 236,996-1002 (1996). 99. R. M. Denney, N. T. Patel, R. R. Fritz, and C. W. Abell, Mol. Pharmucol. 22, 500-508 (1982). 100. R. M. Denney, R. B. Fritz, N. T. Patel, S. G. W’1d en, and C. W. Abell, Mol. Pharmacol. 24, 60-68 (1983). 101. L. W. Thorpe, K. N. Westhmd, L. M. Kochersperger, C. W. AbeIl, and R. M. Denney,J. Histochem. Cytochem. 35,23-32 (1987). 102. L. A. Riley, M. A. Waguespack, and R. M. Denney, Mol. Pharmucol. 36,54-60 (1989). 103. A. Dahlstriim and K. Fuxe, A& Physiol. &and. Suppl. 62,3-55 (1964). 104. K. N. Westlund, T J. Krakower, S.-W. Kwan, and C. W. Abell, Bruin Res. 612, 223-230 (1993). 105. J. M. Luque, S.-W. Kwan, C. W. Abell, M. Da Prada, and J. G. Richards, J. Camp. Neural. 363,665-680 (1995). 106. J. A. Roth, A. J. Rivett, A. Francis, L. B. Pearce, and D. Jeffrey, In “Monoamine Oxidase and Disease: Prospects for Therapy With Reversible Inhibitors” (K. F. Tipton, P. Dostert, and M. S. Benedetti, eds.), pp. 459-468. Academic Press, London, 1984. 107. V. Glover, M. Sandier, F. Owen, and G. J. Riley, Nature (London) 265, 80-81 (1977). 108. J. W. Langston, P. Ballard, J. W. Tetrud, and I. Irwin, Science 219,979-980 (1983). 109. R. S. Bums, C. C. Chiueh, S. P. Markey, M. H. Ebert, D. M. Jacobowitz, and I. J. Kopin, Proc. N&l. Acud. Sci. U.S.A. 80,4546-4550 (1983). 110. R. R. Fritz, C. W. Abell, N. T. Patel, W. Gessner, and A. Brossi, FEBS Lett. 186,224-228 (1985). 111. K. Chiba, A. J. Trevor, and N. Castagnoli, Jr., Biochem. Biophys. Res. Common. I28,12281232 (1985). 112. R. E. Heikkila, W. J. Nicklas, I. Vyas, and R. C. Duvoisin, Neurosci. Lett. 62, 389-394 (1985). 113. G. Cohen, P. Pasik, B. Cohen, A. Leist, C. Mytilineou, and M. D. Yahr, Eur. J. Pharrnacol. 106,209-210 (1984). 114. J. W. Langston, 1. Irwin, and E. B. Langston, Science 225,1480-1482 (1984). 115. J. Grimsby, M. Toth, K. Chen, T. Kumazawa, L. Klaidman, J. D. Adams, F. Karoum, J. Gal, and J. C. Shih, Nature Genet. 17,206-210 (1997). 116. The Parkinson Study Group, N. Engl. J. Med. 321, 1364-1371 (1989). 117. J. Knoll, Mech. Ageing Da. 46,237-262 (1988). 118. I. M. Gallagher, A. Glow, and V. Glover,]. Neural Trunsm. 52,315-320 (1998). 119. W. Birkmayer, J. Knoll, P. Riederer, M. B. H. Youdim, V. Hars, and J. Marton, J. Neural Transm. 64, 113-127 (1985). 120. E. M. Marcotte, M. Pellegrini, H.-L. Ng, D. W. Rice, T. 0. Yeates, and D. Eisenberg, Science 285,751-753 (1999).

Squalene

Synthase:

Structure

and Regulation TERESE R. TANSEY AND I~HAIAHU SEXECHTER~ Department of Biochemistry and Molecular Biology Uniformed Services University of the Health Sciences Bethesda, Maryland 20814 History of SQS Research ................. Purification of Rat SQS .................. Intracellular Location of SQS ............. Isolation of Mammalian SQS cDNA Clones . Reaction Mechanism .................... VI. Identification of Functional Domains in SQS .......................... A. Protein Sequence Comparisons B. Mutagenic Analysis of SQS ............................. VII. SQS Genes and mRNAs ................................... A. Location and Structure of SQS Genes .................... B. Size and Organ Distribution of SQS mRNA ............... VIII. SQS Regulation .......................................... A. Regulation of SQS Activity and Protein Levels ............. .......................... B. Significance of SQS Regulation C. Sterol Regulation of SQS mRNA in Rats and in Human HepG2 Cells ......................................... D. SQS Promoter Analysis ................................. E. Regulation of SQS by Lipopolysaccharide and Cytokines References ..............................................

159 161 163 164 166 166 166 168 174 174 176 177 I77 178

I. II. III. IV. V.

179 179 189 191

....

Squalene synthase (SQS) catalyzes the first reaction of the branch of the isoprenoid metabolic pathway committed

specifically to sterol biosynthesis. Regula-

tion of SQS is thought to direct proximal intermediates

in the pathway into either

sterol or nonsterol branches in response to changing cellular requirements. importance

of SQS in cholesterol

mechanism,

structure,

C,,

isoprenoid,

and regulation

has stimulated

of the enzyme.

head to head. Site-directed

research

SQS produces

in a two-step reaction in which two molecules

phate are condensed tified conserved

metabolism

The

on the

squalene,

a

of farnesyl diphos-

mutagenesis of rat SQS has iden-

Tyr, Phe, and Asp residues that are essential for function. The

aromatic rings of Tyr and Phe are postulated

1 To whom correspondence

should be addressed.

to stabilize carbocation

intermedi-

158

TERESE

R. TANSEY AND ISHAIAHU

SHECHTER

ates of the first and second half-reactions, respectively; the acidic Asp residues may he required for substrate binding. SQS activity, protein level, and gene transcription are strictly and coordinately regulated by cholesterol status, decreasing with cholesterol surfeit and increasing with cholesterol deficit. The human SQS (hSQS) gene has an unusually complex promoter with multiple binding sites for the sterol regulatory element binding proteins SREBP-la and SREBP-2, and for accessory transcription factors known to he involved in the control of other sterolresponsive genes. SREBP-la and SREBP-2 require different subsets of hSQS regulatory DNA elements to achieve maximal promoter activation. Current research is directed at elucidating the precise contribution made by individual SREBPs and accessory transcription factors to hSQS transcriptional control. 0 2000Academic

Cholesterol plays several essential roles in mammalian cell biology. It modulates the properties of cell membranes and serves as the precursor for steroid hormones, bile acids, and vitamin D (I, 2). Addition of a cholesterol moiety to members of the Hedgehog family of developmental signaling proteins is required for proper embryonic patterning (3 - 5). Because high plasma cholesterol contributes to atherosclerotic disease (6), whereas cholesterol deficit causes developmental defects (5, 7), blood and body cholesterol levels must be carefully controlled (I). farSqualene synthase (SQS) (f amesyl-diphosphate:famesyl-diphosphate nesyltransferase, EC 2.5.1.2 l), one of the key enzymes in cholesterolgenesis, catalyzes the first reaction of the branch of the isoprenoid matabolic pathway committed primarily to cholesterol biosynthesis (Fig. 1) (8). SQS produces squalene, a C,, isoprenoid, from the C,, allylic compound, farnesyl diphosphate (FPP), in a two-step reaction (Fig. 2A). Initially, two molecules of FPP are condensed to form presqualene diphosphate (PSPP), which is then rearranged and reduced by NADPH to form squalene (9, IO). Although the two products of the SQS reaction are recognized primarily as cholesterol precursors, both PSPP and squalene have additional functions. PSPP is a mediator of the inflammatory response in neutrophils, where it inhibits oxygen radical production (11). Squalene lies immediately proximal to the branchpoint in sterol biosynthesis at squalene epoxide. Squalene epoxide can be converted either to cholesterol by the standard sterol biosynthetic pathway, or to oxysterols via the squalene dioxide pathway (Fig. 1) (12). 24(S),25-Epoxycholesterol, one of the oxysterols believed to be produced using the squalene dioxide pathway, may be involved in the regulation of bile acid synthesis (13, 14). FPP, the substrate for SQS, lies at a major branchpoint in isoprenoid biosynthesis (Fig. 1). Regulation of SQS activity is thought to play a critical role in directing FPP into either sterol or nonsterol branches in response to changing cellular requirements (1516). The importance of the SQS reaction

SQUALENE

159

SYNTHASE

Acetyl CoA * * HMG CoA HMG CoA Reductase 4

Mevalonate * * Isopentenyl-PP

+

Isopentenyl tRNA

+

FPP Synthase

Ubiquinones Dolichols

+ Farnesyl-PP

Heme A Farnesylated Proteins

Squalene Synthase

4

Squalene

Geranylgeranylated Proteins

3

4 Squalene Epoxide *

+

Squalene Dioxide Ir

* * Cholesterol

4 Bile Acids

4 Steroid Hormones

* 1' 24(S),25-Epoxycholesterol

4 Vitamin D

FIG. 1. The isoprenoid biosynthetic

pathway.

as a regulatory step in isoprenoid metabolism has sparked investigation of both the structure of the enzyme and the regulation of its activity. Our laboratory has focused on the structure, function, and regulation of mammalian SQS, which is the subject of this review. Because SQS catalyzes the first committed step in sterol production, it has been targeted for the development of inhibitors to serve as drugs to lower blood and body cholesterol. These studies will not be covered here, but readers are referred to other publications for an introduction to this literature (17-23).

I. History of SQS Research Much of the early work on SQS was done in the laboratories of Bloch, Popjak, Porter, and Rilling (24). Initial advances came primarily from stud-

we+ / PPO

cm,

j

Ionxeation

of

Donor

FPP

c3

C2

PPI,

H+ 1'-2-3

Presqualene

Condensation

Diphosphate

4

Ionization

Rearrangement Reduction

Phytoene

FIG. 2. (A) Proposed reaction mechanisms for the two half-reactions catalyzed by SQS. (B) The structure of phytoene.

SQUALENE

161

SYNTHASE

ies of the yeast enzyme, which proved more active than mammalian SQS isolated from normal liver microsomes. Many investigations of the 1960s and early 1970s focused on the structure of the SQS reaction intermediates. A C,, intermediate was proposed as early as 1964 (25). Free C,, diphosphate esters were then shown to accumulate in reactions lacking NADPH (26-29). The structure of PSPP remained controversial (26-28, 30) until firmly established in 1971 (31-33). The discovery of PSPP lead to numerous proposals regarding the reaction mechanism and the involvement of SQS in catalysis (24,30,31,34). SQS partially purified from yeast using detergents enabled the first kinetic studies (35), but difficulties completely solubilizing the enzyme hampered progress (36). It was not until the late 1980s that Rilling and colleagues successfully solubilized and purified active yeast SQS (ySQS) to homogeneity (37, 38). The purified protein produced both PSPP and squalene, and migrated as a single band of Mr 47 kDa on denaturing polyacrylamide gels, suggesting that the two-step reaction is carried out by a single polypeptide. The purification of a truncated, active form of mammalian hepatic SQS was achieved by 1992 (Section II) (39). In yeast and other fungi, squalene is a precursor of ergosterol, a cholesterol-like molecule (40). A collection of ergosterol auxotrophs (41, 42) contributed to the isolation of the first cloned SQS genes in 1991. erg9 mutants were suspected to have a defect in SQS because they are unable to convert mevalonate to squalene, but instead accumulate farnesol, a product of FPP hydrolysis. The Saccharomyces cerevisiae ERG9 gene encoding SQS was cloned by phenotypic complementation of an erg9 mutant (43, 44). Three laboratories reported the cloning of rat and human squalene synthase cDNAs in 1993 (45-48). Molecular cloning of SQS enabled the overexpression and subsequent purification of large quantities of recombinant enzyme from Escherichia coli or insect cells (49-53). This has facilitated more detailed study of SQS catalysis and enzyme structure (Section VI). SQS cloning also has been crucial to molecular dissection of the regulation of enzyme activity (Section VIII).

II. Purification of Rat SQS Purification of active SQS from rat liver microsomes three experimental observations (39).

was facilitated by

1. Hepatic SQS activity could be induced ll- to 25-fold by including statins and cholestyramine in the diet. The stains (e.g., lovastatin, fluvastatin) are inhibitors of hydroxymethylglutaryl (HMG) CoA re-

162

TERESE

R. TANSEY AND ISHAIAHU

SHECHTJZR

ductase (HMGR), which is considered the rate-limiting enzyme of cholesterolgenesis (8). Cholestyramine is a bile acid sequestrant that decreases the return of bile acids to the liver via the enterohepatic circulation (2). The combined actions of the statins and cholestyramine deplete cellular cholesterol and induce many of the enzymes of the cholesterol biosynthetic pathway (1, 8). 2. The activity of microsomal SQS could be enhanced by preparing microsomes in the presence of dithiothreitol, and including it and the detergent CHAPS in the reaction mixture. This allowed the SQS assay to be carried out in a reaction volume of 100 ~1. Before the optimized protocol was introduced, the standard SQS assay had been relatively laborious, requiring a large reaction volume and extraction of the reaction with organic solvents prior to chromatographic separation of squalene (35). Alternatively, squalene had been quantitated by measuring protons released during the reaction after exchange into methanol and distillation (29). Squalene produced in the microassay could be measured by spotting an aliquot of the reaction directly onto thinlayer chromatography plates, without prior organic extraction. Many samples, such as the multiple fractions from enzyme purification columns, could be assayed easily by carrying out reactions in microtiter plates. 3. In initial studies of fractionated hepatic cell lysates, SQS activities were found in both microsomal and postmicrosomal (cytosolic) fractions. However, when protease inhibitors were present during cell lysis and fractionation, SQS activity was exclusively microsomal. These observations suggested that a relatively stable, soluble form of active SQS could be released from the microsomal membrane by proteolyic cleavage, and prompted the successful use of controlled trypsinolysis to obtain active soluble SQS. Thus, a step that often had frustrated earlier work with SQS, i.e., development of a method to solubilize the membrane-bound enzyme with detergents and retain activity, was avoided. Rat SQS (rSQS) re 1eased from microsomal membranes by controlled trypsinolysis was purified 156-fold by sequential chromatography on DEAEcellulose, hydroxylapatite, and phenylSepharose columns (39). Proteins present at various stages in the purification are illustrated in Fig. 3A. The purified, active enzyme obtained from the phenylSepharose column migrated as a single band on sodium dodecyl sulfate (SDS) polyacrylamide gels (Fig. 3A, lane 4). This indicated that rSQS, similar to ySQS, is a single polypeptide capable of carrying out both catalytic steps. The sizes of the microsomal and soluble forms of rSQS were determined by Western blotting, using antibodies raised against the purified protein (Fig. 3B). The estimated M, of the

SQUALENE

163

SYNTHASE

B

A MS

MS+ttyp SlSO

DEAE- phenylSeph. cell.

Fluvastatin +

1

2

phenylSeph.

f

47 kDa

+

32 kDa

3

FIG. 3. Purification of rSQS. (A) Coomassie Blue-stained SDS-polyacrylamide gel of proteins at various stages of rSQS purification. Lane 1, liver microsomes from fluvastatin-treated rats; lane 2, S,,,, supematant of trypsin-treated microsomes; lane 3, rSQS fraction following chromatography on DEAE-cellulose; lane 4, rSQS fraction following chromatography on phenyl-Sepharose. (B) Western blot treated with rSQS antibody. Lane 1, 12 pg of protein from normal liver microsomes; lane 2, 12 kg of protein from fluvastatin-treated liver microsomes; lane 3, trypsin-truncated rSQS following purification on phenyl-Sepharose. Adapted from Shechter et nl. (39).

microsomal enzyme was 47 kDa, similar to that reported for ySQS. The trypsinized, truncated form of the enzyme was 32 kDa.

III. Intracellular location of SQS All intermediates of cholesterol biosynthesis up to FPP are water soluble, whereas squalene and subsequent intermediates are very hydrophobic. These substrate properties seem to be reflected in the different intracellular locations of the cholesterolgenic enzymes acting prior to and after SQS. Enzymes immediately proximal to SQS are located primarily in peroxisomes (e.g., FPP synthase, mevalonate kinase). Squalene epoxidase and oxidosqualene cyclase, the enzymes that catalyze the two reactions subsequent to SQS, are located in the membrane of the endoplasmic reticulum (ER) (54). Until an SQS antibody became available, attempts to localize SQS were limited to assays of enzyme activity in subcellular fractions that were subject to cross-contamination. Although the cellular fractionation experiments performed during the purification of rSQS clearly showed the majority of SQS activity in microsomes (Section II) (39), conflicting reports made it uncer-

164

TERESE

R. TANSEY AND ISHAL4HU

SHECHTER

tain whether SQS might also be associated with peroxisomes (55, 56). To resolve this issue, the SQS antibody generated against purified, truncated rSQS was used to localize hepatic rSQS in situ (57). Immunoelectron microscopy of normal rat livers showed SQS associated only with the ER membrane; no SQS was detected in peroxisomes. The ability to release an active form of SQS from microsomal membranes by trypsinolysis suggests that SQS is anchored to the ER membrane with its catalytic domain projecting into the cytosol. This orientation would enable SQS to accept water-soluble FPP and NADPH from the cytosol and release lipophilic squalene to the ER membrane, thus reflecting the enzyme’s unique position in the cholesterol biosynthetic pathway.

IV. Isolation of Mammalian SQS cDNA Clones Cloning of an rSQS cDNA began with the design of degenerate polymerase chain reaction (PCR) primers, based on the amino acid sequences of the N terminus and an internal tryptic peptide of the purified, truncated rSQS described in Section II. The initial 8 72-bp cDNA, amplified from RNA prepared from the H35 rat hepatoma cell line, was extended using 5’ and 3’ rapid amplification of cDNA ends (RACE) to obtain the complete proteincoding region (45). The same degenerate PCR primers were used to amplify a human SQS (hSQS) cDNA from the HepG2 human hepatoma cell line (46). Additional hSQS cDNAs were described soon after the report on the rSQS cDNA appeared (47, 48, 51). The rSQS cDNA contained a 1248bp open reading frame that encoded a protein of 416 amino acids (Fig. 4). The protein had a calculated molecular mass of 48.1 kDa, similar to the estimated 47 kDa of intact microsomal rSQS. Based on the known amino acid sequence of its N terminus, the trypsin-truncated enzyme began at Asn-34 following the trypsin cleavage site at Arg-33. The truncated enzyme also appeared to be missing ~100 Cterminal amino acids, based on its estimated M,of 32 kDa, as determined by SDS-polyacrylamide gel electrophoresis. However, recombinant hSQS has been found to migrate unusually rapidly through SDS-polyacrylamide gels after deletion of either 47 or 82 C-terminal amino acids, leading to underestimates of M, by several kilodaltons (52). Because the hSQS and rSQS sequences are very similar (Fig. 4), tryp sin-truncated rSQS may be larger than 32 kDa. Based on the observation that the C-terminal 47-residue deletion of hSQS was enzymatically active, whereas the 82-residue deletion was not (52), trypsin-truncated active rSQS probably arose by cleavage at one of the five possible trypsin cleavage sites between Arg-348 and Arg-3 79, which would remove between 38 and 69 C-terminal residues.

rsas

192 192 189 199

278 278 277 285

378 378 371 322

rSQS hSQS atSQS

rSQS hSQS atSQS

rSQS hSQS atSQS

L

J

,.SPIYLSFI ..SPIYLSFV QPNSVFIIW NETPIFLK.V

'B,we II ER

SRSHT..... SRSHY..... NRKSYVNDKG QDKLPPNVKP

++

SRLRN SRLRN VSLRD ASIHE

288

286

283

&Section

I I R. I I I LY E

FEDPIVGEDT FEDPLVGEDT SE..VLTPDW FANESLYSN.

208

301

223

MLLAALSWQl MLLAALSWQY VILLAIVFAY KERSRYDDEL

LSTLSQVTED LTTLSQVTED LRAN 410 VPTQQEEEYK

LL 295

416 417

FNMVLSIILS

YVQR.EH YVQTGEH

228

KMDRNSLSNS KMDQDSLSSS AEKQIPPEPH IY-DQ.STSPY

Sianal-Anchor

C L

219

RMGGRRNFIP RIGGKRKVMP . . ..KRAIEK KF.CRTPLFS

336

':LLGFYfIYT

J

J

LHXA

4-14

Section

I

4

17i L

85

Section

174 3-

60

A

370

186 k

FSASE FSASE FLAAG IVIAK 448

191 191 188 198

FIG:. 4. Alignment of SQS proteinsequences.The rat (rSQS), h moan (hSQS), A. thaliana (atSQS), and S. cerevisiae (ySQS) SQS sequences were aligned using the Pileup Program of the Wisconsin Package. Sequences consewed in N. c7(1ssa phytoene synthase (ncPHS) are shown below sections I, A, and B. Residues conserved in all four SQS sequences are highlighted in black. Arrows indicate specific residues referred to in the text.

YSQS

YSQS

YSQS ncPHS

100 100 97 99

1 MGKLLQLALH

rSQS hSQS atsas YSQS ncPHS

YSQS nCPHS

hSQS atSQS

51

34

166

TERESE

R. TANSEY AND ISHAIAHU

SHECHTER

V. Reaction Mechanism Neither the SQS reaction mechanism nor the catalytic process is fully understood. Details of one model of the reaction mechanism, based primarily on the investigations of Poulter and Rilling (9, IO, 31,58- 61; reviewed in 62), will be presented here to aid the summary of attempts to localize functional domains of the enzyme (Section VI). As mentioned earlier, the SQS reaction occurs in two stages (Fig. 2A). In the first half-reaction, ionization of the donor FPP generates an allylic carbocation and a diphosphate anion. The carbocation adds across the 2,3 double bond of the acceptor FPP, as the pro-S proton from the C-l’ allylic methylene group is eliminated. This creates the cyclopropane ring of PSPP The second half-reaction is believed to proceed by ionization of PSPP, forming an initial carbocation at C-l. Rearrangement and reduction by NADPH follow to form squalene, in which the 1’ and 1 carbons of the two farnesyl moieties are joined. SQS requires a divalent cation, preferably Mg2+. In other prenyltra n sferases whose reaction mechanisms are understood in more detail, Mg2+ is thought to be involved in binding of diphosphate substrates and in assisting the extraction of the diphosphate group (63 - 68). The SQS reaction differs from that of most other prenyltransferases in several respects: condensation of two identical substrate moieties to form a 1’ 4 1, rather than a 1’ + 4, linkage, formation of a stable cyclopropylcarbinyl diphosphate intermediate, and the requirement for NADPH. The prenyltransferase most similar to SQS is phytoene synthase (PHS), which catalyzes the conversion of two molecules of geranylgeranyl diphosphate, a C,, isoprenoid, to the C,, compound phytoene (Fig. 2B). Phytoene is an intermediate in the biosynthesis of carotenoids, which are produced by plants and certain photosynthetic bacteria (69). The first half-reaction of PHS, thought to be similar to that of SQS, produces the cyclopropylcarbinyl diphosphate intermediate, prephytoene diphosphate, which is analogous to PSPP The second half-reaction differs from that of SQS in that prephytoene diphosphate is converted to phytoene by a nonreductive rearrangement, and thus NADPH is not required. As with SQS, the two-half-reactions of PHS are catalyzed by a single polypeptide (70 - 74).

VI. Identification of Functional Domains in SQS A. Protein Sequence

Comparisons

As cDNA sequences for SQS and PHS became available, the deduced amino acid sequences were compared to identify conserved, and thus potentially functional, domains. Figure 4 shows an alignment of SQS sequences

SQUALENE

SYNTHASE

167

from mammals (rSQS and hSQS), the plant Arubidopsis thaliana (atSQS), and the yeast S. cerevisiue (ySQS). Th e se q uences of three regions of Neurospora crassa PHS (ncPHS) h omologous to sections of SQS also are shown. Conserved residues (highlighted in black) are found throughout the central portion of SQS; the N and C termini are not conserved. Most attention has focused on four clusters of conserved residues (Fig. 4, sections I, A, B, and C). Regions similar to sections I, A, and B exist in PHS, suggesting that these domains serve functions common to SQS and PHS catalysis. Section C is unique to SQS (43-45, 47, 58, 51). Several enzymes that catalyze cyclization or chain elongation reactions using allylic diphosphate substrates such as FPP contain one or more repeats of the aspartate-rich motif DDXXD 5 to 10 residues away from one or more basic residues (67, 68). Mutagenesis of FPP synthase demonstrated the need for all or a subset of the aspartates and basic amino acids for catalysis (7579). Analyses of the three-dimensional structures of three enzymes that contain the DDXXD motif have helped define its function. In FPP synthase and the two isoprenoid cyclases, pentalenene synthase and 5-epiaristolochene synthase, the aspartates of the DDXXD sequence are located on the wall of the active site cavity near the cavity entrance. The acidic side chains project into the cavity to coordinate Mg2+ ions, which in turn stabilize binding of the substrate’s diphosphate group and possibly assist in substrate ionization. The essential basic residues located adjacent to DDXXD have been proposed to play similar roles by interacting directly with the diphosphate groups (63-68). Two aspartate-rich domains would be expected in SQS, because structural analyses of FPP synthase and the two isoprenoid cyclases indicate that each domain interacts with one allylic diphosphate (63 - 66). SQS contains no perfect match to the DDXXD motif. However, two related regions (Asp80 to Asp-85 and Asp-219 to Asp-223 in rSQS and hSQS; Fig. 4) have been proposed as binding sites for the Mg 2+-diphosphate moieties of the two FPP substrates. Basic amino acids follow both of the aspartate-rich regions within five to seven residues. The aspartates are part of sections I and B (Fig. 4) that are conserved in PHS. It has been hypothesized that the conservation of these two regions in SQS and PHS reflects a role in positioning the substrates for 1’ + 1 condensation (43, 45, 47, 48, 51, 53). As shown in Fig. 4, section A, the third region conserved in SQS and PHS, has a core of aliphatic hydrophobic residues flanked by polar or basic residues. Mutagenesis studies have confirmed the importance of this region in the first half-reaction, which is expected to be similar in SQS and PHS (Section VLB) (53). No region homologous to conserved section C of SQS is found in PHS. This highly nonpolar region thus may be involved in aspects of the second

168

TERESER.TANSEYANDISHAIAHUSHECHTER

half-reaction that are unique to SQS, e.g., the requirement for NADPH (43, 45, 47, 48, 51). S ome investigators have speculated that section C may be involved in binding NADPH, but sequence analysis and secondary structure prediction found no similarities to known pyridine dinucleotide binding motifs in SQS (43, 47, 48, 50). In addition to domains involved in substrate and cofactor binding and catalysis, SQS should contain sequences responsible for its localization on the endoplasmic reticulum (ER) membrane. Initial hydrophobicity analysis of the ySQS sequence suggested that the enzyme has a single membranespanning domain of ~20 amino acids near the C terminus (43). As more SQS sequences became available, the hydrophobicity of the C-terminal region was found to be conserved, although the specific amino acid sequence was not. The C-terminal regions of rSQS and hSQS between residues 379 and 4 10 meet the general criteria for type II ER signal-anchor sequences, which contain a hydrophobic core of 19 -2 7 amino acids flanked by polar residues, with the N-terminal flanking section more positively charged than the C-terminal section (Fig. 4) (81, 82). As a type II membrane protein, SQS would be predicted to have a large N-terminal catalytic domain facing the cytosol, a single membrane-spanning domain, and a short C-terminal domain projecting into the ER lumen. This prediction is supported by the finding that a large fragment of catalytically active rSQS can be released from microsomal membranes by trypsinolysis (Section II) (39).

B. Mutagenic

Analysis

of SQS

1. DELETIONSTUDIESDEFINETHEBOUNDARIES OFTHE CATALYTIC DOMAIN The first 30-33 amino acids at the N-terminus of mammalian SQS play little or no role in catalysis, based on the activity of truncated forms of hSQS and rSQS that lack these sequences (39, 52). C-Terminal deletion analysis demonstrated that hSQS missing the last 47 amino acids (residues 3 71- 4 17) is active, whereas hSQS missing the last 82 amino acids (residues 336-417) is inactive. Thus, the hSQS catalytic domain ends between residues 336 and 370 (52). The only conserved element in this region is the sequence DPXXX(K/R) (Fig. 4). L oss of this sequence may be responsible for the inactivity of hSQS truncated at residue 335. In summary, the N- and C-terminal deletion analysis suggests that the catalytic domain of mammalian SQS is contained between residues 33 and 3 70. 2. SPECIES-SPECIFICFUNCTIONALDOMAINS A role for the nonconserved C-terminal domain of SQS was suggested by experiments in which expression of intact human or A. thaliana SQS in yeast

SQUALENESYNTHASE

169

erg9 mutants failed to rescue the lethal phenotype (47,51,83). Lethality persisted even though the human or plant enzyme was associated with the yeast microsomal membrane in an active form (51,83). When the C-terminal onefourth to one-third of hSQS or atSQS was replaced with the yeast C-terminal sequence, ergosterol prototrophy was restored (47,83). Further experiments showed that squalene produced by erg9 microsomes expressing full-length atSQS could not be converted to subsequent compounds in the ergosterol biosynthetic pathway. In contrast, squalene produced by the chimeric atySQS was converted to later sterol intermediates (83). These observations lead to the hypothesis that yeast cells specifically require the C-terminal region of ySQS to “channel” squalene to squalene epoxidase, the next enzyme in the sterol pathway. The ySQS C terminus might serve this function either by direct physical association with squalene epoxidase or by targeting SQS to a subdomain of the ER specialized for sterol biosynthesis (83). 3. SITE-DIRECTED MUTAGENESISIDENTIFIES RESIDUES INVOLVED IN CATALYSIS a. Experimental Design. The functional importance of conserved residues in sections A, B, and C (Fig. 4) of rSQS has been tested by site-directed mutagenesis (53). In these experiments wild-type and mutant rSQSs were expressed in E. coli as thioredoxin-rSQS (TRSS) fusion proteins, with thioredoxin attached to the N terminus of full-length rSQS. Bacterial lysates were tested for TRSS protein expression by immunoblotting and were assayed for SQS activity. The first half-reaction was assayed by measuring PSPP produced from FPP in the absence of NADPH. The second half-reaction was assayed by measuring squalene produced from PSPP in the presence of NADPH. The complete reaction assay measured squalene, and in some cases PSPP, produced from FPP in the presence of NADPH. Wild-type TRSS resembled native rSQS in being able to carry out first and second half-reactions independently. In a complete reaction assay with wild-type TRSS, squalene accumulated, but PSPP did not. b. The Role of Aromatic Residues in Sections A and C. Evidence from structural studies of FPP synthase and three isoprenoid cyclases (squalenehopene cyclase, 5-epiaristolochene synthase, and pentalenene synthase) suggested that aromatic residues might play two important roles in SQS function. In FPP synthase, a pair of phenylalanines forms the floor of the isoprenoid binding pocket, and thus determines the length of the final product (64, 67). In the three cyclases, the positions of aromatic residues in the active sites suggest that the IT electrons of the aromatic rings stabilize carbocation intermediates formed during the reactions (65, 66, 84-88). Although both aromatic and nonaromatic amino acids in sections A and C (Fig. 4) of

170

TERESE

R. TANSEY AND ISHAIAHU

SHECHTER

rSQS were targeted for mutagenesis by Gu et al. (53), here the focus is on the results obtained by altering conserved aromatic residues. Section A (Fig. 4), found in both SQS and PHS, has been hypothesized to contain residues required to form the cyclopropylcarbinyl diphosphate intermediates, PSPP and prephytoene diphosphate (Section VI,A). The only aromatic residues in section A of rSQS are Tyr-171, which is absolutely conserved in all known SQSs and PHSs, and Tyr-174, which is present in most SQSs but is often replaced by His in PHS. The two Tyr residues of rSQS were converted to the aromatic nonpolar residues Phe and Trp, and to Ser, an aliphatic residue with a hydroxyl group. In assays for the complete reaction, Tyr-171 mutants did not convert FPP to squalene or to PSPP (Fig. 5A). The three Tyr-171 mutants also produced no PSPP in assays of the first half-reaction alone (Fig. 5B). However, mutants with aromatic amino acid replacements ofTyr-171 (Y171F, Y171W) were able to convert added PSPP to squalene at a very low rate in assays of the second half-reaction (Fig. 5B). In contrast, Tyr-174 mutants produced relatively high levels of squalene in assays of the complete reaction. No PSPP was detected in these assays, indicating efficient conversion of the intermediate to squalene (Fig. 5A). Assays of the separate half-reactions confirmed that the Tyr-174 mutants retained substantial first and second half-reaction activities (Fig. 5B). Thus, Tyr-171 but not Tjr-174 has an essential role in both the first and second halfreactions of rSQS. The function of the phenol ring of Tjr-171 in the first halfreaction cannot be fulfilled by an aromatic or hydroxyl moiety alone. In the second half-reaction, the aromatic nature of Tyr-171 seems more important than the hydroxyl moiety, because a low level of second half-reaction activity is retained in the Y171F and Y171W mutants. The results obtained with the Tyr-171 mutants suggest that both the r electrons and the hydroxyl group of tyrosine’s phenol ring are essential for the catalytic process. Based on these observations, a model for the function of Tyr-171 in the catalytic mechanism of the first half-reaction was proposed (Fig. 6). In this model Tyr-171 facilitates the initial ionization of FPP in two ways. The hydroxyl group provides a proton to assist separation of the diphosphate moiety, and the r system stabilizes the incipient carbocation (Fig. 6B).

FIG. 5. Activities of wild-type and mutant TRSS proteins. (A) Lysates prepared from bacteria expressing the indicated TRSS were incubated with a complete reaction mixture containing FPP and NADPH. Production of PSPP and squalene (SQU) are shown. (B) Bacterial lysates were tested for their ability to carry out the two SQS half- reactions independently. To assay the first halt-reaction, lysates were added to a reaction mix containing FPP, but no NADPH, and PSPP production was monitored. To assay the second half-reaction, lysates were added to a reaction mix containing PSPP and NADPH, and squalene production was monitored. TRSS+, Wild-type TRSS. Adapted from Cu et al. (53).

SQUALENE

171

SYNTHASE

A. Complete ‘*’ ++?T+jl

F286

Reaction F288

B. First and Second Half-Reactions

120 I

4

~rqY171)

Y174

jl

F286

II

F2i38

_

172

TERESE R. TANSEY AND ISHAIAHU

C

SHECHTER

D

FIG. 6. The proposed role of Tyr-171 in the conversion of FPP to PSPP. During the SQS first half-reaction, it is hypothesized that the donor FPP ionizes to form an allylic carbocation (A --t B), which is attacked by the nucleophilic 2,3 double bond of the acceptor FPP (B -+ C). The pro-S proton is then abstracted from the intermediate tertiary carbocation to form PSPP (C + D). If appropriately situated, Tyr171 could facilitate ionization of FPP by stabilizing the incipient carbocation through interaction with the n electrons of the phenol ring and by providing a proton to attract the diphosphate (B). 0 nce the tertiary carbocation has formed (C), the phenolate anion of Tyr-171 could serve as a relatively strong base to remove the proton, leading to formation of the cyclopropane ring in PSPP (D). From Gu et al. (53).

The phenolate anion of the resulting intermediate ion pair then serves as a relatively strong base to abstract the pro-S proton (Fig. 6C), resulting in formation of the cyclopropane ring of PSPP (Fig. 6D). Section C (Fig. 4) is conserved among SQSs but is not found in PHSs, and thus has been proposed to be involved in aspects of the second half-reaction unique to SQS (Section VI,A). Section C of rSQS contains two aromatic residues, Phe-286 and Phe-288. Each Phe was replaced with charged (Asp, Arg), aromatic (Tyr, Tip), or aliphatic hydrophobic (Leu) residues. Replacement of either Phe with an acidic or basic residue completely inactivated both half-reactions (Fig. 5). The intolerance of charge at positions 286 and 288 is not surprising, considering the hydrophobic nature of section C. All of the aromatic or aliphatic replacement mutants of Phe-286 and Phe-288, except F288W, had reduced but detectable activity in the two half-reactions. Mutant F288W was completely inactive, and was the extreme example of the trend toward lower activities for mutations introduced at position 288 compared to position 286. These results indicate that both Phe-286 and Phe-288 are required for proper SQS function in the first and second half-reactions, with Phe-288 playing a more essential role.

SQUALENE

SYNTHASE

173

PSPP accumulated in assays of the complete reaction with several of the Phe-286 and Phe-288 mutants, but it was never detected with the wild-type enzyme or with mutants carrying substitutions in sections A or B (Fig. 5A). Mutations introduced at two other positions in section C (Gln-283 and Gln293) also generated PSPP in the complete reaction assay (53). The accumulation of PSPP in reactions provided with FPP and NADPH indicates that the second half-reaction has become rate limiting, supporting the hypothesis that section C is particularly essential for this step in catalysis. The second halfreaction was most specifically compromised in the mutant F288L, which had the highest first half-reaction activity of the Phe-286 and Phe-288 mutants, but made very little or no squalene in assays of the complete or second half-reactions. The activities of mutants F288L and F288Y were consistent with a role for Phe-288 in stabilizing second half-reaction carbocations. Second halfreaction activity of F288L was hardly detectable, whereas that of F288Y was approximately 20% of wild type (Fig. 5B). Thus, a phenol group, but not a hydrophobic aliphatic moiety, can partially substitute for the benzene ring of Phe in the second half-reaction. If indeed the residue at position 288 is required to stabilize carbocations, the reason for the drastically reduced activity of F288W in both first and second half-reactions is unclear. Compared to the phenol and benzene rings of Tyr and Phe, the indole ring of Tip has been hypothesized to be the most attractive carbocation binding site because it provides the largest region of negative electrostatic potential (86, 89). The positions of Tip residues in the active sites of squalene-hopene cyclase, 5epiaristolochene synthase, and pentalenene synthase strongly suggest that Tip does stabilize carbocation intermediates (65, 66, 85). In the case of rSQS, the larger size of the indole ring in F288W may introduce a steric hindrance that inhibits both first and second half-reactions. In summary, mutational analysis has demonstrated the functional importance of conserved Tyr and Phe residues from sections A and C (Fig. 4), respectively, of rSQS. A preliminary report on the crystal structure of hSQS has revealed that conserved Tyr and Phe residues are located in a functionally significant region of the enzyme. In hSQS the active sites of both half-reactions are located in a large channel in the center of the protein. The conserved Tyr and Phe residues surround an enclosed pocket at one end of the channel (90). Whether these residues include Tyr-171, Phe-286, and Phe-288 awaits a complete report on hSQS crystal structure.

c. The Role of Acidic Residues in Section B. Section B (Fig. 4) of rSQS was one of two regions postulated to contain a binding site for the diphosphate group of FPP. The sequence 219-DXXED-223 in section B, with its three conserved acidic residues, seemed likely to be functionally equivalent

174

TERESE

R, TANSEY AND ISHAIAHU

SHECHTER

to the DDXXD domain previously shown to bind diphosphate moieties in FPP synthase, pentalenene synthase, and 5-epiaristolochene synthase (Section VIA). To test this hypothesis, Asp-219 and Asp-223 were replaced with Asn and Glu. Asp-223 also was converted to the basic Lys, as was Glu-222. The activities of the mutant proteins were compared in assays of the complete reaction (Fig. 5A). All mutations of Asp-2 19 and Asp-223 reduced enzyme activity by more than 98%. Because only squalene production was monitored for this series of mutants, it is not known whether the first or second half-reaction was differentially affected. The lack of significant activity on replacement of the two Asp residues with either Asn and Glu indicates that both the size and the acidic nature of the Asp side chain contribute to Asp-219 and Asp-223 function. The inactivity of the D223K mutant further supports the strict requirement for an acidic residue in this position. In contrast to the severe consequences of mutations at positions 219 and 223, replacement of Glu-222 with Lys reduced enzyme activity by only -50%. Thus the acidic nature of Glu-222 contributes relatively little to rSQS activity. In pentalenene synthase and 5-epiaristolochene synthase, only Dl and D5 of the DDXXD motif coordinate the Mg2+ ions that stabilize binding of the substrate’s diphosphate group (65, 66). The importance of Asp-219 and Asp223, and the lack of a specific requirement for an acidic residue at position 222, suggest that the same structural relationship may exist for Dl and D5 in the DXXED motif of SQS.

VII. SQS Genes and mRNAs A.

Location

and Structure

of SQS Genes

Both human and mouse SQS are encoded by single-copy genes, symbolized FDFTl and Fdftl, respectively (47,QI). The human gene was localized to the 8p22-~23.1 region on the short arm of chromosome 8 using fluorescence in situ hybridization (92). Linkage analysis placed the mouse SQS (mSQS) gene on chromosome 14 (93). The mSQS gene is essential. Mice lacking a functional Fdftl allele die between days 9 and 12 of fetal development. Fdftl +/Fdftl - heterozygotes are apparently normal, with normal hepatic cholesterol synthesis, despite a 50% reduction in hepatic SQS enzyme activity (94). The structures of the human and mouse SQS genes are similar (Fig. 7). Both genes span over 30 kb and contain eight exons. Introns interrupt the protein-coding region at the same position in the two genes. The only major difference between the hSQS and mSQS genes is the larger size of exon eight in the hSQS gene, due to an increase in the length of the 3’ untranslated re-

176

TERESE

R. TANSEY AND ISHAIAHU

SHECHTER

gion (UTR). The hSQS 3’ UTR is -700 bp long and contains three polyadenylation signals (Section V&B), whereas the mSQS 3’ UTR is -350 bp long and appears to contain a single polyadenylation signal (T. Tansey and G. Guan, unpublished results). The organization of the two mammalian SQS genes is quite different from that of the only other metazoan SQS genes reported to date, those of the plants A. thaliana and Nicotiana tabacum. The plant genes contain 13 exons, and are only 3.5 kb (A. thaliana) and 7 kb (N. tabacum) long, due to the presence of relatively small introns. Intron positions with respect to the proteincoding sequence are not generally conserved between the plant and mammalian genes (83, 95).

B. Size and Organ

Distribution

of SQS mRNA

Humans and rats produce SQS mRNAs of several distinct sizes. hSQS mRNAs of 2.0, 1.8, and 1.4 kb are coexpressed in a variety of organs and in the HepG2 cell line (46,48). The sequences of cDNAs representing each size class of hSQS transcript differ only in the length of the 3’ UTR. The 3’ UTR of the 2.0-kb cDNA contains potential polyadenylation signals at nucleotide positions =1.3,1.6, and 2.0 kb. These data suggest that the three hSQS transcripts are encoded by the same gene, and arise by use of alternative polyadenylation sites (46-48, 51). Rats coexpress SQS mRNAs of -3.4 and 1.9 kb in different organs; whether the two mRNAs derive from separate genes or are alternatively processed transcripts of a single gene is not known (62, 96). No specific functional properties have been attributed to the different human or rat transcripts, which appear to be coordinately regulated by cholesterol status (Section VIII,C) (46, 62, 96). The mouse is the only species known to express different forms of SQS mRNA under different cholesterolgenic conditions and in different cell types. The sizes of the transcripts range from ~1.6 to 1.8 kb (91) (T. Tansey, unpublished results). Animals fed cholesterol-supplemented chow produce a larger hepatic transcript than do animals fed normal chow. The nature of the difference in the two size classes of mRNA is not known (13). Cell typespecific differences have been found in the mSQS mRNAs from liver and the BNL-CL.2 embryonic liver cell line. Liver mSQS transcripts have a shorter 5’ UTR than do BNL-CL.2 mSQS transcripts, due to the use of different transcription initiation sites in the two cell types. Almost all mSQS transcripts in liver initiate 30 - 3 1 nucleotides 5’ of the translation initiation codon, whereas most BNL-CL.2 mSQS transcripts initiate 20-70 nucleotides further upstream (Fig. 8) (T. Tansey, unpublished results). SQS mRNA is expressed in most organs that have been examined, although the levels vary widely. In humans, hSQS transcripts are most abundant in testis and skeletal muscle, moderately abundant in brain, and low in liver (48) (B. Collins, personal communication). In rats fed normal chow,

SQUALENE

177

SYNTHASE

60 nt 50 nt

40 nt

30 nt

5’ UTFt: Probr : ENL cL.2: mSQS

Liver:

-30 -51 5 I G&txcmrGAGccccGcGcc~-tg ----AUCNXTGGGAGCBTCCC acacgaCACCCTCGCCCTCTAGGOCTGTCCACTCGGGGCGCGGGTCGT~GCGTTCC-5’-3’P ?? * ?? *. tt

??*

FIG. 8. Nuclease protection assay demonstrating cell type-specific differences in the length of the mSQS mRNA 5’ UTR. An antisense oligonucleotide probe labeled at its 5’ end with a2P was incubated with yeast tRNA (lanes 2-3), with total RNA from BNL-CL.2 cells that had been treated with sterols (S) or lovastatin (L) (lanes 4-6), or with total liver RNA isolated from mice fed normal chow (S) or chow supplemented with lovastatin and cholestyramine (L) (lanes 7-8). The sample in lane 2 was not treated with nuclease to show the size of the intact probe. All other samples were treated with DNase; M, size marker. Shown below the autoradiogram are the sequences of the mSQS 5’ UTR and the antisense oligonucleotide probe. The final six nucleotides at the 3’ end of the probe (lowercase letters) are not complementary to the 5’ UTR. Asterisks indicate the 3’ ends of fragments protected from DNase digestion when hybridized to BNL-CL.2 or liver mRNA. The majority of mSQS transcripts in BNL-CL.2 cells initiate at 5’ of position -51. In contrast, most liver mSQS transcripts begin at -30 or -31. The choice of transcription initiation site does not appear to depend on cholesterol status in either cell type.

testis also expresses the highest level of rSQS mRNA. Moderate amounts are found in skeletal muscle, brain, and liver (96).

VIII. SQS Regulation A. Regulation of SQS Activity and Protein Levels Early evidence that SQS is a regulated enzyme came from studies of cholesterol synthesis in liver homogenates prepared from fasted rats or from rats fed a high-cholesterol diet. In these homogenates, the rate of cholesterol synthesis from mevalonate was much higher than the rate of synthesis from ac-

178

TERESE

R. TANSEY AND ISHAIAHU

SHECXTER

etate (97-99). This reflects the reduced activity of HMGR, the rate-limiting enzyme of the pathway (Fig. 1) (8, 98, 100-104). However, the rate of cholesterol synthesis from mevalonate in the homogenates from the fasted or cholesterol-fed rats was lower than in homogenates from animals fed normal chow, indicating that the activities of enzymes distal to HMGR also are suppressed by fasting and cholesterol feeding (97- 99, 105). The demonstration that cholesterol feeding reduced the conversion of FPP, but not of squalene, to cholesterol by liver homogenates indicated that SQS is one of the distal, regulated enzymes (105). Later studies measured SQS activity directly. Rat liver SQS activity decreases with fasting and increases on refeeding a fat-free diet (99). In cultured human fibroblasts, SQS activity is regulated by the level of cholesterol in the culture medium. Removal of cholesterol in the form of low-density lipoprotein (LDL) from the medium caused an eight-fold increase in SQS activity, which was gradually suppressed by 90% when LDL was reintroduced to the medium (15). SQS activities in rat liver and in the human HepG2 cell line also have been shown to be induced up to 25-fold by HMGR inhibitors (39, 106,107). Generation of an antibody to purified rSQS allowed comparison of rSQS activity and protein levels (39). In all cases that have been examined, changes in rSQS activity are accompanied by changes in the level of the protein. Levels of hepatic rSQS protein are low in animals fed normal diets, undetectable in cholesterol-fed animals, and greatly induced in animals treated with statins (Fig. 3B). Thus, modulation of rSQS activity in response to alteration of cellular cholesterol level is caused primarily by changes in the level of rSQS protein (39, 57).

B. Significance

of SQS Regulation

Although both SQS and HMGR activities are coordinately regulated by cholesterol status, SQS activity changes more gradually and to a lesser degree than HMGR activity. Because changes in the rate of cholesterol synthesis parallel the changes in HMGR activity and not the more gradual modulation of SQS activity, SQS regulation is not considered to have a significant effect on the rate of cholesterol synthesis in response to change in cholesterol status (15). Instead, the importance of SQS regulation is thought to be related to the enzyme’s unique position at the major branchpoint of isoprenoid biogenesis (Fig. 1). SQS commits its substrate FPP specifically to sterol biosynthesis. Because FPP is required by other branchpoint enzymes to make a variety of nonsterol products, modulation of SQS activity has been proposed to play a crucial role in directing the flow of FPP into sterol and nonsterol branches of the pathway (15, 16). Evidence for this role of SQS will be presented in Section VIII,E in relation to the immune response.

179

SQUALENESYNTHASE

C. Sterol Regulation of SQS and in Human HepG2

mRNA in Rats

Cells

Hepatic rSQS mRNA levels differ by over lo-fold in rats on different cholesterolgenic diets. Animals fed cholesterol-rich chow have 70% less hepatic rSQS mRNA than do animals fed normal chow. Animals on diets supplemented with a statin, or with a statin plus cholestyramine, have three and four times more rSQS mRNA, respectively, compared to animals fed normal chow (62,96).Th e c h an g es in rSQS mRNA are accompanied by similar alterations in the amount of rSQS protein. rSQS protein and mRNA levels are regulated primarily by changes in the rate of transcription (62). The human hepatoma cell line HepG2, which retains many of the characteristics of normal hepatocytes (108), has facilitated study of the molecular mechanisms underlying regulation of hSQS activity. The cells can be induced to different cholesterolgenic states by manipulating the culture medium. The standard growth medium for HepG2 cells contains 10% fetal bovine serum (FBS), which provides cholesterol in the form of LDL. Cellular sterol levels can be increased further by adding free sterols (most commonly cholesterol and 25-hydroxycholesterol) to the growth medium. Cellular cholesterol can be depleted by replacing FBS with lipid-depleted serum (LDS), from which LDL-cholesterol and other lipids have been removed by organic extraction, or with LDS plus a statin inhibitor of HMGR. Changes in hSQS mRNA in HepG2 cells grown under different cholesterolgenic conditions have been quantitated. The amount of hSQS mRNA is extremely low in cells grown in medium supplemented with 25-hydroxycholesterol. Cells grown in medium containing FBS or LDS have 8 - 16 times more hSQS mRNA. Addition of lovastatin to the medium increases the amount of hSQS mRNA a further 3 - 4 times. Thus, the hSQS mRNA level is highly responsive to cellular sterol status, undergoing 24- to 64-fold change from fully suppressed by sterol supplementation to fully induced by sterol depletion (46).

D. SQS

Promoter Analysis

1. INTRODUCTION:TRANSCRIPTIONALREGULATION OFSTEROL-RESPONSIVEGENESBYSREBPS The promoter DNA elements and transcription factors that mediate sterol-regulated transcription have been defined for a variety of genes involved in cholesterol homeostasis (109, 110). On cholesterol depletion, genes such as those encoding the LDL receptor, HMG CoA synthase, HMG CoA reductase, and FPP synthase are activated by binding of members of the sterol regulatory element binding protein (SREBP) family of transcription factors to sterol regulatory elements (SREs) in the promoter DNA. SREs

180

TERESER.TANSEYANDISHAIAHUSHECHTER

generally span 10 bp and often include one or two copies of the sequence (TC)CA(T/C) (III), although SREs with little similarity to the consensus also have been described (112). SREBPs are only weak transcriptional activators by themselves. Maximal stimulation of transcription by the SREBPs requires binding of one or more accessory transcription factors to the promoter in the vicinity of the SRE. The accessory transcription factors appear to act by stimulating SREBP binding to the promoter (113 -116). The DNA-binding factors now known to enhance SREBP-mediated transcription include members of the Spl, NF-YCBF, and CREB families. Different genes require different accessory factors, For example, the LDL receptor promoter uses Spl (113), whereas the FPP synthase promoter employs NF-Y (114,117). Activation of the HMG CoA synthase promoter by SREBPs requires both NF-Y and CREB (115,118,119). In addition to being positively regulated by accessory transcription factors, SREBP-mediated transcription of the LDL receptor, FPP synthase, and HMG CoA synthase genes also can be suppressed by the transcription factor YY1(120,12l). SREBPs are synthesized as tripartite precursor proteins that are anchored to the ER membrane by two central membrane-spanning regions. The N- and C-terminal domains face the cytosol. The N-terminal domain is a basic helix-loop helix leucine zipper transcription factor with an acidic transcriptional activation domain. Sterol deprivation sequentially activates a pair of proteases that cleave the precursor in the membrane-spanning region to release the N-terminal fragment, which then enters the nucleus (109). The three known members of the SREBP family are SREBP-la, SREBPlc, and SREBP-2. Although the SREBPs were initially characterized by their ability to regulate genes involved in cholesterol synthesis and uptake, more recently they have been found to control genes responsible for fatty acid and triglyceride biosynthesis as well. Evidence suggests that the two isoforms of SREBP-1 preferentially activate genes associated with fatty acid biosynthesis, whereas SREBP-2 preferentially activates cholesterolgenic genes (109, IlO, 122-124). 2. IDENTIFICATIONOFREGULATORYELEMENTS After demonstrating that production of hSQS mRNA in HepG2 cells is regulated by cholesterol status (Section VIII,C) (46), the sterol-responsive DNA elements in the promoter region of the hSQS gene were identified (125-127). In preparation for these studies, a fragment of genomic DNA containing 1.5 kb of sequence 5’ to the hSQS translation initiation codon was cloned and sequenced. The transcription initiation site was located 98 bp 5’ to the translation initiation codon by primer extension analysis of hSQS mRNA (125,126).

SQUALENE

-198

181

SYNTHASE

M6

M4

Ml

M5

-148

-149

TCTAGAGTGTTATCACGCCAGTCTCCTTCCGCGACTGATTGGCCGGGGTC Y-Box HSS-SRE-1

TTCCTAGTGT Inv-SRE-3

TCAGCGCCCGTC CRE Inv-Y-Box

-99 sRJ?a-1(8/10)

J@ -98

-49

GAGGCCGCAGCTA l? -88

GC Box

2'<&

L

$$g@ Ml1 j;~~~~~~~~,;~~~~~:~ca:~~~i~ z,, .T,...,.rr. i:I_ -13

-48

GCGGCGGGCGGGGCGTCGCCGTACTAGGCCTGCCCCCTGTCCGGCCAGC GC Box

+1

FIG. 9. Sequence of the hSQS promoter and location of promoter mutations. The sequence of 198 nucleotides immediately 5’ of the hSQS transcription initiation site (indicated by the arrow at + 1) is shown. Potential regulatory elements are underlined and in bold. Grey bars span the residues converted en bloc to the corresponding transversion mutations for the experiments shown in Figs. 11 and 13. Upward arrows indicate the 5’ ends of the - 126 and -88 promoter constructs described in Fig. 10. Adapted from Guan et al. (126).

Inspection of the DNA sequence flanking the 5’ end of the hSQS gene revealed three SRE-like sequences, HSS-SRE-1, Inv-SRE-3, and SRE-1(8110), within 200 bp of the transcription initiation site (Fig. 9). The same region also contains sequences similar to known binding sites for three of the factors that cooperate with SREBPs in other genes. The Y-Box and Inv-Y-Box, two potential NF-Y binding sites with opposite orientations (128), flank the Inv-SRE3 sequence. A CAMP response element, or CRE, the binding site for CREB (129), is located adjacent to SRE-1(8/10). A series of potential Spl binding sites is found between - 55 and - 10, with two classic Spl binding sites, the GC boxes, located between -55 and -38. The hSQS promoter contains no match to the YY-1 binding site (120). The hSQS promoter elements required for sterol regulation were initial-

182

TERESE

R. TANSEY AND ISHAIAHU

SHECHTER

pHSS1 kb-Luc

*

5’)

-934

B Fold Activation:

43

51

3.4 ___~

2.4 ~~~

_

r 1+ ‘sl+

-934

-459

-198

5’ End of hSQS Promoter

-126

StemIs Low

-00

DNA

FIG. 10. Analysis of hSQS promoter truncation mutations. (A) Structure of pHSSlkbLuc, the wild-type hSQS promoter-luciferase reporter, and four promoter deletion constructs. (B) Activation of hSQS promoter-luciferase reporter genes by sterols and lovastatin. Promoterreporter plasmids were transiently transfected in HepG2 cells along with the plasmid pCMVp-gal, which constitutively expresses p-galactosidase. Transfected cells were treated with cholesterol and 25-hydroxycholesterol (sterols) or lovastatin (lova) for 24 hr. Luciferase and pgalactosidase activities were measured in cell lysates. To control for differences in transfection efficiency, luciferase activity was normalized to P-galactosidase activity to obtain relative luciferase activity. Fold activation by lovastatin was calculated by dividing the relative luciferase activity in lovastatin-treated cells by that in sterol-treated cells. Adapted from Guan et al. (125).

ly localized by sequential 5’ truncation of hSQS 5’ flanking DNA. The starting point was a 1-kb fragment containing 934 bps of hSQS 5’ flanking sequence plus the first 73 bp of the 5’ UTR. A set of chimeric hSQS promoter-luciferase reporter plasmids was made by fusing either the entire 1-kb DNA fragment or a series of 5’ deletions of this fragment to the luciferase protein coding region (Fig. 1OA). The plasmids were transfected into HepG2 cells, which were then treated with either lovastatin or a mixture of choles-

SQUALENE

SYhTTHASE

183

terol plus 25-hydroxycholesterol in medium containing LDS. Luciferase activity was measured in cell lysates (Fig. 10B). The plasmid pHSSlkb-Luc, which contains the complete 1-kb promoter fragment, produced ~48 times higher luciferase activity in cells treated with lovastatin compared to cells treated with sterols, indicating higher promoter activation by sterol depletion. Lovastatin-induced activation of truncated constructs containing 459 and 198 bp of hSQS 5’ flanking DNA was similar to that of pHSSlkb-Luc. However, additional deletion of 72 bp, leaving 126 bp of 5’ flanking DNA, caused a dramatic decrease in the response to lovastatin. A final deletion that left only 88 bp of 5’ flanking DNA had only limited additional effect on promoter activation. These results localize essential hSQS sterol-responsive regulatory elements to the small 72-bp region between - 198 and - 12 7. None of the deletions affected the low level of luciferase expression in sterol-treated cells, suggesting that the small segment of hSQS DNA between -88 and + 73 contains sequences sufficient for basal transcription (125). The hSQS promoter deletion analysis suggests that one or more of the three elements located between - 198 and - 127 (HSS-SRE-1, Y-Box, and Inv-SRE-3) play a critical role in sterol regulation of hSQS transcription (Fig. 9). Regulatory DNA elements were defined more precisely by scanning mutagenesis, in which small segments of promoter sequence in the promoterreporter plasmid pHSSlkb-Luc were successively replaced by mutations (regions Ml-M9, Fig. 9). Promoter activities of the mutant constructs were assessed in transiently transfected HepG2 cells grown in medium containing sterols or lovastatin. Activation of the hSQS promoter in response to lovastatin was decreased only by mutation of regions containing sequences similar to promoter elements involved in sterol regulation of other genes (Fig. 11A). Mutations of HSS-SRE-1, the Y-Box, Inv-SRE-3, or SRE-l(8110) decreased activation by -lo-300/o, whereas mutations that affected the Inv-Y-Box (M5) or the Inv-Y-Box and the CRE (M6) had more severe consequences, decreasing promoter activation by =60%. The only recognized element whose mutation had no effect was that of 5’ GC box (M9). Additional experiments have shown that at least one of the two GC boxes must be intact for full promoter activation in response to sterol depletion (G. Guan, unpublished results). As was observed for the deletion mutations, none of the replacement mutations affected basal transcription in the presence of sterols (126). The advantage of scanning mutagenesis over progressive 5’ deletion analysis can be seen in the results with mutant M6, which includes the InvY-Box + CRE. Although this region contributes significantly to promoter activity, its effects were masked in the deletion analysis once upstream elements had been removed. Thus, the - 126 and -88 promoter deletion-luciferase reporter constructs, which, respectively, contained and lacked the Inv-Y-Box + CRE, had similar low activities in lovastatin-treated cells (Fig. 10).

TERESE

R. TANSEY AND ISHAIAHU

SHECHTER

60

WT

Ml

M2

M3

M4

M5

M6

M7

M8

M9

hSQS Promoter Mutant

hSQS Promoter Mutant FIG. 11. Identification of hSQS promoter elements by scanning mutagenesis. Luciferase reporter plasmids with 1 kb of wild-type hSQS 5’ flanking DNA (WI) or 1 kb of 5’ flanking DNA carrying the mutagenized regions shown in Fig. 9 were transiently transfected into HepG2 cells and treated with sterols or lovastatin as described in the legend to Fig. 10. Fold activation by lovastatin was calculated by dividing the relative luciferase activity in lovastatin-treated cells by that in sterol-treated cells. Results are expressed as percentage of the fold activation of the WT hSQS promoter. Basal promoter activity in the presence of sterols was similar for all constructs. Potential regulatory elements within the mutagenized regions are shown above the bars. (A) Single-segment mutations. (B) Multiple-segment mutations. Adapted from Guan et al. (126).

SQUALENESYNTHASE

185

The effects of mutations in the hSQS SREs and the Inv-Y-Box + CRE implicate the SREBPs, NF-Y, and/or CREB as important transcription factors regulating the response of the hSQS gene to sterols. The larger decrease in activation caused by mutation of the Inv-Y-Box + CRE (M6), compared to mutations that affect individual SREs (Ml, M4, M7) (Fig. llA), suggests that NF-Y and/or CREB are essential for maximal activation by the SREBPs, as has been shown for several other sterol-regulated genes (Section VIII,D,l). Of the three hSQS SREs, SRE-l(8110) contributes least to promoter activation by sterol depletion. None of the replacement mutations studied in Fig. 11A completely prevented increased hSQS promoter activity following sterol depletion, suggesting that multiple cis elements contribute additively to the response. Evidence supporting this hypothesis was obtained by combining single-region mutations to create promoters with mutations covering two or three regions (Fig. 11B). All combinations of HSS-SRE-1, Inv-SRE-3, and Inv-Y-Box + CRE double mutations decreased promoter activation to a greater degree than any of the single mutations, indicating that each of these regions can make an independent contribution to promoter activity. However, once HSS-SRE-1 and the Inv-Y-Box + CRE have been mutated, Inv-SRE-3 is unable to exert a stimulatory effect (compare double-mutant Ml + M6 with triple-mutant Ml + M4 + M6) (126). 3. DIFFERENTIALEFFECTSOFSREBP-~~ANDSREBP-2 ONSQSPROMOTERACTMTY a. Response of the hSQS Promoter. Transcription factor-DNA binding assays gave the first clue that the SREs associated with the hSQS gene might not respond equivalently to different SREBPs. In gel retardation studies with oligonucleotides containing HSS-SRE-1, Inv-SRE-3, or SRE-l(8110) sequences, SREBP-la bound with high affinity only to the HSS-SRE-1 probe, while SREBP-2 bound relatively strongly to all three SREs (126). The differential binding of SREBP-la and SREBP-2 to hSQS SREs led to the investigation of possible differences in the activation of the hSQS promoter by the two transcription factors (126,130). In these experiments wildtype or mutant hSQS p romoter-luciferase reporter constructs were cotransfected into HepG2 cells with increasing amounts of expression vector for the transcriptionally active, N-terminal fragment of SREBP-la or SREBP-2. A wide range of SREBP vector concentrations was used to ensure that the response to expression of transfected SREBPs was within the physiological range of endogenous SREBPs in sterol-depleted cells. Cotransfected cells were grown in the presence of sterols to suppress processing of endogenous SREBPs, thus ensuring that activation of the luciferase reporter gene would be due strictly to expression of the transfected SREBPs. Luciferase activities were measured 24 hr after transfection.

186

TERESE

R. TANSEY

AND

ISHAIAHU

--

SHECHTER

cSRE

b

4=

kg

$+ &P

hSQS Promoter Mutant

FIG. 12. Activation of the wild-type hSQS promoter by SREBP-la and SREBP-2. (A) HepG2 cells were transiently transfected with pHSSlkb-Luc, the wild-type hSQS promotere-luciferase reporter plasmid, and increasing amounts of a plasmid that constitutively expresses the nuclear, transcriptionally active form of either SREBP-la or SREBP-2. Cells were grown in medium supplemented with cholesterol and 25-hydroxycholesterol to suppress activation of endogenous SREBPs. All transfections included the plasmid pCMV-B-gal to normalize for differences in transfection efficiency. Fold activation is the relative luciferase activity in sterol-treated cells cotransfected with the hSQS-luciferase reporter and SREBP expression plasmids divided by the relative luciferase activity in sterol-treated cells transfected with the reporter plasmid alone. (B) Activation of the wild-type hSQS promoter by endogenous SREBPs. To compare the level of activation by SREBP expression vector with the level of activation achieved by lovastatin treatment, an aliquot of cells transfected with pHSSlkb-Luc, but not with an SREBP expression vector, was treated with lovastatin. It was found that 89 ng SREBP-la DNA and 46 ng SREBP-2 DNA activate pHSSlkb-Luc to =50% of the level obtained by lovastatin treatment. These amounts of SREBP expression vectors were used in the experiments shown in Fig. 13 to illustrate the effects of the SREBPs within a physiological response range. From Guan et al. (130).

The wild-type hSQS promoter in the plasmid pHSSlkb-Luc was activated by both SREBP-la and SREBP-2 (Fig. 12A). The dose-response curves for the two transcription factors were similar, but SREBP-2 was somewhat more potent, requiring less expression vector to achieve the same level of activation as SREBP-la. Whether the increased potency of SREBP-2 reflects

SQUALENE

SYNTHASE

187

a physiological difference between the two transcription factors is unclear. The different response to SREBP-la and SREBP-2 in the transfection experiments may be related to the differences in affinity for the hSQS SRE elements observed in in vitro DNA binding assays (126). However, because the amount of SREBP-la and SREBP-2 expressed in the transfected cells was not quantitated, the greater potency of SREBP-2 may be due to more efficient expression of the transfected SREBP-2 vector or to greater stability of the truncated SREBP-2 protein. In contrast to the ability of pHSSlkb-Luc to respond to SREBP-2 in HepG2 cells, the endogenous hSQS promoter in a HeLa cell line that inducibly expresses the nuclear form of SREBP-2 did not respond to SREBP2. This conclusion was based on the relatively small increase in hSQS mRNA 6 hr after induction of SREBP-2 expression, a time when the mRNAs for HMGR and the LDL receptor had increased two- to three-fold (131). Differences in experimental design make it difficult to assess the significance of the different responses of pHSSlkb-Luc and the endogenous hSQS promoter to SREBP-2. For example, it is unclear whether activation of pHSSlkbLuc measured 24 hr after cotransfection with the SREBP-2 expression vector can be compared in a meaningful way to the endogenous hSQS mRNA level measured 6 hr after SREBP-2 induction. Experiments similar to that shown in Fig. 12 for the wild-type hSQS promoter were performed with a variety of promoter mutants (130). This review focuses on the consequences of mutations in the hSQS SREs, the Inv-Y-Box + CRE, and the GC boxes. To illustrate the effects of the SREBPs within a physiological response range, Fig. 13 compares the activation of various hSQS promoter constructs in cells cotransfected with the amount of SREBP expression vector that activated the wild-type hSQS promoter in pHSSlkbLuc (Fig. 12A) to 50% of the level obtained by lovastatin treatment (Fig. 12B). Although the wild-type hSQS promoter-luciferase reporter responded similarly to SREBP-la and SREBP-2 (Fig. 12), transcriptional activation by the two SREBPs was differentially affected by mutation of the hSQS SRE elements (Fig. 13). Promoter activation by SREBP-la was reduced by =4060% on mutation of either HSS-SRE-1 or Inv-SRE-3, and was decreased by =90% when both of these elements were mutated. SREBP-la lost all ability to activate transcription when the three hSQS SREs were mutated. In striking contrast, promoter activation by SREBP-2 was unaffected by mutation of HSS-SRE-1, and actually was enhanced by mutation of Inv-SRE-3. Mutation of both HSS-SRE-1 and Inv-SRE-3 reduced SREBP-2-mediated activation by only =400/o. Mutation of the three hSQS SREs reduced activation by SREBP2 by ~85%. These results indicate that SREBP-la activates transcription primarily through additive interactions with HSS-SRE-1 and Inv-SRE-3. Both of these elements must be intact for the maximal response to SREBP-la.

188

g '3 p 'J k"

TERESE

R. TANSEY

AND

ISHAIAHU

SHECHTER

60 50 40

z al 30 : 2 20 10

FIG. 13. Differential effect of SREBP-la and SREBP-2 on mutant hSQS promoter activity. HepG2 cells were transiently transfected with luciferase reporter plasmids containing the wildtype hSQS promoter (WT) or mutant promoters with one or more of the mutations shown in Fig. 9. The promoter-reporter plasmids were cotransfected with increasing amounts of SREBP expression vector and treated with sterols as described in the legend to Fig. 12. The results obtained with cells transfected with 89 ng SREBP-la or 46 ng SREBP-2 DNA are shown. These amounts of SREBP DNA activated the wild-type hSQS p romoter to 50% of the level obtained by lovastatin treatment of cells transfected with the wild-type hSQS-luciferase reporter vector alone (Fig. 12). Fold activation was calculated by dividing the relative luciferase activity in the cotransfected cells by that in cells transfected with the promoter-reporter plasmid alone. Results are expressed as percentage of the fold activation obtained with the wild-type hSQS promoter in lovastatin-treated cells. To compare the level of activation by transfected SREBPs with that achieved by lovastatin treatment, an aliquot of cells transfected with each promoterreporter plasmid alone was treated with lovastatin (black bars). Half-maximal percentage activation is plotted (Lova,sa). Regulatory elements within the mutagenized regions are shown above the bars. Adapted from Guan et a2. (130).

SREBP-2, on the other hand, can activate transcription through interactions with each of the three hSQS SREs. Approximately half of SREBP-2-mediated transcriptional activation appears to result from interaction with SRE-l(81 lo), which contributes little to the SREBP-la response. This analysis demonstrates that SREBP-la and SREBP-2 differ substantially in their mode of promoter activation through use of the three hSQS SREs. Similar experiments using mutations of the Inv-Y-Box + CRE and the GC boxes were performed to determine if SREBP-la and SREBP-2 differ in their requirements for accessory transcription factors (Fig. 13). Mutation of the Inv-Y-Box + CRE decreased the transcriptional response to SREBP-la by -70% and to SREBP-2 by 40%. Mutation of the 5’ or 3’ GC box separately

SQUALENE

SYNTHASE

189

(Ml0 and Mll, Fig. 9) had little effect on activation by either factor (results not shown) (130). A mutation encompassing both CC boxes decreased activation by SREBP-2 by 40%, but had no effect on activation by SREBP-la (Fig. 13). Thus, at least one intact CC box is required for maximal transcriptional activation by SREBP-2, but not by SREBP-la. These results suggest that both SREBP-la and SREBP-2 require NF-Y and/or CREB for full hSQS promoter activation, but only SREBP-2 also requires Spl. The effects of promoter mutations on promoter activation by transfected SREBPs can be compared with their effects on promoter activation by endogenous SREBPs in response to sterol depletion by examining the halfmaximal activation level obtained with the mutant constructs in lovastatintreated cells in the absence of cotransfected SREBP expression vector (Fig. 13, black bars). The endogenous response to mutation of the SREs and the Inv-Y-Box + CRE is similar to the SREBP-la response, suggesting that SREBP-la may be more important than SREBP-2 in mediating transcriptional activation of the hSQS promoter in response to sterol depletion in HepG2 cells. The experiments described b. Response of Nonhuman SQS Promoters. in the preceding section indicate that the wild-type hSQS promoter in the promoter-reporter construct pHSSlkb-Luc can be activated by SREBP-la or SREBP-2 (Fig. 12). Th e endogenous mSQS promoter also responds to both factors, based on the increase in hepatic mSQS mRNA levels in mice that overexpress SREBP-la or SREBP-2 in liver (122, 124, 132). In contrast to these results, the endogenous SQS promoter in a Chinese hamster ovary cell line responds quite differently to SREBP-la and SREBP-2. This observation was made in a study that compared the ability of the two transcription factors to activate a variety of genes involved in cholesterol and fatty acid biosynthesis (133). The SQS g ene was unique in that it was activated only by SREBP-2. The inability of SREBP-la to activate the SQS promoter in the CHO cell line cannot currently be reconciled with its clear stimulatory effects on mouse and human SQS promoters. The contrasting results may be due to differences in experimental design or to physiologically relevant species differences in SQS promoter response.

E. Regulation of SQS by Lipopolysaccharide and Cytokines In addition to responding to dietary cholesterol levels, cholesterol metabolism is substantially altered by infection or by administration of lipopolysaccharide (LPS), which mimics infection by gram-negative bacteria.

190

TERESE

02

4

8

Time (hours)

16

R. TANSEY

02

AND

4

ISHAIAHU

8

SHECHTER

16

Time (hours)

FIG. 14. Effect of LPS on hepatic SQS mRNA and enzyme activity. Syrian hamsters were injected with saline or LPS (100 *g/100 g body weight). At the indicated times, animals were killed and livers were removed. SQS mRNA (A) and activity (B) were determined. Data are presented as mean + SEM; n = 5 for each time point; *I’< 0.01; **P< 0.001. From Memon et al. (147).

Many of the changes induced by LPS can also be triggered by the proinflammatory cytokines, tumor necrosis factor-o (TNF-a) and interleukin-1 R (IL-ll3) (134). The effects of LPS and cytokines on cholesterol metabolism differ in primates and rodents. In primates, LPS and TNF-o, but not IL-ll3, decrease serum cholesterol (135-137). IL-1R inhibits cholesterol synthesis in HepG2 cells (138). In rodents, serum cholesterol and hepatic cholesterol synthesis increase in response to LPS, TNF-a, and IL-l R (139 - 142). Although LPS produces opposite effects on overall cholesterol homeostasis in primates and rodents, one common feature of the response seems to be differential or discordant regulation of the cholesterol metabolic genes. Thus, in HepG2 cells, TNF-a and IL-ll3 increase the production of LDL receptor mRNA (138, 143--145), but have little effect on SQS mRNA levels (145). More extensive evidence for discordant regulation of cholesterolgenic genes has been found in Syrian hamsters, in which administration of LPS has minimal effects on HMG CoA synthase, FPP synthase, and the LDL receptor, but triggers rapid and opposite changes in HMGR and SQS (141, 146, 147). Hepatic HMGR activity increases S- to lo-fold after LPS administration, due to increased transcription of HMGR mRNA (141,146). In contrast, hepatic SQS mRNA levels drop precipitously following LPS treatment. By 4 hr after LPS injection, SQS mRNA levels decline to 12% of control levels; by 8 hr, SQS mRNA is less than 5% of control (Fig. 14A). SQS activity (Fig. 14B) and protein levels also decrease with LPS treatment, although not as much

SQUALENE

191

SYNTHASE

as the mRNA. Similar results were obtained with IL-1 R or with a mixture of IL-1R and TNF-a (147). The reduced activity of SQS may explain the initially puzzling observation that the rate of hepatic cholesterol synthesis in Syrian hamsters increases by only e-fold following LPS administration, despite the 8- to lo-fold increase in HMGR activity (141). The decrease in SQS activity, coupled with the increase in HMGR activity, may direct the isoprenoid intermediates between mevalonate and squalene into nonsterol pathways, whose products may be important during the acute phase response to infectious and inflammatory stimuli (147). A possible mode of action of LPS and the cytokines is through a signaling cascade that activates the transcription factor NF-KB (148). A sequence that resembles an NF-KB binding site is found in the hSQS promoter between nucleotides - 154 and - 145. Whether NF-KB mediates the dramatic effects of LPS, IL-IR, and TNF-cx on SQS expression in Syrian hamsters is the subject of ongoing research.

ACKNOWLEDGMENT Much of the work describing SQS protein structure and function and SQS transcriptional regulation has been supported by National Institutes ofHealth, grants HL50628 and HL48540.

REFERENCES 1. P A. Edwards and R. Davis, In “Biochemistry of Lipids, Lipoproteins, and Membranes” (D. E. Vance and J. E. Vance, eds.), pp. 34 l-362. Elsevier, Amsterdam and New York, 1996. 2. D. W. Russell and K. D. Setchell, Biochemistq 31,4737 (1992). 3. J. A. Porter, S. C. Ekker, W. J. Park, D. P. van Kessler, K. E. Young, C. H. Chen, Y. Ma, A. S. Woods, R. J. Cotter, E. V. Koonin, and P. A. Beachy, Cell 86,21 (1996). 4. J. A. Porter, K. E. Young, and P. A. Beachy, Science 274,255 (1996). 5. R. V. Farese, Jr. and J. Herz, Trends Genet. 14,115 (1998). 6. D. M. Small and G. G. Shipley, Science 185,222 (1974). 7. M. K. Cooper, J. A. Porter, K. E. Young, and P. A. Beachy, Science 280,1603 (1998). 8. J. L. Goldstein and M. S. Brown, Nature (London) 343,425 (1990). 9. C. D. Poulter and H. C. Rilling, In “Biosynthesis of Isoprenoid Compounds” (J. W. Porter and S. L. Spurgeon, eds.), p. 414. Wiley, New York, 1981. 10. C. D. Poulter, Act. Chem. Res. 23, 70 (1990). 11. B. D. Levy, N. A. Petasis, and C. N. Serhan, Nature (London) 389,985 (1997). 12. T. Spencer, Ace. Chem. Res. 27,83 (1994). 13. D. J. Peet, S. D. Turley, W. Ma., B. A. Janowski, J. M. Lobaccaro, R. E. Hammer, and D. J. Mangelsdorf, Cell 93, 693 (1998). 14. S. A. Kliewer, J. M. Lehmann, and T. M. Willson, Science 284, 757 (1999). 15. J. R. Faust, J. L. Goldstein, and M. S. Brown, Proc. N&l. Acad. Sci. U.S.A. 76, 5018 (1979).

192

TERESE

R. TANSEY AND ISHAIAHU

SHECHTER

16. M. S. Brown and J. L. Goldstein, J. Lipid Res. 21,505 (1980). 17. C. P. Ciosek, Jr., D. R. Magnin, T. W. Harrity, J. V. Logan, J. K. Dickson, Jr., E. M. Gordon, K. A. Hamilton, K. G. Jolibois, L. K. Kunselman, R. M. Lawrence, et al.,J. Biol. Chem. 268, 24832 (1993). 18. J. D. Bergstrom, C. Dufresne, G. F. Bills, M. Nallin-Omstead, and K. Byrne, Annu. Rev. Microbial. 49,607 (1995). 19. A. Nadin and K. Nicolaou, Angers. Chem. Int. Ed. (EngZ.j 35,1623 (1996). 20. N. S. Watson and P. A. Procopiou, Prog. Med. Chem. 33,331(1996). 21. S. Biller, K. Neuenschwander, M. Ponpipom, and C. D. Poulter, Curr. Pharm. Des. 2, 1 (1996). 22. I. Schechter, P. Gu, G. Jiang, T. J. Onofrey, R. 0. Cann, A. Castro, and T. Spencer, Bioorg. Med. Chem. Leti. 2585-2588 (1996). 23. H. J. Harwood, Jr., E. G. Barbacci-Tobin, S. F Petras, S. Lindsey, and L. D. Pellarin, Biochem. Pharmacol. 53,839 (1997). 24. E. Beytia, A. A. Qureshi, and J. W. Porter,J. Biol. Chem. 248,1856 (1973). 25. G. Krishna, D. H. Feldbruegge, and J. W. Porter, Biochem. Biophys. Res. Commun. 14,363 (1964). 26. H. C. Rilling,]. Biol. Chem. 241,3233 (1966). 27. G. Popjak, J. Edmond, K. Clifford, and V. Williams,J. Biol. Chem. 244, 1897 (1969). 28. W. W. Epstein and H. C. Rilling, J. Biol. Chem. 245,4597 (1970). 29. H. C. rilling,]. Lipid Res. 11,480 (1970). 30. J. Edmond, G. Popjak, S. M. Wong, and V. P. Williams, J. Biol. Chem. 246, 6254 (1971). 31. H. C. Rilling, C. D. Poulter, W. W. Epstein, and B. Larsen, J. Am. Chem. Sot. 93, 1783 (1971). 32. R. M. Coates and W. H. Robinson,J. Am. Chem Sot. 93,1785 (1971). 33. L. J. Altman, R. C. Kowerski, and H. C. Rilling,J. Am. Chem. Sot. 93,1782 (1971). 34. E. E. Tamelen and M. A. Schwartz,J. Am. Chem. Sot. 93,178O (1971). 35. I. Shechter and K. Bloch,J. Biol. Chem. 246, 7690 (1971). 36. W. S. Agnew and G. Popjak,J. Biol. Chem. 253,4574 (1978). 37. G. Kuswik-Rabiega and H. C. Rilling,J. Biol. Chem. 262,1505 (1987). 38. K. Sasiak and H. C. Rilling, Arch. Biochem. Biqnhys. 260, 622 (1988). 39. I. Shechter, E. Khnger, M. L. Rucker, R. G. Engstrom, J. A. Spirito, M. A. Islam, B. R. (1992). Boettcher, and D. B. Weinstein, J. Biol. Chem. 267,8628 40. L. W. Parks and W. M. Casey, Annu. Reu. Microbial. 49, 95 (1995). 41. l? Karst and F. Lacroute, Mol. Gen. Genet. 154,269 (1977). 42. B. M’Baya, M. Fegueur, M. Servouse, and F. Karst, Lipids 24,102O (1989). 43. S. M. Jennings, Y. H. Tsay, T. M. Fisch, and G. W. Robinson, Proc. N&l. Acad. Sci. U.S.A. (1991). 88,6038 44. M. Fegueur, L. Richard, A. D. Charles, and F Karst, Cut-r Genet. 20,365 (1991). 45. T. L. McKenzie, G. Jiang, J. R. Straubhaar, D. G. Conrad, and I. Shechter, J. Biol. Chem. 267,21368 (1992). 46. G. Jiang, T L. McKenzie, D. G. Conrad, and I. Shechter,J. Biol. Chem. 268,12818 (1993). 47. G. W. Robinson, Y. H. Tsay, B. K. Kienzle, C. A. Smith-Monroy, and R. W. Bishop, Mol. Cell Biol. 13,2706 (1993). 48. C. Summers, F. Karst, and A. D. Charles, Gene 136,185 (1993). 49. D. Zhang and C. D. Poulter, Anal. Biochem. 213,356 (1993). 50. P. V. LoGrasso, D. A. Soltis, and B. R. Boettcher, Arch. Biochem. Biophys. 307,193 (1993). 51. D.A. Soltis, G. McMahon, S. L. Caplan, D. A. Dudas, H. A. Chamberhn, A. Vattay, D. Dottavio, M. L. Rucker, R. G. Engstrom, S. A. Cornell-Kennon, et al., Arch. Biochem. Biophys. 316, 713 (1995).

SQUALENE

SYNTHASE

193

52. J. F. Thompson, D. E. Danley, S. Mazzalupo, P. M. Milos, M. E. Lira, and H. J. Harwood, (1998). Jr., Arch. Biochem. Biophys. 350,283 53. P. Gu, Y. Ishii, T. A. Spencer, and I. Shechter,J. Biol. Chem. 273, 12515 (1998). 54. S. K. Krisans, Ann. N.Y. Acud. Sci. 804, 142 (1996). 55. L. H. Cohen, M. Griffioen, C. W. van Roermund, and R. J. Wanders, Biochim. Biophys. A&z 1126, 114 (1992). 56. J. Ericsson, E. L. Appelkvist, A. Thelin, T. Chojnacki, and G. Dallner, J. Biol. Chem. 267, 18708 (1992). 57. K. D. Stamellos, J. E. Shackelford, I. Shechter, G. Jiang, D. Conrad, G. A. Keller, and S. K. Krisans, J. Biol. Chem. 268, 12825 (1993). 58. G. Popjak and W. S. Agnew, Mol. Cell. Biochem. 27,97 (1979). 59. H. C. Rilling, Biochem. Sot. Trans. 13,997 (1985). 60. K. A. Mookhtiar, S. S. Kalinowski, D. Zhang, and C. D. Poulter,J. Biol. Chem. 269,112Ol (1994). 61. M. Jarstfer, B. Blagg, D. Rogers, and C. D. Poulter, J. Am. Chem. Sot. 118,13089 (1996). 62. I. Shechter, G. Guan, and B. Boettcher, In “Isoprenoids Including Carotenoids and Steroids” (D. Barton, K. Nakanishi, 0. Me&Cohn, and D. E. Cane, eds.), p. 245. Elsevier, New York, 1999. 63. L. C. Tarshis, M. Yan, C. D. Poulter, and J. C. Sacchettini, Biochemistry 33,10871 (1994). 64. L. C. Tarshis, P J. Proteau, B. A. Kellogg, J. C. Sacchettini, and C. D. Poulter, Proc. Nutl. Acud. Sci. U.S.A. 93, 15018 (1996). 65. C. A. Lesburg, G. Zhai, D. E. Cane, and D. W. Christianson, Science 277,182O (1997). 66. C. M. Starks, K. Back, J. Chappell, and J. P. Noel, Science 277, 1815 (1997). 67. B. A. Kellogg and C. D. Poulter, Cum. Opin. Chem. Biol. l, 570 (1997). 68. J. C. Sacchettini and C. D. Poulter, Science 277,1788 (1997). 69. G. A. Armstrong and J. E. Hearst, FASEBJ. 10,228 (1996). 70. B. L. Jones and J. W. Porter, Methods Enzymol. 110,209 (1985). 71. 0. Dogbo, A. Laferriere, A. D’Harlingue, and B. Camara, Proc. Nutl. Acud. Sci. U.S.A. 85, 7054 (1988). 72. G. E. Bartley, P. V. Viitanen, K. 0. Bacot, and P A. Scolnik,]. Biol. Chem. 267,5036 (1992). (1992). 73. D. Chamovitz, N. Misawa, G. Sandmann, and J. Hirschberg, FEBS Lett. 296,305 74. J. van Lintig, R. Welsch, M. Bonk, G. Giuliano, A. Batschauer, and H. Kleinig, PZuntJ. 12, 625 (1997). 75. M. N. Ashby and P. A. Edwards,J. Bid. Chem. 265,13157 (1990). 76. M. N. Ashby, S. Y. Kutsunai, S. Ackerman, A. Tzagoloff, and l? A. Edwards, J. Biol. Chem. 267,4128 (1992). 77. P. E Marrero, C. D. Poulter, and P. A. Edwards,]. Biol. Chem. 267,21873 (1992). 78. A. Joly and P. A. Edwards,J. Biol. Chem. 268,26983 (1993). 79. L. Song and C. D. Poulter, Proc. Natl. Acad. Sci. U.S.A. 91,3044 (1994). 80. C.-I. Brand&n and J. Tooze, “Introduction to Protein Structure.” Garland, New York, 1999. 81. J. M. Wahlberg and M. Spiess, I. Cell Biol. 137,555 (1997). 82. B. Martoglio and B. Dobberstein, Trends Cell Biol. 8,410 (1998). 83. R. Kribii, M. An-o, A. Del Arco, V. Gonzalez, L. Balcells, D. Delourme, A. Ferrer, F. Karst, and A. Boronat, Eur. J. Biochem. 249,61 (1997). 84. K. U. Wendt, K. Poralla, and G. E. Schulz, Science 277,1811 (1997). 85. K. U. Wendt, A. Lenhart, and G. E. Schulz,J. Mol. Biol. 286,175 (1999). 86. D. A. Dougherty, Science 271,163 (1996). 87. J. C. Ma and D. A. Dougherty, Chem. Rev. 97,1303 (1997). 88. C. Jenson and W. Jorgenson,J. Am. Chem. Sot. 119,10846 (1997).

194

TIZRESE R. TANSEY AND ISHAIAHU

SHECHTER

89. S. Mecozzi, A. l? West, Jr., and D. A. Dougherty, Proc. Natl. Acad. Sci. U.S.A. 93, 10566 (1996). 90. J. Pandit, D. E. Danley, G. K. Schulte, S. M. Mazzalupo, T. A. Pauly, C. M. Hayward, E. S. Hamanaka, J. F. Thompson, and H. J. Harwood, FASEB]. 13, Al338 (1999). 91. T Inoue, T. Osumi, and S. Hata, Biochim. Biuphys. Acta I260,49 (1995). 92. I. Shechter, D. G. Conrad, I. Hart, R. C. Berger, T. L. McKenzie, J. Bleskan, and D. Patterson, Genomics 20, 116 (1994). 93. C. L. Welch, Y. R. Xia, I. Shechter, R. Farese, M. Mehrabian, S. Mehdizadeh, C. H. Warden, and A. J. Lusis,J. Lipid Res. 37, 1406 (1996). 94. R. Tozawa, S. Ishibashi, J. Osuga, H. Yagyu, T. Oka, Z. Chen, K. Ohashi, S. Perrey, F. Shionoiri, N. Yahagi, K. Harada, T Gotoda, Y. Yazaki, and N. Yamada,J. BioZ. Ch.em. 274, 30843 (1999). 95. T P. Devarenne, D. H. Shin, K. Back, S. Yin, and J. Chappell, Arch. Biochem. Biophys. 349, 205 (1998). 96. R. K. Keller, A. Cannons, E Vilsaint, Z. Zhao, and G. C. Ness, Arch. Biochem. Biqwhys. 302,304 (1993). 97. N. Bucher, K. McGarrahan, E. Gould, and A. Loud, J. Biol. Chem. 234,262 (1959). 98. M. Siperstein and M. Guest,]. Clin. Invest. 39,642 (1960). 99. L. L. Slakey, M. C. Craig, E. Beytia, A. Briedis, D. H. Feldbruegge, R. E. Dugan, A. A. Qureshi, C. Subbarayan, and J. W. Porter,J. Biol. Chem. 247,3014 (1972). 100. R. G. Gould and G. Popjak, Biochem. J. 66,51P (1957). 101. N.L.R. Bucher, P. Overath, and F. Lynen, Biochim. Biuphys. Acta 49,491 (1960). 102. M. D. Siperstein and V. M. Fagan,J. Biol. Chem. 241,602 (1966). 103. T. C. Linn,J. Biol. Chem. 242,990 (1967). 104. L. W. White and H. Rudney, Biochemistry 9,2725 (1970). 105. R. G. Gould and E. A. Swyryd,]. Lipid Res. 7,698 (1966). 106. L. H. Cohen, A. M. Griffioen, R. J. Wanders, C. W. Van Roermund, C. M. Huysmans, and H. M. Princen, Biochem. Biophys. Res. Commun. 138,335 (1986). 107. L. H. Cohen, E. van Miert, and M. Griffioen, Biochim. Biophys. Actu 1002,69 (1989). 108. N. B. Javitt, FASEBJ. 4,161 (1990). 109. M. S. Brown and J. L. Goldstein, Cell 89,331 (1997). 110. J. D. Horton and I. Shimomura, Cum Opin. Lipidol. 10,143 (1999). 111. M. M. Magana, S. S. Lin, K. A. Dooley, and T. F. Osborne, J. Lipid Res. 38, 1630 (1997). 112. D. E. Tabor, J. B. Kim, B. M. Spiegelman, and P. A. Edwards,]. Biol. Chem. 274,20603 (1999). 113. H. B. Sanchez, L. Yieh, and T. F. Osborne,]. Biol. Chem. 270,116l (1995). 114. S. M. Jackson, J. Ericsson, R. Mantovani, and P. A. Edwards, 1. Lipid Res. 39,767 (1998). 115. K. A. Dooley, S. Millinder, and T. F. Osborne,J. Biol. Chem. 273, 1349 (1998). 116. P. A. Edwards and J. Ericsson, Cum Opin. Lipidol. 9,433 (1998). 117. J. Ericsson, S. M. Jackson, and F! A. Edwards,J. Biol. Chem. 271,24359 (1996). 118. S. M. Jackson, J. Ericsson, T. F. Osborne, and P. A. Edwards, I. Biol. Chem. 270, 21445 (1995). 119. K. A. Dooley, M. K. Bennett, and T E Osborne,]. Biol. Chem. 274,5285 (1999). 120. J. Ericsson, A. Usheva, and P. A. Edwards, J. Biol. Chem. 274,14508 (1999). 121. M. K. Bennett, T. T. Ngo, J. N. Athanikar, J. M. Rosenfeld, and T. F. Osborne,]. Biol. Chem. 274, 13025 (1999). 122. H. Shimano, J. D. Horton, R. E. Hammer, I. Shimomura, M. S. Brown, and J. L. Goldstein, J. Clin. Znuest. 98, 1575 (1996). 123. H. Shimano, J. D. Horton, I. Shimomura, R. E. Hammer, M. S. Brown, and J. L. Goldstein, et al.,]. Clin. Invest. 99,846 (1997).

SQUALENE

SYNTHASE

195

124. J. D. Horton, I. Shimomura, M. S. Brown, R. E. Hammer, J. L. Goldstein, and H. Shimano, J. Clin. Invest. lOl, 2331 (1998). 125. G. Guan, G. Jiang, R. L. Koch, and I. Shechter,]. Biol. Chem. 270,21958 (1995). 126. G. Guan, P. H. Dai, T. F. Osborne, J. B. Kim, and I. Shechter, J. Biol. Chem. 272, 10295 (1997). 127. J. Inoue, R. Sato, and M. Maeda,]. Biochem. (Tokyo) 123, 1191 (1998). 128. W. Bi, L. Wu, F. Coustry, B. de Crombrugghe, and S. N. Maity,J. Biol. Chem. 272,26562 (1997). 129. J. S. Fink, M. Verhave, S. Kasper, T. Tsukada, G. Mandel, and R. H. Goodman, Proc. Natl. Acad. Sci. U.S.A. 85,6662 (1988). 130. G. Guan, P Dai, and I. Shechter,J. Biol. Chem. 273,12526 (1998). 131. Y. Kawabe, T. Suzuki, M. Hayashi, T. Hamakubo, R. Sato, and T. Kodama, Biochim. Biophys. Acta 1436,307 (1999). 132. H. Shimano, I. Shimomura, R. E. Hammer, J. Hen, J. L. Goldstein, M. S. Brown, and J. D. Horton, J. Clin. Invest. 100,2 115 (1997). 133. J. T. Pai, 0. Guryev, M. S. Brown, and J. L. Goldstein, J. Biol. Chem. 273,26138 (1998). 134. I. Hardardottir, C. Grunfeld, and K. R. Feingold, Cur-r. @in. Lipidol. 5,207 (1994). 135. B. Feinberg, R. Kurzrock, M. Talpaz, M. B&k, S. Sak s, and J. U. Gutterman,J. Clin. Oncol. 6, 1328 (1988). 136. B. J. Auerbach and J. S. Parks,J. Biol. Chem. 264, 10264 (1989). 137. W. H. Ettinger, L. A. Miller, T. K. Smith, and J. S. Parks, Biochim. Biophys. Acta 1128,186 (1992). 138. C. D. Moorby, E. Gherardi, L. Dovey, C. Godliman, and D. E. Bowyer, Atherosclerosis 97, 21 (1992). 139. K. R. Feingold and C. Grunfeld,J. Clin. Invest. 80, 184 (1987). 140. S. Kitagawa, Y. Yamaguchi, N. Imaizumi, M. Kunitomo, and M. Fujiwara, Jpn. J. Pharmacoz. 58,37 (1992). 141. K. R. Feingold, I. Hardardottir, R. Memon, E. J. Krul, A. H. Moser, J. M. Taylor, and C. Gnmfeld, J. Lipid Res. 34,2147 (1993). 142. R. A. Memon, C. Grunfeld, A. H. Moser, and K. R. Feingold, Endocrinology 132, 2246 (1993). 143. W. Liao and C. H. Floren, Hepatology 17,898 (1993). 144. A. T. Stopeck, A. C. Nicholson, F. P. Mancini, and D. P. Hajjar,J. Biol. Chem. 268, 17489 (1993). 145. A. Kumar, A. Middleton, T. C. Chambers, and K. D. Mehta, J. Biol. Chem. 273, 15742 (1998). 146. K. R. Feingold, A. S. Pollock, A. H. Moser, J. K. Shigenaga, and C. Gnmfeld, J. Lipid Res. 36, 1474 (1995). 147. R. A. Memon, I. Shechter, A. H. Moser, J, K. Shigenaga, C. Grunfeld, and K. R. Feingold, J. Lipid Res. 38,162O (1997). 148. S. Ghosh, M. J. May, and E. B. Kopp, Annu. Rev. Immunol. 16,225 (1998).

Yeast Chromatin

Structure

and Regulation Gene

of GAL

Expression R. BASH*

AND D. LoHR*‘~

*Department of Chemistry and Biochemistry ~Molecular/Cellular Biology Program Arizona State Unioersity Tempe, Arizona 85287 I. Yeast Genomic-Level Chromatin Structure ......................... A. Basic Organization .......................................... B. Novel Core Nucleosome Features in Situ, Involving the Linker-Core Boundary .................................................. C. The Uncertain State of Yeast Higher Order Chromatin Structure .... II. Yeast GAL Gene Regulation ..................................... A. TheGALFamily ............................................ B. Structural Gene Regulation ................................... C. Regulatory Gene Control ..................................... III. GAL Gene-Specific Chromatin Structure .......................... A. Chromatin Structure in the Uninduced State (Glucose or Gly/Lac) B. Chromatin Structure in the Induced State (Galactose) ............. C. Regulatory Factors, Nucleosomes, and GAL Gene Activation ...... References ....................................................

Yeast genomic DNA is covered by nucleosome length linkers. The short linkers, reinforced number

of unique

nent unpeeling

of DNA

from the ends of the core, an inability to bind even full a conformational

the changes found in active chromatin.

large-scale

packaging

transition that

These features probably

plain how yeast can maintain most of its genome in a transcribable regulated

create a

structural features in aioo: perma-

147 bp core DNA lengths, and facility to undergo resembles

203 212 214 214 215 220 223 224 233 246 254

cores spaced by short, discrete

by novel histone properties,

and dynamic nucleosome

198 198

away of inactive genes. The GAL genes provide

system in which to study gene-specific

a closely

chromatin structure. GAL struc-

tural genes are inactive without galactose hut are highly transcribed ence; the expression patterns of the regulatory

ex-

state and avoid

in its pres-

genes can account for many of the

features of GAL structural gene control. In the inactive state, GAL. genes demonstrate a characteristic activation sequence

promoter (UAS,)

as the TATA and transcription hle ps implement

chromosomal

the major upstream

start sites are in nucleosomes.

gene regulation

regions, where-

This organization

in this state and may benefit the organism.

duction of GAL expression triggers

Progress in Nucleic Acid Research and Molecular Biology, Vol. 65

organization;

elements lie in open, hypersensitive

Gal4p-dependent

197

upstream

nucleosome

Indis-

Cupynght 0 2001 by Academic Press. All rights of reproduction in any form reserved 0079.6603#01 $35.00

198

R. BASH ,4ND D. LOHR ruption. Disruption dent nucleosome

is transient and can readily be reversed by a GaltiOp-depen-

deposition process. Both are sensitive to the metabolic

the cell. Induction triggers sequences,

perhaps

sus promoter

different

reflecting

the differing

bending

roles of nucleosomes

positive and negative

and chromosomal

also play a role in GAL regulation. cleosomes

state of

changes on the coding on coding ver-

regions. GAL gene activation is a complex process involving multi-

ple Gal4p activities, numerous tails. DNA

kinds of nucleosome

architecture

cofactors, and the histone

of the promoter

Regulator-mediated

competition

regions may between

nu-

and the TATA binding protein complex for the TATA region is proba-

bly a central aspect of GAL regulation and processes that contribute

to it.

and a focal point for the numerous

factors

D ~000 Academic PESS.

I. Yeast Genomic-level Chromatin Structure The early 1970s saw the first appearance of experimental results that would ultimately define the nucleosome and thus revolutionize our understanding of the foundation of eukaryotic chromosome structure. Baker’s yeast (Saccharomyces cerevisiae) was already a well-established genetic model by this time, so it is not surprising that yeast chromatin soon came under study. Results from yeast have fundamentally influenced how we view nucleosome structure/function in eukaryotes (I- 3) and continue to provide pioneering insights (4 - 8). The basic chromatin organization in yeast resembles that in metazoans, but there are some structural differences. We will first discuss yeast chromatin structure at the genomic level. This provides both an important background for the gene-specific studies that are the major topic of this chapter and insights that help interpret the single-gene chromatin transitions that accompany changes in gene activity.

A.

Basic Organization

1. A CONSERVED NUCLEOSOME CORE/A SHORT LINKER: MOST OF THE GENOME Is NUCLEOSOMAL Yeast chromatin has a nucleosomal organization (9) that is indistinguishable from that of metazoans in protected core DNA length, 147 bp, and DNA-histone structure (IO-II). Particles containing the yeast histones and this length of DNA can be isolated after micrococcal nuclease (MNase) (Table I) digestion of yeast nuclei (12). Thus the basic or core nucleosomal particle first found in metazoans is also present in yeast. In excess of 90% of the yeast genome is estimated to be nucleosomal(13) and the nucleosomal character remains unchanged in stationary phase cells (14) or in cells undergoing meiosis (R. Malone and D. Lohr, unpublished results). On the other hand, linker DNA, the DNA that lies between the core nucleosomes along the chromatin fiber, is significantly shorter in yeast, by some 30-40 bp, than the intemu-

YEAST CHROMATIN AND GAL GENE REGULATION

199

cleosomal DNA in metazoan chromatins (13, 15). Linker DNA in yeast, as in all eukaryotes, shows some length heterogeneity (13). Based on the nucleosomal repeat length, core + linker, yeast core nucleosomes would be spaced on average by - 15 bp (h owever, see Section 1,B). How far apart two nucleosomes are on DNA without any regard to their specific locations on the underlying sequence will be referred to as nucleosome spacing. Specific location of nucleosomes on particular DNA sequences is called nucleosome positioning. 2. DISCRETE LENGTH LINKERS: ABUNDANT IN YEAST DNase I analyses provide novel and important information about both gene-specific and bulk chromatin structure, as described in Table I and illustrated in Fig. 1. For example, DNase I patterns from nuclear chromatin di-

gests show clearly that the heterogeneity in yeast linker length is not random; linkers have a preference for lengths that are multiples of 10 bp (16) plus a 5-bp increment, i.e., 5, 15, 25 bp (17). A s a result of this linker quantization, core nucleosomes will have a tendency to be discretely spaced, by -0.5, -1.5, -2.5, etc. turns of DNA along the yeast chromatin fiber. Drosophila (18) and rat liver chromatins (19) were subsequently found to possess discrete length linkers by different experimental approaches, as was chicken erythrocyte by DNase I analysis (20). Even chromatin reconstituted in vitro shows discrete intemucleosomal spacing; alternative nucleosome positions on the popular sea urchin 5s DNA arrays (21) vary by multiples of 10 bp (22-23). Th us, discrete linkers seem to be a general chromatin preference. Such linker regularity probably has important functional consequences. For example, it may reflect higher order packing requirements. Also, specific nucleosome positioning on promoters, replication origins, etc. may be needed for proper regulation (24-25). However, there is a thermodynamic preference for variability in nucleosome locations over specific location, whose entropic price is high. Linker lengths that vary by 10 bp will allow a DNA sequence to maintain at least a constant rotational orientation on a nucleosome, i.e., the proper phase relation between the helical setting of the DNA and its contact sites on the histone surface (26), even if the precise location of the nucleosome on the DNA varies. Thus, discrete linker length variation may be a compromise between the need for specific chromosome organization and the high thermodynamic price of absolutely precise nucleosome location. The 5-bp increment (10n + 5 bp) in the preferred linker length may reflect a favored exit/entry relationship of the DNA on neighboring core nucleosomes. For example, discrete linkers with a 5-bp increment lead naturally to a particular type of alternating zigzag arrangement (Fig. 2). This struc-

200

R. BASH

AND D. LOHR

TABLE I CHROMATINCLEAVAGEAGENTSANDHOWTHEYCLEAVE~ Cleavage agent

Cleavage preference

DNase I

Both linker and core

MPE

MNase

- Fe(H)

Comments

Strong linker preference

Linker

preference

Useful for detecting hypersensitive regions, DNA bound to a nucleosome surface, specific factor binding to DNA; low to moderate sequence specificity Useful for detecting hypersensitve regions, specific factor binding to DNA, nucleosome locations; low sequence specificity Useful for detecting nucleosome patterns and profiles; moderate to high sequence specificity

lntercore \

I lntracore 3

I

lntracore

aDNase I digestion analysis can provide structural information about both the core and linker regions of chromatin. DNase I cleavage (nicking) within core particle DNA occurs maximally at sites that lie - 10 bp apart, reflecting the periodic maximal exposure of DN,4 strands on the core histone surface (see illustration above showing genesis of fragments for intracore and intercore DNase I patterns). This structural feature has also been verified with chemical cleavage agents (cf. 34). Because of this periodic accessibility, major bands in the intracore digest pattern, i.e., the pattern of bands composed of DNA fragments arising from nicks within the same core nucleosome, shows an - lo-bp band periodicity, or “ladder,” on denaturing gels. In principle, this pattern can extend up to the full core DNA length of 147 bp. The evidence for discrete linker lengths comes from analysis of the pattern composed of DNA fragments arising from a cleavage event in each of two neighboring (or next-nearest neighbor) core particles and thus including an intact length(s) of linker DNA. This intercore pattern still forms a regular ladder of discrete bands up to DNA lengths in excess of 300 bp (16). As noted above, the cleavages with in core nucleosomes occur at discrete sites, but only if the intevening linker were also discrete in length would these fragments still form a regular ladder. Measurement of the precise spacing of the intercore bands shows that they occur at - IO-bp intervals; measurement of their size shows that they contain a 5-bp increment relative to the bands from the intracore pattern. Thus, yeast linkers must be 10n + 5 bp in length, where n = 0, 1,2, etc. The 5. bp offset in the intercore bands allows them to be distinguished from the intracore bands where the two patterns overlap (Fig. 1). The intracore pattern provides information about linker accessibility and the linker-core transition region (see text). DNase I cleavage within the linker probably occurs randomly, i.e., not at discrete sites, and produces the diffuse and general background smear of DNA underlying the discrete band pattens in these digests. DNase 1 can also be used for footprinting analysis, to detect specific factor binding in chromatin. Factor binding produces a smaller and more uniformly protected stretch of DNA, and because there is no periodic exposure, no lo-bp ladder pattern.

YEAST CHROMATIN

AND GAL GENE REGULATION

MOBILITY

(cm)

FIG. 1. (A) Intracore and intercore DNase I patterns from yeast. Cleavage of isolated core nucleosornes with DNase I in vitro produces a pattern of 14 bands, the sizes of which form a regular series when plotted against electrophoretic mobility (cf. 1). The bands in this intracore pattern can be designated by band numbers. The size of a band is equal to (band number) X (average band periodicity, in base pairs). In situ DNase I digests also produce an intracore pattern (Table I); the yeast nuclear digests depicted here show a clear intracore pattern (A) to about band #lo, at which point the bands become more closely spaced. Above this region the intercore pattern (Table I) again forms a regular series of bands (A). The region of closely spaced bands is the overlap region between the intercore and intracore series. With this realization, alternate bands in the overlap region can be clearly assigned to either the intracore or the intercore series (vertical arrows). These two series can be distinguished in the overlap region because the intercore bands contain a 5-bp offset. To denote this they contain 0.5 in their designation (e.g. 12.5). (B) A scan of the yeast intercore pattern. The smallest band (12.5) corresponds to 131 bp in size and the 24.5 band to 257 bp. The discrete band pattern extends beyond 340 bp. Above band 24.5 there is a discontinuity. Fragments below this point (lower molecular weight/to the right in the scan) arise from nicks in neighboring nucleosomes, n and n + 1. Fragments above the discontinuity (higher molecular weight /to the left in the scan) arise from nicks in core nucleosomes, n and n + 2, i.e., next-nearest neighbors. These fragments would contain two linkers and thus 2 X 5 bp increments, or 10 bp, and thus be back in helical phase with the intracore fragments. The discontinuity probably reflects this phase change. A similar discontinuity is seen in the transition from intracore (10n) pattern to intercore (10n + 5) pattern. From Lohr and van Holde (17).

R. BASH AND D. LOHR

FIG. 2. Yeast linker lengths suggest a zigzag chromatin organization. DNA winds around the nucleosome in a left-handed supercoil. If the core nucleosome contains an integral number of turns of DNA, then if DNA starts to wrap around the nucleosome with a particular strand orientation on the histone core, the one-half turn in the linker (+5 bp) would cause this strand orientation to switch every other nucleosome and repeat every dinucleosome (17). The core DNA is shown to be wrapped - 1.5 times around the octamer, based on results presented in the text (Section 1,B).

ture has a dinucleosomal repeat (17), a feature that may be an important element of chromatin short-range order (27, 28). The arrangement might also lessen the torsional stress in a fiber of nucleosomes spaced by short linkers. Zigzag structural models have also become popular for chromatin with longer average linker lengths, e.g., chicken erythrocyte (cf. 29, 30). In these chromatins, a zigzag structure may reflect other constraints associated with higher order packing (cf. 30,31). Thus di screte linkers and the associated regularity might be beneficial in many ways for chromatin organization. It was suggested that the 5-bp increment in the intercore pattern actually results from a systematic variation in the angle of DNase I attack within core nucleosomes rather than a linker increment (19). However, that suggestion is inconsistent with the constant slope of both intra- and intercore band plots (Fig. 1). Moreover, very small intercore bands, which arise from DNase I cleavage near the ends of neighboring core nucleosomes and thus contain linker DNA as a major component, lie in the same relative size position to intracore bands, and therefore contain the same 5-bp increment, as the larger series of intercore bands that extends to beyond dinucleosome size (20). The simplest explanation, that linkers actually contain a 5-bp increment, is also the explanation that is most consistent with this constant intercore fragment size.

YEAST CHROMATINAND GAL GENE REGULATION

203

The precise abundance of discrete linkers in chromatin has always been in question. The abundance can be estimated from the fraction of DNA in the intercore portion of the nuclear DNase I pattern (Table I). However, this provides only a minimum estimate for several reasons, principally because whenever DNase I happens to cleave in a linker, it destroys the evidence of discrete length in that linker. Only pairs of cleavages that occur in cores (neighboring or next-neighbor) will produce a DNA fragment with an intact linker and thus allow a test for discrete length. DNase I is the most useful enzyme for these analyses because it readily cleaves within the core and thus, as it digests chromatin, leaves the most linker DNA intact. The relative preferences for core versus linker cleavage vary with nuclear digestion conditions (20). Under conditions favoring cleavage within the core, the yeast intercore pattern becomes very prominent, suggesting that discrete-length linkers are abundant, perhaps prevalent, in yeast. Discrete lengths even appear to be significant in long linker-length chromatin (chicken) when analyzed under these conditions. One might anticipate that such significant linker regularity would be reflected in very distinct nucleosomal ladder patterns in yeast. Indeed, genomic-level patterns are typically quite clear and single-gene ladders are even clearer; on inactive GALI, the pattern is clearly resolved up to 15 or more nucleosomes, i.e., >2.5 kb of chromosomal DNA (D. Lohr, unpublished observations).

B. Novel

Core Nucleosome

the Linker-Core

Features

in Situ, Involving

Boundary

1. CONFORMATIONALCHANGES In nuclear digests done in very low-salt/low-divalent cation concentrations, both the MNase and DNase I patterns are altered compared to digests done in more traditional salt concentrations (20). For DNase I, the intercore pattern becomes exceptionally prominent and the diffuse DNA background that usually underlies the discrete band patterns becomes almost undetectable. Both of these changes suggest that there is relatively less linker cleavage/relatively more intracore cleavage in nuclear chromatin under conditions of very low ionic strength. The changes in the diffuse background are especially revealing because this feature arises from, and thus directly reflects, linker cleavage (Table I). In the very low-ionic-strength MNase digest patterns, a significant amount of DNA is found between the major nucleosome bands, much more than in a typical nucleosomal pattern, although the major bands are still prominent (Fig. 3A). Th e major bands reflect linker cleavage; their continued presence indicates that nucleosomes remain present. The increased interband intensity indicates that nucleosome structure is altered to allow

B

FIG. 3. The efkcts of ionic strength on MNase digests of yeast nuclear chromatin. (A) Gel pro&s and scans of these profiles for a low-ionic-strength (pattern 3) and a 150 mM NaCl digest (pattern 4) are shown The direction of electrophoresis is top to bottom in the gels and left to right in the scans. The roman numerals II and IV locate di- and tetranucleosomal length DNA. (B) A low-ionic-strength bulk chromatin digest (0) is compared to a gene-specific GAL1 pattern obtained by MNase digestion of nuclei from induced, transcriptionally active cells (+). Roman numerals I and II locate mono- and dinucleosomal length DNA. From Lohr (20).

YEAST CHROMATIN

AND GAL GENE REGULATION

205

cleavage within normally protected regions. Nucleosome loss would be unlikely under these low-ionic-strength conditions and the interband DNA confirms this expectation. It does not form a smear, as would be produced by cleavage of protein-free DNA, but a discrete pattern of up to 15 bands spaced at - lo-bp intervals between the major bands (Fig. 3A), suggesting that it arises from cleavage in exposed but still protein-involved intranucleosomal regions. Thus, very low-ionic-strength causes a change in core nucleosome structure that results in increased intracore cleavage in situ by DNase I and MNase. Under very low-ionic-strength conditions, the effects of DNA phosphatephosphate and possibly histone-histone repulsive interactions would be heightened, and this could destabilize the core structure. Indeed, core nucleosome unfolding is observed in I.&-Oin very low salt, and models for this unfolding have been proposed (1). The models suggest a loosening of the compact core particle to a more open, extended structure by a change involving relative histone movement but maintaining the histonejcore DNA contacts (32). These features are consistent with the pattern changes observed in the low-ionic-strength in situ digests: an opening up of the compact core should enhance intracore nuclease sensitivity; the maintenance of DNA-histone contacts would explain the discrete DNase I and MNase intranucleosomal band patterns observed. This type of nucleosome unfolding could also enhance linker protection. Even in the compact core nucleosome, histone contacts can apparently extend beyond the core. For example, presumably through their tails, the core histones alone can protect up to 166 bp of DNA (7, 33, 34), well beyond the core DNA length of 147 bp. H3 and H2A tails have been shown to make linker contact. Under very low-ionic-strength conditions, interactions between the highly positive N-terminal tails and linker DNA would be strengthened; the loosening or unfolding of the compact core structure could place the tails (or other histone regions) in an even better position to interact with linker DNA than in a compact particle. The same types of digestion changes seen in bulk chromatin at very low ionic strength are observed in gene-specific chromatin digests when the target genes are transcriptionally activated (Fig. 3B). Thus, the low-salt changes may be an appropriate model for understanding the structure of transcriptionally active chromatin and single-gene nucleosome changes (Section IILB). This analogy would suggest that features that change the local electrostatic balance around nucleosomes, such as histone phosphorylationacetylation (8) or the presence of negatively charged factors such as the hyperphosphorylated carboxy-terminal domain (CTD) ofRNA polymerase II, which is preferentially associated with elongating pol II (34u), might be able to trigger these same kinds of changes in specific nucleosomes and thus produce the

206

R.BASHANDD.LOHR

structure associated with transcriptionally active chromatin. The low-salt structural change has the effect of decreasing the distinction between core and linker regions in bulk chromatin (20), resulting in a more uniform fiber structure. This could provide a structural explanation for the smooth fiber often detected in electron microscopy of active chromatin (1). Chicken erythrocyte chromatin digests done under very low-ionicstrength conditions show the same qualitative changes as yeast-enhanced DNase I intercore pattern/lower backgrounds and enhanced MNase intranucleosomal cleavage (20). Thus, metazoan chromatin can undergo this type of in situ conformational transition. However, the changes in chicken are quantitatively less striking. Several features make yeast chromatin especially disposed to transcription-associated chromatin changes (Section I,B,4). Perhaps that is why this transition can be seen especially clearly in yeast. 2. UNPEELINGOFDNAFROMTHECOREENDS:APREFERENCE IN SHORTLINKER-LENGTH~HROMATIN Under any in situ digestion conditions, the intracore DNase I pattern in yeast does not extend beyond - 126 bp (20), i.e., it terminates well below the full 147-bp core DNA length. More bands would be detected in these patterns if they were present (see below). These results suggest that the DNA lengths associated with yeast core nucleosomes under in situ conditions are less than the prototypical 147-bp core length. The largest intracore bands, 126 bp and greater, arise from pairs of DNase I cleavages that span the core. The absence of the larger bands suggests that yeast core nucleosomes exist in z&o with at least 20 bp of DNA unpeeled from their termini. End-unpeeling is distinct from the low-salt conformational transition described in the previous section because end-unpeeling in yeast is even present in physiological salt, conditions that favor the compact core structure. Of course, end-unpeeling could help the conformational transition to occur. The association of only 126 bp of DNA with the core means that the average length bp, much closer to the averof intemucleosomal DNA in yeast is -35-45 age metazoan intemucleosomal DNA length. We have obtained results from a recent collaboration with the lab of Stuart Lindsay (Arizona State University) (34b) that confirms end-unpeeling and indicates that it is a preference of nucleosomes in short linker-length arrays. The work used an in vitro reconstituted chromatin fragment consisting of 12 tandemly repeated, 172-bp 5s DNA units (2 l), reconstituted with chicken histones. Because each 172-bp unit can position a nucleosome, at least with moderate precision (22, 23), this DNA template allows one to reconstitute a defined nucleosomal array, in this case a short linker-length, yeastlike array. Atomic force microscopy (AFM) was used to measure the average internucleosomal DNA length (image in Fig. 4).Thislengthsubtracted from the

YEAST

CHROMATIN

AND GAL GENE

REGULATION

207

FIG 4. An atomic force microscopy image of reconstituted 172-12 chromatin. The image was obtained by tapping mode in air on a mica surface. The field shown in this magnification is 660 X 880 nm (courtesy S. Leuba).

unit length of 172 bp yields the average amount of DNA associated with the cores in this chromatin array. The result is -100 bp, significantly less than the 147 bp associated with the prototype core nucleosome. As a control, we also analyzed a chromatin array consisting of 12 tandemly repeated, 20%bp 5S DNA units, each of which contains the same nucleosome positioning sequence but 36 bp more of linker. In this array, an average of - 125 bp of DNA is associated with the core nucleosomes. Thus, a short linker increases the tendency for unpeeling of DNA from the core termini. Surface forces inherent in AFM are probably responsible for some artifactual loss of DNA from the ends of the core in both arrays, explaining why we do not see full 14 7-bp core association with the 208-12 template (S. Leuba and S. Lindsay, unpublished results). Therefore, the in situ DNase I results from the previous section probably estimate the amount of end-unpeeling in yeast chromatin more accurately. However, these in. vitro results (1) strengthen the conclusion that the in situ intracore DNase I pattern in yeast terminates at 126 bp instead of 147 bp because DNA is unpeeled from the core termini and (2) indicate that this is due, at least partly, to the short linker length in yeast. End-unpeeling should facilitate the transit of DNA from one core to the next in a chromatin

R. BASHANDD. LOHR

208

fiber with short linkers; it has also been observed in various in vitro studies (35). In yeast, unique histones properties may also contribute to this in situ terminal instability (Section I,B,5). 3. WINDING MORE DNA ON THE CORE: NOT IN YEAST In erythrocyte nuclear digests in low ionic strength, the intracore DNase I pattern terminates at 126 bp (Fig. 5), indicating that there is also end-unpeeling in metazoan cores under these conditions in situ. However, in physiological salt (150 mM), th e intracore pattern extends to 147 bp, full core nu-

Y FIG. 5. The effects of ionic strength on chicken nuclear DNase I digests. Scans of chicken chromatin digests done at low ionic strength (LO) and at 350 mM NaCl(350) are compared to a yeast digest (Y). The yeast digest shows only the transition region from intracore (A) to intercore (A) patterns. The largest resolved intracore band in this yeast pattern is band 11; 12 is a shoulder. The yeast intracore bands are slightly larger than the chicken bands because they are spaced on average by 10.5 bp rather than 10.35 bp (17). Th e intracore bands (A) in the 350 mA4 NaCl chicken digest are numbered. In these digests, the intracore band pattern continues to band 15, 155 bp, with shoulders at 166 and 176 bp. In the low-salt chicken digests, the largest band in the intracore series is 124 bp (band 12). Beyond that band, the pattern switches to a series in which the bands are offset from the intracore series by 5-7 bp. The position of the first offset band beyond 12 is marked by a dashed vertical line. This switch is analogous to the switch from intracore to intercore pattern in yeast above band 12, except yeast intercore bands always contain a 5-bp offset relative to the intracore series. As a result of the offset of 5-7 bp in chicken, the low-salt digest chicken bands in this transition region do not always line up with the yeast bands. However, above 190 bp (far left in the scans), bands in both chicken digests line up with the yeast intercore pattern and stay in alignment to 300 bp. From Lohr (20).

YEASTCHROMATINANDGALGENEREGULATION

209

cleosome length, and in 350 mM NaCl to 157 bp, with detectable shoulders corresponding to 168 and 179 bp (Fig. 5). The 168 bp band, or two full turns of DNA wound around the core, corresponds to a chromatosome structure (36), an important higher order structural feature in metazoan chromatin (33). Thus, increasing salt concentration in this range causes increasing amounts of DNA to associate with the core in chicken nuclear chromatin. However, the yeast intracore pattern never extends beyond 126 bp, even in 350 mM NaCl, and at this [NaCl] yeast nuclei show a decided tendency to lyse (20). The ability to detect the large intracore bands in erythrocyte digests verifies that they would be seen in yeast digests if they were present (Fig. 5). Thus, yeast and chicken chromatins demonstrate significantly different responses in their core terminal regions to increasing salt concentration. Chicken can wrap up to two full turns of DNA on its nucleosomes; yeast cannot even wrap the full length of DNA usually associated with the core. The inability to wind more DNA onto the core argues that yeast cannot form a chromatosome in vivo and thus does not use this metazoan chromatin folding intermediate. The abundant chicken linker histones could play a role in the different abilities of chicken and yeast to associate DNA with the core. However, the results indicating that the core histones alone can protect up to 166 bp (34) suggests that linker histones are not entirely responsible for these differences. 4. CONSEQUENCESOFTERMINALLIABILITY The terminal regions of the core nucleosome have the fewest histoneDNA contacts (I, 26) and the greatest excess of negative charge in the parti cle (1) and thus should be the most sensitive to ionic strength changes. Indeed, end-unpeeling and winding more DNA on the core (in chicken) involve this region. Also, the biggest changes associated with the conformational transition at low salt occur at the core termini. For example, small intercore bands, which arise from DNase I cleavage near the termini, are readily detectable in low-ionic-strength digests (20) and intranucleosomal cleavage by MNase is most prominent near the termini, with less cleavage in regions corresponding to the center of the nucleosome (Fig. 3). In addition to the above changes, the presence of millimolar concentrations of either divalent cations or spermine/spermidine during digestion cause (unique) changes that involve only the terminal 20-30 bp on each end of the core (20). Thus, the chromatin structure of the core termini (and linker) regions is very sensitive to solvent conditions, ligands, etc., and its structure is quite labile. In vitro studies have shown that specific DNA-binding proteins can most easily invade the nucleosome at its termini (7, 33, 37), results that are quite consistent with the in situ terminal lability described above. The in situ results demonstrate that terminal accessibility can vary significantly. In chro-

210

R. BASH AND D. LOHR

matins with long linkers, nucleosomes have a range of possible core structures, from 126 to 168 bp or more associated with the core, i.e., from 1.5 to 2 turns of DNA or more wrapped around the histone octamer. Such variations should affect the ability of transcription factors, and perhaps RNA polymerase, to invade the nucleosome. It seems likely that the end-unpeeled structure would be associated with transcribable chromatin; factor invasion ought to be much facilitated in this structure. The chromatosome structure is associated with folding and gene inactivity. These variations in terminal association should also affect higher order packing, due to the variation in DNA entry-exit trajectories on the nucleosome (35). However, yeast lacks such diversity. At low or physiological ionic strength, no more than 126 bp of DNA associates with the core (20). Thus yeast chromatin in situ only exists in an end-unpeeled structure, the structure that should facilitate transcription factor and RNA polymerase invasion. The lowsalt conformational transition, a possible model for transcription-associated nucleosome changes (Section III,B), might also be more easily triggered in a permanently end-unpeeled nucleosome. Therefore, the short linker lengths in yeast, perhaps aided by histone features (see below), help create a chromatin structure that is inherently favorable for gene activity and unfavorable for folding and inactivating genes, at least in the style used by metazoans. The crystal structure of the core nucleosome shows that the repeating and strongest DNA-interacting motifs of the histone octamer organize 12 1 bp of DNA (26). Thus, the DNA lengths stably associated with yeast nucleosomes correspond to the regions of strongest DNA contact with the core. Given the above differences, why did the yeast core nucleosome initially appear to be identical to metozoan core nucleosomes (JO)? In that work, the yeast and metazoan DNase I digests were all done under identical and fairly low salt concentrations. Thus, the intracore DNase I pattern in all digests terminated at roughly the same size, 116-126 bp, and the salt concentrations that would drive more DNA onto the metazoan cores were not tested. MNase was also used in those studies. It strongly prefers to cleave linker DNA under the conditions used. Cleavage of the linker is most likely at or near its center and even in yeast could produce a core length of 147 bp because cutting the linker will relax the torsional stress of the chromatin fiber and allow the freed DNA to associate with the flanking cores. Therefore, while the basic nucleosomal structure is similar, there are significant dynamic differences between yeast and metazoan chromatins. These differences depend on constraints associated with in viva internucleosomal organization and thus are most clearly manifested under conditions that maintain the intact organization of the chromatin fiber, such as in situ. These differences will undoubtedly affect how and how well nucleosomes undergo the structural transitions

YEASTCHROMATINANDGALGENEREGULATION

required for the functional processes of replication, transcription, nation and repair.

211 recombi-

5. HISTONECONTRIBUTIONSTOTHEUNIQUEYEASTCHROMATIN FEATURES?

Core histones (H3, H4, H2A, and H2B) are rather conserved proteins (1). Each core histone consists of at least two distinct domains, a central region that contains the histone fold (38), the basic assembly motif that forms the globular compact core structure as well as the binding sites for DNA wrapping around the nucleosome, and the N-terminal tails that extend out from this compact structure (26, 39). The N-terminal tails are not essential for nucleosome formation because both assembly and positioning in vitro are unaffected by complete tail removal by trypsin, but these tails can affect the accessibility of nucleosomal DNA to other proteins (7, 40). The tails are very basic and have been suggested to function in internucleosomal contact leading to chromatin compaction, by interacting with linker DNA and/or with other core nucleosomes (7, 26, 40~). Whether the tails adopt a particular structure is uncertain and, in fact, they were not well visualized in the nucleosome crystal structure (26). The tails are the target of acetylation at specific lysine residues, a modification associated with replication and transcriptional regulation (7, 41, 42). H2A al so contains a sizable C-terminal extension (43). Yeast core histones are similar in sequence to other eukaryotic histones, but not identical. Even the most highly conserved histones, H3 and H4, show 15 of 135 (H3) and 8 of 102 (H4) amino acid differences compared to calf (1, 44). Most of these differences, 12 of 15 and 7 of 8, respectively, are located in the globular central region and most are conservative (43, 44). Yeast H2A and H2B contain - 30 or more changes each compared to calf. Less than half of these changes lie in the globular region; most fall in the N-terminal or Cterminal tail domains. There are two H3 and two H4 genes encoding identical proteins (44) and two H2A and two H2B genes that encode proteins that differ slightly (45, 46). There are indications that inherent histone features give yeast nucleosomes unique properties that are consistent with the core terminal instability observed in situ. Thus, histone features could contribute to the in situ instability. For example, a much higher fraction of the nucleosomal DNA is involved in the thermal denaturation “premelting” transition in isolated yeast versus erythrocyte core mononucleosomes (47). This premelting fraction corresponds to DNA unpeeling from the core termini (1). In plasmid chromatin in vivo, supercoiling assays show that DNA is less constrained from thermal untwisting on yeast nucleosomes than on metazoan nucleo-

212

R.BASH

AND

D. LOHR

somes (48). Unpeeling of DNA from the termini of the core is a possible explanation for at least some of this decreased constraint. Both the enhanced premelt observed in vitro and the reduced DNA constraints observed in viva were ascribed to unique histone properties, in the former case H3 or perhaps H2A or H2B (47). CD measurements also suggest that DNA in isolated yeast mononucleosomes is less constrained than in chicken mononucleosomes. In situ DNase I results show that DNA winding on yeast core nucleosomes is also looser than on metazoan nucleosomes, averaging 10.5 bp/turn versus 10.35 bp/tum (17). Isolated yeast nucleosomes readily dissociate in low-ionic-strength solutions and thus are much more unstable than calf or chicken nucleosomes (47). Yeast nucleosomes also undergo significant histone dissociation in 0.5 M NaCl, whereas chicken nucleosomes show no dissociation at these salt concentrations (49). Therefore, yeast nucleosomes are also less stable in elevated salt. Whether this observation explains yeast nuclear instability in these salt ranges (20) is unclear, The tails of H2A and H2B form contacts with DNA toward the ends of the core particle (7, 33, 34). In yeast, these tails contain many sequence changes and several fewer basic residues per nucleosome than calf H2A and H2B tails. Being less basic, these tails might stabilize the nucleosome termini less well, thus facilitating end-unpeeling, for example. Also, the (YN helix of H3, the region that binds the terminal 13 bp on each core nucleosome (26), has a sequence change in yeast that could affect its binding properties, a Phe instead of a Tyr Specific histone tail regions are crucial for proper DNA wrapping on the nucleosome in viva (43). The numerous sequence changes in the yeast H2A and H2B tails could contribute, at least partly, to the lessened ability of yeast nucleosomes to constrain DNA. Thus unique histone features may promote the core terminal instability and conformational changes that are characteristic of yeast nucleosomes in vivo, working in concert with torsional constraints arising from short intemucleosomal spacing.

C. The Uncertain Chromatin

State of Yeast Higher

Order

Structure

Short intemucleosomal spacing should be a major determinant in yeast higher order chromatin structure. For example, by preventing association of chromatosome or even full 147-bp core lengths of DNA with the nucleosome, this feature will certainly affect how yeast cores pack into a chromosome fiber. Short intemucleosomal spacing is consistent with the absence of another major determinant of metazoan higher order structure, linker-specific histones such as Hl or H5 (33,34). However, a yeast Hl candidate with -35% identity to the most conserved sequence motif of human Hl was identified from yeast genomic sequence analysis (50). Deletion of this protein

YEAST CHROMATIN

AND GAL GENE REGULATION

213

does not affect cell growth, viability, or mating but can affect expression of a reporter gene (51). The protein is apparently unable to modulate gene activation at a distance or to mediate silencing (52), two putative functions of metazoan Hl. In fact, yeast genes can be packaged into inactive structures without Hl (53). If the yeast candidate is a bona fide Hl, it may function quite differently from metazoan Hl and/or be restricted in its genomic role. The first level of higher order folding in metazoan chromatins is generally regarded to be the 30-nm fiber, although its precise structure and role remain controversial (cf. 31,33,54). Yeast chromatin can form a 30-nm fiber, either in nuclear spreads (55) or when isolated (56). In the latter case, it resembled the 30-nm fiber found in chromatin containing linker histone, leading those authors to suggest Hl presence. Forming a 30-nm fiber would require folding the basic zigzag structure shown in Fig. 2. Models for the folding of zigzag, long linker-length chromatins have been suggested (30,57) but these involve linker DNA cross-overs and thus are probably not feasible for short linker chromatins, even with end-unpeeling (58). Formation of the metazoan 30-nm structure depends on Hl and physiological levels of salt (59). G iven the different responses of yeast and chicken nuclear chromatins to physiological salt concentrations, the inability of yeast to form a chromatosome, and the uncertain status of yeast Hl, it is conceivable that the structure of a yeast 30-nm fiber may differ from a metazoan one, even though the fiber diameters are similar. Core histones can cause chromatin folding (40) and in yeast could account for at least some of the folding observed. The amount of DNA associated with the yeast core in situ, 126 bp, would put DNA entry and exit points at opposite sides of the core. In such a structure, nearest-neighbor nucleosome associations, i.e., 1 -+ 2 in Fig. 2, would probably be disfavored because they would create a small, torsionally constrained looping out of the linker. Next-nearest-neighbor contacts, i.e., 1 + 3 in Fig. 2, ought to result in more than 126 bp of DNA being associated with the intervening core, which should lead to a more extended DNase I intracore pattern than is observed. Therefore, nucleosome-nucleosome contacts between more distant parts of the fiber might be preferred. These could be short range or even long range, perhaps analogous to a psheet structure in polypeptides. DNase I analysis was able to demonstrate a higher order, genomic-level feature unique to yeast. Metazoans sequester inactive genes into highly inaccessible (to DNase I) and transcriptionally repressed higher order structures (60). The exact nature of these structures remains an active research area (cf. 61) but a lack of domain-level sensitivity to DNase I is a signature feature. However, yeast does not sequester detectable amounts of DNA into such structures because virtually the whole genome is uniformly and highly

214

R. BASH AND D. LOHR

DNase I sensitive (62). This analysis would probably not detect specific silenced genes (cf. 53), which in yeast may be few in number. The genomic DNase I result suggests that most genes in yeast remain potentially expressible, and therefore accessible to transcription factors, even when inactive. A study based on individual nucleosome chemical accessibilities has come to a similar conclusion (63). For reasons discussed above, we suggest that the unique structural features of yeast chromatin (Section 1,B) are responsible for maintaining this expressibility and hindering the packaging of yeast DNA into the types of transcriptionally repressed, folded structures found in metazoan genomes. In addition to maintaining its genes in an expressible and structurally open state, -40% of the yeast genome is actually being expressed at any given time (64). These features make yeast an exceptionally attractive model for studying the chromatin structure of transcriptionally active genes.

II. Yeast GAL Gene Regulation To uncover the functional roles of chromatin structure, it will be necessary to apply detailed genetic and biochemical analyses to specific control regions such as gene promoters or replication origins, to learn how cells organize these critical regions and how the chromatin structure is linked to the functional activities occurring at these key sites. We have been interested in these types of questions for over 20 years, with analysis of the chromatin structures of yeast replication origins (65), ribosomal gene promoters (6668), and GAL1 genes.

A. The GAL Family This family consists of the set of yeast genes involved in the utilization of galactose as a carbon source for growth. This nutrient-regulated family is extremely well suited for regulation studies because gene activity is tightly controlled and easily manipulated. Pioneering work by Douglas, Hawthorne, and co-workers had established the genetic foundations of GAL regulation by the late 1960s (69). Continuing genetic and biochemical analyses have combined to produce a level of understanding that places this family at the forefront of eukaryotic gene regulation model systems (70 - 74). We will discuss the basics of GAL transcriptional regulation in Section II and GAL chromatin structure and its relation to gene control in Section III. More detailed descriptions of GAL regulation can be found elsewhere (69, 70-74). 1 We will follow the convention of capitalizing and italicizing wild-type genes, listing mutant genes in reman type, and identifying gene products by “p” after the name. ’

215

YEAST CHROMATINAND GAL GENE REGULATION TABLE II THE GAL FAMILY Expression in

Gene

Function

Glut

Gly/Lac

Gal

Gal transport Gal --) Gal-l-P Gal-l-P Glu-1-P

_

-

++

2

_

-

-

_

++

++

4” 2

-

_

++

4”

0 1 1

structural GAL2 GAL1 GAL7

UDP-Gl

7%

UDP-Gal

#UAS,;

GAL10 Regulatory

GAL4 GAL80 GAL3

Transcription activator Transcription inhibitor Signal transducer

+

+

+

+ _

+ +

+ +

“Shared behveen GALIO and GAL1

1. STRUCTURAL GENES-GAL],

GALB, GAL7,

AND

GAL10

The structural gene products carry out the various steps required for galactose usage by cells. The GAL2 product transports galactose into cells. GALI, GAL7, and GAL10 gene products convert galactose to glucose-lis converted to the glycolytic inphosphate (Table II). Gl ucose-l-phosphate termediate glucose-6-phosphate by the GAL5 gene product, a mutase. 2. REGULATORYGENES--GAL&

GAL80,

AND

GAL3

Three regulatory gene products are largely responsible for implementing the strict carbon source dependence of GAL structural gene expression (see below): (1) Gal4p, the major and essential transcription activator, (2) Gal80p, the major inhibitor of transcription, and (3) Gal3p, which has traditionally been classed as a regulator but acts like a galactose signal transducer. Gene regulation also involves various pleiotropic factors that help the specific regulators control GAL. expression (Section 111,C) and of course general factors such as TBP and RNA polymerase II.

B. Structural

Gene

Regulation

GALl, 2, 7, and 10 are tightly and efficiently regulated by carbon source: completely repressed (undetectable expression) in glucose, poised for expression but still completely inactive in the nonfermentable carbon source glycerol-lactate, and very strongly induced in galactose.

216

R. BASH AND D. LOHR

1. THE INDUCED

STATE-GALACTOSE

Galactose induction of the structural genes achieves extremely high levels of expression. Under induced conditions, GALI, 7, and 10 mRNA comprise 0.25-l% of total poly(A) mRNA and similar percentages of total cell protein (70). Gal4p activates this high-level expression while bound to the major GAL-specific promoter elements, the upstream activation sequence (UAS), located upstream of the GALl, 2, 7, and 10 genes. A UAS, element suffices to confer galactose inducibility on any gene, and all genes known to be galactose inducible contain one or more of these elements. GAL2 and 7 have two UAS, each; the divergently transcribed GAL1 -10 share four UAS,, asymmetrically located in the intergenic region between the two genes (Fig. 6). Gal4p binding to GAL1 -10 UAS, h as b een detected in vivo by various techniques (75 - 77). Th e same protection pattern on the GAL1 -10 UAS,

UAS_ v I 200

I

I 400

I

UASo r,Gl rq 600

iii% I

I 800

I

I 1000

I

FIG. 6. Maps of the CALJ -10 and GAL80 upstream regions. DNA regulatory elements are located to scale along the thick line. The numbers are base pairs measured from a reference EcoRI site in GAL10 and from an Mb011 site upstream of GAL80. The transcription start sites lie at the origins of the horizontal, labeled arrows. The chromatin structure is illustrated below each sequence map, with nucleosomes located to scale. Individual nucleosomes referred to in the text are identified (A-D and o).

YEASTCHROMATINANDGALGENEREGULATION

217

can be observed in Escherichia coli that express Gal4p as the only yeast protein, demonstrating that the UAS, protection observed in yeast can be totally accounted for by Gal4p binding (75). The yeast in viva DNA protection pattern is also consistent with the protein-DNA contacts found in the X-ray crystal structure of a Gal4p-UAS, complex (78). The crystal structure confirms a number of previous suggestions (70) that a Gal4p dimer binds a single UAS, element. The DNA-binding domain of Gal4p lies near the N terminus of the polypeptide, residues 14-5 7, while the major transcription activation domain resides at the C terminus, residues 767-88 1 (70). Thus one end of Gal4p is bound to the upstream UAS, while the other end activates transcription. The UAS, lie 200-300 bp away from the sites where transcription starts. Therefore the three-dimensional shape of Gal4p, which is unknown, and the chromosomal organization of the promoter regions (Section 111,B) may play roles in the activation process. Possible activation mechanisms of Gal4p will be discussed in Section 111,C. 2. THEPOISEDSTATE-GLYCEROL/LACTATE In glycerol/lactate (gly/lac), the GAL structural genes are completely inactive. Nevertheless, Gal4p is bound to the GALI-IO UAS, in this carbon source (75 - 77) and the UAS, appear to be as strongly protected as they are in galactose (Fig. 7). The presence of Gal4p on the UAS, in the absence of gene expression demonstrates that activator-DNA binding is not sufficient for GAL1 -10 expression. This can be confirmed in other ways. For example, in a superrepressed (SOS) strain in which GAL1 -10 are permanently and completely uninducible in galactose, Gal4p is still bound as usual to the UAS, (76). Gal4p also binds to the GAL2 UAS, in glyilac (79), indicating that this feature is characteristic of GAL structural genes. Even though bound to the UAS,, Gal4p cannot activate structural gene transcription in gly/lac because Gal80p inhibits Gal4p in nongalactose carbon sources, by directly binding to the Gal4p C-terminal activation domain and masking its activity (70). Galactose triggers the release of GalSOp inhibition and allows Gal4p to activate its target genes. A variety of recent work, both in viva and in vitro, has been able to demonstrate that Gal3p mediates this release of inhibition by directly binding to Gal80p (Fig. 8), in a reaction that depends on galactose and ATP (80-83). A ternary Gal3p-Gal80pGal4p complex has been observed in d-o (83). Gallp, the structural gene product, can also mediate the galactose-dependent induction of GAL stmctural gene expression (84). This activity probably involves a Gal3p-like domain in Gallp and also occurs by a direct Gallp interaction with Gal80p (83, 85, 86). Gal3p probably has to carry out the initial induction of structural gene expression when galactose first appears, and Gallp may help to maintain the high expression level characteristic of the induced state.

218

R.BASHANDD.LOHR

FIG. 7.Gal4p footprints on the GAL1 -10 and GAL80 upstream regions. DNase I nuclear digestion profiles from the GALI -10 and GAL80 upstream regions are shown. The locations of the UAS, in the GAL1 -10 intergenic region and in the GAL80 upstream region are shown just to the left of the GAL1 -10 profiles (profiles 1) and the GAL80 profiles (the rightmost two profiles). The chromatin digests are from wild-type cells (wt) grown in galactose (G) or glyllac (g) for the GAL1 -10 profiles or from galactose-grown wild-type or Gal4disrupted (44 cells for the GAL80 profiles. The Gal4p dependence of GALIUAS, protection has been demonstrated (see text). Transcription start sites are located at the origin of the wavy arrows (T=TATA). The GALI -10 sites are mapped from an EcoRI site in GALIO. The GAL80 sites are mapped from a TaqI site upstream of G&80. From Lohr (99). Copyright 1993 National Academy of Sciences, U.S.A.

3. THEREPRESSEDSTATE-GLUCOSE Glucose repression of the GAL, structural genes is part of the general, glucose-dependent repression system that operates in yeast (70, 71, 73). This general system works through multiple mechanisms and numerous gene products. On GAL structural genes, three major mechanisms ensure structural gene inactivity in glucose (87, 88): (1) no Gal4p-UAS, binding due to lowered Gal4p levels, (2) Gal80p-dependent inhibition of Gal4p, and (3) re-

219

YEASTCHROMATINANDGALGENEREGULATION induced

non-induced

ATP gal

UAS

UAS

FIG. 8. Galactose-dependent Gal4p activation. GalSOp masks the Gal4p activation domain in nongalactose carbon sources such as glyilac. Gal3p triggers the release of the Gal80p-dependent inhibition in a galactose- and ATPdependent reaction (see text), allowing Cal4p to activate transcription. Gal80p and Gal4p remain in contact.

pression acting through specific DNA sequence elements on the structural genes. 4. THEPOISEDSTATEANDTHENATUREOF

GAL

GENEACTIVATION

GALI, 2, 7, and 10 expression is undetectable in glucose or in glyllac. However, these genes can be induced to full expression within minutes in gly/lac, whereas galactose induction from the glucose-repressed state requires hours (70). In the former state, Gal4p is present on the UAS,; in the latter it is not. Thus in gly/lac, Gal4p is poised to activate expression if galactose becomes available and only Gal80p inhibition prevents activation. Gal4p occupation of the UAS, is probably crucial to the rapid inducibility of GAL structural genes in gly/lac. The existence of a distinct poised but unexpressed state for these genes reflects the independence of the DNA binding and transcription activation functions of Gal4p (89,90) and their differential regulation. Gal4p levels determine Gal4p-UAS, binding (91) whereas Gal80p controls the Gal4p activation function (70). Recent results suggest that the Gal4p DNA-binding domain may also play a more direct role in transcription activation (92). A significant feature in GAL gene regulation is the presence of a constitutive Gal4p-Gal80p complex. In this complex, Gal4p and Gal80p are bound rather tightly, K, = 5 x 10WgM, or about half as strong as the affinity of Gal4p for a typical GAL UAS, (see 72). Gal80p is probably always severalfold more abundant than Gal4p and thus in sufficient excess to bind the

220

R.BASHANDD.LOHR

available activator, resulting in little or no free Gal4p in the nucleus. The steps of GAL gene activation subsequent to Gal4p-UAS, binding involve changes in the Gal4p-Gal80p interaction (Fig. 8). For example, Gal3p binding to the Gal4p-Gal80p complex in galactose must alter the nature of the Gal80p-Gal4p interaction because it unmasks the Gal4p activation function. However, this presumably involves only a conformational change because Gal4p and Gal80p remain in contact in galactose (93). Gal3p interacts with Gal80p (80-83) but apparently not with Gal4p (83). Therefore, Gal80p is the mediator of the changes associated with the activation response. In addition to the above protein-protein interactions, Gal4p probably contacts other transcription factors in the course of activating transcription (Section 111,C) and both Gal4p and Gal8Op mediate chromatin changes and thus may contact nucleosomes or remodeling factors (Section 111,B). Protein-protein interactions are vital for GAL activation, as for many eukaryotic genes (94).

C. Regulatory Gene Control The GAL family provides some unique insights into this important but poorly understood aspect of eukaryotic transcriptional regulation. I. GAL4-A

NOVELPROMOTER

GalLipis the key regulator in GAL gene activation and it is constitutively produced. However, GAL4 expression does vary with carbon source (95). Transcription levels are lowest in glucose, due to global glucose repression mechanisms (91). The low cellular levels of Gal4p resulting from this repression are probably responsible for the major structural gene repression mechanism in glucose, a lack of Gal4p-UAS, binding (87, 88). Indeed, increasing the Gal4p levels (via plasmids) results in Gal4p-UAS, binding in this carbon source (79). GAL4 mRNA levels are two- to threefold higher in galactose and threeto fivefold higher in glyllac than in glucose. Gal4p-UAS, binding appears to be highly cooperative (96- 98) and thus should be very sensitive to Gal4p concentration (72). The modest GAL4 transcriptional increases in galactose and gly/lac may therefore increase Gal4p levels sufficiently to trigger activator-UAS, binding on the structural genes in these carbon sources. It does seem odd that GAL4 expression is actually higher in the inactive glyllac state than in galactose, where Gal4p activates GAL transcription. However, there is evidence that Gal4p-UASG affinity in viva may be somewhat lower in gly/ lac than in galactose (99) and the elevated transcription may be necessary to provide enough Gal4p to ensure its occupation of the structural gene UAS, in gly/lac, a crucial feature of the poised state. GAL4 transcription is mediated by promoter elements that are novel, at

YEAST

CHROMATIN

AND

GAL GENE

REGULATION

221

least within the GAL family (100). Th ere is a gene-proximal upstream essential sequence (UES) that is not a TATA element (it cannot replace or be replaced by a TATA) and a more distal activation sequence that is not a UAS, (it cannot replace or be replaced by a UAS,). The use of novel regulatory elements sets up a distinct hierarchy of control; cells can regulate GAL4 expression specifically and allow Gal4p to control the other GAL genes. GAL4 does not regulate itself nor is it regulated by Gal80p. The GAL4 promoter is inherently weak, producing very low levels of GAL4 mRNA, and resulting in low levels of gene product (70). How these low levels of Gal4p manage to carry out multigene regulation is a major mystery (72). The role played by posttranscriptional control of Gal4p is uncertain. For example, Gal4p is phosphorylated (101) and at least some phosphorylation was reported to be required for Gal4p transcriptional competence (102). However, that conclusion has been questioned (103). Phosphorylation does not affect Gal4p-UAS, binding (104). At least some phosphorylation of Gal4p is carried out by a cyclin-dependent kinase associated with RNA polymerase II (105). 2. GALgO-DUAL

PROMOTERS

Gal80p inhibits Gal4p activation activity in glucose and in gly/lac. To carry out this function, GAL80 is expressed at a similar basal level in both carbon sources. Basal expression is Gal4p independent (106) and is promoted via an Initiator motif located at or near the start site of GAL80 transcription (107). Basal transcription suffices to keep Gal80p levels in severalfold excess of Gal4p levels (72). In galactose, GAL80 expression is induced to levels that are 5 to lo-fold higher than the basal level of transcription. Induced expression depends on Gal4p (106), the GAL80 UAS,, and TATA (107), and Gal4p strongly protects the GAL80 UAS, in galactose (Fig. 7). Thus, GAL80 also has an inducible promoter that is regulated like GAL structural gene promoters. The presence of the inducible promoter on GAL80 indicates that cells must require more Gal80p under induced conditions. This is surprising because these conditions cause Gal80p inhibition of Gal4p function to be relaxed. However, it appears that Gal80p continues to modulate GAL promoters in galactose, keeping induced expression levels lower than they might otherwise be, because GAL1 expression rises to 150% of wild-type induced levels in a Ga180- (disrupted) strain grown in galactose (108). Modulating GAL1 expression might be particularly important because Gallp can cause induction and thus has the potential to autoinduce. The presence of the typical GAL inducible promoter directly links GAL80 induced expression, and higher Gal80p levels, to structural gene activity. The nature of this modula-

222

R.BASHANDD.LOHR

tory role of Gal80p in induced cells and how increased Gal80p levels might contribute to it will be discussed further (Section 111,C). GAL80 also contains a far upstream DNA element called UAS,,,,, which is required for full levels of both basal and induced transcription (107). It does not resemble any known eukaryotic promoter elements nor contain any known protein-binding motifs, but it does bind a protein from a yeast extract. The UASG,,, lies -100 bp upstream of the UAS, (Fig. 6). GAL80 is also regulated by its own gene product. For example, in Ga180s mutants, which are uninducible due to mutations in Gal80p, GAL80 becomes uninducible in galactose, just like GAL1 -10. In 4c mutants, Gal80p cannot inhibit Gal4p activation activity, thus allowing structural gene expression to occur in gly/lac. In these mutants, GAL80 expression in glyllac rises to fully induced levels. Therefore Gal80p must normally regulate its own gene in gly/lac by inhibiting Gal4p activation, much like it does on GAL1 -10, 7, and 2. 3. GAL3-A

SIGNALTRANSDUCER?

The expression pattern of GAL3 fits the role of a galactose signal transducer. In glucose, where there is no need for signaling galactose presence, there is no detectable GAL3 expression, in contrast to the GAL4 and GAL80 genes, which are expressed in all carbon sources. There is low-level GAL3 expression in glyllac and expression is induced by three- to fivefold in galactose (109). Induced expression is Gal4p and UA&-dependent. Elevated expression of GAL3 in galactose probably indicates that Gal3p does function in the maintenance of the induced state, perhaps in concert with Gallp (see above). 4. REGULATORYVERSUSSTRUCTURALGENEINDUCTION Induced expression of GAL80 and GAL3 is promoted via a Gal4p-dependent pathway requiring UAS, and TATA elements and thus resembles induction on GALl, 2, 7, and 10. However, GAL3 and GAL80 are only modestly induced while GALI, 2, 7, and 10 are very highly induced. GALI, 2, 7, and 10 have two UAS, per gene and thus will have more promoter-bound Gal4p than GAL3 and GAL80 with one UAS, each. In some studies, the number of UAS, strongly affects reporter gene expression levels (96) but in others the effects are quite modest (110). So differing numbers of UAS, may not completely explain the sizable induction differences between the GAL structural and regulatory genes. At least for the GAL80 UAS,, chromosomal context appears to affect efficiency because this element can produce higher levels of reporter gene induction when removed from its normal chromosomal location (110).

YEAST

CHROMATIN

AND

GAL

GENE

REGULATION

223

A decreased involvement of auxiliary factors in induced expression might contribute to the lower induction levels of the GAL regulatory genes. For example, deletion of the coactivators Gall1 or Med2 (see Section II&C) decreases induced structural gene expression severalfold but is reported to have little effect on GAL80 expression (111, 112). How such selectivity might be effected is an intriguing question. In 4c mutants, GAL80 expression in gly/ lac rises to fully induced levels. However in these same mutants, GAL1 -10 expression rises to levels that are two- to threefold higher than their fully induced levels. Such results are consistent with an intrinsic difference between regulatory and structural gene induction pathways and further suggest that these intrinsic differences do not depend completely on the galactose-dependent activation machinery.

III. GAL Gene-Specific Chromatin Structure The extensive genetic/biochemical knowledge base of the GAL. family provides a powerful setting for the study of chromatin structure and its role in gene regulation. My lab has been studying chromatin structure in this system since the early 1980s. Our studies use haploid yeast strains in which the analyzed gene is present in one copy per cell and in its natural chromosomal location. This provides the best assurance that the analyzed structures correspond to functional gene copies. Studies since 1985 have used an isogenic set of strains containing various “disruption mutants,” strains lacking a specified (disrupted) regulatory factor, like Gal4p, but otherwise genetically identical to the wild-type strain. These strains are used to test the involvement of specific regulatory factors in observed chromatin changes. To analyze the chromatin structure, nuclease and chemical cleavage studies are carried out on nuclei isolated as quickly as possible from early log-phase cultures (cf. 113). G ene-specific chromatin studies typically rely on some form of indirect end-label analysis (114,115). In this approach, the sites of nuclease or chemical cleavage in chromatin are mapped relative to a particular chromosomal restriction enzyme site, using blot/hybridization techniques or more recently primer extension (cf. 116). From the resulting profile of DNA accessibilities, one makes inferences about the chromatin structure. For example, an array of nucleosomes would produce bands (accessible DNA) located at nucleosome repeat-length intervals; the absolute band size would locate the nucleosomes relative to the reference chromosomal restriction site. Data of this type, which map absolute cleavage site locations, will be referred to as di gestion profiles. The pattern of nuclease accessibilities without reference to

224

R.BASHANDD.LOHR

the absolute chromosomal location of the cleavage sites can also provide useful information, particularly about structural interrelationships. This approach yields the gene-specific equivalent of the type of information obtained from bulk chromatin analyses. Data of this type will be referred to as digestion patterns.

A. Chromatin Structure in the Uninduced State (Glucose or Gly/Lac) All the known regulation of GAL gene transcription occurs through the 5’ upstream regions, within a few hundred base pairs, at most, of the transcription start site. Thus, chromatin studies have focused on those regions. 1. PROMOTERREGIONCHROMATINORGANIZATION: NONNUCLEOSOMALUA~QNUCLEOSOMALTATA The GAL1 -10 genes have been extensively studied. These genes are diintergenic region (Fig. 6). vergently transcribed from a common, -700-bp The intergenic chromatin region contains a sizable open stretch of DNA, i.e., DNA that is not protected by protein. The region is defined by its strong and uniform accessibility to nucleases such as DNase I (117) or chemical cleavage agents such as methidiumpropylethylenediammine tetracetate*iron(II) [MPE*FE(II)] (Fig. 9). All f our UAS,, as well as DNA flanking these elements, lie within this -170-bp open, hypersensitive region (HR, Fig. 6). The HR is present in all carbon sources and in the presence or absence of Gal4p, GalSOp, or both (76). Th us it is a permanent feature of this regulatory chromosomal locus. The DNA regions more proximal to the GAL1 and GALlO genes are protected from nucleases in -170-bp intervals (Fig. lo), suggesting that these sequences are nucleosomal (117-119). These nucleosomes would cover (GALIO) or closely surround (GALI) the TATA and cover (GALI) or closely surround (GALIO) the transcription start sites (Fig. 6). The structure of the entire region, HR and flanking nucleosomes, is identical in either the repressed or poised inactive state (117). The same intergenic chromatin organization is also observed when GAL1 -10 is present on a multicopy plasmid (119, 120). Thus, the region itself must contain sufficient information to determine this structure, regardless of its precise chromosomal location. Digestion profiles like these are highly suggestive but not totally conclusive. For example, the hypersensitive region could actually be covered by an unfolded and therefore exposed nucleosome, or the upstream regions that intervals could be covered by no&stone comare protected in -170-bp plexes that happen to protect a nucleosomal-like length of DNA. DNase I patterns can be used to test these possibilities. DNase I cleavage produces a char-

YEAST CHROMATIN

AND GAL GENE REGULATION

225

FIG. 9. Hypersensitive regions upstream of GAL1 -10 and GAL80. Chromatin profiles across the GAL1 -10 intergenic region from wild type (wt) are shown in the left two tracks: a DNase I digest of chromatin from glucose-grown cells (D); an MPE.Fe(II) digest of chromatin from cells grown in gly/lac (g). The protection in the center of the hypersensitive region in gly/ lac is due to Gal4p binding to the UAS,. Maps locating the DNA regulatory features on the GALI -10 intergenic region are shown to the right of each profile. The rightward three tracks (1,2,4) show MPE.Fe(II) profiles from the GAL80 upstream region for a naked DNA digest (n), a chromatin digest from wild-type cells grown in glyjlac (g), and a chromatin digest from Ga14disrupted cells (4”) grown in galactose. Maps locating the regulatory elements in the GAL80 upstream region are shown to the left of track 1. U, UAS,; T, TATA; the origins of the wavy arrows locate the transcription start sites. The GALI -10 sites are mapped from an EcoRI site in GALIO. The GAL80 sites are mapped from an EcoRI site in GAL80 (at 1050 bp). From Lohr (99). Copyright 1993 National Academy of Sciences, U.S.A.

acteristic - lo-bp intracore ladder pattern because the DNA in core nucleosomes lies on (wraps around) a surface created by the histone octamer (cf. 34), rather than being more uniformly covered or wrapped as in regulatory

226

R. BASH AND D. LOHR

-I

FIG. 10. MNase profiles across the GALI-IO intergenic region. Profiles l-4 are mapped from an EcoRI site in G&IO. Profile n is a naked DNA digest. The chromatin profiles 1,2, and 4 are from cells grown in (1) glucose (-), (2) galactose (+), and (4) galactose into stationary phase (St). Profiles 6 and 7 are mapped toward the intergenic region from an EcoRI site located in GALI, -1.9 kb from the GAL10 RI site. These chromatin profiles correspond to growth in (6) glucose (-) and (7) galactose (+). The maps to the left (l-4) and right (6-7) locate the various DNA regulatory features and transcription units (arrows with numbers). The arrows to the left of profile 1 locate the hypersensitive region; A, B, and C correspond to the nucleosomes in the intergenic region (Fig. 6). Filled circles (0) to the left of profile 1 and to the right of profile 6 correspond to accessible DNA lying in the intergenic region; open circles (0) correspond to accessible DNA lying in the coding region. The stitched arrows at the bottom of the maps represent probe DNA locations. From Lohr (117), by p ermission of Oxford University Press.

protein-DNA complexes. Thus, the intracore pattern is diagnostic for the presence of nucleosomes, even unfolded nucleosomes (see Section 1,B). To apply this diagnosis, I challenged a DNase I intracore ladder pattern [trans-

YEAST CHROMATIN

I

“R.. I

227

AND GAL GENE REGULATION

---.+

9

-

a

4

b

FIG. 11. DNase I ladder patterns from the GALI-IO intergenic region, The intercore/intracore transition region is shown in patterns 2 and 4. These patterns were obtained with probes from the hypersensitive region (2) and nucleosome C region (4), i.e., the Hinfl and 6 probes, respectively (upper map). More of the intracore region is shown in patterns 5-7 and the patterns are as follows: (5) nucleosome A region (l3 probe), (6) the hypersensitive region (Hinfl probe), and (7) nucleosome C (6 probe). Band 9, corresponding to 95 bp, is located in the patterns. The symbols (A) to the right of pattern 7 mark the intracore bands. All patterns were from inactive GALI chromatin (-). From Lohr (117), by permission of Oxford University Press.

ferred to a diazobenzyloxymethyl (DBM) filter] with three very small probes: (1) from putative nucleosomal region A, (2) from the hypersensitive region, and (3) from putative nucleosomal region C. The three patterns were obtamed by repetitive hybridizations of the same DBM filters, minimizing the possibility of artifactual differences. Because the intracore pattern is small, to -126 bp at most in yeast (see above), and because the probes are small, the information obtained in the autoradiogram reflects the chromatin struc-

228

R. BASH AND D. LOHR

ture in these very local areas. To maximize the possibility of detecting a nucleosome on the HR, profiles from glucose-grown cells were analyzed. The results show quite typical nucleosomal intracore ladder patterns from putative nucleosomal regions A and C but a smear from the hypersensitive region (Fig. 11). These data provide strong evidence that protected regions A and C are indeed nucleosomal, not merely nucleoprotein covered, and that the hypersensitive region is nonnucleosomal. There is also independent supporting evidence for the nucleosomal nature of protected region B (121,122). Thus, GALI -10 intergenic chromatin consists of a nonnucleosomal hypersensitive region in which the UAS, reside, with nucleosomes on the surrounding sequences. Other GAL genes have the same promoter region chromatin organization. On GAL7, the UAS, lie in an open hypersensitive region surrounded by nucleosomes, one of which covers the TATA (119). Although its chromatin structure has not been determined, the GAL2 UAS, are constitutively accessible to Gal4p (79), suggesting that they also lie in an open region. Even the regulatory gene GAL80 has the same basic upstream chromatin organization (99). Both DNase I and the chemical cleavage agent MPE*Fe(II) detect an - 150-bp, upstream hypersensitive region surrounded by nucleosomes (Fig. 9). The HR is constitutively present, uniformly hypersensitive, and contains the single UAS, found on GAL80 (Fig. 6). The gene-specific UAS GAL8o also lies in the HR, at its upstream end. The downstream nucleosome (D, Fig. 6) covers the GAL80 TATA and transcription start sites. Thus, it appears that this basic chromatin organization is conserved and therefore characteristic of GAL gene upstream regions, whether divergently organized like GAL1 -10 or present as single genes. The coding regions of GAL genes are nucleosome covered. For example, MNase digestion produces an exceptionally clear nucleosomal ladder pattern from the coding region of GAL1 in the inactive state (123). This very distinct pattern probably arises from regularity in intemucleosomal spacing (117) rather than specific nucleosome positioning. MNase and MPE*Fe(II) profiles also demonstrate regular nucleosomal organization on the coding regions of GALI, 7, and 10 in the uninduced state (127, 119, 120).

a. What Keeps the Hypersensitive Region Nucleosome Free? The open nature of the HR is enigmatic. This stretch of DNA is very accessible to large exogenous reagents, e.g., enzymes, yet completely excludes nucleosomes, even though it is large enough to be nucleosome occupied. To some extent the HR and upstream nucleosomal regions behave like autonomous domains. The upstream nucleosomes around the GAL1 -10 HR undergo rapid, expression-dependent nucleosome association/dissociation reactions during gene activation (see below) without detectable effects on the HR structure,

YEAST CHROMATIN

AND GAL GENE

REGULATION

229

and Gal4p can bind to the UAS, within the HR (in gly/lac) without disturbing the structure of the surrounding nucleosomal regions. Several explanations for nucleosome absence from the HR can probably be rejected. 1. No affinity for histones. In vitro, nucleosomes form on the HR DNA sequences even more readily than on the surrounding DNA sequences that are nucleosome covered in vivo (124). Thus, a lack of DNA-histone affinity is not responsible for the nonnucleosomal nature of the HR in vivo. Also, there is no nucleosome positioning preference on the intergenic region in vitro, so positioning is determined by in vivo features. 2. Specific regulatory protein presence. It was suggested that a pleiotropit yeast factor, now called Reblp (or Grf2p), keeps the GALI-IO HR nucleosome free in vivo and determines nucleosome organization on the surrounding sequences (125, 126). Reblp has a binding site that overlaps two of the GAL1 -10 UAS,. However, it has now been shown that the positioning of nucleosome B and the distinctive MNase profile across the intergenic region are unaffected by mutations in the Reblp binding site that abolish in vivo binding of the protein (122). This calls into question the role of Reblp in upstream nucleosome organization and in HR maintenance on GAL1 -10. Also, Reblp can only bind to its site in glucose, yet the HR is present in every carbon source (117). Furthermore, GAL7 and GAL80 have HR that are similar to the GAL1 -10 HR but no Reblp binding site, so at least there must be alternative mechanisms to maintain such open structures. 3. HR contacts with the nuclear superstructure (matrix, nucleoskeleton) such that nucleosomes are excluded. Given the complete and uniform hypersensitivity of the HR, it is hard to imagine that the HR could be bound to the nuclear superstructure or to any large structural protein. For example, when Gal4p binds to the UAS,, this binding can be detected within the HR as a protected region (cf. Fig. 9). Also, high-resolution DNase I footprinting detected Reblp protection of DNA with in the HR (in glucose) but otherwise the profile from this region looked much like naked DNA (122). Th us, major protein binding seems unlikely. Perhaps in vivo organization keeps the HR region in a higher order structure that disfavors nucleosome formation. For example, right-handed coiling of the HR DNA might be inhibitory to formation of the leftward-coiling nucleosome. Features intrinsic to DNA, such as bending or torsional con-

230

R. BASH AND D. LOHR

&mints, or features implemented by proteins or specific ligands might produce a structure that disfavors nucleosome formation. Partly for these kinds of reasons, we are involved in trying to determine the intrinsic DNA structure of this region and its higher order organization (Section 111,B).

b. Nucleosome Positioning on GAL Gene Promoters. Nucleosome occupation of specific positions on gene control regions has generated a great deal of biological interest. However, precise positioning is thermodynamically expensive to maintain, and the more precise the positioning, the higher the cost (cf. 33). Th us it is more likely, at least in general, that promoter region nucleosome locations will simply be biased toward preferred positions, resulting in distributions of actually occupied locations around the preferred ones. A careful analysis of nucleosome locations on the mouse mammary tumor virus long terminal repeat (MMTV LTR) in mouse chromatin has detected just such behavior (127). Alternate nucleosome positions in such distributions might vary by 10 bp to preserve a particular rotational setting; this would be consistent with the discrete linker variation observed in yeast and other genomes. Many of the positions on MMTV do vary by -10 bp or multiples thereof (127). For the GAL upstream regions, there is as yet no compelling evidence that nucleosomes occupy absolutely precise positions, but a great deal of evidence suggests that there is strong bias toward preferred positions. It would certainly be tempting to infer precise positioning based on the very distinctive MNase band profiles across the GAL1 -10 intergenic region (cf. Fig. 10). Unfortunately, the similarity between naked DNA and chromatin digests in some parts of the profiles suggests that DNA sequence-specific cleavage may contribute significantly to the chromatin profiles. MNase commonly shows this tendency, which makes it a risky probe for positioning analysis. The chemical cleavage agent MPE*Fe(II) generally shows very low sequence specificity and a strong preference to cleave within linker regions (128) and thus can provide a more stringent test for positioned nucleosomes. If nucleosomes are absolutely positioned, the only accessible DNA will be the short stretch of linker, and bands in an MPE*Fe(II) digestion profile will form a very distinct ladder. The less precise the positioning, the wider the bands and the more diffuse the ladder. MPE*Fe(II) analysis of GAL1 -10 chromatin in multicopy plasmids was able to demonstrate a distinct nucleosomal ladder profile across the intergenic region and into the 5’ end of GAL1 (120). However, that work was done at low resolution and really cannot be interpreted as demonstrating absolutely precise positioning for nucleosomes AC, although it certainly indicates there is at least a strong bias toward preferred positions on the intergenic region. We have been unable to obtain a distinct MPE*Fe(II) ladder profile from a higher resolution, single-copy

231

YEASTCHROMATINANDGALGENEREGULATION

analysis of the GAL1 -10 intergenic region (J. Lopez and D. Lohr, unpublished observations). Those nucleosomes that abut a chromosome structural boundary, like the HR, may demonstrate the most precise positioning because such boundaries can provide nucleosome positioning constraints (129). On GAL1 -10, these would be nucleosomes A and B. In fact, specific positioning of nucleosome B in the single genomic copy of GAL1 - 10 has been suggested from highresolution MNase (130) and DNase I profile analyses (122). The difficulties with MNase mapping were discussed above. DNase I lo-bp ladder profiles also have a drawback; they cannot distinguish nucleosome occupation at a unique position from occupation at multiple positions differing by 10 bp. The only sure way to confirm a particular nucleosome position from a DNase I ladder profile would be to match the ladder band intensities from the chromatin profile to DNase I intracore cleavage frequencies. However, intracore frequencies have only been determined in vitro (cf. 134, and not with yeast nucleosomes, and those frequencies could not predict the intensities of the in viva intracore genomic pattern from yeast (17). For example, to match the genomic intensity pattern it was necessary to enhance the in vitro cutting frequency around the nucleosome dyad. Thus, GAL1 -10 nucleosomes A-C clearly have strong locational preferences, perhaps B in particular, but the precise degree of their positioning remains uncertain. MPE*Fe(II) analysis has been able to detect a distinct profile, and thus a positioning preference, for nucleosome D on GAL80 (99). Nucleosome D abuts a chromosomal boundary, an HR, but its positioning is clearly not absolute. Based on bandwidths in a moderate resolution analysis, nucleosome D location varies over at least a 20- to 30-bp range (Fig. 9). Even though not absolute, this level of positioning is significant; the TATA and start site would remain within nucleosome D throughout this range of positions. The MPE*Fe(II) profiles also suggest that the first nucleosome lying upstream of the GAL80 HR, nucleosome (Y (Fig. 6), may have some positioning preference (Fig. 9). However, the region further upstream from 01 is uniformly cleaved and does not show a nucleosomal profile at all (Fig. 9), so it is difficult to estimate the degree of positioning for (Y. The size of the protected cx region is about 160 bp, which suggests that the CYcore of 126 bp (Section 1,B) may reside within an -30-bp-wide window of possible positions. DNase I produces a lo-bp ladder profile from the 01 region (Fig. 7), consistent with some degree of positioning for this nucleosome. 2. STRUCTURE/FUNCTIONRELATIONSHIPSFOR

GALPROMOTER

REGIONCHROMATININTHEINACTIVESTATE

a. The Hypersensitive Region-Constitutive UAS, Accessibility. Location of the UAS, within a nonnucleosomal region appears to be a general fea-

232

R.BASHANDD.LOHR

ture on GAL genes. The analyzed genes (GALI, 10, 7, and 80) were single copy; thus the structures observed clearly reflect functional genes, not redundant or silent copies. An absence of nucleosomes from the UAS, and nearby DNA probably ensures that Gal4p will always have access to these elements. Thus, nucleosome removal in order to allow activator-DNA binding will not be required for GAL. transcriptional activation, as it is on genes such as PHO5 or MMTV (6, 132), nor does competition between Gal4p and nucleosomes for the DNA region that contains the UAS, play a role in GAL regulation. This constitutive chromosomal accessibility undoubtedly facilitates the binding of Gal4p to the UAS, in gly/lac and thus enables the very rapid induction possible in this carbon source. Rapid inducibility from poorer carbon sources such as gly/lac would allow yeast growing in the wild to utilize quickly the better carbon source galactose whenever it is encountered, and might be especially advantageous in brief encounters. Thus, this type of chromatin organization on the GAL promoter regions probably provides advantages to the organism. Other rapidly inducible genes, e.g., the heat-shock genes, also maintain constitutively accessible promoter elements (cf. 133). Genes such as PHO5 and MMTV at least partly cover their major promoter elements with nucleosomes (132) and thus can use competition between activators and nucleosomes as part of a regulatory strategy. GAL gene activation does involve competitive binding between factors and nucleosomes for a particular stretch of DNA, but for the TATA rather than the gene-specific promoter elements (Section 111,C). 3. THEUPSTREAMNUCLEOSOMES-BASALREPRESSION The nucleosomes on the GAL upstream regions help maintain the inactive state of expression. Because those nucleosomes cover (G&7,10, and 80) or closely surround (GALI) the TATA, and TBP cannot bind to TATA that lie within nucleosomes (34), they will hinder TBP access to this crucial promoter element. Indeed, partial nucleosome depletion allows some TATAdependent GAL1 expression in glucose (134), presumably due to upstream nucleosome loss. Partial loss of nucleosome B in these mutants has been directly demonstrated (121). Interestingly, the transcription observed when histones are depleted does not require the presence of a UAS, (134). Apparently, even under repressing conditions and without the aid of processes mediated through the UAS,, at least some transcription machinery lies near enough to GAL1 to access it when nucleosome interference is removed. In a perhaps related observation, run-on transcription can be observed at the 5’ end of GAL1 in glucose-grown cells that have been detergent permealyzed, a treatment that probably destroys the chromatin structure (135). The repressive function of these upstream nucleosomes depends on their

YEASTCHROMATINANDGALGENEREGULATION

233

N-terminal tails (43). Tails from all four core histones are required; the loss of any one pair, even H2A or H2B, results in diminished repression. The regions in the tails that are most crucial for repression lie close to the boundary between the tails and the structured central regions. Perhaps the loss of a pair of regions that lie so close to the folded histone core destabilizes it. However, specific removal of the basal repression domains does not cause significant chromatin structural change (43). Alternatively, these repression domains could hinder transcription factor access to the nucleosome. In vitro, histone tails have this ability (7). Partial loss of nucleosome A-C protection (118) and low-level GAL1 expression (108) are also observed in Gal80- (disruption) mutants in glucose, suggesting that Gal80p plays a role in maintaining upstream nucleosome presence, a feature more directly indicated by results discussed in Section 111,B. The repressive function of the histone tails may involve their interaction with specific factors such as Gal80p that are involved in maintaining nucleosome presence. Nucleosome D on GAL80 functions in a somewhat different situation than nucleosomes A-C on GAL1 -10. Nucleosome D will certainly block TBP access to the GAL80 TATA in nongalactose carbon sources, thus inhibiting induced expression. However, its presence would also be expected to interfere with basal expression, which is promoted through an initiator element located at + 1, well within nucleosome D. How basal expression deals with nucleosome D is unclear.

B. Chromatin

Structure

in the Induced

State (Galactose) 1. PROMOTERREGIONS-TRANSIENTNUCLEOSOME DISRUPTION/DEPOSITION

a. Gal4p-Dependent Disruption. Th e h ypersensitive region containing the GAL1 -10 UAS, shows no carbon source dependence. However, growth in galactose causes large changes on the surrounding nucleosomal regions (117,119, 221,122,136, 237). Instead of being protected in 170-bp multiples, regions A, B, and C show high accessibility to DNase I or MNase and the chromatin digestion profiles become very similar to naked DNA profiles (Fig. 10, tracks 2 and 3). The extensive exposure of the DNA within these regions suggests that GAL1 -10 activation is accompanied by the complete removal of upstream nucleosomes A-C. This should facilitate transcription by exposing the TATA and transcription start sites to cellular factors such as TBP. Also, the sites of initial DNA unwinding for GAL1 -10 transcription (138) lie in the regions from which nucleosomes A and C are removed; the release of the supercoiling restrained by those nucleosomes could aid this initial unwinding (however, see 139).

234

R. BASH

AND D. LOHR

These extensive changes are always observed in all three (A-C) regions, indicating that all three are equally and simultaneously affected. Thus these changes may reflect the action of a cooperative process involving all three nucleosomes, like the changes in a four-nucleosome stretch upstream of PHO5 (6,132). Induction also results in extensive changes in GAL7 upstream nucleosome structure (119). The term “nucleosome disruption” will be used to refer to promoter region nucleosome changes, large or small, that occur in association with gene activation. Disruption is common on eukaryotic genes (cf. 4) and its features on PHO5 and MMTV (132) and PHO5 and GAL genes (6) extensively discussed. Nucleosome disruption on GAL genes is Gal4p dependent (119, 121, 137). Disruption is linked to the active state of expression but does not depend on transcription (121, 140). For example, disruption is unaffected in a GAL1 TATA mutant or even when the TATA is deleted, either of which will largely abolish transcription. Nor does disruption depend on sequences downstream of the TATA. Promoter region nucleosome disruption is a dedicated function of the activator (121) and is not simply an indirect effect of the transcription process. Gal4p-dependent disruption was also observed on GAL1 -10 multicopy plasmids in galactose-induced cells, but the change was interpreted to reflect only a partial nucleosome loss (120). The observation of partial changes might be due to copy number effects, such as heterogeneity in activation. For example, less extensive upstream disruption was noted in low-copy-number plasmid chromatin compared to a single integrated copy of the same construct (119).

b. GalBOp-Dependmt Nucleosome Deposition. When we first switched to the isogenic set of strains, we did not observe any upstream nucleosome disruption on GAL1 -10 under induced conditions (UB), in striking contrast to earlier work in other strains (117). We eventually realized that this is due to certain steps in the nuclear isolation protocol. To make nuclei, the yeast cell wall must first be removed, to produce osmotically sensitive spheroplasts that can be lysed to release nuclei. Cells are essentially starved during this -1-hr cell harvest and spheroplast pretreatment. This produces a general shutdown in transcription that can be reversed by metabolic reactivation of the spheroplasts with a carbon source and casamino acids (141). When these cells were metabolically reactivated (during the final step of spheroplast formation), GAL1 -10 nucleosomes A-C were indeed found to be disrupted in the chromatin digests from induced cells (Fig. 12). The disrupted profiles resemble those observed earlier in other strains (117) and are consistent with results obtained from techniques applied to intact cells (121, 136). This disruption is both Gal4p and galactose dependent, in both the reactivation step and during preliminary cell growth (Fig. 12). The galactose

YEAST CHROMATIN

AND CAL GENE REGULATION

235

x

A

FIG. 12. Metabolic reactivation produces Gal4p-dependent nucleosome disruption. MNase profiles were mapped from the EcoRI site in GALlO. Profiles 1-4 are as follows: (1) a naked DNA digest (n) and (2-4) chromatin digests from wild-type cells grown in galactose and not reactivated (G) or reactivated in galactose (GR) or a chromatin digest from a gal4- and gal80-disrupted strain (4D/80D) grown and reactivated in galactose (GR). The map to the left locates the regulatory DNA features; A-C correspond to upstream nucleosomes in Fig. 6. Note that the nucleosome profile on the intergenic region is similar in patterns 2 and 4 but the coding sequence profile is not, for example, at the nucleosome boundary downstream of the GAL1 transcription start site. This demonstrates the differential responses of these two regions (see text). T=TATA, and the left ends of the wavy arrows locate transcription start sites. The hypersensitive region is located to the right of track 3 (HR). Th e 1ocations of sites that undergo induction-dependent structural changes in vivo are shown (X). (136). From Lohr and Lopez (137).

236

R. BASH AND D. LOHR

dependence in reactivation indicates that this disruption process is specific; only galactose can trigger it. The requirement for galactose during cell growth probably reflects the dependence of disruption on earlier steps in the galactose induction pathway. Taken together, these results suggest that the nucleosome disruption triggered in spheroplasts by reactivation produces the usual disrupted upstream chromatin structure and thus reflects activity of a bona fide in vivo disruption process. That it is necessary to regenerate (by a reactivation process) the disrupted upstream region structure in chromatin from induced cells indicates that nucleosomes are being redeposited on these regions, probably in response to the brief starvation during or after cell harvest. Reactivation then removes them. This nucleosome redeposition does not occur, i.e., the in vivo disrupted structure persists through harvest/isolation, in a strain that lacks Gal80p (Fig. 13). Thus, th e nucleosome redeposition on the GAL1 -10 upstream region in wild-type cells depends on Gal80p. This Gal80p-dependent nucleosome deposition process produces the same upstream chromatin structure seen in inactive, noninduced cells (cf. Fig. 12, profiles 2 and 4). Therefore, Gal80p mediates an in vivo process that deposits nucleosomes and thus generates the promoter region nucleosomal structure associated with the inactive state of GAL1 -10. The nucleosome disruption triggered by reactivation is rapid-as few as 3-4 min of reactivation is sufficient to observe it. Redeposition takes place within an -1-hr protocol but may also occur within minutes because analogous changes on PHO5 are triggered in minutes by spheroplast treatments (242). Disruption on PHO5 can also be rapidly triggered in spheroplasts. Apparently, mechanisms to remodel upstream nucleosomal regions rapidly and reversibly are general features. The rapidity indicates that these mechanisms are DNA replication independent (see also 142). On PHO5, c. Metabolic Competence Affects Disruption/Deposition. nucleosome disruption in spheroplasts requires the addition of glucose, probably to provide energy. An energy requirement for nucleosome disruption is consistent with in vitro observations that nucleosome remodeling machinery requires ATP to function (143). Galactose, the signal that triggers disruption on GAL genes, will also provide an energy source. However, energy alone is not sufficient because other carbon sources such as glucose do not trigger the GAL1 -10 disruption (137). An energy requirement should link nucleosome disruption to cellular energy states and ATP availability. Indeed, the disrupted GALJ -10 upstream nucleosome structure disappears when induced cells are grown (in galactose) into stationary phase (Fig. lo), where metabolic activity slows down significantly. Thus, disruption depends on activators, activating signals, and metabolic competence; all must be favorable. Energy

YEAST CHROMATIN

AND

GAL GENE

REGULATION

237

W n

G

FIG. 13. GaBOp-dependent nucleosome deposition. MNase profiles were mapped from the EcoRI site in GALIO. Profiles are as follows: (1) a naked DNA digest (n) and (2-3) chromatin digests from wild-type cells grown in galactose and not reactivated (G) or reactivated (CR) in galactose. Profile 5 is a chromatin digest of a Gal80-disrupted strain grown in galactose and not reactivated (G). The map to the left locates the DNA sequence features on the intergenic region (Fig. 6). The left ends of the wavy arrows locate transcription start sites (T=TATA). From Lohr and Lopez (137).

is required for other steps in the activation pathway, for example, the Gal3pdependent activation of Gal4p. d. GAL80 Disruption May Involve Modest Changes. Like nucleosomes A-C on GAL1 -10, GAL80 nucleosome D provides typical nucleo-

238

R. BASH AND D. LOHR

1

2

3

FIG. 14. Nucleosome disruption on GALRO. MNase profiles were mapped from an EcoRI site in GAL80 at 1055 bp. Profiles l-3 are chromatin digests from wild-type cells grown in galactose and not reactivated (G) or reactivated (GR) in galactose. The GAL80 upstream sequence features are located in the map on the left; U, UAS,; T, TATA; the origins of the wavy arrows locate transcription start sites. The location of the hypersensitive region is shown between profiles 1 and 2 (HR). The symbol (+) to the right of profile 2 marks the region of enhanced nuclease sensitivity in reactivated chromatin. Note also the increased protection near the UAS, region in reactivated chromatin. From Lohr and Lopez (137).

somal protection to the GAL80 TATA/start site regions in digests from induced, unreactivated cells, but becomes nuclease sensitive when spheroplasts are reactivated (Fig. 14). The changes are Gal4p and galactose dependent (137). However, the degree of DNA exposure in region D is not as extensive as the exposures in GALI -10 regions A-C. In fact, nuclease sensitivity is localized to the region of the GAL80 transcription start site, which lies near the middle of nucleosome D (Fig. 14). These results suggest that disruption on GAL80 may not involve complete nucleosome loss. A modest in-

YEAST CHROMATIN

AND

GAL

GENE

REGULATION

239

crease in nuclease sensitivity could reflect the loss of an H2A-H2B dimer, a change known to be associated with transcriptionally active chromatin (144) or a conformational transition that exposes the central region of the nucleosome (6). Nucleosome OL,which lies upstream of the HR (Fig. 6) and contains no known GAL80 regulatory elements, also shows expression-linked behavior. In chromatin from induced cells, nucleosome (Y protection shrinks by -40 bp on its upstream side, suggesting a loss of histones or a change in position under induced conditions (99). This feature is Gal4p dependent and thus linked to induced GAL80 expression. However, the change does not depend on and is not affected by reactivation; thus it differs from the changes on nucleosome D. Nucleosome (Yalso shows an -20-bp-wide alteration near its downstream (gene-proximal) end in cells that lack Gal80p, indicating a Gal80p involvement with this far-upstream nucleosome. The two affected DNA regions would be in proximity on the termini of nucleosome ct. Thus, although distal from the promoter, nucleosome (Yappears to play a role in induced GAL80 expression. This may involve contacts with promoter elements in the HR, such as the UAS, or UASGALso, or the TATA. The DNase I lo-bp ladder observed in the (Y region profiles (Fig. 7) actually appears to extend beyond the o-HR boundary, which is mapped by hypersensitivity analysis (Fig. 9), into the hypersensitive region (99). This may reflect an interaction between proteins on the HR and nucleosome CL 2. CODING REGIONS-NUCLEOSOME CONFORMATIONAL CHANGES/REARRANGEMENT MNase digestion produces a very clear nucleosomal ladder pattern from the GAL1 coding region in the inactive state, whether analyzed on native (nondenaturing) or denaturing gels. When GAL1 expression is induced, this clear pattern is altered (123). 0 n native gels, the pattern becomes quite smeared and uninterpretable but on denaturing gels a residual nucleosomal ladder can still be seen. However, there is significant intensity between the ladder bands (Fig. 15). The active GAL1 patterns are strikingly similar to the low-ionic-strength genomic patterns (Fig. 15) and, as was the case for those patterns, the simplest explanation is that nucleosomes are still present on the DNA but have undergone a conformational transition that exposes the intranucleosomal regions to nuclease cleavage. This internal cleavage involves single-strand nicking (123); thus the patterns on native gels are smeared. Denaturing gels were also used in the genomic analyses (20). As in the low-salt genomic digests, the enhanced interband intensity in the active, gene-specific MNase digests forms a discrete pattern of bands spaced by - 10 bp, much like a DNase I pattern (Fig. 15). Again, this suggests that the DNA remains protein associated in the altered nucleosome and also

240

R. BASH AND D. LOHR

1

4 4 4 4

@ .::

gj D ‘Tg];, +

-

+

FIG. 15. Gene-specific patterns from the GALI coding region. Patterns 4 and 5 show MNase digests from the GAL1 coding region chromatin in the inactive (-) or induced (+) expression state. Electrophoresis is from top to bottom. The scans to the right compare the gene active pattern from 5 (+) to a bulk chromatin low-salt digest (0). Roman numerals II and I identify di- and mononucleosomal DNA peaks, respectively. Direction of electrophoresis is left to right in scans. Pattern 1 is a DNase I digest from the GAL1 coding region in the induced state (+). Intracore bands (A) and intercore bands (A) are identified. Intracore bands 8 and 11 correspond to 84 and 115 bp, respectively. From Lohr (20).

argues against template heterogeneity as an explanation for the pattern changes, i.e., some genes (inactive ones) still contain nucleosomes and some (the active ones) do not. In that case, the interband DNA should form a smear, not specific bands. Results discussed below are also inconsistent with a template heterogeneity model. Thus, it appears that the nucleosomes on the active GAL1 gene undergo changes but remain present under induced conditions. Other results confirm this suggestion. DNase I digests continue to produce the -lo-bp ladder patterns characteristic of nucleosomes and do not

YEAST CHROMATIN

AND GAL GENE REGULATION

241

show smears as they would if nucleosomes were lost (Fig. 15). Both the intracore and intercore patterns are present in the gene-specific digests and they resemble patterns obtained using a total genome probe (123). Psoralen cross-linking approaches, which provide a different and very powerful test for nucleosome presence, also indicate that nucleosomes are not lost from the active GAL1 coding region (119). GAL1 is . single copy in these cells, its expression is required for cell growth (in galactose), and it is transcribed at an exceptionally high level. GAL1 thus provides one of the best possible templates for a determination of the fate of nucleosomes on transcribed genes in viva, making the conclusion that nucleosomes can remain present on actively transcribed chromatin especially compelling. This conclusion is consistent with the more recent in vitro observations that RNA polymerase can transcribe through nucleosomes without detectable octamer loss from the DNA template (145, 146). MNase profiles from the GAL1 coding region also show a loss of nucleosomal character in the induced state (Fig. 10) (117, 119, 137). The profile changes and the DNase I ladder patterns would be consistent with either a conformational change or simply a reshuffling of nucleosome locations on the actively transcribed GAL1 gene. However, the MNase pattern changes described above require that the nucleosome undergoes a conformational transition. Both, reshuffling and a conformational change, may occur. A reshuffling of nucleosome positions is observed in vitro after RNA polymerase transit through nucleosomes (145). The response of GAL1 coding-region nucleosomes to gene activation is fundamentally different from the response of the promoter region nucleosomes A-C. First, upstream nucleosomes seem to be completely dissociated from the DNA; coding-region nucleosomes undergo conformational changes locational rearrangements but do not disappear. Complete nucleosome removal may be necessary for the massive RNA polymerase II (pal II) initiation apparatus to operate while RNA polymerase can elongate through the nucleosomes on genes without the need for such drastic alterations. Second, metabolic reactivation, which causes striking changes on the promoter nucleosomal regions, has absolutely no effect on the altered coding-region structure (137). Those alterations are present to the same extent with or with out reactivation. Thus, unlike the upstream changes, the altered nucleosome structure of the coding region is not subject to rapid reversal. The coding region changes are probably consequences of expression whereas the upstream changes are a facet of gene regulation and thus are tightly controlled. The control of promoter region nucleosome presence is mediated, directly or indirectly, by the major GAL-specific regulators, Gal4p and Gal80p. Their constitutive association should help the regulators carry out these reciprocal nucleosome transactions, and their promoter region location should place them

242

R. BASH AND D. LOHR

in good position to do so. Thus, promoter region nucleosomes have unique regulatory functions and as a result different regulatory constraints than coding-region nucleosomes. 3. CHROMOSOMALARCHITECTURE OF PROMOTER REGIONS IN THE INDUCED STATE a. intrinsic DNA Structure. DNA bending plays a prominent role in prokaryotic transcriptional regulation (147,148). However, except for a few examples, such as TBP bending of the TATA (cf. 149), DNA bending is not as strongly implicated in eukaryotic transcriptional control. Bending can be an intrinsic, sequence-dependent feature of DNA (150). Intrinsically bent DNA is permanently curved and more rigid than random DNA. We have looked for intrinsically bent DNA on the GAL1 -10 and GAL80 promoter regions, first by applying several computational algorithms developed to predict DNA bending (X1-153). Th e various algorithms all predict bends in the upstream regions but give quite different predictions for the actual locations and extents of the bending. Based on the predicted bending sites, we carried out restriction enzyme digestions of cloned DNA to cut the regions into various small fragments and electrophoresed the fragments under conditions (low temperature and low voltage) that maximize the anomalously slow migration of bent DNA (cf. 154). Fragments with bends were identified by comparing their gel mobility to molecular weight standards. Candidates demonstrating anomalous migration in the above test were examined by circular permutation analysis (154a). In this technique, a dimeric concatamer of the fragment of interest is made by cloning; the dimeric DNA is isolated and cleaved with various restriction enzymes that cut once within a monomer unit. Thus, each digest contains a unit length of DNA but in the various digests the intrinsic bend will lie at various relative positions within the unit. The nearer the bend is to the middle of the fragment, the slower the fragment migrates. A plot (Fig. 16) locates the center of bending and allows an estimation of the bending angle (155). The results are quite suggestive (Fig. 17). On CAM-IO, we have confirmed two strong bends. The strongest is centered at 645 bp, 30 bp upstream of the TATA, and at the downstream edge (gene proximal) of nucleosome B. The center of this bend thus appears to be in phase (three helical turns) with the strong bend that will be formed on the TATA when TBP binds to it. It has been shown that insertion of an intrinsically bent sequence upstream and in helical phase with a TATA can strongly increase TBP-TATA affinity (156). Th e p resence of this intrinsic bend could thus enhance TBP affinity at the GALJ TATA. Its calculated bending angle is 70” and, if bending occurs in the same direction as the bend induced by TBP binding, would result in a total bend of ~150” in this small region when TBP binds to the

YEAST CHROMATIN

550

243

AND GAL GENE REGULATION

580

610

640

670

700

730

760

790

820

850

880

Enzyme

cut

!sita

FIG. 16. Circular permutation analysis of a GAL1 -10 bend. The relative mobility of various circularly permuted monomers is plotted versus restriction enzyme cut site, for two separate experimental determinations. When the restriction enzyme cuts at the intrinsic bend, the fragment has the highest relative mobility (0) Run 1; (0) Run 2.

TATA. The second strongest bend (-50”) lies at -525 bp, which is near the upstream edge of nucleosome B. Thus, nucleosome B, a critical nucleosome in disruption, is bracketed by intrinsically bent DNA. The Goodsell and Dickerson algorithm has been quite successful for the GAL genes. It predicts five strong bends on the GAL1 -10 and GAL80 upstream regions. Our initial screen did not detect evidence for other intrinsically bent sequences besides those and we have verified three of the predicted bends. The two remaining are predicted to lie at or near the edges of nucleosome A (Fig. 17), another GAL1 -10 nucleosome that is disrupted during galactose induction. We are currently testing for those bends. From the above results, it appears that intrinsically bent DNA sequences on the GAL1 -10 upstream region lie near nucleosome termini. It has been suggested that nucleosomes might prefer to locate intrinsic bends in their

244

R. BASH AND D. LOHR 330 \$

170 + I 0

I w GAL10

~;..,~.~..:.~,~.::.;..:;.~ ..,...;..~...;..,..,; :,.. ;:.,.

I 200

WI

LIJ 400

.... :; .... ;.:...:..:;., ‘,.I;;,.,. .,;:,.:: ;,.”

....;..

----

.;..,.

s

A

I 200

I

rq ‘4 520

‘&

I 800

I

a% B

I 400

I

‘.,~,;.,~,;..~.,~.:..,~..:,.~~.: ..,.~.~..~.~,~..~:. ..:.,,:.: ....‘.,I....~~~.~.;;,~.,:...~.,:..,~..~,,‘ .~~..,~~~~~.;~~~..,;..~..~~~~~.~~~~ ..~.,~. ..,.‘.:,.,.‘.;::.,; ..,.i

HR

=

I 600

LJI

I T

HR L Hypersensitive Region

I

C

I

,$

I

I

1000

800

=TATAbox

FIG. 17. Intrinsically bent DNA on the GAL1 -IO and GAL30 upstream regions. The sites of predicted (dashed arrows) and confirmed (solid arrows) bends on the GAL1 -10 and GAL80 promoter regions are located on the map shown in Fig. 6, with the positions of the bend center given by the number above (GALI-IO) or below (GAL80) the arrow.

central region, where DNA is kinked by nucleosome structure (25, 34). Intrinsic bends near termini might make a nucleosome inherently less stable and easier to disrupt than a nucleosome that contains a bend near its center, or no intrinsic bend at all. This feature might aid the complete removal of the upstream nucleosomes that occurs during GALI -10 induction. Nucleosome D on GAL80, which is not completely removed during induction, contains no intrinsic bends. We have verified the presence of the only predicted (Goodsell and Dickerson) bend on the GAL80 upstream region. It is located at -500 bp (Fig. 17), which would place it in the GAL80 HR, downstream of nucleosome (Y and the UAS,,,,. As noted above, the UASGALso plays an important role in all GAL80 transcription, both basal and induced, and nucleosome OLmay communicate with downstream elements during induced GAL80 expression. The bend may help bring the UAS,,,,, and nucleosome (Y closer to the downstream promoter region or at least help orient them toward the 5’ end of GAL80. Another possible function of these intrinsic bends is suggested by the ob-

YEAST

CHROMATIN

AND

GAL

GENE

REGULATION

245

servation that in a double null mutant of NHPGA/B proteins, which are yeast homologs of HMGl/B, the induced expression of a reporter gene driven by the GALI-IO promoter region is decreased by -60-fold (157). HMG1/2 proteins bind preferentially to bent DNA (158) and thus might bind to any of the intrinsic DNA bends we have identified in the GAL1 -10 promoters, solidifying and perhaps even enhancing the curvature at these sites and compressing the linear distance between the UAS, and the targets of Gal4p (see below). In vitro, NHPGA promotes the formation of a complex between TBP and TFIIA at the TATA (157). If bound at the GALI-IO 645 bend, NHPGA would be in good position to do the same in viva, thus aiding transcription initiation in that way. NHPGA/B affect both UAS, and TATA function on GAL1 (157). b. Higher order Structure. On GAL1 -10, the closest UAS, are -110 bp (GALIO) and -220 bp (GALI) from the sites where DNA is first unwound for transcription. This corresponds to 40 and 75 nm, respectively, of linear DNA. Gal4p contains 88 1 residues, with DNA binding and transcription activation regions located at opposite ends of the protein. A spherical protein of this molecular weight would be 9 nm or so in diameter. Several features could help bridge these distances between the DNA sites of Gal4p binding and transcription initiation. 1. Gal4p might fold in an extended structure. Residues 51-64 form a coiled-coil that extends away at right angles from the DNA (78) and residues 64 - 9 7 also form a helical coiled-coil structure (159). Beyond this, the shape of Gal4p is unknown. 2. The involvement of large multiprotein complexes in activation (Section II&C) should also shorten the effective distance between bound Gal4p and its activation targets. 3. The three-dimensional organization of these upstream regions in the chromosome might also help shorten these distances and thus may play a role in the expression and regulation of GAL genes. A knowledge of this three-dimensional organization would be very useful, but its experimental determination is probably an impossible goal at this time. However, the GAL genes are prime candidates for an attempt to calculate such a structure, due to their relative simplicity and extensive structural knowledge base. To achieve this goal, we have entered into a collaboration with the lab of Shing Ho at Oregon State University (159u). The GALI-IO genes are particularly favorable for this calculation for several reasons: (1) the crystal structures of the protein-DNA complexes known to be present on the upstream region in the induced state, Gal4pUAS, and TBP-TATA, have been determined (78,160) and can thus be part

R. BASH AND D. LOHR

246

of the calculation. In situ footprinting of this region under induced conditions does not detect evidence for other bound proteins (118,122). Certainly there could be others, but the inability to detect them suggests they may not play major structural roles. (2) Bending analysis will have determined the intrinsic DNA structure of the intergenic region (see above). (3) Chromatin analysis indicates that nucleosomes are absent from the intergenic region under induced conditions. Thus, issues such as precise nucleosome locations and orientations, which would make such calculations exceedingly difficult, need not be addressed. We consider this project to be only a first step toward the long-term goal of determining the structure of a eukaryotic promoter region. Nevertheless, it is revealing to see what this first approach will yield. In these large-scale models, the conformations of the naked DNAs are relatively unremarkable, with the exceptions of the intrinsic bends described above. As expected from the crystal structure, TBP-induced distortions place a sharp kink at the TATA site. The model for the TBP-bound form of the GAL80 gene suggests a possible three-dimensional relationship between the Gal4p binding site and the RNA polymerase initiation site for this gene and provides a structural rationale for the alternate transcriptional start sites observed under induced and basal conditions of expression (107). Building a model for the induced GAL.1 -10 gene is more complex, requiring some assumptions as to how the DNA conformation at the UAS, sites is perturbed to fit the eight Gal4p binding there. The most interesting aspect of the resulting GAL1 -10 model is the topological coordination of the divergent GAL1 -10 promoters in the induced state (159u).

C. Regulatory

Factors,

Nucleosomes,

and GAL

Gene Activation 1. How DOES Gal4p ACTIVATETRANSCRIPTION? Precisely how an activator brings about transcriptional activation is a central question in eukaryotic regulation. Because Gal4p can function throughout eukaryotes, in microbes as well as in plants and animals (70), it must utilize common, conserved mechanisms and can therefore provide general insights on this critical question. Widespread functionality suggests that Gal4p targets a universal component(s), for example, a general transcription factor. Indeed, peptides from the Gal4p activation domain demonstrate an ability to bind specifically to TBP (161, 162) and TFIIB (162). Moreover, the binding affinity of various activation domain peptides for TBP correlates directly with their ability to activate transcription, indicating that the GallipTBP interaction has relevance for in viuo gene activation. Activation domain peptides also bind Gal80p as well as two other factors in a cellular extract, Sugl and Ada2 (161). Sugl, initially isolated as a suppressor of a Gal4 muta-

YEAST CHROMATIN

AND CAL GENE

REGULATION

247

tion, is a component of the 26s proteasome, a complex that carries out ATPdependent degradation of ubiquitin-protein conjugates (163). Ada2 is a component of a yeast complex that acetylates histones (7). Gal4p also promotes transcription in other ways. For example, it may recruit RNA polymerase II (164, 165). R ecruiting polymerase to active genes may be easier than generally imagined; experiments described earlier (Section 111,A) suggest that RNA pol II may be located rather near the GALZ promoter even under repressed conditions. Some features of its biochemical behavior suggest that Gal4p might be associated with the nuclear superstructure (72). This could be an aspect of its recruitment function, by preferentially bringing (or localizing) genes to nuclear sites of facilitated transcriptional activity (cf. 166). Disruption of TATA-bound nucleosomes is another Gal4p activation function. Recent evidence suggests that Gal4p uses both the Swi-Snf and the Gcn5-Ada histone acetyltransferase complexes during GALZ induction (167). Th ese complexes could function in upstream nucleosome disruption on GAL1 -10. The extremely high levels of induced structural gene expression indicate that Gal4p is a strong activator. At least in part, this strength could be due to the ability of Gal4p to recognize coactivators. For example, Galllp is a coactivator for expression of many genes including GAL genes (168). Without Galll, GAL structural gene transcription is decreased by severalfold (HZ). Galllp is a component of Mediator, a multiprotein complex now thought to mediate RNA pol II-activator interactions for many yeast genes and throughout eukaryotes (8, 112, 169). Deletion of other Mediator components, e.g., Med6 or Pgdl or Medl or Med2, also decreases GAL structural gene expression severalfold (112, Z69-Z7Z), indicating that this multiprotein complex is involved in Gal4p-mediated activation. Direct Gal4p contact has not been demonstrated for the above factors but an interaction between the Gal4p activation domain and Srb4, another Mediator component required for Gal4p-mediated transcription, has been suggested (172). Deletion of the Medl component of Mediator also partially relieves GAL1 repression in noninducing carbon sources (Z69), suggesting that this factor plays both a positive role, in induction, and a negative role, in maintaining gene inactivity, as do two RNA pol II-associated factors, SrblO and 11. Deletion of the RNA pol II-associated factor xtclp leads to increases in the induced levels of Gal4pdriven reporter gene expression, suggesting that this factor is an inhibitor of Gal4p-mediated activation (173). In w&-o, xtclp can bind to a Gal4p activation domain peptide. Clearly, a large number of factors affect the ability of Gal4p to activate transcription, both positively and negatively, and many of these may do so by directly interacting with the activator. The Gal4p activation function is rather tolerant of single amino acid changes. For example, only 4 of 42 Gal4p activation domain mutants detected in a random mutagenesis study were missense mutants (174). This tol-

248

R. BASH AND D. LOHR

erance may reflect some redundancy in activation mechanisms. Furthermore, cysteine and even proline substitutions introduced into a Gal4p activation domain peptide do not affect its transcription activation ability, measured in viva (175). This result was interpreted to indicate that the activation domain does not have to adopt any particular ordered structure in order to carry out its activation function. The cysteine or proline substitutions do diminish or abolish the interaction of this activation domain peptide with Gal80p. Apparently, Gal80p binding to this Gal4p region does require an ordered structure, which Gal80p probably induces when it binds to the region. Thus, the release of Gal8Op inhibition associated with Gal4p activation might be accompanied by a significant conformational change in Gal4p. 2. HISTONE PARTICIPATION IN GAL REGULATION Deleting the N-terminal tails of histone H4 results in an -20-fold decrease in the levels of GALI, 7, and 10 induced expression (176) but deletion of H3 tails produces a severalfold hyperinduction of those genes (177). Apparently, the H4 tails have a positive role in induction, but the H3 tails have an inhibitory role. The H3 and H4 deletions do not affect GAL4 expression, indicating that they exert their effects at the structural gene level. The roles of the H3 and H4 tails are quite specific. GAL1 remains hypoinduced, as in the H4 tailless strain, when an H3 tail replaces the endogenous tail on histone H4, or even if the H3 and H4 tails are exchanged, the H3 tail attached to the H4 central core region and the H4 tail attached to the H3 central region (178). Likewise, an H4 tail attached to the H3 core does not reduce the hyperinduction seen in an H3 tailless mutant. Deletion of H2A or H2B tails has no effect on GAL expression. Promoter swapping experiments indicate that the specific effects mediated by H3 and H4 tails function through different GAL promoter elements (179). The hyperinduction associated with H3 tail loss depends on the UAS,, and thus presumably Gal4p, because it is abolished when a PII05 UAS replaces the UAS,. However, hyperinduction still occurs if the region downstream of the GAL UAS,, including the TATA, is replaced by the analogous PHO5 downstream region. The H4 deletion effect shows the reverse dependence; it requires the GAL1 downstream region but not the UAS,. Because specific exchanges of the PHO5 and GAL1 TATA elements do not result in GAL1 hypoinduction, the H4 deletion effect appears to be linked to the nucleosome B region. The H3 and H4 tail deletions cause structural effects on the GAL1 upstream chromatin region that are consistent with these differing functional dependencies. For example, deletion of the H4 tails decreases the accessibility of DNA just upstream from the TATA, within nucleosome B, and was interpreted to cause a shift in nucleosome B position (130). The H4 tails may

YEAST CHROMATIN

AND GAL GENE REGULATION

249

help keep this DNA region accessible by controlling nucleosome positioning, perhaps to mediate TATA accessibility. Whether the intrinsic DNA bend in this region, and any effects associated with it (Section III,B,3), are involved in these features remains to be determined. Deletion of H3 tails does not affect GALI upstream nucleosome structure, consistent with their UAS,linked dependence. H3 and H4 tail deletions produce similar chromatin changes within the GAL1 coding region. These effects are probably linked to histone acetylation because mutation of the specific acetylation sites in the H3 and H4 tails achieves the same effects as complete tail removal (176, 177). Acetylation decreases the charge of the tails and causes some structural changes within the core nucleosome and in nucleosomal arrays (7), but it seems unlikely that H3 and H4 tail loss could produce opposing functional effects based on these structural changes. Acetylation also increases factor access to nucleosomal DNA in. vitro (7) and could even create sites that attract regulatory components-for example, nucleosome remodeling complexes or other factors (180). Thus, the H3 and H4 tails could be contact sites for nucleosome deposition and/or disruption machinery. For example, the loss of Gal80p function or the loss of H3 tails each result in roughly similar hyperinduction of GALl. Gal80p modulates the GALI-induced promoter (Section II,C), perhaps through its nucleosome deposition activity on the GALl-10 upstream region (Section III,B), and the H3 tails might be required to carry out this activity. The H4 tails could be important for disruption because they maintain nucleosome accessibility or because they contact disruption machinery or even factors such as NHPGA. At least for H4, histone acetylation levels on GAL1 are the same in repressed or induced chromatin (181). Thus, th e role of acetylation on GAL genes may be more subtle than a simple correlation between hyperacetylation and gene activation. For example, the opposing H3 and H4 effects discussed above may both require acetylated histone tails. 3. CHROMATIN

STRUCTURE

AND

GAL

GENE EXPRESSION

a. Competition for the TATA. Gal4p does not have to remove nucleosomes to access its DNA binding sites on GAL genes, but it does mediate nucleosome removal from the TATA and transcription start sites during gene activation, an activity that is a dedicated function of the activation domain (121). As a result of this disruption, the TATA become very accessible to exogenous probes and should therefore be accessible to TBP. Binding of TBP2 2 Whether the formation of the transcription preinitiation complex (PIC) follows a stepwise sequential assembly pathway from TBP or involves preformed holocomplexes containing TBP is not unambiguously clear. The arguments presented here should apply to either case equally well so we will simply use the term TBP to include whatever sort of core complex is involved.

250

R. BASH AND D. LOHR

to the TATA is a crucial step in transcription initiation on pol II genes (cf. 149, 182). Thus nucleosome removal to expose these elements is a key aspect in GAL gene activation. On GAL genes, occupation of the TATA/start site regions may be determined by a competition, mediated by the gene-specific regulators GalLip and Gal80p, between nucleosomes and TBP (Fig. 18). For example, occupation of these upstream regions by nucleosomes or TBP depends on regulator efficiency, as indicated by the following observations. The degree of nucleosome B disruption is proportional to the strength of the Gal4p activator (121). Gal4p activator strength also correlates directly with its ability to bind TBP (161, 162) and TBP mutants with a decreased ability to bind the TATA, and therefore a decreased ability to compete with nucleosomes for TATA-contaming DNA, decrease the ability of Gal4p to activate transcription (183). Gal80p modulation of the induced GAL1 -10 promoter, i.e., this promoter is not as active as it could be, is another suggestion of such a competition. Whether the TATA/start site regions are occupied by nucleosomes or by TBP will be thus determined by the competition between nucleosome disruption and nucleosome deposition. Activation conditions, activators, and metabolic energy favor disruption, allowing TBP to occupy the region (Fig. 18). Negative factors, inhibitory signals, or low cellular energy favor Ga180dependent nucleosome deposition and TBP displacement. The presence of Gal4p-Gal80p together in a complex could influence this competition, by regulating the stoichiometry of the opposing regulators. The large increase in Gal80p levels under induced conditions may impact this balance (see below). b. A Dynamic Znduced State? The nucleosome disruption/deposition reactions on GAL1 -10 occur rapidly and apparently reversibly. Therefore, the competition between nucleosomes or TBP to occupy these regions may be dynamic, with either occupant present so long as, and perhaps to the extent that, the signals corresponding to the associated functional state prevail. In particular, a transient disrupted (gene-active) state will ensure that this chromosomal structure exists only so long as the activation criteria are fulfilled. When any necessary signals cease, such as galactose presence or sufficient cellular energy, the structure will revert to the intact nucleosome (gene-inactive) structure. Both a transient nature and a requirement for ongoing signaling are features common to signal transduction pathways that activate other cellular responses. The presence of Gal4p and Gal80p in a complex should facilitate the rapid reversibility of these nucleosome reactions by keeping the opposing regulators in contact, a conformational change away from favoring the opposite reaction. Again, this is a model for upstream, not coding sequence, nucleosomes.

YEAST CHROMATIN

AND GAL GENE

REGULATION

I-L!

ITJ--

TATA

Box

FIG. 18. Regulator-mediated competition between nucleosomes and TBP for the TATA/ start site region. In the top panel, activating conditions favor TBP occupation of the TATA and disruption of the upstream nucleosomes. In the bottom panel, inactivating signals or a lack of required activation criteria favor nucleosome deposition on the TATA region and TBP absence. The activation domain of Gal4p is represented by the hatched arrow point (4). The disk represents a nucleosome. Heavy arrows denote favored processes under a given set of conditions and the double arrows are used to denote the reversibility of these nucleosome reactions. This model is developed from data obtained with GAL1 -10 but is presumed to apply to the other GAL, structural genes and, in perhaps a modified form, to GAL80 (see text).

From a chemical perspective, nucleosome occupation of upstream promoter regions can be viewed as a bidirectional equilibrium, between a nucleosome intact and a disrupted structure (Fig. 19). The two structures are

252

Fi. BASH AND D. LOHR lntrin lnstab Acetylation TBz:TA Nut Rem Coactivators Gal4p

.

. .

Intact , ~~~~]Ro\’ Disrupted t Hist/DNA Gal8Op

FIG. 19. Nucleosome transactions as a bidirectional reaction. The processes, disruption or deposition, producing the nucleosome-disrupted structure (right) or the nucleosome-intact structure (left) are indicated. Factors or processes thought to cooperate to favor a process are listed above or below that process.

usually presumed to reflect stable preferences associated with the induced and uninduced states of GAL expression. However, any chemical reaction is inherently dynamic and the rapid reversibility of this one is consistent with such a nature. Therefore, especially under induced conditions, a disruption deposition competition might be ongoing and continuous. If the energy difference between the two structural organizations is not large, individual promoters could flip back and forth between the two structures, as in any chemical equilibrium. Even the act of promoter clearance by RNA pol II could affect the competition, by generating negative supercoiling (184) in regions A and C, behind the departing polymerase; negative supercoiling favors nucleosome formation (deposition) (185). Whether any such contribution actually affects the structure depends on the energetic balance among all the contributions (Section 111,C). A dynamic model of gene regulation views the existence of a given promoter region in the nucleosome-disrupted state as only a transient preference in this bidirectional competition. Due to the cell cycle dependence of histone synthesis, new histones will not typically be available for deposition reactions. Rapid and reversible nucleosome transactions would therefore suggest that a histone acceptor(s) functions to keep the displaced histones near enough to the promoter region to allow rapid nucleosome redeposition. Possible acceptor candidates are Gal80p, which might accept displaced histones and bind them until conditions signal deposition; structural proteins such as nuclear lamins or matrix proteins, which can bind histones (186); or nucleosome-remodeling com-

YEAST CHROMATIN

AND GAL GENE

REGULATION

253

plexes (cf. 143). Remodeling complexes probably interact with histones in the course of destabilizing histone-DNA contacts and may in fact function by competitive binding to histones. For example, the abundant RSC complex can transfer a histone octamer from one DNA molecule to another (187) and should bind histones during the course of this reaction. c. Cooperating Factors and Processes. The precise balance in this bidirectional reaction (Fig. 19), i.e., which structure predominates on a given promoter, will be determined by the intrinsic energetics of the two structures and by other factors or processes that interact with them. Because the favored structural state in this reaction can be rapidly and reversibly shifted (see above), other factors or processes must play a major role in determining the favored structure. There are undoubtedly many factors or processes that make thermodynamic contributions to disruption or deposition. Individually they may be insufficient to trigger a nucleosome structural change, but by cooperating with one another can provide the energy to cause a transition. Contributors to disruption include the activities of the Gal4p activation domain; a favorable cellular energy state/high ATP availability; inherent nucleosome stability; acceptors for histones released by disruption; and the energy of TBP binding to a TATA exposed by disruption. The correlation of TBPTATA affinity with extent of activation (183) would seem to support the importance of this last contribution. The importance of inherent nucleosome stability has been directly demonstrated on PHO5, where replacement of an endogenous nucleosome by a hyperstable one inhibited the disruption reaction and gene activation (188). Intrinsic DNA bends located near the termini on GALI-IO upstream nucleosomes A and B may make these nucleosomes less stable and easier to disrupt. Gal4p itself is probably the target of a competition between positive and negative factors, which determine whether, and perhaps to what extent, the activities of the Gal4p activation domain can be exerted. For example, cross-linking studies indicate that GalSOp and TBP contact a common region in the Gal4p activation domain and binding of Gal80p to this region prevents TBP binding to it (175). Many other factors that might affect Gal4p activity, e.g., Gal3p and Mediator, were discussed above. The contributors to nucleosome deposition are less clear. Obviously, histone octamer-DNA affinity makes a contribution, as does GalSOp and any other factors that help Gal8Op carry out deposition. Gal80p could affect the bidirectional reaction directly, through its involvement in nucleosome deposition, or indirectly, by inhibiting Gal4p activation and disfavoring disruption. The direct role might account for the elevated GalSOp levels in induced cells; higher Gal80p levels could help to keep the deposition reaction competitive, through a mass action effect, with the disruption reaction that might

254

R. BASH AND D. LOHR

otherwise be too strongly favored in the induced state due to numerous positive contributions. Nucleosome transactions on GAL structural gene upstream regions can be viewed as a bidirectional process whose role is to control TATA accessibility and whose equilibrium position, i.e., whether disruption or the intact structure is favored, is determined by all the factors and processes involved in gene expression: specific regulators such as Gal4p and Gal8Op; TBP; histone tails; coactivators; negative factors; energy; acetylation; nucleosome-remodeling complexes; intrinsic nucleosome stability; phosphorylation; and ligand or protein binding. This competition need not work precisely the same way on every gene. For example, on GAL1 -10 the upstream nucleosomes seem to be completely removed, but on GAL80 there appears to be a more modest change, either partial histone loss or a conformational transition. Such differences may reflect intrinsically different mechanisms operating on the two genes or merely differing extents of progression along a single multistep pathway, due to differences in expression levels. The basic view that promoter region nucleosome structure can be transient and subject to rapid and reversible disruption/deposition reactions that result from the cooperation of multiple factors and processes is thermodynamically realistic and chemically sound and thus provides a very relevant view of this fascinating and complex area of eukaryotic molecular biology.

ACKNOWLEDGMENTS The NIH has supported the work in my lab. I thank Linda Russett for patient and dedicated manuscript preparation.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.

K. van Holde, “Chromatin.” Springer-Verlag, New York, 1988. M. Grunstein, Annu. Reu. Cell Biol. 6,643-678 (1990). M. Grunstein, Sci. Am. 267,68-74B (1992). G. Felsenfeld, Cell 86, 13-19 (1996). G. Hartzog and F. Winston, Cum. @in. Genet. Ileo. 7, 192-198 (1997). D. Lohr,J. BioE. Chem. 272,26795-26798 (1997). J. L. Workman and R. E. Kingston, Annu. Rm. Biochm. 67,545-79 (1998). S. Bjiirklund, G. Almoumi, G. Davidson, K. Nightingale, and K. Weiss, Cell 96,759-767 (1999). 9. D. Lohr and K. E. van Holde, Science 188,165-166(1975). 10.D. Lohr, J. Corden, R. T Kovacic, K. TatcheII, and K. E. van Holde, Proc. Natl. Acad. Sci. U.S.A. 74, 79-83 (1977). 11. S. G. Bavykin, S. I. Usachenko, A. I. Lishanskaya, V. V Shick, A. V Belyavsky, I. M. Un-

YEAST

CHROMATIN

AND GAL GENE

REGULATION

255

dritsov, A. A. Strokov, I. A. ZaIenskaya, and A. D. Mirzabekov, Nucleic Acids Res. 13, 3439-3459 (1985). 12. D. A. Nelson, W. R. Beltz, and R. L. Rill, Proc. N&Z. Acud. Sci. U.S.A. 74, 1343- 1347 (1977). 13. D. Lohr, R. T. Kovacic, and K. E. van Holde, Biochemistry 16,463-471 (1977). 14. D. Lohr and G. Ide, Nucleic Acids Res. 6,1909-1927 (1979). 15. J. 0. Thomas and V. Furber, FEBS Lett. 66,274-280 (1976). 16. D. Lohr, K. Tatchell, and K. E. van Holde, Cell R&829-836 (1977). 17. D. Lohr and K. E. van Holde, Proc. Natl. Acad. Sci. U.S.A. 76,4285-4288 (1979). 18. V L. Karpov, S. G. Bavykin, 0. V. Preobrazhenskaya, A. V. Belyavsky, and A. D. Mirzabekov, Nucleic Acids Res. 10,4321-4337 (1982). 19. F. Strauss and A. Prunell, EMBOJ. 2, 51-56 (1983). 20. D. Lohr,]. BioZ. Chem. 261,9904-9919 (1986). 21. R. T. Simpson, F. Thoma, and J. M. Brubaker, Cell 42,799-808 (1985). 22. F. Dong, J, C. Hansen, and K. E. van Holde, Proc. N&l. Acad. Sci. U.S.A. 87,5724-5728 (1990). 23. G. Meersseman, S. Pennings, and E. M. Bradbury,J. Mol. BioZ. 220,89-100 (1991). 24. R. T. Simpson, Prog. Nucleic Acid Res. Mol. BioZ. 40,143-183 (1991). 25. F. Thoma, Biochim. Biophy. Actu 1130, 1-19 (1992). 26. K. Luger, A. Mader, R. Richmond, D. Sargent, and T. Richmond, Nature &ondon) 389, 251-260 (1997). 27. R. J. Arceci and P. R. Gross, Deu. BioZ. 80, 186-209 (1980). 28. L. A. Burgoyne and J. D. Skinner, Biochem. Biophys. Res. Commun. 99,893-899 (1981). 29. C. L. F. Woodcock, L. L. Frado, and J. B. Rattner,J. Cell Biol. 99,42-52 (1984). 30. R. A. Horowitz, D. A. Agard, J. W. Sedat, and C. L. Woodcock,]. Cell BioZ. I25,1-10 (1994). 31. J. Bednar, R. Horowitz, S. Grigoryev, L. Carruthers, J. Hansen, A. Koster, and C. Wood(1998). cock, Proc. N&Z. Acad. Sci. U.S.A. 95, 14173-14178 32. V. M. Zayetz, S. G. Bavykin, V. L. Karpov, and A. D. Mirzabekov, Nucleic Acids Res. 9, 1053-1068 (1981). 33. J. Widom, Annu. Reo. Biqrhys. Biomol. Struct. 27,285-327 (1998). 34. A. Wolffe and H. Kurumizaka, Prog. Nucleic Acid Res. Mol. Biol. 6 1,3 79-42 1 (1998). 34a. G. Otero, J. Fellows, Y. Li, T. de Bizemont, A. Dirac, C. Gustafsson, H. Erdjument-Bromage, P Tempst, and J. Svejstrup, Mol. Cell 3, 109-118 (1999). 34b. S. Leuba, R. Bash, S. Lindsay, and D. Lohr, In preparation. 35. K. van Holde and J. Zlatanova, BioEssays 21,776-780 (1999). 36. R. ‘I Simpson, Biochemistry 17,5524-5531 (1978). 37. K. Polach and J. Widom,J. Mol. BioZ. 258,800-812 (1996). 38. G. Arents and E. N. Moudrianakis, Proc. N&Z. Acad. Sci. U.S.A. 90,10489-10493 (1993). 39. K. van Holde, J. Zlantanova, G. Arents, and E. Moudrianakis, “Chromatin Structure and Gene Expression” (S.C.R. Elgin, ed.), pp. l-21, Oxford University Press, Oxford, 1995. (1996). 40. T. M. Fletcher and J. C. Hansen, Crit. Reu. Eukayot. Gene Expr. 6,149-188 40~. K. Luger and T. J. Richmond, Cur-r. Opin. Gen. G Dec. 8,140-146 (1998). 41. M. Grunstein, Nature (London) 389,349-352 (1997). 42. K. Struhl, Genes Dew. l2,599-606 (1998). 43. F. Lenfant, R. Mann, B. Thomsen, X. Ling, and M. Gnmstein, EMBO]. 15,3974-3985 (1996). 44. M. M. Smith and 0. S. Andresson,J. Mol. BioZ. 169,663-690(1983). 45. J. W. WaIlis, L. Hereford, and M. Grunstein, Cell 22, 799-805 (1980). 46. J. Choe, D. Kolodrubetz, and M. Grunstein, Proc. N&l. Acad. Sci. U.S.A. 79, 1484-1487 (1982).

256

FL BASH

AND D. LOHR

47. K. P Lee, H. J, Baxter, J. G. Guilemette, H. G. Lawford, and P. N. Lewis, Can. _Z.Biochem. 60,379-388 (1982). 48. R. H. Morse, D. S. Pederson, A. Dean, and R. T Simpson, Nucleic Acids Res. 15,1031110330 (1987). 49. M. Piiiero, C. Puerta, and E. Palacian, Biochemistry 30,5805-5810 (1991). 50. D. Landsman, Trends Biochem. Sci. 21,287-288 (1996). 51. S. Ushinsky, H. Bussey, A. Ahmed, Y. Wang, J. Friesen, B. Williams, and R. Storms, Yeast 13,151-161 (1997). 52. D. Escher and W. Schaffner, Mol. Gen. Genet. 256,456-461 (1997). 53. A. Hecht, T. Laroche, S. Strahl-Bolsinger, S. M. Gasser, and M. Grunstein, Cell 80, 583 592 (1995). (1999). 54. J. Zlatanova, S. H. Leuba, and K. van Holde, Biophys. J. 74,2554-2566 55. J. B. Rattner, C. Saunders, J. R. Davies, and B. A. Hamkalo, 1. Cell. Biol. 92, 217-222 (1982). (1989). 56. P. T Lowary and J. Widom, Proc. Natl. Acad. Sci. U.S.A. 86,8266-8270 57. J. Bednar, R. A. Horowitz, J. Dubochet, and C. L. Woodcock,]. BioZ. Chem. 131,13651376 (1995). 58. J. S. Godde and J. Widom,J. Mol. BioZ. 226,1009-1025 (1992). 59. F. Thoma, T Keller, and A. Klug,J. Cell Biol. 83, 403-427 (1979). 60. H. Weintraub and M. Groudine, Science 193,848-856 (1976). 61. C. Bonifer, U. Jagle, and M. Huber,]. BioZ. Chem. 272,26075-26078 (1997). (1979). 62. D. Lohr and L. Hereford, Proc. NatZ. Acad. Sci. U.S.A. 76,4285-4288 63. T. Chen, M. Smith, S. Le, R. Stemglanz, and V. Allfrey, J. BioZ. Chem. 266, 6489-6498 (1991). 64. L. Hereford and M. Rosbash, Cell 10,453-462 (1977). 65. D. Lohr and T. Torchia, Biochemistry 27,3961-3965 (1988). 66. D. Lohr, Biochemistry 22,4527-4534 (1983). 67. D. Lohr, Biochemistry 22,927-935 (1983). 68. D. Lohr and G. Ide,]. BioZ. Chem. 258,4668-4671 (1983). 69. Y. Oshima, Regulatory circuits for gene expression: The metabolism off galactose and phosphate. In “The Molecular Biology of the Yeast Saccharomyces, Metabolism and Gene Expression” (J. Strathern, E. Jones and J. R. Broth, eds.), Vol. 1, pp. 159-180. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1982. 70. M. Johnston, Microbial. Rm. 51,458-476s (1987). 71. M. Johnston and M. Carlson, Regulation of carbon and phosphate utilization. In “The Molecular and Cellular Biology of the Yeast Sacchuromyces: Gene Expression” (E. Jones, J. Pringle, and J. Broach, eds.), Vol. 2, pp. 193-281. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1992. 72. D. Lohr, P. Venkov, and J. Zlatanova, FASEBJ. 9,777-787 (1995). 73. H. Ronne, Trends Genet. 11, 12-7 (1995). (1997). 74. R. J. Reece and A. Platt, BioEssays 19,1001-1010 75. E. Giniger, S. Varnum, and M. Ptashne, Cell 40, 767-774 (1985). 76. D. Lohr and J, Hopper, Nucleic Acids Res. 13, 8409-8423 (1985). 77. S. B. Selleck and J. M. Majors, Mol. Cell. BioZ. 7,3260-3267 (1987). 78. R. Marmorstein, M. Carey, M. Ptashne, and S. C. Harrison, Nature (London) 356, 408414 (1992). 79. J. Huibregtse, P. Good, G. Marczynski, J. Jaehning, and D. Engelke, J. BioZ. Chem. 268, 22219-22222 (1993). 80. T. Suzuki-Fujimoto, M. Fukuma, K. I. Yano, H. Sakurai, A. Vonika, S. A. Johnston, and T Fukasawa, MoZ. Cell. BioZ. 16,2504-2508 (1996).

YEAST CHROMATIN

81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123.

AND GAL GENE REGULATION

257

K. Yano and Fukasawa, T. Proc. Natl. Acad. Sci. U.S.A. 94,1721-1726 (1997). (1997). T. Blank, M. Woods, C. Lebo, P. Xin, and J. Hopper, Mol. Cell. Biol. 17,2566-2575 A. Platt and R. Reece, EMBOJ. 17,4086-4091 (1998). P. Bhat and J. Hopper, Mol. Cell. Biol. 12,2701-2702 (1992). F. Zenke, R. Engles, V Vollenbroich, J. Meyer, C. P. Hollenberg, and K. D. Breunig, Science 272,1662-1665 (1996). V. Vollenbroich, J, Meyer, R. Engels, G. Cardinali, R. Menezes, and C. Hollenberg, Mol. Gen. Genet. 261,495-507 (1999). M. S. Lamphier and M. Ptashne, Proc. Natl. Acad. Sci. U.S.A. 89,5922-592 (1992). M. Johnston, J. Flick, andT. Pexton, Mol. Cell. Biol. 14,3834-3841 (1994). R. Brent and M. Ptashne, Cell 43,729-736 (1985). L. Keegan, G. Gill, and M. Ptashne, Science 231699-704 (1986). D. W. Griggs and M. Johnston, Proc. Natl. Acad. Sci. U.S.A. 88,8597-8601 (1991). J. Corton, E. Moreno, and S. Johnston,J. Biol. Chem. 273,13776-13780 (1998). K. Leuther and S. Johnston, Science 256,1333-1335 (1992). L. Chen, Cum Opin. Struct. Biol. 9,48-55 (1999). A. Laughon and R. F. Gesteland, Proc. Natl. Acad. Sci. U.S.A. 79,6827-6831 (1982). E. Giniger and M. Ptashne, Proc. Natl. Acad. Sci. U.S.A. 85,382-86 (1988). T. Kang, T. Martins, and I. Sadowski,J. Biol. Chem. 268,9629-9635 (1993). H. Xu,T. Kodadek, andS. A. Johnston,Proc. Natl. Acad. Sci. U.S.A. 92,7677-7680 (1995). D. Lohr, Proc. Natl. Acad. Sci. U.S.A. 90,10628-10632 (1993). D. W. Griggs and M. Johnston, Mol. Cell. Biol. 13,4999-5009 (1993). L. M. Mylin, M. Johnston, and J. E. Hopper, Mol. Cell. Biol. 10,4623-4629 (1990). I. Sadowski, C. Costa, and R. Dhanawansa, Mol. Cell. Biol. 16,4879-4887 (1996). W. Ding and S. Johnston, Mol. Cell. Biol. 17,2538-2549 (1997). M. R. Parthun and J. A. Jaehning, Mol. Cell. Biol. I2,4981-4987 (1992). M. Hirst, M. Kobor, N. Kuriakose, J. Greenblatt, and I. Sadowski, Mol. Cell 3, 673-678 (1999). H. Shimada and T. Fukasawa, Gene 39,1-9 (1985). (1994). H. Sakurai, T. Ohishi, and T. Fukasawa, Mol. Cell. Biol. 14,6819-6828 T. E. Torchia, R. W. Hamilton, C. L. Cano, and J. E. Hopper, Mol. Cell. Biol. 4,1521-1527 (1984). W. Bajwa, T. E. Torchia, and J. E. Hopper, Mol. Cell Biol. 8,3439-3447 (1988). M. Igarashi, T. Segawa, Y. Nogi, Y. Suzuki, and ‘I? Fukasawa, Mol. Gen. Genet. 207,273279 (1987). Y. Suzuki, Y. Nogi, A. Abe, and T. Fukasawa, Mol. Cell. Biol. 8,4991-4999 (1988). L. Myers, C. Gustafsson, K. Hayashibara, P. Brown, and R. Kornberg, Proc. Natl. Acad Sci. (1999). U.S.A. 96,67-72 D. Lohr, In “Yeast Practical Methods” (Duffus and Campbell, eds.). IRL Press, Oxford (1988). S. Nedospasov and P. Georgiev, Biochem. Biophys. Res. Commun. 92, 532-539 (1980). C. Wu, Nature (London) 285, 854-860 (1980). J. Axelrod and J. Majors, Nucleic Acids Res. 17, 171-183 (1989). D. Lohr, Nucleic Acids Res. 12,8457-8474 (1984). D. Lohr, T. Torchia, and J. Hopper,J. BioZ. Chem. 262,15589-15597 (1987). G. Cavalli and F. Thoma, EMBOJ. I2,4603-4613 (1993). M. J. Fedor and R. D. Komberg, Mol. Cell. Biol. 9, 1721-1732 (1989). J. D. Axelrod, M. S. Reagan, and J. Majors, Genes Dev. 7, 857-869 (1993). M. Reagan and J. Majors, Mol. Gen. Genet. 259,142-149 (1998). D. Lohr, Nucleic Acids Res. lI,6753-6773 (1983).

258

R. BASH AND D. LOHR

124. R. Rainbow, J. Lopez, and D. Lohr, Biochemistry 287486-7490 (1989). 125. M. J. Fedor, N. F. Lue, and R. D. Komberg,J. Mol. BioZ. 204,109-127 (1988). 126. D. Chasman, N. Lue, A. Buchman, J. LaPointe, Y. Larch, and R. Komberg, Genes Dev. 4, 503-514 (1990). 127. G. Fragoso, S. John, M. S. Roberts, and G. L. Hager, Genes Dev. 9,1933-1947 (1995). 128. L. Cartwright, R. P. Hertzberg, P B. Dervan, and S. C. Elgin, Proc. N&Z. Acad. Sci. U.S.A. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154.

11,3213-3217 (1983). R. Komberg and L. Stryer, Nucleic Acids Res. 16,6677-6690 (1988). G. Fisher-Adams and M. Grunstein, EMBOJ. 14, 1468-1477 (1995). (1979). L. Lutter, Nucleic Acids Res. 6,41-56 G. Hager, C. Smith, J. Svaren, and W. H&z, “Chromatin Structure and Gene Expression” (S. C. R. Elgin, ed.), pp. 89-99. Oxford University Press, London, 1995. (1986). L. Cartwright and S. C. R. Elgin, Mol. Cell. BioZ. 6,779-791 M. Han and M. Grunstein, Cell 55, 1137-1145 (1988). Akhtar, G. Faye, and D. Bentley, EMBO]. 15,4654-4664 (1996). S. B. SeIIeck and J. Majors, Nature (London) 325,173-177 (1987). D. Lohr and J. Lopez,J. BioZ. C/rem. 270,27671-27678 (1995). C. Giardina and J. Lis, Science 261,759-761 (1993). C. Liang and W. Garrard, Mol. Cell. Biol. 17,2825-2834 (1997). S. SeIIeck and J. Majors, Proc. N&Z. Acud. Sci. U.S.A. 85,5399-5403 (1988). W. Beltz, RNA synthesis in isolated yeast nuclei. Ph.D. thesis, Florida State University, TaI lahassee, Florida (1977). Schmid, K. D. Fascher, and W. Hi&-z,CeZZ71,853-864 (1992). Cairns, Trends Biochem. Sci. 23,20-24 (1998). R. W. Baer and D. Rhodes, Nature (LowZot$3OI, 482-488 (1983). V St&it&y, D. Clark, and G. Felsenfeld, Cell 76,371-382 (1994). V. M. St&t&y, G. A. Kassavetis, E. P. Geiduschek, and G. Felsenfeld, Science 278,19601963 (1997). J. Perez-Martin, E Rojo, and V. de Lorenzo, Microbial. Rev. 58,268-290 (1994). J. Perez-Martin and V de Lorenzo, Annu. Rev. Microbial. 51,593-628 (1997). S. Burley and R. Roeder, Annu. Rev. Biochem. 65,769-799 (1996). P Hager-man, Biochim. Biophys. A& 1131,125-132 (1992). Bolshoy, P. McNamara, R. Harrington, and E. Tiifonov, Proc. Nutl. Acud. Sci. U.S.A. 88, 2312-2316 (1991). D. Goodsell and R. Dickerson, Nucleic Acids Res. 22,5497-5503 (1994). (1995). N. UIyanov and T. James, Methods Enzymol. 261,90-120 G. Schroth, J. Siino, C. Cooney, J. Th’ng, P. Ho, and E. Bradbury,J. Biol. Chem. 267,9958-

9964 (1992). 154a. H.-M. Wu and D. M. Crothers, Nature 308,509-513 (1984). 155. J. Thompson and A. Landy, Nucleic Acids Res. 16,9687-9705 (1988). 156. J. A. Parvin, R. J. McCormick, P. A. Sharp, and D. E. Fisher, Nature (London) 373,724727, (1995). 157. T PauII, M. Carey, and R. Johnson, Genes Dev. 10,2769-81(1996). 158. M. Bustin and R. Reeves, Prog. Nucleic Res. Acid Mol. BioZ. 54,35-100 (1996). 159. P. Gadhavi,J. Protein Chem. 17,591-598 (1998). 159a. J. Vargason D. Lohr, and P Shing Ho, In preparation. 160. Y. C. Kim, J. H. Geiger, S. Hahn, and P. B. Sigler, Nature (London) 365,512-520 (1993). 161. K. Melcher and S. Johnston, Mol. Cell. BioZ. 15,2839-2848 (1995). 162. Y. Wu, R. Reece, and M. Ptashne, EMBOJ. 15,3951-3963 (1996).

YEAST

CHROMATIN

AND

GAL GENE

REGULATION

259

163. D. Rubin, 0. Coux, I. Wefes, C. Hengartner, C. Young, A. Goldberg, and D. Finley, Nature (London) 379,655-657 (1996). 164. M. Ptashne and A. Gann, Nature (London) 386,569-577 (1997). 165. L. Gaudreau, A. Schmid, D. Blaschke, M. Ptashne, and W. Horn, Cell 89(l), 55-62 (1997). 166. P R. Cook, Science 284,1790-1795 (1999). 167. S. Biggar and G. Crabtree, EMBOJ. 18,2254-2264 (1999). 168. M. Nishizawa, Y. Suzuki, Y. Nogi, K. Matsumoto, and T Fukasawa, Proc. Natl. Acad. Sci. U.S.A. 87,5373-5377 (1990). 169. D. Balciunas, C. Galman, H. Ronne, and S. Bjorklund, Proc. Natl. Acad. Sci. U.S.A. 96, 376-381 (1999). 170. Y. C. Lee, S. Min, B. S. Gim, and Y. J. Kim, Mol. Cell. BioZ. 17, 4622-4632 (1997). 171. J. I. Piruat, S. Chavez, and A. Aguilera, Genetics 147,1585-1592 (1997). 172. S. S. Koh, A. Z. Ansari, M. Ptashne, and R. A. Young, Mol. Cell 1,895-904 (1998). 173. Emili, R. Kobayashi, and C. A. Ingles, Proc. N&Z. Acad. Sci. U.S.A. 95, 11122-11127 (1998). 174. M. Johnston and J. Dover, Genetics 120,63-74 (1988). 175. Ansari, R. Reece, and M. Ptashne, Proc. NatZ. Acad. Sci. U.S.A. 95,13543-13548 (1998). 176. L. K. Durrin, R. K. Mann, P S. Kayne, and M. Grunstein, Cell 65,1023-1031 (1991). (1992). 177. R. K. Mann and M. Grunstein, EMBOJ. 11,3297-3306 178. X. Ling, T. Harkness, M. Schultz, G. Fisher-Adams, and M. Grunstein, Genes Dev. 10(6), 686-99 (1996). 179. J. Wan, R. Mann, andM. Grunstein, Proc. NatZ. Acad. Sci. U.S.A. 92,5664-5668 (1995). 180. J. E. Brownell and C. D. Allis, Cum. Opin. Genet. Deu. 2,176-184 (1996). 181. S. Rundlett, A. Carmen, N. Suka, B. Turner, and M. Grunstein, Nature (London) 392,83 l835 (1998). 182. S. Burley, Curr. Opin. Stmc. BioZ. 6,69-75 (1996). 183. K. Arndt, S. Ricupero-Jovasse, and F. Winston, EMBO]. 14,1490-1497 (1995). 184. L. F. Liu and J. C. Wang, Proc. NatZ.Acad. Sci. U.S.A. 84, 7024-7027 (1987). 185. K. van Holde, D. Lohr, and C. Robert,J. BioZ. Chem. 267,2837-2840 (1992). 186. P. Traub and R. Shoeman, BioEssays 16,349-355 (1994). 187. Y. Larch, M. Zhang, and R. Komberg, CeZZ96,389-392 (1999). 188. Straka and W. Horn, EMBOJ. 10,361-368 (1991).

A Coordinated Interplay: Proteins with Multiple Functions in DNA Replication, DNA Repair, Cell Cycle/ Checkpoint Control, and Transcription MANUEL 100~

STUCKI,

STAGLJAR,

ZOPHONIAS ULRICH

0. JONSSON,

AND

H~~BSCHER~

Department of Veterinary Biochemistry University of Ziirich-lrchel CH-8057 Ziirich, Switzerland ,.............,,.,....... I. DNA Polymerases . A. DNA Polymerase cx ... . . B. DNAPolymerasesSande .................................... II. DNA Polymerase Accessory Proteins .............................. A. ReplicationFactorC ......................................... B. Proliferating Cell Nuclear Antigen ............................. III. Transcription Factors and Their Role in Activation of DNA Replication A. Introductory Remarks ........................................ B. Lessons from Viral Systems ................................... .......................................... C. LessonsfromYeast D. How Might Transcription Factors Help in Activating DNA Replication? ...................................................... IV Perspectives and Conclusions .................................... References ....................................................

In eukaryotic scription require

cells, DNA

transactions such as replication,

263 263 271 279 279 281 286 286 286 287 290 291 292

repair, and tran-

a large set of proteins. In all of these events, complexes of more

than 30 polypetides

appear to function in highly organized

and structurally well-

defmed machines. We have learned in the past few years that the three essential macromolecular

events, replication, repair, and transcription, have common fimc-

tional entities and are coordinated by complex regulatory mechanisms. This can he documented

for replication and repair, for replication and checkpoint control, and

for replication

and cell cycle control, as well as for replication

’ To whom correspondence Progress in Nncle~c Acid Research and Molecul;ur Riology, Vol. 65

and transcription.

should be addressed.

261

Copyright 0 2001 by Academic Press. All rights of reproduction in any form resewed. OOiY-6603Lll $35.00

262

MANUEL

STUCK1 ET AL.

In this review we cover the three different protein classes: DNA polymerases, DNA polymerase

accessory proteins, and selected transcription factors. The “common

enzyme-different

pathway strategy” is fascinating from several points of view: fmt,

it might guarantee that these events are coordinated; an evolutionary

second, it can he viewed from

angle; and third, this strategy might provide

mechanisms for essential physiological tasks.

6 zoooAcademic

cells with backup

press.

Maintenance of genetic stability is a key issue for any form of life. Consequently highly sophisticated mechanisms for maintenance of the genome were well established before the three kingdoms of life (Archaea, Eubacteria, and Eukaryotes) separated. DNA replication is the event leading to the duplication of DNA in advance of mitosis (or meiosis) and cell division. It occurs in vivo in an ordered and highly organized way, in which all enzymes and proteins involved have their exact roles in a replication complex called the replisome, which is located in so-called nuclear replication factories. Models have been proposed on how the enzymatic machinery might be spatially arranged at replication forks (I). The models were based on the idea that DNA polymerases (~01s) dimerize and that the DNA forms loops on the lagging strand such that the “directionality” for the pol is the same. If one postulates that the replisome is fixed to structures in the nuclear replication factories, it would thread the DNA through itself. The assembly and the events in a replisome might occur as follows (2): 1. An initiator protein complex, the origin recognition complex (ORC), is bound to an origin of replication. The ORC has to be activated by other proteins, such as minichromosomal maintenance (MCM) and cell division cycle (Cdc6 and Cd7/Dbf4) proteins, by mechanisms such as phosphorylation or possibly other posttranslational modifications. This leads to the formation of an initiation complex that is able to alter DNA structures in its vicinity, presumably by activating the intrinsic helicase activity of an MCM hexamer or by attracting other DNA helicases to the origins. The single-stranded DNA thus produced must be protected and stabilized by the single-stranded DNA binding protein, called replication protein A (RP-A); RP-A can help to unwind the DNA by its unwinding activity and, possibly, through its interactions with DNA helicases and pol (Yprimase (pal (Y),which acts as the initiating pol. 2. After very limited DNA synthesis a DNA polymerase switch from pol (Y to the processive ~016 holoenzyme occurs, most likely mediated by the pol auxiliary protein replication factor C (RF-C). The pol (Ycomplex subsequently acts at the discontinuously synthesized lagging strand, where

PROTEINSWITH FUNCTIONSIN DNAPROCESSES

263

it initiates Okazaki fragments. While the pol S holoenzyme [pol 6, proliferating cell nuclear antigen (PCNA), and RF-C] is engaged in processive leading-strand DNA synthesis, the situation at the lagging strand is more complex. A second pol (pal S or E) holoenzyme is formed for completion of each Okazaki fragment initiated by pol (Y.DNA syntheses on the leading and on the lagging strand are perhaps coordinated by dimerization of the two processive pol holoenzymes, possibly on physical interaction of two pols or via a clamp factor. 3. The initiator RNA at the lagging strand is removed by RNase H and by a 5’ -+ 3’ exonuclease such as flap endonuclease (Fen 1) or by the Dna2 endonuclease. After complete synthesis the Okazaki fragments are sealed by DNA ligase I (Lig I). Topological constraints are released by DNA topoisomerase I and the replicated DNA can finally be separated by DNA topoisomerase II. Duplication of the genetic information is not the only DNA transaction. Integrity of the genome in nondividing cells is maintained by various DNA repair mechanisms (3), including nucleotide excision repair (NER) (4), base excision repair (BER) (5), mismatch repair (MMR) (6), and double-strand break repair (DSBR) (7). Th ese mechanisms, together with control processes such as checkpoint control (8), guarantee that either newly replicated DNA goes without mutation into mitosis or that nondividing cells maintain the DNA at a level of mutations that guarantees proper function. Most of the various DNA repair processes and DNA replication might not function in an “authistic” way, but rather are connected to each other. Controlled transcription is the event that fullfils the requirements for a timely and coordiated expression of the genetic information to yield proteins with specific functions. Transcriptional events have also been found to be of importance for initiation of DNA replication. In the following three sections we focus on multiple functions of the three pols, (Y,6, and E; the two pol auxiliary proteins, RF-C and PCNA; and on the four transcription factors, Ga14, ~53, BRCAl, and ABFl, which can function in DNA replication via their activation domains (see Table I).

I. DNA Polymerases A.

DNA

Polymer-use 01

1. GENERAL DESCRIPTIONOF THE ENZYME In 1957 the first eukaryotic polymerase, pol (Y,was discovered (9). Pol (Y still holds a special position in the growing family of eukaryotic pols, because it is the only enzyme that can start DNA synthesis de nowo. It first synthesizes

TABLE

I

break

see text.

Double-strand

“PO1a/plimase complex.

“DSBR,

aFor references,

Gal4

repair;NER, nucleotide

+ +

+ + + +

P53 BRCAI ABFl

excision repair; BER, base excision

+ + -

-

repair;MMR,

mismatch

repair.

+ + + +

Apoptosis Recombination _

140.kDa subunit; binds to telomerase _

_

-

3 7-kDa subunit

-

NER, BER, MMR, DSBR NER, BER, MMR, DSBR

PCNA RF-C

_

_

C terminus of large subunit

NER, BER, DSBR

PO1E

Other functions Telomer length (2) -

Primase ?

DSBR NER, BER, MMR, DSBR

Initiating pol Leading-s&and pal; lagging-strand pol? Leading strand pol ?; lagging-strand pol + +

Pol a” PO16

Transcription _ _

DNA repa?

DNA replication

Protein

Checkpoint control

Functior8’

PROTEINSWITHMULTIPLEFUNCTIONS”

PROTEINS

WITH

FUNCTIONS

IN DNAPROCESSES

265

an RNA primer of about 10 nucleotides followed by extension of the primer to produce a short piece of DNA. Therefore, the major task of pol (Yis thought to be the initiation of DNA synthesis during replication. The human enzyme is composed of four individual subunits with molecular masses of 180, 70, 58, and 48 kDa, and a similar subunit composition was found for all eukaryotes examined (IO). The DNA polymerizing activity is located on the largest (~180) subunit whereas the p48 subunit contains the catalytic center of the primase function. Although isolated ~48, devoid of any detectable p58 subunit, is sufficient for RNA primer synthesis, primase activity associated with free ~48 is highly unstable, indicating that its association with the other polypeptides of the pol LXcomplex is crucial for stabilizing the enzyme activity (11). The 180-kDa subunit exists as a tight complex with the p70 subunit and it was found that this interaction is essential for both the synthesis of the ~180 protein and its translocation into the nucleus (12). No catalytic activity was found in p70 subunit per se, but it is thought to play an important role in modulating the activity of the whole complex as well as in controlling its ability to associate with the chromatin at the origins of replication. This guarantees a tightly controlled and coordinated initiation of DNA synthesis on both the leading and the lagging strand. These effects are most probably mediated through cell cycle-specific phosphorylation of pol (Y and/or interaction with other protein partners that are involved in DNA metabolism (summarized in Table II). 2. ROLE OF POL (YIN VIRAL DNA REPLICATION Biochemical studies based on plasmids containing the simian virus 40 (SV40) origin of replication have uncovered the fundamental mechanisms leading to the recognition of the origin sequence, the local unwinding of the DNA, and the loading of the pol (Ycomplex onto the DNA. Only two proteins, the virus-encoded SV40 large T antigen (TAg) (13) and RP-A, are required. They interact with each other and with pol (Y,tether the pol to the DNA, and cooperate to initiate DNA synthesis at the SV40 origin. (13, 14). Data obtained using the human papillomavirus (HPV) replication system suggest a similar mechanism of initiation. The HPV El helicase together with the E2 protein bind cooperatively to the viral origin of replication, forming an ElE2-ori complex similar to the SV40 TAg-ori complex. The El protein interacts with RP-A and with the ~180 and p70 subunits of pol cx and might bring pol cx to the DNA in a way similar to that of TAg (15-17). As soon as pol (Yis associated with the initiation complex, it synthesizes a short RNA-iDNA (i stands for initiator) fragment, approximately 10 nucleotides of RNA followed by 30 nucleotides of DNA, which then serves as a primer for extension by another pol(18 -21). This process (called polymerase

II

Viral and cellular DNA replication; primosome assembly ? Lagging-strand synthesis? Lagging-strand synthesis? ? Okazaki fragment processing Cell cycle regulation Cell cycle regulation Cell cycle regulation Tumor suppressor, DNA replication; proofreading? Tumor suppressor Tumor suppressor DNA damage survey?

Human Calf Mouse S. s. S. s. s. cerevisiae; xenopus Human Human s. cer-evisiae Human Human Human; mouse Human

pi; fi

pi; fi

pi; @

pi pi pi

pi; fi

fi fi fi

pi pi pi

pi

RP-A

AAF

Cdc68p/SptlGp PoblpKtf4p Pob3p Dna2p

Cdc45piXCdc45p

Cyclin A-Cdk2p Cyclin E-Cdk2p Cdc28p-Clb

P53 Rb Doe-1

PARP

aFor references, see text. bAbbreviations:Ag, Antigen; AAF, alpha accessoryfactor; Rb, retinoblastomaprotein; PARP, poly(ADP)ribose polymerax “pi, Physical interaction;fi, functional interaction; gi, genetic interaction.

Primosome assembly

Viral DNA replication; primosome assembly Viral DNA replication; primosome assembly

Simian virus 40 Papillomatis

pi; fi pi; fi

SV40 large T Ag El

cerevisiae cerevisiae cerevisiae cerevisiae

Presumed functional task with pol u

Species

U/fiIMASEa

Type of interactionC

WJTH POL

TABLE hTERACTINC

Protei&

PROTEINS

PROTEINSWITHFUNCTIONSINDNAPROCESSES

267

switch) is not yet fully understood. Although SV40 or&dependent replication in vitro with highly purified enzymes showed that pol cx is only moderately processive (22), it is unlikely that the limited size of the RNA-iDNA primer synthesized by pol (Yin vivo is the result of a spontaneous dissociation of the enzyme from the nascent DNA due to its lack of processivity. Indirect evidence suggests that the clamp loader RF-C plays an important role for the polymerase switch (21, 23) and additional results have confirmed an active role of RF-C in this process (see Section II). On the leading strand, pol (Yhas to initiate DNA synthesis only once per round of viral replication, whereas on the lagging strand, it has to reinitiate DNA syntheis for every Okazaki fragment. This poses a problem, because pol (Ydoes not have a proofreading activity and thus is unable to remove any errors that it inserted into the iDNA on the lagging strand. Although these mismatches might cause only minor implications for viral replication, they must be repaired on the over 20 million Okazaki fragments initiated during one round of cellular DNA replication (in human cells). One interesting and yet unsolved question is why pol o. contains a DNA pol fuction at all, because the other replicative pols, S and E, are able to elongate RNA primers in vitro (24). However, mutation analysis in yeast revealed that an operative pol fuction is essential (25). Ob viously, the cell is somehow able to deal with this problem. 3. POL 01MIGHT BE PARTOF A MULTIPROTEINCOMPLEX FORCOORDINATEDANDERROR-FREE LAGGING-STRANDREPLICATION

It has been proposed that the synthesis of an RNA-iDNA primer to start an Okazaki fragment and maturation of the previous Okazaki fragment are coordinately carried out during lagging-strand synthesis by a multiprotein complex, and that this complex might contain all the required enzymatic functions to remove RNA-iDNA primers that contain mismatches (14, 26, 27). Studies now support this hypothesis: mutations in the essential DNA2 nuclease-helicase gene from the yeast Saccharomyces cerevisiae were found to interact genetically with POLl and CTF4/POBl, which encode the large subunit of yeast pol cx and a protein involved in DNA metabolism in vivo (28). DNA2 has previously been shown to interact genetically and physically with RAD27, the yeast homolog of the human structure specific nuclease Fen 1 (29) that is involved in Okazaki fragment maturation. It has been proposed that Fen 1 might be able to remove the mismatch containing iDNAs endonucleolytically (2 7). H owever, Fen 1 does not act here as a single player because the gene is not essential in yeast. There is growing evidence that the task of quick and error-free synthesis as well as maturation of the Okazaki

268

MANUELSTUCKIETAL.

fragments are carried out by a complex multiprotein machinery containing at least pol (Y, pol 6, PCNA, RF-C, RP-A, Dna2, Fen 1, and Lig I (K.-S. Seo, personal communication). This suggests that the iDNA could be replaced by rereplication of the resulting gap by one of the proofreading pols, 6 or E. Such multiprotein complexes have been isolated from S. cerevisiae (30) and calf thymus (23). 4. ROLEOFPOL~YINTHECELLCYCLECONTROL OFDNAREPLICATION Most of the mechanisms of replication fork progression have been studied with viral in vitro replication systems such as SV40. Therefore, much less is known about cellular DNA replication. Nevertheless, in recent years, genetic studies using the yeasts S. cerevisiae and Schizosaccharomyces pombe, as well as in vitro and in vivo protein-protein interaction studies, have provided exciting insights into the complex network of biochemical signaling pathways that control chromatin replication and couple it tightly to the S phase during the cell cycle (31). The unique ability of pol (Y to initiate DNA synthesis makes it a perfect downstream target for cell cycle control pathways. It has been observed that polo is phosphorylated in a cell cycle-specific manner in yeast (32) and human (33) cells. Most of the phosphorylation occurs late in the cell cycle, suggesting that it does not play a role for initiation control. However, phosphorylation of both the ~180 subunit and the p70 subunit could also be achieved in vitro, using cyclin-dependent kinases (cdks). This phosphorylation did not influence the pol activity on primed templates, but it slightly stimulated primase activity, and specific initiation of replication on SV40 ori-containing plasmids was affected. Cyclin A/Cdk2 inhibited the ability of pol (Y to initiate SV40 DNA replication, whereas phosphorylation by cyclin E/Cdk2 stimulated its initiation activity (34). Tiyptic phosphopeptide mapping of the in vitro phosphorylated p70 subunit and p70 from human cells that were synchronized and labeled in G,S and in G, showed similar patterns: a cyclin E/ cdk2-like pattern in G,/S and a cyclin A/cdk2-like pattern in G, (35). This suggests that replication activity of pol (Yis regulated on phosphorylation by cdks in vivo. Another important regulatory pathway that is mediated through cdks limits DNA replication to once per cell cycle because these kinases can also prevent rereplication. Assembly of prereplicative complexes at origins can occur only during G,, because assembly is blocked by cdkl together with B cyclins (Clbs). Desdouets et ~2. describe the recruitment of pol (Yto chromatin as a second separate process that can occur only during G,, because recruitment is also blocked by cdkl/Clb, indicating that this might be an additional layer of control (36).

PROTEINS

WITH

FUNCTIONS

IN DNA

PROCESSES

269

Several proteins have been found to interact either physically or genetically with pol OL(see Table II). Among these, some are thought to modulate pol (Y activity in a cell cycle-specific manner or support loading of the complex onto chromatin. Almost 10 years ago a protein isolated and purified from mouse cell extracts was called alpha accessory factor (AAF), because it was able to stimulate pol (Y activity severalfold on synthetic templates (37, 38). However, its physiological function remains unclear. CDC68SPT16, a gene from S. cerevisiae, is required for passage through the cell cycle control point START, and Cdc68p was shown to bind to pol (Yby affinity chromatography (39). The same method has led to the identification of two other proteins, PoblpKtf4p (40) and Pob3p, a protein with significant amino acid similarity to a HMGl-like protein from vertebrates. Cdc68p and Pob3p seem to compete with Ctf4p for binding to pol CLCDC68 also interacts genetically with POLl and CTF4, indicating that the proteins are involved in the same biochemical pathway (39). Interestingly, a CDC68 mutation was also synthetic lethal with a DNA2 mutation, suggesting that the proteins may rather be part of a complex that is responsible for coordinating lagging-strand DNA syn thesis (see above), rather than supporting recruitment of pol o_to the origins. Moreover, another protein, called Cdc45p (XCdc45), has been identified in budding yeast and Xenopus egg extracts. It is essential for the initiation of replication and has been shown to interact physically with pol CLThere is growing evidence that this protein plays a pivotal role in the loading of pol (Y onto chromatin under the control of S phase cdks (41). Finally, three human tumor suppressor proteins have been shown to interact physically with pol ct as well: p53 (42), DOC-1 (43), and the human retinoblastoma protein Rb (44). The interaction between pol OLand p53 is of particular interest, because p53 is believed to have an intrinsic 3’ + 5’ exonuclease activity and colocalizes with DNA synthesis in uiuo. It has been suggested that p53 might increase pol (Yfidelity during replication (45, 46). 5. POLCIASAPOTENTIALPART

OF

CELLCYCLE

CHECKPOINTPATHWAYS Cell cycle checkpoints are regulatory pathways that control the order and timing of cell cycle transitions and ensure that critical events such as DNA replication and chromosome segregation are completed with high fidelity. In addition, checkpoints respond to DNA damage by either arresting the cell cycle or slowing down S phase to provide time for repair and by inducing transcription of genes that facilitate repair (8, 47). In S. cerevisiae, two essential genes from the central conduit for checkpoint signal transduction were identified: MECI (related to the ATM gene that is defective in the human disease ataxia telangiectasia) and RAD53; both encode protein kinases.

270

MANUEL

STUCK1 ET AL.

The first evidence that pol OLcould be involved in one or several checkpoint pathways came from genetic studies with the yeast S. pombe. The gene cd.cl +, the homolog of RAD53, has been identified as a multicopy suppressor of a temperature-sensitive mutant in the pol (Y gene (48). Another group showed that germinating S. pombe spores, disrupted for the POLl gene, could enter mitosis despite defects in DNA synthesis (49). Possibly pol 01 is required for a checkpoint signal that is activated as cells traverse START, and is essential to prevent mitosis until S phase has been completed. Moreover, genetic studies with S. cerevisiae revealed that the primase function of pol (Y might be a final target for the S phase checkpoint. A temperature-sensitive mutation in the primase gene was defective in DNA synthesis at the permissive temperature in the absence of DNA-damaging agents, whereas the same mutant, in the presence of the DNA-damaging agents proceeded faster through S phase than did wild type (50). An explanation for this apparent paradox could be that this mutant somehow fails to respond properly to a regulatory mechanism that is required to inhibit G, + S transition and S phase progression in the presence of DNA damage. The primase may be inhibited on DNA damage in order to prevent reinitiation of DNA synthesis downstream to the lesion, which then could lead to subsequent slowing down of S phase progression. Interestingly, mutations in the MECl gene were shown to result in a similar phenotype (51), indicating that yeast pol cx could indeed be part of the Meclp checkpoint pathway. Finally, the yeast protein Cdc45p, which is believed to have a key regulatory function for association of pols onto the origins (see above), is downregulated on induction of the S phase checkpoint, revealing evidence that the loading of pol OLonto chromatin is an additional target for checkpoint regulation (52). 6. Is POL a INVOLVEDIN DNA REPAIR? Bulky DNA lesions such as cyclobutane pyrimidine dimers or 6-4 photoproducts are efficiently repaired by nuclear extracts from Xenopus oocytes. This repair process seems to be dependent on pol CY,because neutralization of the enzyme with a specific antibody resulted in reduced repair effeciecy that could be restored by the addition of purified pol (Y(53, 54). However, a direct role of pol (Y in one of the excision repair pathways has not yet been reported. A novel model for double-strand break (DSB) repair has been proposed (55). Holmes and Haber analyzed mitotic DSB-induced gene conversion at the MAT locus in mutant S. cerevisiae strains that were temperature sensitive for essential replication factors. Surprisingly, a pol (Y mutant turned out to be greatly defective for completion of gene conversion at the MAT locus, indicating that lagging-strand synthesis is required for this process.

PROTEINS WITH FUNCTIONSIN DNA PROCESSES

B. DNA

Polymerases

271

6 and E

1. GENERAL DESCRIPTION OF THE ENZYMES Pol6 and pol E share two common features: they contain intrinsic 3’ -+ 5’ exonuclease functions that contribute a useful proofreading activity and they are responsive to the accessory proteins PCNA and RF-C (see also Section II). Pol S is thought to be the major replicative and repair pol(56, 57). Its activities have been characterized extensively by in vitro studies using the SV40 replication system and broad genetic and biochemical studies were carried out with the yeasts S. cerevisiae and S. pombe. The subunit composition of pol 6 is complex and remains ambiguous. Most likely, the S. cerevi.siae enzyme is composed of three individual subunits with apparent molecular masses of 125, 58, and 55 kDa, encoded by the genes POL3 (or CDC2), PO1531 (or HYSB), and POL32, respectively, which may form a dimer or a heterotrimer (58). However, an active heterodimeric form composed of Pol3p and Po13lp that showed a biochemical behavior somewhat different than that of the trimer could be isolated as well (59). Furthermore the POL32 gene is not essential for viability in budding yeast. In contrast, ~016 from S. pombe seems to be composed of at least four, and perhaps even five, subunits that migrate on sodium dodecyl sulfate (SDS) gels; the subunits have molecular masses of 125 (PO/~+), 55 (C&l+), 54 (C&27+), and 42 and 22 kDa (Cdml’) (60). The mammalian enzyme has been isolated and characterized from various sources. Highly purified preparations were described as a two-subunit enzyme, with a 125-kDa subunit (~125) and a subunit from calf thymus (24, 61) and mouse cell extracts (62) that was 48-50 kDa (~50). The genes of these two subunits have been isolated and cloned from mouse, calf, and human (56) cells; although the gene of the largest ~016 subunit shows high homology throughout the species, the genes for the small subunits are much less conserved. The ~125 subunit carries the catalytic centers of both pol and exonuclease functions whereas the functional tasks of the small subunit(s) are not understood. Mouse ~018 can be purified in two forms: the single catalytic subunit that is not responsive to PCNA and the two-subunit enzyme that is stimulated by PCNA (62). Similar observations were made with highly purified pol 6 fractions from calf thymus and HeLa cells (M. Stucki and U. Hiibscher, unpublished results) and for recombinant ~125 from either Es&et-i&a coli (63) or from recombinant baculovirus-infected insect cells (64). A study using PCNA affinity chromatography as an alternative approach to search for additional subunits provided more evidence that mammalian ~018 is indeed composed of more than two subunits. A new polypeptide from a mouse cell line extract had an apparent molecular mass of 66 kDa and

272

MANUELSTUCKIETAL.

coeluted from a PCNA affinity column with ~125 and ~50. Sequence characterization and a sequence data base search led to the identification of an open reading frame, and the carboxy-terminal part of the corresponding clone showed significant homology to the carboxy terminus of C&27+ from S. pombe, although overall homology between these two sequences is low (65). Assessing all of these data, it seems that mammalian pol 6 can be isolated in at least three forms: as a monomer (p125), as a dimer (~125 together with p50), and as a multisubunit complex that contains at least three, perhaps more, polypeptides. However, the question of a physiological significance of these subforms has not yet been addressed and therefore an exact characterization of the components of the mammalian ~016 complex and their biochemical roles is still elusive. Pol E is the second PCNA-responsive pol. Four to five subunits have been identified for S. cweuisiae pol E. The genes POL2, encoding the large 256kDa catalytic subunit, and DPB2, encoding one of the small subunits (80 kDa), are essential (66, 67), wh ereas the genes DPB3, encoding one or two subunits of 34 and 3 1 kDa, respectively, and DPB4, which encodes the fourth 29-kDa subunit, are not essential (68). The S. pombe gene Cdc20+ encodes the catalytic pol E subunit of fission yeast (70). In human cells the enzyme consists of two polypeptides of 2 15 and 55 kDa (71), whereas it was purified from calf thymus as a complex of 140, 125, 48, and 40 kDa (72). The 140kDa polypeptide is most likely a degradation product of a large precursor, because the enzyme has a high tendency to degrade during purification procedure due to a protease-sensitive site within the large catalytic subunit. However, the degraded 140-kDa form contains all the catalytic fuctions required for DNA synthesis and proofreading activity (73). Pol E genes from some mammalian cells have been cloned and characterized (74, 75). 2. THEPOL~ANDPOLEHOLOENZYMES The complex of ~016 or pol E with PCNA and RF-C is referred to as the holoenzyme form of the corresponding enzyme (76- 78). Assembly of the holoenzyme at a template-primer junction and pol switching are most probably orchestrated by RF-C (see Section II), which might leave the complex after assembly of a processive pol-PCNA clamp (79). The exact architecture of these processive pol machines is complex and has not yet been resolved to molecular details. The pol activities of pol6 and pol E appear to depend strongly on the geometry of the template (80) and on the assay conditions. On a linear primed homopolymer poly(dA)-oligo(dT)] pol E is fully processive in the absence of PCNA, whereas on the same template pol S is processive only in the presence of PCNA and under low-pH conditions. In this case, RF-C is not required because PCNA is able to slide

PROTEINS

WITH

FUNCTIONS

IN DNA PROCESSES

273

on the linear DNA spontaneously. On primed circular templates, however, both pols form stable processive holoenzyme complexes only in the presence of RF-C, PCNA, ATP, and a single-strand binding protein such as RP-A or E. cc& SSB (81,82). Th e same requirement for accessory factors can be achieved on linear DNA in the presence of ATP and under physiological pH conditions (83). One important difference between ~016 and pol E holoenzymes was reported for gap-filling synthesis (84). On double-stranded DNA circles containing a defined gap, pol 6 and pol E in the presence of PCNA and RF-C could fill the gap, and pol S, but not pol E, was able to synthesize up to 150 nucleotides into the double-stranded region (84, 85). This observation, called “limited strand displacement activity,” could have strong implications for the pol usage either for Okazaki fragement processing or for gap-filling synthesis during repair such as NER (86, 87) or BER (88). The mechanism of how PCNA stabilizes the pol-template complex, thus transforming the distributive enzyme into a processive form, is not yet clear. Several studies report a direct physical interaction between the N terminus of mammalian ~016 ~125 to the so-called domain-connecting loop of PCNA (89- 92). However, this interaction does not stimulate, or only slightly stimulates, processivity of the ~125 subunit alone. The ~50 subunit seems to be required for full stimulation (64), although this protein most probably does not interact directly with PCNA (92). S accharomyces cerevisiae pol &PCNA interaction has been confined to the Po132p subunit using both the yeast twohybrid system and a PCNA overlay method (58), but deletion of the gene for this subunit is not a lethal mutation (see above). Moreover, the question of the relative positions of components of the mammalian pol S/PCNA complex has been addressed: it seems that PCNA is located “behind” ~016 in regard to the DNA synthesis (3’ -+ 5’) direction and that only the large catalytic subunit contacts the DNA (93). Taken together, these data suggest that at least one (perhaps more) of the pol 6 subunits directly contacts PCNA and that this interaction stabilizes the enzyme on the DNA and might also induce a conformational change of ~125 and p50 relative to each other, which could be crucial for the transformation of the catalytic subunit into a processive form. The biochemical impact of PCNA on ~016 is different from that on pol E. First of all, pol E is processive on primed linear homopolymer DNA in the absence of PCNA. However, under conditions of high ionic strength, pol E becomes highly dependent on PCNA (94). Exact kinetic analysis revealed that PCNA is able to increase both the affinity for the primer and the rate of nucleotide incorporation by pol E, thus stimulating its activity (95). By using four characterized mutants of PCNA (91) containing three to four alanine

274

MANUEL

STUCKI ET AL.

residue substitutions, either on the C-terminal side (which is located on the “front side” in regard to the DNA synthesis direction) or on the back side of the trimer, the kinetics of primer binding and nucleotide incorporation by pol E were tested on linear DNA (96). In contrast with what had been found in interaction studies between pol S and PCNA, these data suggested that stimulation of pol E primer binding involves interactions with both the C-terminal side (frontside) and the backside of PCNA, whereas for stimulation of pol E DNA synthesis, exclusively the C-terminal side appears to be sufficient (96). 3.

POL~ANDPOLE IN VIRALANDCELLULAR DNAREPLICATION

SV40 DNA replication has been completely reconstituted with highly purified enzymes (14), and it has been shown that ~016 is required and sufficient for primer elongation on both the leading and lagging strands during SV40 replication in vitro (97). There is increasing evidence that pol S (like pol a) is part of a multiprotein complex that is responsible for a coordinated lagging-strand synthesis. First, pol 6 copurified with RF-C and pol CYin a replication-competent complex that also contained minor amounts of Fen 1 and Lig I (23; M. Stucki and U. Hiibscher, unpublished results). Second, pol S has been shown to interact genetically with the Fen 1 homolog RAD27 in budding yeast (98). Because ~016 is able to form a dimer in budding yeast via the Po132p subunit, it is interesting to speculate that a ~016 dimer that contains two catalytic subunits might be involved in coordinated leading- and lagging-strand synthesis during eukaryotic replication fork progression. Several genetic and biochemical studies suggest pol E to be the third replicative pol(66, 84, 99, IOO), whereas no essential role for human pol E in the in vitro SV40 DNA replication system was detected (94). Additional data provide strong evidence that this pol is not involved in SV40 replication (101, 102). However, there are several lines of evidence that indicate there is a function of pol E in DNA replication in eukaryotic cells (57). Interestingly, this function does not require DNA synthesis, based on work showing that the amino-terminal domain of the large pol E subunit containing both the pol and 3’ -+ 5’ exonuclease motifs is dispensable for DNA replication, DNA repair, and cell viability in budding yeast (103). Conversely, the carboxy-terminal region, lacking any known catalytic activity, is both necessary and sufficient for all of the essential functions of pol E (see below). Of course this does not mean that pol E is not participating in DNA synthesis events during chromosomal DNA replication, because there is ample evidence for this (57, 102) but it means that pol S (at least in budding yeast) is sufficient and can replace every catalytic function provided by pol E, thus representing a nice example of molecular redundancy in the cell.

PROTEINSWITHFUNCTIONSINDNAPROCESSES

275

4. ROLEOFPOLEINSPHASECHECKPOINTCONTROL It has been proposed that pol E acts as a sensor of DNA replication that coordinates the transcriptional and cell cycle responses to replication blocks (204). In response to DNA damage and replication blocks, yeast cells arrest at distinct points in the cell cycle and induce the transcription of genes encoding products that facilitate DNA repair. Mutant yeast cells that have a defect in pol E fail to arrest and to activate transcription of certain damage response genes. As a consequence, these cells enter into mitosis without correctly completing DNA replication. The checkpoint function of pol E has been mapped to the essential carboxy-terminal domain (104). Moreover, examination of the induction of certain genes in response to W damage revealed that Pol2p (pal E) is required to sense W damage and replication blocks when cells are in the S phase, whereas other checkpoint genes (RADS, RAD24, MEC3) are responsible for a cell cycle arrest in G, or G, (105). Finally, the S. cerevisiae gene DPBII, which has a checkpoint function, was shown to interact genetically with POL2 (106). However, the notion that such a checkpoint function by pol E is common to all eukaryotes turned out to be rather unlikely, because opposite results were reported in the fission yeast S. pombe (70). In this organism, pol E does not seem to have a role as a checkpoint sensor, in contrast to pol CY(49) (see above), suggesting that this checkpoint signal is generated prior to the elongation stage of DNA synthesis. Finally, as mentioned above, it has been shown that the essential role of the pol E carboxy terminus in budding yeast does not depend on induction of a checkpoint, which suggests that this pathway might not be essential in S. cerevisiae either (103). 5. ROLEOFPOL~ANDPOLEINEXCISIONREPAIR DNA synthesis during replication is not the only important task of ~016 and pol E (Fig. 1). The three major excision repair pathways, BER, NER, and MMR, involve a DNA synthesis step that replaces damaged or mismatched bases or nucleotides excised during repair. The BER pathway is essential in all organisms and is the main strategy to correct both spontaneous base loss or base damage and small DNA adducts. In mammalian cells, two different BER pathways have been identified. After spontaneous base loss or enzymatic removal of a damaged base by a DNA glycosylase, leaving a so-called apurinic/apyrimidinic site (AI’ site), a specific endonuclease incises the DNA double helix immediately 5’ to the deoxyribose-phosphate residue. Most probably, the main route for completion of this repair pathway is mediated by pol B in mammalian cells (107). This pol seems to be designed for this task because it contains an 8-kDa basic domain that has B-elimination activity that serves to excise the 5’-terminal deoxyribose-phosphate residue. Consequently, the same enzyme can excise and resynthesize the lesion by replacing only one

RNaseH I ?

\

1 DNA2 ?

5

3

3

5

Fen 1

F‘en 1

C RF;C

PCNA

Pol a/E

I

Fe\n 1

PROTEINS

WITH

FUNCTIONS

IN DNA PROCESSES

277

nucleotide (the so-called short-patch BER). In budding yeast, the situation seems to be different, because a deletion of the yeast pol B homolog POL4 did not significantly result in increased sensitivity to monoalkylating agents (108). In contrast, a POL3 (~016) mutant is highly sensitive to the same treatment (109). In another yeast genetic study, pol E has been suggested to be required for BER in viva (110). However, it is also possible that alkylation damage is repaired by a different pathway (which necessitates DNA synthesis by ~016 or pol ?? ) in yeast. This is a good example of the difficulty in assigning specific pols to particular pathways using either only biochemical or only genetic approaches (57). Biochemical analysis with human and Xenopus-oocyte cell extracts or fractionated extracts of a pol B-deleted mouse fibroblast cell line indicated that BER can be completed by an alternative route when more than one nucleotide is replaced. This pathway, called long-patch BER, is strictly PCNA dependent (III-113), indicating the involvement of a PCNA-dependent pollike ~016 or pol E in the resynthesis step. Another interpretation of these data is possible based on the finding that PCNA interacts and stimulates Fen 1 (114). This enzyme is thought to be required for excision of the damaged DNA strand 5’ to the lesion (115, 116) (see below). Long-patch BER has been reconstituted with highly purified human enzymes (88, 116,117) (see Fig. 1C). It appears that pol B, ~016, and pol E are able to complete long-patch AP site repair in vitro. PCNA and Fen 1 are required for both the production and ligation of long-patch repair intermediates created by ~016 and pol E, suggesting an important role of this complex in both excision and resynthesis steps. The repair intermediates in the absence of a ligase are significantly different when in vitro assays were performed using either ~016 or pol E, which could mirror the inefficiency of pol E to perform strand displacement synthesis (88). However, the question of the physiological significance of ~016 and pol E in mammalian long-patch BER is difficult to address and remains elusive. Nevertheless, taking into account that BER is an essential repair pathway (spontaneous base loss alone occurs up to lo4 times per cell per day) and that pol B-deleted mouse fibroblast cells are viable and show normal growth characteristics and only moderate sensitivity to alkylating agents (107), we believe that long-patch BER by ~016 and

FIG. 1. Gap-filling reactions during DNA replication (A), nucleotide excision repair (B), and long-patch base excision repair (C) by DNA polymerase 8/e. Three gap-filling reactions in DNA replication and DNA repair share identical enzymes that are associated in holoenzymes for elongation and processing of the DNA synthesis patches. All of these proteins (except the singlestrand binding protein RP-A) were shown to interact directly with PCNA. (C) The apuriniw apyrimidnic (A) site is indicated. (See text and references for details.)

278

MANUELSTUCKIETAL.

pol E may be a less efficient but nevertheless important and sufficient redundancy pathway for cell viability. The NER pathway acts most effectively on bulky or helix-distorting lesions. The main function of this repair pathway in humans is the removal of W-induced DNA photoproducts caused by sunlight. Several protein factors are required to recognize the damage and remove it within an oligonucleotide of about 30 residues in length (4). The resulting gap subsequently gets resynthesized and ligated. A role of ~016 or pol E in the synthesis step of this repair pathway is consistent with the strict requirement of PCNA for NER in human cell extracts, and, indeed, several lines of evidence suggest a participation of either ~016 or pol E in the gap-filling synthesis process during NER in vitro and in viva (118). Mammalian NER was reconstituted in vitro using highly purified enzymes (119). These and other in vitro data revealed that a combination of pol E, PCNA, RF-C, RP-A, and Lig I is well suited to the task of creating NER patches. Although ~016 seems to be able to replace pol E, it gives rise to a small portion of ligated products. Addition of Fen 1 increases substantially the yield of ligated repair products, which is consistent with the ability of pol S to perform limited strand displacement synthesis (Fig. 1B).It seems, however, that pol S is unable to reach the same repair efficiency seen with pol E in vitro (86). MMR is the third essential excision repair pathway in eukaryotes. Like the other repair pathways, MMR involves two steps: damage recognition and incision, followed by resynthesis and ligation. Although the damage recognition step has been studied in some detail (6, 120), relatively little is known about the excision and resynthesis steps. There is some evidence that at least pol 6 is involved in both steps. Human cell extract fractions that are defective in supporting mismatch repair in vitro could be complemented with a partially purified fraction that contained ~016 and was free of pol (Yand pol E. Purified calf thymus ~016 also fully restored mismatch repair ability of the depleted extract, indicating that pol 6 is required for MMR in human cells. However, due to the presence of pol (Y and pol E in the depleted extract, potential involvement of one or both pols in the reaction could not be excluded (121). Another report based on genetic studies in budding yeast suggested the involvement of the 3’ + 5’ exonuclease function of ~016 and pol E in the excision step (122). Clearly, much more work remains to be done to understand in detail the enzymatic requirements and mechanisms of the excision and synthesis steps in eukaryotic MMR. 6. POLBANDPOLEINOTHERREPAIRPATHWAYS DNA lesions in the template strand cause a block to the replication machinery. Replication across such lesions occurs by a mutagenic bypass process in which a wrong base is inserted opposite the lesion, or involves

PROTEINSWITHFUNCTIONSINDNAPROCESSES

279

processes that are relatively error free. Work by several groups has indicated the requirement of two pols, one for mutagenic (pal 5) and one for error-free bypass (pal I$ (9). Genetic studies implicating PCNA in this process have suggested that either ~016 or pol E may be involved as well. By the use of temperature-sensitive mutations of S. cerevisiae POL2 and POL3 it was shown that postreplicative bypass of W-damaged DNA is severely inhibited in the POL3 (pal 6) mutant but not in the POL2 (pal ?? ) mutant at the restrictive temperature. From these observations, the authors suggested a requirement of ~016 in postreplicative bypass of W-damaged DNA (123). However, a direct participation of ~016 in translesion synthesis is rather unlikely, because this enzyme is inefficient in carrying out lesion bypass synthesis in vitro (124). A study by Holmes and Haber (55) suggests pol F and pol E to be involved in double-strand break repair in vivo. MAT switching in budding yeast provides a good model for studying the involvment of enzymatic functions in DSBR via homologous recombination. Temperature-sensitive mutants of pol 6 and pol E were analyzed for their ability to complete MAT switching in vim The pol E mutant strain was almost as effective as the wild-type strain at both the permissive and the nonpermissive temperature, whereas the pol 6 mutant strain exhibited delayed and reduced MAT switching. These findings suggest that ~016 might be the major pol involved in this process, but some functional redundancy appears to exist between ~016 and pol E in DSBR as well (55).

II. DNA Polymerase Accessory Proteins A. Replication Factor C 1. GENERALDESCRIPTIONOFTHEENZYME RF-C is a heteropentameric complex composed of one large subunit (pl40/RFCl) and four smaller ones (p40/RFC4, p38/RFC5, p37lRFC2, and p36/RFC3), which interestingly all share considerable sequence similarity with each other as well as with their bacterial clamp loader counterparts in the pol III y complex (78). That all four small RF-C subunits share a common ancestor is further suggested by the fact that the genome of the archaeon Methanococcus jannaschii (125) contains only two RF-C genes, one encoding a large subunit and one encoding a small subunit homolog. The bacteriophage T4 also has a functional clamp (gp45) and a heteropentameric clamp loader composed of one gp44 and four gp62 subunits (78). Although derived from a common ancestor, the different subunits of RF-C are all essential for viability in yeast (126-129) and have been shown to have distinct functions, which involve the assembly of the RF-C complex, interaction with PCNA and DNA, and ATP hydrolysis (79,130 - 133). Some subcomplexes of

280

MANUELSTUCKIETAL.

RF-C subunits as well as the pentameric ing activity. 2. RF-CIs

complex also have PCNA unload-

ACLAMPLOADER

RF-C was first isolated by fractionation for activities that were required for SV40 replication in vitro (134). Later, by using an assay that measured stimulation of pol6 activity on primed ssDNA in the presence of PCNA, ATP, and RP-A, RF-C was also identified in yeast (135, 136) and calf thymus (82). Footprinting experiments revealed that RF-C bound DNA at templateprimer junctions in an ATP-dependent manner (137). However, RF-C on its own had little or no effect on DNA synthesis by pol CX,6, or E. An explanation for this lack of activity by RF-C on pols (Y, 6, and E is that its main function is to load the pol processivity factor PCNA onto the DNA (21, 138-140). RF-C-catalyzed PCNA loading is obligate for assembly of pol6 onto the DNA template to form the processive holoenzyme that acts during DNA synthesis of both leading and lagging strands at the replication fork (20,82,135 137, 139) (see also Section I). This is, however, by no means the only pathway requiring RF-C. As already discussed above and in more detail below, PCNA is involved in a number of DNA repair pathways, distinct from DNA replication, and because PCNA loading is required for these functions as well, RF-C is also an important component of these pathways. Two well-documented examples are NER (86,119) andlong-patch BER (88,112,113,117). Furthermore, RF-C may have separate roles in cell cycle regulation by virtue of its phosphorylation (78; A. Fotedar, personal communication) or interaction with other proteins. The yeast Rfc5p has been shown to interact with Spklp, an essential protein kinase for the transition from S phase to mitosis (141); the yeast RFC2 gene is required for an S phase checkpoint (142) and the S. pombe Rfc2p has been shown to play a key role in a DNA replication checkpoint (143). The exact mechanism of PCNA loading by RF-C is not yet understood in detail. It has been shown that RF-C dissociates from PCNA after loading it onto the DNA and does not remain directly associated with the pol 6 core (79), although the proteins can be isolated together as components of higher order complexes (144, 23). How exactly RF-C opens up the PCNA torus to load it onto the DNA is not known in detail. Studies on the homologous systems from baceriophage T4 (146-150) and E. co.& (I%), as well as the crystal structure of the 6’ subunit of the E. coli y complex (152), have given insights into how this may take place. The y complex can, on ATP binding, bind and open up the B-clamp (the structural and functional homolog of PCNA). The y/B/ATP complex then associates with the primer terminus and forms a ternary complex with the DNA. The DNA binding stimulates ATP-hydrolysis, which ejects the y complex and leaves B on the DNA (151). The mecha-

PROTEINS WITH FUNCTIONS IN DNA PROCESSES

281

nism of T4 clamp loading seems to be slightly different because

ATP hydrolysis is required prior to opening up the gp45 clamp (147, 149, 150). Early studies of RF-C showed that in the presence of the nonhydrolyzable ATP analog, ATP+, a stable RF-C/PCNA/DNA complex, is formed (81, 140, 153), which might implicate that the eukaryotic system more closely resembles the y complex in respect to how the PCNA clamp is loaded by RF-C (151). 3. RF-C Is LIKELY RESPONSIBLE FOR THE POLYMERASE SWITCH As discussed in Section I an important event in DNA replication is the switch from the pol (Yto ~016. An initial observation (21) suggested a role for RF-C in this mechanism. In very recent work we have shown that RF-C can displace pol (Yprior to PCNA loading and that RF-C can abrogate primer syrthesis by pol (Yat a critical length of 30 nucleotides (G. Maga, R. Mossi, and U. Hiibscher, unpublished results).

B. Proliferating

Cell Nuclear

Antigen

1. GENERAL DESCRIPTION OF THE PROTEIN Although the role of PCNA

in replication is important, it is by no means

its only known role. Indeed, as its name implies, it was first discovered as an antigen characteristically found in the nucleus of dividing cells (154, 155);

only a short while later PCNA was discovered independently by cell cycle researchers performing two-dimensional gel electrophoresis with radiolabeled cell extracts, and they named it cyclin because of the characteristic variation in synthesis during the cell cycle (156, 157). Later PCNA was shown to be an auxiliary factor for DNA replication (158-160). It can thus be argued that the role of PCNA as a cell cycle regulator was the first such discovery. The early history of PCNA research and studies of its cell cycle regulation and intracellular localization were summarized previously (161). Subsequently, PCNA has been showing up in new pathways, which may reflect the need to coordinate cellular functions such as cell cycle regulation and DNA repair with the fundamental process of DNA replication. In multicellular organisms development and differentiation add an additional level of complexity to the regulatory mechanisms required, and it seems that PCNA has a role there as well (77). PCNA is a homotrimeric ring-shaped protein with a molecular mass of 29 kDa for each monomer that occupies 120” in the ring. The crystal structure of PCNA (162, 163) revealed how PCNA can carry out its sliding clamp function on DNA by forming a trimeric ring that encircles the DNA strand, without making direct contact. In addition to its crystal structure being solved, a wealth of information about the involvement of PCNA in replication has accumulated and the list of PCNA interactors continues to grow

282

MANUEL

STUCK1 ET AL.

rapidly, although the exact function of most of these interactions has not been clarified. Here we discuss only the best characterized functions of PCNA and must mention most of its interacting partners only briefly. 2. PCNA Is

A

CWP

THAT COMMUNICATES

WITH MANY OTHER PROTEINS The exact sites of interaction on the PCNA molecule with the three multisubunit proteins ~018, pol E, and RF-C have been studied by several groups (g&164-166); the sites have been mapped to a region on the outer front surface of PCNA involving the loop that connects the two domains of each PCNA monomer and a loop immediately preceding the C terminus (Table III). In addition, pol E has been shown to interact with a loop on the backside of PCNA (96). Th e interaction sites on the other interaction partners have not been characterized in as much detail. An interesting question involves the site of PCNA interaction in the ~016 holoenzyme, and apparently there are multiple sites of interaction (see Section I). The most intensively studied PCNA-interacting protein is the ~21”~~~’ wafl kinase inhibitor of mammalian cells (167); this kinase can inhibit DNA replication in vitro by competing for PCNA binding. Although the interaction of the two proteins has been studied in great detail by many groups, the consequences of the interaction in vivo have still not been conclusively determined. The most popular theory is that p21 can mediate cell cycle arrest in response to DNA damage by its interaction with cyclincdks but achieves an acute stop of DNA synthesis by occupying the same binding sites on PCNA as ~016. It has, however, been shown that p2 1 does not interfere with repair DNA synthesis by pol S (168). The p21 protein belongs to a family of kinase inhibitors with two identified homologs in mammalian cells (p2 7K’P1 and ~57~‘~~) as well as homologs in other species. Two of these have been shown to engage in PCNA interaction. The cyclin-dependent kinase inhibitor p57 contains a PCNA-binding site within its C-terminal region, which is necessary for p57 to suppress fully myc/RAS-mediated transformation in primary cell lines (169). The Drosophila Dacapo protein, which probably represents the insect counterpart of mammalian p21, also interacts with Drosophila PCNA (270). Another well-characterized function of PCNA is to stimulate the activity of Fen 1 (171, 172). Fen 1 has been shown to play important roles in the maturation of Okazaki fragments during replicative DNA synthesis (173, 174), in long-patch BER (115) and NER by pol 6 (86). Although Fen 1 has a potent nuclease activity in the absence of PCNA, the interaction is important for its substrate recognition and it has been demonstrated that mutation of the PCNA binding site of Fen 1 impairs its activity in long-patch BER (117, 175).

PROTEINS

WITH FUNCTIONS

283

IN DNA PROCESSES

TABLE III PROTEINS~NTERACTINGWITH

PCNA”

Protein

Presumed functional tasks

PO16

P21 Drosophila Dacapo p27 and p57 Gadd45 MyD118 MSH2 and MLHl UDG2 UNG2 XP-G MCMT CAFl YB-1

Replicative pol, gap-filling pol in nucleotide excision repair, base excision repair and postreplication mismatch repair, double-strand break repair Gap-filling pol in nucleotide excision repair, base excision repair, double-strand break repair, replication, checkpoint control Adapter for PCNA, clamp loader and possible unloader; binds to telomerase Processing of Okazaki fragments, nucleotide excision repair, base excision repair, and recombination Ligation of Okazaki fragments, nucleotide excision repair, base excision repair Control of cell cycle cdk inhibitor; up-regulated in response to DNA damage Homolog of p2 1 cdk inhibitors Unknown role; up-regulated in response to DNA damage Terminal differentiation of some cell types Mismatch repair Base excision repair Base excision repair Nucleotide excision repair Transcription Chromatin assembly factor, repair Binds cisplatin-damaged DNA

Structure Chromatin Replisome Repairsome

PCNA role Assembly of chromatin Scaffold for pols and other enzymes Scaffold for pols and other enzymes

PO1

E

RF-C Fen 1 Lig 1 Cyclins and cdk complexes

"Forreferences, see text.

Besides the roles that PCNA plays in DNA replication and BER, it has been shown to be involved in most other types of DNA repair. Reconstitution of NER in vitro with purified proteins (119) showed a requirement for RF-C and PCNA for the DNA synthesis step, and, in addition, PCNA has been shown to interact directly with XP-G, an endonuclease that functions in the incision step of the reaction (176). What the function of this interaction might be is unknown, but it could provide a physical link between the steps of incision and gap filling. An alternative possibiliy is that PCNA is engaged in recognition of some types of DNA lesions. This is suggested by the fact that YB-1, a protein that binds to cisplatin-damaged DNA, has been shown to bind to PCNA as well (177).

284

MANUEL

STUCK1 ET AL.

Yet another type of DNA repair reaction in which PCNA has been demonstrated to play an active role is postreplicative MMR. Mutations in yeast PCNA have been shown to cause MMR defects, and direct interactions between PCNA and the yeast MMR proteins MSH2 and MLHl have also been shown (178). The exact role of PCNA in MMR is still unknown but it has been proposed to provide the signal that allows the mismatch proteins to distinguish between the parent and newly synthesized strand. In addition to these important roles in DNA repair, PCNA has been shown to interact with DNA glycosylases that are part of the normal BER machinery in mammalian cells. Both a putative cyclinlike minor uracil DNA glycosylase (UDGS) (179) and th e major nuclear uracil DNA glycosylase (UNGB) (180) h ave been shown to interact with PCNA. UNG2 localizes in replication foci during S phase and catalyzes rapid removal of dUMP residues that have been incorporated during DNA replication, probably as a part of a BER complex containing UNGB, RP-A, and PCNA that closely follows the advancing replication fork. PCNA may be responsible for guiding UNG2 to the sites of replication in the nucleus. All the DNA repair events outlined above, as well as the maturation of Okazaki fragments, must be concluded by sealing nicks in the doublestranded DNA. Mammalian cells contain four distinct ligase activities but the major replicative and repair ligase is the abundant Lig I (181). Lig I contains a consensus PCNA binding site at its N terminus and indeed binds to PCNA (182,183). As for UNGB, it has been demonstrated that PCNA binding is important for the localization of Lig I to sites of replication in the nucleus (182); additional data suggested that the replication factory targeting sequence/ PCNA binding site is required in G, to control the phosphorylationdephosphorylation status of Lig I (145). In eukaryotic cells the methylation state of the base cytosine can be inherited in a quasiepigenetic manner (184). DNA methylation is absent in Drosophila, Caenorhabditis elegans, and yeast, but in mammals DNA metbylation is involved in transcriptional regulation (185) and through that in several important regulatory processes, such as regulation of development (186) and genomic imprinting (187). Abnormal methylation can lead to several diseases, including cancer and fragile X syndrome, and it is therefore crucial that methylation patterns are correctly reproduced following DNA replication. DNA (cytosine-5)-methyltransferase (MCMT) is responsible for methylating newly replicated mammalian DNA, but how its activity is regulated is unknown. The interaction between MCMT and PCNA (188) has, however, provided some hints toward an understanding of the regulation of cytosine methylation. It was shown that a region between amino acids 163 and 174 of MCMT binds to PCNA. These residues lie inside a previously determined region that targets MCMT to sites of DNA replication (189), which is remi-

PROTEINS

WITH

FUNCTIONS

IN DNA PROCESSES

285

niscent of what has been observed with UNG2 and Lig I. PCNA does not seem to alter the activity of MCMT on synthetic substrates but competition with p21 seems to affect methylation in vivo (188). One possible interpretation of these results is that inhibition by p21 could prevent methylation of damaged DNA, which would lead to mutations. Another form of epigenetic inheritance that is conserved in eukaryotic cells is mediated by the chromatin assembly factor 1 (CAF-l), a complex of three subunits, ~150, ~60, and p48 (190). In vitro CAF-1 promotes assembly of chromatin during DNA replication and in vivo it is required for inheritance of epigenetically determined chromosomal states. It has been shown that PCNA provides a signal for marking recently replicated DNA for chromatin assembly and does so through a direct interaction with the CAF-1~150 subunit (191). The role of the CAF-1 PCNA interaction does, however, not seem to be limited to postreplicative chromatin assembly, given that a mechanistic link has also been observed between DNA repair and chromatin assembly. Incubation of UV-damaged plasmid DNA in cell-free extracts revealed that de novo nucleosome assembly occurs concomitantly with NER (192, Z93), and this chromatin assembly pathway is stimulated by CAF-1 (192). As we have already discussed, PCNA is recruited to sites of DNA damage and could therefore be expected to recruit CAF-1 to these sites. That this is indeed the case has been demonstrated (J. G. Moggs and G. Almouzni, personal communication). The PCNA interactions that we have discussed so far mostly involve PCNA bound to DNA, but in replicating cells PCNA continuously cycles between a chromatin-bound detergent-insoluble form in S phase and a soluble form when DNA replication is not taking place. A large population of unbound PCNA thus also exists and some of this PCNA has been found to associate in complexes with p2 1 and several pairs of cyclincdks, including cyclin D/cdk4, cyclin E/cdk2, cyclin A/cdk2, and cyclin B/cdc2 (194). It seems almost certain that these interactions play a role in coordinating the cell cycle with DNA replication and repair, but their exact role is still relatively poorly understood. On one hand it may be that PCNA can attract some of its interaction partners to active cyclincdk complexes for being phosphorylated; on the other hand it has been suggested that excessive levels of cyclin Dl can repress cell proliferation by inhibiting DNA replication and cdk2 activity through binding to PCNA and cdc2 (195). We have shown a direct interaction between PCNA and Cdk2 that involves the C-terminal part of PCNA and the catalytic site of Cdk2. Binding of PCNA to Cdk2 resulted in an inhibition of the kinase activity and in inhibition of pol 6 in a PCNA-dependent assay. PCNA and Cdk2 form a complex in S phase. Based on our results we propose that PCNA brings Cdk2 to proteins involved in DNA replication and can act as a

286

MANUEL

STUCK1 ET AL.

“stirrup” for cyclin A/cdk2 to further target proteins involved in DNA transactions (S. Koundrioukoff and U. Hiibscher, unpublished results). The list of reported PCNA interactors presented here is by no means complete and the collection keeps on growing. It should, however, be evident that PCNA has more than a dual role in DNA transactions; the paradigm could sound more like “PCNA has a role in every cellular pathway that involves DNA.” The notable exception whereby a role for PCNA has not been demonstrated is RNA transcription. Although the PCNA homolog of the phage T4, the gp45 protein, has been shown to play an important role in transcription of genes that are expressed late in the viral life cycle (196), and other examples of the involvement of viral PCNA homologs in transcription exist (197, 198), eukaryotic PCNA has not been shown to be directly involved in tran scription. This possibility has, however, not been formally excluded either.

III. Transcription Factors and Their Role in Activation of DNA Replication A.

Intrpductory

Remarks

An important parameter that determines the order of chromosome replication is transcriptional activity: most, but not all actively transcribed genes are replicated early in the S phase, whereas quiescent genes are replicated at later times (199). This is true even when two copies of the same gene reside in the same cell and one is transcriptionally active while the other is inactive (200, 201). Due to this observation it was proposed that the transcription process may play an important role in determining the order of replication of genes during S phase. We know that transcription by RNA polymerase II (pal II) in eukaryotes is carried out with the aid of many accessory proteins, including the general transcription factors (GTFs) TFIIA, -B, -D, -E, -F, and -H and the components of the MediatorSRB complex, which interacts with the C-terminal domain (CTD) of the pol II major subunit (202, 203). Transcription factors form a large family of regulatory proteins that can positively or negatively influence transcription by binding to regulatory elements in promoter elements. In the past 8 years, it has become clear that many transcription factors are multifunctional and also influence DNA replication. B. Lessons from Viral Systems Most of our knowledge about the role of transcription factors in activating DNA replication comes from studies of DNA tumor viruses (204-206). Replication origins are classified and viral origins belong to so-called simple origins, together with those origins found in mammalian (human, mouse) mitochondria, in yeasts (S. cerevisiae and S. pornbe), and in slime mould (207).

PROTEINS

WITH

FUNCTIONS

IN DNA PROCESSES

287

A typical viral origin is from 50 to about 1000 bp long and is defined by cis-acting mutations that prevent DNA replication. It consists of at least four elements, three of which [origin recognition element (ORE), DNA unwinding element (DUE), and A/T-rich element] are essential and therefore referred to as the origin core. The fourth element contains transcription factor binding sites (aux-1 and aux-2) that flank one or both sides of origin core (199, 207). Aux-1 is p roximal to the DUE element, and aux-2 is proximal to the A/T element. Origin auxiliary components stimulate replication only when they bind one or more transcription factors, and only when the transcription factor contains an activation domain that specifically interacts with the replication machinery (199,207,208). In some viral genomes, such as SV40, polyomavirus and Epstein-Barr virus genomes, the same sequence elements that function as promoters or enhancers in transcription also function as auxiliary components in replication. Substitution of the polyomavirus enhancer with the immunoglobulin gene enhancer conferred tissue-specific replicator-y ability to the virus, which showed for the first time the importance of these transcriptional elements in regulating viral DNA replication (204). In addition, the auxiliary sequences of the SV40 and polyoma virus origins of replication located adjacent to the binding site for the large T antigen contain elements recognized by cellular transcription factors such as Spl, APL and ~53. These transcription factors can increase viral origin activity up to lOOO-fold (204, 206, 209). Heterologous transcription factors can also activate viral replication when tethered to the origin of replication. For example, factors such as NF-KB, VP16, ElA, bovine polyomavirus E2, and yeast Gal4 stimulate polyomavirus DNA replication (209-211). H ow do transcriptional elements regulate viral origin activity? First, transcription factors could regulate the temporal order of DNA replication during S phase, just as they initiate transcription of different genes at different times during the cell division cycle (212). Second, the ability of a particular transcription factor to stimulate a particular origin may be limited to specific members of a transcription factor family, and to the availability of specific coactivator proteins (199,212, 213). Due to the lack of space, we refer the reader to several reviews discussing the role of transcriptin factors in viral DNA replication (208, 214, 215).

C. Lessons from Yeast Compared with the experiments in viral systems, the role of transcription factors in cellular DNA replication is less well understood. Nevertheless, analyses of eukaryotic cellular origins of DNA replication have been greatly facilitated by the biochemically and genetically tractable yeast S. cerevisiae.

288

MANUEL

STUCKI

ET AL.

DNA sequences of S. cerevisiae were identified that enabled the extrachromosomal replication of plasmids (216). These sequences are termed autonomously replicating sequences (ARSs) and they act as genuine origins of replication (217, 218). A detailed mutational analysis of one origin, ARSl, has led to identification of two essential elements, A and B (213). Element A contains an 11-bp consensus sequence that is conserved among all origins in S. cerevisiae. It is the binding site for the initiator protein called origin recognition complex (ORC) (219). The B element is composed of three functional sequences: Bl, B2, and B3, which are collectively essential for origin function (213). The Bl element is important for ORC binding and additional functions in replication initiation (220). The function of the B2 element is not clear. The B3 element contains a binding site for the yeast protein Abfl that functions as a transcriptional activator and repressor protein (221, 222). It was shown that, like enhancers in viral replication, the function of the B3 element of ARSl in plasmid replication can be replaced by the binding site for other yeast transcription factors such as Gal4 and RAPl. The ability of Gal4 and RAP1 to activate DNA replication resides in their transacting transcriptional activation domains (213). In addition, others have analyzed which type of transcriptional activation domain can activate a yeast cellular origin of replication in a chromosomal context (223). It has been reported before that acidic trans-activation domains derived from herpes simplex virus VP16 and from the tumor suppressor ~53 can stimulate viral DNA replication (224-226). Acidic activation domains can also activate a cellular origin of replication in a chromosomal context: when tethered to the yeast ARSl origin of replication, both VP16 and p53 acidic activation domains enhanced origin function (223). In addition, the C-terminal acidic region of the ABFl was sufficient for activating ARSl function when tethered to the origin. These findings strongly suggested functional conservation of the mechanisms used by the acidic activation domains to activate viral DNA replication in mammalian cells and chromosomal replication in yeast. Moreover, these experiments showed that activation of DNA replication and transcription by acidic activation domains has many common characteristics. A 50-amino acid C-terminal region of the Abflp seems to be sufficient to stimulate the function of ARSl21 origin when tethered to it. When tested for transcriptional activation of a EacZ reported gene, the same 50-amino acid C-terminal region of the Abflp had negligible transcriptional activation potential, suggesting that activation of DNA replication at ARS12 1 may occur independently of a transcriptional activation domain (227). Unlike ARSl, activation of ARSl21 by ABFl cannot be replaced by other replication factors, suggesting a special function of Abflp at ARS12 1.

PROTEINS

WITH FUNCTIONS

IN DNA PROCESSES

289

What is the exact mechanism by which transcription factors activate cellular DNA replication in viva? Results suggest that chromatin remodeling might be an important pathway utilized by transcription factors to activate yeast chromosomal replication (228). This extends and substantiates previous in vitro studies of the transcription factor’s role in activation of viral DNA replication (229,230). When tethered to a cellular replication origin, a fusion between the Gal4 DNA-binding domain and the C-terminal region of the breast cancer protein BRCAl, which has been previously shown to activate transcription in yeast as well as in mammalian cells, alters the local chromatin structure and stimulates chromosomal DNA replication. Cancer-predisposing mutations in BRCAl that abolish transcriptional activation function also prevented chromatin remodeling and activation of DNA replication (228). BRCAl is implicated in transcription (231,232), repair (233), replication, and recombination (234, 235). BRCAl is associated with chromatin-modified proteins such as BRCAB (236), ~300, and hBRG1 (237). In addition, BRCAl also interacts with components of transcription and repair machineries, such as mammalian pol II holoenzyme (238) and RAD51 (234). Although replication assays showed that the BRCAl activation domain is capable of activating an ARSl origin of replication in yeast, a direct involvement of BRCAl in mammalian DNA replication has to be established. Additional results suggested that DNA replication in yeast can be activated by recruitment of the pol II transcription complex (239). A defined, single interaction between a DNA-bound derivative of the activator Gal4 and GalllP, a mutant form of the pol II holoenzyme component Galll, suffices for stimulating DNA replication, as it does for transcription. Moreover, our results showed that recruitment of TATA-binding protein (TBP), which can activate transcription from a gene promoter, also stimulates DNA replication from ARSl origin (239). Th us, a DNA-binding protein does not need to interact directly with replication factors in order to activate replication from ARSl. In analogy with the implications of the Ga14-GalllP interaction in gene activation, stimulation of replication in these experiments does not require direct interaction of the DNA-binding protein with the machinery that helps in removing inhibitory chromatin structures (240). This model is also consistent with work (241) in which it was shown in Xenopus oocytes and HeLa cells that MCM proteins copurify with the pol II holoenzyme complex and GTFs. In addition, antibodies against MCM2 specifically inhibited transcription by pol II in microinjected Xenopus oocytes. The MCM 2-7 proteins, a family of six conserved proteins, are essential for the initiation of DNA synthesis at replication origins in S. cerevisiae (242, 243). Each of these proteins has a corresponding homolog in all eukaryotes examined so far (244), suggesting that their role in DNA replication initiation is universal.

290

MANUEL

STUCK1 ET AL.

Studies in S. cerevisiae indicate that mutations in the MCM 2-7 genes prevent initiation of DNA synthesis at replication origins in plasmids and chromosomes (245,246). Transactivators could recruit MCM proteins to origins of replication via contacts with pol II holoenzyme complex and thereby stimulate DNA replication (241). Furthermore, it could well be that MCM proteins play a previously unsuspected role in transcription. This hypothesis is in agreement with observations of a correlation between transcriptional activation by Statla and its ability to bind MCM5 (247). Taken together, all of these results raise the possibility that natural transcriptional activators binding near replication origins also activate replication by recruitment of the pol II transcription complex through direct interactions with one or more of its components (239, 241).

D. How Might DNA

Transcription

Factors Help in Activating

Replication?

The important and direct role of transcription factors in the initiation of DNA replication has been well substantiated through several experiments performed with animal viruses and S. cerevisue (208, 218, 228, 230). It has been postulated that involvement of the same transcription factors in the activation of DNA replication from origin sites as well as transcription from gene promoters requires that these proteins have either two different biochemical activities or one single activity that stimulates two distinct processes (213). That some transcriptional activators might indeed be endowed with different biochemical activities has been indicated by results of in vitro experiments showing that these proteins can interact not only with components of the pol II transcription complex but also with replication factors, e.g., the activation domains of VP16 and p53 can interact with RP-A, an essential component of the replication machinery (225,248,249). Therefore, these results indicate that transcriptional activators might stimulate DNA replication from cellular origins by recruitment of replication factors to origin sites, i.e., the same mechanism by which some activators have been shown to stimulate viral DNA replication (208). However, results of in vivo experiments showed that replication from a cellular origin site (ARSl) in yeast can be stimulated by the same biochemical activity (proteinprotein interaction) of a DNA-binding protein that also activates transcription from a gene promoter (239). Because this molecular interaction has been shown to activate transcription by recruiting the pol II transcription complex to DNA (250), th ese results strongly suggested that the same mechanism, i.e., recruitment of the pol II transcription complex, applies to this instance of activation of DNA replication. These results more broadly raise the possibility that natural transcriptional activators binding near replication origins also activate replication by

PROTEINS

WITH

FUNCTIONS

IN DNA PROCESSES

291

recruitment of the pol II transcription complex through direct interactions with one or more of its components. The recruited transcription complex may cause stimulation of DNA replication in a number of ways. One attractive possibility is that recruitment of this complex by transcriptional activators modifies chromatin structures to facilitate formation of the replication complex at the origin site. In fact, it has been demonstrated that acid activators can counteract the nucleosomal repression of transcription and that the antirepression function may be mediated by several chromatin remodeling systems (240, 251). These data strongly suggest that chromatin reconfiguration might be a highly conserved feature of transcription factors in activation of eukaryotic replication as well as transcription. In this respect it will be important to determine which nuclear machinery a transcription factor has to stimulate at a particular time and location. This may include the chromosomal region with which the transcription factor is associated (replication origin versus transcriptional promoter), the physiological state of the cell (proliferating versus nonproliferating), and specific association of transcription factors with highly condensed chromatin. In this way, origin sequences simply may not be accessible to transcription factors in highly condensed regions of the genome. In conclusion, it is now clear that there are several mechanisms by which transcription factors can facilitate initiation of DNA replication in eukaryotic cells. The mechanisms are not mutually exclusive, and might be, as in the case of different gene activation mechanisms, origin specific. The findings in yeast are highly interesting, and future studies should prove to be extraordinarily exciting for understanding the regulation of DNA synthesis by transcription factors.

IV. Perspectives and Conclusions We have exemplified that nature has evolved complex intertwined proteins that can maintain the integrity of the genome. Our simple-minded view that a particular enzyme machinery is responsible for a certain event has to be revised completely. We do not know, however, why, when, and how a particular enzyme protein machinery is called to a particular DNA transaction. The few selected examples, pols CL,6, and E, RF-C, PCNA, Ga14, ~53, BRCAl, and ABFl, clearly indicate that proteins/enzymes are multifunctional and that their different tasks must be coordinated depending on the physiological state of a cell or an organism and its response to environmental disturbances (e.g., DNA damage). The most fascinating versatile example among them is PCNA, which can interact with almost all important protein families involved in DNA metabolism and thus can act as a coordinator (Table III), a

292

MANUEL

STUCK1 ET AL.

stirrup, or a bridge in DNA replication, in several DNA repair events, in cell cycle control, in checkpoint control, and possibly even in transcription. We have now to learn in great detail how all these enzymes and proteins can interact and what the triggers are that engage a particular enzyme in a particular event. The combined efforts of biochemical, genetic, and cell biological approaches will be required to obtain a complete understanding of these most essential events in life. Last but not least, important as-yet unsolved medical problems such as cancer, aging, and other genetic diseases will profit from this basic understanding of how enzymatic machineries can maintain the software of the cell with an integrity that guarantees life, from unicellular organisms to humans.

ACKNOWLEDGMENTS We thank A. Barberis for exciting suggestions. Y.-S. Seo and A. Fotedar are acknowledged for communicating unpublished observations. The work in the authors’ lab has been supported by grants from the Swiss National Science Foundation to ZOJ and UH, the Swiss Cancer League to MS and UH, the EU-TMR project ERBMRXCT CT970125 to IS and UH, and the Kanton of Ziirich to IS, ZOJ, and UH.

REFERENCES 1. R. Laskey and M. Madine, In “DNA Replication in Eukaryotic Cells” (M. L. DePamphilis, ed.), pp. 119-130. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. 2. M. L. DePamphilis, “DNA Replication in Eukaryotic Cells.” Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. 3. H. Naegeh, “Mechanisms of DNA Damage Recognition in Mammalian Cells.” Springer, Heidelberg, 199 7. 4. R. D. Wood,J. Biol. Chem. 272,23465-23468 (1997). 5. ‘l’. Lindahl, P. Karran, and R. D. Wood, Cum Opin. Genet. Dev. 7,158169 (1997). 6. J. Jiricny, Mutat. Res. 409, 107-121 (1998). 7. R. Kanaar, J. H. Hoeijmakers, and P. C. van der Vliet, Trends Cell. Biol. 8,483-489 (1998). 8. M. Foiani et al., Biol. C&m. 379,1019-1023 (1998). 9. U. Hiibscher, H.-P. Nasheuer, and J. Syvaoja, Trends Biochem. Sci. 25, 143- 147 (2000). 10. T. S. Wang, Annu. Rev. Biochem. 60,513-552 (1991). 11. C. Santocanale, M. Foiani, G. Lucchini, and P. Plevani, J. BioZ. Chem. 268, 1343-1348 (1993). 12. T. Mizuno et aZ., Mol. Cell. BioZ. l&3552-3562 (1998). 13. P. A. Bullock, 0%. Rev. Biochem. Mol. BioZ. 32,503-568 (1997). 14. S. Waga and B. Stillman, Annu. Rev. Biochem. 67,721-751 (1998). 15. P. Park etal., Proc. N&Z. Acud. Sci. U.S.A. 91, 8700-8704 (1994). 16. l? J. Masterson, M. A. Stanley, A. P Lewis, and M. A. Romanos, J. Viral. 72, 7407-7419 (1998).

PROTEINS

17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61.

WITH FUNCTIONS

IN DNA PROCESSES

293

Y. Han, Y. M. Loo, K. T. Militello, andT. Melendy,]. Viral. 73,4899-4907 (1999). G. Prelich and B. Stillman, Cell 53, 117-126 (1988). D. H. Weinberg and T. J. Kelly, Proc. Natl. Acad. Sci. U.S.A. 86,9742-9746 (1989). T. Tsurimoto, T. Melendy, and B. Stillman, Nature &on&n) 346,534-539 (1990). T. Tsurimoto and B. Stillman,]. Biol. Chem. 266, 1961-1968 (1991). T. Eki, T. Matsumoto, Y. Murakami, and J. Hmwitz,]. Biol. Chem. 267,7284-7294 (1992). (1996). G. Maga and U. Hiibscher, Biochemistry 35,5764-5777 T Weiser et al.,]. Biol. Chem. 266, 10420-10428 (1991). G. Lucchini et al., Gene 90,99-104 (1990). R. A. Bambara, R. S. Murante, and L. A. Hemicksen,]. Biol. Chem. 272, 4647-4650 (1997). J. A. Rumbaugh, L. A. Hemicksen, M. S. DeMott, and R. A. Bambara,]. Biol. Chem. 274, 14602-14608 (1999). T. Formosa and T. Nittis, Genetics 151, 1459-1470 (1999). M. E. Budd and J. L. Campbell, Mol. Cell. Biol. 17,2136-2142 (1997). E. E. Biswas et al., Biochemistry 32,3013-3019 (1993). B. Stillman, Science 274, 1659-1664 (1996). (1995). M. Foiani, G. Liberi, G. Lucchini, and P Plevani, Mol. Cell. Biol. 15, 883-891 H.-P Nasheuer, A. Moore, A. F. Wahl, and T. S. Wang, ]. Biol. Chem. 266, 7893-7903 (199 1). C. Voitenleitner, E. Fanning, and H.-P. Nasheuer, Oncogene 14,1611-1615 (1997). C. Voitenleitner et al., Mol. Cell. Biol. 19,646-656 (1999). C. Desdouets et al., EMBOJ. 17,4139-4146 (1998). M. Goulian, C. J. Heard, and S. L. Grimm,]. Biol. Chem. 265,13221-13230 (1990). M. Goulian and C. J. Heard,]. Biol. Chem. 265, 13231-13239 (1990). J. Wittmeyer and T. Formosa, Mol. Cell. Biol. 17,4178-4190 (1997). J. Miles and T Formosa, Mol. Cell. Biol. b&5724-5735 (1992). S. Mimura and H. Takisawa, EMBOJ. 17,5699-5707 (1998). C. Kuhn et al., Oncogene l&769-774 (1999). T. Tsuji et aZ.,J. Biol. Chem. 273,6704-6709 (1998). M. Takemura et al., Oncogene 15,2483-2492 (1997). T. Mummenbrauer et al., Cell 85, 1089-1099 (1996). P. Huang, Oncogene 17,261-270 (1998). S. J. Elledge, Science 274,1664-1672 (1996). H. Murakami and H. Okayama, Nature @ondon) 374,817-819 (1995). G. D’Urso, B. Grallert, and P. Nurse,]. Cell Sci. 108,3109-3118 (1995). F. Marini et al., EMBOJ. 16, 639-650 (1997). A. G. Paulovich and L. H. Hartwell, Cell 82,841-817 (1995). 0. M. Aparicio, A. M. Stout, and S. P. Bell, Proc. Natl. Acad. Sci. U.S.A. 96, 9130-9135 (1999). J. K. Saxena, J. B. Hays, and E. J. Ackerman, Nucleic Acids Res. 18, 7425-7432 (1990). N. Oda et al., J. Biol. Chem. 271,13816-13820 (1996). A. M. Holmes and J. E. Haber, Cell 96,415-424 (1999). R. Hindges and U. Hiibscher, Biol. Chem. 378,345-362 (1997). P. M. Burgers, Chromosoma 107,218-227 (1998). K. J. Gerik, X. Li, A. Pautz, and P. M. Burgers,]. Biol. Chem. 273,19747-19755 (1998). P. M. Burgers and K. J. Gerik,]. Biol. Chem. 273,19756-19762 (1998). S. Zuo et al., Proc. Natl. Acad. Sci. U.S.A. 94, 11244-11249 (1997). M. Y. Lee, C. K. Tan, K. M. Downey, and A. G. So, Biochemistry 23, 1906-19 13 (1984).

294

MANUEL

STUCK1 ET AL.

62. M. Goulian, S. M. Herrmann, J. W. Sackett, and S. L. Grimm,J. Biol. Chem. 265,1640216411 (1990). 63. R. Hindges and U. Hiibscher, Gene 158,241-246 (1995). 64. J. Q. Zhou, H. He, C. K. Tan, K. M. Downey, and A. G. So, Nucleic Acids Res. 25, 10941099 (1997). 65. P. Hughes, I. Tratner, M. Ducoux, K. Piard, and G. Baldacci, Nucleic Acids Res. 2 7,2 108 2114 (1999). 66. A. Morrison, H. Araki, A. B. Clark, R. K. Hamatake, and A. Sugino, Cell 62, 1143-1151 (1990). 67. H. Araki, R. K. Hamatake, L. H. Johnston, and A. Sugino, Proc. N&l. Acad. Sci. U.S.A. 88, 4601-4605 (1991). 68. H. Araki et al., Nucleic Acids Res. 19,4867-4872 (1991). 69. T. S. Wang, In “DNA Replication in Eukaryotic Cells” (M. L. DePamphilis, ed.), pp. 461493. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. 70. G. D’Urso and P Nurse, Proc. Natl. Acad. Sci. U.S.A. 94, 12491-1246 (1997). (1989). 71. J. Syviaoja and S. Linn, J. Biol. Chem. 264,2489-2497 72. F. Focher et al., Nucleic Acids Res. 17, 1805-1821 (1989). 73. T. Kesti and J. E. Syvaoja,J. BioZ. Chem. 266,6336-6341 (1991). 74. D. Huang, R. Knuuti, H. Palosaari, H. Pospiech, and J. E. Syvaoja, Biochim. Biophys. Acta 1445,363-371 (1999). 75. D. Huang, H. Pospiech, T. Kesti, and J. E. Syvaoja, BiochemJ. 339,657-665 (1999). 76. U. Hiibscher, G. Maga, and V N. Podust, In “DNA Replication in Eukaryotic Cells” (M. L. DePamphilis, ed.), pp. 525-543. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. 77. Z. 0. Jbnsson and U. Hiibscher, BioEssays 19,967-975 (1997). 78. R. Mossi and U. Hiibscher, Eur. J. Biochem. 254,209-216 (1998). 79. V N. Podust, N. Tiwari, S. Stephan, and E. Fanning, J. Biol. Chem. 273, 31992-31999 (1998). 80. L. M. Podust, V. N. Podust, C. Floth, and U. Hiibscher, Nucleic Acids Res. 22,2970-2975 (1994). 81. V. Podust, V Mikhailov, A. Georgaki, and U. Hiibscher, Chromosome 102, 133-141 (1992). 82. V. N. Podust, A. Georgaki, B. Strack, and U. Hiibscher, Nucleic Acids Res. 20,4159-4165 (1992). 83. F. Uhlmann, E. Gibbs, J. Cai, M. O’Donnell, and J. Hurwitz,~. Biol. Chem. 272,1006510071 (1997). 84. V. N. Podust and U. Hiibscher, Nucleic Acids Res. 21,841-846 (1993). 85. V. N. Podust, L. M. Podust, F. Miiller, and U. Hiibscher, Biochembtq 34, 5003-5010 (1995). 86. M. K. Shivji, V. N. Podust, U. Hiibscher, and R. D. Wood, Biochemistry 34,5011-5017 (1995). 87. R. Mossi, E. Ferrari, and U. Hubscher,J. Biol. C/rem. 273,14322-14330 (1998). 88. B. Pascucci, M. Stucki, Z. 0. Jonsson, E. Dogliotti, and U. Hiibscher, J. Biol. Chem. 274, 33696-33702 (1999). 89. W. C. Brown and J, L. Campbell,]. Biol. Chem. 268,21706-21710 (1993). 90. P. Zhang et aZ.,J. Biol. Chem. 270, 7993-7998 (1995). 91. Z. 0. Jonsson, R. Hindges, and U. Hiibscher, EMBOJ. 17,2412-2425 (1998). 92. P. Zhang et al., J. Biol. Chem. 274,26647-26653 (1999). 93. D. J. Mozzherin, C.-K. Tan, K. M. Downey, and P. A. Fisher,]. Biol. Chem. 274,1986219867 (1999).

PROTEINS

WITH

FUNCTIONS

IN DNA PROCESSES

295

94. S. H. Lee, Z. Q. Pan, A. D. Kwong, P. M. Burgers, and J. Hurwitz, J. Biol. Chem. 266, 22707-22717 (1991). 95. G. Maga and U. Hiibscher, Biochemistry 34,891-901 (1995). 96. G. Maga, Z. 0. Jhnsson, M. Stucki, S. Spadari, and U. Hiibscher, J. Mol. Biol. 285, 259-

267 (1999). 97. S. Waga and B. Stillman, Nature (London) 369,207-212 (1994). 98. R. Gary et al., Mol. Cell. Biol. 19,5373-5382 (1999). 99. M. E. Budd and J. L. Campbell, Mol. Cell. Biol. 13,496-505 (1993). 100. J. J. Turchi, L. Huang, R. S. Murante, Y. Kim, and R. A. Bambara, Proc. h’atl. Acad. Sci. U.S.A. 91,9803-9807 (1994). 101. T. Zlotkin et al., EMBOJ. 15,2298-2305 (1996). 102. H. Pospiech et al., Nucl. Acids Res. 27,3799-3804 (1999). 103. T. Kesti, K. Flick, S. Keranen, J. E. SyvHoja, and C. Wittenberg, Mol. Cell 3, 679-685 (1999). 104. T. A. Navas, Z. Zhou, and S. J. Elledge, Cell 80,29-39 (1995). 105. T. A. Navas, Y. Sanchez, and S. J. Elledge, Genes Dec. 10,2632-2643 (1996). 106. H. Araki, S. H. Leem, A. Phongdara, and A. Sugino, Proc. Natl. Acad. Sci. U.S.A. 92, 11791-11795 (1995). 107. R. W. Sobol et al., Nature jlondon) 379, 183-186 (1996). 108. S. H. Leem, P. A. Ropp, and A. Sugino, Nucleic Acids Res. 22,3011-3017 (1994). 109. A. Blank, B. Kim, and L. A. Loeb, Proc. Natl. Acad. Sci. U.S.A. g&9047-9051 (1994). 110. Z. Wang, X. Wu, and E. C. Friedberg, Mol. Cell. Biol. 13,1051-1058 (1993). 111. Y. Matsumoto, K. Kim, and D. F. Bogenhagen, Mol. Cell. BioZ. 14, 6187-6197 (1994). 112. G. Frosina et al.,]. Biol. Chem. 271,9573-9578 (1996). 113. M. Stucki et al., Oncogene 17,835-843 (1998). 114. E. Warbrick, D. P. Lane, D. M. Glower, and L. S. Cox, Oncogene 14,2313-2321 (1997). 115. K. Kim, S. Biade, and Y. Matsumoto,J. BioZ. Chem. 273, 8842-8848 (1998). 116. A. Klungland and T. Lindahl, EMBOJ. 16,3341-3348 (1997). 117. Y. Matsumoto et al., J. Biol. Chem. (in press) (1999). 118. R. D. Wood and M. K. Shivji, Carcinogenesis 18,605-610 (1997). 119. A. Aboussekhra et al., Cell 80,859-868 (1995). 120. R. D. Kolodner and G. T. Marsischky, Curr. Opin. Genet. Deu. 9,89-96 (1999). 121. M. J. Longley, A. J. Pierce, and P. Modrich,J. BioZ. Chem. 272,10917-10921 (1997). 122. H. T. Tran, D. A. Gordenin, and M. A. Resnick, Mol. Cell. BioZ. 19,2000-2007 (1999). 123. C. A. Torres-Ramos, S. P&ash, and L. Prakash,J. Biol. C/rem. 272,25445-25448 (1997). 124. D. J. Mozzherin, S. Shibutani, C. K. Tan, K. M. Downey, and P. A. Fisher, Proc. Natl. Acad. Sci. U.S.A. 94, 6126-6131 (1997). 125. C. J. Bult et al., Science 273, 1058-1073 (1996). 126. G. Cullmann, K. Fien, R. Kobayashi, and B. Stillman, hfol. Cell. BioZ. 15, 4661-4671 (1995). 127. X. Li and P M. Burgers, Proc. NatZ. Acad. Sci. U.S.A. 9l,868-872 (1994). 128. X. Li and P M. Burgers,J. Biol. Chem. 269,21880-21884 (1994). 129. V. Noskov et al., Nucleic Acids Res. 22, 1527-1235 (1994). 130. J. Cai et al., Proc. N&l. Acad. Sci. U.S.A. 93, 12896-12901 (1996). 131. V. Ellison and B. Stillman,J. BioZ. Chem. 273,5979-5987 (1998). 132. V. N. Podust and E. Fanning,J. BioZ. Chem. 272,6303-6310 (1997). 133. l? Uhlmann et al., Proc. NatZ. Acad. Sci. U.S.A. 93,6521-6526 (1996). 134. T. Tsurimoto and B. Stillman, Mol. Cell. BioZ. 9, 609-619 (1989). 135. B. L. Yoder and P. M. Burgers,J. Biol. Chem. 266,22689-92267 (1991). 136. K. Fien and B. Stillman, Mol. Cell. BioZ. 12, 155-163 (1992).

296

MANUEL

STUCK1 ET AL.

137. T. Tsurimoto and B. Stillman,]. Biol. Chem. 266,1950-1960 (1991). 138. L. M. Podust, V. N. Podust, J. M. Sogo, and U. Hiibscher, Mol. Cell. Biol. 15,3072-3081 (1995). 139. 140. 141. 142. 143. 144. 145. 146. 147.

148. 149. 150. 151. 152. 153. 154.

155. 156. 157. 158.

159. 160. 161. 162. 163. 164.

165. 166. 167. 168.

169. 170.

171.

S. H. Lee, A. D. Kwong, Z. Q. Pan, and J. H urwitqj. Biol. Chem. 266,594-602 (1991). P. M. Burgers,J. Biol. Chem. 266,22698-22706 (1991). K. Sugimoto et al., Proc. Natl. Acad. Sci. U.S.A. 93,7048-7052 (1996). (1998). V. N. Noskov, H. Araki, and A. Sugino, Mol. Cell. Biol. 18,4914-4923 N. Reynolds, P. A. Fantes, and S. A. MacNeill, Nucleic Acids Res. 27,462-469 (1999). C. M. Simbulan-Rosenthal et al., Biochemistry 35,11622-11633 (1996). R. Rossi etal., EMBOJ. 18,5745-5754 (1999). G. J. Latham, D. J. Bacheller, P Pietroni, and P H. von Hippel,J. Biol. Chem. 272,3168531692 (1997). G. J. Latham, P. Pietroni, F. Dong, M. C. Young, and P H. von Hippel, J. Mol. Biol. 264, 426-439 (1996). G. J. Latham, D. J. Bacheller, P. Pietroni, and P. H. von Hippel,J. Biol. Chem. 272,3167731684 (1997). l? Pietroni, M. C. Young, G. J. Latham, and P. H. von Hippel,J. Biol. Chem. 272,3166631676 (1997). D. J. Sexton, T. E. Carver, A. J. Berdis, and S. J. Benkovic,J. Biol. Chem. 27l, 28045-28051 (1996). J. Turner, M. M. Hingorani, Z. Kelman, and M. O’Donnell, EMBOJ. l&771-783 (1999). B. Guenther, R. Onmst, A. Sali, M. O’Donnell, and J. Kuriyan, Cell 91,335-345 (1997). S.-H. Lee and J. Hurwitz, Proc. Natl. Acad. Sci. U.S.A. 87,5672-5676 (1990). K. Miyachi, M. J. Fritzler, and C.-K. Tan,]. Immunol. l21,2228-2234 (1978). Y. Takasaki, J. S. Deng, and E. M. Tan,I. Exp. Med. 154,1899-1909 (1981). R. Bravo and J. E. Celis,]. Cell. Biol. 84,795-802 (1980). R. Bravo et al., Exp. Cell. Res. 136,311-319 (1981). C.-K. Tan, C. Castillo, A. G. So, and K. M. Downey,]. Biol. Chem. 26l, 12310-12316 (1986). G. Prelich et al., Nature (London) 326,517-520 (1987). R. Bravo, R. Frank, P. A. Blundell, and H. Macdonald-Bravo, Nature (London) 326,515517 (1987). M. B. Mathews, In “Growth Control during Cell Aging” (E. Wang and H. R. Warner, eds.), pp. 89-120. CRC Press, Boca Raton, 1989. J. M. Gulbis, Z. Kelman, J, Hun&z, M. O’Donnell, and J. Kuriyan, Cell 87, 297-306 (1996). T. S. R. Krishna, X.-P. Kong, S. Gary, P. M. Burgers, and J. Kuriyan, Cell 79,1233-1243 (1994). Z. Kelman, S. Zuo, M. P. Arroyo, T. S. Wang, and J. Hurwitz, Proc. N&l. Acad. Sci. U.S.A. 96,9515-9520 (1999). J. C. Eissenberg, R. Ayyagari, X. V. Gomes, and P M. Burgers, Mol. Cell. Biol. 17,63676378 (1997). T. Oku et al., Genes Cells 3,357-369 (1998). S. Waga, G. J. Hannon, D. Beach, and B. Stillman, Nature (London) 369,574-578 (1994). M. K. Shivji, E. Fermri, K. Ball, U. Hiibscher, and R. D. Wood, Oncogene 17,2827-2838 (1998). H. Watanabe et al., Proc. N&l. Acad. Sci. U.S.A. 95,1392-1397 (1998). E. Warbrick, W. Heatherington, D. P. Lane, and D. M. Glover, Nucleic Acids Res. 26, 3925-3932 (1998). X. Wu et al., Nucleic Acid..sRex 24,2036-2043 (1996).

PROTEINS

WITH

FUNCTIONS

IN DNA

PROCESSES

297

172. U. Chen, S. Chen, I? Saha, and A. Dutta, Proc. Natl. Acad. Sci. U.S.A. 93,11597-11602 (1996). 1, and R. A. Bambara,J. Biol. Gem. 269,1191-1196 173. R. S. Murante, L. Huang, J. J. T urc h’ (1994). 174. S. Waga, G. Bauer, and B. Stillman,J. Biol. C&m. 269,10923-10934 (1994). 175. R. Gary, K. Kim, H. L. Cornelius, M. S. Park, and Y. Matsumoto,J. Biol. Chem. 274,43544363 (1999). 176. R. Gary, D. L. Ludwig, H. L. Cornelius, M. A. MacInnes, and M. S. Park, J. Biol. Chem. 272,24522-24529 (1997). 177. T. Ise et al., Cancer Res. 59,342-346 (1999). 178. R. E. Johnson et al.,J. Biol. Chem. 271,27987-27990 (1996). 179. S. J. Muller-Weeks and S. Caradonna, Exp. Cell. Res. 226,346-355 (1996). 180. M. Otterlei et al., EMBOJ. 18, 3834-3844 (1999). 181. A. E. Tomkinson and D. S. Levin, BioEssays 19,893-901 (1997). 182. A. Montecucco et al., EMBOJ. 17,3786-3795 (1998). 183. D. S. Levin, W. Bai, N. Yao, M. O’Donnell, and A. E. Tomkinson, Proc. Natl. Acad. Sci. U.S.A. 94,12863-12868 (1998). 184. A. Bird, Cell 70,5-8 (1992). 185. X. Nan, F. J. Campoy, and A. Bird, Cell 88, 471-481 (1997). 186. E. Li, T. H. Bestor, and R. Jaenisch, Cell 69,915-926 (1992). 187. E. Li, C. Beard, and R. Jaenisch, Nature (Londonj 366,362-365 (1993). 188. L. S. Chuang et al., Science 277,1996-2000 (1997). 189. H. Leonhardt, A. W. Page, H. U. Weier, and T. H. Bestor, Cell 71, 865-873 (1992). 190. S. Smith and B. Stillman, Cell 58,15-25 (1989). 191. K. Shibahara and B. Stillman, Cell 96, 575-585 (1999). 192. P. H. Gaillard et al., Cell 86,887-896 (1996). 193. P. H. Gaillard et al., EMBOJ. 16,6281-6289 (1997). 194. Y. Xiong, H. Zhang, and D. Beach, Cell 7l, 505-514 (1992). 195. J. Fukami-Kobayashi and Y. Mitsui, Exp. Cell Res. 246,338-347 (1999). 196. D. R. Herendeen, G. A. Kassavetis, and E. P. Geiduschek, Science 256,1298-1303 (1992). 197. A. M. Crawford and L. K. Miller,J. Viral. 62,2773-2781 (1988). 198. D. R. O’Reilly, A. M. Crawford, and L. K. Miller, Nature (London) 337,606 (1989). 199. M. L. DePamphilis, Trends Cell. Biol. 3, 161-167 (1993). 200. D. M. Gilbert, Proc. Natl. Acad. Sci. U.S.A. 83,2924-2928 (1986). 201. M. Schmidt and B. R. Migeon, Proc. NatZ. Acad. Sci. U.S.A. 87,3685-3689 (1990). (1998). 202. V. E. Myer and R. A. Young,]. Biol. Chem. 273,27757-27760 203. G. Orphanides, T. Lagrange, and D. Reinberg, Genes Dev. 10,2657-2683 (1996). 204. J. de Villiers, W. Schaffner, C. Tjmdall, S. Lupton, and R. Kamen, Nature (London) 312, 242-246 (1984). 205. K. A. Jones, J. T. Kadonaga, P. J. Rosenfeld, T. J. Kelly, and R. Tjian, Cell 48,79-89 (1987). 206. Z.-S. Guo, C. Gutierrez, V. Heine, J. M. Sogo, and M. L. DePamphilis, Mol. Cell. Biol. 9, 3593-3602 (1989). 207. M. L. DePamphilis, In “DNA Replication in Eukaryotic Cells” (M. L. DePamphilis, ed.), pp. 45-86. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. 208. P. C. van der Vliet, In “DNA Replication in Eukaryotic Cells” (M. L. DePamphihs, ed.), pp. 87-118. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. 209. Z.-S. Guo and M. L. DePamphilis, MoZ. Cell. Biol. l2,2514-2524 (1992). 210. L. Gutierrez, Z.-S. Guo, J. Roberts, and M. L. DePamphilis, Mol. Cell. Biol. 10,1719-1728 (1990). 211. M. Baru, M. Shlissel, and H. Manor,]. Viral. 65,3496-3503 (1991).

298

MANUEL

STUCKI

ET AL.

212. N. H. Heintz, Cum @in. Cell Biol. 4,459-467 (1992). 213. Y. Marahrens and B. St&mm, Science 255,817-23 (1992). 214. A. Stenlund, In “DNA Replication in Eukaryotic Cells” (M. L. DePamphilis, ed.), pp. 679697. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. 215. J. A. Hassel and B. T. Brinton, In “DNA Replication in Eukaryotic Cells” (M. L. DePamphilis, ed.), pp. 639 - 679. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1996. 216. D. T. Stinchcomb, K. Struhl, and R. W. Davis, Nature (London) 282,39-43 (1979). 217. B. J. Brewer and W. L. Fangman, Cell 5I,473-481 (1987). 218. C. S. Newlon, Cell 9I, 717-720 (1997). 219. S. P. Bell and B. Stillman, Nature 357, 128-134 (1992). 220. A. Rowley, J. H. Cocker, J. Harwood, and J. F. X. Diffley, EMBOJ. 14,2631-2641(1995). 221. J. F. X. Diffley and B. Stilhnan, Proc. N&Z. Acad. Sci. U.S.A. 85,2120-2124 (1988). 222. C. S. Newlon and J. F. Theis, Curr. @in. Genet. Dev. 3,752-758 (1993). 223. R. Li et al., Mol. Cell. Biol. 18, 1296-1302 (1998). 224. R. Li and M. R. Botchan, Cell 73,1207-1221 (1993). 225. Z. He, B. T. Brinton, J. Greenblatt, J. A. Hassell, and C. J. Ingles, Cell 73,1223-1232 (1993). 226. L. Z. Cheng, J. L. Workman, R. E. Kingston, and T. J. Kelly, Proc. N&l. Acud. Sci. U.S.A. 89,589-593 (1992). 227. S. Wiltshire, S. Raychaudhuri, and S. Eisenberg, Nucleic Acids Res. 25, 4250-4256 (1997). 228. Y.-E Hu, Z. L. Hao, and R. Li, Genes Deu. 13,637-642 (1999). 229. L. Cheng and T. J. Kelly, Cell 59,541-551 (1989). 230. R. Li and M. R. Botchan, Proc. N&Z. Acud. Sci. U.S.A. g&7051-7055 (1994). 231. M. S. Chapman and I. M. Verma, Nature 382,678-679 (1996). 232. A. N. Monteiro, A. August, and H. Hanafusa, Proc. Natl. Acad. Sci. U.S.A. 93, 1359513599 (1996). 233. L. C. Gowen, A. M. Avrutskaya, A. M. Latour, B. H. Koller, and S. A. Leadon, Science 28l, 1009-1012 (1998). 234. R. Scully et al., Cell 88,265-275 (1997). 235. R. Scully et al., Cell 90, 425-435 (1997). 236. J. Chen et al., Mol. Cell 2,317-328 (1998). 237. A. S. Neish, S. F. Anderson, B. P. Schlegel, W. Wei, and J. D. Parvin, Nucleic Acids Res. 26, 847-853 (1998). 238. R. Scully et al., Proc. N&l. Acad. Sci. U.S.A. 94, 5605-5610 (1997). 239. I. Stagljar, U. Hiibscher, and A. Barberis, Biol. Chem. 380,525-530 (1999). 240. A. Barber-is and L. Gaudreau, Biol. Chem. 379,1397-1405 (1998). 241. K. Yankulov et al., Mol. Cell. Biol. 19,6154-6163 (1999). 242. C. S. Newlon, In “DNA Replication in Eukaryotic Cells” (M. L. DePamphilis, ed.), pp. 873914. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 1997. 243. S. E. Kearsey and K. Labib, Biochim. Biophys. Actu 1398,13-136 (1998). 244. B. K. Tye, Trends Cell. BioZ. 4, 160-166 (1994). 245. K. M. Hennessy, A. Lee, E. Chen, and D. Botstein, Genes Deu. 5,958-969 (1991). 246. H. Yan, A. M. Merchant, and B. K. Tye, Genes Dev. 7,2149-2160 (1993). 247. J. J. Zhang et al., EMBOJ. 17, 6963-6971 (1998). 248. R. Li and M. R. Botchan, Cell 73,1207-1221 (1993). 249. L. M. Leiter, J. Chen, T. Marathe, M. Tanaka, and A. Dutta, Oncogene 12, 2661-2668 (1996). 250. A. Barberis et al., Cell Sl, 359-368 (1995). 251. C. L. Peterson and J. W. Tamkun, Trends Biochem. Sci. 20,143-146 (1995).

Signal Transduction Pathways and the Modification of Chromatin Structure JAMES

R. DAVIE AND

VIRGINIA

A. SPENCER

Manitoba Institute of Cell Biology and Department of Biochemistry and Medical Genetics University of Manitoba Winnipeg, Manitoba, Canada R3E OV9 I. Chromatin Organization ....................................... II. Core Histone Tails ............................................. III. Transcriptionally Active Chromatin .............................. IV. Histone Variants and Modifications .............................. A. Core Histone Variants ....................................... ....................................... B. HistoneHlSubtypes C. Core Histone Modifications .................................. ...................................... D. HistoneUbiquitination E. Histone Acetylation ......................................... F. Histone Methylation ........................................ G. Histone Phosphorylation and Mitosis .......................... H. Histone Phospholylation, Transcription, and Signal Transduction V Karyoskeleton and Organization of Chromatin ..................... VI. Karyoskeleton and Transcription Factories ........................ .......... VII. Transcriptionally Active Chromatin and the Karyoskeleton VIII. Transcription Factors and the Karyoskeleton ....................... IX. HATS, HDACs, and the Karyoskeleton ............................ X. Mechanical Signaling Pathways and Organization of Nuclear DNA XI. Future Directions ............................................. References ...................................................

Mechanical formation

...

300 302 303 305 305 305 306 306 307 321 322 322 326 327 327 328 329 329 331 332

and chemical signaling pathways are involved in transmitting in-

from the exterior of a cell to its chromatin. The mechanical

pathway consists of a tissue matrix system that links together sional skeletal networks,

the extracellular

matrix, cytoskeleton,

signaling

the three-dimenand karyoskele-

ton. The tissue matrix system governs cell and nuclear shape and forms a structural

and functional

Further,

connection

this mechanical

progression

between

signaling

and gene expression. Chemical

mitogen-activated

protein kinase (MAPK)

kinases that modify transcription

Propas in Nucleic Acid Research and Molecular Biology, Vol. 65

the cell periphery

pathway

and chromatin.

has a role in controlling

pathway can stimulate the activity of

factors, nonhistone chromosomal

299

cell cycle

signaling pathways such as the Ras/ proteins, and

Copyight 0 2001 try Academc Prrs, All rights of reproduction ,,I any form reselved. 0078-6603~01 s35.00

300

JAMES R. DAVIE AND VIRGINIA

histones. Activation of the Ras/MAPK

pathway results in the alteration

matin structure

The tissue matrix and chemical

and gene expression.

pathways are not independent

A. SPENCER

of chrosignaling

and one signaling pathway can affect the other. In

this chapter, we will review chromatin organization,

histone variants and modifi-

cations, and the impact that signaling pathways have on chromatin structure and fUnCtiOn.

0 2000 Academic Press.

I. Chromatin Organization Nuclear DNA exists as a hierarchy of chromatin structures, resulting in compaction of the nuclear DNA about lO,OOO-fold. The histones provide an essential role in the compaction of nuclear DNA. Eukaryotes have five classes of histones, H2A, H2B, H3, H4, and Hl. H2A, H2B, H3, and H4 are referred to as the core histones. The core histones are arranged as an octamer in the nucleosome core particle, the basic repeating structural unit in chromatin. The core histone octamer is organized as an (H3-H4), tetramer and two H2A-H2B dimers positioned on both sides of the tetramer. Around the histone octamer core are wrapped 146 bp of DNA, forming the nucleosome core particle. The crystal structure of the nucleosome core particle was reported in 1997 (I). The core histones (H2A, H2B, H3, and H4) have a similar structure, with a basic N-terminal domain, a globular domain organized by the histone fold, and a C-terminal tail (Fig. 1). The distribution of basic amino acids in the core histones is asymmetric, with the N-terminal portion of the molecule having a high amount of the basic amino acid residues. Some core histones (e.g., H3 and H2A) also have a short C-terminal tail. H3 and H4 are evolutionarily conserved, as is the histone fold portion of the H2A and H2B. The histone fold domains of the four core histones mediate histone-histone and histoneDNA interactions (1). The nucleosomes are joined by linker DNA, which is of varying length. The Hl histones or linker histones bind to the linker DNA and to core histones. Hl stabilizes the higher order compaction of chromatin. Hl has a tripartite structure consisting of a central globular core and lysine-rich N- and C-terminal domains (Fig. 2). The globular domain site I, which has a winged helix DNA-binding domain found in the hepatocyte transcription factor HNF3 (2), binds to one linker DNA strand as it exits or enters the nucleosome, whereas the globular domain site II, which consists of basic amino acids, binds to nucleosomal DNA near the dyad axis of symmetry of the nucleosome (3). Not all Hl histones have the globular domain. Tetruhymena thermophila macronuclear Hl, for example, lacks a globular domain. In contrast, yeast Hl (Hholp) has two globular domains.

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

301

K

HDAC 1)

HATA

FIG. 1. Sites of postsynthetic modifications on the core histones. The structures of the H2Atetramers, Hlb, and the sites of modification are shown. The modifications shown are acetylation (AC), phosphorylation (P), ubiquitination (Ub), and methylation (Me). The enzymes catalyzing reversible acetylation and phosphorylation are shown (HAT, histone acetyltransferase; HDAC, histone deacetylase; PPl, protein phosphatase). H2B dimers, the (H3-H4),

Chromatin in the interphase nucleus is highly compact and is organized into chromosome territories (4, 5). Decondensation of the chromatin fiber to the 30-nm form, the structure of which remains controversial (2), is rarely observed in the interphase nucleus (4, 6). Chromosome territories are visualized by fluorescent in situ hybridization with chromosome-specific DNA whole library probes. However, a novel method to monitor chromosome territories

FIG. 2. Sites of postsynthetic modifications on mouse histone Hlb. The sites of phosphorylation (P) and the enzymes catalyzing reversible phosphorylation (Cdk2, cyclin-dependent protein kinase 2; PPl, protein phosphatase) are shown.

302

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

and their dynamics has been developed by Cook and colleagues. By incorporating a fluorescent analog of thymidine triphosphate into nuclear DNA, direct imaging of chromatin in living cells was made possible (7).

II. Core Histone Tails The N-terminal tails of the core histones emerge from the core particle in all directions (1). The lengths of the N-terminal tail domains vary from 16 to 44 amino acids (H3, 44 amino acids; H4, 26 amino acids; H2B, 32 amino acids: H2A, 16 amino acids) (I). Removal of the histone tails by protease digestion does not affect the structural integrity of the nucleosome. Thus, the core histone tails are not required to maintain nucleosome structure. However, removal of the N-terminal tails prevents the chromatin fiber from establishing a stable 30-nm fiber in the presence of Hl histones, showing that the tails are essential for condensation of chromatin. Genetic studies in yeast have revealed the importance of core histone tails in repression (8) and of individual tails in repression or activation of specific genes (9). The core histone tails have been thought to lack structure. However, it has been reported that in the nucleosome core particle, half of the residues in the H3 and H4 tails adopt an a-helical structure when bound to nucleosomal DNA (IO). It has been proposed that the N-terminal tails undergo an induced folding when in contact with other proteins or DNA (II). Invitro chromatin structural studies have identified a spectrum of chromatin structural states, including unfolded, moderately folded, and extensively folded conformations (12,13). The histone tails are involved in the genesis of these chromatin structural states. In the absence of Hl, the H3 and H4 tails are needed for the formation of the moderately folded chromatin conformation, whereas all core histone tails are required to mediate extensive chromatin folding at physiological ionic strength. Either H2A and H2B tails or H3 and H4 tails are needed for interfiber interactions, which result in oligomerization, to occur at physiological ionic strength (14, 15). At low ionic strength the chromatin fibers appear as irregular, threedimensional structures (13). The globular domain of Hl and either the Hl tails or the H3 tail domain are needed to stabilize this three-dimensional arrangement of nucleosomes (13). The other core histone tails cannot substitute for the H3 tail. It is thought that the length of the H3 tails and their position of exit from the nucleosome enable the H3 tails to contribute to the three-dimensional structure of chromatin. The exit of H3 tails from the nucleosome near the linker DNA entry-exit points would position the H3 tail to make interactions with the linker DNA (13). The core histone N-terminal tails are available for interaction with other histones and nonhistone chromosomal proteins. Richmond and colleagues

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

303

observed that the H4 tail (K16 to N25) binds to the H2A-H2B dimer of a neighboring nucleosome. This interaction would contribute to the folding of the chromatin fiber, and it may be involved in nucleosome positioning (1, 8). This region of yeast H4 also has an important role in telomeric silencing (9). Nonhistone chromosomal proteins interact with the tails of H3 and/or H4 to form a transcriptionally competent or repressive chromatin structure. In yeast, the H3 and H4 tails bind to the trans-acting repressors, Sir3 and Sir4, leading to the formation of a transcriptionally repressed chromatin domain (16, 17). The N-terminal tails of yeast H3 and H4 also bind to the global repressor SsnG/Tupl (18). Drosophila Groucho and its mammalian homologs, the transducin-like Enhancer of split proteins, are transcriptional repressors that bind to the N-terminal tail of H3 (19,20). High-mobility-group1 (HMG)14 and -17 proteins bind to nucleosomes and unfold the higher order chromatin fiber, facilitating transcription (21). The C-terminal domain of HMG14, which is involved in chromatin unfolding, binds to the N-terminal tail of H3 (amino acid residues 20 to 50) (21).

III. Transcriptionally Active Chromatin Transcriptionally active DNA is associated with nucleosomes. However, nucleosome structure is perturbed on transcription with RNA polymerase II i Abbreviations: HMG, high mobility group; DNase I, deoxyribonuclease I; IGC, interchromatin granule cluster; HAT, histone acetyltransferase; CBP, CREB (CAMP response element binding protein)-binding protein; PCAF, p300/CBP-associated factor; Esal, essential SAS2-related acetyltransferase; Elp3, elongator protein 3; SAGA, Spt-Ada-GcnS-acetyltransferase; TAF,,, RNA polymerase II-specific TAT-binding protein-associated factor; TRRAP, human transformation/transcription domain-associated protein; TipGO, Tat-interacting protein 60; MORF, monocytic leukemia zinc finger protein-related factor; MOZ, monocytic leukemia zinc linger protein; SRC-1, steroid receptor coativator 1; RACS, receptor-associated coactivator 3; AIBl, amplified in breast cancer 1; TRAM-l, thyroid hormone receptor activator molecule 1; MAPK, mitogen-activated protein kinase; HDAC, histone deacetylase; NuRD, nucleosomeremodeling histone deacetylase complex: SAP, Sin3-associated polypeptide; Rb, retinoblastoma protein; RbAP, Rb-associated protein; N-CoR, nuclear receptor corepressor; SMRT, silencing mediator for retinoid and thyroid hormone receptors; MBD, methyl-CpG-binding domaincontaining protein; RBPl, Rb-binding protein 1; ER, estrogen receptor; RAR, retinoic acid receptor; TR, thyroid hormone receptor: MEF2, myocyte enhancer factor 2; MITR, MEF-2 interacting transcription repressor: PML, promyelocytic leukemia; PLZF, promyelocytic leukemia zinc finger; BTB/POZ, bric-a-brac tramtrack broad complex/pox viruses and zinc fingers; LAZB/BCLG, lymphoma-associated zinc finger 3/B cell lymphoma 6; CHIP, chromatin immunoprecipitation; CDK; cyclin-dependent kinase; SRE, serum response element; TPA, 12. O-tetradecanoylphorbol-13-acetate; EGF, epidermal growth factor; S/MARS, scaffold/matrix-attached regions; NMTS, nuclear matrix targeting sequence; GFP, green fluorescent protein; ARBP, attachment region binding protein.

304

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

(22, 23). Nucleosomes associated with transcribed chromatin have an unfolded structure, exposing a normally buried cysteine at position 110 in H3 (24, 25). N UC1eosomes associated with newly replicated chromatin do not have exposed H3 thiol groups. Transcriptional elongation is required to form the unfolded nucleosome (23). Nucleosomes associated with RNA polymerase II-transcribed genes in mammalian cells cease to be thiol reactive when RNA polymerase II transcription is arrested with a-amanitin. Garrard and colleagues reported that the process of transcription did not cause disruption of chromatin structure, because transcription by T7 RNA polymerase did not disrupt nucleosome structure (22). Why transcription by RNA polymerase II, but not T7 RNA polymerase, perturbs chromatin structure is not clear. However, possible factors include the duration of torsional stress induced by RNA polymerase II transcription (22) and that the process of RNA polymerase II transcription occurs at the karyoskeleton (26, 27). The chromatin of transcriptionally active genes differs from the bulk of the genome in susceptibility to digestion by nucleases, including micrococcal nuclease and deoxyribonuclease I (DNase I) (28). The nucleosome disruption is largely confined to the DNA sequences of the transcribed region. However, the preferential DNase I sensitivity of active genes is not restricted to the coding portion of the gene but extends far upstream and downstream into adjacent nontranscribed DNA sequences before converting to a DNase I-resistant conformation. Typically the DNase I sensitivity of transcriptionally active domains is 2- to 3-fold greater than that of DNase I-resistant chromatin. However, in one study of the p-globin gene domain in 12-day embryonic red blood cells, a DNase I sensitivity of 20-fold was observed. This increase in DNase I sensitivity is due to the preservation of torsional stress within the R-globin domain. Nicking the DNA within the domain by y rays causes a reversal of DNase I sensitivity from 20-fold to 2- to S-fold (29). These studies found that a single nick in the DNA of the domain was sufficient to dissipate the torsional stress within the domain. Extended chromatin loops with decondensed (30-nm fiber) regions, which are presumably transcribed, have been observed in G, phase nuclei of Chinese hamster ovary cells (4). Several studies show that sites of transcription are found at or near the surface of compact chromosome subdomains (5, 30). Interestingly, these decondensed chromatin regions are often found near interchromatin granule clusters (IGCs), ribonucleoprotein structures that serve as storage sites for splicing factors (6, 30). Highly transcribed genes are located near IGCs (30), and highly, dynamically acetylated histones are associated with the chromatin positioned near these structures (6).

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

305

IV. Histone Variants and Modifications Histone variants and particularly their modified isoforms modify higher order chromatin packaging and are tightly linked to the transcription process. In the following sections, histone variants and modifications are reviewed.

A. Core Histone

Variants

Variants of the core histones generate considerable complexity in the histone octamers of the nucleosomes, impacting on chromatin structure. There are multiple forms of H3, H2A, and H2B, which have microheterogeneity in their primary sequence. For example, in the DT40 chicken B cell line there are three variants of H2A, two variants of H3, and four variants of H2B (31). The population of nucleosome histone variants changes during development, differentiation, and lymphocyte activation (32). There is evidence that the histone variants may be of importance structurally and functionally (31, 33). Deletion of an H2B variant in DT40 cells results in changes in the cellular protein profile, suggesting gene-specific effects (34). The amino acid sequence of some histone variants can differ markedly from the major histone. H2A.Z, a member of the histone H2A family, is evolutionarily conserved and appears to be a component of transcriptionally active chromatin (35). The amino acid sequences of histone H2A.Z. and H2A differ by 40%. Deletion of the Drosophila gene coding for the histone variant HBAvD, which is similar to mammalian H2A.Z, is lethal. Similarly, expression of the H2A.Z gene in ?: thennophila and mice is required for viability (33). MacroH2A is a novel histone H2A variant that is expressed in mammalian and avian cells (36). The N-t erminal third of macroH2A is 64O/o identical to H2A, whereas the remainder of the protein has a segment that resembles a leucine zipper, a dimerization motif found in many transcription factors. In female mammalian cells, macroH2A is preferentially located with the inactive chromosome (37). Cse4P and CENP-A are H3 variants that are found in centromeric chromatin of Saccharomyces cerevisiae and mammals, respectively (38, 39). The N-t erminal domains of these H3 variants are unique, but the C-terminal histone folds of Csep4 and CENP-A are about 60% identical to the histone-fold domain of H3.

B. Histone

Hl

Subtypes

The Hl histones are a heterogeneous group of several subtypes that differ in amino acid sequence. The Hl subtypes differ in their abilities to condense DNA and chromatin fragments. Thus, the differential distribution of the Hl histones with chromatin domains may generate chromatin regions with different degrees of compaction (28). Most nuclei typically have more than one Hl subtype. However, there are exceptions. Trout testis, for example,

306

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

has one Hl subtype. The relative amounts of the Hl subtypes vary with cell type within a particular species, as well as among various species. For example, mouse tissues contain various levels of Hl subtypes, Hla, Hlb, Hlc, Hld, Hle, and Hl’. A nomenclature for the mammalian Hl histones has been proposed (40). The expression of the subtypes is differentially regulated throughout development, through the cell cycle, and during differentiation (41). Changes in the expression of Hl subtypes affect gene expression (41, 42).

C. Core Histone Modifications The ability of the core histones to promote chromatin intrafiber and interfiber interactions is modulated by histone modifications. The core histone tails are susceptible to a wide range of postsynthetic modifications, including acetylation, phosphorylation, methylation, ubiquitination, glycosylation, and ADP-ribosylation (Fig. 1). We have known about these modifications since the 1960s but it has only been recently that we have come to appreciate the impact of these modifications on nuclear processes such as transcription. Most modifications occur on the N-terminal basic tail domain, with histone ubiquitination being the exception. In the following sections we will review histone ubiquitination, acetylation, methylation, and phosphorylation and their roles in gene expression.

D. Histone Ubiquitination Histones H2A, H2B, H3 and their variant forms are reversibly ubiquitinated (28). The carboxyl end of ubiquitin, a highly conserved 76-amino acid protein, is attached to the e-amino group of lysine (K119 in H2A; K120 in H2B) (Fig. 1). The linkage of ubiquitin to H3, which has only been observed in elongating spermatids of rat testis, is not known (43). In multicellular eukaryotes, H2A is typically ubiquitinated to a greater extent than H2B (approximately 10% of H2A versus about l-2% of H2B). H2A, H2B, and their variants are also polyubiquitinated, with H2A having the greater levels of polyubiquitinated isoforms. The major arrangement of ubiquitin in polyubiquitinated H2A is a chain of ubiquitin molecules joined to each other by isopeptide bonds to a ubiquitin molecule that is attached to the e-amino group of K119 of H2A (28). Ubiquitinated H2B and to a lesser extent ubiquitinated H2A are associated with transcriptionally active DNA. Ubiquitination of H2B is the only core histone modification that is dependent on ongoing transcription (44). The C-terminal sequence of H2B, but not H2A, is buried in the nucleosome (1). It is thought that the process of transcription disrupts nucleosome structure, exposing the C terminus of H2B to become accessible to the enzymes catalyzing the addition of ubiquitin (22, 44). Another mechanism, which

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

307

does not require ongoing transcription, is by exchange of newly synthesized ubiquitinated H2B and H2A with histones that were in transcriptionally active nucleosomes of Go phase cells (45). The introduction of ubiquitinated H2B into the nucleosome may result in an alteration in nucleosome and/or higher order chromatin structure. In H2B the tyrosine residue positioned next to the site of ubiquitination interacts with H2A N-terminal tail (residues 17 to 20) just before it exits the nucleosome (1). H2B ubiquitination would likely interfere with this interaction.

E. Histone

Acetylation

The core histones are reversibly acetylated at specific lysine residues located in the N-terminal tail domains. With the exception of H2A, the core histones are acetylated at four to five sites. Thus, a nucleosome typically has 26 sites of acetylation. Over the past few years, the role of histone acetylation in the process of transcription has been solidified (46). Histone acetylation is a dynamic process that is governed by the net activity of histone acetyltransferases and histone deacetylases. The process of reversible histone acetylation is not dependent on ongoing transcription. Acetylation occurs at more than one rate, as does the subsequent deacetylation. In mammalian cells and mature avian erythrocytes, one population of core histones is characterized by rapid hyperacetylation (t,,, = 7 to 15 min for monoacetylated histone H4) and rapid deacetylation (t,,, = 3 to 7 min) (47-49). This highly dynamic acetylation-deacetylation is limited to 15% (hepatoma tissue culture cells) of the core histones (47). A second population is acetylated (t,,, = 140 to 300 min for monoacetylated H4) and deacetylated at a slower rate (t,,, = 30 min) (47-49). Approximately 2% of the chromatin of adult chicken immature erythrocytes has core histones that participate in active acetylation and deacetylation. The bulk of the chicken erythrocyte chromatin has core histones frozen in unacetylated and monoacetylated states. There is one rate of acetylation for the immature erythrocyte dynamically acetylated histones (t,,, = I2 min for monoacetylated H4), which is in contrast to two rates of acetylation in mature cells (49). One population of dynamically acetylated H4 is rapidly acetylated to the mono- and diacetylated states, and is slowly deacetylated. Another population of dynamically acetylated H4 is rapidly acetylated to the tetraacetylated state, and the tetraacetylated H4 isoform is rapidly deacety lated (t,,, = 5 min). Thus, the chicken erythroid tetraacetylated H4 isoform is short lived. Of the four core histones, H2B is the most rapidly deacetylated (49, 50). The rates of deacetylation of monoacetylated H4 are also different in adult chicken mature and immature erythrocytes (145 versus 90 mm, respectively) (50). I n c h’ ic k en immature erythroid cells, the dynamically highly acetylated histones appear to be associated with transcriptionally active

308

JAMESR.DAVIEANDVIRGINLAA.SPENCER

chromatin, including transcribed chromatin associated with the karyoskeleton (49- 52). In yeast S. cerevisiae there is a high level of highly acetylated core histones (53). These highly acetylated histones are slowly acetylated and deacetylated (53). Regions of the yeast genome that are transcriptionally silent are associated with hypoacetylated histones (54). In contrast to yeast, approximately 20% of the chromatin of the unicellular green alga Chlamydomonas reinhardtii has core histones that are dynamically multiacetylated with a half-life of about 2 min (55). Alterations at all levels of chromatin structure are invoked by acetylation of the core histones. Acetylation of the histone tails disrupts higher order chromatin folding (56), p romotes the solubility of chromatin at physiological ionic strength, and maintains the unfolded structure of the transcribed nucleosome (57). Analysis of chromatin fibers enriched in transcribed genes and acetylated histones revealed that these chromatin fibers underwent compaction but not oligomerization as the concentration of NaCl was raised to 150 mM. Additional studies show that nucleosomes do not have to be maximally acetylated to prevent higher order chromatin folding. Hansen and colleagues demonstrated that acetylation to 46% of maximal site occupancy was sufficient to prevent higher order folding and stimulation of transcription by RNA polymerase III (58). It has been proposed that acetylation of core histone tails interferes with folding of the N-terminal tail and interactions with proteins and/or DNA, destabilizing higher order chromatin organization (11, 56). These combined effects of histone acetylation on the destabilization of chromatin structure facilitate transcription (58, 59). Histone acetylation can affect the interaction of nonhistone chromosomal proteins with chromatin in at least two ways. First, histone acetylation facilitates the interaction of transcription factors with nucleosomal DNA (60). Partial acetylation of the core histone tails is sufficient to expose nucleosomal DNA for transcription factor binding without displacement of the N-terminal tail domains from DNA (61). S econd, for proteins that interact with the N-terminal tail domain, acetylation may modulate these interactions. For example, acetylation disrupts interactions between the tail domain and the repressor Tupl (18). 1. HISTONEACETYL,TRANSFERASESANDGENEACTIVATION The first histone acetyltransferase (HAT) gene cloned (Tetrahymena nuclear HAT ~55) was found to be homologous to yeast Gcn5, a transcriptional adaptor/coactivator with HAT activity (62). This pivotal discovery told us how HATS were directed to transcribed chromatin regions. Following the discovery that the transcriptional activator, Gcn5, had HAT activity, many other coactivators with HAT activity have since been identified, including ~3001 CBP-associated factor (PCAF), CBP/pSOO, Esal, NuA4, steroid receptor

HISTONE

MODIFICATIONS

AND SIGNALING

309

PATHWAYS

TABLE I HETONE ACETYLTRANSFERASES ANDTHEIR SUBSTRATES~ HAT A (organism; proteins in complex)

Free histone or other substrate

Nucleosomal histone substrate

Gcn5 (yeast, human, Drosophila)

_

Ada (yeast; Gcn5 and Ada proteins)

H3 > H4 (K14 of H3; K8, 16 of H4) _

SAGA (yeast; Gcn5, Ada, and Spt proteins, TAFns, and Tra-1, a homolog of the TRRAP)

_

H3 > H2B

H3 > H4

_

TFTC (mammalian; Gcn5-L, hAda3, hSpt3, hTRAPPm TAF,,s)

H3

H3

PCAF (human; human counterparts of yeast ADA proteins, Spt proteins, human TAF,,s, and PAF400, a 400-kDa protein almost identical to TRRAP)

H3 > H4, TFIIF, TFIIE

H3 -

STAGA (human; Gcn5-L, Spt3, TAF,,31)

H3, H2B

Esal (yeast)

H4 > H3 > H2A (K5 > K8, 12,16 of H4; K14 of H3; K5 of H2A)

NuA4 (yeast, Esal)

_

H4, H2A

H3, H4 > H2A, H2B (K5, 8, 12, 16 of H4), TFIIF, TFIIE, ~53, EKLF, ACTR, SRC-1 H3 > H4 (K14 of H3), TFIIE

H3, H4, H2A, H2B

NuA3 (yeast) CBP/p300

(human)

TAF,,250 (human), Drosophila, yeast)

H3

_

Tip60 (human)

H4 > H3 > H2A (K5 > K8, 12,16 of H4)

-

MORF (human)

H4 > H3 >> H2B (K5 > K8, 12,16 of H4)

H4 > H3

Elp3 (yeast, elongating RNA polymerase II holoenzyme)

H4, H3, H2A, H2B

Not known

SRC-1 (human)

H3 > H4 (Kg, K14 of H3)

H3, H4, H2A, H2B

ACTR (human)

H3, H4 > H2B

H3 > H4

“For references, see histone acetyltransferaseInternet site at http://u?Yw.mdanderson.org/genedev/ Bone/hathome.html.

coactivators, and most recently Elp3 (63; for review see 64, 65) (Table I). We have now come to appreciate the mechanistic connections between histonemodifying activities and the RNA polymerase II machinery. The solution and crystal structures of the HAT domain of T’truhymena Gcn5, yeast Gcn5, and PCAF have been reported (66- 69). Glu-173 in yeast Gcn5, Glu-122 in Tetruhymena Gcn5, and Glu-570 in PCAF are essential

310

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

residues in catalysis (66, 69, 70). It is thought that the structural and functional properties of the catalytic domains of other HATS will be similar. The crystal structure of another domain, the bromodomain, commonly found in HATS has been presented (71). Interestingly, this domain in Gcn5 and PCAF interacts with the N terminus of H3 and H4 and may be involved in targeting the coactivator to chromatin (71, 72). The bromodomain of human Gcn5 binds also to the DNA-dependent protein kinase. The recruited kinase phosphorylates Gcn5, inhibiting HAT activity (73). The substrate specificities of the HATS differ. Further, many of the HATS are in multiprotein complexes, and the substrate specificity of the HAT will vary depending on whether the enzyme is free or in a complex (for review see 73). For example, yeast Gcn5 acetylates free H3, but inefficiently acetylates histones in nucleosomes. In mammalian cells, differentially spliced forms of Gcn5 transcripts generate different Gcn5 isoforms (74). A 98kDa-long form of mammalian Gcn5 has an N-terminal extension not found in yeast Gcn5. Although the presence of the N-terminal extension does not change the substrate specificity of Gcn5 toward free histones, it does enable the mammalian Gcn5 to acetylate nucleosomal H3 (74). The N-terminal extension is similar to that of human PCAF; both proteins bind to other coactivators with HAT activity, e.g., CBP/p300. The C-terminal domain of PCAF’ is similar to that of yeast Gcn5. Yeast Gcn5 will efficiently acetylate histones in nucleosomes only when it is in high-molecular-weight multiprotein complexes such as Spt-Ada-Gcn5acetyltransferase (SAGA) (1.8 MDa) and Ada (0.8 MDa) (75). Although both complexes contain Ada3 and Ada2, which binds to yeast Gcn5, the two complexes are distinct (76). The SAGA complex contains Ada, Spt proteins [SptBO (Ada5), 3, 7, and 81, TAFr,s (TAF,, 90, 68/61, 60, 25123, 20/17), and Tra-1, a homolog of the human transformation/transcription domain-associated protein (TRRAP) (73, 77). TAF,,68, which is homologous to human TAF,,BO and related in structure and sequence to H2B, is required for the integrity, nucleosomal acetylation, and transcriptional enhancing activities of SAGA (77). Yeast TAF,,GO and TAF,,17 have sequence similarities to H3 and H4 and interact with each other as a heterotetramer through a histone fold. These observations suggest the presence of a histone octamer-like structure within the SAGA complex. Similar to yeast SAGA, human PCAF is in large multiprotein complexes consisting of human counterparts of yeast Ada proteins, Spt proteins, human TAF,,s (TAF,, 31, 20/15, and two TAF-like proteins, PAF65cx and PAF65R), and PAF400, a 400-kDa protein almost identical to TRRAP (73). Human TAF,,31 and TAF,,20/15 have the histone fold structure found in histones H3 and H2B, respectively. PAF65 (Yh as similarity to human TAF,,80 and has an H4-like region, whereas PAFR has similarity to the WD40 repeatcontaining TAF,,lOO (78).

HISTONE

MODIFICATIONS

.4ND SIGNALING

PATHWAYS

311

Esal (essential SAS2-related acetyltransferase; the ESAl gene is essential for yeast growth) and Tip60 (Tat-interacting protein 60) are members of the MYST family of proteins (named after founding members, MOZ, YBFBI SAS3, SASB, and TipGO). Esal and Tip60 are HATS with similar substrate specificities (79). Neither protein can acetylate chromatin substrates. The monocytic leukemia zinc finger protein-related factor (MORF) shares significant sequence similarity with MOZ and has HAT activity (80). MORF, TipGO, and Esal share a conserved HAT domain. Unlike Esal and TipGO, MORF acetylated H4 and H3 in oligonucleosomes (80). However, when yeast Esal is assembled into a multiprotein complex called NuA4, the complex will acetylate nucleosomal H2A and H4 (see Table I) (81, 82). Similar to other HATS, NuA4 is associated with Tral (82). A HAT with properties similar to those of yeast NuA4 (about 1.3 MDa) was isolated from Tetruhymena. One difference, however, is that the Tetruhymena HAT (80 kDa) appears to exist as a single protein or as a small multiprotein complex (81). CBP/p300, a coactivator with HAT activity, is an integrator of multiple signaling pathways (Fig. 3). Transcription factors, including hormone receptors, CREB, and fos-jun, loaded onto promoters or enhancers bind directly or indirectly to CBP/p300. Further, CBP is a component of the RNA polymerase II holoenzyme. RNA helicase A binds to CBP and is thought to mediate an interaction between CBP and RNA polymerase II (83). The steroid receptor coactivators SRC-1 and ACTR (and related proteins RAC3, AIBl, and TRAM-l) bind to a variety of nuclear receptors in a ligand-dependent manner. These coactivators associate with CBP/p300 and PCAF. Thus, a ligand-activated nuclear receptor could recruit multiple coactivators with HAT activity (e.g., TipGO, SRC-1, CBP, and PCAF) (84, 85) (Figs. 3 and 4). CBP is a phosphoprotein. CBP is phosphorylated by ERKl, enhancing the HAT activity of CBP in vitro (86). This observation suggests that the activity of CBP may be regulated by the Rasmitogen-activated protein kinase (MAPK) pathway (87, 88). The acetyltransferase activity of several HATS is not limited to histones. PCAF acetylates the nonhistone chromosomal protein HMG-17 (89). Acetylation of HMG-17 reduces the protein’s binding affinity to the nucleosome. CBP/p300 acetylates the four core histones in nucleosomes and a variety of transcription factors (Table I). F or example, CBP acetylates p53 and GATA1 and potentiates the activities of these transcription factors (73). CBP also acetylates other HATS (e.g., ACTR, SRC-l), which disrupts the interaction of the coactivator (ACTR) with the estrogen receptor (90). Transcription factors may recruit one or more HATS. Transcriptional activators with an acidic activation domain (e.g., VP16) or helix-loop-helix proteins with the LDFS motif (e.g., yeast transcription factor Rtg3) recruit SAGA, resulting in localized acetylation and transcriptional stimulation of

312

JAMES FL DAVIE AND VIRGINIA

NUCLEAR RECEPTORS

: PLASMA 1 TGF-beta

stat2 Ets-1 PCAF complex*

Nuclear receptor stat1 ~65 (NF-kB)

CREB Jun Myh Sapla TAX Elk1 SREBP ~65 (NF-kB)

MEMBRANE GROWTH

A. SPENCER

RECEPTORS FACTORS.

EIA SRC-1 Fos JunB MyoD PS3 TFIIB Cdk2 clEBPb%mad2,3 Sv40 IST ACTR (RACI, Ez-1 AlBl,TRAM-1) __: E2F ;;to& RNA helicase PCAF*

FIG. 3. The histone acetyltransferase/coactivator, CPB/pBOO, cointegrates diverse signaling pathways. A variety of sequence-specific transcription factors and coactivators bind to different regions of the CBP (CREB-binding protein)/p300 protein. Kinases such as JAK (Janus kinase), PKA (protein kinase A), MAPK (mitogen-activated protein kinase), and JNK (Jun amino-terminal kinase), once activated, will phosphorylate a variety of transcription factors. The signaldependent transcription factor, once bound to its DNA binding site [e.g., HRE (hormone responsive element, CRE (CREB-responsive element), GAS (interferon-stimulated gene response element), and SBE (Smad b’ m din g e 1ement)], will recruit CBP/p300 (shown by arrows from CBP/p300). CBP/p300 and steroid receptor coactivators have histone acetyltransferase (HAT) activity (*).

nucleosomal substrates in viva and in vitro (75, 91- 93). Importantly, the transcriptional stimulatory activity of the recruited SAGA complex is dependent on its HAT activity (75). Similar to ligand-activated nuclear receptors (e.g., estrogen receptor bound to estradiol; Fig. 4), VP16 appears to recruit several HATS in situ, including Gcn5, PCAF, and CBP/p300 (94). NF-KB recruits several coactivators with HAT activity, including CBP, PCAF, and SRC-1 (95). Interestingly, on phosphorylation by a CAMP-independent

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

313

HISTONE DEACETYLATION

HISTONE

ACETYLATION

SFc-1

/ 0

ESTRADIOL

/

/

F CBp og

FIG. 4. Recruitment of coactivators/HATs and corepressor/HDACs by karyoskeletonbound estrogen receptor. The estrogen receptor is shown associated with the nuclear matrix and its estrogen response element in a nucleosome. When bound to estradiol, the estrogen receptor will recruit coactivator/HATs, resulting in acetylation of histones and activation of transcription. When bound to hydroxytamoxifen, the estrogen receptor will recruit corepressors/HDACs, resulting in histone deacetylation and gene repression. PIC, Preinitiation complex.

protein kinase A, the p65 subunit of NF-KB undergoes a conformational change that forms a bivalent interaction with CBP (96) (see Fig. 3). 2. ROLE

OF

HATS IN TRANSCRIPTION

Coactivator multiprotein complexes with HAT activity can stimulate transcription at several levels, including stimulating the formation of the preinitiation complex and by remodeling chromatin (73, 88, 92, 97-99) (Fig. 4). Several lines of evidence support the view that recruited coactivators with HAT activity acetylate surrounding histones in nucleosomes, leading to the destabilization of higher order chromatin structure and stimulation of transcription (65, 75, 81, 92,100). However, it has been questioned whether the histones are the bona fide in viva substrates of some HATS. The activity of some HATS may be directed toward transcription factors, affecting transcriptional processes. Studies with yeast and Tetrahymena HATS provide evidence that histone acetylation has a role in the transcription process. When recruited to a promoter, yeast HATS (SAGA, NuA4, NuA3, Ada) and Tetrahymena NuA4 facilitated transcription in vitro from nucleosomal, but not naked DNA, templates (75, 81, 92, 100, 101). Importantly, the HAT stimulation was observed only when acetyl CoA was present.

314

JAMES FL DAVIE AND VIRGINIA

A. SPENCER

Acetylation of chromatin components can activate or repress transcription. The activity of the interferon-R (IFN-R) enhanceosome is regulated, in part, by acetylation. The enhanceosome consists of NF-KB, IRFl, ATFB/ c-Jun, and HMGI(Y), an essential architectural protein involved in the stereospecific assembly of this complex. Once assembled, the complex effectively recruits CBP (102), which then acetylates H3 and H4 in neighboring nucleosomes, resulting in the remodeling of chromatin and the recruitment of the RNA polymerase II holoenzyme (103). The net result is the turning on of IFN-R gene expression. However, CBP can also acetylate HMGI(Y) at a site important in DNA binding. The result of HMGI(Y) acetylation is disruption of the enhanceosome and the turning off of IFN-R gene expression (104). 3. HISTONE DEACETYLASE AND GENE REPRESSION In 1996, the cloning of mammalian histone deacetylase 1 (HDACl) revealed that it was related to yeast transcription regulator RPD3, providing a link between transcription regulation and histone deacetylation. Several HDACs have since been reported, including HDACB (the mammalian homolog of RPD3) an d mammalian HDACS (reviewed in 46). HDACs, bacterial acetoin utilization proteins, and acetylpolyamine amidohydrolases appear to be members of an ancient protein superfamily. These proteins share nine blocks of sequence similarity, with 20 amino acids being invariant in these alignments (105). S ome of these conserved amino acids could be involved in binding a metal atom, e.g., zinc; there is evidence that HDACl is a metalloenzyme (106,107). The crystal structure of the HDAC catalytic core based on the hyperthermophilic bacterium Aqu@r aeolicus HDAC was reported (108). Th e active site consists of residues that are conserved, a zincbinding site, and two Asp-His charge relay systems. In contrast to HATS, recruitment of HDACs can lead to repression. It is important to note, however, that chromatin regions engaged in transcription are associated with dynamically acetylated histones (64). Thus, both HATS and HDACs are recruited to these regions. When the balance of activity of these two enzymes favors deacetylation, the chromatin region will take on a repressive higher order structure. The HDACs have been categorized into two classes. The first class consists of yeast histone deacetylases Rpd3, HOSl, and HOS2 and mammalian HDACs, HDACl, HDAC2 (the mammalian homolog of yeast RPD3), and HDACS (64). Class 2 consists of yeast HDAl and mammalian HDAC4 (HDAC-A), HDAC5 (mHDA1, NY-CO-g, HDAC-B), and HDACG (mHDA2) (109-111). Mammalian HDACl and HDACB, but not HDACS, are in large multiprotein complexes, e.g., mSin3A and NuRD (Fig. 5). The mSin3A complex contains mSin3, N-CoR or SMRT (corepressors), SAP18, SAP30, RbAp48,

HISTONE

MODIFICATIONS

AND SIGNALING

315

PATHWAYS

DNA

REMODELING AND HISTONE DEACETYLATION

NUCLEAR RECEPTORS

MAD-MAX

FIG. 5. The HDAC complexes are involved in several signaling pathways. The Sin3A and NuRD complexes may be recruited to specific sites by nonliganded hormone receptors to hormone response elements or by Mad-Max to E-box-related sequences. MeCP2 and MBD2 recruit the HDAC complexes to methylated DNA, resulting in silencing. ARBP/MeCP2 may also recruit the HDAC complex to scaffold/matrix-associated regions. Rb, YYL, and several other karyoskeleton-bound transcription factors could recruit HDAC complexes to the karyoskeleton.

RbAp46, and c-Ski (112). Another complex, the nucleosome remodeling histone deacetylase complex (NuRD), consists of N-CoR, MTA2 (highly related to metastasis-associated protein MTAl), Mi2, RbAP46/48, and MBD3 (methyl-CpG-binding d omain-containing protein), and has both ATPdependent chromatin remodeling and HDAC activities (113). HDAC3 and the class II HDACs are not found in the mSin3A and NuRD complexes (111). However, HDAC4 and HDAC5 bind to HDAC3 (111). Like human HDACB, yeast Sin3 and RPD3 are in a large 2-MDa protein complex. The other yeast HDACs, HDAl, HOSl, HOS2, and HOS3, are also in higher molecular mass complexes (114). Class I and class II histone deacetylases can deacetylate the four core histones. However, in vitro studies show that HDACs have site preferences. For example, yeast HOS3 preferentially deacetylates yeast histones at K5 and K8 of H4, at K14 and K23 of H3, at K7 of H2A, and at Kll of H2B (114). Substrate preference is regulated by components of the multiprotein complexes. For example, free avian HDACl preferentially deacetylates H3, but not nucleosomal H3. HDACl in a multiprotein complex associated with the karyoskeleton preferentially deacetylates free H2B and will deacetylate histones in nucleosomes (107). NuRD has both ATP-dependent chromatin remodeling

316

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

and histone deacetylase activities (115-218) (Fig. 5). ATP stimulation of deacetylation of chromatin templates by NuRD varied from no stimulation to about threefold. HDAC or HDAC complexes are recruited to specific genomic sites by transcription factors (repressors). HDACs 1, 2, and 3 bind to YYl, whereas Rb and E2F form a complex with HDACl (64, 65). Evidence has been presented that the IXCXE motif located in the C-terminal region of HDACl and HDACS associates with the Rb “pocket” domain of hypophosphorylated Rb, ~107, and ~130 (119, 120). More recent studies, however, provide evidence that binding of HDACs 1, 2, and 3 to the pocket proteins (Rb, ~107, ~130) requires an intermediary protein, RBPl (121). The recruitment of the E2FRb-HDACl complex is partly responsible for the repression of the cyclin E promoter in G, phase of the cell cycle. Phosphorylation of Rb by CDK4 and CDK6 results in the dissociation of Rb from HDACl and E2F (122). The promoter-associated E2F may now associate with CBP/p300, recruiting a HAT activity that stimulates transcription of the cyclin E gene. HDACl can also bind to the C-terminal domain of Spl, repressing this transcription factor’s activity (123). E2Fl binds to the Cterminal domain of Spl and displaces HDACl. Binding sites for Spl and E2Fl are found on S phase-specific promoters. Thus, there is an interesting relationship between Spl, E2F, pocket proteins, and histone deacetylases in the repression of these growth-regulated genes (123). The methyl-CpG-binding protein 2 (MeCP2) recruits the mSin3A complex, and methyl-CpG-binding domain-containing protein (MBDS) binds to the NuRD complex, providing mechanisms for coupling DNA methylation and histone deacetylation in gene silencing (113,124). Interestingly, mutations in the X-linked gene encoding MeCP2 were found to be the cause of some cases of Rett syndrome, a progressive neurodevelopmental disorder (125). Several signal transduction pathways regulate the recruitment of the HDAC corepressor complex to specific loci. The Sin3A-N-CoR-HDACl/B complex, for example, is recruited by unliganded nuclear receptors, the Mad family of basic helix-loop-helix zipper proteins, and ~53 (64, 65, 126) (Fig. 5). SAP30, which binds to mSin3 and N-CoR, is required for N-CoR/mSin3mediated repression of hydroxytamoxifen-bound estrogen receptor (ER) and homeodomain protein Rpx, but not unliganded retinoic acid receptor (RAR) and thyroid hormone receptor (TR) (127, 128) (Fig. 4). Interestingly, microinjection of anti-N-CoR or anti-SMRT IgG into cells can convert hydroxytamoxifen-bound ER and RU486-bound progesterone receptor from repressors to activators (129). Further, there is intriguing evidence that decreasing the levels of N-CoR can lead to tamoxifen-resistance in breast cancer (129). BRCAl was shown to bind to RbAP46, RbAp48, HDACl, and HDAC2, suggesting that BRCAl may be a component of one or more of the HDACl/ 2 multiprotein complexes (130). BRCAl functions as a transcriptional coac-

HISTONE MODIFICATIONSAND SIGNALINGPATHWAYS

317

tivator that associates With the RNA polymerase II holoenzyme, and is also involved in transcription-coupled DNA repair. Thus, BRCAl may recruit HDAC complexes to sites of transcription and repair. Transcription factors recruiting the class II HDACs are becoming known. Human myocyte enhancer factor 2 (MEFS) recruits HDAC4, resulting in the repression of MEF2 transcriptional activation. MEF2 also binds to the corepressor MEF-2 interacting transcription repressor (MITR), which shares sequence similarity with class II HDAC family members. MITR binds to HDACl; thus, MEF2 is able to recruit both class I and class II HDACs (131). Similar to the situation with E2F1, MEF2 associates also with the coactivatoriHAT CBP/pSOO. Thus, the association of MEF2 with corepressors or coactivators governs the transcriptional response of MEF2 (132). 4. ROLE OF HDACs IN TRANSCRIPTION HDAC has a principal role in transcription repression (64). Once recruited to a specific promoter, HDAC deacetylates histones in nucleosomes, leading to the condensation of chromatin (65). However, acetylated HMG proteins and transcription factors may also be targets of the HDAC activity. The HDAC corepressor complex can also repress transcription by mechanisms that do not require deacetylation. N-CoR and mSin3A of the HDAC complex interact with components of the preinitiation complex. Thus, the HDAC complex may interfere with the generation of a functional initiation complex (Fig. 4). 5. HATS, HDACs, AND CANCER In humans loss of one allele of CBP is the underlying defect in Ruberstein-Tabyi syndrome. Patients with this syndrome are more prone to cancer, consistent with the suggestion that CBP/pSOO may function as a tumor suppressor. Further somatic translocations involving the CBP gene are found in various types of hematological malignancies (133,134). PML-RARcx, PLZF-RARo, and AML-l-ETO, oncoproteins in acute promyelocytic leukemia (PML) generated by chromosomal translocations, recruit SMRT-mSinSA-HDACl and N-CoR-mSin3A-HDAC1/2 complexes (135-137). The SMRT-mSin3A-HDACl complex is recruited by the BTB/ POZ domain found in the oncoprotein LAZ3/BCL6 (46). The recruitment of HDACl is crucial to the transforming potential of these oncoproteins. Inhibiting the HDAC activity with new-generation HDAC inhibitors appears to be a promising approach to the treatment of these cancers (135, 138). 6. LOCATION OF ACETYLATED HISTONES The exact role of histone acetylation in transcription has been questioned for several decades. Researchers have accumulated substantial amounts of indirect evidence from chromatin fractionation and pulse chase labeling

318

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

experiments to show a relationship between this modification and transcription. Evidence elucidating the direct involvement of histone acetylation in transcription has also been obtained by the chromatin immunoprecipitation (CHIP) assay. In this assay, cells are fixed with formaldehyde, sonicated, and then regions of DNA associated with histones are isolated by immunoprecipitation with antibodies recognizing specific modified histone isoforms (139, 140). The use of this assay with an antibody that recognizes the Eacetyllysine residues of all acetylated core histones showed that histone hyperacetylation corresponds to DNase I-sensitive, transcriptionally active and competent but not inactive regions of the chicken R-globin domain (52). Such a finding suggested that histone acetylation functions to maintain an open chromatin structure within transcriptionally active or competent genes, thereby increasing the access of transcription factors to these specific DNA target sequences. However, the antibody used in this study detected all acetylated histone isoforms. Therefore, the results of this study could not identify the role of specific acetylated histone isoforms in transcription. Elucidating the role of acetylated histone isoforms in transcription is important due to the recent discovery that several transcriptional cofactors containing intrinsic HAT or HDAC activity are targeted to specific histone substrates. Realizing this, several investigators have produced antibodies specifically directed against different acetylated histone isoforms such as H4 acetylated at K5, KS, K12, and K16 (54) and H3 acetylated at K9 and K14 (141). Antibodies recognizing specific H4 acetylation sites have also been produced to further understanding of the role of site-specific acetylation in transcription (142). The use of these various antibodies in CHIP assays has lead to several important observations. In one study, antiacetylated H3 and H4 antibodies were used in a CHIP assay to further define the role of insulator DNA sequences in transcription. The results of this study showed that insulator DNA sequences prevent histone deacetylation, and a loss of transcriptional activity in transgene expression (143). In addition, the CHIP assay has shown that transcriptional repression by proteins such as the Ikaros lineage-determining factor correlates with a reduction in H3 acetylation at the promoter of genes involved in lymphocyte development (144). Likewise, CHIP studies have shown that transcriptional silencing is correlated with a decrease in H3 and H4 hyperacetylation in yeast (54, 145). As well, the CHIP assay has shown that the nature of active and inactive gene structure in X. lads is such that transcriptionally active, somatic genes are packaged with hyperacetylated H4, whereas transcriptionally inactive oocyte genes are packaged with hypoacetylated H4 (146). Similarly, a study on transgene expression in zebrafish demonstrated a correlation between H4 acetylation and transgene expression (147). In another study, antiacetylated H3 and H4 antibodies were used to immunoprecipitate endogenous DNA sequences associated with the estrogen

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

319

receptor when bound to estrogen and antiestrogen ligand (90). This assay showed that the binding of estrogen to its receptor caused an induction of histone hyperacetylation at the promoter of several estrogen receptor target genes, whereas the binding of the antiestrogen tamoxifen to the estrogen receptor caused a dramatic reduction in histone acetylation at these same target genes (90) (Fig. 4). Contrary to the acetylation events caused by ligand-activated estrogen receptor, the CHIP assay with antibodies to antiacetylated H4 and H3 has shown that genes (e.g., thyroid-stimulating hormone o) negatively regulated by the TR require the recruitment of an unliganded TR along with transcriptional corepressors for transcriptional stimulation and histone acetylation (148). As well, binding of ligand to the TR causes transcriptional repression along with a decrease in histone acetylation at the promoter of target genes (148). Thus, the CHIP assay has recently become an important tool for studying in vivo hormone-receptor action in transcription. The CHIP approach has also been used to determine the involvement of histone acetylation in events related to cell cycle and development. Immunoprecipitation of DNA sequences associated with antiacetylated H3 antibodies has shown that the retinoblastoma protein represses the transcription of cell cycle genes containing E2F sites by recruiting HDACs to the promoter, where they are able to deacetylate associated H3 (149). In addition, the FMRl gene, a gene mutated in people with fragile X mental retardation, shows higher levels of H3 and H4 acetylation in cells from normal individuals when compared to cells from fragile X patients (150). The CHIP assay has also shown a link between signal-regulated H4 hyperacetylation events and transcription (151). More specifically, the hyperacetylation of H4 positioned along a reporter gene is induced by extracellular signals such as growth factor, stress, and cdc42 (151). Th e evidence accumulated from various CHIP studies combined with the identification that several transcription factors contain intrinsic HAT activity suggests that histone acetylation is directly and more actively involved in the transcriptional process than previously postulated. The CHIP assay has also become a useful approach for fine-mapping the distribution of hyperacetylated histones along a gene. To fine-map histone isoforms to particular DNA sequences, nuclear lysates are extensively sonicated into fragments of approximately 350- 500 bp in average size (141,152). This allows the resolution of histone hyperacetylation over a range of 1 to 3 nucleosomes. Following this, acetylated histone isoforms along with their cross-linked DNA sequences are immunoprecipitated and the DNA is analyzed for particular sequences using quantitative polymerase chain reaction (PCR). To date, the results of several fine-mapping studies of histone hyperacetylation along transcriptionally active DNA regions show that the pro-

320

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

moter region of a transcriptionally active gene is enriched in hyperacetylated histones H3 and H4, whereas the coding regions and regions upstream of the promoter are depleted in hyperacetylated histones (103, 141, 153). In yeast, the Sin3-Rpd3 HDAC complex causes histone deacetylation over a range of 1 to 2 nucleosomes within the promoter of a repressed gene (141). Similarly, transcriptional activation of the human interferon gene by virus infection induces histone hyperacetylation over 2 to 3 nucleosomes within the promoter region (103). Furthermore, the yeast Gcn5 HAT complex acetylates histones only in the HO gene promoter (153). Based on these studies, researchers postulate that HATS and HDACs undergo a targeted recruitment to the promoter. When at the promoter, these enzymes modify the acetylation state of the histones within a limited number of nucleosomes positioned on the promoter. This, in turn, causes localized perturbations in chromatin structure that enable transcription factors to gain access to their target DNA sequences, Although the studies mentioned above provide a strong argument for promoter-targeted histone acetylation, conflicting data exist to suggest that HATS and HDACs are recruited to both the promoter and coding regions of transcriptionally active genes. First, a 60-kDa subunit of the elongator/RNA polymerase II holoenzyme referred to as Elp3 has been identified as a HAT and is able to acetylate all four core histones in vitro (63). Second, histone hyperacetylation is required to maintain the transcriptionally active nucleosome in an open conformation for transcriptional elongation (57). As well, when an antibody recognizing all acetylated histone isoforms is used in a CHIP assay, analysis of the DNA sequences associated with the immunoprecipitated histones shows a widespread distribution of histone hyperacetylation along the transcriptionally active c-myc and B-globin genes (52, 98). Thus, the distribution of acetylated histones or acetylated lysine residues within a gene may not be uniform (98). More specifically, the hyperacetylation observed throughout the c-myc and B-globin coding regions may represent hyperacetylated isoforms of H2B, as well as the hyperacetylated isoforms of H3 and H4 that are not recognized by the currently used antiacetylated H3 and H4 antibodies. This contradiction in experimental evidence also suggests that a cell may contain two types of HATS with respect to the transcriptional process: those involved in transcriptional initiation, and those involved only in transcriptional elongation. HATS required for the initiation process would most likely function to alter the chromatin structure of the promoter, making the DNA more accessible to transcriptional initiation factors, whereas HATS required for elongation would increase the accessibility of elongation factors to the DNA within coding regions. In support of this, the ~300 HAT interacts specifically with the form of RNA polymerase II involved in transcriptional initia-

HISTONE

MODIFICATIONS

AND

SIGNALING

PATHWAYS

321

tion, whereas the PCAF is associated only with the RNA polymerase II form involved in transcriptional elongation (154). Cell type and the type of gene studied are also important factors that influence the distribution of acetylated histone isoforms within a cell’s genome. The HDAC Rpd3 preferentially deacetylates H4 at K5 (155). This preference, however, is evident for only a select number of genes. In addition, the types of HATS and HDACs that target histones within the promoter of a gene can vary with cell type (156). Th us, the differences in cellular context between cell types most likely have a significant influence on the types of HATS and HDACs recruited to specific promoters. Considering these studies, it is evident that the CHIP assay has made a significant contribution toward our understanding of the involvement of histone acetylation in various nuclear and cellular events. F. Histone Methylation The core histones H2B, H3, and H4 are modified by methylation. Heat shock of Drosophila melanogaster Kc cells induces methylation of H2B at an N-terminal proline residue (157). With the exception of plants, H4 is methylated at K20 (158) (Fig. 1). K20 of mammalian H4 is 70-100% methylated at this site. H3 may be methylated at K4, K9, K27, and K36, but the site utilization varies. Mammalian H3 is typically methylated at K9 and K2 7, being modified to 3 5 and 70 - lOO%, respectively (159,160). Cycad, Chlamydomonas, and Tetrahymena H3 are methylated at K4 to 20,81, and 50010, respectively. Cycad and Chlamydomonas H3 are also methylated at K9, K2 7, and K36 but to varying extents (e.g., K9, 100 versus 16%; K27, 50 versus 25%). Tetrahymena H3 is methylated at K27 (40%) but not at K9 or 36. Chick H3 is methylated at K9, K27, and K36 to 20, 100, and 20%, respectively. Acetylated isoforms of H3 and H4 are often the targets of ongoing methylation (161-164). In chicken immature erythrocytes, rapidly acetylated and deacetylated H3 and H4 are selectively methylated, whereas in HeLa cells dynamically acetylated H3, but not H4, is methylated (164, 165). H4 that is slowly acetylated and deacetylated is methylated in HeLa (164). The processes of histone methylation and dynamic acetylation are not directly coupled: neither modification predisposes H3 or H4 to the other (165). The association of dynamically acetylated histones with transcribed chromatin suggests that methylated H3 and (in some cases) methylated H4 are bound to transcriptionally active DNA (161,164). Histone methylation is a relatively stable modification with a slow turnover rate. However, there is evidence of methyl group turnover for HeLa H3 (164). It remains to be shown if this histone demethylase activity is present in transformed but not normal cells. Very little is known about the

322

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

histone methyltransferases. Histone-lysine methyltransferase is a chromatinbound enzyme that catalyzes the addition of methyl groups onto the e-amino groups of chromatin-bound H3 and H4 (161).

G. Histone Phosphorylation

and Mitosis

The core histones and Hl undergo phosphorylation on specific serine and threonine residues. Hl can be phosphorylated on Ser/Thr residues on the Nterminal and C-terminal domains of the molecule, and H3 can be phosphorylated on SerThr residues on its N-terminal domain. The phosphorylation of both Hl and H3 is cell cycle dependent, with the highest level of phosphorylation of both histones occurring in M phase. In G, phase of the cell cycle, the lowest number of Hl sites is phosphorylated, and there is a gradual increase in the number of sites phosphorylated throughout S and G, phases of the cell cycle. In M phase, when chromatin is highly condensed, the maximum number of sites is phosphorylated. The strong correlation between highly phosphorylated Hl and chromatin condensation at mitosis leads to the assumption that Hl phosphorylation drives mitotic chromatin condensation; however, chromatin condensation can occur in the absence of Hl phosphorylation (166). Hl phosphorylation destabilizes chromatin structure and weakens its binding to DNA. Therefore, Hl phosphorylation may lead to decondensation of chromatin and access of the DNA to factors involved in transcription and replication in G, and S and to condensing factors present in mitosis (167 and references therein). Studies on H3 phosphorylation during mitosis have revealed that Ser-10 phosphorylation of H3 is correlated with both mitotic and meiotic divisions in Tetrahyma micronuclei (168), and that phosphorylation at this site is required for proper chromosome condensation and segregation (169). In mammalian cells, mitosis-specific phosphorylation of H3 on Ser-10 initiates primarily within percentromeric heterochromatin during late G, and spreads in an ordered fashion throughout the condensing chromatin and is complete just prior to the formation of the prophase chromosomes (170). Phosphorylation of H3 at Ser-10 weakens the association of the H3 tail to DNA, which may promote the binding of factors that drive chromatin condensation as cells enter mitosis (171). Phosphorylation of H3 at Ser-28 was shown to occur in mammalian cells during early mitosis, suggesting that H3 phosphorylation at sites Ser-10 and Ser-28 are involved in mitotic chromosome condensation (172).

H. Histone Phosphorylation,

Transcription,

and Signal Transduction Studies by Lee and Archer show an involvement of Hl phosphorylation in gene transcription. Inactivation of the mouse mammary tumor virus (MMTV) promoter is associated with dephosphorylation of Hl, and reacti-

HISTONE

MODIFICATIONS

AND SIGNALING PATHWAYS

323

vation of the promoter is associated with rephosphorylation of Hl (173). Further, mouse Hlb phosphorylation is dependent on ongoing transcription and replication processes; the inhibition of these processes may alter accessibility of Hlb to the Hlb kinase, which would result in decreased levels of phosphorylated Hlb (174). The modification of this mouse histone is unique in this regard. No other histone modification has been shown to be dependent on these processes. Oncogene-transformed mouse fibroblasts have a more decondensed chromatin structure than do parental cell lines (167). We found that levels of phosphorylated Hl were elevated in oncogene (TYZ.S, j&s, mos, myc) and aberrantly expressed MAPKK (MEK) transformed mouse fibroblasts, which have elevated activities of MAF’K (ERKl and ERK2) (167) (Fig. 6). Subsequently,

ruj

+ Histone ..*

cyclin E-CDK2 HI kinase

DECONDENSATION FIG. 6. Activation of the Ras-MAPK signaling pathway increases the level of phosphorylated Hl. Mitogen-activated protein kinase kinase (MAPKK) and mitogen-activated protein kinase (MAE’K) are also called MEK and ERK, respectively. Activation of ERK leads to the increased activity of cyclin E-&2, an Hl kinase. Hl phosphorylation results in decondensation of chromatin.

324

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

Weinberg’s group found that Rb-deficient human fibroblasts have increased levels of phosphorylated Hl and a relaxed chromatin structure (175). These authors had evidence that cyclin E/cyclin-dependent kinase 2 (cdk2) was directly involved in increasing the levels of phosphorylated Hl. Our unpublished results also suggest that elevated cyclin E-associated Hl kinase (cdk2) activity resulting from persistent activation of the Ras-MAPK pathway is responsible for the increased level of phosphorylated Hl in oncogene-transformed mouse fibroblasts (D. N. Chadee, C. P. Peltier, and J. R. Davie, unpublished observations). Persistent activation of the Ras-MAPK signaling pathway also results in elevated levels of phosphorylated (Ser-10) H3 in oncogene-transformed mouse fibroblasts (Fig. 7). The remodeling of chromatin structures resulting from increased H3 phosphorylation may contribute to aberrant gene expression (176). Though most studies on H3 phosphorylation have focused on the phosphorylation that occurs in mitosis, H3 phosphorylation also occurs in G,.

c-fos

Gene GO Phase

Gl Phase

FIG. 7. Regulation of the expression of the immediate-early C-$X gene. In Go phase cells, the serum response element (SRE) of the c-j& promoter is loaded with transcription factors SRF and Elkl, which recruit the coactivator/HAT CBP Stimulation of the cells with growth factors or phorbol esters results in the activation of the Ras-MAI’K pathway and activation of MAI’K/ ERK. MAPK phosphorylates the SRE-bound transcription factors, CBP and MSKl, an H3 kinase. CBP would acetylate the nucleosomal histones and MSKl would phosphorylate H3. The net result of these modifications would be the decondensation of chromatin and the release of the elongation block.

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

325

Phosphorylation of H3 has been implicated in the establishment of transcriptional competence of immediate-early response genes. H3 is rapidly phosphorylated when the Ras-MAPK pathway of serum-starved cells is stimulated with growth factors and phorbol esters. H3 phosphorylation is concurrent with the transcriptional activation of the early response genes c-fos and c-&n (176). We demonstrated that the newly phosphorylated H3 is located in numerous small foci scattered throughout the interphase nuclei of 12-0-tetradecanoylphorbol-13-acetate (TPA)-treated cells; the foci were found outside condensed chromatin regions (176). Highly acetylated H3 is also observed in similarly positioned numerous small foci, which agrees with the observation that H3 phosphorylation is restricted to a small fraction of H3 histones that are dynamically highly acetylated (6). Using the CHIP assay, we provided direct evidence that the newly phosphorylated H3 is associated with induced c-fos and c-myc genes (176). The observation of numerous foci of newly phosphorylated H3 in TPA-treated cells suggests that many other induced genes, such as those described by Brown and colleagues, are associated with phosphorylated H3 (177). The c-j& gene is transcribed in quiescent cells; however, elongation of the gene is blocked approximately 100 nucleotides from the site of initiation. Stimulation of the Ras-MAPK pathway results in the release of this block in elongation (Fig. 7). Activation of the MAPK signaling pathway results in the phosphorylation and activation of transcription factors, such as the Ets transcription factor family (178). The c-fos p romoter serum response element (SRE) is continuously occupied by SRF and Ets proteins of the TCF family; both of these factors are targets of signaling pathways (178). p62 TCF and Elk-l (members of the Ets family) are direct targets of the Ras-MAPK signaling pathway and are phosphorylated by MAPK (p42/p44; also named ERKl/2) (179, 180). Phosphorylation of the Ets proteins is thought to contribute to the induced expression of the C-$X gene. CBP binds to both Elk-l and SRF (179). Elk-l phosphorylation results in a functional interaction between Elk-l and CBP (181). Th us, these transcription factors recruit a coactivator with HAT activity. It is possible that phosphorylation and, likely, acetylation of H3 associated with the c-fos gene are also involved in release of the elongation block, by allowing the chromatin fiber to be less compact. Consistent with this hypothesis, the c-fos chromatin becomes more DNase I sensitive following activation of the Ras-MAPK pathway (182). As the H3 tail contributes to the folding and interassociation of chromatin fibers, modification of the H3 tail by acetylation and phosphorylation may destabilize higher order compaction of the chromatin fiber and contribute to maintaining the unfolded structure of the transcribing nucleosome. The steady-state level of H3 phosphorylation is dependent on a balance of phosphatase and kinase activities in the cell. Protein phosphatase 1 appears

326

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

to be the H3 phosphatase (176). Allis and colleagues have presented evidence that the activity of Rsk2, a member of the pp90 rsk kinases, is required for the mitogen-stimulated phosphorylation of H3 (183). Coffin-Lowry patients have a mutation in the Rsk2 gene. Fibroblasts from these patients do not exhibit epidermal growth factor (EGF)- or TPA-stimulated phosphorylation of H3 and, interestingly, growth factor-induced expression of the c-fos gene is severely impaired. However, Mahadevan and colleagues presented evidence that MSKl is the H3 kinase (184). Both Rsk2 (MAPKAP kinase-1B) and MSKl are members of a subfamily of MAPK-activated protein kinases with two distinct protein kinase domains. MSKl, but neither ERKs nor Rsk2, is inhibited by H89, a protein kinase inhibitor. H89 inhibits TPA- and EGFstimulated H3 phosphorylation and expression of c-fos and c-&n (184). In our in vitro studies, we found that Rskl and Rsk2 efficiently phosphorylated H2B, which has two Rsk consensus sequences (RXXS), but failed to phosphorylate H3, which has an RXS sequence at Ser-10 and Ser-28 (D. N. Chadee, C. P Peltier, I. S. Sterlkov, J. R. Davie, unpublished observations). Our data concur with those of Mahadevan and colleagues that Rsk2 is not the physiological relevant TPA- or EGF-stimulated H3 kinase.

V. Karyoskeleton and Organization of Chromatin The chromatin fiber is organized into loops such that the base of the loop is attached to proteins of the karyoskeleton (also called nuclear matrix and nucleoskeleton) (28, 185, 186). The DNA sequence binding to the karyoskeleton is called the matrix attachment region (MAR) or scaffold attachment region (SAR). The S/MARS delineate the loop domain in different cell types regardless of the transcriptional activity of the gene(s) within the domain. The loop domain containing transcribed genes has a less condensed, more DNase I-sensitive structure than that of loops with repressed genes. For transcriptionally active chromatin loops, the boundaries of the several DNase I-sensitive gene domains comapped with S/MARS (28). A comparison of the DNA sequences of S/MARS shows that they do not share extensive sequence homology; however, S/MAR DNA sequences have high bending potential and may act as topological sinks (187). The karyoskeleton is the nuclear structure that is present following the salt extraction of nuclease-digested nuclei (188). It is the dynamic structural framework of the nucleus composed of a meshwork of core filaments linked to the nuclear lamina proteins. The diameters of core filaments of the internal matrix are similar to diameters of cytoskeletal intermediate filaments (188, 189). The composition of the core filaments is not yet known (190). However, there is an interesting observation that NuMA is capable of forming a scaffold (191).

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

327

There has been much controversy over the existence of the karyoskeleton. The rather harsh methods (high salt extraction of nuclease-digested nuclei) originally used to isolate karyoskeletons were criticized as generating artifacts. However, additional studies have defused these concerns. Karyoskeletons isolated by 0.2 M ammonium sulfate extraction of DNase I-digested nuclei maintained the organization of nuclear components such as transcription sites and nuclear speckles, which are involved in storage of RNA splicing factors (27). Further formaldehyde fixation of cells or nuclei followed by isolation of karyoskeletons yielded structures similar to unfixed karyoskeletons (192). More gentle methods in which fragmented chromatin is removed from nuclei at physiological ionic strength yielded karyoskeleton structures essentially identical to those prepared by more harsh methods (26). A study of major importance was published by Hendzel and colleagues, in which they visualized the karyoskeleton fiber network in cells (193).

VI. Karyoskeleton and Transcription Factories The karyoskeleton has a central role in transcription (26, 27). There are about 2000 transcription sites in a HeLa nucleus, with each of these sites containing about 25 active polymerases and greater than 10 active genes (194). These sites are referred to as transcription factories. The transcription machinery is associated with the karyoskeleton. Using a gentle extraction method, Roeder, Cook and colleagues presented evidence that active RNA polymerases are attached to the karyoskeleton (26). Cells were lysed with saponin in a physiological buffer, and the release of transcription factors and RNA polymerase II into soluble and insoluble cellular fractions were analyzed. RNA polymerase II was observed in both fractions, with unengaged RNA polymerase II (predominantly form HA) being found in the soluble fraction and engaged RNA polymerase II (predominantly for 110) being present in the insoluble (karyoskeleton) fraction.

VII. Transcriptionally Active Chromatin and the Karyoskeleton Transcribed and nontranscribed sequences are precisely compartmentalized within the nucleus (28, 195). Actively transcribed, but not inactive, chromatin regions are immobilized on the nuclear matrix by multiple dynamic attachment sites. When histones are removed by high salt, loops of DNA are seen emanating from a central nuclear skeleton, forming a halo around this nuclear structure. Transcriptionally inactive genes are found in the halo, whereas DNA loops with transcriptionally active genes

328

JAMES FL DAVIE AND VIRGINIA

A. SPENCER

remain associated with the residual nucleus (186, 195). The transcription machinery, transcription factors, and nuclear enzymes are thought to mediate the dynamic attachments between transcribing chromatin and nuclear matrix.

VIII. Transcription Factors and the Karyoskeleton The karyoskeleton has a role in the expression of genes by concentrating a subset of transcription factors at specific nuclear sites (196, 197). Transcription factors associated with the karyoskeleton include estrogen receptor, mutant p53, YYl, AML-1, Spl, and Rb (198-201). The karyoskeleton targeting sequence (nuclear matrix targeting sequence, or NMTS) has been identified for several factors, including AML-1, YYl, glucocorticoid receptor, and Pit-l (196,201-204). At present there is no consensus sequence or structure for the NMTS of these factors. For AML-1, the factor’s NMTS functions as a transactivation domain and targets AML-1 to karyoskeleton sites containing a hyperphosphorylated active form of RNA polymerase II (196). The association of a transcription factor with the karyoskeleton is dynamic (201,204), and the equilibrium between karyoskeleton bound and unbound states may be influenced by protein modification. For example, hypophosphorylated Rb associates with the karyoskeleton only during early G, phase of the cell cycle. Throughout the remainder of the cell cycle Rb becomes phosphorylated and is dephosphorylated in late mitosis. Thus, the association of Rb with the karyoskeleton may be determined by phosphorylation. Identification of the karyoskeleton acceptor for the various transcription factors is currently under investigation. Several karyoskeleton acceptors for hypophosphorylated Rb have been reported. Durfee et al. identified an 84kDa karyoskeleton Rb acceptor that binds to the N-terminal region of Rb, in monkey kidney CV-1 cells (205). The 84-kDa protein colocalized with BlC8 to speckles or interchromatin granule clusters. NRP/B (nuclear restricted protein/brain), which is expressed in primary neurons, is also a karyoskeleton acceptor for hypophosphorylated Rb (206). The karyoskeleton has a central role in steroid hormone action (207,208). The estrogen receptor (ER) is associated with the karyoskeleton of estrogenresponsive tissues (209-212). In in vitro reconstitution studies with karyoskeletons and hormone receptors (e.g., ER and androgen receptor), it has been shown that nuclear acceptor sites for the hormone receptors are associated with the karyoskeleton (207, 209, 213) (Fig. 4). The binding of ER to the karyoskeleton was saturable, of high affinity, target tissue specific, and receptor specific (209). In studies with estrogen-responsive tissues, including human breast cancer cell line MCF-7, evidence for the presence of ER acceptor proteins has been reported (207,214). However, we do not know the identity of the karyoskeleton acceptor proteins for ER in human cells. Bind-

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

329

ing studies with ER provide evidence that some karyoskeleton acceptors for ER are ligand dependent (214). To observe ER subcellular trafficking and localization in living cells, we tagged the N terminus of ER with the S65T variant of the green fluorescent protein (GFP) (212). The GFP-ER fusion protein was a functional protein, as determined by transient transfection assays. Independent of ligand, GFPER protein was found in the nucleus of human breast cancer cells, with the exception of ICI 182780, with which some cytoplasmic GFP-ER was observed. In ligand-free conditions, a diffuse pattern of nuclear localization was seen for GFP-ER in MCF-7 cells. When e&radio1 was added to the media, the GFP-ER pattern became more punctate or speckled. The same effects were observed when 4-hydroxytamoxifen and ICI 182780 were added. Transiently expressed GFP-ER was associated with the nuclear matrix. We are currently testing the idea that the GFP-ER nuclear patterns reflect the location of nuclear matrix acceptors for ER.

IX. HATS, HDACs, and the Karyoskeleton Both HATS and HDACs are associated with the karyoskeleton (64). We proposed that karyoskeleton-associated HAT and HDACs mediate a transient attachment of transcriptionally active chromatin to the karyoskeleton. Many of the transcription factors that associate with the HATS or HDACs are karyoskeleton associated (e.g., YYl, hypophosphorylated Rb, Spl, GATA-1, ER) (215) (Figs. 4 and 5). HATS (e.g., TAFrr250, CBP), HDACs, and dynamically acetylated histones are found gathered around interchromatin granule clusters (6). The karyoskeleton may have a role in establishing this organization. Importantly, these interactions between HATS, HDACs, transcription factors, and active chromatin should be viewed as dynamic. We reported that HDACl is associated with matrix-associated region DNA in human breast cancer cells (185). These results suggest that HDACl may have a role in the organization of nuclear DNA. It is interesting to note that the attachment-region binding protein (ARBP), a karyoskeleton protein that binds to S/MARS, is homologous to MeCP2 (216). Thus, the N-CoRSin3A-HDACl complex could be recruited to the karyoskeleton and to S/MAR DNA by MeCP2iARBP (Fig. 5).

X. Mechanical Signaling Pathways and Organization of Nuclear DNA There is ample evidence that the tissue matrix system, which consists of the extracellular matrix, cytoskeleton, and karyoskeleton, sends signals from the cell exterior to the nuclear interior (217-219). Changes in the shape of the nucleus and the cytoskeleton most likely alter chromatin structure and

330

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

perturb the karyoskeleton. Changes in cell and nuclear shape regulate cell proliferation and gene expression (220, 221). A series of molecular events are put into play when a cell receives suitable extracellular cues to go from quiescence to a proliferative state. Activation of the Ras-MAPK pathway is necessary, but not sufficient, for cells to proliferate. Alterations in cell shape as defined as degree of cell extension or spreading are also required for cells to enter the cell cycle (220, 222, 223). The initial transient activation of the Ras-MAPK pathway results in the expression of p2 lcipI, an inhibitor of cyclin E-&2 (224). Cell adhesion to the ECM is required for the sustained activation of the Ras-MAPK pathway and expression of cyclin Dl (220, 222, 225). The cytoskeleton, particularly the actin network, plays a pivotal role in these events (220, 223). The active cyclin Dl-cdk4/6 complex then phosphorylates Rb, resulting in the dissociation of HDAC (122). A s cells reach middle to late G, phase of the cell cycle, the expression of p21ciPr declines. Cell adhesion to the ECM is required for down-regulation of p21cip1 and p2 7Kip1, which are inhibitors of cyclin E-cdk2 (224). Following the reduction of p21cip1 and ~27~@r, active cyclin E-cdk2 further phosphorylates Rb, preventing Rb from binding to E2Fl. Phosphorylation of Rb also results in its liberation from the karyoskeleton. EQF-regulated genes, such as cyclin A, are then expressed. Cells have entered into S phase with activation of cyclin A-cdk2. Thus, chemical and mechanical signaling pathways are required for quiescent Go phase cells to move through G, phase of the cell cycle (220). The B-casein gene is an excellent example of the importance of cell shape and structure in gene regulation. The extracellular matrix and prolactin activate the BCE-1 enhancer of this gene through at least three transcription factors binding to the enhancer and, perhaps, by altering the acetylation state of the histones. The chromatin context of the enhancer is critical because BCE1 on a nonintegrated template will not respond to extracellular and prolactin signals (226). The authors proposed two mechanisms by which the extracellular matrix induces the expression of the B-casein gene. Changes in the three-dimensional architecture of the cell by the extracellular matrix could alter the three-dimensional structure of the nucleus and the structure and/or composition of the karyoskeleton. Perturbation in the karyoskeleton could reposition karyoskeleton-associated HATS and/or HDACs (227) resulting in the remodeling by histone acetylation and transcriptional activation of the Bcasein chromatin template. Alternatively, the extracellular matrix could induce or modify cofactors that have HDAC or HAT activity (64). Thus this study shows how the structure of a cell can play a role in the regulation of gene expression. Current evidence suggests that intermediate filaments relay signals from the plasma membrane to nuclear DNA, resulting in changes in chromatin organization and perhaps function. Intermediate filaments, a component of the

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

331

cytoskeletal network, extend from the plasma membrane and penetrate the nuclear lamina (228-231). Further, intermediate filament proteins (cytokeratins and vimentin) are positioned to make contact with nuclear DNA in viva (228, 232, 233). Using the cross-linking agent cis-diamminedichloroplatinum (cisplatin), it has been shown that cisplatin preferentially cross-links karyoskeleton proteins to S/MAR DNA in situ (234). The interactions between intermediate filament proteins and nuclear DNA are dynamic. For example, the interaction between cytokeratin intermediate filaments and nuclear DNA is regulated by estrogens in ER-positive, hormone-dependent breast cancer cells (232, 235). Our studies suggested that estrogen regulated the levels of cytokeratins in the cells (232). In hormone-dependent cells estrogen-regulated interactions between intermediate filaments and nuclear DNA could manipulate the organization of chromatin. However, in ER-positive, hormonenonresponsive breast cancer cell line T5-PRF, the estrogen regulation of cytokeratins with nuclear DNA was not observed (232, 235). It was reported that activity of ERIC1 and ERK2 was elevated in the T5PRF cell line (236). ER is a phosphoprotein that may be phosphorylated at multiple sites by a variety of kinases. Stimulation of the RAS-MAPK pathway results in the phosphorylation of ER at Ser-167 and Ser-118 (237, 238). This phosphorylation event enhances DNA binding and transcription of ER (237, 238). Thus, enhanced activity of the Ras-MAPK pathway in the T5PRF cell line may negate the requirement for estrogen. There is evidence that the cytokeratin genes may be estrogen responsive (239). The constitutive activity of ER in the T5-PRF cell line may therefore result in deregulated expression of the cytokeratin genes. It is not known which DNA sequences are associated with the intermediate filaments and what are the consequences of the interaction between the cytoskeleton and chromatin. Because intermediate filaments are located at the periphery of the nucleus, they likely interact with heterochromatin. However, because transcribed genes are associated with the nuclear periphery, it is also possible that the intermediate filaments are in contact with transcribed regions of chromatin (240, 241). It was shown that a subset of karyoskeleton proteins may be involved in chromosome territory organization (242). Thus, the interactions between intermediate filaments, the karyoskeleton, and chromatin may have profound effects on the organization of chromosome territories and surrounding nuclear components.

Xl. Future Directions In the past 4 years, we have seen major breakthroughs in our understanding of the relationships between histone-modifying enzymes, transcription,

332

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

and signal transduction pathways. Future studies will further illuminate the role of the chemical and mechanical (tissue matrix system) signaling pathways in altering chromatin structure and gene expression. How HATS and kinases activated by various signal transduction pathways are recruited and modify specific regions of chromatin will be realized. Many histone modifications occur in concert, e.g., mitogen-stimulated histone H3 phosphorylation and dynamic acetylation. Future studies will reveal the mechanisms by which HATS, HDACs, histone kinases, and other histone-modifying enzymes are recruited to specific sites in nuclear space. We believe that the karyoskeleton has an important role in this process. Transiently assembled huge multiprotein complexes assembled on the karyoskeleton could coordinately catalyze chromatin remodeling and transcription. There is still much to be learned about the function of the histone tails and the consequences of modifications on tail function. Further, there are several histone-modifying enzymes that remain to be purified and characterized, e.g., enzymes catalyzing histone methylation and demethylation. In understanding the function of the activated histone-modifying enzymes, we need to know the precise gene location of the modified histone isoforms. The CHIP technology with antibodies recognizing specific modified histone isoforms will be pivotal to leaming which segments of a gene are targets of histone-modifying enzymes. Increasingly there are examples in the literature of how misdirecting and/or deregulating the activity of a histone-modifying enzyme leads to abnormal gene expression and cancer. In future studies, we will see the development of exciting new agents to inhibit the activity of these rogue enzymes, leading to novel approaches to treat cancer.

ACKNOWLEDGMENTS Research supportby grants from the Medical Research Council of Canada (MT-9186, RO15183), Manitoba Health Research Council and U.S. Army Medical and Materiel Command Breast Cancer Research Program (#DAM17-97-l-7175) and the Cancer Research Society, Inc. and a Medical Research Council of Canada Senior Scientist to JRD are gratefully acknowledged.

REFERENCES 1. K. Luger, A. W. Mader, R. K. Richmond, D. E Sargent, and T. J. Richmond, Nature &ondon) 389,251-260 (1997). 2. J. 0. Thomas, Cutx Opin. Cell Biol. ll,312-317 (1999). 3. W. An, S. H. Leuba, K. Van Holde, and J. Zlatanova, Proc. Nutl. Acad. Sci. U.S.A. 95,33963401 (1998). 4. A. S. Belmont and K. Bruce,J. Cell Biol. I27,287-302 (1994).

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

333

5. P J. Verschure, D. K. van, I, E. M. Manders, and R. van Driel, J. Cell Bid. 147, 13-24 (1999). 6. M. J. Hendzel, M. J, Kruhlak, and D. P. Bazett-Jones, Mol. Bid. Cell 9,2491-2507 (1998). 7. E. M. Manders, H. Kimura, and P. R. Cook,J. Cell Biol. 144,813-821 (1999). 8. F Lenfant, R. K. Mann, B. Thomsen, X. Ling, andM. Grunstein,EMBOJ. 15,3974-3985 (1996). 9. G. Fisher-Adams and M. Grunstein, EMBOJ. 14,1468-1477 (1995). 10. J. L. Baneres, A. Martin, and J. Parello,J. Mol. Biol. 273,503-508 (1997). 11. J. C. Hansen, C. Tse, and A. P. Wolffe, Biochemistry 37,17637-17641 (1998). 12. T. M. Fletcher and J. C. Hansen, @it. Reu. Eukaryot. Gene Expr. 6,149-188 (1996). 13. J, Zlatanova, S. H. Leuba, and K. Van Holde, Biophys. J. 74,2554-2566 (1998). 14. C. Tse and J, C. Hansen, Biochemistry 36,11381-11388 (1997). 15. S. C. Moore and J. Ausio, Biochem. Biophys. Res. Commun. 230,136-139 (1997). 16. A. Hecht, T Laroche, S. St&l-Bolsinger, S. M. Gasser, and M. Grunstein, Cell 80, 583592 (1995). 17. M. Grunstein, Cell 93,325-328 (1998). 18. D. G. Edmondson, M. M. Smith, and S. Y. Roth, Genes Deu. 10,1247-1259 (1996). 19. A. Palaparti, A. Baratz, and S. Stifani,J. Biol. Chem. 272,26604-26610 (1997). 20. A. L. Fisher and M. Caudy, Genes Deu. 12,1931-1940 (1998). 21. M. Bustin, Mol. Cell Biol. 19, 5237-5246 (1999). 22. U. G. Sathyanarayana, L. A. Freeman, M. S. Lee, and W. T. Garrard, J. Biol. Chem. 274, 16431-16436 (1999). (1987). 23. T. A. Chen andV. G. Allfrey, Proc. Natl. Acad. Sci. U.S.A. 84,5252-5256 24. D. P. Bazett-Jones, E. Mendez, G. J. Czarnota, F. P. Ottensmeyer, and V. G. Allfrey, Nucleic Acids Res. 24,321-329 (1996). 25. 6. J. Czamota, D. P. Bazett-Jones, E. Mendez, V G. Allfrey, and F. P. Ottensmeyer, Micron. 28419-431 (1997). 26. H. Kimura, Y. Tao, R. G. Roeder, and P. R. Cook, Mol. Cell Biol. 19, 5383-5392 (1999). 27. X. Wei, S. Somanathan, J. Samarabandu, and R. Berezney, J. Cell Biol. 146, 543-558 (1999). 28. J. R. Davie, Znt. Reu. Cytol. 162A, 191-250 (1995). 29. B. Villeponteau and H. G. Martinson, Mol. Cell Biol. 7, 1917-1924 (1987). 30. T. Misteli and D. L. Spector, Cum Opin. Cell Biol. 10,323-331 (1998). 31. Y. Takami, S. Takeda, and T. Nakayama,J. Mol. Biol. 265,394-408 (1997). 32. N. Louzalen, J. Moreau, and M. Mechali, Nucleic Acids Res. 24,3947-3952 (1996). 33. M.J. Clarkson, J. R. Wells, F. Gibson, R. Saint, and D. J. Tremethick, Nature @ondon) 399, 694-697 (1999). 34. Y. Takami, S. Takeda, and T. Nakayama,J. Biol. Chem. 270,30664-30670 (1995). 35. M. Clarkson and R. Saint, DNA Cell Biol. l&457-462 (1999). 36. J. R. Pehrson and R. N. Fuji, Nucleic Acids Res. 26,2837-2842 (1998). 37. C. Costanzi and J. R. Pehrson, Nature &ondonj 393,599-601 (1998). 38. P. E. Warburton, C. A. Cooke, S. Bourassa, 0. Vafa, B. A. Sullivan, G. Stetten, G. Gimelli, D. Warburton, C. Tyler-Smith, K. F. S u11’ wan, G. G. Poirier, and W. C. Eamshaw, Cum Biol.

7,901-904 (1997). 39. K. C. Keith, R. E. Baker, Y. Chen, K. Harris, S. Stoler, and M. Fitzgerald-Hayes, Mol. Cell Biol. 19, 6130-6139 (1999). 40. M. H. Parseghian, A. H. Henschen, K. G. Krieglstein, and B. A. Hamkalo, Protein Sci. 3, 575-587 (1994). 41. 0. C. Steinbach, A. P Wolffe, and R. A. Rupp, Nature &ondonj 389,395-399 (1997). 42. A. Gunjan and D. T Brown, Nucleic Acids Res. 27,3355-3363 (1999).

334

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

43. H. Y. Chen, J.-M. Sun, Y. Zhang, J. R. Davie, and M. L. Meistrich, J. Biol. Chem. 273, 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75.

13165-13169 (1998). J. R. Davie and L. C. Murphy, Biochemistry 29,4752-4757 (1990). M. J. Hendzel and J. R. Davie, Biochem. J. 271,67-73 (1990). J. R. Davie, Cm-r. Opin. Genet. Dev. 8,173-178 (1998). J. Covault and R. Chalkley,J. Biol. Chem. 255,9110-9116 (1980). M. R. Duncan, M. J. Robinson, and R. T. Dell’Orco, Biochim. Biophys. Actu 762,221-226 (1983). D.-E. Zhang and D. A. Nelson, Biochem. J. 250,233-240 (1988). D.-E. Zhang and D. A. Nelson, Biochem. J. 250,241-245 (1988). M. J. Hendzel, G. P. Delcuve, and J. R. Davie,J. Biol. Chem. 266,21936-21942 (1991). T. R. Hebbes, A. L. Clayton, A. W. Thome, and C. Crane-Robinson, EMBOJ. 13, 18231830 (1994). D. A. Nelson,]. Biol. Chem. 257, 1565-1568 (1982). M. Braunstein, A. B. Rose, S. G. Holmes, C. D. Allis, and J. R. Broach, Genes Dev. 7,592604 (1993). J. H. Waterborg,J. BioZ. Chem. 273,27602-27609 (1998). M. Garcia-Ramirez, C. Rocchini, and J. Ausio, J. Biol. Chem. 270, 17923-17928 (1995). H. Walia, H. Y. Chen, J.-M. Sun, L. T. Holth, and J. R. Davie,J. Biol. Chem. 273, 1451624522 (1998). C. Tse, T Sera, A. P. Wolffe, and J. C. Hansen, Mol. Cell. Biol. 18,4629-4638 (1998). K. P. Nightingale, R. E. Wellinger, J. M. Sogo, and P B. Becker, EMBOJ. 17,2865-2876 (1998). J. L. Workman and R. E. Kingston, Annu. Rev. Biochem. 67,545-579 (1998). V. Mutskov, D. Gerber, D. Angelov, J. Ausio, J, Workman, and S. Dimitrov, Mol. Cell BioZ. l&6293-6304 (1998). J. E. Brownell, J. X. Zhou, T Ranalli, R. Kobayashi, D. G. Edmondson, S. Y. Roth, and C. D. Allis, Cell 84,843-851 (1996). B. 0. Wittschieben, G. Otero, T de Bizemont, J. Fellows, H. Erdjument-Bromage. R. Ohba, Y. Li, C. D. Allis, P Tempst, and J. Q. Svejstrup, Mol. Cell 4, 123-128 (1999). J. R. Davie and D. N. Chadee,J. Cell Biochem. 30/3l, 203-213 (1998). T. Kouzarides, Curt-, Opin. Genet. Deu. 9,40-48 (1999). R. C. Trievel, J. R. Rojas, D. E. Sterner, R. N. Venkataramani, L. Wang, J. Zhou, C. D. Al(1999). lis, S. L. Berger, and R. Marmorstein, Proc. N&Z. Acad. Sci. U.S.A. 96,8931-8936 A. Clements, J. R. Rojas, R. C. Trievel, L. Wang, S. L. Berger, and R. Marmorstein, EMBO J. l&3521-3532 (1999). Y. Lin, C. M. Fletcher, J. Zhou, C. D. Allis, and G. Wagner, Nature jlondon) 400,86-89 (1999). J. R. Rojas, R. C. Tiievel, J. Zhou, Y. MO, X. Li, S. L. Berger, C. D. Allis, and R. Marmorstein, Nature (London) 401,93-98 (1999). K. G. Tanner, R. C. Trievel, M. H. Kuo, R. M. Howard, S. L. Berger, C. D. ALlis, R. Marmorstein, and J, M. Denu,]. BioZ. Chem. 274, 18157-18160 (1999). C. Dhalluin, J. E. Carlson, L. Zeng, C. He, A. K. Aggarwal, and M. M. Zhou, Nature (London) 399,491-496 (1999). P. Omaghi, P. Ballario, A. M. Lena, A. Gonzalez, and P. Filetici, J. Mol. BioZ. 287, l-7 (1999). S. L. Berger, Cum Opin. Cell Biol. 11,336-341 (1999). W. Xu, D. G. Edmondson, and S. Y. Roth, Mol. CeZZBiol. 18,5659-5669 (1998). R. T. Utley, K. Ikeda, P. A. Grant, J. Cote, D. J. Steger, A. Eberharter, S. John, and J. L. Workman, Nature (London) 394,498-502 (1998).

HISTONE

MODIFICATIONS

AND

SIGNALING

PATHWAYS

335

76. A. Eberharter, D. E. Sterner, D. Schieltz, A. Hassan, J. R. Yates III, S. L. Berger, and J. L. Workman, Mol. Cell Biol. 19,6621-6631 (1999). 77. I? A. Grant, D. Schieltz, M. G. Pray-Grant, D. J. Steger, J. C. Reese, J. R. Yates III, and J. L. Workman, Cell 94,45-53 (1998). 78. V. V. Ogryzko, T. Kotani, X. Zhang, R. L. Schlitz, T. Howard, X. J. Yang, B. H. Howard, J. Qin, and Y. Nakatani, Cell 94,35-44 (1998). 79. E. R. Smith, A. Eisen, W. Gu, M. Sattah, A. Pannuti, J. Zhou, R. G. Cook, J. C. Lucchesi, and C. D. Allis, Proc. Natl. Acad. Sci. U.S.A. 95,3561-3565 (1998). HO. N. Champagne, N. R. Bertos, N. Pelletier, A. H. Wang, M. Vezmar, Y. Yang, H. H. Heng, and X. J. Yang,]. Biol. Chem. 274,28528-28536 (1999). 81. R. Ohba, D. J. Steger, J. E. Brownell, C. A. Mizzen, R. G. Cook, J. Cote, J. L. Workman, and C. D. Allis, Mol. CeZZBiol. 19,2061-2068 (1999). 82. S. AIlard, R. T. UtIey, J. Savard, A. Clarke, P. Grant, C. J. Brandl, L. Pillus, J. L. Workman, and J. Cote, EMBOJ. 18, 5108-5119 (1999). 83. T. Nakajima, C. Uchida, S. E Anderson, C. G. Lee, J. Hun&z, J. D. Parvin, and M. Montminy, Cell 90, 1107-1112 (1997). 84. M. E. Brady, D. M. Ozanne, L. Gaughan, I. Waite, S. Cook, D. E. Neal, and C. N. Robson, J. BioZ. Chem. 274, 17599-17604 (1999). 85. J. C. Blanco, S. Minucci, J. Lu, X. J, Yang, K. K. Walker, H. Chen, R. M. Evans, Y. Nakatani, and K. Ozato, Genes Dee. B&1638-1651 (1998). 86. S. Ait-S-Al& D. Carlisi, S. Ramirez, L. C. Upegui-Gonzalez, A. Duquet, P. Robin, B. Rudkin, A. Harel-Bellan, and D. Trouche, Biochem. Biophys. Res. Commun. 262, 157-162 (1999). 87. Y. Z. Liu, N. S. Thomas, and D. S. Latchman, Neuroreport 10,1239-1243 (1999). 88. Y. Z. Liu, J. C. Chrivia, and D. S. Latchman,J. BioZ. Chem. 273,32400-32407 (1998). 89. J. E. Hen-era, K. Sakaguchi, M. Bergel, L. Trieschmann, Y. Nakatani, and M. Bustin, Mol. Cell Biol. 19,3466-3473 (1999). 90. H. Chen, R. J. Lin, W. Xie, D. Wilpitz, and R. M. Evans, Cell 98,675-686 (1999). 91. M. E. Massari, P. A. Grant, M. G. Pray-Grant, S. L. Berger, J. L. Workman, and C. Murre, I&Z. CeZZ4,63-73 (1999). 92. K. Ikeda, D. J. Steger, A. Eberharter, and J. L. Workman, Mol. Cell Biol. 19, 855-863 (1999). 93. M. H. Kuo, J. Zhou, P. Jambeck, M. E. Ch urchill, and C. D. Allis, Genes Dez;. 12,627-639 (1998). 94. T. Tumbar, G. SudIow, and A. S. Belmont, J. Cell BioZ. 145,1341-1354 (1999). 95. K. A. Sheppard, D. W. Rose, Z. K. Haque, R. Kurokawa, E. McInerney, S. Westin, D. Thanos, M. G. Rosenfeld, C. K. Glass, andT. Collins, Mol. Cell Biol. 19,6367-6378 (1999). (1998). 96. H. Zhong, R. E. Voll, and S. Ghosh, Mol. Cell l, 661-671 97. W. L. Kraus and J. T. Kadonaga, Genes Da. E&331-342 (1998). 98. L. Madisen, A. Krumm, T. R. Hebbes, and M. Groudine, Mol. Cell BioZ. 18, 6281-6292 (1998). 99. Z. Liu, J. Wong, S. Y. Tsai, M. J. Tsai, and B. W. O’Malley, Proc. NatZ. Acad. Sci. U.S.A. 96, 9485-9490 (1999). 100. D. J. Steger, A. Eberharter, S. John, P. A. Grant, and J. L. Workman, Proc. NatZ. Acad. Sci. U.S.A. 95,12924-12929 (1998). 101. A. E. WaIlberg, K. E. Neely, J. A. Gustafsson, J. L. Workman, A. P. Wright, and P. A. Grant, Mol. Cell BioZ. 19, 5952-5959 (1999). 102. T. K. Kim, T. H. Kim, and T. Mania& Proc. NatZ. Acad. Sci. U.S.A. 95, 12191-12196 (1998). 103. B. S. Parekh and T. Mania&, 1MoZ.Cell 3, 125-129 (1999).

336

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

104. N. Munshi, M. Merika, J. Yie, K. Senger, G. Chen, and D. Thanos, Mol. Cell 2,457-467 (1998). 105. D. D. Leipe and D. Landsman, Nucleic Acids Res. 25,3693-3697 (1997). 106. C. A. Hassig, J. K. Tong, T. C. Fleischer, T. Owa, P G. Grable, D. E. Ayer, and S. L. Sci. U.S.A. g&3519-3524 (1998). Schreiber, Proc. Nutl. Ad. 107. J.-M. Sun, H. Y. Chen, M. Moniwa, S. Samuel, and J. R. Davie, Biochemistrq 38,59395947 (1999). 108. M. S. Finnin, J. R. Donigian, A. Cohen, V. M. Richon, R. A. Rifkind, P. A. Marks, R. Breslow, and N. P Pavletich, Nature 401, 188-193 (1999). 109. A. Verde1 and S. Khochbin, Biol. Chm. 274,2440-2445 (1999). 110. W. Fischle, S. Emiliani, M. J. Hendzel, T. Nagase, N. Nomura, W. Voelter, and E. Verdin, J. Biol. Chem. 274,11713-11720 (1999). 111. C. M. Grozinger, C. A. Hassig, and S. L. Schreiber, Proc. N&l. Acad. Sci. U.S.A. 96,48684873 (1999). 112. T Nomura M. M. Khan, S. C. Kaul, H. D. Dong, R. Wadhwa, C. Colmenares, I. Kohno, and S. Ishii, Genes Deu. 13,412-423 (1999). 113. Y. Zhang, H. H. Ng, H. Erdjument-Bromage, P Tempst, A. Bird, and D. Reinberg, Genes Dew. 13,1924-1935 (1999). 114. A. A. Carmen, P. R. Griffin, J, R. Calaycay, S. E. Rundlett, Y. Suka, and M. Grunstein, Proc. Nutl. Aced. Sci. U.S.A. 96, 12356-12361 (1999). 115. J. K. Tong, C. A. Hassig, G. R. Schnitzler, R. E. Kingston, and S. L. Schreiber, Nature (Lon(1998). don) 395,917-921 116. l? A. Wade, P. L. Jones, D. Vermaak, and A. P. Wolffe, Cum Biol. 8, 843-846 (1998). 117. Y. Zhang, G. LeRoy, H. P. Seelig, W. S. Lane, and D. Reinberg, Cell 95,279-289 (1998). 118. Y. Xue, J. Wong, G. T. Moreno, M. K. Young, J. Cote, and W. Wang, Mol. Cell 2,851-861 (1998). 119. L. Magnaghi-Jaulin, R. Groisman, I. Naguibneva, P Robin, S. Lorain, J. P. Le Villain, F Troalen, D. Trouche, and A. Harel-Bellan, Nature (London) 391,601-605 (1998). 120. R. Ferreira, L. Magnaghi-Jaulin, P. Robin, A. Harel-Bellan, and D. Trouche, Proc. Nutl. Acud. Sci. U.S.A. 95,10493-10498 (1998). 121. A. Lai, J. M. Lee, W. M. Yang, J. A. DeCaprio, W. G. Kaelin, Jr., E. Seto, and P. E. Branton, Mol. Cell Biol. 19,6632-6641 (1999). 122. J. W. Harbour, R. X. Luo, S. A. Dei, A. A. Postigo, and D. C. Dean, Cell 98, 859-869 (1999). 123. A. Doetzlhofer, H. Rotheneder, G. Lagger, M. Koranda, V. Kurtev, G. Brosch, E. Wintersberger, and C. Seiser, Mol. Cell Biol. 19,5504-5511 (1999). 124. H. H. Ng, Y. Zhang, B. Hendrich, C. A. J o h nson, B. M. Turner, H. Erdjument-Bromage, P. Tempst, D. Reinberg, and A. Bird, Nature Genet. 23,58-61 (1999). 125. R. E. Amir, d. V. Van, I, M. Wan, C. Q. Tran, U. Francke, and H. Y. Zoghbi, Nature Genet. 23,185-188 (1999). 126. M. Murphy, J. Ahn, K. K. Walker, W. H. Hoffman, R. M. Evans, A. J. Levine, and D. L. George, Genes Dew. 13,2490-2501 (1999). 127. C. D. Laherty, A. N. Billin, R. M. Lavinsky, G. S. Yochum, A. C. Bush, J.-M. Sun, T.-M. Mullen, J. R. Davie, D. W. Rose, C. K. Glass, M. G. Rosenfeld, D. E. Ayer, and R. N. Eisenman, Mol. Cell 2,33-42 (1998). 128. Y. Zhang, Z. W. Sun, R. Iratni, H. Erdjument-Bromage, P. Tempst, M. Hampsey, and D. Reinberg, Mol. Cell l, 1021-1031 (1998). 129. R. M. Lavinsky, K. Jepsen, T Heinzel, J. Torchia, T. M. Mullen, R. Schiff, A. L. Del-Rio, M. Ricote, S. Ngo, J. Gemsch, S. G. Hilsenbeck, C. K. Osborne, C. K. Glass, M. G. Rosenfeld, and D. W. Rose, Proc. N&Z. Acad. Sci. U.S.A. 95,2920-2925 (1998).

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

337

130. R. I. Yarden and L. C. Brody, Proc. Natl. Acad. Sci. U.S.A. 96,4983-4988 (1999). 131. D. B. Sparrow, E. A. Miska, E. Langley, S. Reynaud-Deonauth, S. Kotecha, N. Towers, G. Spohr, T. Kouzarides, and T. J. Mohun, EMBOJ. l&5085-5098 (1999). 132. E. A. Miska, C. Karlsson, E. Langley, S. J. Nielsen, J. Pines, and T. Kouzarides, EMBOJ. 18, 5099-5107 (1999). 133. R. H. Giles, D. J. Peters, and M. H. Breuning, Trends. Genet. 14, 178-183 (1998). 134. S. Jacobson and L. Pillus, Cum Opin. Genet. Dev. 9,175-184 (1999). 135. R. J. Lin, L. Nagy, S. Inoue, W. Shao, W. H. Miller, Jr., and R. M. Evans, Nature (London) 391,811-814 (1998). 136. F. Grignani, S. De Matteis, C. Nervi, L. Tomassoni, V. Gelmetti, M. Cioce, M. Fanelli, M. Ruthardt, F. F. Ferrara, I. Zamir, C. Seiser, M. A. Lazar, S. Minucci, and I? G. Pelicci, Nature &ondon) 391,815-818 (1998). 137. B. Lutterbach, J. J. Westendorf, B. Linggi, A. Patten, M. Moniwa, J. R. Davie, K. Huynh, V. J. Bardwell, R. M. Lavinsky, M. G. Rosenfeld, E. Seto, and S. W. Hiebert, Mol. Cell. Biol. 18,7176-7184 (1998). 138. V. M. Richon, S. Emiliani, E. Verdin, Y. Webb, R. Breslow, R. A. Rilkind, and P. A. Marks, Proc. Natl. Acad. Sci. U.S.A. 95,3003-3007 (1998). 139. V. Orlando, H. Strutt, and R. Paro, Methods ll,205-214 (1997). 140. I? C. Dedon, J. A. Soults, C. D. All’ is, and M. A. Gorovsky, Anal. Biochem. 197, 83-90 (1991). 141. D. Kadosh and K. Struhl, Mol. Cell BioZ. B18, 5121-5127 (1998). 142. B. M. Turner, A. J. Birley, and J. Lavender, Cell 69,375-384 (1992). (1998). 143. M. J. Pikaart, F. Recillas-Targa, and G. Felsenfeld, Genes Dev. l2,2852-2862 144. J. Koipally, A. Renold, J. Kim, and K. Georgopoulos, EMBOJ. 18,3090-3100 (1999). 145. M. Braunstein, R. E. Sobel, C. D. Allis, B. M. Turner, and J. R. Broach, Mol. Cell. BioZ. 16, 4349-4356 (1996). (1998). 146. L. Howe, T. A. Ranalli, C. D. Allis, and J. Ausio,J. BioZ. Chem. 273,20693-20696 147. P Collas, M. R. Liang, M. Vincent, and P. Alestrom, J. Cell Sci. 112, 1045-1054 (1999). 148. T. Tagami, Y. Park, and J. L. J ameson,J. Biol. Chem. 274,22345-22353 (1999). 149. R. X. Luo, A. A. Postigo, and D. C. Dean, Cell 92,463-473 (1998). 150. B. Coffee, F. Zhang, S. T. Warren, and D. Reines, Nature Genet. 22,98-101 (1999). 151. A. S. Alberts, 0. Geneste, and R. Treisman, Cell 92,475-487 (1998). 152. P. B. Meluh and J. R. Broach, Methods Enzymol. 304,414-430 (1999). 153. J. E. Krebs, M. H. Kuo, C. D. Allis, and C. L. Peterson, Genes Dev. 13,1412-1421 (1999). 154. H. Cho, G. Orphanides, X. Sun, S. J. Yang, V. Ogryzko, E. Lees, Y. Nakatani, and D. Reinberg, Mol. Cell Biol. 18,5355-5363 (1998). 155. S. E. Rundlett, A. A. Carmen, N. Suka, B. M. Turner, and M. Grunstein, Nature jlondon)

392,831-835 (1998). 156. S. M. Burgess, M. Ajimura, and N. Kleckner, Proc. NatZ. Acad. Sci. U.S.A. 96,6835-6840 (1999). 157. R. Desrosiers and R. M. Tanguay,]. BioZ. Chem. 263,4686-4692 (1988). 158. J. H. Waterborg, A. J. Robertson, D. L. Tatar, C. M. Borza, and J. R. Davie, PZant Physiol. 109,393-407 (1995). 159. Y. Ohe and K. Iwai, Biochem. (Tokyoj 90,1205-1211 (1981). 160. W. F Brandt, W. N. Strickland, M. Morgan, and C. Von Holt, FEBS Lett. 40, 167-172 (1974). 161. M. J. Her&e1 and J. R. Davie,J. BioZ. Chem. 264,19208-19214 (1989). 162. J. H. Waterborg,]. BioZ. Chem. 268,4912-4917 (1993). 163. J. Reneker and T. W. Brotherton, Biochemistry 30,8402-8407 (1991). 164. A. T. Annunziato, M. B. Eason, and C. A. Perry, Biochemistry 34,2916-2924 (1995).

338

JAMES FL DAVIE AND VIRGINIA

A. SPENCER

165. M. J. Hendzel and J. R. Davie, Biochem. Biophys. Res. Commun. 185,414-419 (1992). 166. K. Ohsumi, C. Katagiri, and T Kishimoto, Science 262,2033-2035 (1993). 167. D. N. Chadee, W. R. Taylor, R. A. R. Hurta, C. D. Allis, J. A. Wright, and J. R. Davie,J. BioZ. Chem. 270,20098-20105 (1995). 168. Y. Wei, C. A. Mizzen, R. G. Cook, M. A. Gorovsky, and C. D. Allis, Proc. N&Z. Acad. Sci. U.S.A. 95, 7480-7484 (1998). 169. Y. Wei, L. Yu, J. Bowen, M. A. Gorovsky, and C. D. Allis, Cell 97,99-109 (1999). 170. M. J. Hendzel, Y. Wei, M. A. Mancini, A. Van Hooser, T. RanaUi, B. R. Brinkely, D. P. BazettJones, and C. D. Allis, Chromosoma 106,348-360 (1997). 171. D. M. Sauve, H. J. Anderson, J. M. Ray, W. M. J ames, and M. Roberge, J. CeZZBioZ. 145, 225-235 (1999). 172. H. Goto, Y. Tomono, K. Ajiro, H. Kosako, M. Fujita, M. Sakurai, K. Okawa, A. Iwamatsu, T. Okigaki, T. Takahashi, and M. Inagaki,]. BioZ. Chem. 274,25543-25549 (1999). 173. H. L. Lee and T. K. Archer, EMBOJ. 17,1454-1466 (1998). 174. D. N. Chadee, C. D. Allis, J. A. Wright, and J. R. Davie,J: BioZ. Chem. 272,8113-8116 (1997). 175. R. E. Herrera, F. Chen, and R. A. Weinberg, Proc. NatZ.Acad. Sci. U.S.A. 93,11510-11515 (1996). 176. D. N. Chadee, M. J. Hendzel, C. I? Tylipski, C. D. Allis, D. P. Bazett-Jones, J, A. Wright, and J. R. Davie,J. Biol. Chem. 274,24914-24920 (1999). 177. V. R. Iyer, M. B. Eisen, D. T. Ross, G. Schuler, T. Moore, J. C. F. Lee, J. M. Trent, L. M. Staudt, J. Hudson, Jr., M. S. Boguski, D. Lashkari, D. ShaIon, D. Botstein, and P. 0. Brown, Science 283,83-87 (1999). 178. B. Wasylyk, J. Hagman, and A. Gutierrez-Hartmann, Trends. Biochem. Sci. 23,213-216 (1998). 179. S. Ramirez, S. A. S. Ah, P. Robin, D. Trouche, and A. Harel-Bellan, J. Biol. Chem. 272, 31016-31021 (1997). 180. H. GiIle, M. Kortenhann, T. Strahl, and P. E. Shaw, Mol. Cell Biol. 16, 1094-1102 (1996). 181. R. Janknecht and A. Nordheim, Biochem. Biophys. Res. Commun. 228,831-837 (1996). 182. J. Feng and B. ViIIeponteau, Biochim. Biophys. Acta 1130,253-258 (1992). 183. P. Sassone-Corsi, C. A. Mizzen, F’.Cheung, C. Crosio, L. Monaco, S. Jacquot, A. Hanauer, and C. D. Allis, Science 285,886-891 (1999). 184. S. Thomson, A. L. Clayton, C. A. HazzaIin, S. Rose, M. J. Barratt, and L. C. Mahadevan, EMBOJ. 18,4779-4793 (1999). 185. S. K. Samuel, V. A. Spencer, L. Bajno, J.-M. Sun, L. T. Holth, S. Oesterreich, and J. R. Davie, Cancer Res. 58,3004-3008 (1998). 186. I. de Belle, S. Cai, and T. Kohwi-Shigematsu,J. Cell BioZ. 141335-348 (1998). 187. J. Bode, M. Stengert-Iber, V. Kay, T. Schlake, and A. Dietz-Pfeilstetter, Crit. Rev. Eukaryot. GeneExpr. 6,115-138 (1996). 188. D. C. He, J. A. Nickerson, and S. Penman,]. Cell BioZ. 110,569-580 (1990). 189. D. A. Jackson and P. R. Cook, EMBOJ. 7,3667-3677 (1988). 190. J. A. Nickerson,J. Cell Biochem. 70, 172-180 (1998). 191. J. Harbor&, J. Wang, C. Gueth-Hallonet, K. Weber, and M. Osbom, EMBOJ. 18,16891700 (1999). 192. J. A. Nickerson, G. KrockmaInic, K. M. Wan, and S. Penman, Proc. iVatZ.Acad. Sci. U.S.A. 94,4446-4450 (1997). 193. M. J. Hendzel, l? Boisvert, and D. P. Bazett-Jones, Mol. BioZ. Cell 10,2051-2062 (1999). (1996). 194. E J. Iborra, A. Pombo, D. A. J ac k son, and P. R. Cook,j. Cell Sci. 109,1427-1436 195. M. G. Gerdes, K. C. Carter, P. T. Moen, Jr.. and J. B. Lawrence,]. Cell BioZ. l26,289-304 (1994).

HISTONE

MODIFICATIONS

AND SIGNALING

PATHWAYS

339

196. C. Zeng, S. McNeil, S. Pockwinse, J. Nickerson, L. Shopland, J. B. Lawrence, S. Penman, S. Hiebert, J. B. Lian, A. J. van Wijnen, J. L. Stein, and G. S. Stein, Proc. Natl. Acad. Sci. U.S.A. 95,1585-1589 (1998). 197. G. S. Stein, A. J. van Wijnen, J. L. Stein, J. B. Lian, S. M. Pockwinse, and S. McNeil, J. Cell Biochem. Suppl. 30-31,220-231 (1998). 198. A. J. van Wijnen, J. P. Bidwell, E. G. Fey, S. Penman, J. B. Lian, J. L. Stein, and G. S. Stein, Biochemistry 32,8397-8402 (1993). 199. B. F. Miiller, D. Paulsen, and W. Deppert, Oncogene l2,1941-1952 (1996). 200. M. K. Kim, L. A. Lesoon-Wood, B. D. Weintraub, and J. H. Chung, Mol. Cell. Biol. 16, 4366-4377 (1996). 201. D. Stenoien, Z. D. Sharp, C. L. Smith, and M. A. Mancini, J. Cell Biochem. 70,213-221 (1998). 202. S. McNeil, B. Gou, J. L. Stein, J. B. Lian, S. Bushmeyer, E. Seto, M. L. Atchison, S. Penman, A. J. van Wijnen, and G. S. Stein,J. Cell. Biochem 68,500-510 (1998). 203. Y. Tang, R. H. Getzenberg, B. N. Vietmeier, M. R. Stallcup, M. Eggert, R. Renkawitz, and (1998). D. B. DeFranco, Mol. Endocrinol. l2,1420-1431 204. M. G. Mancini, B. Liu, Z. D. Sharp, and M. A. Mancini, J. Cell Biochem. 72, 322-338 (1999). 205. T. Durfee, M. A. Mancini, D. Jones, S. J. Elledge, and W.-H. Lee, J. Cell Biol. 127, 609622 (1994). 206. T. A. Kim, J. Lim, S. Ota, S. Raja, R. Rogers, B. Rivnay, H. Avraham, and S. Avraham,J. Cell Biol. 141, 553-566 (1998). 207. A. H. Lauber, N. l? Sandhu, M. Schuchard, M. Subramaniam, and T C. Spelsberg, ht. Reu. Cytol. 162B, 337-376 (1995). 208. Y. T. Tang and D. B. DeFranco, Mol. Cell. Biol. 16, 1989-2001 (1996). 209. D. A. Metzger and K. S. Korach, Endocrinology 126,2190-2195 (1990). 210. D. A. Metzger, S. Curtis, and K. S. Korach, Endocrinology 128,1785-1791 (1991). 211. R. B. Alexander, G. L. Greene, and E. R. Barrack, Endocrinology 120,1851-1857 (1987). 212. H. Htun, L. T. Holth, D. Walker, J. R. Davie, and G. L. Hager, Mol. Biol. Cell 10,471-486 (1999). 213. E. R. Barrxk,]. SteroidBiochem. 27,115-121 (1987). 214. R. K. Singh, M. F. Ruh, W. B. Butler, and T S. Ruh, Endocrinology 118,1087-1095 (1986). 215. J. R. Davie, S. Samuel, V. Spencer, L. Bajno, J.-M. Sun, H. Y. Chen, and L. T. Holth, Gene Ther. Mol. Biol. 1,509-528 (1998). 216. J. M. Weitzel, H. Buhrmester, and W. H. Stratling, Mol. Cell Biol. 17,5656-5666 (1997). 217. M. E. Chicurel, C. S. Chen, and D. E. Ingber, Cum @in. Cell Biol. 10,232-239 (1998). 218. N. Boudreau and M. J. Bissell, Cum @in. Cell Biol. 10, 640-646 (1998). (1998). 219. D. Ingber,]. Cell Biochem. Suppl. 30-31232-237 220. S. Huang, C. S. Chen, and D. E. Ingber, Mol. Biol. Cell 9,3179-3193 (1998). 221. C. S. Chen, M. Mrksich, S. Huang, G. M. Whitesides, and D. E. Ingber, Science 276,14251428 (1997). 222. R. K. Assoian,J. Cell Biol. 136, 1-4 (1997). 223. R. K. Assoian and X. Zhu, Cum. Opin. Cell Biol. 9,93-98 (1997). 224. M. E. Bottazzi, X. Zhu, R. M. Bohmer, and R. K. Assoian, J, Cell Biol. 146, 1255-1264 (1999). 225. K. Roovers, G. Davey, X. Zhu, M. E. Bottazzi, and R. K. Assoian, Mol. Biol. Cell 10,31973204 (1999). 226. C. A. Myers, C. Schmidhauser, J. Mellentin-Michelotti, G. Fragoso, C. D. Roskelley, G. Casperson, R. Mossi, P. Pujuguet, G. Hager, and M. J. Bissell, Mol. Cell Biol. 18, 21842195 (1998).

340

JAMES R. DAVIE AND VIRGINIA

A. SPENCER

227. J. R. Davie, Mol. Biol. Rep. 24,197-207 (1997). 228. W. S. Ward, W. N. Schmidt, C. A. Schmidt, and L. S. Hnilica, Proc. Natl. Acud. Sci. U.S.A. 81,419-423 (1984). 229. J. A. Nickerson, G. Krockmalnic, D. C. He, and S. Penman, Proc. Natl. Acad. Sci. U.S.A. 87,2259-2263 (1990). 230. S. Penman, Proc. Natl. Acad. Sci. U.S.A. 92,5251-5257 (1995). 231. X. Wang, G. Tolstonog, R. L. Shoeman, and P. Traub, DNA Cell Biol. 15,209-225 (1996). 232. V. A. Spencer, A. S. Coutts, S. K. Samuel, L. C. Murphy, and J. R. Davie,J. Biol. Chem. 273, 29093-29097 (1998). 233. R. Olinski, A. Wedrychowski, W. N. Schmidt, R. C. Briggs, and L. S. Hnilica, Cancer Res. 47,201-205 (1987). 234. L. T Holth, D. N. Chadee, V. A. Spencer, S. K. Samuel, J. R. Safneck, and J. R. Davie, ht. J. Chcol. 13,827-837 (1998). 235. A. S. Coutts, J. R. Davie, H. Dotzlaw, and L. C. Murphy,J. Cell. Biochem. 63, 174-184 (1996). 236. A. S. Coutts and L. C. Murphy, Cancer Res. 58,4071-4074 (1998). 237. G. Bunone, P A. Briand, R. J. Miksicek, and D. Picard, EMBOJ. 15,2174-2183 (1996). 238. P. B. Joel, J. Smith, T W. Sturgill, T. L. Fisher, J. Blenis, and D. A. Lannigan, Mol. Cell Biol. l&1978-1984 (1998). 239. I. Choi, J. Hoeksema, L. J. Gudas, and B. S. Katzenellenbogen, Endo 98, Pl-11 (1998). 240. P C. Park and U. De Boni, Li$ Sci. 64,1703-1718 (1999). 241. P C. Park and U. De Boni, Chromosome 107,87-95 (1998). 242. H. Ma, A. J. Siegel, and R. Berezney,J. Cell Biol. 146,531-542 (1999).

Regulation of Mammalian Cell Membrane Biosynthesis ATHANASIOSLYKIDISAND SUZANNEJACKOWSKI*~~ Department of Biochemistry St. Jude Children’s Research Hospital Memphis, Tennessee 38105 and *Department of Biochemistry University of Tennessee Memphis, Tennessee 38163 I. Introduction.. ........................................... II. Regulation of Phosphatidylcholine Biosynthesis ............... A. Choline Kinase ....................................... B. CTP:Phosphocholine Cytidylyltransferase ................. ........ C. CDP-Choline:l&diacylglycerol Phosphotransferase III. Regulation of Phosphatidylethanolamine Biosynthesis .......... A. Ethanolamine Kinase .................................. B. CTP: Phosphoethanolamine Cytidylyltransferase ........... C. CDP-Ethanolamine: 1,2diacylglycerol Phosphotransferase IV. Regulation of Phosphatidylinositol Biosynthesis ............... A. Diacylglycerol Kinase .................................. B. CDP-Diacylglycerol Synthase ........................... C. Phosphatidylinositol Synthase ........................... . Regulation of Phosphatidylserine Biosynthesis ................ VI. Cells Maintain a Constant Amount of Membrane Phospholipid VII. Inhibition of Phosphatidylcholine Biosynthesis Induces Apoptosis VIII. Future Directions ........................................ References ..............................................

. .

362 364 364 365 373 374 375 375 376 376 377 377 378 379 380 382 386 388

This review explores current information on the interrelationship between phospholipid biochemistry and cell biology. Phosphatidylcholine is the most abundant phospholipid and it biosynthesis has been studied extensively. The choline cytidylyltransferase regulates phosphatidylcholine production, and recent advances in our understanding of the mechanisms that govern cytidylyltransferase include the discovery of multiple isoforms and a more complete understanding of the lipid regulation of enzyme activity. Similarities between phosphatidylcholine formation and the phosphatidylethanolamine and phosphatidylinositol biosynthetic pathways are discussed, together with current insight into control mechanisms. Membrane phospholipid doubling during cell cycle progression is a function of

1 To whom correspondence Progress in Nucleic Acid Research and Molecular Biology, Vol. 65

should be addressed.

361

Copylight 0 2001 by Academic Preaa. All rights of reproduction in any form resewed. 0079.6603/01 $35 00

362

ATHANASIOS

LXKIDIS

AND SUZANNE

JACKOWSKI

periodic biosynthesis and degradation. Membrane bomeostasis is maintained by a pbospbolipase A-mediated degradation of excess pbospbolipid, whereas insufficient pbospbatidylcbolhe triggers apoptosis in cells. o wooAcademic PESS.

I. Introduction Membranes separate the interior of the cell from the environment and also delineate subcellular organelles. Membranes consist mainly of phospholipids, which are amphipathic small molecules that spontaneously aggregate into bilayers in aqueous media. Phospholipids have a glycerol backbone, two fatty acyl chains, and a phosphorylated alcohol headgroup. The exception is sphingomyelin, in which the backbone is sphingosine, an aminoalcohol that contains a long, unsaturated hydrocarbon chain linked by an amide bond to a fatty acid. The primary hydroxyl group of sphingosine is linked to phosphorylcholine. The most abundant membrane phospholipids are phosphatidylcholine (PtdCho),2 phosphatidylethanolamine (PtdEtn), and sphingomyelin (SM), whereas phosphatidylserine (PtdSer) and phosphatidylinositol (PtdIns) are present in very small quantities. The biosynthetic pathways for PtdCho, PtdEtn, and PtdIns are strikingly similar (Fig. 1). Each of the three pathways is initiated by a kinase reaction (CK, EK, or DGK) that phosphorylates choline, ethanolamine or diacylglycerol. The next step, catalyzed by a cytidylyltransferase (CCT, ECT, or CDS), is characterized by the transfer of a @dine moiety donated by CTP to the phospho derivative to yield the cytidylyldiphospho intermediate. Finally, a phosphotransferase (CPT, EPT, or PIS) catalyzes the formation of the final phospholipid molecule. The enzymes involved in the PtdCho and PtdEtn pathways are highly related structures. Choline kinase also functions as an ethanolamine kinase (I 3). CPT and EPT from yeast have a high degree of homology between their

’ Abbreviations: Cho, choline; P-Cho, phosphorylcholine; CDP-Cho, cytidylyldiphosphocholine; PtdCho, phosphatidylcholine; Etn, ethanolamine; P-Etn, phosphorylethanolamine; CDP-Etn, cytidylyldiphosphoethanolamine; PtdEtn, phosphatidylethanolamine; DG, diacylglycerol; PtdOH, phopshatidic acid; CDP-DG, cytidylyldiphosphodiacylglycerol; PtdIns, phosphatidylinositol; PtdSer, phosphattidylserine; SM, sphingomyelin; CK, choline kinase; CCT, CTP:phosphocholine cytidylyltransferase; CPT, CDP-Cho:1,2-diacylglycerol phosphotransferase; EK, ethanolamine kinase; ECT, CTP:phosphoethanolamine cytidylyltransferase; EPT, CDP-Ethanolamine:l,2diacylglycerol phosphotransferase; DGK, diacylglycerol kinase; CDS, CDP-diacylglycerol synthetase; PIS, phosphatidylinositol synthase; PEMT, phosphatidylethanolamine methyltransferase; PSS, phosphatidylserine synthase; PSD, phosphatidylserine decarboxylase; SMS, sphingomyelin synthase; GPC, glycerophosphocholine; GPE glycerophosphoethanolamine.

MAMMALIAN

CELL MEMBRANE

363

BIOSYNTHESIS

SM

y Cho *

P-Cho

2

CDP-Cho F

PtdCho ~ t pemt

Etn

+

P-Etn

9~

CDP-Etn

y

CTP

ATP

DG

AT

PtdOH

Fb

ATP

CTP

kinase

cytidylyltransferase

PtdLtn

.

PtdSer

&

DG

CDP-DG

+

Ptdlns I”\

phosphotransferase

FIG. 1. Pathways of phospholipid biosynthesis in mammalian cells. Phosphatidylcholine, phosphatidylethanolamine, and phosphatidylinositol are synthesized by a series of similar reactions. Sphingomyelin (SM) is synthesized from phosphatidylcholine whereas phosphatidylserine can be synthesized from either phosphatidylcholine or phosphatidyletbanolamine. Abbreviations in italic characters indicate the enzymes that catalyze each reaction.

primary sequences (4) and CPT utilizes CDP-ethanolamine and functions as an EPT in vitro (5). The first two enzymes of both the PtdCho and PtdEtn biosynthetic pathways are water soluble and catalyze reactions using watersoluble substrates. In contrast, the enzymes of the PtdIns pathway are exclusively membrane bound and utilize hydrophobic intermediates. The water-soluble headgroup, inositol, is the last moiety added to PtdIns (Fig. 1). PtdCho is the most abundant phospholipid, constituting 40-80% of the total membrane mass, depending on the cell type. PtdCho is also a precursor to PtdEtn, SM, and PtdSer. The CDP-choline pathway is the major route for PtdCho biosynthesis. In liver tissue, the methylation of PtdEtn by PtdEtn methyltransferase (PEMT) is an alternative pathway to PtdCho formation and is of quantitative significance under conditions of choline starvation (6, 7). PtdCho donates its phosphocholine headgroup to ceramide to form SM in a reaction catalyzed by sphingomyelin synthase (SMS). PtdCho is converted to PtdSer by a base exchange reaction catalyzed by PtdSer synthase I (PSSI). PtdEtn can also be converted to PtdSer by the same reaction catalyzed by PtdSer synthase II (PSSII). Decarboxylation of PtdSer by the PtdSer decarboxylase (PSD) yields PtdEtn. The cellular content of PtdEtn is maintained at a constant level for each cell type, just as the PtdCho content is controlled, and so the relative contribution of the two pathways of PtdEtn synthesis to the PtdEtn pool, either via the CDP-ethanolamine intermediate or by PtdSer decarboxylation, has been a long-standing interest.

364

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

The CDP-choline pathway of PtdCho biosynthesis has been a major focus of research in the area of mammalian phospholipid metabolism. PtdCho is a dynamic metabolic pool that undergoes rapid turnover. Cells coordinate the biosynthesis and degradation of PtdCho to double the membrane phospholipid prior to cell division and it is likely that cells also regulate PtdCho metabolism to maintain a constant composition of each phospholipid class. In this review the discussion focuses on recent progress concerning the mechanisms that regulate PtdCho biosynthesis and metabolism and current ideas that quantitatively and qualitatively define the phospholipid content of cell membranes.

II. Regulation of Phosphatidylcholine Biosynthesis PtdCho biosynthesis has been studied more extensively than the other pathways of phospholipid biosynthesis. PtdCho is not found in prokaryotic cells with few exceptions, but it is a major component of cell membranes in yeast and higher organisms. Choline is not synthesized by mammalian cells and is a nutritional requirement. Choline is also necessary for acetylcholine biosynthesis, and PtdCho can supply the precursor for acetylcholine when the choline supply is limited (8). PtdCho is not only found in cell membranes, but it is also secreted from selected tissues. For example, it is a component of lung surfactant, bile, and serum lipoprotein particles. Regulation of PtdCho synthesis is important to maintain phospholipid homeostasis both inside and outside the cells.

A. Choline

Kinase

The CK reaction commits choline to PtdCho biosynthesis. CK is a soluble enzyme that has been purified to homogeneity from several sources (1, 2), and multiple isoforms have been cloned from several species. A rat liver cDNA, CKRl, encodes a protein of 435 amino acids (9) and a splice variant, CKRS, contains an extra 54 bp of internal nucleotide sequence encoding an l&amino acid insertion (IO). A third cDNA from rat kidney yields CKR3, a protein of 394 amino acids, the product of a second gene (11). Two CK clones from mouse embryo are homologous with the two rat genes and both isoforms are expressed ubiquitously (12). Th ere fore, there are at least two genes encoding rat and mouse CK, but only one human CK is characterized thus far (13). The hCK cDNA is similar to the rat CKRl and CKR2 isoforms. CK appears to be a dual-specificity enzyme because it catalyzes both the phosphorylation of choline and ethanolamine in vitro (1, II), but the contribution of CK to PtdEtn biosynthesis has not been evaluated in vivo. CK regulation has not been studied, although its activity correlates with mitogenesis. Treat-

MAMMALIANCELLMEMBRANEBIOSYNTHESIS

365

ment of 3T3 fibroblasts with serum or growth factors increases CK activity (14) corresponding to elevated phosphocholine levels in cells (15). The phosphocholine level in Ha-ru.s-transformed NIH 3T3 cells is also elevated (16) and was initially thought to result from PtdCho degradation. However, later studies suggest that activation of CK accounts for the phosphocholine elevation in transformed cells (17, 18).

B. CTP:Phosphocholine

Cytidylyltransferase

1. Two GENESENCODINGCCT The primary pathway for PtdCho synthesis in all mammalian tissues is the CDP-choline pathway (7), and metabolic labeling establishes CCT as the rate-determining step (19). Until recently, only one isoform of CCT, called CCTcx, has been identified and cDNAs from rat (ZO), mouse (21), hamster (22), and human (23) cells reveal only minor differences in primary sequence among these mammalian proteins. The murine CCTol gene, called Ctpct, is localized on chromosome 16 spans approximately 26 kb, and is composed of 9 exons and 8 introns (21), (24). The 5’ flanking region upstream from the first exon contains GC-rich regions but lacks TATA or CAAT boxes. The first exon contains only untranslated sequence and the remaining 8 exons correspond to functional domains of the CCT protein. Exon 2 encodes a domain characterized by a nuclear localization signal, exons 4 through 7 correspond to the catalytic domain, exon 8 codes for a regulatory domain consisting of a repeated a-helical protein structure, and exon 9 corresponds to the reversibly phosphorylated CCT carboxy-terminal domain. Analysis of the promoter region indicates that Spl, Apl, and an unknown transcription factor regulate expression of the CCTo gene (25). The pattern of transcriptional regulation is characteristic of a housekeeping gene, consistent with its ubiquitous expression (26). The CCTa proteins have four distinct domains (Fig. 2). A nuclear targeting sequence includes a characteristic cluster of five positively charged amino acids and is located at the amino terminus of the protein, This domain directs the accumulation of the protein in the nucleus (27) but disruption of the nuclear localization sequence does not affect the ability of CCTo to complement the temperature-sensitive growth defect in cells containing a conditionally defective CCT (27). These data suggest that nuclear CCTa may not have an essential role. However, overexpression is a blunt tool to address the function of nuclear CCTa, because a portion of the protein lacking the CCTo targeting sequence is still found within the nucleus (27). The catalytic core of the protein resides near the center of the protein primary sequence (Fig. 2) and has a high degree of homology with a similarly

366

ATHANASIOS

c&&tic I 0

28

I

72

LYKIDIS

AND SUZANNE

HellI

core

Domah I

233

2%2~

JACKOWSKI

PhOsphU@tiOll

Domain I 315

CCTplI

CCTf32 FIG. 2. Domain structure and comparison of the predicted amino acid sequences of CCT isoforms. Regions in black represent identical residues. The arrow depicts the point of splicing in the p isoforms.

located region in yeast CCT (28) and the amino terminus of Bacillus sub& glycerol-3-phosphate cytidylyltransferase (29). An HXGH motif is conserved among the enzymes in the cytidylyltransferase superfamily (30) and is an important component of the CTP binding site (31, 32). Mutation of the glycine residue in this motif results in a 25-fold decrease in the affinity of the enzyme for CTP (31), and the catalytic defects exhibited by CCT enzyme with mutated histidine residues point to their involvement in the stabilization of the transition state of the CCT reaction (32). The crystal structure of the B. subt&s glycerol3-phosphate cytidylyltransferase has been solved (33), and, presents the best model for understanding the molecular arrangement of the amino acids that make up the CCT active site. The third domain of CCT is adjacent to the catalytic core and consists of three 11-residue amphipathic a-helical repeats (20) whose secondary structure is promoted by interaction with lipid (34). One helical repeat is all that is necessary for CCT binding to lipid bilayers (35), but all three helices participate in the interaction (36). CCT enzyme activation is absolutely dependent on lipid interaction and the helical region is the major regulatory domain that mediates the regulation of CCT enzyme by lipids embedded in PtdCho vesicles (3 7- 39). The carboxy terminus of CCTa is enriched in amino acids that are potential phosphorylation sites. Within a stretch of 52 amino acids, there are 16 serines, 2 threonines, and 1 tyrosine. Seven of the serines are adjacent to proline residues, thus suggesting these as substrate sites for phosphorylation by specific kinases such as MAP kinase or cdc2 kinase. CCTol is, in fact, extensively phosphorylated in viva (40). Phosphorylation occurs exclusively in the car-boxy-terminal domain and exclusively on serine residues (41). The

I 367

MAMMALIAN

CELL

MEMBRANE

BIOSYNTHESIS

367

role of phosphorylation in modulating the in viva function of CCT is still unclear, although several publications report an inverse correlation between enzyme phosphorylation and membrane-associated enzyme activity in viva (42). One hypothesis is that CCT phosphorylation regulates its membrane association and activity. Under conditions where PtdCho biosynthesis is reduced, the phosphorylated enzyme is released from digitonin-permeabilized cells and presumably cytosolic (19). Additional studies challenge this idea and indicate that either there is a lack of strict correlation between CCT phosphorylation status and activity (43) or that alteration of the membrane lipid composition triggers CCT binding and dephosphorylation (44). Also, we now know that the soluble CCT is largely nuclear (26). Hyperphosphorylation of the CCT carboxy-terminal domain is reminiscent of a similar modification of RNA polymerase II (45), suggesting that both enzymes may be substrates for the same kinases or may be modified in response to the same signal. A second isoform, called CCTR has been identified (46) and it is encoded by a second gene, called PCYTlB, located on the X chromosome in mice and humans (M. Karim and S. Jackowski, unpublished results). The CCTP amino terminus does not resemble that of CCTo and lacks a nuclear localization sequence. Fluorescent in situ immunolocalization studies show that both R isoforms are associated exclusively with the endoplasmic reticulum, whereas CCTo is found both in the nucleus and in association with the endoplasmic reticulum (26). The catalytic core of CCTR is nearly identical to the catalytic core of rat and human CCTa and 64% identical to the catalytic region of yeast CCT. Th e (Yand l3 isoforms are also almost identical in the helical domain adjacent to the catalytic core. CCTor and CCTp share 88% amino acid identity between residues 256 and 288, with conservative substitutions at the three remaining residues. CCTR activity is regulated by association with lipids, as would be predicted by the high degree of homology with CCTa in the helical region, and overexpression of CCTP accelerates PtdCho biosynthesis (26, 46). Two splice variants of CCTl3 are expressed, CCTl31 and CCTP2; these are identical for the first 320 residues but differ in the length and primary sequence of their carboxy termini (26). The CCTf31 protein is 330 amino acids long whereas CCTp2 is 369 amino acids long. The shorter CCTRl contains only two phosphorylation sites whereas the phosphorylation domain of CCTl32 contains 21 potential phosphorylation sites (19 serines and 2 threonines). CCTP2 is highly phosphorylated in vizjo and CCTRl is also a phosphoprotein, although the degree of its modification is barely detectable compared to CCTR2 and CCTa. CCTo is ubiquitously expressed in all tissues examined thus far and, by contrast, CCTR is highly expressed in testis, ovary, and brain. The CCTp2

368

ATHANASIOSLXKIDISANDSUZANNEJACKOWSKI

isoform is expressed at a high level in brain whereas CCTBl predominates in most other adult and fetal tissues. Neither CCTBl nor CCTB2 is detected in adult lung tissue, which is surprising in light of the large demand for PtdCho synthesis for surfactant production. A specific cellular function for the CCTp isoforms is unknown at this point, but the (Y, Bl, and B2 proteins, when overexpressed, complement a cell line, CT58, conditionally defective in CCT activity (26). Th ese data suggest that the B isoforms function to supply CDP-Cho, which is necessary for PtdCho biosynthesis, survival, and growth of the cells. The CCTCY mutation in CT58 cells is known (47); however, it is not known whether CCTB is also mutated in this cell line. By contrast, overexpression of PtdEtn iV-methyltransferase (PEMT) activity, which produces PtdCho but by a different mechanism, does not rescue the cellular phenotype (48). CCTB proteins have lower activity in v&o compared to CCTa (26,46), h owever, and overexpression may compensate for what otherwise might be insufficient activity expressed from the endogenous gene copy number. Supporting this concept, overexpression of truncated CCTol proteins with significantly reduced activity in vitro (37, 39) stimulates PtdCho biosynthesis (49) and complements the CT58 conditionally defective cell line (50). Possible functions for the CCT isoforms can be hypothesized from their tissue-specific distribution and subcellular localization. CCTf32 is highly expressed in brain and may play a role in neuronal development and function because active PtdCho biosynthesis is critical for axon elongation (51), which often occurs at great distance from the nucleus. CCTBl is highly expressed in placenta and may be involved in providing PtdCho for the developing embryo. Until genetically engineered cell lines or animals that do not express the CCTB isofonns are developed, we can only speculate about the specific function(s) of these proteins. The ubiquitous expression of CCTCI suggests that this isoform performs a basic function in cellular regulation. The significance of its accumulation in the nucleus has not been established. The nuclear localization may be necessary for synchronization of CCTa activity with cell cycle-regulated events or participation in nuclear signal transduction pathways. The fact that the nuclear CCTo is not associated with membranes might imply an alternative hypothesis: the nucleus may be a repository of inactive enzyme that, on stimulation, exits the nucleus and associates with membranes and becomes active (52). 2. SUBCELLULARLOCALIZATIONOF

CCT ISOFORMS

Conflicting data have spawned a debate concerning the subcellular localization of CCT. Biochemical and gradient sedimentation experiments indicate that CCT activity is associated with the endoplasmic reticulum and Golgi membranes (53, 54). Later indirect immunofluoresent light microscopy, using a specific antipeptide antibody directed against the amino

MAMMALIAN

CELL MEMBRANE

BIOSYNTHESIS

369

terminus of CCTa, shows that the majority of the CCT is within the nucleus (55). Using an antibody raised against a peptide corresponding to the CCTo membrane interaction domain and immunoelectron microscopy, CCT is found both in the nucleus and the endoplasmic reticulum (56). The discovery of the CCTR isoforms reveals that the membrane interaction domains of CCTa and CCTR are identical, suggesting that the antibody raised against CCTa membrane interaction domain may cross-react with CCTR. Most recently, antibodies specific for each CCT isoform were developed to investigate their subcellular localization using direct immunof luorescent confocal microscopy (26). Confocal microscopy is superior to conventional techniques because it allows the observations to be made at the same focal plane, resulting in a significant increase in the resolution of the subcellular structures. Also, direct labeling of the affinity-purified immunological probes with multiple copies of a fluorescent tag enhances sensitivity, allowing detection of low endogenous protein levels. Using this approach, CCTa is found both in the nucleus and associated with the endoplasmic reticulum. In contrast, the CCTR isoforms are exclusively associated with the endoplasmic reticulum. Experiments using an antibody that recognizes the helical region of both CCTa and CCTR show that the amount of CCTCY in the nucleus decreases relative to the CCT signal associated with the endoplasmic reticulum when IIC9 cells are stimulated with serum (52). These data suggest that fluctuations in the nuclear CCT level may be associated with changes in growth status. The CCTp isoforms, on the other hand, are maintained outside of the nucleus with or without serum (52)~their specialized function, if any, remaining obscure. 3. LIPID REGULATION

OF

CCT

Purified CCT in the presence of Triton X-100 requires addition of anionic lipid in the assay for maximum catalytic activity (57). Removal of the detergent by diethylaminoethyl (DEAE) c h romatography inactivates CCT, and readdition of mixtures of PtdCho with either anionic or neutral lipids reconstitutes an active enzyme (38,58,59), thus demonstrating a strict dependence on lipid for maximal activity. These data suggest that soluble CCT assayed in cytosolic fractions of crude cell extracts retains tightly bound lipid, and, in fact, purified CCT preparations often contain lipid bound to the protein (58). Anionic lipid activators include oleic acid, phosphatidic acid, phosphatidylglycerol, or phosphatidylinositol. PtdChoanionic lipid mixtures are considered to be the most potent combination whereas PtdCho alone is not effective (58). Similarly, diacylglycerol alone or in detergent cannot activate CCT but synergizes with anionic lipids in PtdCho vesicle mixtures to support CCT enzymatic activity at lipid concentrations that more closely approximate physiological levels (60). CCT interaction with PtdCho lipid mixtures is mediated by the helical re-

370

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

gion of the protein spanning residues 257 to 297. Proteolytic removal of this region from CCTol results in an enzyme refractory to lipid activation (61). Also, the activities of CCTa proteins truncated either at residue 236 or residue 257 are lipid independent (37, 49) but can be assayed at high CTP concentrations (37, 50). Th ere is no evidence indicating a significant difference between a CCTol protein truncated at residue 236 compared to a protein truncated at residue 257, and so it is unlikely that the intervening sequence has a role in lipid binding and activation. There are two models to explain the mechanism of CCT activation by lipid (37,49,50). Kent and coworkers propose that the helical domain is inhibitory to CCT catalysis and that occupation of the residues in this domain with lipid removes them from constraining the catalytic domain (49). In this model, the enzyme activity is essentially the same for the full-length protein in the presence of activating lipids compared to the truncated protein measured in lysates from stable transfectants. In contrast, Jackowski and co-workers, using CCT protein isolated from either mammalian cells following transient overexpression or from baculovirus-infected cells prior to purification to near homogeneity, find that the truncated protein is less active than the full-length CCTol plus lipid at CTP concentrations near the kinetic Km. Higher CTP concentrations promote the activity of both the truncated protein and the full-length protein in the absence of lipid. Thus the second model proposes that the binding of the helical domain to lipid induces a conformational change that, in turn, alters the enzyme’s affinity for CTP substrate (37). The difference between the two models is based on the interpretation of in vitro enzyme specific activity measurements that are dependent on the CTP concentration and other conditions in the assay, the amount and the nature of endogenous lipid that is bound to the CCT, and the accuracy of the CCT protein concentration. Future work on the structure of CCT will establish the validity of each model. Although the function of the phosphorylated carboxy-terminal region of CCT in vivo is not clear, kinetic analysis of a truncated protein lacking this domain reveals its influence on enzyme activity in vitro (62). Lipid activation of the full-length phosphoprotein exhibits negative cooperativity. Removal of the carboxy-terminal domain removes the negative cooperativity of lipid activation, indicating that this domain is responsible for interference with lipid binding. The protein truncated after residue 3 11 responds to increasing lipid concentrations in a classical Michaelis-Menten fashion, resulting in a more active protein at a given concentration of activator. The full-length protein used for these experiments is isolated from baculovirus-infected cells and thus the carboxy-terminal domain is highly phosphorylated (41). It is likely that multiple phosphorylation sites on the carboxy-terminal domain act in concert to accelerate or deccelerate the enzymatic response to lipid, rather

MAMMALIAN

CELL MEMBRANE

371

BIOSYNTHESIS

than providing an on/off switch to regulate activity. Extensive phosphorylation does not completely inactivate CCT, and dephosphorylation increases activity of the full-length protein (39, 63). Unfortunately, extensive kinetic analysis of the dephosphorylated enzyme and its regulation by lipids is not available. 4. CELL CYCLE

REGULATION

OF PTDCHO

METABOLISM

Cells maintain a characteristic composition of phospholipids and meet the challenge of doubling the phospholipid mass while maintaining the correct composition prior to cytokinesis. Disturbance of the amount or composition of phospholipids inherited by the daughter cells by even a fractional percent would rapidly result in cells with a major excess or deficit of membrane surface, or with membranes having altered composition of phospholipids and, consequently, altered biophysical properties. Therefore, regulatory mechanisms exist that allow cells to coordinate their phospholipid metabolism with their life cycle. These molecular mechanisms are largely unknown, although recent progress in our understanding of the coordination of phospholipid metabolism with the cell cycle enables the development of a working hypotheses and suggests likely points of regulation (see Fig. 3). The stimulation of growth in quiescent cells accelerates PtdCho biosynthesis (52, 64, 65), CCTa mRNA abundance (65, 66), and total CCT specif-

Cell Division Start Synthesis Start Degradation

Rapid Phospholipid Turnover

ion

FIG. 3. Phospholipid metabolism through the cell cycle. High rates of phospholipid s)mthesis and degradation characterize the G, phase. At the G,/S boundary degradation ceases whereas synthesis continues. This results in phospholipid accumulation during the S phase. Phospholipid metabolism pauses during G,/M phases.

372

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

ic activity in vitro (65). The stimulation of PtdCho biosynthesis is accompanied by a loss of CCTa in the nucleus and an increased CCT protein signal in the endoplasmic reticulum in serum-starved IIC9 cells (52). Newly translated CCTol and CCTR protein associates first with the endoplasmic reticulum and, whereas CCTP protein remains with the endoplasmic reticulum, CCTol protein migrates and accumulates in the nucleus (52). The stimulation of PtdCho biosynthesis immediately after mitogenic stimulation does not result in accumulation of membrane, but rather increased PtdCho degradation maintains a constant membrane mass (63). The G,/G, phase represents both the cellular response to growth factor stimulation and the initiation of cell cycle progression. Rapid PtdCho turnover is a periodic event characteristic of the G, phase during cell cycle progression and has been demonstrated to be independent of growth factor stimulation in cells entering the second G, phase and those released from an M phase block (63). Equilibrium phospholipid labeling and mass measurements indicate that, in contrast to the events of G, phase, the net accumulation of membrane phospholipid occurs during S phase due to significant reduction of PtdCho degradation while PtdCho synthesis continues (63). The regulation of phospholipid degradation, particularly at the G,S boundary, is a major requirement for the maintainance of phospholipid homeostasis together with the regulation of phospholipid synthesis. During the later G, and M phases, phospholipid metabolism diminishes progressively as biosynthesis reaches the lowest rate during the cell cycle (63). Experiments with several mammalian cell lines such as CSHlOT,,, preadipocytes (64), HeLa cells (67), mast cells (68) thymocytes (69), and fibroblasts (70) indicate also that membrane phospholipid accumulation occurs during the S phase. CCTo enzyme activity oscillates during the cell cycle (63), increasing to a maximum level in mid-G,, then declining gradually during S phase to reach its lowest point when cells enter G,/M phase. CCTa activity likely also increases in response to mitogenic stimulation due to an immediate early elevation of the mRNA level (65), contributing to the enhanced CCT activity during the first cell cycle. The long-term periodic variation in CCT activity is not caused by changes in the expression of CCT mRNA or alterations of the CCT protein levels (63). Rather, CCTa protein is dephosphorylated early in G, and is then progressively phosphorylated on multiple sites within the carboxy-terminal domain, beginning in late G, and continuing through to S phase and into G,. Virtually all of the protein is hyperphosphorylated in M phase (63). The correlation between variation in CCTa enzyme activity and the modification of the protein strongly suggests that a lack of phosphorylation enhances enzyme activity whereas phosphorylation reduces activity. These observations are consistent with the in vitro kinetic analysis of CCTa, whereby truncation of its phosphorylated carboxy-terminal domain enhances

MAMMALIAN

CELL MEMBRANE

BIOSYNTHESIS

373

activity (62), or in vitro enzymatic dephosphorylation of the protein increases activity (39). As th e cells enter the G, phase of the second cell cycle, CCTol is dephosphorylated and activated. The cell’s decision to double its phospholipid mass is made early in G, phase because aphidicolin, an agent that inhibits DNA replication, does not prevent accumulation of phospholipids, although it effectively blocks S phase (63). Also, cyclic AMP analogs that cause growth arrest in mid-G, (71) d o not cancel phospholipid accumulation (63), indicating that cells make the decision to double their membrane mass either early in G, or in the later stages of the preceding cell cycle (63). The precise molecular signals that govern the coordination of phospholipid biosynthesis with the cell cycle remain unknown. The periodic nature of CCTo activity and phosphorylation suggests that cell cycle-regulated kinases and phosphatases modulate the activity of the enzyme. Consistent with this concept, CCTo is phosphorylated in vitro by a cyclin kinase and a MAP kinase (72, 73). These data do not prove the concept, though, because protein kinase A can use CCTa as a substrate in vitro (74), whereas in vivo CCT is not phosphorylated at the predicted site (75). The discovery of the p isoforms adds a new dimension to the problem and their contribution to the regulation of PtdCho biosynthesis is unknown. CCTf32 is a phosphoprotein and its activity is likely regulated by phosphorylation in a manner similar to that of the (Yisoform. On the contrary, CCTPl lacks the phosphorylation domain, suggesting that it is refractory to regulation by kinases and phosphatases. Reevaluation of a large body of previous data is necessary in light of the existence of the CCTR isoforms, because many of the studies used an antibody to the helical region of CCTcx that cross-reacts with the identical region of CCTR. In particular, those investigations that biochemically address the translocation of CCTo between cytosol and membrane may have measured different expression levels of CCTa, which is primarily nucleosolic (55), compared to CCTR, which is primarily associated with the membrane of the endoplasmic reticulum (26). To date, the available data suggest that the molecular control points for phospholipid homeostasis reside with CCT to supply an abundance of PtdCho molecules for increasing the membrane, and with a phospholipase for the controlled degradation that limits the number of PtdCho molecules.

C. CDP-Choline: 1,2-diacylglycerol Phosphotransferase CPT catalyzes the last reaction in the PtdCho pathway. CPT is a membrane-bound enzyme that has resisted purification efforts. The first mammalian cDNA reported encodes a predicted protein of 416 amino acids with seven membrane-spanning domains (76). Mutational analysis of the

374

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

yeast qt gene defines a CDP-alcohol phosphotransferase motif, DG(X), AFI(X),G(X),D(X),D, which is also present in the mammalian protein (77). The subcellular localization of the CPT protein is of special interest because it defines the cellular site of membrane biogenesis. Most subcellular fractionation studies conclude that the bulk of the CPT activity resides on the endoplasmic reticulum, but CPT activity is also detected in association with Golgi, mitochondrial, and nuclear membranes, depending on the method used to isolate these organelles (54). More detailed information awaits the development of antibodies based on the reported cDNA sequence. The role of CPT in the regulation of PtdCho biosynthesis is poorly understood. The CPT reaction is readily reversible (78) and th us could diminish the PtdCho content and produce diacylglycerol. Elevation of diacylglycerol levels following growth factor stimulation are generally thought to be due to activation of a PtdCho phospholipase C activity (79- 81), but the reverse CPT reaction would yield the same diacylglycerol product (78) that is monitored in such experiments. One would predict that the substrate specificity of the CPT would reflect the acyl-chain composition of the predominant diacylglycerol (80- 82) and PtdCho molecular species (83). Inhibition of PtdCho synthesis via the inhibition of CPT by treatment of cells with famesol (84) or shortchain ceramides (85) triggers apoptosis, thus demonstrating the essential role of this enzyme in maintaining phospholipid homeostasis.

III. Regulation of Phosphatidylethanolamine Biosynthesis There are two biosynthetic routes to PtdEtn: the CDP-ethanolamine pathway and the decarboxylation of PtdSer (Fig. 1). The relative contribution of each route to the total amount of PtdEtn is an enigma (86). The mammalian gene encoding PtdSer decarboxylase (PSD) has been cloned and posttranslational processing is necessary to obtain a functional protein (87). PSD is associated with the inner mitochondrial membrane (88) and PtdSer is imported into the mitochondria (89-91) by an ATP-dependent translocation (92) prior to PSD-catalyzed conversion to PtdEtn (93). The CDP-ethanolamine pathway is similar to the CDP-choline pathway for PtdCho biosynthesis but has not been studied in as much detail. The CDP-ethanolamine pathway is dependent on addition of ethanolamine to culture medium and so most cell lines produce PtdEtn by PtdSer decarboxylation (94). On the other hand, PtdEtn biosynthesis following partial hepatectomy is dependent on the exogenous ethanolamine concentration (95). Increasing ethanolamine concentrations elevate phosphoethanolamine levels, while the CDP-ethanolamine level remains constant (96), suggesting that

MAMMALIAN

CELL

MEMBRANE

BIOSYNTHESIS

375

the ethanolamine cytidylyltransferase (ECT) is rate limiting to the pathway. The CDP-ethanolamine pathway of PtdEtn biosynthesis is similar to the CDP-choline pathway for PtdCho (Fig. 1). Three enzymes make up the pathway: ethanolamine kinase, CTP:phosphoethanolamine cytidylyltransferase, and CDP-ethanolamine:1,2-diacylglycerol phosphotransferase.

A.

Ethanolamine

Kinase

It has been proposed that the EK and CK activities in mammalian cells reside on the same protein because nearly homogeneous CK preparations from rat kidney (2) and liver (1) exhibit significant EK activity. However, numerous studies report the biochemical characterization of kinases with high affinity for ethanolamine. A 42-kDa EK has been purified from human liver (97) and a high- mo 1ecular-mass (84-kDa) EK has also been described (98). Drosophila (99) and yeast (100) p assess specific EK enzymes, underlining the possibility that mammals also possess specific EK proteins. The final answer awaits the characterization of mammalian cDNAs that encode ethanolaminespecific kinases. Isolation of the Drosophila easily-shocked (us) cDNA (99) presents a clue hinting at a critical role for PtdEtn in neural function and provides a tool for identifying a mammalian EK clone. The eas gene encodes a 495-amino acid protein that exhibits EK activity. A Drosophila mutant with a defective eas gene has a slow response to loud noises but the mechanism by which malfunction of the EK could cause the phenotype is not known, although the PtdEtn level is reduced.

B. CTP: Phosphoethanolamine

Cytidylyltransferase

ECT is a cytoplasmic protein that preferentially localizes to areas rich in rough endoplasmic reticulum (101). The cDNA encoding human ECT is known (102) and North em blot analysis reveals that ECT is widely expressed in human tissues. Human ECT (102) has 389 amino acids and contains two repetitive sequences, one at the amino terminus and the other at the carboxy terminus. The repetitive sequences have high homology with the catalytic core of the CCT proteins and probably constitute the catalytic core(s) of the ECT enzyme. These two catalytic domains in ECT are connected by an intervening region of 44 amino acids located in the center of the primary sequence. The rat cDNA sequence (103) predicts a protein that is homologous with the human ECT but contains an additional 18 amino acids in the connector region between the catalytic domains. There is no information about the regulation of ECT. In contrast to CCT, ECT is not regulated by lipids (104). The connector domain between the catalytic portions of the protein could be envisioned as a regulatory region, perhaps having a role in controlling the interaction between the two catalytic domains. Active ECT may adopt a conformation that resembles a dimer be-

376

ATHANASIOS LYKIDIS AND SUZANNE JACKOWSKI

tween the catalytic regions, similar to CCT, which functions as a homodimer (105). The connector region includes 8 serines for human ECT and 12 serines for rat ECT; these potential phosphorylation sites may govern the conformation of the connector region. It is not yet known whether ECT protein is phosphorylated in vivo. C.

CDP-Ethanolamine:

1,2-diacylglycerol

Phosphotransferase EPT catalyzes the final step in PtdEtn biosynthesis. Mammalian EPT has not been purified from any source nor has a cDNA been cloned. The mammalian CPT cDNA also exhibits EPT activity (76) when expressed in yeast, although its involvement in PtdEtn biosynthesis in mammalian cells remains to be determined. The yeast gene encoding EPT utilizes both CDPethanolamine and CDP-choline in vitro (106), but labeling studies reveal that the EPT enzyme uses CDP-ethanolamine almost exclusively when expressed in cells (107). The yeast EPT accounts for only 5% of the net PtdCho biosynthesis and this percentage decreases when ethanolamine is added to the growth medium (108). Th e mammalian EPT activity associates with the microsomal fraction (109), but a more precise localization awaits the development of specific antibodies.

IV. Regulation of Phosphatidylinositol Biosynthesis The PtdIns biosynthetic pathway (Fig. 1) consists of three enzymatic activities: a diacylglycerol kinase (DGK), a CDP-diacylglycerol synthase (CDS), and phosphatidylinositol synthase. Diacylglycerol and phosphatidic acid (PtdOH) are intermediates in this pathway and also function in signaling pathways. DG is the endogenous activator of protein kinase C (110) and PtdOH is the metabolic derivative of lysophosphatidic acid, an extracellular mitogenic ligand (111,112). DG is also a precursor in both PtdCho and PtdEtn biosynthetic pathways (Fig. 1) as well as in triacylglycerol synthesis (113), and these three pools are very large compared to PtdIns. PtdOH arises not only from exogenous lysophosphatidic acid but also from endogenous phospholipid synthesis preceding the formation of PtdCho, PtdEtn, and triacylglycerol. Another source of PtdOH is the hydrolysis of phospholipids by phospholipase D (PLD). PLD activity is widely distributed in mammalian tissues and there are at least two isoforms, termed PLDl and PLD2 (114). There is a considerable body of evidence indicating that PLDl is regulated by protein kinase C, heterotrimeric G proteins, and ADP-ribosylation factors (114). PtdIns biosynthesis would thus be under the control of agonists and signal transduction pathways that stimulate PLD activity. PtdIns constitutes less

MAMMALIAN

CELL MEMBRANE

BIOSYNTHESIS

377

than 1% of the total phospholipid in most cells, and so diversions of DG and PtdOH to PtdIns are likely control points, restricting the incorporation of these intermediates and resulting in a very small PtdIns pool.

A. Diacylglycerol

Kinase

The mammalian DGK family of enzymes includes nine different isoforms that have been cloned and characterized (01,R, y, 6, E, 5, +r-l, 8, L). DGK contains distinct domains postulated to regulate the catalytic activity and mediate protein:protein interactions, and the isoforms have been classified into five different subtypes based on their structural similarities. A review by Topham and Prescott describes current information regarding DGK (115) and so its regulation will not be discussed at length in this text. Most reports study DGK from the perspective of signaling because the conversion of DG to PtdOH represents a potential regulatory switch that can govern cellular responses. However, little information is available on the role of DGK in PtdIns biosynthesis.

B. CDP-Diacylglycerol

Synthase

CDS is the second enzyme in PtdIns biosynthesis and catalyzes the formation of CDP-diacylglycerol (CDP-DG) from PtdOH and CTP CDS protein is at a branchpoint of phospholipid metabolism and is proposed to be at two subcellular locations: the inner mitochondrial membrane, where it is involved in cardiolipin and phosphatidylglycerol biosynthesis (116), and the cytoplasmic aspect of the endoplasmic reticulum, where it functions in the PtdIns biosynthetic pathway. The CDS that associates with the endoplasmic reticulum constitutes about 90% of the total cellular activity (117) and is activated by GTP (117-119). H owever, all nucleotides, in fact, activate CDS1 expressed from the cloned cDNA, ruling out a specific GTP-mediated regulatory scheme (120). The CDS1 cDNA was cloned from human (121) and rodent (122) sources and a second cDNA, called CDS2, encodes a protein with high homology to CDS1 (123). CDS1 and CDS2 are both highly similar to the Drosophila isoform selectively expressed in photoreceptor cells (124). The human CDS1 gene is located on chromosome 4q2 1.1 whereas the CDS2 gene is found on chromosome 20~13 (125). Northern blot analysis reveals CDS2 to be widely expressed at approximately the same levels in all tissues examined, and thus appears to have a housekeeping function (125). In situ hybridization shows high expression in differentiating neuroblasts of the neural retina and in the central nervous system during embryonic development. However, CDS2 is not expressed in the adult retina. In contrast, CDS1 is highly expressed in fetal lung, kidney, and brain but its mRNA levels are lower in fetal liver (120). In situ hybridization indicates that CDS1 is the isoform highly expressed in the photoreceptor layer of adult retina (125), and the enzyme

378

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

localizes to the endoplasmic reticulum, vesicles, and nuclear envelope consistent with the observation that CDS1 lacks a mitochondrial signal (122). It will be interesting to determine whether CDS2 is localized specifically to mitochondria, although sequence analysis does not reveal a recognizable mitochondrial localization signal. CDS, the second enzyme in the PtdIns biosynthetic pathway, catalyzes a cytidylyltransferase reaction (Fig. 1). The striking similarity with the PtdCho and PtdEtn pathways led to the proposal that CDS is rate determining in PtdIns biosynthesis by analogy (126). However, CDS1 overexpression does not alter either the rate of PtdIns synthesis or the total amount of cellular PtdIns (120), ar gu’ m g a g ainst a regulatory role for this enzyme. The CDP-DG level also remains constant when CDS1 is overexpressed, although the specific activity of the CDS1 enzyme is increased significantly. Coexpression of the CDS1 and PIS cDNAs also does not alter PtdIns metabolism. These data suggest an alternative hypothesis, that the supply of PtdOH precursor is limiting to the pathway and that DGK is the candidate enzyme for overall regulation of PtdIns biosynthesis. Among the many DGK isoforms, however, it is difficult to predict the isoform that would be important for this task. CDS may have a role in maintaining PtdIns for use in cellular signaling. Cell lines overexpressing CDS1 by twofold secrete more tumor necrosis factor-o and interleukin-6 in response to interleukin-1R stimulation (127). These results suggest that elevated CDS1 activity, although not sufficient for altering homeostatic PtdIns metabolism, can perhaps amplify a cellular signalling response. Experiments in Drosophila support this idea. Mutants defective in an eye-specific CDS isoform cannot sustain a light-activated current because of rapid depletion of PtdIns-4,5-bisphosphate (124). Future work with the mammalian CDS clones may reveal specific roles for these isoforms in cells wherein the PtdIns cycle is a major component of the signaling response.

C. Phosphatidylinositol

Synthase

PIS catalyzes the final step in PtdIns biosynthesis, attaching inositol to the hydrophobic CDP-DG to yield PtdIns. PIS, a membrane-bound enzyme, is located primarily at the cytoplasmic aspect of the endoplasmic reticulum (128) but it is also found in plasma membrane preparations (129-132). PIS protein has been purified from human placenta with an apparent molecular size of 24 kDa (133) and from rat liver with a size of 21 kDa (134). The PIS cDNA was isolated from rat (135) and human (120) sources, and a single 2.1kb PIS mRNA is expressed uniformly in all tissues, consistent with the idea that PIS is a housekeeping gene. A PtdIns:inositol base-exchange reaction (132) mediated by the PIS working in reverse may play a role in PtdIns biosynthesis. In the exchange reac-

MAMMALIAN

CELL MEMBRANE

BIOSYNTHESIS

379

tion the inositol moiety of the PtdIns product is replaced by a different inositol molecule in the absence of the CDP-DG intermediate. The reverse PIS reaction of the mammalian protein is the same as that described in yeast, which is CMP dependent (136-138). A CMP-independent exchange activity can also be detected but is due to endogenous CMP that is tightly bound to the PIS protein (120), arguing against the existence of a second isoform (138, 139). PIS overexpression does not increase the rate of PtdIns synthesis nor the amount of PtdIns in the membrane, even when expressed in combination with the CDS1 clone (120). Also, PIS activity is not inhibited by its product PtdIns (130) b ecause overexpression of CDS1 does not result in accumulation of the CDP-DG intermediate (120). On the other hand, PIS overexpression results in an increase in the exchange reaction in membranes, and this activity is significantly stimulated by CMP (120). These results support the idea that PIS does not have a direct regulatory role in the PtdIns pathway, but do suggest that cellular CMP levels may modulate PtdIns biosynthesis.

V. Regulation of Phosphatidylserine Biosynthesis Cellular PtdSer is present in low amounts relative to PtdCho, PtdEtn, and SM and is located primarily in the cytoplasmic leaflet of membranes (140). PtdSer can be synthesized by a base-exchange reaction either from PtdCho or from PtdEtn (Fig. 1) in a reaction catalyzed by phosphatidylserine synthase (PSS). There are two genes encoding PSS activity. PSSl catalyzes the exchange reaction between serine and both PtdCho and PtdEtn, whereas PSS2 uses only PtdEtn as substrate. The enzymes have not been purified but the cDNAs from CHO cells (142, 142) an d mouse cells have been cloned (143). Addition of exogenous PtdSer to the culture medium inhibits de nova PtdSer biosynthesis (144), suggesting that a feedback control mechanism regulates PtdSer production. A mutant cell line was isolated that is highly resistant to feedback inhibition by PtdSer (144~) and it exhibits a threefold higher rate of PtdSer biosynthesis. The resistance to PtdSer inhibition is due to a mutation in the PSSl gene (144b). Overexpression of the PSSl protein that is resistant to PtdSer inhibition causes a fivefold increase in the PtdSer biosynthetic rate and a twofold increase in the cellular PtdSer level. Corresponding mutation of the PSS2 protein results in a protein that is also refractory to PtdSer inhibition, and overexpression of this mutant protein causes a fourfold increase in the biosynthetic rate of PtdSer (145). Thus, both PSSl and PSSB, which catalyze PtdSer biosynthesis, are controlled by feedback inhibition by PtdSer and are key factors that control cellular PtdSer levels,

380

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

VI. Cells Maintain a Constant Amount of Membrane Phospholipid The phospholipid pool normally undergoes rapid turnover in proliferating cells as the result of active degradation coupled with resynthesis (63). Disruption of synthesis and the inability to replenish membrane mass lead to apoptosis (38,146-X2), demonstrating that both metabolic processes are essential for cell survival. Cells maintain their membrane characteristics throughout this dynamic turnover and retain their phospholipid content and composition. At the same time, cells readily take up lipids from serum and from culture medium (153), and so the question arises as to the pathway for phospholipid degradation and the metabolic fate of the excess phospholipid. When cells are supplemented with exogenous phospholipid, the response is an efficient breakdown of the extra lipid (153). Elevated PtdCho synthesis is also balanced by increased degradation in a number of other physiological situations. Transformation of C3HlOT,,, mouse fibroblasts with the H-ras oncogene results in both accelerated PtdCho formation and accelerated degradation in keeping with the accelerated growth rate (154). Both phospholipid synthesis and degradation increase during G, phase without a significant change in the total phospholipid content of cells (63). Overexpression of CCTor or CCTB increases the rate of PtdCho synthesis but also does not result in a significant change in phospholipid content (26, 46, 153, 155). Phospholipids are degraded three different ways and each process is mediated by a phospholipase that is defined by its function. Phospholipase C (PLC) enzymes yield DG and a phosphorylated base; phospholipase D (PLD) enzymes yield PtdOH and a free base; phospholipase A (PLA) enzymes cleave the fatty acyl moieties from the glycerol backbone to yield the glycerophospho base. PLA, enzymes remove the fatty acid from the m-2 position of the phospholipid and PLA, enzymes remove the fatty acid from the an-1 position. GPC is a metabolic end-product of PtdCho metabolism whereas phosphocholine and choline are also intermediates in the biosynthetic pathway leading to PtdCho (Fig. 1) and thus can be reincorporated into membrane phospholipid. Recent data show that GPC and phosphocholine are the two principal PtdCho breakdown products in cells (153), indicating that PLAand PLC-type mechanisms mediate basal turnover. Consistent with this idea, both the phosphocholine and the GPC levels are elevated in ras-transformed cells compared to nontransformed cells of the same derivation (18), although the 5-fold increase in phosphocholine has been attributed primarily to an increase in choline kinase activity (18). The GPC level increases about lo-fold in the ru.s-transformed cells (18) and is a better indicator of the elevation in PtdCho turnover rate because it is not reincorporated into the phospholipid pool (153). Additional information also demonstrates that the GPC exits cells

MAMMALIAN

CELL

MEMBRANE

BIOSYNTHESIS

381

and accumulates in the medium (153), therefore earlier measurements of cellular GPC levels underestimate the extent of PtdCho breakdown (18, 154, 155). The rate of phosphocholine release from PtdCho when cells are challenged with excess phospholipid remains the same, whereas the production of GPC increases dramatically (153). Thus, a deacylation reaction functions to maintain phospholipid homeostasis (153). If cells are supplemented with phospholipids with an ether linkage between the sn-1 fatty acid and the glycerol backbone rather than an ester linkage, these are no longer substrates for PLA (or lysophospholipase) enzymes and the phospholipid is not degraded (153). Rather, accumulation of the nonmetabolizable phospholipid causes growth arrest, although the cells retain their morphology and do not enter programmed cell death (153). The results suggest that degradation of excess phospholipid is essential for cell proliferation. The PLA, family of enzymes is continuing to grow and PLA, activities participate in signal transduction, eicosanoid synthesis, platelet-activating factor synthesis, and phospholipid remodeling. PLA, enzymes are classified in nine groups based on their molecular properties (156, 157). Groups I, II, III, V, and IX comprise the secretory PLA, (sPLA,) proteins. Groups VII and VIII consist of the PLA, proteins involved in platelet-activating factor (PAF) metabolism, group IV includes the cytosolic Ca”+-dependent PLA, (cPLA,) enzymes, and group VI is characterized by the Ca2+-independent PLA, (iPLA,). Which of th ese PLA, enzymes is responsible for the degradation of excess phospholipid? The candidate enzyme is not likely secretory, because excess phospholipid generated from within the cell is controlled as well as phospholipid taken up from the medium. The PLA, enzymes involved in PAF metabolism probably are not involved in this homeostatic mechanism because PAF is synthesized in selected tissues. Also, these enzymes have specificity for the PAF molecule, which differs from most phospholipids in that it has an ether-linked fatty acid at the sn-1 position and an acetate moiety at the sn-2 position. The ideal enzyme would recognize PtdEtn as well as PtdCho (153), would not require a signaling mechanism for activation, and would exhibit lysophospholipase activity in addition to PLA, activity. The cPLA, is a mitogen-responsive enzyme and is not responsible for the enhanced phospholipid turnover accompanying increased biosynthetic rates, based on a lack of correlation between the kinetics of activation of the enzyme and measurements of PtdCho turnover (158). The group VI iPLA, enzymes are involved in phospholipid remodeling, whereby a fatty acid is cleaved at the sn-2 position from a preexisting phospholipid molecule to generate a lysophospholipid that subsequently is reacylated by a different fatty acid to yield a new phospholipid. This cycle is believed to be of special importance for the incorporation of arachidonic acid into phospholipids. The

382

ATHANASIOS

LYKIDIS AND SUZANNE

JACKOWSKI

iPLA, also has a low specific activity in cells and is speculated to be a basal, housekeeping enzyme (159). Bromoenol lactone (BEL) is a specific inhibitor of the iPLA, (160) and is diagnostic for the identification of the iPLA, in contrast to other cellular phospholipases. The iPLA, can also utilize either PtdCho or PtdEtn as substrate (159, 161). BEL inhibits the formation of GPC from cell lines overexpressing CCT in a dose-dependent manner (X3), suggesting that the calcium-independent iPLA, mediates the breakdown of PtdCho and the excess phospholipid response. Stable cell lines overexpressing CCT constitutively exhibit increased iPLA, activity and an increase in the mass of the protein, in contrast to the cPLA, activity that remains constant (162). Experiments thus far point to the iPLA, as the enzyme that regulates membrane phospholipid content in response to elevated levels of CCT and provides an adaptation mechanism for long-term cell culture. However, the case for iPLA2 is largely circumstantial and more definitive experiments will be needed to establish its role in the excess phospholipid response. The PLA,-mediated breakdown of phospholipid is a generalized mechanism that is stimulated when cells are challenged with excess phospholipid. Cells have the ability to uptake exogenous lysophospholipids and convert them to the respective phospholipid species via an acylation step (153). Loading cells with 1ysoPtdCho stimulates production of both GPC and glycerophosphoethanolamine (GPE) (153). Also, loading cells with either IysoPtdCho or 1ysoPtdEtn results in GPC production (153), suggesting that PtdEtn is regulated in a manner similar to PtdCho. Both GPC and GPE can exit the cells but cells can also retain some level of GPC (or GPE) to maintain osmotic balance with the extracellular medium (163-165). Thus, the cellular shape changes induced by osmotic stress are buffered by the products of bulk phospholipid turnover and the rate of GPC exit from cells may be governed by a physiological response independent of membrane phospholipid metabolism.

VII. Inhibition of Phosphatidylcholine Biosynthesis Induces Apoptosis Evidence is accumulating that inhibition of PtdCho biosynthesis triggers apoptosis. It is generally thought that under most conditions CCT is the major regulatory enzyme of the CDP-choline pathway, and irreversible CCT inhibition results in a delay of cell cycle progression at the G, phase and apoptosis (146, 147, 152). Depletion of cellular PtdCho due to either a conditional defect in CCT (152) or inhibition of CCT by 1ysoPtdCho analogs (146,147) causes the characteristic DNA fragmentation and cytological degeneration associated with programmed cell death. A class of synthetic ether-linked

MAMMALIAN

CELL MEMBRANE

BIOSYNTHESIS

383

1ysoPtdCho analogs are selectively cytotoxic to cancer cells (166-172), and there is growing interest in these compounds because they do not directly target DNA and could potentially complement existing DNA-directed anticancer compounds. ET-18-OCH,, or 1-0-octadecyl-2-O-methyl-rac-glycero3-phosphocholine, is a nonmetabolizable 1ysoPtdCho analog and belongs to the first generation of ether lipids tested as growth inhibitors (173). A number of investigations have suggested primary targets for the antineoplastic ether-linked 1ystoPtdCho analogs (174 -177), but the inhibition of PtdCho biosynthesis can be distinguished as causative in their cytotoxic action (153). The block in the CDP-choline pathway can be bypassed by supplementing cells with PtdCho or 1ysoPtdCho in quantities sufficient to maintain membrane mass. Both PtdCho and 1ysoPtdCho are quantitatively taken up by cells and 1ysoPtdCho is readily converted to PtdCho by an acyltransferase following uptake (153). LysoPtdCho is more soluble in culture media and thus is the preferred supplement. Apoptosis resulting from the inhibition of the CDP-choline pathway is prevented by 1ysoPtdCho supplementation (147), but the 1ysoPtdCho must be added repeatedly on a regular basis because the rapid catabolism of membrane PtdCho (63) requires replenishment in proliferating cells. LysoPtdCho is used not only as a supplement to prevent apoptosis but also acts as a regulator of the CDP-choline pathway. LysoPtdCho inhibits CCT (38) but the rapid metabolic conversion of 1ysoPtdCho to PtdCho eliminates the need for the CDP-choline pathway, thus proliferation continues. The synthetic 1ysoPtdCho ether analogs, typified by ET-18-OCH,, inhibit CCT by the same biochemical mechanism (38,146), but are not metabolized (178), thus conferring persistent inhibition and concomitant arrest of cell growth. Negative regulation of the CDP-choline pathway is specific for IysoPtdCho and its synthetic analogs, because lysophospholipids, either with a different headgroup or lacking the choline headgroup, do not inhibit PtdCho synthesis (38). Metabolic radiolabeling patterns and enzymatic analysis suggest that CCT is the relevant physiological target for 1ysoPtdCho or a 1ysoPtdCho analog in the CDP-choline pathway, because both choline and phosphocholine, precursors to CDP-choline, accumulate while the CDP-choline decreases in cells treated with these compounds (38, 179). LysoPtdCho inhibits purified CCT enzyme activity in vitro, by kinetically competing with the lipid activator in the assay (38). CCT protein truncated at amino acid 256 lacks the helical membrane-binding domain and is resistant to inhibition by IysoPtdCho or 1ysoPtdCho analogs in vitro (38), supporting the idea that a direct physical interaction between the bilayer and the helical region of CCT modulates the inhibition of activity. Lipid activation or inhibition of CCT alters the affin-

384

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

ity of the enzyme for CTP substrate (37), and one might speculate that the action of 1ysoPtdCho analogs may be augmented by inhibitors of CTP synthesis in a chemotherapeutic context. CCT associates with the cytoplasmic aspect of the membrane bilayer (Fig. 4) and its activity is modulated by the interaction of an amphipathic helical domain with bilayer lipids (19,36,37,180,181). Positive or negative lipid regulators modify the curvature elastic stress of the bilayer, either promoting or preventing the intercalation of the CCT helical domain (18.2). Fatty acid or diacylglycerol activates CCT (58, 183), whereas 1ysoPtdCho or synthetic ether analogs of 1ysoPtdCho inhibit CCT (38,146) when presented to the enzyme as constituents of PtdCho lamellar vesicles. The influence of these regulatory lipids on both the positive and negative curvature elastic stress of PtdCho bilayers correlates well with CCT activity measurements (182) and with the ability of the enzyme to bind to PtdCho vesicles (58,183-187). This hypothesis proposes that the degree of physical interaction between the membrane structure and CCT protein has a determining role in the extent of activation In contrast to previously proposed mechanisms that address only the

FIG. 4. CCT interaction with phospholipid. Phospholipid promotes the formation of the amphipathic o-helical structure in the CCT membrane interaction domain. The hydrophobic residues are clustered together on one side of the helix and associate with the core hydrophobic acyl chains of the phospholipid molecules aggregated in the membrane leaflet or a lamellar vesicle The charged residues on the helix are also clustered and positioned opposite the countercharged moieties of the phospholipid, facing the aqueous environment. The polar headgroup of PtdCho provides both a negative charge associated with the phosphate and a positive charge associated with choline.

MAMMALIAN

CELL MEMBRANE

BIOSYNTHESIS

385

stimulation of CCT by lipids (60, 283), the curvature elastic stress hypothesis explains both positive and negative regulation of CCT. In contrast, electrostatic interaction between anionic lipids and the positively charged amino acid residues in the amphipathic helix (58, 60) does not explain the stimulation of CCT activity by uncharged lipids such as diacylglycerol and PtdEtn. The alteration of phospholipid headgroup packing (183) does not explain inhibition by 1ysoPtdCho or the synthetic ether analogs of IysoPtdCho, which have the same headgroup as PtdCho. Enforced overerexpression of CCT confers resistance to ET-18-OCH, (188), providing compelling genetic evidence that CCT is the relevant physiological target mediating the apoptosis induced by this class of antineoplastic agents. Heightened CCT expression or 1ysoPtdCho supplementation is equally effective in reversing the cytotoxic effect, but neither method of rescue allows cells to proliferate in the presence of the synthetic 1ysoPtdCho analog ET-18-OCH,. On the other hand, 1ysoPtdCho supplementation of a cell line with conditionally defective CCT restores proliferation as well as viability and morphology at the nonpermissive condition (189). Whereas interruption of PtdCho synthesis is the primary physiological imbalance accounting for the cytotoxicity of ET-18-OCH,, the drug likely interacts with a second target to mediate the cytostatic effect. The second target is anticipated to be independent of PtdCho biosynthesis. The proliferative capacity of HL60 cells treated with hexadecylphosphocholine, a less potent alkylphosphocholine, can be restored on 1ysoPtdCho supplementation (146), suggesting that the second target has a lower affinity for hexadecylphosphocholine compared to ET-18-OCH,. The increased sensitivity of transformed cells to inhibition of PtdCho biosynthesis can be explained several possible ways. Transformed cells are generally characterized by elevated rates of PtdCho metabolism. For example, cells transformed with the STC(190,191) or rus (154,192,193) oncogenes have significantly higher rates of PtdCho breakdown and resynthesis. These high rates of PtdCho turnover place an increased demand on the PtdCho biosynthetic pathway to maintain membrane phospholipid mass, thereby making the cells more sensitive to inhibition of the CDP-choline pathway. Tumorigenic cells may also have enhanced uptake or accumulation of the 1ysoPtdCho synthetic analogs (194, 195), or may express lower amounts of CCT (196) and thereby exhibit increased sensitivity. The last enzyme in the CDP-choline pathway is 1,2diacylglycerol cholinephosphotransferase (CPT), and inhibition of this step results in accumulation of CDP-choline and correlates with cellular apoptosis (84, 85,197199). A variety of cytotoxic drugs that induce apoptosis by other primary mechanisms also cause CPT inhibition, such as camptothecin, etoposide, and chelerythrine, and at least partial rescue from the cytotoxicity is achieved by

386

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

PtdCho supplementation (84, 197). The PtdCho protection was much greater in cells treated with chelerythrine or farnesol, which inhibit CPT directly (84, 198). Cellular acidification inhibits CPT indirectly (84) and often accompanies the onset of the apoptotic program, suggesting that inhibition of PtdCho biosynthesis may be a general feature of this important biological phenomenon. The Bcl-2 and Bcl-XL proteins are involved in the apoptotic response and, although their cellular functions not clearly defined, overexpression of these proteins protects cells from ET-18-OCHs-induced apoptosis (194), implying that there may be some interaction with a component of PtdCho biosynthesis. It is possible that the apoptosis results directly from a loss of membrane integrity due to reduction in the amount of PtdCho. Gross alteration of membrane structures, including fragmentation or blebbing and reduction in cell size, is a component of the morphological transition that characterizes the process. On the other hand, PtdCho is a metabolic precursor to sphingomyelin (SM), b ecause a phosphocholine headgroup is transferred from PtdCho to SM. Choline is an essential nutrient and deficiency induces apopReduction of the suptosis both in vitro and in whole animals (148,150,200). ply of PtdCho from choline deficiency, in turn, inhibits the biosynthesis of SM, resulting in an accumulation of ceramide (148). In support of this scheme, treatment of cells with hexadecylphosphocholine results not only in reduced PtdCho synthesis but also increased ceramide levels (201). Interestingly, treatment of cells with short-chain ceramides induces apoptosis (85, 148), activates a caspase (148), and also inhibits CPT and reduces the CPT activity in membranes (85). Addition of a caspase inhibitor protects cholinedeficient cells from apoptosis together with restoring the SM and ceramide levels. PtdCho levels, however, remain depressed (148), arguing that the inhibition of SM synthesis is critical to the apoptotic process.

VIII. Future Directions The availability of numerous cDNA clones encoding phospholipid biosynthetic enzyme will enable major advancement in our understanding of the enzymology and cellular regulation of membrane phospholipid metabolism. Purification of membrane-associated proteins is difficult and has been largely unsuccessful for mammalian enzymes of phospholipid metabolism. Heterologous cDNA expression, however, resolves the dilemma and provides large amounts of starting material for purification and kinetic analysis. Sitedirected mutagenesis also facilitates the identification of protein regulatory domains and can reveal their physiological relevance. The significant progress that has been made concerning the unique regulation of CCT activity

MAMMALIAN

CELL MEMBRANE

BIOSYNTHESIS

387

by lipids since the first cDNA clone was isolated is a clear example of the impact of molecular biology on the field. The identification of cDNAs encoding multiple CCT isoforms also raises the question as to their multiple function(s) in cells, if any, and their specialized roles in tissues of whole animals. Whereas CCT overexpression studies in cell lines can be a useful but blunt tool to probe its physiological role, isolation of the CCT genomic clones followed by homozygous deletion of the murine PCYTlA and PCYTlB genes will likely uncover the most critical, nonredundant tasks for the CCTa and CCTB isoforms, respectively. The first example of a gene knockout of a lipid biosynthetic enzyme is the PEMT (- / -) mouse, in which its absence exacerbates liver dysfunction when the choline supply is deficient. This result indicates the very specialized but important role for the alternative route of PtdCho biosynthesis mediated by PEMT PtdCho has multiple uses in animals, not only as the major component of cell membranes but also as a secretory product of lung and liver tissue, which underlines the importance of understanding the function of multiple CCT isoforms in whole animals. CCTa is expressed ubiquitously and so a PCYTlA(-/-) mouse may be unable to progress through fetal development. However, cell type-specific deletion using conditional gene targeting via the Cre-1oxP system will permit the analysis of CCTol isoform function in selected tissues. These examples of experimental approaches used to investigate CCT structure and function are a paradigm for applications to understanding the CK isozymes, the ECT and CPT isoforms, and other enzymes of phospholipid metabolism. It is anticipated that the function of nuclear CCTo may also be revealed using cell lines established from tissues lacking its expression. One intriguing idea is that nuclear CDP-choline has a role in cell signaling. Defects in nuclear signaling events, such as the generation and recycling of DG in response to thrombin or the nuclear PtdIns turnover associated with S phase, may be altered in cells lacking nuclear CCTa. Alternatively, defects in membrane synthesis or trafficking would indicate that the nucleus is an important site for CCTCY regulation of these processes. Perhaps CCTo is inactive when located in the nucleus, although the evidence for this is circumstantial, and the substrates for CCT, as well as the CDP-choline enzymatic product, are water-soluble, diffusible molecules. It should be noted that cells or tissues can compensate for loss of essential functions through enhanced expression or increased activity of alternative isoforms. Thus, nuclear CCTa may be a resevoir to provide necessary CCT to the endoplasmic reticulum when the supply of active enzyme is depleted. Loss of either CCT isoform by degradative mechanisms needs to be examined more actively as a mechanism for regulation, together with an investigation of the effect of CCT phosphorylation on the half-life of the protein. PtdCho biosynthesis is necessary for cell survival and inhibition of this

388

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

pathway is a likely component of the network of effector mechanisms that mediate programmed cell death. The mechanism of inhibition may be indirect in many cases, as with apoptosis caused by DNA-damaging agents or FAS ligand, but the inhibition of PtdCho synthesis may also be directly involved in some cases. Treatment of cells with nonmetabolizable CCT inhibitors causes a delay in cell cycle progression at the G, phase that is reminiscent of the checkpoint control exhibited by cells treated with ionizing radiation, arresting damaged cells with less than adequate DNA. By analogy, the PtdCho content must also be monitored to prevent cell division in the absence of sufficient membrane phospholipid for functional daughter cells. Characterization of the molecular mechanisms that survey cellular phospholipid content as well as coordinate phospholipid metabolism with the cell cycle remains a challenge.

ACKNOWLEDGMENTS We are indebted to Charles Rock and Richard Heath for their helpful comments during the preparation of the manuscript. This work was supported by NIH Grant GM45737, Cancer Center (CORE) Grant CA 21765, and the American Lebanese Syrian Associated Charities.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

T. J. Porter and C. Kent,J. Biol. Chem. 265,414 (1990). K. I&date, K. Furusawa, and Y. Nakazawa, Biochim. Biophys. Actu 836,119 (1985). K. Ishidate, Biochim. Biophys. Acta 1348,70 (1997). R. H. Hjelmstad and R. M. Bell,J. Biol. C/rem. 266,5094 (1991). R. H. Hjelmstad and R. M. Bell,J. Biol. Chem 265, 1755 (1990). C. J. Walkey, L. Donohoe, R. Bronson, L. Agellon, and D. E. Vance, Proc. Natl. Acad. Sci. U.S.A. 94, 12880 (1997). (1998). C. J. Walkey, L. Yu, L. B. Agellon, and D. E. Vance, j. Biol. Chem. 273,27043 J. K. Blusztajn, M. Liscovitch, C. Mauron, U. I. Richardson, and R. J. Wurtman,]. Neural Transm. St&. 24,247 (1987). T. Uchida and S. Yamashita,]. Biol. Chem. 267, 10156 (1992). T. Uchida,J. Biochem. (‘Bookyoj116,508 (1994). C. Aoyama, K. Nakashima, M. Matsui, and K. Ishidate, Biochim. Biophys. Acta 1390, 1 (1998). C. Aoyama, K. Nakashima, and K. Ishidate, Biochim. Biophys. Acta 1393,179 (1998). K. Hosaka, S. Tanaka, J. Nikawa, and S. Yamashita, FEBS Leti. 304,229 (1992). C. H. Warden and M. Friedkin,J. Bid. Chem. 260,6006 (1985). A. Carnero, A. Cuadrado, L. Del Peso, and J. C. Lacal, Oncogene 9,1387 (1994). J. C. Lacal, J. Moscat, and S. A. Aaronson, Nature (London) 330,269 (1987). I. G. Macara, Mol. Cell. Biol. 9,325 (1989). S. Ratnam and C. Kent, Arch. Biochem. Biophys. 323,313 (1995). C. Kent, Biochim. Biophys. Acta 1348, 79 (1997).

MAMMALIAN

CELL MEMBRANE

BIOSYNTHESIS

389

20. G. B. KaImar, R. J. Kay, A. Lachance, R. Aebersold, and R. B. Cornell, Proc. Natl. Acad. (1990). Sci. U.S.A. 87,6029 21. M. S. Rutherford, C. 0. Rock, N. A. Jenkins, D. J. Gilbert, T. G. Tessner, N. G. Copeland, and S. Jackowski, Genomics 18,698 (1993). 22. H. Arita, K. Hanasaki, T. Nakano, S. Oka, H. Teraoka, and K. Matsumoto, J. Biol. Chem. 266, 19139 (1991). 23. G. B. Kalmar, R. J. Kay, A. C. LaChance, and R. B. Cornell, Biochim. Biophys. Acta 1219, 328 (1994). 24. W. Tang, G. A. Keesler, and I. Tabas,]. Biol. Chem. 272,13146 (1997). 25. M. Bakovic, K. Waite, W. Tang, I. Tabas, and D. E. Vance, Biochim. Biophys. Acta 1438, 147 (1999). 26. A. Lykidis, I. Baburina, and S. Jackowski,J. Biol. Chem. 274,26992 (1999). 27. Y. Wang, J. I. S. MacDonald, and C. Kent,J. Biol. Chem. 270,354 (1995). 28. Y. Tsukagoski, J. Nikawa, and S. Yamashita, Eur. J. Biochem. 169,477 (1987). 29. Y. S. Park, P. Gee, S. Sanker, E. J. Schurter, E. R. P. Zuiderweg, and C. Kent, J. Biol. Chem. 272, 15161 (1997). 30. P. Bork, L. Holm, E. V. Koonin, and C. Sander, Proteins 22,259 (1995). 31. D. P. Veitch and R. B. Cornell, Biochemistry 35, 10743 (1996). 32. D. P. Veitch, D. Gilham, and R. B. Cornell, Eur. J. Biochem. 255,227 (1998). 33. C. H. Weber, Y. S. Park, S. Sanker, C. Kent, and M. L. Ludwig, Structure. Fold. Des. 7,1113 (1999). 34. J. E. Johnson, N. M. Rao, S.-W. Hui, and R. B. Cornell, Biochemistry 37,9509 (1998). 35. J. Yang, J. Wang, I. Tseu, M. Kuliszewski, W. Lee, and M. Post, Biochem. J. 325,29 (1997). 36. J. E. Johnson, R. Aebersold, and R. B. Cornell, Biochim. Biophys. Actu 1324,273 (1997). 37. W. Yang, K. P Boggs, and S. Jackowski,J. Biol. Chem. 270,239Sl (1995). 38. K. P. Boggs, C. 0. Rock, and S. Jackowski,J. Biol. Chem. 270, 7757 (1995). 39. R. B. Cornell, G. B. Kalmar, R. J. Kay, M. A. Johnson, J. S. Shanghera, and S. L. Pelech, Biochem. J. 3 10,699 (1995). 40. J. D. Watkins and C. Kent,]. Biol. Chem. 265,219O (1990). 41. J. I. S. MacDonald and C. Kent,]. Biol. Chem. 269,10529 (1994). 42. J. D. Watkins and C. Kent,J. Biol. Chem. 266,21113 (1991). 43. Y. Wang and C. Kent,]. Biol. Chem. 270,17843 (1995). 44. M. Houweling, H. Jamil, G. M. Hatch, and D. E. Vance, J. Biol. Chem. 269,7544 (1994). 45. M. E. Dahmus,J. BioZ. Chem. 27l, 19009 (1996). 46. A. Lykidis, K. G. Murti, and S. Jackowski,]. Biol. Chem. 273,14022 (1998). 47. T. D. Sweitzer and C. Kent, Arch. Biochem. Biophys. 311,107 (1994). 48. M. Houweling, Z. Cui, and D. E. Vance,J. Biol. Chem. 270,16277 (1995). 49. Y. Wang and C. Kent,J. Biol. Chem. 270,18948 (1995). 50. J. A. Friesen, H. A. Campbell, and C. Kent,J. Biol. Chem. 274,13384 (1999). 51. E. Posse de Chaves, D. E. Vance, R. B. Campenot, and J. E. Vance, J. Cell. Biol. 5, 913 (1995). 52. I. C. Northwood, A. H. Tong, B. Crawford, A. E. Drobnies, and R. B. Cornell,,,, Biol. Chem. 274,26240 (1999). 53. J. A. Higgins and J. K. Fieldsend,J. Lipid Res. 28,268 (1987). 54. J. E. Vance and D. E. Vance, J. Biol. Chem. 263,5898 (1988). 55. Y. Wang, T. D. Sweitzer, P. A. Weinhold, and C. Kent, J. Biol. Chem. 268,5899 (1993). 56. M. Houweling, Z. Cui, C. D. Anfuso, M. Bussiere, M. H. Chen, and D. E. Vance, Eur. J. Cell. Biol. 69, 55 (1996). 57. P. A. Weinhold, M. E. Rounsifer, and D. A. Feldman, J. Biol. Chem. 261,5104 (1986). 58. R. B. Cornell, Biochemistry, 30, 5873 (1991).

390 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 7.2. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99.

ATJXANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

M. M. Luche, C. 0. Rock, and S. Jackowski, Arch. Biochem. Biophys. 301114 (1993). R. S. Arnold and R. B. Cornell, Biochemistry 35,9917 (1996). L. Craig, J. E. Johnson, and R. B. Cornell,J. Biol. Chem. 269,3311 (1994). W. Yang and S. Jackowski,J. BioZ. Chem. 270,16503 (1995). S. Jackowski, J. Biol. C&n. 269,3858 (1994). E Terce, H. Bum, and D. E. Vance, J. Lipid Rex 35,213O (1994). T. G. Tessner, C. 0. Rock, G. B. KaImar, R. B. Cornell, and S. Jackowski,J. BioZ. Chem. 266, 16261 (1991). M. Houwehng, L. B. M. Tijburg, W. J. Vaartjes, J. J. Batenburg, G. B. Kahnar, R. B. Cornell, and L. M. G. van Golde, Eur. J. Biochem. 2 14,927 (1993). E. Robbins and M. Scharff, “Cell Synchrony” (I. L. Cameron and G. M. Padilla, eds.). Academic Press, New York, 1966. J. J. M. Bergeron, A. M. H. Warmsley, and C. A. Pastemak, Biochem. Z. 119,489 (1970). G. Gross, M. Danzel, W. Fischer, and K. Brand, Biochim. Biophys. Actu 962,220 (1988). A. J. R. Habenicht, J. A. Glomset, M. Goerig, R. Gronwald, J. Gruhch, U. Loth, and G. Schettler,J. Biol. Chem. 260,137O (1985). C. 0. Rock, J. L. Cleveland, and S. Jackowski, Mol. Cell. BioZ. l2,2351 (1992). M. Wieprecht, T. Wieder, C. C. Geilen, and C. E. Orfanos, FEBS. Lett. 353,221 (1994). M. Wieprecht, T. Wieder, C. Paul, C. C. Geilen, and C. E. Orfanos, J. BioZ. Chem. 271, 9955 (1996). S. L. Pelech and D. E. Vance,J. BioZ. Chem. 257,14198 (1982). (1992). J. D. Watkins, Y. Wang, and C. Kent, Arch. Biochem. Biophys. 292,360 A. L. Henneberry and C. R. McMaster, Biochem. I. 339,291 (1999). J. G. Williams and C. R. McMaster,J. BioZ. Chem. 273,13482 (1998). C. R. McMaster and R. M. Bell, Biochim. Biophys. A& 1348,100 (1997). J. H. Exton, Biochim. Biophys. Actu 12 l2,26 (1994). D. M. Raben, M. S. Pessin, L. A. Rangan, andT. M. Wright,J. CeZZ.Biochem. 44,117 (1990). M. S. Pessin, J. J. BaIdassare, and D. M. Raben, I. BioZ. Chem. 265, 7959 (1990). T. M. Wright, H. S. Shin, and D. M. Raben, Biochem. J. 267,501 (1990). C. J. DeLong, Y. J. Shen, M. J. Thomas, and Z. Cui,J. BioZ. Chem. 274,29683 (1999). M. L. Anthony, M. Zhao, and K. M. Brindle, J. BioZ. Chem. 274,19686 (1999). B. A. Bladergroen, M. Bussiere, W. Klein, M. J. Geelen, L. M. van Golde, and M. Houweling, Eur. J. Biochem. 264,152 (1999). M. A. Miller and C. Kent,J. BioZ. Chem. 261,9753 (1986). 0. Kuge, K. Saito, M. Kojima, Y. Akamatsu, and M. Nishijima, Biochem. J. 319 (Pt. l), 33 (1996). J. Zborowski, A. Dygas, and L. Wojtczak, FEBS Lett. 157, 179 (1983). D. R. Voelker,J. BioZ. Chem. 264, 8019 (1989). (1990). D. R. VoeIker,J. BioZ. Chem. 265,14340 J. E. Vance,J. BioZ. Chem. 265, 7248 (1990). D. R. VoeIker,_Z. BioZ. Chem. 268,7069 (1993). D. R. VoeIker and E. B. Golden, Methods Enzymol. 209,360 (1992). D. R. VoeIker, Proc. NutZ. Acad. Sci. U.S.A. 8% 2669 (1984). M. Houweling, L. B. Tijburg, W. J. Vaartjes, and L. M. van Golde, Biochem. J. 283 (Pt. l), 55 (1992). R. Sundler and B. .&esson,J. BioZ. Chem. 250,3359 (1975). E. Draus, J. Niefind, K. Vietor, and B. Havsteen, Biochim. Biophys. Actu 1045,195 (1990). T. Uchida, Biochim. Biophys. Acta 1349, 13 (1997). P. Pavhdis, M. Ramaswami, and M. A. Tanouye, CeZZ79,23 (1994).

MAMMALIAN

CELL MEMBRANE

BIOSYNTHESIS

391

100. K. Kim, K. H. Kim, M. K. Storey, D. R. Voelker, and G. M. Car-man, 1. Biol. Chem. 274, 14857 (1999). 101. S. A. Bladergroen and L. M. van Golde, Biochim. Biophys. Acta 1348,91 (1997). 102. A. Nakashima, K. Ho&a, and J. Nikawa,J. Biol. Chem. 272,9567 (1997). 103. B. A. Bladergroen, M. Houweling, M. J. Geelen, and L. M. van Golde, Biochem. J. 343,107 (1999). 104. P. S. Vermeulen, L. B. M. Tijburg, M. J. H. Geelen, and L. M. G. van Golde, J. Biol. Chem. 268, 7458 (1993). (1989). 105. R. B. Cornell, J. Biol. Chem. 264,9077 106. R. H. Hjelmstad and R. M. Bell,J. Biol. Chem. 263,19748 (1988). 107. C. R. McMaster and R. M. Bell,J. Biol. Chem. 269(45), 28010 (1994). 108. J. P. Merlie, Biochimie 56, 717 (1974). 109. C. R. McMaster and R. M. Bell, B&him. Biophys. Actu 1348, 117 (1997). 110. Y. Nishizuka, Science 258,607 (1992). 111. Z. Guo, K. Liliom, D. J. Fischer, I. C. Bathurst, L. D. Tomei, M. C. Kiefer, and G. Tigyi, Proc. Nutl. Acad. Sci. U.S.A. 93, 14367 (1996). 112. W. H. Moolenaar,J. Biol. Chem. 270, 12949 (1995). 113. L. B. Tijburg, M. J, Geelen, and L. M. van Golde, Biochim. Biophys. Actu 1004, 1 (1989). 114. J. H. Exton, Biochim. Biophys. Actu 1439, 121 (1999). 115. M. K. Topham and S. M. Prescott,]. Biol. Chem. 274,11447 (1999). 116. M. Schlame and D. Haldar,J. Biol. Chem. 268,74 (1993). 117. A. Y. Mok, G. E. McDougall, and W. C. McMurray, Biochem. Cell Biol. 71,183 (1993). 118. A. Y. Mok, G. E. McDougall, and W. C. McMurray, FEBS Lett. 312,236 (1992). 119. R. G. Liteplo and M. Sribney, Biochim. Biophys. Actu 619,660 (1980). (1997). 120. A. Lykidis, P D. Jackson, C. 0. Rock, and S. Jackowski,J. Biol. Chem. 272,33402 121. A. M. Heacock, M. D. Uhler, and B. W. Agranoff,]. Neurochem. 67,220O (1996). 122. S. Saito, K. Goto, A. Tonosaki, and H. Kondo, J. Biol. Chem. 272,9503 (1997). 123. A. M. Heacock and B. W. Agranoff, Biochim. Biophys. Actu 1348,166 (1997). 124. L. Wu, B. Niemeyer, N. Colley, M. Socolich, and C. S. Zuker, Nature (London) 373, 216 (1995). 125. M. Volta, A. Bulfone, C. Gattuso, E. Rossi, M. Mariani, G. G. Consalez, 0. Zuffardi, A. Ballabio, S. Ban& and B. France, Genomics 55,68 (1999). 126. C. Kent, Annu. Reti. Biochem. 64,315 (1995). 127. R. Weeks, W. Dowhan, H. Shen, N. Balantac, B. Meengs, E. Nudelman, and D. W. Leung, DNA Cell Biol. 16,281 (1997). 128. L. M. Ballas and R. M. Bell, Biochim. Biophys. A& 665,586 (1981). 129. A. Imai and M. C. Gershengom, Nature (LondonJ 325,726 (1987). 130. A. Imai and M. C. Gershengom, J. Biol. Chem. 262,6457 (1987). 131. 0. M. Santiago, L. I. Rosenberg, and M. E. Monaco, Biochem. J. 290,179 (1993). 132. C. Vaziri, C. P Downes, and S. C. Macfarlane, Biochem. J. 294,793 (1993). 133. B. E. Antonsson, Biochem. J. 297,517 (1994). (1994). 134. M. E. Monaco, M. Feldman, and D. L. Kleinberg, Biochem. J. 304,301 135. S. Tanaka, J. Nikawa, H. Imai, Y. Yamashita, and K. Hosaka, FEBS Lett. 393,89 (1996). 136. A. S. Fischl, M. J. Homann,M.A. Poole,andG. M. Cannan,J. Biol. Chem. 261,3178 (1986). 137. G. M. Car-man and A. S. Fischl, Methods Enzymol. 209,305 (1992). 138. F. McPhee, G. Lowe, C. Vaziri, and C. P. Downes, Biochem. J. 275,187 (1991). 139. T. Takenawa and K. Egawa, Arch. Biochem. Biophys. 202,601 (1980). 140. F Basse, J. G. Stout, P. J. Sims, and T. Wiedmer,J. Biol. Chem. 271, 17205 (1996). 141. 0. Kuge, M. Nishijima, and Y. Akamatsu,J. Biol. Chem. 266, 24184 (1991).

392

ATHANASIOS

LYKIDIS

AND SUZANNE

JACKOWSKI

142. 0. Kuge, K. Saito, and M. Nishijima,J. Biol. Chem. 272, 19133 (1997). 143. S. J. Stone and J. E. Vance, &o&em. J. 342,57 (1999). 144. M. Nishijima, 0. Kuge, and Y. Akamatsu, J. Biol. Cha. ZSI, 5784 (1986). (1989). 144a.K. Hasegawa, 0. Kuge, M. Nishijima, and Y. Akamatsu,J. BioZ. Chem. 264,19887 144b.O. Kuge, K. Hasegawa, K. Saito, and M. Nishijima, Proc. N&l. Acad. Sci. I% S. A. 95,4199 (1998). 145.0. Kuge, K. Saito, and M. Nishijima,J. Biol. Chem. 274,23844 (1999). 146. K. P. Boggs, C. 0. Rock, and S. Jackowski, Biochim. Biophys. Actu 1389,l (1998). (1995). 147. K. P. Boggs, C. 0. Rock, and S. Jackowski,J. Biol. Chem. 270,11612 148. C.-L.E. Yen, M.-H. Mar, and S. H. Zeisel, FASEBJ. 13,135 (1999). 149. C. D. Albright, S. H. Zeisel, and R. I. SaIganik, Pathobiology 66,71 (1998). 150. M. Q. Holmes-McNary, R. Loy, M. H. Mar, C. D. Albright, and S. H. Zeisel, Bruin Res. Deu. Bruin Re.s. 101,9 (1997). 151. C. D. Albright, R. Liu, T C. Bethea, K. A. da Costa, R. I. SaIganik, and S. H. Zeisel, FASEB J. 10, 510 (1996). 152. Z. Cui, M. Houwehng, M. H. Chen, M. Record, H. Chap, D. E. Vance, and F. Terce,J. BioZ. Chem. 271,14668 (1996). 153. I. Baburina and S. Jackowski,J. Biol. Chem. 274,940O (1999). 154. D. Teegarden, E. J. Taparowsky, and C. Kent,J. Biol. Chem. 265,6042 (1990). 155. C. J. WaIkey, G. B. KaImar, and R. B. Cornell,,,, BioZ. Chem. 269,5742 (1994). 156. E. A. Dennis, Trends Biochem. Sci. 22, 1 (1997). 157. J. Balsinde and E. A. Dennis,J. BioZ. Chem. 272,16069 (1998). 158. X.-X. Xu, C. 0. Rock, Z.-H. Qiu, C. C. Leslie, and S. Jackowski,]. Biol. Chem. 269,31693 (1994). 159. E. J. Ackermann, E. S. Kempner, and E. A. Dennis,]. BioZ. Chem. 269,9227 (1994). 160. J. Balsinde, I. D. Bianco, E. J. Ackermann, K. Conde-Frieboes, and E. A. Dennis, Proc. N&l. Acud. Sci. U.S.A. 93, 8527 (1995). 161. J. Tang, R. W. Kriz, N. Wolfman, M. Shaffer, J. Seehra, and S. S. Jones, J. BioZ. Chem. 272, 8567 (1997). 162. S. E. Barbour, A. Kapur, and C. L. Deal, Biochim. Biophys. Actu 55465,l (1999). 163. K. S. Zablocki, P E Miller, A. Garcia-Perez, and M. B. Burg, Proc. NutZ. Acud. Sci. U.S.A. 88, 7820 (1991). 164. A. Garcia-Perez and M. B. Burg, Physiol. Rev. 7I, 1081 (1991). 165. H. Bauemschmitt and R. Kinne, Biochim. Biophys. Actu 1148,331 (1993). 166. R. Andreesen, M. ModoleII, H. U. Weltzien, H. Eibl, H. H. Common, G. W. Lohr, and P. G. Munder, Cancer Re.s. 38,3894 (1978). 167. R. Andreesen, M. ModoleII, and P G. Munder, Blood 54,519 (1979). 168. W. E. Berdel, W. R. E. Bausert, H. U. Weltzein, M. L. ModoleII, K. H. Sidman, and P. G. Munder, Eur. J Cancer 16,1119 (1980). 169. M. H. Runge, R. Andreesin, A. Pfleidrer, and P G. Munder, J. NatZ. Cancer Inst. 64, 1301 (1980). 170. W. R. Vogler, A. C. Olson, S. Okamoto, M. Shoji, R. L. Raynoe, J. F. Kuo, W. E. Berdel, H. Eibl, J. Hajdu, and H. Nomura, Lipids 26,141s (1991). 171. D. S. VaIlari, Z. L. Smith, and E Snyder, Biochem. Biophys. Res. Commun. 156,1 (1988). 172. D. R. Hoffman, L. H. Hoffman, and F. Snyder, Cancer Res.46,5803 (1986). 173. 0. WestphaI, Lipid.s, 22,787 (1987). 174. L. W. Daniel, In “Cancer Chemotherapy” (J. A. Hickman and T. R. Tritton, eds.). Blackwell Scientific, Oxford, 1993. 175. M. Lohmeyer and R. Bittman, Drugs ofthe Future 19,102l (1994). 176. P. Principe and P Braquet, Crit. Reu. Oncol. HemutoZ. 18, 155 (1995).

MAMMALIAN

CELL

177.W. J. Hot&an,

MEMBRANE

BIOSYNTHESIS

393

M. Lohmeyer,P. Workman, andS. H. Cheon, Med. Res. Rev. 15,157 (1995). 178. C. C. Geilen, T Wieder, A. Haase, W. Reutter, D. M. MO&, and D. J. Morn?, Biochim. Biophys. Actu 1211,14 (1994). 179. R. Haase, T. Wieder, C. C. Geilen, and W. Reutter, FEBS Leti. 288, 129 (1991). 180. J. E. Johnson and R. B. Cornell, Biochem. J. 33,4327 (1994). 181. S. J. Dunne, R. B. Cornell, J. E. J oh nson, N. R. Glover, and A. S. Tracey, Biochemistry 35, 11975 (1996). 182. G. S. Attard, W. S. Smith, R. H. Templer, A. N. Hunt, and S. Jackowski, Biochem. Sot. Trans. 26, S230 (1998). 183. R. B. Cornell, Biochemisby 30,588l (1991). 184. P. C. Choy, S. B. Farren, and D. E. Vance, Can. J. Biochem. 57,605 (1979). 185. R. G. Sleight and H. N. Dao, Lipids 25,100 (1990). 186. H. Jamil, Z. M. Yao, and D. E. Vance,J. Biol. Chem. 265,4332 (1990). 187. R. B. Cornell and D. E. Vance, Biochim. Biophys. A& 919,37 (1987). 188. I. Baburina and S. Jackowski,J. Biol. Chem. 273,2169 (1998). 189. J. D. Esko, M. Nishijima, and C. R. H. Rae& Proc. Nutl. Acud. Sci. U.S.A. 79,169s (1982). 190. J. Song, L. M. Pfeffer, and D. A. Foster, Mol. Cell. Biol. 11,4903 (1991). 191. A. W. Wyke, S. J. Cook, E. E. MacNulty, and M. J. 0. Wakelam, Cell Signal. 4(3), 267 (1992). 192. H. Cai, P. Erhardt, J. Troppmair, M. T. Diaz-Meco, G. Sitanandam, U. R. Rapp, J. Moscat, and G. M. Cooper, Mol. Cell. BioZ. 13, 7645 (1993). 193. H. Cai, P. Erhardt, J. Szeberenyi, M. T. Diaz-Meco, T. Johansen, J. Moscat, and G. M. Coop(1992). er, Mol. Cell. BioZ. l2,5329 194. l? Mollinedo, J. L. Femandez-Luna, C. Gajate, B. Martin-Martin, A. Benito, R. MartinezDalmau, and M. Modolell, Cancer Res. 57,132O (1997). 195. E. A. Fleer, D. Berkovic, H. Eibl, and C. Unger, Lipids 28,731 (1993). 196. W. R. Vogler, M. Shoji, D. J. Hayzer, Y. P. Xie, and M. Renshaw, Led. Res. 20,947 (1998). 197. S. N. Williams, M. L. Anthony, and K. M. Brindle, Mugn. Reson. Med. 40,411 (1998). 198. K. Miquel, A. Pradines, F. Terce, S. Selmi, and G. Favre,]. BioZ. Chem. 273,26179 (1998). 199. A. Voziyan, C. M. Goldner, and G. Melnykovych, Biochem. J. 295,757 (1993). 200. 0. H. Shin, M. H. Mar, C. D. Albright, M. T. Citarella, K. A. da Costa, and S. H. Zeisel, J. CeZZBiochem. 64, 196 (1997). (1998). 201. T. Wieder, C. E. Orfanos, and C. C. Geilen,]. BioZ. Chem. 273,11025

RGS Proteins: Lessons from the RGS9

Subfamily CHRISTOPHER W. COWAN,~ WEI HE, AND

.,

THEODORE G. WENSEL

Department of Biochemistru and Molecular Biology Baylor College of Medicine Houston, Texas 77030

I.

I. Discovery of Photoreceptor GAP and RGS Protein Family . .. A. GTPase Acceleration in Phototransduction .... . . B. Recognition of RGS Proteins as Ubiquitous G Protein GAPS II. Some Outstanding Problems in RGS Protein Function and Structure A. Problem of RGS Specificity . .. B. Problem of Role of Additional Domains . . . C. Alternative RNA Processing and Functions of Splice Variants ... .. D. Problem of Membrane Targeting E. Problem of Regulation and Modulation of GAP Activity-How Is Correct Timing Achieved and Varied According to Cellular Conditions? III. Contributions to Solving These Problems from RGS9 and Its Subfamily ofRGSProteins ................................................ A. Intrinsic Selectivity of RGS Domains toward Various Ga May Account for Some Specificity but Is Unlikely to Account for Most of It B. RGS Specificity Can Be Conferred through Cell Type-Specific Expression by Transcriptional Control ...... .. . C. RGS Specificity Can Be Conferred through Cell Type-Specific Ex.. .. .. pression by Alternative RNA Processing D. RGS Specificity Can Be Conferred through Cell Type- and OrganelleSpecific Stability by Requisite Association with Another Subunit . E. RGS Specificity Can Be Conferred through Membrane Targeting by Domains outside the Catalytic Domain . . . . . F. RGS Specificity Can Be Conferred through Specific Enhancement or Inhibition by Effecters . . . G. Knockouts....................................,..,.,...,.,,. H. Regulation Likely Involves Multiple Other Proteins . IV FutureProspects References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.. . ............................,

l Present address: Children’s Hospital, Division of Neuroscience, School, Department of Neurobiology, Boston, MA 02115. Progress in Nucleic Acid Reseaxh and Molecular Biology, Vol. 65

341

Copyright All rights

342 342 344 344 344 344 348 348 348 349 349 350 351 351 352 353 354 354 355 356

and Harvard Medical

0 2001 by Academic Press. form reserved. 0079.6603,Ol 135.00

ofreproduction in any

342

CHRISTOPHER

W. COWAN

ET AL.

RGS proteins enhance the time resolution of G protein signaling cascades by accelerating GTP hydrolysis of Go subunits of heterotrimeric G proteins. RGSS1, a photoreceptor-specific RGS protein, is the first vertebrate member of this sizeahle family whose physiological function in a well-defined G protein pathway has been identified. It is essential for normal subsecond recovery kinetics of the light responses in retinal photoreceptors. Understanding this role allows RGSS-1 to serve as a useful model for understanding how specificity and regulation of RGS function are achieved. In addition to the catalytic RGS domain, shared among all members of this family, RGSS-1 contains several other domains, which are also found in a closely related subset of RGS proteins, the RGS9 subfamily. One of these domains, the Gy-like (GGL) d omain, has been identified as the attachment site for G as proteins, which act as obligate subunits for this subfamily. Results from RGSS-1 and other subfamily members suggest that specificity is achieved by cell type-specific transcription, RNA processing, and G -dependent protein stabilization. In addition, membrane localization via specs*P rc targeting domains likely plays an important role. 8 2000Academic PRSS.

I. Discovery of Photoreceptor GAP and RGS Protein Family The RGS gene family encodes GTPase-accelerating proteins (GAPS) for heterotrimeric G proteins. These genes are ubiquitous in eukaryotes, and in metazoans display considerable variety, with close to 50 different kinds now known. A subset of these, which we call the RGS9 subfamily, has been identified in vertebrates, nematodes, and insects, and is found primarily, although not exclusively, in neural tissue or at neuromuscular junctions. It is named the RGS9 subfamily after the photoreceptor-specific RGSS-1, which was the first RGS protein to make its presence known in biochemical assays of GTP hydrolysis, and the first mammalian member of the RGS family to have its physiological function clearly identified. Its identification and characterization represent the coming together of two major lines of research: the effort in several laboratories to understand the molecular mechanisms for rapid recovery from visual excitation, and the effort of a different set of laboratories to understand the functions of the RGS family, the existence of which began to be recognized in 1995 and 1996. S ever-al very useful reviews of RGS proteins have been published (I- 3), and these provide excellent overviews. Our focus here is on RGS9 and its closest relatives, in the hope that our understanding of the function of RGSS-1 in phototransduction can provide some useful insights into how other RGS proteins might function and be regulated.

A. GTPase Acceleration

in Phototransduction

When vertebrate photoreceptor cells respond to light, they must do so with a high degree of temporal resolution, to allow detection of motion and

RGS PROTEINS

343

rapid changes in scene. It has been known since the 1930s from psychophysical studies (4,5), that changes in intensity occurring in less than 100 msec can be detected and processed by the human retina and visual system. By the early 1990s a combination of exquisite electrophysiological studies and careful biochemical characterization had made it possible to account for all the essential features of the rapid activation of rods and cones using the measured or estimated physicochemical properties of a small set of wellcharacterized protein molecules (6). These included, in addition to rhodopsin or the closely related cone pigments, heterotrimeric G proteins called transducins (G,), cGMP phosphodiesterases (PDE) and cGMP-gated cation channels, In either rods or cones, a total of nine polypeptides appears to be sufficient to account for the activation kinetics observed by electrophysiology. In recent years, the inactivation phase of the light response has received increasing attention. It contrasts with the activation phase in two important ways: it is much slower than the activation phase, and therefore rate limiting for the temporal resolution of photodetection, and it is more complicated than activation, involving a larger set of polypeptides and second messengers. Because phototransduction is G protein mediated, it has been widely taken for granted that GTP hydrolysis by G, would be important for inactivation. Experiments with hydrolysis-resistant GTP analogs seemed to confirm that this is the case, at least for responses to bright flashes of light, and possibly for dim flashes as well (7-9). However, biochemical experiments with G, made it clear that, by itself, the (Y subunit of G, hydrolyzes GTP much too slowly to account for recovery kinetics observed in intact photoreceptors. Amphibian rods recover with a time constant of about 2 s, whereas mammalian rods recover about 10 times as fast (6, IO). Cones recover even faster than rods. Purified Gal proteins hydrolyze bound GTP on a time scale of tens of seconds, in general (11, 12) and Giu is no exception. This discrepancy led several groups to search for GTPase-accelerating proteins with activities toward Gal similar to those of GAPS that had been identified for Ras and other small monomeric G proteins (13), and at least one group looked for GTPase-independent mechanisms for cascade inactivation (14). Using crude membrane preparations, evidence was found first for a dependence on membrane concentration of single-turnover GTP hydrolysis by G, (L-17), and second, for an enhancement of this hydrolysis by the effector subunit PDEy (18). Subsequent work (12) revealed that a tightly membrane-associated protein (now known to be the RGS9-l/Ga,, complex), distinct from PDE and its subunits, accounted for the GTPase-accelerating activity. Further, it was found that PDEy on its own had no effect on GTP hydrolysis by G,, despite its formation of a tight complex with Gta-GTP, but that it dramatically enhances the activity of the membrane-bound GAP (19, 20).

344

CHRISTOPHER

B. Recognition of RGS Proteins as Ubiquitous

W. COWAN

ET AL.

G

Protein GAPS Shortly after these discoveries, the RGS family came to the forefront when it was noted (21-23) that a domain present in a protein, Sst2p (24), involved in regulating pheromone responses in yeast, was present in a number of proteins in mammals, in Caenorhabditis elegans (21) and in Aspergillus niduZans (25). The observation that these proteins acted as negative regulators of G protein signaling in yeast and nematodes led to the suggestion (21) that they may function as GTPase accelerators, and to the name, “regulator of G protein signaling,” or RGS, for the shared domain. The demonstration (26) that indeed RGS proteins can accelerate Gu-GTP hydrolysis has been followed by a flurry of activity in this field. Soon after, it was recognized that a member of this GAP family, RGSS-1, was expressed with high specificity in photoreceptor cells, and could account for the G, GAP activity in rods, and probably cones as well (27, 28). M ore recently, the roles of two non-RGS domains and of G,, proteins that appear to act as obligate subunits have been identified for the RGS9 subfamily, as described below.

II. Some Outstanding Problems in RGS Protein Function and Structure A. Problem of RGS Specificity One of the puzzling features of RGS protein function has been the observation that many different RGS domains can accelerate GTP hydrolysis by many different Gal, including members of the G, Go, G,, Gq,il, and GZ classes of GOI.Outside of an apparently complete lack of activity by all tested RGS protein toward Gs,, and by most toward G1sa and G1aa, there seems to be little specificity built into the Ga-RGS domain interaction. Of course different RGS-Gal pairs do show different affinities and catalytic efficiencies when these are carefully quantified (see discussion below), but these are relatively small differences compared, for example, to the discrimination that receptors in the same signaling pathways (29 - 31) display toward various potential ligands. Because it is hard to account for the extensive diversity within the RGS family unless each member is assumed to have specific physiological roles in specific cell types, additional mechanisms for conferring specificity seem likely. In the case of the RGS9 subfamily, it is clear that this is the case, as discussed below.

B. Problem of Role of Additional

Domains

An additional puzzle for understanding RGS proteins is the role of the domains of various sizes, outside the RGS domain, found in these proteins. Two

RGS PROTEINS

345

examples are the fungal RGS proteins, Sst2p from budding yeast (24) and the FlbA gene product from A. nidulans (25). Less than 20% of their structures consist of the RGS domain. Of the remaining nearly 600 amino acids of Sst2p, approximately 300 form a domain identified as homologous to RasGAP (32), and both the Ras-GAP and RGS domains are necessary to complement the pheromone hypersensitivity phenotype of sst2. A number of metazoan RGS proteins also have multiple domains (Fig. 1). The RGSl2 splice variants contain up to 1447 amino acids, with an active RGS domain, an active PDZ (PSD-95, Disc-large, and ZO-1) domain, a Cterminal PDZ binding domain, and a phosphotyrosine binding (PTB) domain (I), in addition to over 1000 amino acid residues of unknown function (33). RGS3 (23) has a large domain N terminal to the RGS domain, the function of which is unknown but with a sequence that suggests its involvement in coiled-coil protein-protein interactions. The subfamily containing GAIP, RGSZl (34), and RetRGS-1 (35) h ave cysteine-rich sequence elements proposed to be sites of palmitoylation (36), and GAIP has a C-terminal PDZ-domain binding peptide as well (37). Axin (38), an inhibitor of Wnt signaling important in axis development, includes an RGS domain within its nearly lOOO-amino acid sequence, in addition to a domain homologous to the Dishevelled protein (DIX), and a domain that recognizes glycogen synthase kinase 3 (GSK-3) (39). The ~115 Rho GEF protein has a domain that catalyzes guanine nucleotide exchange by the small GTP binding protein Rho in a Glaa-dependent way (40), but in addition has an RGS domain that stimulates GTP hydrolysis by G1sa and Gizol (41). The protein kinase anchoring A protein D-AKAPB (42) has been proposed to contain an RGS domain in addition

FIG. 1. Variety of domain structures in RGS domain-containing proteins. RGS, -125-amino acid conserved catalytic domain defining RGS proteins; DEP, DEP domain; R14, a domain of homology between RGSl2 and RGS14; R4, a domain of homology between RGS4 and RGSS; R9, a domain of unknown function common to the RGS9 subfamily; PDZ, PDZ domain; DM, domain of Axin with homology to a domain in Dishevelled outside the DEP domain; GGL, G protein y-like domain; PTB, phosphotyrosine-binding domain. The domain structure of RGS6 is very similar to that of RGS7, and the domains of RGSll and EAT-16 (103) are very similar to those of RGSS. RGS9-2 contains an alternative C-terminal domain of 205 amino acids.

CHRISTOPHER W. COWAN ET AL.

346 Alternative

RNA processing

of human RGS9

Exons

Bnd of K(iS domain

? ?100b

a

RGS9 retina isoform, RGS9-1 (484 a.a.)

RCSJ brain isoform, RGS9-2 (671 a.a. )

4

Key start codon

*

stop codon

.“ ;..’

incomplete exon

? ?alternativeexon Start of DEP domain n

Alternative

RNA processing

of RGSll

0

100 b

800 b

Human RGSll isoform 2 ----

FIG. 2. Alternative RNA processing of RGS9 subfamily members. Gene structures (RGSS, RGSll, and RGSG) are for human genes (47) and likely alternative processing schemes are based on human (RGSS, RGSll, and RGSG) or mouse (RGSG and RGS7) cDNAs. The GenRank accession numbers retrieved are as follows: human RGSll genomic sequence (269667); human RGS6 genomic sequence (ACO05993, AC005477, AC005157, and AC005533); human RGSll cDNAs (ARO16929, AF035153, and AF035154); human RGS6 cDNAs (AF156932, AF107620, AF107619, AF073921, and AFO73920).

347

RGS PROTEINS

Alternative

RNA processing

Eaons

of RGSB End of RCrS domain

End DEP domain

100b ??

a

a

806 h

Human RGSC isofonn 1 (490 aa.)

Hum+

RGS6 isofok

ml

........

2 (472 a.a.)

v

8oq~opom

9B

Human RGSQ isoform 3, (346 aa)

....... .......

Potential alternative

RNA processing

of mouse RGS7 a.a. after the

End of KGS domain

~~100

RCiS domain

b

l-l

40

B*

....... .......

*

.......

1... . . . .

39

. .. . . . . ... ..

22

..I

21

FIG. 2 (continued).

to its recognition domain for the regulatory subunit of CAMP-dependent protein kinase. One of the identifying characteristics of the RGS9 subfamily, apart from

348

CHRISTOPHER

W. COWAN

ET AL.

sequence similarity within the RGS domain, is a common set of domains in the N-terminal portion of these proteins. Two of these display homology to non-RGS proteins: the Gy-like, or GGL, domain (43) and the DEP domain (44). The former is named for its similarity to G protein y subunits, and the latter for the first three proteins in which it was recognized: Dishevelled (a Drosophila protein involved in wingless signaling), EGL-10 (a C. eleguns RGS protein of the RGS9 subfamily), and pleckstrin. The conservation of these domains suggests they have important regulatory roles in RGS function, but for the most part these remain poorly understood. An important exception is the role of the GGL domain in the RGS9 subfamily, as discussed below.

C. Alternative RNA Processing and Functions of Splice Variants Diversity in RGS proteins is evident not only in the numerous genes encoding them found in higher organisms, but also in the different mRNA products of these genes resulting from alternative processing. Alternative mRNAs or cDNAs have been identified for RGS9 (45-47), RGSll (43), RGS7 (W. He, unpublished observations), RGS6 (48), RGSl2 (33), RGS3 (21,23), RGS2 (49), GAIP (22) and pll5GEF (50). A number of these have been determined to give rise to alternative protein products, suggesting important roles for this diversity of RNA processing events, but we are just beginning to understand what some of these roles might be. Figure 2 shows our current understanding of alternative RNA processing in the RGS9 subfamily.

D. Problem of Membrane Targeting G proteins are, in general, firmly attached peripheral membrane proteins. When GAP activity was detected in photoreceptor membranes, it was found to be so tightly bound as to behave in many ways like an integral membrane protein (12). Analyses of the sequences of RGS proteins have failed to support the idea of their having transmembrane domains, with the possible exception of RET-RGSl (35). S everal RGS proteins have been localized to specific subcellular membranes, such as GAIP on clathrin-coated vesicles (51). Therefore, important questions remain regarding the mechanisms of membrane binding and targeting of RGS proteins to those membranes where they are needed.

E. Problem of Regulation and Modulation Activity-How

of GAP

Is Correct Timing Achieved and

Varied According to Cellular Conditions? Because every other aspect of G protein signaling seems to be subject to multiple levels of regulation, it would be surprising if the same were not true

349

RGS PROTEINS

of the GAP activity of RGS proteins. Excessive constitutive GAP activity toward any GOIwould be likely to preclude signaling through that G protein, whereas insufficient GAP activity at the right place and the right time could allow the signal to continue for tens of seconds, which is already known to be incompatible with certain rapid responses. For example, when G proteinregulated K+ channels (GIRK channels) are coexpressed with muscarinic acetylcholine receptors, but without RGS proteins, the responses are greatly slowed. However, coexpression of RGS proteins accelerates the kinetics of the response to more nearly physiological rates (52-56). Obviously a fine tuning of protein levels, or of protein activity, is needed for RGS proteins, and one of the more important questions in this field is how this regulation is achieved.

III. Contributions to Solving These Problems from RGS9 and Its Subfamily of RGS Proteins A. Intrinsic

Selectivity

GOLMay Account Unlikely

to Account

of RGS Domains for Some Specificity

toward

Various

but Is

for Most of It

Of course one possible mechanism for specificity in RGS signaling is through different affinities of each RGS protein for different GOI.RGS proteins, in general, accelerate the GTPase kinetics of most G proteins; however, several studies (5 7- 59) h ave compared the potency of different RGS proteins for a particular Gal and found significant differences in relative potency. Similarly, many studies have compared the GTPase acceleration levels of many different Gal subunits to a single RGS protein in hopes of identifying its cellular partner (34, 60). It is tempting to assume that the most effective RGS protein (determined either by highest maximal velocity, lowest effective concentration for GAP activity, or binding affinity to the AlF, transition state analog) for a given G protein indicates the relevant physiological partner in the cell. However, results from RGSS-1 indicate that this mode of specificity does not necessarily point to the relevant RGS:Gol pair operative in cells. An instructive example comes from comparing results from the GAIP/ RGSZl/Ret-RGSl subfamily with those from the RGS9 subfamily. RGSZl and GAIP potently accelerate GTPase rates of Gzol (>40O-fold) as compared to Gia or Goal (34, 57, 61). Nonetheless, RGSZl does still accelerate GTPase for Goa and Gi,, stimulating their hydrolysis rates approximately 2.5-fold at an RGS concentration of only 21 nM. If the activities of different RGS proteins are compared by stimulation of Gzol GAP activity, RGSZl, RET-RGSl, and GAIP have much greater Gzol GAP activity than does RGS4. However, RGS4 is nonetheless an effective GAP for Gz,, stimulating its hydrolysis

350

CHRISTOPHER

W. COWAN

ET AL,.

about &fold at an RGS concentration of only 12 nM (62) in one report, and demonstrating readily measurable GZa GAP activity at only 300 pM in another (34). Based on these comparisons, alone, the RGSZl family proteins would appear to be the physiological GAPS for Gzol in cells due to this in vi& ro specificity. However, the case of Gtol and RGSS-1 argues that in vitro selectivity alone does not necessarily indicate biological specificity. The activity of the recombinant RGS domain of RGS9 toward its cognate Ga protein Gta (2 7, 63) is much lower than either the activity of RGS4 toward GZa or the activity of RGSZl f amily members toward some Gia family members. Moreover, of all RGS proteins tested on Gtol,which now include RGS9 (27, 63, 64), RGSG, RGS7, RGSll (W. He, unpublished observations), RGS4, GAIP (58), and RGS16 (65, 66), th e most potent GAP appears to be RGS16, or possibly RGS4. However, it is clear that neither RGS16 nor RGS4 is the physiological partner for G,. In fact, most of the RGS proteins tested with Gta for GAP activity are more potent than RGSS; it is the combination of tissue-specific expression, subcellular localization, and cofactor potentiation that confer specificity to the RGSS-l-G, pair in photoreceptors. Thus, even in the unusual case (e.g., for the RGSZl subfamily) whereby RGS-Gal coupling shows striking specificity in vitro, caution must be exercised in drawing conclusions about in vivo function. Intrinsic differences in activity or affinity must be considered only one of many factors governing specificity of biological function. Moreover, results with RGS9 as well as other RGS proteins make it clear that in native membranes, with all domains properly folded and all necessary protein (and possibly lipid) cofactors in place, activity of an RGS protein toward a particular Gel can be dramatically different from what is observed for the same RGS-Ga pair when both are examined as purified recombinant proteins or fragments.

B. RGS Specificity Type-Specific

Can Be Conferred Expression

through

by Transcriptional

Cell Control

One of the first features of RGS family specificity that was examined was the pattern of RNA expression in different tissues. For example, RGS7, RGSS, RGSZl, RGSS, and RGS4 were found to have expression largely restricted to the central nervous system (21, 23, 27, 53, 61), whereas others are found in multiple tissues, each displaying its own pattern of specificity. Relatively few studies have addressed expression patterns at the level of specific cell types, although detailed immunohistochemical analyses have been carried out on RGS7 (67) and RGS9 (2847). F rom studies of expression carried out so far, it is obvious that expression of different RGS genes is tightly controlled, and varies from cell type to cell type, and varies with time as well. A functional role for temporal regulation of transcription has been most clearly documented for the SST2 gene in budding yeast. Its transcription is increased in

351

RGS PROTEINS

response to mating pheromone, providing a mechanism for desensitization. There is also evidence for temporal regulation or regulation in response to stimuli (68 - 73) of RGS gene expression in mammals. In mammals the clearest example of functional specificity conferred by cell type specificity of transcription is the RGS9 gene. It is transcribed almost exclusively in only two tissues, the brain and the retina, with some low level expression also detectable in the lung. Moreover, within the brain expression is found only in a small subset of neurons, primarily restricted to the striatum (27, 45, 47, 74, 75). Within the retina it is expressed only in rod and cone photoreceptors (2 7, 28, 47). A clear correlation with function comes from the finding that the only phenotype detected so far in mice with their RGS9 genes inactivated is a dramatic slowing of recovery in the light responses of rod photoreceptors (76).

C. RGS Specificity Type-Specific

Can Be Conferred Expression

through

Cell

by Alternative

RNA Processing The RGS9 gene provides another example of cell type-specific regulation, with even greater selectivity. Northern analysis reveals that the RGS9 mRNA species produced in the striatum is considerably shorter (2.5 kb in rodents, cattle, and humans) than the ones produced in the retina (8.5-9.5 kb) (27, 45-47, 75). s e q uence analysis of the cDNA and the genes reveals that these different mRNAs result from alternative polyadenylation, which leads to differential exon usage and to different proteins being encoded by the brain and retinal messages. All of the consequences of these different protein structures are not yet known, but there is evidence for functional differences in downregulation of G protein pathways in transient transfection assays (45), and the C-terminal domain, which includes the divergent residues, has been implicated in membrane binding of RGSS-1 (w. He, unpublished observations).

D. RGS Specificity

Can Be Conferred

Type- and Organelle-Specific Association

with Another

through

Stability

Cell

by Requisite

Subunit

Obviously, confining an RGS protein to a particular cell type or subcellular organelle can provide a measure of specificity, because it can only interact with those Gal with which it is in close proximity. As discussed above, some cell type specificity can be achieved at the mRNA level, but there is some evidence that selection can occur at the level of protein translation, stability, or trafficking as well. One means for achieving this is by interaction with another protein, such as a requisite subunit. Two such subunits have been identified, both products of the same gene and homologous to G pro-

352

CHRISTOPHER

W. COWAN ET AL.

tein R subunits. These are G,, and GPJ_L, two splice variants differentially expressed in different cell types. G,, is found in numerous cell types in the nervous system and elsewhere; GP5_L is found exclusively in photoreceptor cells (77). When a weak similarity was noted between a domain of RGSll (now called the GGL domain) found also in other RGS9 subfamily members, and G protein y subunits, the hypothesis was proposed and tested that these might bind to G, subunits. Heterologous expression studies (43, 48, 78) demonstrated that this interaction could occur with high affinity and specificity for Gas, and endogenous RGS7 and RGSS-1 in the retina were found to be tightly bound to G,, (79) and GPB_L(SO), respectively. Mice lacking the RGS9 gene produce no functional Gps_L despite the presence of normal levels of mRNA encoding it (76). Th e simplest explanation is that RGSS-1 is required for proper folding or stability of GPB_L. The converse appears to be true as well. In Sf9 cells, RGSS-1 cannot be expressed in stable active form without coexpression of G Likewise, attempts to express bovine RGSS1 as an EGFP fusion in Xa5-L* enopus lads using transgenesis and a Xenopus opsin promoter produced no detectable RGSS-1 until a GP5_L transgene construct was introduced simultaneously (X. Zhang, unpublished observations). Thus it would appear that even if RGSS-1 mRNA were produced in an inappropriate cell type, it would be unlikely to generate a large amount of functional stable protein, unless GpB_L (or possibly Gp5) were expressed there as well. A study on kinetics of G protein-regulated potassium channels (56) confirms that G,, enhances the activity of either RGS7 or RGSS.

E. RGS Specificity Membrane the Catalytic

Can Be Conferred

Targeting

by Domains

through outside

Domain

One potential mechanism for achieving specificity is to target RGS domains to sites of action where specific signaling proteins are found, typically on membrane surfaces. There is some evidence that domains specific for certain RGS proteins, or subsets of RGS proteins, help target them to membranes, or at least allow them to bind the membranes tightly. In the case of RGSS-1, limited proteolysis reveals that specific removal of a C-terminal 3kDa peptide abolishes tight binding to rod outer segment membranes (W. He, unpublished observations). An intriguing possibility, not yet tested, is that the alternative RNA processing that gives rise to different C-terminal domains in brain RGS9-2 and retinal RGSS-1 confers different membrane anchoring specificity on the two tissue-specific protein products. The altemative C-terminal domain of RGS9-2 is rich in proline residues, which might play a role in recognizing tethering proteins in the brain. This C-terminal domain of RGSS-1 is unique among RGS proteins, so it

RGS PROTEINS

353

seems likely that different domains are responsible for membrane targeting in other RGS proteins. RGS-GAIP (51) and RGSZl (34), which are also membrane bound in cells, have been proposed to be tethered via palmitoylation of clusters of cysteine residues near their N termini. RGSZl binds membranes in brain very tightly, and is resistant to detergent and high ionic strength. The N-terminal 380 amino acids of RGS3 seem to be important for translocating it from the cytosol to the plasma membrane on agonist stimulation (81). Although endogenous RGS4 in NG108 cells is predominantly soluble (82), and the same is true for RGS4 or RGS16 transfected into mammalian cells (83,84), in yeast expressing RGS4 (85) and RGS16 (84), they are bound to membranes. Their palmitoylation (at either of two cysteine residues) was not required for membrane association in either yeast and mammalian cells. It has been proposed that an amphipathic helix N terminal to the RGS domain mediates their membrane binding (84). There are probably many additional motifs for RGS protein-membrane recognition that remain to be determined. F.

RGS Specificity Can Be Conferred through Specific Enhancement or Inhibition by Effecters

There are apparently at least three modes by which effecters can modulate GAP activity of RGS proteins. In one mode, they can substitute for the RGS protein, directly accelerating GTP hydrolysis by binding to GOLin the absence of RGS. Phospholipase Cl3 (PLC) has been shown to do this (86), and there is some evidence that adenylyl cyclase may exert a similar effect on Gsol (87). Alternatively, effecters may inhibit RGS action. In the case of PLC, there appears to be competition with RGS domains for binding to Gqa (59, 88,89), so that the effector must apparently be displaced in order for the RGS domain to bind. The physiological role of this antagonism is unclear. For the other effector for which these interactions have been analyzed, the PDEy subunit of cGMP PDE, it appears that inhibitory action can be exerted in a ternary complex involving Ga-GTP, PDEr, and the RGS domain (58, 65). PDEy exerts an inhibitory effect in this mode on all RGS domains tested except for RGSS. These include RGS4, GAIP, and RGS16 (58, 65). In contrast, PDEy stimulates the activity of the RGS protein with which it is found in nature: RGSS-1 of photoreceptor cells (19, 27, 90). Structurefunction studies have revealed that part of the specificity of the PDEy effect is encoded in the RGS domain, and part of it is conferred by interactions with regions of the GAP outside of the RGS domain proper; specifically, the GGL domain and its complex with G p5_L play an important role in inhibiting RGSS-1 GAP activity in the absence of PDEy and in facilitating enhancement of GAP activity when PDEy is added. These effects of PDEy on RGS9

354

CHRISTOPHER

W. COWAN

ET AL.

are very robust in vitro, but require free PDEy, i.e., PDEy not in its usual state of tight complexation with the catalytic PDEoB unit. How this activity of PDEy is made available in intact photoreceptors remains a mystery.

G. Knockouts Gene inactivation is powerful approach to determining physiological function, and indeed the first clues to RGS domain function came from genetic studies of mutants in Saccharomyces cerevisiae (24, 32, 91) and C. elegans (21). More recently, genetic approaches have revealed the functions of additional RGS proteins in S. cerevisiae (92), Schizosaccharomyces pombe (93), C. elegans (93), and Drosphila melanogaster (94). Gene inactivation in mammals is somewhat more difficult, and usually requires starting with the gene, and then discovering the phenotype. This method has been very valuable in elucidating the functions of some Gcy proteins (95); however, the results have sometimes been difficult to interpret at first, because of redundancy of function in some cases (96), or unexpected subtlety of phenotype in others (97, 98). On the other extreme, genes whose deletion leads to lethality early in embryogenesis [e.g., as observed for the Fused gene encoding the RGS protein Axin (QQ)]also pose difficulties in analysis. Because there are so many RGS proteins, redundancy or subtlety of a phenotype can be expected to present even greater problems in analysis of RGS gene knockouts. Results from inactivation of the RGS9 gene in mice provide an illustrative example (76). Based on earliest reports of expression of this gene in the brain, one might have looked at brain development or obvious behavioral traits and concluded that there was no phenotype of RGS9 knockouts. Examination of retinal morphology or of sensitivity to light would have led to the same conclusion. Only careful recording of the kinetics of the recovery phase of the photoresponse and specific biochemical assays revealed the striking phenotype of these mice: slowed GTP hydrolysis for a specific G protein, G,, and slowed recovery of photocurrents (76). A complementary study showed that replacement of the gene for the effector subunit that stimulates RGSS-1 GAP activity, PDEy, with a mutated version producing a defective protein also leads to slowed recovery (100). The most obvious phenotype in this case was a profoundly reduced sensitivity of rods to light, and the careful kinetic analysis that revealed the slowed recovery was guided by the prior biochemistry (101). As results are obtained from knockouts of additional RGS genes in mammals, it will be important to have biochemical data on the activity of endogenous RGS proteins in specific cell types to guide analysis of the phenotypes.

H. Regulation Likely Involves Multiple Other Proteins In the short time that RGS proteins have been the objects of serious scrutiny, several proteins with which they interact, either through the RGS domain

355

RGS PROTEINS

or through other domains, have been identified. Based on what is known about other key regulators of signal transduction, and based on the presence of domains whose most likely functions involve protein-protein interactions, it seems reasonable to expect that many more proteins will be found to interact with and regulate (or be regulated by) RGS proteins. Likely candidates are PDZ domain-containing proteins, or proteins with PDZ-binding C termini, protein kinases and phosphatases, receptors, and calcium-binding proteins. The search for such novel interactions is underway in several laboratories and is already bearing fruit. For example, RGS7 has been found to interact with polycystin, and with components of the proteasome pathway (102).

IV. Future Prospects The field of RGS protein study is still in its early stages. Perhaps the biggest “gap” in our understanding is uncertainty about the physiological roles of the individual proteins containing RGS domains. There are several keys to filling in this gap: 1. To focus on biochemical and cell biological studies of endogenous RGS proteins in the tissues and cells in which they are normally expressed. Heterologous expression, and especially overexpression, in tissue culture cells can provide useful information, but can be very misleading with respect to functions in specific pathways. Studies of expression patterns do not solve the problem, but they can help to guide such studies by providing candidate genes and proteins for further studies in particular cells; much more needs to be done in the area of cataloguing these patterns. However, it needs to be kept in mind that as potentially potent catalysts, RGS proteins do not have to be present at high levels to exert large effects on signaling. 2. To study effects of gene inactivation. To date, studies of mutants with interesting phenotypes whose mutated genes turn out to encode RGS proteins have been among the most informative. In the future, it seems likely that some of the biggest advances in understanding RGS domain function in mammals will come from gene inactivation in mice. Another area of focus will be on structure and function of domains outside the RGS domains. How the different modules fit together in three dimensions and how they work together to regulate signaling remain some of the most intriguing mysteries in this field. Finally, discovery of new interactions between RGS proteins and other regulatory proteins will be a very active area of research in the next few years.

CHRISTOPHER

356

W. COWAN ET AL.

REFERENCES

1. D. P. Siderovski, B. Strockbine, and C. I. Behe, Crit. Rev. Biochem. Mol. Biol. 34, 215 (1999). 2. T. Wieland and C. K. Chen, Naunyn Schmiedebergs Arch. Pharmacol. 360,14 (1999). 3. L. De Vries and M. Gist Farquhar, Trends Cell BioZ. 9, 138 (1999). 4. S. Hecht and C. D. Verrijp,J. Gen. Physiol. 17,251 (1933). 5. S. Hecht and S. Shlaer, J. Gem Physiol. 19,965 (1936). 6. E. N. Pugh, Jr. and T D. Lamb, Biochim. Biophys. Acta 1141,111 (1993). 7. W. A. Sather and P B. Detwiler, Proc. NatZ. Acad. Sci. U.S.A. 84,929O (1987). 8. T. D. Lamb and H. R. Matthews,J. Physiol. 407,463 (1988). (1997). 9. M. S. Sagoo and L. Lagnado, Nature (London) 389,392 10. V. Y. Arshavsky and E. N. Pugh, Jr., Neuron 20, 11 (1998). 11. ‘I G. Wensel, Heterotrimeric G proteins: Structure, regulation and signaling mechanisms. In “Introductory Signal Transduction” (A. Sitaramayya, ed.), pp. 29-46. Birkhauser, Boston, 1999. 12. J. K. Angleson and T. G. Wensel, Neuron ll, 939 (1993). 13. M. S. Bog&i and F. McCormick, Nature (London) 366,643 (1993). 14. M. A. Erickson, P Robinson, and J. Lisman, Science 257,1255 (1992). 15. V Y. Arshavsky, M. I? Antoch, and l? P. Philippov, FEBS Lett. 224,19 (1987). 16. E. A. Dratz, J. W. Lewis, L. E. Schaechter, K. R. Parker, and D. S. KIiger, Biochem. Biophys. Res. Commun. 146,379 (1987). 17. R. Wagner, N. Ryba, and R. Uhl, FEBS Lett. 234,44 (1988). 18. V. Arshavsky and M. D. Bownds, Nature (London) 357,416 (1992). 19. J. K. Angleson and T. G. Wensel,J. BioZ. Chem. 269,16290 (1994). 20. V. Y. Arshavsky, C. L. Dumke, Y. Zhu, N. 0. Artemyev, N. P. Skiba, H. E. Hamm, and M. D. Bownds,J. BioZ. Chem. 269,19882 (1994). 21. M. R. Koelle and H. R. Horvitz, Cell 84,115 (1996). 22. L. De Vries, M. MousIi, A. Wurmser, M. G. Farquhar, Proc. Natl. Acad. Sci. U.S.A. 92, 11916 (1995). 23. K. M. Druey, K. J. Blumer, V. H. Kang, and J. H. Kehrl, Nature &on&n) 379,742 (1996). 24. C. Dietzel and J. Kurjan, Mol. Cell. BioZ. 7,4169 (1987). 25. B. N. Lee and T H. Adams, Mol. Microbial. 14,323 (1994). 26. D. M. Berman, T M. W&e, and A. G. Gilman, Cell 86,445 (1996). 27. W. He, C. W. Cowan, and T G. Wensel, Neuron 20,95 (1998). 28. C. W. Cowan, R. N. Fariss, I. Sokal, K. Palczewski, and T. G. Wensel, Proc. NatZ.Acad. Sci. U.S.A. 95, 5351 (1998). (1999). 29. P. G. Strange, Biochem. Pharmucol. 58,108l 30. G. Milhgan, D. A. Groarke, A. McLean, R. Ward, C. W. Fong, A. Cavalli, and T. Drmota, Biochem. Sot. Trans. 27,149 (1999). 31. A. Chollet and G. Turcatti,J. Comput. Aided Mol. Des. 13,209 (1999). 32. H. G. Dohlman, J. Song, D. Ma, W. E. Courchesne, and J. Thomer, Mol. Cell. Biol. 16,5194 (1996). 33. B. E. Snow, R. A. Hall, A. M. Krumins, G. M. Brothers, D. Bouchard, C. A. Brothers, S. Chung, J. Mangion, A. G. Gilman, R. J. Leikowitz, and D. P. Siderovski,J. Biol. Chem. 273, 17749 (1998). 34. J. Wang, A. Ducret, Y. Tu, T. Kozasa, R. Aebersold, and E. M. Ross, J. Biol Chem. 273, 26014 (1998). 35. E. Faurobert and J. B. Hurley, Proc. Natl. Acad. Sci. U.S.A. 94,294s (1997).

RGS PROTEINS

357

36. L. De Vries, E. Elenko, L. Hubler, T. L. Jones, and M. G. Farquhar, Pmt. N&l. Acud. Sci. U.S.A. 93,15203 (1996). 37. L. De Vries, X. Lou, G. Zhao, B. Zheng, and M. G. Farquhar, Proc. N&Z. Acad. Sci. U.S.A. 95, 12340 (1998). 38. L. Zeng, F. Fagotto, T. Zhang, W. Hsu, T. J. Vasicek, W. L. Perry, 3rd, J. J. Lee, S. M. Tilghman, B. M. Gumbiner, and F Costantini, Cell 90,181 (1997). 39. C. M. Hedgepeth, M. A. Deardorff, K. Rankin, and. P. S. Klein, Mol. Cell. Biol. 19, 7147 (1999). 40. M. J. Hart, X. Jiang, T Kozasa, W. Roscoe, W. D. Singer, A. G. Gilman, P. C. Stemweis, and G. Bollag, Science 280, 2112 (1998). 41. T Kozasa, X. Jiang, M. J. Hart, P. M. Stemweis, W. D. Singer, A. G. Gilman, G. Bollag, and P C. Stemweis, Science 280,2109 (1998). 42. L. J. Huang, K. Durick, J. A. Weiner, J. Chun, and S. S. Taylor, Proc. Nutl. Acud. Sci. U.S.A. 94, 11184 (1997). 43. B. E. Snow, A. M. Krumins, G. M. Brothers, S. F. Lee, M. A. Wall, S. Chung, J. Mangion, S. Arya, A. G. Gilman, and D. P. Siderovski, Proc. Nutl. Acud. Sci. U.S.A. 95, 13307 (1998). 44. C. P. Ponting and P Bork, Trend.s Biochem. Sci. 21,245 (1996). 45. Z. Rahman, S. J. Gold, M. N. Potenza, C. W. Cowan, Y. G. Ni, W. He, T. G. Wensel, and E. J. Nestler,J. Neurosci. 19,2016 (1999). 46. J. G. Granneman, Y. Zhai, Z. Zhu, M. J. Bannon, S. A. Burchett, C. J. Schmidt, R. Andrade, and J. Cooper, Mol. Phurmucol. 54,687 (1998). 47. K. Zhang, K. Howes, W. He, M. Pettenati, K. Palczewski, T. G. Wensel, and W. Baehr, Gene 240,23 (1999). 1d erovski, Proc. Nutl. Acud. Sd. 48. B. E. Snow, L. Betts, J. Mangion, J. Sondek, and D. P. S’ U.S.A. 96,6489 (1999). 49. C. Chen, B. Zheng, J. Han, and S. C. Lin,J. Biol. Chem. 272, 8679 (1997). 50. M. J. Hart, S. Sharma, N. elMasry, R. G. Qiu, P. McCabe, P. Polakis, and G. Bollag, J. Biol. Chem. 271,25452 (1996). 51. L. De Vries, E. Elenko, J. M. McCaffery, T. Fischer, L. Hubler, T. McQuistan, N. Watson, and M. G. Farquhar, Mol. Biol. Cell. 9, 1123 (1998). 52. 0. Saitoh, Y. Kubo, M. Odagiri, M. Ichikawa, K. Yamagata, and T. Sekine, J. Biol. Chem. 274,9899 (1999). 53. 0. Saitoh, Y. Kubo, Y. Miyatani, T. Asano, and H. Nakata, Nature (Londonj 390,525 (1997). 54. C. A. Doupnik, N. Davidson, H. A. Lester, and P. Kofuji, Proc. Nutl. Acad. Sci. U.S.A. 94, 10461 (1997). 55. H. H. Chuang, M. Yu, Y. N. Jan, and L. Y. Jan, Proc. Nutl. Acud. Sci. U.S.A. 95,11727 (1998). 56. A. Kovoor, C.-K. Chen, W. He, T. G. Wensel, M. I. Simon, and H. A. Lester,]. Biol. Chem. 275,3397 (2000). 57. S. Popov, K. Yu, T. Kozasa, and T. M. Wilkie, Proc. Nutl. Acud. Sci. U.S.A. 94,7216 (1997). 58. E. R. Nekrasova, D. M. Berman, R. R. Rustandi, H. E. Hamm, A. G. Gilman, and V. Y. Arshavsky, Biochemistry 36,7638 (1997). 59. J. R. Hepler, D. M. Berman, A. G. Gilman, and T. Kozasa, Proc. N&Z. Acud. Sci. U.S.A. 94, 428 (1997). 60. N. Watson, M. E. Linder, K. M. Druey, J. H. Kehrl, and K. J. Blumer, Nature &ondon) 383, 172 (1996). 61. J. L. Glick, T. E. Meigs, A. Miron, and P. J. Casey,J. Biol. Chem. 273,26008 (1998). 62. D. M. Berman, T. Kozasa, and A. G. Gilman,J. BioZ. Chem. 271,27209 (1996). 63. N. l? Skiba, C. S. Yang,T. Huang, H. Bae, andH. E. Hamm,J. Biol. Chem. 274,877O (1999). 64. R. L. McEntaffer, M. Natochin, and N. 0. Artemyev, Biochemistry 38,493l (1999). 65. M. Natochin, A. E. Granovsky, and N. 0. Artemyev,]. Biol. Chem. 272,17444 (1997).

358

CHRISTOPHER

W. COWAN

ET AL.

66. C. K. Chen, T. Wieland, and M. I. Simon, Proc. N&Z. Acad. Sci. U.S.A. 93, 12885 (1996). 67. X. Z. Khawaja, J. J. Liang, J. A. Saugstad, P G. Jones, S. Harnish, I? J. Corm, and M. I. Cockett,J. Neurochem. 72,174 (1999). 68. R. Panetta, Y. Guo, S. Magder, and M. T. Greenwood, Biochem. Biophys. Res. Commun. 259, 550 (1999). 69. S. A. Burchett, M. J. Bannon, and J. G. Granneman, J. Neurochem. 72,1529 (1999). 70. C. Beadling, K. M. Druey, G. Richter, J. H. Kehrl, and K. A. Smith,J. Immunol. 162,2677 (1999). 71. L. Buckbinder, S. Velasco-Miguel, Y. Chen, N. Xu, R. Talbott, L. Gelbert, J. Gao, B. R. Seizinger, J. S. Gutkind, and N. Kley, Proc. Nutl. Acad. Sci. U.S.A. 94, 7868 (1997). 72. T. Kardestuncer, H. Wu, A. L. Lim, and E. J. Neer, FEBS Lett. 438,285 (1998). 73. Y. G. Ni, S. J. Gold, P. A. Iredale, R. Z. T erwilliger, R. S. Duman, and E. J. Nestler, J. Neurosci. 19,3674 (1999). 74. S. J. Gold, Y. G. Ni, H. G. Dohlman, and E. J. Nestler,J. Neurosci. 17,8024 (1997). 75. E. A. Thomas, P E. Danielson, and J. G. Sutcliffe, J. Neurosci. Res. 52, 118 (1998). 76. C. K. Chen, M. E. Bums, W. He, T. G. Wensel, D. A. Baylor, and M. I. Simon, Nature (London) 403,557 (2000). 77. A. J. Watson, A. M. Aragay, V. Z. Slepak, and M. I. Simon,J. Biol. Chem. 271,28154 (1996). 78. K. Levay, J. L. Cabrera, D. K. Satpaev, and V. Z. Slepak, Proc. Nutl. Acud. Sci. U.S.A. 96, 2503 (1999). 79. J. L. Cabrera, F. de Freitas, D. K. Satpaev, and V. Z. Slepak, Biochem. Biophys. Res. Com(1998). nun. 249,898 80. E. R. Makino, J. W. Handy, T. Li, and V. Y. Arshavsky, Proc. Nutl. Acud. Sci. U.S.A. 96,1947 (1999). 81. N. 0. D&n, A. Sorokin, E. Reed, S. Elliott, J. H. Kehrl, and M. J. Dunn, Mol. Cell. Biol. 19, 714 (1999). 82. K. M. Druey, B. M. Sullivan, D. Brown, E. R. Fischer, N. Watson, K. J. Blumer, C. R. Gerfen, A. Scheschonka, and J. H. Kehrl,J. Biol. Chem. 273,18405 (1998). 83. K. M. Druey, 0. Ugur, J. M. Car-on, C. K. Chen, P. S. Backlund, and T. L. Jones, J. Biol. Chem. 274, 18836 (1999). 84. C. Chen, K. T Seow, K. Guo, L. I? Yaw, and S. C. Lin, J. Biol. Chem. 274, 19799 (1999). 85. S. P. Srinivasa, L. S. Bernstein, K. J. Blumer, and M. E. Linder, Proc. Nutl. Acud. Sci. U.S.A. 95,5584 (1998). 86. G. Bernstein, J. L. Blank, D. Y. Jhon, J. H. Exton, S. G. Rhee, and E. M. Ross, Cell 70,411 (1992). 87. K. Scholich, J. B. Mullenix, C. Wittpoth, H. M. Poppleton, S. C. Pierre, M. A. Lindorfer, J. C. Garrison, and T. B. Patel, Science 283, 1328 (1999). 88. Y. Yan, P. P. Chi, and H. R. Boume,J. Biol. Chem. 272, 11924 (1997). 89. S. P Heximer, N. Watson, M. E. Linder, K. J. Blumer, and J. R. Hepler, Proc. Natl. Acud. Sci. U.S.A. 94, 14389 (1997). 90. V. Y. Arshavsky and M. D. Bownds, Nature (London) 357,416 (1992). 91. R. K. Chan and C. A. Otte, Mol. Cell. Biol. 2, 11 (1982). 92. M. Versele, J. H. de Winde, and J. M. Thevelein, EMBO]. l&5577 (1999). 93. P Watson, K. Davis, M. Didmon, P. Broad, and J. Davey. Mol. Microbial. 33,623 (1999). 94. S. Granderath, A. Stollewerk, S. Greig, C. S. Goodman, C. J. O’Kane, and C. Klambt, Development I26,1781 (1999). 95. S. Offermanns, Nuunyn Schmiedebergs Arch. Phuvmucol. 360,5 (1999). 96. S. Offermanns, L. P. Zhao, A. Gohla, I. Sarosi, M. I. Simon, and T. M. Wilkie, EMBOJ. 17, 4304 (1998).

RGS PROTEINS

359

97. D. Valenzuela, X. Han, U. Mende, C. Fankhauser, H. Mashimo, P Huang, J. Pfeffer, E. J. Neer, and M. C. Fishman, Proc. N&l. Acad. Sci. U.S.A. 94, 1727 (1997). 98. U. Mende, B. Zagrovic, A. Cohen, Y. Li, D. Valenzuela, M. C. Fishman, and E. J. Neer, J. Neurosci. Res. 54, 263 (1998). 99. T. J. Vasicek, L. Zeng, X. J. Guan, T. Zhang, F. Costantini, and S. M. Tilghman, Genetic.s, 147, 777 (1997). 100. S. H. Tsang, M. E. Bums, P. D. Calvert, P. Gouras, D. A. Baylor, S. P. Goff, and V Y. Arshavsky, Science 282,117 (1998). 101. V Z. Slepak, N. 0. Artemyev, Y. Zhu, C. L. Dumke, L. Sabacan, J. Sondek, H. E. Hamm, M. D. Bownds, and V. Y. Arshavsky, J. Bid. Chem. 270, 14319 (1995). 102. E. Kim, T. Amould, L. Sellin, T. Benzing, N. Comella, 0. Kocher, L. Tsiokas, V. P Sukhatme, and G. Walz, Proc. Natl. Acud. Sci. U.S.A. 96, 6371 (1999). 103. Y. M. Hajdu-Cronin, W. J. Chen, G. Patikoglou, M. R. Koelle, and P. W. Sternberg, Genes DCG. 13, 1780 (1999).

Index

B

A

Base excision repair (BER), see DNA repair BER, see Base excision repair BRCAl

ABFl autonomously replicating sequence binding, 288 functions, overview, 264 AFM, see Atomic force microscopy Amphibian metamorphosis, thyroid hormone regulation comparison with mammalian models, 9192 interconversion of hormone forms, 5859 intestine development, see Intestinal remodeling, Xenopus studies of thyroid hormone levels in development, 56 overview, 54 - 55 precocious metamorphisis induction, 5657 prospects for study, 93 thyroid inhibition effects, 57-58 Apoptosis intestinal development role of matrix metalloproteinases in amphibians cell fate determination mechanisms, 91 collagenase, 87 differential expression analysis, 87 extracellular matrix remodeling, 85 GelA, 87 stromelysin-3 apoptosis association, 86-87, 89 developmental expression pattern, 86 substrates, 89-90 thyroid hormone regulation, 84-85 phosphatidylcholine biosynthesis inhibition and apoptosis induction, 382386 Atomic force microscopy @FM), intemucleosomal DNA length measurement, 206207

functions, overview, 264 histone deacetylase recruitment, transcriptional activation, 289

3 16-3 17

C B-Casein, cell shape and structure in gene regulation, 330 CCT, see CTP:phosphocholine cytidylyltransferase CDP-choline:l,&-diacylglycerol phosphotransferase (CPT) inhibition and apoptosis induction, 385386 regulation, 3 74 structure, 3 73 subcellular localization, 374 CDP-diacylglycerol syntbase (CDS), features, 377-378 CDS, see CDP-diacylglycerol synthase Cell membrane, see Membrane, cell Ceramide, phosphatidylcholine metabolism relationship in apoptosis induction, 386 CHIP assay, see Chromatin immunoprecipitation assay Cholesterol synthesis, see Squalene synthase Choline kinase (CK) activation in mitogenesis, 364-365 genes and isoforms, 364 Chromatin, see also Histone; Karyoskeleton DNA winding on nucleosome core, 208209 imaging in living cells, 301-302 mechanical signaling pathways and nuclear DNA organization, 329-331 proliferating cell nuclear antigen role in assembly, 285

395

396 Chromatin (cont.) remodeling in thyroid hormone-responsive gene regulation chromatin disruption assays, 66,68 histone acetylation regulation, 65 - 66 Xenopus oocyte system advantages, 63, 65,93 structure overview, 300-301 transcriptionally active chromatin, 303304 transcription factors in replication, 29 1 yeast chromatin properties discrete length linkers, 199, 202-203 DNA unpeeling from core ends, linker length effects, 206-208 GAL gene-specific chromatin structure, see GAL higher-order structure, 2 12-2 14 histone contributions to unique features, 211-212 metazoan core nucleosome similarity, 198-199,210-211,213 nucleosome conformational changes in situ, 203, 205-206 nucleosome core with short linkers, 198-199 terminal core nucleosome DNA liability and consequences, 209 -2 11 Chromatin immunoprecipitation (CHIP) assay, acetylated histones caveats in studies, 320-321 cell cycle studies, 319 fine-mapping, 3 19 - 320 transcription studies, 3 18 -3 19 Circular permutation analysis, GAL DNA bends, 242-243 CK, see Choline kinase CPT, see CDP-choline:l&diacylglycerol phosphotransferase CTP:phosphocholine cytidylyltransferase (CCT) cell cycle regulation of expression and activity, 371-373,388 domains, 365-366 genes and isoforms, 365,367,387 inhibition and apoptosis induction, 382386 knockout mouse, 387 lipid regulation, 369-371,384-385

INDEX

membrane contact and regulation, 384385 nuclear signaling, 38 7 overexpression effects, 380 phosphorylation, 366-367,370,373 subcellular localization of isoforms, 368369 tissue distribution, 367-368 Cyclic nucleotide phosphodiesterase (PDE) catalytic mechanism chemistry, 9 conserved residues, 12-14 metal requirements, 14 - 16 structural determinants, 11 classes, 3, 5 domains alignment in class I superfamily, 3 - 5, 7 catalytic domain, 7- 8 regulatory domains, 8 familes, see PDEl-PDEll functions, 2-3 inhibitors, 16 - 18 photoaffinity labeling, 7- 8 phototransduction, 343,353-354 phylogeny of class I superfamily, 5 -6 regulation extracellular signaling pathways, 19 long-term regulation, 20 overview, 2, 18 short-term feedback, 19-20 substrate features for catalysis, 8 - 9 substrate specificity amino acid residues in binding, 10 cyclic AMP versus cyclic GMP, 9-11 hydropathy analysis of catalytic domains, 10 syn versus anti configuration, 9 - 10

D Deprenyl, monoamine oxidase inhibition, 151-152 DGK, see Diacylglycerol kinase Diacylglycerol kinase (DGK), features, 377 DNA footprinting, see GAL DNA methylation, proliferating cell nuclear antigen role, 284-285 DNA polymerase cx (Pal cx) cell cycle checkpoint role, 269 -2 70

INDEX

cell cycle control of DNA replication, role, 268-269 DNA repair role, 2 70 functions, overview, 264 mechanism, 263,265 protein interaction cell cycle control, 269 coordinated and error-free laggingstrand replication, 267-268 stable of interacting proteins, 266 subunits, 265 viral DNA replication, 265,267 DNA polymerase f3 (Pol f3), base excision repair role, 2 75, 2 77 DNA polymerase S (Pol S) functions base excision repair, 2 77 double-strand break repair, 2 79 mismatch repair, 278 nucleotide excision repair, 278 overview, 264,271 postreplicative bypass of W-damaged DNA, 279 gap-filling synthesis, 2 73 holoenzyme assembly, 272-274 protein interactions, 271-272 subunits, 2 7 1 viral DNA replication, 2 74 ) DNA polymerase E (Pol ?? functions base excision repair, 277 double-strand break repair, 2 79 gap-filling synthesis, 273 nucleotide excision repair, 278 overview, 264 S phase checkpoint control, 275 holoenzyme assembly, 272-274 subunits, 2 72 viral DNA replication, 2 74 DNA repair DNA polymerase 01role, 270 DNA polymerase 6, base excision repair role, 275, 277 DNA polymerase 6 roles base excision repair, 277 double-strand break repair, 2 79 mismatch repair, 2 78 nucleotide excision repair, 2 78 postreplicative bypass of W-damaged DNA, 2 79

397 DNA polymerase E roles base excision repair, 2 7 7 double-strand break repair, 279 nucleotide excision repair, 278 overview of mechanisms, 263,275 proliferating cell nuclear antigen role, 282-284 DNA replication events at replisome, 262-263 polymerases, see DNA polymerase cx; DNA polymerase 6; DNA polymerase E transcription factor roles, 289-291

E ER, see Estrogen receptor Estrogen receptor (ER), karyoskeleton association, 328-329 Ethanolamine cytidylyltransferase, features, 375-376 Ethanolamine Ethanolamine 376

kinase, features, 375 phosphotransferase, features,

F Ferredoxin-NADP+ reductase (FNR), flavinbinding site, 141, 143 FNR. see Ferredoxin-NADP+ reductase

G GAL chromatin structure in induced state coding regions DNase I digestion, 240-241 micrococcal nuclease digestion, 239, 241 nucleosome conformational changes and rearrangement, 239-242 digestion profiles versus patterns, 223 224 disruption mutants, 223 promoter regions Gal4p-dependent disruption, 233234

398 GAL (cont.) GAL80 nucleosome disruption, 237239 Gal80p-dependent nucleosome deposition, 234,236,280-281 metabolic competence and nucleosome disruption/deposition, 236-237 chromatin structure in uninduced stateof promoter region digestion profiles versus patterns, 223224 disruption mutants, 223 hypersensitive region constitutive UAS, accessibility, 23 l232 DNase I cleavage, 224-228 factors keeping region free of nucleosomes, 228-230 regulatory GAL genes, 224 intracore ladder patterns, 228 nucleosome covering of coding regions, 228 nucleosome positioning, footprinting analysis, 230-231 upstream region nucleosomes and basal repression, 232-233 chromosomal architecture of promoter regions in induced state circular permutation analysis, 242-243 higher-order structure, 245-246 intrinsic DNA structure, 242-245 Gal4 functions, overview, 264 transcriptional activation, 289 Gal4p in transcriptional activation coactivator utilization, 247 cooperating factors and processes, 253254 nucleosome disruption, 233-234 nucleosome removal from TATA site, 249-250 protein interactions, 246-247 protein mutagenesis studies, 247-248 RNA polymerase II recruitment, 247 TATA-binding protein interaction, 246 histone role in gene regulation, 248-249, 252-253 history of study, 214

INDEX

nucleosome binding dynamics in gene regulation, 250-254 regulatory genes regulation of expression comparison with structural gene induction, 222-223 GAL3 as galactose signal transducer, 222 GAL4 promoter, 220-221 GAL80 dual promoters, 221-222 types, 2 15 structural genes regulation activation steps, 219-220 comparison with regulatory gene induction, 222-223 galactose induction, 216-217 glucose repression, 218-219 glycerol/lactate and poising for expression, 217-218 types, 2 15 GAPS, see GTPase-accelerating proteins GTPase-accelerating proteins (GAPS), see also Regulator of G protein signaling phototransduction, 343 types, 342

H Histone, see also Chromatin; GAL; Karyoskeleton acetylation acetyltransferases types and substrates, 308-310 Gcn5,308-310 SAGA, 3 10 PCAF, 310-311 transcription factor recruitment, 3 ll313,320 transcriptional activation, 3 13 -3 14 Esal, 3 11 CBP/pSOO, 309 substrate specificity, 3 10 structure, 309-310 cancer role, 3 17 chromatin immunoprecipitation assay in localization caveats in studies, 320-321

INDEX

399

cell cycle studies, 319 fine-mapping, 319-320 transcription studies, 3 18 -3 19 chromatin structure effects, 308 deacetylases gene repression, 3 14,3 17 structure, 3 14 substrate specificity, 3 I5 transcription factor recruitment, 316317,320 types, 314-315 distribution in core histones, 307 susceptibility of histone types, 307-308 thyroid hormone-responsive gene regtlation, 65-66 core histones modifications overview, 306 signal transduction, 332 N-terminal tails and protein interactions, 302-303 structures, 300 variants, 305 linker Hl histones overview, 300 subtypes, 305-306 methylation, 321-322 phosphorylation cell cycle dependence, 322 early response gene transcription correlation, 325 kinases and phosphatases, 325-326 sites, 322 transcription role, 322-325 types, 300 ubiquitination, 306-307

I IL-l, see Interleukin-1 Insulin, S6 kinase role in production, 119 Interleukin-1 (IL-l), squalene synthase regu lation, 190-191 Intestinal remodeling, Xenopus studies of thyroid hormone autonomous response to thyroid hormone cell culture studies, 76-77 epithelial apoptosis, 75-76

organ culture studies, 74-76 connective tissue remodeling, 73-74 epithelial transformation, 73 larval intestine features, 70, 73 matrix metalloproteinases cell fate determination mechanisms, 91 collagenase, 8 7 differential expression analysis, 87 extracellular matrix remodeling, 85 GelA, 87 stromelysin-3 apoptosis association, 86-87, 89 developmental expression pattern, 86 substrates, 89 - 90 thyroid hormone regulation, 84-85 muscle thickening, 74 thyroid hormone receptor messenger RNA levels, 77 TRR genes cell type-dependent temporal regulation, 79 upregulation by thyroid hormone, 77-79,82-83 thyroid hormone response genes differential screening, 80 early response genes, 80,82-83 gene regulation cascade model, 92-93 late response genes, 83-84 table, 8 1

K Karyoskeleton cisplatin cross-linking, 33 1 core filaments, 326 estrogen receptor association, 328-329 histone acetyltransferase and deacetylase association, 329 isolation, 326-327 matrix attachment region, 326 mechanical signaling pathways and nuclear DNA organization, 329-331 transcription role transcriptionally active chromatin, 327328 transcription factor association, 328329 transcription factories, 327

400

INDEX

L Lipopolysaccharide (LPS), squalene synthase regulation, 189 - 19 1 LPS, see Lipopolysaccharide

M Mammalian target of rapamycin (mTOR) essential amino acids in S6 kinase activation, 116-117 mechanism of S6 kinase effector activity, 115-116 phosphorylation, 115 sequence analysis, 115 MAO A, see Monoamine oxidase A MAO B, see Monoamine oxidase B Matrix attachment region, see Karyoskeleton Matrix metalloproteinases (MMPs) functions, 84 intestinal development role in amphibians cell fate determination mechanisms, 91 collagenase, 8 7 differential expression analysis, 87 extracellular matrix remodeling, 85 GelA, 87 stromelysin3 apoptosis association, 86-87, 89 developmental expression pattern, 86 substrates, 89-90 thyroid hormone regulation, 84 - 8 5 proteolytic processing, 84 MCM proteins, DNA replication role, 289290 Mediator, Gal4p as component, 247 MEF2, see Myocyte enhancer factor 2 Membrane, cell composition, 362-363 phosphatidylcholine biosynthesis inhibition and apoptosis induction, 382-386 regulation, 364 -3 74 phosphatidylethanolamine biosynthesis regulation, 3 74 - 3 76 phosphatidylinositol biosynthesis regulation, 376-379 phosphatidylserine biosynthesis regulation, 379

phospholipid biosynthesis homeostasis, 380-382 overview, 362-363 Mismatch repair (MMR), see DNA repair MMPs, see Matrix metalloproteinases MMR, see Mismatch repair Monoamine oxidase A (MAO A) deficiency in aggression, 130, 132 functions catalytic reaction, 130 central nervous system, 151 gene cloning, 13 1 locus, 132 organization, 132, 135 monoamine oxidase B chimera properties, 140,148 prospects for study, 152 sequence homology with type B, 131-132 tissue and cell distribution, 149 - 150 Monoamine oxidase B (MAO B) active site and catalytic residue mutagenesis, 148-149 deprenyl inhibition, 151-152 flavin adenine dinucleotide binding comparison with other flavin-binding enzymes, 141, 143, 146 dot-blot assay of noncovalent binding, 140 flavinylation assay, 140, 142 steps in flavinylation, 146-147 stoichiometry, 135 functions catalytic reaction, 130 central nervous system, 15 1 gene cloning, 13 1 locus, 132 organization, 132,135 sequence homology between species, 131-134 monoamine oxidase A chimera properties, 140,148 Parkinson’s disease role, 130-131,151 prospects for study, 152 sequence homology with type A, 131-132 site-directed mutagenesis of flavin-binding sites covalent-binding site, 147-148

INDEX

401

dinucleotide-binding site, 136- 13 7, 140 fingerprint site, 144-147 overview, 135 second flavin adenine dinucleotidebinding site, 14 1- 144 table of mutants, 139 target selection, 135-136, 141 transient transfection system, 136 tissue and cell distribution, 149-150 mTOR. see Mammalian target of rapamycin Myocyte enhancer factor 2 (MEF2), histone deacetylase recruitment, 317

N N-CoR, thyroid hormone receptor corepressor, 62,65 NER, see Nucleotide excision repair Nitrate reductase, flavin-binding site, 141 Nucleosome, see Chromatin; GAL Nucleotide excision repair (NER), see DNA repair

P P53 functions, overview, 264 viral replication stimulation, 288 Parkinson’s disease, monoamine oxidase B role, 130-131, 151 PCNA, see Proliferating cell nuclear antigen PDE, see Cyclic nucleotide phosphodiesterase PDEl cafcium/cafmodulin binding, 20-22 inhibitors, 17 phosphorylative regulation, 21 proteolytic properties, 22 subtypes and kinetic properties, 20-21 tissue distribution, 22-23 PDE2 cyclic nucleotide stimulation and binding sites, 23-24 functions, 24 inhibitors, 17 isoforms, 23 kinetic parameters, 23 tissue distribution, 24

PDE3 functions, 26 inhibitors, 17 kinetic parameters, 24-25 regulation, 25-26 structures, 25-26 tissue distribution, 25-26 PDE4 amino-terminal domain, 27-28 functions, 29 genes, 26-27 metal studies of catalysis, 14, 16 regulatory domain functions, 28-29 rolipram inhibition, 17-18, 26, 29 tissue distribution, 26 PDE5 aflosteric cGMP-binding sites, 31-33 genes, 29-30 inhibitors, 17-18, 31, 33 kinetic parameters, 30 metaf studies of catalysis, 14, 16,30-3 phosphorylative regulation, 3 1-32 site-directed mutagenesis of catalytic residues, 12-13,30 tissue distribution, 33 PDE6 allosteric cGMP-binding sites, 34-35 inhibitors, 17 metal studies of catalysis, 14, 16 structure, 34 transducin-PDE-y-PDEnB interactions in phototransduction, 36-37 vision role, 33-35 PDE7 inhibitors, 17, 37-38 tissue distribution, 38 variants, 38 PDE8 inhibitors, 17,39 kinetic parameters, 38 tissue distribution, 39 PDE9 inhibitors, 17, 39 kinetic parameters, subtypes, 39 PDElO inhibitors, 17 structure, 40 tissue distribution,

39

40

1

402 PDEll inhibitors, 17 kinetic parameters, 40 regulation, 40 tissue distribution, 40-41 PDKl, S6 kinase as substrate, ill-112,122 PDR, see Phthalate dioxygenase reductase Phosphatidylcholine, biosynthesis enzymes, see CDP-choline:1,2diacylglycerol phosphotransferase; Choline ki nase; CTP:phosphocholine cytidylyltransferase homeostasis, 380-381 inhibition and apoptosis induction, 382386 overview, 362-364 regulation, 364-374 Phosphatidylethanolamine, biosynthesis ethanolamine cytidylyltxuxferase, 375-376 ethanolamine kinase, 375 ethanolamine phosphotmnsferase, 3 76 overview, 362-363,374-375 Phosphatidylinositol, biosynthesis CDP-diacylglycerol synthase, 377-378 diacylglycerol kinase, 377 overview, 362-363,376-377 phosphatidylinositol synthase, 378-379 Phosphatidylinositol3kinase (PI3K), S6 kinase effector studies, 113 Phosphatidylserine, biosynthesis regulation, 379 Phospholipases membrane phospholipid homeostasis maintenance, 380 PLA, functions, 381-382 groups, 38 1 Phthalate dioxygenase reductase (PDR), flavin-binding site, 14 1, 143 PI3K, see Phosphatidylinositol3-kinase PKB, see Protein kinase B PKC, see Protein kinase C PI.&, see Phospholipases Pol u, see DNA polymerase cr Pol B, see DNA polymerase B Pol8, see DNA polymerase 6 Pol ?? , see DNA polymerase E Proliferating cell nuclear antigen (PCNA) DNA polymerase holoenzyme assembly, 272-274,281

INDEX functions cell cycle regulation, 285-286 chromatin assembly, 285 DNA methylation, 284-285 DNA repair, 282-284 DNA replication, 263,282-283 overview, 263-264,291-292 protein interactions, 282-286 replication factor C in clamp loading, 280-281 structure, 281-282 Protein kinase B (PKB), S6 kinase effector studies, 113-114 Protein kinase C (PKC), S6 kinase effector studies, 114

R Regulator of G protein signaling (RGS) activity regulation and modulation, 348349 domains, 344-345,347-348 historical perspective of study, 342-344 membrane targeting, 348 prospects for study, 355 RGS9 cell type-specific expression RNA processing, 351 subunit association and control through stability, 351-352 transcriptional control, 350-351 effector modulation, 353-354 Ga selectivity, 349-350 knockout analysis, 354 membrane targeting by domains outside catalytic domain, 352-353 protein interactions, 354-355 RNA processing and variants, 348 specificity for G protein types, 344 subfamilies, 342 Replication factor C (RF-C) DNA polymerase holoenzyme assembly, 272-273 functions DNA polymerase switching, 28 1 overview, 262-264 proliferating cell nuclear antigen loading, 280-281 subunits, 2 79

403

INDEX

RF-C, see Replication factor C RGS, see Regulator of G protein signaling Ribosome, see also S6 cell cycle progression changes, 102 energy investment in biogenesis, 102 RNA polymerase II accessory factors, 286 recruitment, 247,289 Rohpram, cyclic nucleotide phosphodiesterase inhibition, 17-18, 26, 29

S S6 Drosophila S6 extraribosomal function, 107-108 gene and transcripts, 107 homology between species, 104-105 location in ribosomes, 105 phosphorylation inducers, 103 kinase, see S6 kinase regulatory functions, 105-107 sites, 103 -104 S’TOP messenger RNA regulation, 106-107, 117-118 S6K1, see S6 kinase S6 kinase (S6Kl) cell cycle progression role, 118-119 domains, 108, 110-111 Drosophila mutants advantages of study, 120-121 deficiency phenotype, 121 Minutes mutant comparison, 122 overexpression effects, 121-122 insulin production role, 119 isoforms, 108 knockout mouse phenotype, 119-120 phosphorylative activation PDKl as kinase, ill-112,122 signaling pathways, 113 -114 sites, 111-113 prospects for study, 122-123 S6K2 comparison to S6K1,108,119-120 5’TOP messenger RNA translation regulation, 106-107,117-118 upstream effecters mammalian target of rapamycin essential amino acids in activation, 116-117

mechanism of effector activity, 115116 phosphorylation, 115 sequence analysis, 115 phosphatidylinositol3kinase, 113 protein kinase B, 113 - 114 protein kinase C, 114 Scaffold attachment region, see Karyoskeleton Sildenafn, cyclic nucleotide phosphodiesterase inhibition, 17-18 Sin3 histone deacetylase recruitment, 316 thyroid hormone receptor corepressor, 62, 66 Site-directed mutagenesis monoamine oxidase B active site and catalytic residue mutagenesis, 148 - 149 flavin-binding sites covalent-binding site, 147-148 dinucleotide-binding site, 136- 13 7, 140 fingerprint site, 144 - 14 7 overview, 135 second flavin adenine dinucleotidebinding site, 14 1- 144 table of mutants, 139 target selection, 135-136, 141 transient transfection system, 136 PDE5 catalytic residues, 12-13,30 squalene synthase catalytic residues acidic residues in section B, 173 - 174 aromatic residues in sections A and C, 169-170,172-173 Asp-2 19, 174 Asp-223, 174 Phe-286,172-173 Phe-288,172-173 thioredoxin fusion proteins, 169 Tyr-171, 170,172 Tyr-174, 170 SMRT, thyroid hormone receptor corepressor, 62, 65 SQS, see SquaIene synthase Squalene synthase (SQS) assay, 169 deletion studies of catalytic domain, 168 dietary effects on levels, 161-162, 177179

404 Squalene synthase (cont.) domains, 166 - 168 functional overview, 158-159 genes cloning, 164 loci, 174 structure, 164, 174, 176, 180-181 historical perspective of research, 159, 161 hydropathy analysis, 167-168 interleukin-1 regulation, 190 - 19 1 lipopolysaccharide regulation, 189- 19 1 messenger RNA organ distribution, 176 - 177 sizes, 176 sterol effects on levels, 179 physiological significance of regulation, 178 purification of rat liver enzyme column chromatography, 162 dietary induction, 16 1- 162 microsome preparation, 162 size, 162-163 reaction mechanism, 158,166 sequence alignment of mammalian enzymes, 164-167 site-directed mutagenesis of catalytic residues acidic residues in section B, 173-174 aromatic residues in sections A and C, 169-170,172-173 Asp-2 19, 174 Asp-223, 174 Phe-286,172-173 Phe-288,172-173 thioredoxin fusion proteins, 169 Tyr-171,170,172 Tyr-174,170 species-specific functional domains, 168169 sterol regulatory element binding protein regulation accessory transcription factor binding sites, 181, 185 reporter assays of binding specificity, 185-187 scanning mutagenesis of promoter, 183, 185 SREBP-la accessory transcription factor requirements, 188-189 binding element specificity, 185 - 18 7

INDEX

nonhuman promoter specificity studies, 189 promoter mutation studies of binding, 187-189 sterol depletion studies, 189 SREBP-2 accessory transcription factor requirements, 188-189 binding element specificity, 185-187 nonhuman promoter specificity studies, 189 promoter mutation studies of binding, 187-189 sterol regulatory elements, 181 truncation analysis of promoter, 18 l183 subcellular localization, 163 - 164 therapeutic targeting, 159 tumor necrosis factor-a regulation, 190191 SIG1, thyroid hormone receptor coactivator, 62-63 SREBP, see Sterol regulatory element binding protein Sterol regulatory element binding protein (SREBP) accessory factors, 180 biosynthesis, 180 squalene synthase gene regulation accessory transcription factor binding sites, 181, 185 reporter assays of binding specificity, 185-187 scanning mutagenesis of promoter, 183, 185 SREBP-la accessory transcription factor requirements, 188-189 binding element specificity, 185-187 nonhuman promoter specificity studies, 189 promoter mutation studies of binding, 187-189 sterol depletion studies, 189 SREBP-2 accessory transcription factor requirements, 188 - 189 binding element specificity, 185 - 18 7 nonhuman promoter specificity studies, 189

405

INDEX promoter mutation studies of binding, 187-189 sterol regulatory elements, 181 truncation analysis of promoter, 18 l183 sterol regulatory element features, 179180 target genes, 179 types, 180 Stromelysin-3, intestinal development role in amphibians apoptosis association, 86-87, 89 developmental expression pattern, 86 substrates, 89-90

T TH, see Thyroid hormone Thyroid hormone (TH) amphibian metamorphosis role, see Am phibian metamorphosis, thyroid hormone regulation chromatin remodeling in gene regulation chromatin disruption assays, 66,68 histone acetylation regulation, 65-66 Xenopus oocyte system advantages, 63, 65,93 deficiency in cretinism, 55 - 56 interconversion of forms, 58-59 intestine development, see Intestinal remodeling, Xenopus studies of thyroid hormone mammalian development role, 55-56 receptors DNA binding, 60-61 domains, 59-60

hormone binding, 60 transcriptional regulation of genes activation versus repression, 61-62 coactivators, 62-63 corepressors, 62 model of action, 68,70 TNF-o, see Tumor necrosis factor-o 5’TOP messenger RNA, translation regulation by S6 kinase, 106-107,117-118 Transducin, phototransduction, 343 Tiichostatin A (TSA), histone deacetylase inhibition studies of thyroid hormone action, 66 TSA, see Trichostatin A Tumor necrosis factor-o (TNF-o), squalene synthase regulation, 190 - 19 1

U Ubiquitination,

histones, 306-307

V Villin, thyroid hormone regulation in intestinal development, 83 - 84 Viral replication, transcription factors, 286287

Z Zaprinast, cyclic nucleotide phosphoclesterase inhibition, 17-18 Zinc, cyclic nucleotide phosphodiesterase binding and catalysis, 14-16,30-31