Advances in Genetics, Volume 55
Serial Editors
Jeffery C. Hall Waltham, Massachusetts
Jay C. Dunlap Hanover, New Hampshire
Theodore Friedmann La Jolla, California
Veronica van Heyningen Edinburgh, United Kingdom
Contents Contributors
ix
1 Site-Specific DNA Recombinases as Instruments for 1 Genomic Surgery Aram Akopian and W. Marshall Stark I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII.
Introduction 2 Inadequacy of Current Methods 2 Site-Specific Recombination 4 Mechanisms of Site-Specific Recombination 5 Transposition 8 Applied Site-Specific Recombination and Transposition Systems 9 Changing Recombination Site Specificity 9 Mutagenesis-Selection Strategies 10 Structure-Based Strategies: Z-Resolvases 12 Targeting Transposition to Specific Sequences 15 General Considerations in Applications of Site-Specific Recombinases and Transposases 16 Prospects and Conclusions 18 References 19
2 Human Clinical Trials of Plasmid 25 DNA Vaccines
Margaret A. Liu and Jeffrey B. Ulmer I. II. III. IV. V.
Background 26 Development of DNA Vaccines for Clinical Trials Human Clinical Trials 31 Limitations of DNA Vaccine Potency 33 Prospects and Conclusion 35 References 35
28
v
Contents
vi
3 Genetic and Environmental Influences on Antisocial Behaviors: Evidence from 41 Behavioral–Genetic Research Terrie E. Moffitt
I. Appraising the Quantitative Behavioral–Genetic Evidence Base 44 II. Estimating the Relative Influences of Genes and Environments 58 III. Do the Omnibus Estimates of Genetic and Environmental Influences Always Apply, or do These Influences Vary Under Different Conditions? 66 IV. Testing Developmental Theory of Antisocial Behavior 71 V. Testing Hypotheses About Environmental Causation 74 VI. Testing the Hypothesis of Interaction Between Genes and Environments 80 VII. The Way Forward 88 References 91
4 Genetics of Graviperception in Animals
105
Kathleen M. Beckingham, Michael J. Texada, Dean A. Baker, Ravi Munjaal, and J. Douglas Armstrong I. Introduction 106 II. Vertebrate Genetic Models (M. musculus and D. rerio) 107 III. The Arthropod Genetic Model (D. melanogaster) IV. The Nematode Model System (C. elegans) 135 V. Conclusion 137 References 138
5 Retroviral DNA Integration—Mechanism 147 and Consequences
Mary K. Lewinski and Frederic D. Bushman I. Introduction 148 II. Retroviral Life Cycle 148 III. Mechanism of Integration 149
120
Contents
IV. V. VI. VII. VIII.
Integrase Structure 153 Composition of Integrase Complexes In Vivo 159 Retroviral Integration Targeting 162 Consequences of Integration into Host Chromosomes Conclusions 169 References 170
Index
183
vii
166
Contributors Numbers in parentheses indicate the pages on which the authors’ contributions begin.
Aram Akopian (1) Institute of Biomedical & Life Sciences, University of Glasgow, Glasgow G11 6NU, United Kingdom J. Douglas Armstrong (105) School of Informatics, University of Edinburgh, Edinburgh EH8 9LE, United Kingdom Dean A. Baker (105) School of Informatics, University of Edinburgh, Edinburgh EH8 9LE, United Kingdom Kathleen M. Beckingham (105) Department of Biochemistry and Cell Biology, Rice University, Houston, Texas 77005 Frederic D. Bushman (147) Department of Microbiology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104 Mary K. Lewinski (147) Infectious Disease Laboratory, The Salk Institute for Biological Studies, La Jolla, California 92186 Margaret A. Liu (25) Transgene S.A., 11, rue de Molsheim, 67082 Strasbourg Cedex, France Terrie E. Moffitt (41) Institute of Psychiatry, King’s College London, SE5 8AF, United Kingdom; University of Wisconsin, Madison, Wisconsin 53706 Ravi Munjaal (105) Department of Biochemistry and Cell Biology, Rice University, Houston, Texas 77005 W. Marshall Stark (1) Institute of Biomedical & Life Sciences, University of Glasgow, Glasgow G11 6NU, United Kingdom Michael J. Texada (105) Department of Biochemistry and Cell Biology, Rice University, Houston, Texas 77005 Jeffrey B. Ulmer (25) Chiron Corporation, Emeryville, California 94608
ix
1
Site‐Specific DNA Recombinases as Instruments for Genomic Surgery Aram Akopian and W. Marshall Stark Institute of Biomedical & Life Sciences University of Glasgow, Glasgow G11 6NU United Kingdom
I. II. III. IV. V. VI. VII. VIII. IX. X. XI.
XII.
Introduction Inadequacy of Current Methods Site‐Specific Recombination Mechanisms of Site‐Specific Recombination Transposition Applied Site‐Specific Recombination and Transposition Systems Changing Recombination Site Specificity Mutagenesis‐Selection Strategies Structure‐Based Strategies: Z‐Resolvases Targeting Transposition to Specific Sequences General Considerations in Applications of Site‐Specific Recombinases and Transposases A. Recombinase and transgene “delivery” B. Chromatin C. Reversibility D. Nonspecific reactions Prospects and Conclusions References
Advances in Genetics, Vol. 55 Copyright 2005, Elsevier Inc. All rights reserved.
0065-2660/05 $35.00 DOI: 10.1016/S0065-2660(05)55001-6
2
Akopian and Stark
ABSTRACT Site‐specific DNA recombinases can “cut and paste” DNA. For example, they can promote excision of specific DNA segments or insertion of new DNA segments in specific places. However, natural recombinases act only at their cognate recombination sites, so current applications are limited to genetically modified organisms in which these sites have been introduced into the genome. Transposases also catalyze DNA rearrangements; they promote insertion of specific DNA sequences but at nonspecific locations. Applicability of site‐ specific recombinases and transposases in experimental genetics, biotechnology, and gene therapy would be much wider if they could be reengineered so as to act specifically at chosen sequences within an organism’s natural genome. This review will discuss progress towards the creation of such “designer” recombinases. ß 2005, Elsevier Inc.
I. INTRODUCTION As numerous genomes are sequenced and our understanding of genome function grows, the possibilities for changing genomic DNA sequence to our benefit become more apparent. We could add new useful bits of DNA in specific places and delete or change defective or undesirable bits. This “genomic surgery” could be applied to the treatment of disease as well as in biotechnology and experimental genetics. The development of realistic methods for genomic surgery will bring many opportunities for improving our health and well being, along with dangers and ethical problems that will be faced by science and society. In this review, we will discuss advances in the still‐primitive methods that are under consideration, focusing on the attempts to adapt natural enzymes that “cut and paste” DNA—recombinases and transposases—so that their activity can be directed to chosen sequences.
II. INADEQUACY OF CURRENT METHODS For many years, it has been possible to make artificial alterations to an organism’s genetic material. The alteration may be temporary (e.g., when foreign nonreplicating DNA is introduced into a cell or when gene expression is altered by RNAi) or permanent (e.g., when transgenic DNA integrates into a chromosome). Integration of a foreign DNA segment can be targeted to a specific genomic locus by methods based on homologous recombination (HR; see in later section). Why then do we need new methods to target DNA manipulation to particular genomic sites?
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
3
The currently available methods for directing DNA modification to specific target sequences lack the versatility, efficiency, and specificity, which will be needed if genomic surgery is to take off (Yanez and Porter, 1998). For example, consider a scenario in which a new functional gene is to be integrated into a mammalian genome. Without targeting, a transfected DNA fragment will generally integrate inefficiently and at essentially random positions (Lacy et al., 1983; Smith, 2001). Transgene expression might therefore be subject to locus‐ specific downregulation by epigenetic silencing processes (Pannell and Ellis, 2001) or the integrated DNA might affect expression of nearby genes, thus potentially causing cell malfunction or oncogenesis (Palmiter and Brinster, 1986). Ideally, the transgene should be integrated at a locus where it will be expressed properly, and it will not interfere with other functions of the genome. Use of HR is the only current means of targeting integration to any chosen genomic locus. By embedding the gene to be transfected in the middle of a few kilobase‐pairs (kbp) of target site sequence, the cell’s HR machinery can be co‐ opted to promote integration at that site. This approach has been successfully applied as an experimental tool in some organisms for integration of new genes or to “knock out” resident genes (e.g., Colosimo et al., 2000; Vasquez et al., 2001). Unfortunately, integration by HR is inefficient, and specificity for the chosen target locus is low. Normally, the great majority of cells following transfection have not integrated the transfected DNA, and the transgene lands at the desired target site in only a small fraction of the remainder (Smith, 2001; Vasquez et al., 2001). Homologous recombination is therefore of practical use only if one can select the correctly modified cells and, if necessary, induce them to participate in embryogenesis. The options for improvement of HR‐based methods are limited because HR is crucial for the maintenance of the genome; tampering with its natural components will possibly have serious biological consequences and would probably only be feasible in experimental systems. A further technical limitation of HR is that the transgene must be flanked by long sequences homologous to the target. This can restrict the size of genes to be introduced with viral vectors, which can only carry a limited length of foreign sequence (Monahan and Samulski, 2000; Yanez and Porter, 1998). Methods for enhancing the efficiency of HR at chosen sites have been developed, based on the introduction of factors with sequence recognition specificity (DNA‐binding proteins or oligonucleotides) (Uil et al., 2003). However, it seems that, for more advanced genomic surgery, alternatives to HR will be required. In later sections, we consider whether site‐specific recombinases and transposases may be useful “surgical instruments.” Other strategies for alteration of specific genomic sequences are also being investigated (Belfort et al., 2002; Collins et al., 2003; Epinat et al., 2003; Guo et al., 2000; Lambowitz and Zimmerly, 2004; Monahan and Samulski, 2000; Portlock and Calos, 2003; Uil et al., 2003).
4
Akopian and Stark
III. SITE‐SPECIFIC RECOMBINATION Site‐specific recombinases rearrange DNA sequences by catalyzing cleavage and rejoining of DNA strands at specific short DNA sequences (sites) to which they bind. The outcome of the reaction can be excision of a DNA sequence segment bounded by two sites, or the reverse of this process, integration, or inversion of the orientation of a segment of DNA bounded by two sites (Fig. 1.1). An attractive feature of these systems as potential tools for genomic surgery is that they can be highly specific, efficient, and fast (Nash, 1996). Microbial site‐specific recombination systems have been widely exploited in mammals and other higher eukaryotes for experimental research or projected gene therapy/biotechnology applications (see Section VI). However, their use is currently limited to situations where one or more of the sites recognized by the recombinase have already been introduced into the genome by other methods. It would be a major advance if we could engineer recombinases to act with high efficiency and specificity at sequences that are already there. The best‐characterized site‐specific recombination systems come from bacteria and yeasts. Most of the systems can be assigned to one of two large families, according to the type of recombinase that is used. These two families, the tyrosine recombinases and the serine recombinases, are so‐called because the conserved nucleophilic amino acid residue that attacks and becomes linked to the DNA during strand exchange is either a tyrosine or a serine. The well‐ known enzymes Cre, FLP, and integrase are tyrosine recombinases; examples from the serine recombinase family are resolvase and C31 integrase (Nash, 1996). The two families are unrelated to each other, having different protein structures and reaction mechanisms.
Figure 1.1. Substrates and products of site‐specific recombination reactions. The diagrams show how the relationship of two recombining sites (small black and white arrows) determines the type of recombination product formed. On the left, excision of DNA between directly repeated sites and the reverse reaction, integration; on the right, inversion of the DNA segment between sites in inverted repeat.
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
5
Natural site‐specific recombination systems have evolved to carry out specific “programmed” genetic rearrangements in their host organisms. Therefore, their activity is usually tightly regulated, and the type of rearrangement they bring about (excision, inversion, or integration) is usually specified. These constraints can cause problems for those who wish to subvert the recombinases to become tools for genetic manipulation.
IV. MECHANISMS OF SITE‐SPECIFIC RECOMBINATION The two DNA sites that are to recombine (typically a few tens of base‐pains(bp) long; Fig. 1.2B) must first be recognized by the recombinase protein. The protein‐bound sites must then be brought together (synapsed) prior to strand exchange, which involves cutting and rejoining of the DNA strands. During strand exchange, the recombinase transiently becomes covalently linked to a phosphate of the DNA backbone via the hydroxyl group of the nucleophilic tyrosine or serine residue (Hallet and Sherratt, 1997; Nash, 1996). The strands are broken at fixed points in the site sequence. Tyrosine recombinases exchange strands one pair at a time, and thus the reaction proceeds via an intermediate which is analogous to the Holliday junction of HR (Holliday, 1964). In contrast, serine recombinases make intermediates in which all four DNA strands of the two recombination sites are broken. These mechanisms are summarized in Fig. 1.2A. The recombination site generally has an asymmetric sequence and thus a “left” and a “right” end; each left end is joined to the right end of the partner site, so as to reconstitute similar sites in the recombination products. This polarity is often assured by an asymmetric “overlap” sequence between the staggered points of top and bottom strand exchange, so that recombination of misaligned sites would produce mismatched base pairs in the recombinants (Fig. 1.2B). The connectivity of the two recombining sites in the substrate DNA therefore defines the type of product formed: inversion, if the sites are oriented in opposite directions to each other; excision, if they are directly repeated (“head to tail”); and integration, if they are on separate molecules (Fig. 1.1). Some recombinases will act on pairs of sites in any relationship (e.g., Cre and Flp), whereas others are specific for one type of relationship (e.g., resolvase, which only catalyzes excision). Our understanding of the mechanisms of site‐specific recombination has been greatly advanced by a number of crystal structures showing recombinases bound to their cognate sites, and intermediates in the strand exchange process itself (Chen et al., 2000; Grindley, 2002; Guo et al., 1999; Van Duyne, 2002; Yang and Steitz, 1995). Structures of Cre (Fig. 1.3) and Flp show a recombinase tetramer synapsing two recombination sites. Each subunit of these tyrosine recombinases wraps around the DNA, making many contacts with it.
6
Akopian and Stark
Figure 1.2. (A) Mechanisms of DNA strand exchange by serine recombinases and tyrosine recombinases. The pairs of thick or thin lines represent double‐stranded DNA. The small arrowheads mark the ends of a “site” of a few tens of basepairs (Fig. 1.2B), which is recognized specifically by the recombinase enzyme (not shown). The staggered lines show where the strands are broken at the centres of the sites. Note that the sites are aligned in parallel in the upper row (serine recombinases), but in antiparallel in the lower row (tyrosine recombinases), reflecting the proposed structures of reaction intermediates. See text for further details. (B) Recombination sites. The sequences of two simple recombination sites are shown: loxP, the site for Cre recombinase, and attB, one of the two nonidentical sites for C31 integrase. The loxP site has twofold (palindrome) symmetry, indicated by the long arrows, except in the central 8 bp; in contrast, C31 attB is quite asymmetric. The short black arrows indicate the positions of breakage and rejoining of the DNA strands. The asymmetric ‘overlap’ sequence between the two arrows is important in specifying that left and right half‐sites are joined in the recombinant products. The overlap is typically 2 bp for serine recombinase sites, as in C31 attB, and 5–8 bp for tyrosine recombinase sites (6 bp in loxP).
The basis of sequence‐specific binding by these proteins is still not clear. A structure of the serine recombinase resolvase (Fig. 1.3) shows a recombinase dimer bound to a single site. A small C‐terminal domain of each resolvase subunit, thought to be responsible for much of the sequence specificity, binds in the major groove of the DNA. The larger N‐terminal domains, which contain the active site and make intersubunit interactions, contact the DNA in the minor groove on the opposite side of the double helix from the C‐terminal domains.
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
7
Figure 1.3. Crystal structures of site‐specific recombinases bound to DNA. (A) A dimer of the serine recombinase resolvase, bound to ‘site I’, the part of its recombination site where strand exchange occurs (Yang and Steitz, 1995). (B) A tetramer of the tyrosine recombinase Cre in a synapse with two loxP sites (Guo et al., 1999). The DNA (cream and blue) is in spacefill representation, and the protein (green and yellow) is in a ribbon representation. The two loxP sites in B are in approximately antiparallel alignment (see Fig. 1.1). The pictures were created with the program PYMOL.
8
Akopian and Stark
V. TRANSPOSITION Transposition has much in common with site‐specific recombination. A semiautonomous DNA element called a transposon moves from one locus to another in a reaction catalyzed by a transposase enzyme. The ends of the transposon must be recognized and acted upon by the transposase site‐specifically, but the transposon DNA is then inserted at new positions, which are generally not specific (Fig. 1.4). The mechanisms of transposition are diverse (Craig, 2002a) and can even involve RNA intermediates (the process being then called retrotransposition). DNA transposases (those that catalyze direct transfer of DNA from one locus to another) belong to several different structural groups. Some are related to the serine and tyrosine site‐specific recombinases introduced in Section IV. The best characterized are the so‐called “DDE” transposases, which are very widespread; transposons that encode them are found in bacteria, archaea, and eukarya. Like site‐specific recombinases, transposases have already attracted considerable interest as potential tools for gene therapy and biotechnology (Boeke, 2002). They share with site‐specific recombinases the ability to promote highly efficient, specific, and fast reactions. By their nature, transposases are best adapted to promote integration of a DNA segment, flanked by short “ends” from their cognate transposon, into target DNA. The transposase breaks either one or both strands at each end and then joins the broken ends to the target DNA. The complete process normally involves subsequent “tidying‐up” operations, which require host enzymes (Craig, 2002a). Usually, the transposase has little or no selectivity for specific target sequences, so transposon insertions can occur
Figure 1.4. Transposition. In the simplest case, as shown here, a DNA transposon (stippled rectangle) is excised from the DNA by its transposase enzyme acting at specific end sequences (arrowheads), and then inserted into a random site in target DNA (thicker lines).
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
9
anywhere in the genome; for exceptions that do have strongly preferred target sites or insert in specific regions of the genome, see Bushman, 2003; Craig, 2002b. The long terminal repeat (LTR) retrotransposons and LTR retroviruses reverse‐transcribe their RNA genome into DNA before integrating it into the host genome. Their integrase enzymes belong to the DDE transposase family, and the mechanism of integration is similar to that of typical DNA transposons with DDE transposases (Craig, 2002a). Integration by the non‐LTR retrotransposons involves different types of enzymes and direct interaction of RNA molecules with the target DNA. Non‐LTR retrotransposition systems may also have potential for targeted gene integration (Eickbush, 2002). In Section X, we will briefly review attempts to target insertion mediated by DDE transposases/retroviral integrases.
VI. APPLIED SITE‐SPECIFIC RECOMBINATION AND TRANSPOSITION SYSTEMS Recombinases have become very popular tools for manipulating DNA in vitro and in vivo (Boeke, 2002; Branda and Dymecki, 2004; Gorman and Bullock, 2000; Groth and Calos, 2003; Kilby et al., 1993; Kolb, 2002; Nagy, 2000). The most frequently used enzymes have been Cre recombinase from bacteriophage P1 and Flp recombinase from Saccharomyces cerevisiae. The 38‐kDa Cre protein promotes recombination between two 34‐bp loxP sites, and the 43‐kDa Flp recombinase acts on 34‐bp FRT sites. Advantages of these systems are their short DNA recombination sites, enzyme stability in vivo, and the robustness of their activity even when acting upon chromatin‐associated DNA (Jayaram et al., 2002; Sauer, 2002). Numerous other recombinases have been used in biotechnology or investigated as possible biotechnology tools (Boeke, 2002). Similarly, several DNA transposition systems have become popular for sequencing, mutagenesis, and transgene integration purposes (Boeke, 2002); and retroviral integration is being widely investigated for possible uses in gene therapy (see Sections X and XI).
VII. CHANGING RECOMBINATION SITE SPECIFICITY Hundreds of different natural site‐specific recombination systems have been identified, so there is a large natural “library” of sites with different sequences, acted upon by known recombinases. However, it is still very unlikely that any of these sequences will be found by chance in a useful place in a genome of interest, as most natural recombination sites are at least about 25 bp long. Furthermore,
10
Akopian and Stark
characterization of a new system to the stage where it could be used as a research tool requires a considerable amount of work. A more practical approach is to mutate or redesign a well‐characterized recombinase so that it recognizes a new sequence. Changing the sequence recognition properties of site‐specific recombinases has been a subject of investigation for many years; initially the aims were to provide insight on structure and mechanism and to create useful experimental tools. Recognition by serine recombinases related to resolvase has been altered by point mutation (Grindley, 1993) or by replacing part or all of the entire DNA‐binding domain with the equivalent sequence from a related protein (Ackroyd et al., 1990; Avila et al., 1990; Schneider et al., 2000). Similar mutation and domain‐swap approaches were used to change the specificities of tyrosine recombinases (Dorgai et al., 1995; Shaikh and Sadowski, 2000; Yagil et al., 1995). Recently, researchers have begun to develop more systematic strategies for altering recombinase site recognition, with the long‐term goal of creating more useful and adaptable systems for genetic manipulation in vivo. Two types of strategy are discussed in Sections IV and V. As hinted previously, it would be a great advance if integration could be targeted efficiently and specifically to even one good natural site in a genome of interest (e.g., human), where gene expression is optimal and there would be no adverse effects. One approach is therefore to choose a natural sequence with some similarity to the site for a useful recombinase, often one that has been shown to act as a “pseudosite” (where the recombinase occasionally catalyzes recombination). One can then attempt to produce variants of the recombinase that act there efficiently and specifically by multiple rounds of mutagenesis and selection. A more radical approach is the attempt to design a recombinase whose sequence specificity can be altered at will by changing a “DNA recognition module.”
VIII. MUTAGENESIS‐SELECTION STRATEGIES C31 Int, a serine recombinase, mediates integration by recombination of phage attP (39 bp) and bacterial genomic attB (34 bp) sites (Groth and Calos, 2003). Integration is irreversible in the reconstituted system lacking factors encoded by the phage C31 or its Streptomyces host. The relatively short sites and irreversible reaction make the C31 Int system a very attractive candidate for applications. There is no high‐resolution structural information as yet on C31 Int or any of its close relatives, but current biochemical evidence suggests that DNA recognition involves multiple domains within these large proteins. Systematic redesign of sequence specificity by “protein engineering” is therefore not an option at present. Although there are high‐resolution structures of the tyrosine recombinases Cre and Flp bound to their cognate sites, a protein‐engineering
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
11
approach is still problematic for them because it is difficult to distinguish residues involved in DNA recognition from those involved in catalysis (Section IV). The mutagenesis‐selection approach for altering site recognition, which does not depend on a deep understanding of the recombinase structure, has therefore been adopted for Cre, Flp, and C31 Int. A typical procedure employs an in vivo test substrate, which is designed so that recombination at the sites under investigation alters the expression of an observable marker gene, such as a fluorescent protein or the lacZ‐encoded ‐fragment of ‐Gal (e.g., by excision of the gene, separating it from its promoter; Fig. 1.5). The substrate is established in the organism of choice, then challenged by introduction of DNA encoding a “library” of recombinase mutants (which may be generated by any appropriate method, such as error‐ prone PCR). Cells containing active mutants are then selected (e.g., by a change in color [Santoro and Schultz, 2002; Sclimenti et al., 2001]). A variation in the procedure, which facilitates the isolation of active mutants, is to combine the recombinase gene and the test construct in one plasmid (Buchholz and Stewart, 2001). Recombinase mutants obtained in the first cycle of mutagenesis‐ selection may then be used as the starting point for further cycles, either selecting for increased recombination efficiency on the same site or activity on a site that contains further changes from the wild‐type sequence. Mutants obtained by mutagenesis‐selection procedures, which can recombine at sites at which there are some differences compared with the wild‐ type site, generally have activity on the latter (normal) site (i.e., they have
Figure 1.5. An example of an assay for recombinase activity. Excision separates the reading frame of a marker gene (in this example green fluorescent protein, GFP) from its promoter, abolishing expression of the gene. Cells in which recombination has occurred can thus be detected (in this case, by loss of fluorescence). See text for further details.
12
Akopian and Stark
relaxed specificity). Therefore, some groups have developed methods to select, either simultaneously or sequentially, for activity on a variant site and loss of activity on the wild‐type site (Buchholz and Stewart, 2001; Voziyanov et al., 2002). A frequent elaboration of the mutagenesis‐selection procedure is to use “DNA shuffling,” a method that allows the generation of random combinations of a number of existing mutants (Stemmer, 1994), along with further mutagenesis; this potentially speeds up the selection process (Buchholz and Stewart, 2001; Santoro and Schultz, 2002; Sclimenti et al., 2001; Voziyanov et al., 2002, 2003). Impressive changes to the sequence specificities of Cre, Flp, and C31 Int have been achieved by these methods (see references previously cited), and specific activity at certain natural sequences in mammalian genomic DNA has been demonstrated (Buchholz and Stewart, 2001; Sclimenti et al., 2001; Thyagarajan et al., 2001). However, the strategy is not yet applicable to the targeting of sequences other than those that have been shown to be pseudosites or whose sequence is quite similar to the recombinase’s proper site. One might predict that, as more “specificity mutants” are identified, this approach would highlight the recombinase residues that are most important for sequence recognition. Further mutagenesis could then focus on these residues. Currently, it seems that many residues in Cre and Flp, along the whole length of the primary amino acid sequence, contribute to specificity. Some residues do appear as “hot spots” for mutations in the altered‐specificity variants, but as yet there is no clear pattern in the published data that would allow a more rational approach to the creation of new variants (Hartung and Kisters‐Woike, 1998; Voziyanov et al., 2003).
IX. STRUCTURE‐BASED STRATEGIES: Z‐RESOLVASES Crystallography of the serine recombinase resolvase reveals a modular structure. The 140‐residue N‐terminal domain, which contains all residues known to be involved in catalysis, is structurally and spatially distinct from the 40‐residue C‐terminal “helix‐turn‐helix” domain that is the primary determinant of sequence‐specific DNA binding (Fig. 1.3; Abdel‐Meguid et al., 1984; Grindley, 2002; Yang and Steitz, 1995). The two domains are connected by a short linker sequence that associates with the DNA minor groove. Resolvase and its very close relative Tn3 resolvase have been extensively studied in vitro (Grindley, 2002). The length of the tripartite recombination site (res; 114 bp), together with strong selectivities for supercoiling and pairs of sites that are in direct (“head‐to‐tail”) repeat on the same DNA molecule, renders these enzymes unsuitable for most of the envisaged applications of site‐specific recombinases. However, recent studies have led to “hyperactive” resolvase variants that
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
13
recombine two copies of a short dimer‐binding site (the 28 bp site I of res) and no longer require supercoiling or directly repeated sites (Arnold et al., 1999; Burke et al., 2004). These variants might be put to the same kinds of uses as Cre and Flp. Mutant versions of relatives of Tn3/ resolvase have also been shown to act at simple dimer‐binding recombination sites without the need of additional factors (Johnson, 2002). Some minor alterations in sequence selectivity of this group of serine recombinases have been made by mutation, or by substituting the DNA‐binding domain with that of another member of the group (see Section VII). However, it became apparent that the sequence specificity of hyperactive resolvase variants might be more radically altered by replacing the DNA‐binding domain with one from an unrelated protein. The zinc‐finger DNA‐binding domain of the mouse transcription factor Zif268 was an especially attractive choice for this purpose. The Zif268 DNA‐binding domain is small (~90 amino acids), is monomeric in its functional form, and recognizes a short (9‐bp) sequence with high specificity. Its structure in a complex with DNA has been solved (Elrod‐Erickson et al., 1996; Pavletich and Pabo, 1991). Significantly, it is the focus of a campaign, involving a number of research groups, to create engineered zinc‐finger proteins, which can recognize any defined short DNA sequence (Beerli and Barbas, 2002; Pabo et al., 2001). Variants of the Zif268 domain that recognize some specific sequences in mammalian and viral genomes have already been created (Reynolds et al., 2003). The domain is composed of three similar zinc‐finger modules, each recognizing 3–4 bp of its 9‐bp target sequence. In each finger, only a few residues make base‐specific contacts with the DNA. It is therefore possible to simplify the task of selection for recognition of a new sequence by focusing on one finger at a time, and on the most important residues for recognition (Beerli and Barbas, 2002). The so‐called “phage display” technique has been adapted for the efficient selection of Zif268 domain variants that bind a chosen DNA sequence with very high affinity (Pabo et al., 2001), and other methods are being developed to select for high specificity (Hurt et al., 2003). These Zif268 domain variants with novel sequence specificity can act as artificial transcription factors, repressing or activating transcription from promoters close to their binding sites (Reynolds et al., 2003). They can also be used to tether an enzyme to a specific site on DNA. Novel chimeric nucleases have been created in which a nonspecific DNA cleavage domain is linked to a Zif268‐derived site‐specific DNA‐binding domain (Chandrasegaran and Smith, 1999). Cleavage mediated by the DNA‐bound chimeric enzyme occurs selectively at the chosen recognition site, and this can increase the in vivo efficiency of HR at that site (Bibikova et al., 2002, 2003; Porteus and Baltimore, 2003). Chimeric recombinases (“Z‐resolvases”) in which the catalytic domain of a hyperactive Tn3 resolvase variant is linked to the DNA‐binding domain of
14
Akopian and Stark
Figure 1.6. Cartoon of a Z‐site þ Z‐resolvase complex. Two 9 bp sequences (boxes with three arrows) are each recognized by a 3‐zinc‐finger DNA‐binding domain (dark grey shape) derived from that of Zif268. The 22 bp intervening sequence (hatched) interacts with the dimerized catalytic domains of a mutant version of Tn3 resolvase (light grey ovals) (see text for details). The Zif268 domain and the resolvase catalytic domain are linked by a short peptide (curved lines). The positions of strand breakage and rejoining are indicated by the staggered thin line.
Zif268 have been shown to recombine in vivo and in vitro specifically at “Z‐sites” (Akopian et al., 2003). Z‐sites consist of appropriately spaced pairs of 9‐bp sequence motifs recognized by the Zif268 domain, flanking a central sequence which is acted upon by the catalytic domains (Fig. 1.6). Each Z‐site binds a dimer of Z‐resolvase. The Z‐resolvase/Z‐site specificity was very high, and no activity on the natural target sequence of the catalytic domain (res site I) was detected. The rate of recombination was shown to depend on structural features of the Z‐resolvase and on attributes of the Z‐site sequence. The distance between the two 9‐bp motifs recognized by the Zif268 domains was critical, 22 bp being optimal. The reasons for this distance requirement are unknown as yet. The length and sequence of the peptide linking the two domains of Z‐resolvase also affected activity. In principle, any sequence of about 40 bp might be regarded as a potential Z‐site. Its ends could be recognized by variant Zif268 domains, and its central sequence could be cut and rejoined by resolvase catalytic domains. However, useable sequences might be relatively scarce. Thirteen basepairs of the original 28‐bp Tn3 res site I sequence were retained at the centre of all the Z‐sites tested so far (Fig. 1.6). Although current evidence suggests that resolvase contacts only a few basepairs of this sequence (Hatfull et al., 1988; Yang and Steitz, 1995), very different central sequences might be unsuitable. Sequences resembling that of the centre of res site I are likely to be the best targets, especially those with a central TATA motif, because these 4 bp contain the bonds that are broken and rejoined by resolvase. The 2‐bp “overlap” sequence in the TATA motif (AT) is palindromic, so a TATA‐centered Z‐site does not
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
15
have a polarity. A substrate with two of these sites could recombine to give both excision and inversion products (see Section IV and Fig. 1.2B). Polarity can be restored by an asymmetric overlap sequence (e.g., AC), which prevents ligation of two left half‐sites or two right half‐sites (Stark et al., 1991). A pair of recombining sites must have the same overlap sequence; if not, recombination is blocked because the products would contain mismatched basepairs. Because resolvase assembles on DNA as a dimer, two Zif268 domain variants might have to be created, one to bind to each “half‐site” of a chosen genomic sequence (Fig. 1.6). In vivo targeting of a heterodimeric enzyme, with each subunit being fused to a different Zif268 domain variant, has a precedence in the studies on chimeric nucleases (see earlier section; Bibikova et al., 2003; Porteus and Baltimore, 2003). More generally, up to four Z‐resolvases, each with the same catalytic domain but a different Zif268 domain variant, could theoretically be used to recombine between two different natural Z‐sites (e.g., to excise a chosen segment of genomic DNA). However, strong recognition of just one half‐site by one subunit of a Z‐resolvase homodimer can suffice to target recombination (Akopian, A., unpublished results). Therefore, it is likely to be advantageous to use rare sequences as targets for each of the Zif268‐derived domains to minimize reactions at sequences other than the chosen Z‐site. The as‐yet hypothetical task of selecting a new site specificity for a Z‐resolvase can be split into parts, because of the enzyme’s modular nature. Zinc‐ finger domain variants that recognize the outer sequences of the Z‐site, and catalytic domains with optimal activity on the central sequence, would be created separately. Candidate recombinases targeted to the full chosen sequence would then be assembled by linking the selected domains. Enhancement of recombination activity on the chosen site could be achieved by mutagenesis‐ selection, as described in the Section VIII.
X. TARGETING TRANSPOSITION TO SPECIFIC SEQUENCES The problem of targeting transposition to specific genomic sites is somewhat different from the analogous problem for site‐specific recombination. A recombinase must be redirected from its natural site to a new target sequence, at which it would normally be inactive. In contrast, transposases act specifically at their cognate transposon ends but typically can insert them into many target sequences; the problem is to direct insertion to a chosen target site and ideally prevent insertion at any others. Whereas a “designer” site‐specific recombinase is generally intended to have activity at a single new site, transposase activity can potentially be targeted to larger regions of genomic DNA. One option is to “tether” the transposase to a specific binding site (e.g., by using an attached DNA‐binding domain), in order
16
Akopian and Stark
to favor transposition nearby. Alternatively, the transposase may be redesigned so as to interact with a specific DNA‐binding protein, so transposition occurs in regions of genomic DNA where the target protein is present. This strategy echoes the behavior of some natural retroviral integrases, which interact similarly with endogenous DNA‐binding proteins (Bushman, 2003; Sandmeyer, 2003). The published studies are all based on retroviral integrases, but similar approaches could be adopted for classical transposases. Fusions of HIV integrase with the DNA‐binding protein LexA (Goulaouic and Chow, 1996; Katz et al., 1996), Zif268 (Bushman and Miller, 1997), or E2C (Tan et al., 2004), and of Moloney murine leukemia virus integrase with Sp1 (Peng et al., 2002), have all been studied—the idea being that tethering the transposition intermediate to an appropriate binding site in the DNA will enhance insertion nearby. However, the results have been rather discouraging; the extent of integration near the target site was only a few times higher than that expected if target choice is random. Recent work by Voytas’ group (Zhu et al., 2003) showed how the yeast retrotransposon Ty5 is naturally targeted to heterochromatin by interaction with the protein Sir4p but will also insert close to artificially introduced ectopic Sir4p binding sites. Furthermore, the integrase protein can be redesigned so as to interact with other DNA‐binding protein partners. Target selection by some natural site‐specific DNA transposons also involves protein–protein interactions (e.g., Tn7; Craig, 2002b), but so far there have been no attempts to retarget them to new insertion sites. Despite these advances, it seems likely that competing transposon insertion at random sites will continue to be a problem until strategies that restrict transposase activity to the chosen locus are devised.
XI. GENERAL CONSIDERATIONS IN APPLICATIONS OF SITE‐SPECIFIC RECOMBINASES AND TRANSPOSASES A. Recombinase and transgene “delivery” In current applications of site‐specific recombination, DNA encoding the recombinases(s) and cognate recombination site(s) is either transfected or is preintegrated in the genome of the organism under study. Targeting of natural genomic sequences would normally still require transfection of DNA encoding the modified recombinase(s) along with any foreign sequences to be integrated. Recombinase‐based gene therapies would therefore have to surmount the same DNA delivery problems as all other gene therapies (Pfeifer and Verma, 2001). Of course, the difficulties are considerably less if transfection can be carried out
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
17
in vitro or “ex vivo,” as would be the case for most nontherapeutic applications. Site‐specific recombination may be promoted by direct introduction of recombinase protein into cells (Baubonis and Sauer, 1993). It might even be practicable to introduce protein–DNA complexes in which the recombinase or transposase proteins required for targeted integration are already bound to the DNA (Goryshin et al., 2000). These ambitious approaches would avoid any undesirable side effects of prolonged production of recombinase from a transgenic expression construct (see Section D).
B. Chromatin Sequences that are inaccessible due to chromatin structure are likely to be poor targets for any recombinase‐based system. There is no obvious solution to this problem, although some enzymes may prove to be better than others in reaching protected sequences. Chromatin structure might also have effects on the sequence specificity of site‐specific recombinases or transposases (Portlock and Calos, 2003).
C. Reversibility Many site‐specific recombinases (e.g., Cre and Flp) promote recombination between two identical sites, so there are also two identical sites in the products. The reaction is therefore reversible (Kilby et al., 1993; Fig. 1.2B). The back reaction can be minimized if the recombinase is present only transiently (Baubonis and Sauer, 1993; Kilby et al., 1993). Some site sequences favor the forward reaction (Bouhassira et al., 1997; Hoess et al., 1986; Thomson et al., 2003), so it may be possible to choose pairs of sites for which this is the case. Certain recombinases, such as the phage integrases, recombine nonidentical sites, and their reactions can be essentially unidirectional (Groth and Calos, 2003). Similar problems might arise with DDE transposase‐mediated integration. The same transposase might promote excision of the integrated DNA. Such problems might be alleviated by strategies similar to those mentioned in Section A (e.g., transient availability of the transposase).
D. Nonspecific reactions All recombinases have the potential to cause “collateral damage” if they act at sequences other than the intended targets, and there is evidence that this happens in vivo. Cre can recombine at a number of pseudo‐loxP sites in the human and mouse genomes, which are quite divergent from the standard (“wild‐ type”) loxP sequence (Thyagarajan et al., 2000). Similar illegitimate recombination events, due to overexpression of Cre in transgenic mouse cells, can lead to
18
Akopian and Stark
growth inhibition, male sterility, cell‐cycle arrest, and DNA damage (Adams and van der Weyden, 2001; Loonstra et al., 2001; Schmidt et al., 2000). Such nonspecific activity would be very undesirable and potentially dangerous in gene therapy applications. The dangers of untargeted integration were tragically highlighted by two cases of leukemia in children participating in a trial for gene therapy of X‐linked severe combined immunodeficiency (X‐SCID). The therapeutic transgene was introduced ex vivo into cells that had been temporarily removed from the patients, by infection with a modified retrovirus whose integration was not targeted. Some of the retroviral integration events that ensued activated a nearby gene, resulting in oncogenesis (Kohn et al., 2003). Minimization of risk of analogous genetic damage would entail optimization of target site specificity, cell type‐specific recombinase gene delivery, cell type‐ specific promoters, and temporary recombinase expression. The tendency for retargeted recombinases to have relaxed specificity compared with their parent enzyme might restrict their applicability. Development of methods to select for restoration of high specificity is therefore most important. Of course, therapeutic modification of gene expression might be achieved in many circumstances by targeting but not altering DNA sequences (e.g., with DNA‐binding proteins or triplex‐forming oligonucleotides) (Uil et al., 2003).
XII. PROSPECTS AND CONCLUSIONS We still have a long way to go, but what will we be able to do when we have created efficient and highly specific “designer recombinases?” Perhaps the most exciting (but distant) prospects are in the field of gene therapy, where we can hope to bring about safe and efficient integration of transgenes at chosen genetic locations that support optimal expression. We might also knock out genes, or alter their expression levels, by integration of DNA sequences at suitable positions. Complete removal of a gene or other segment of DNA by recombinase‐ mediated excision will be very demanding technically because sites at both ends of the segment would have to be targeted by different versions of the recombinase. However, the task may be simpler in some special cases where an undesirable DNA segment is flanked by similar or identical sequences—the most obvious of these being the integrated proviral DNA of LTR retroviruses such as HIV. In the field of biotechnology, we can hope to improve and simplify procedures for making transgenic organisms for production of valuable proteins. Integration could be targeted to loci where gene expression will be high and only in the desired tissues (e.g., to a mammalian casein gene‐expression site, for production of proteins in milk).
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
19
The uses of designer recombinases will not be restricted to mammals; similar strategies could be applied in any organism. In view of the concerns about introduction of transgenic DNA into food plants and wild species, the possibility that genetic changes at specific sites might be brought about by injecting designer recombinase proteins into cells, without any foreign DNA at all, may be appealing. Finally, the uses of these systems in experimental genetics, as tools for the specific genetic modification of laboratory organisms, are limited only by the scientists’ imagination.
References Abdel‐Meguid, S. S., Grindley, N. D. F., Templeton, N. S., and Steitz, T. A. (1984). Cleavage of the site‐specific recombination protein resolvase: The smaller of two fragments binds DNA specifically. Proc. Natl. Acad. Sci. USA 81, 2001–2005. Ackroyd, A. J., Avila, P., Parker, C. N., and Halford, S. E. (1990). Site‐specific recombination by mutants of Tn21 resolvase with DNA recognition functions from Tn3 resolvase. J. Mol. Biol. 216, 633–643. Adams, D. J., and van der Weyden, L. (2001). Are we creating problems? Negative effects of Cre recombinase. Genesis 29, 115. Akopian, A., He, J., Boocock, M. R., and Stark, W. M. (2003). Chimeric recombinases with designed DNA sequence recognition. Proc. Natl. Acad. Sci. USA 100, 8688–8691. Arnold, P. H., Blake, D. G., Grindley, N. D. F., Boocock, M. R., and Stark, W. M. (1999). Mutants of Tn3 resolvase which do not require accessory binding sites for recombination activity. EMBO J. 18, 1407–1414. Avila, P., Ackroyd, A. J., and Halford, S. E. (1990). DNA binding by mutants of Tn21 resolvase with DNA recognition functions from Tn3 resolvase. J. Mol. Biol. 216, 645–655. Baubonis, W., and Sauer, B. (1993). Genomic targeting with purified Cre recombinase. Nucl. Acids Res. 21, 2025–2029. Beerli, R. R., and Barbas, C. F., III (2002). Engineering polydactyl zinc‐finger transcription factors. Nat. Biotechnol. 20, 135–141. Belfort, M., Berbyshire, V., Parker, M. M., Cousineau, B., and Lambowitz, A. M. (2002). Mobile introns: Pathways and proteins. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 761–783. ASM Press, Washington, DC. Bibikova, M., Beumer, K., Trautman, J. K., and Carroll, D. (2003). Enhancing gene targeting with designed zinc finger nucleases. Science 300, 764. Bibikova, M., Golic, M., Golic, K. G., and Carroll, D. (2002). Targeted chromosomal cleavage and mutagenesis in Drosophila using zinc‐finger nucleases. Genetics 161, 1169–1175. Boeke, J. D. (2002). Putting mobile DNA to work: The toolbox. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 24–37. ASM Press, Washington, DC. Bouhassira, E. E., Westerman, K., and Leboulch, P. (1997). Transcriptional behavior of LCR enhancer elements integrated at the same chromosomal locus by recombinase‐mediated cassette exchange. Blood 90, 3332–3344. Branda, C. S., and Dymecki, S. M. (2004). Talking about a revolution: The impact of site‐specific recombinases on genetic analyses in mice. Dev. Cell 6, 7–28. Buchholz, F., and Stewart, A. F. (2001). Alteration of Cre recombinase site specificity by substrate‐ linked protein evolution. Nat. Biotechnol. 19, 1047–1052.
20
Akopian and Stark
Burke, M. E., Arnold, P. H., He, J., Wenwieser, S. V. C. T., Rowland, S.‐J., Boocock, M. R., and Stark, W. M. (2004). Activating mutations of Tn3 resolvase marking interfaces important in recombination catalysis and its regulation. Mol. Microbiol. 51, 937–948. Bushman, F. D. (2003). Targeting survival: Integration site selection by retroviruses and non‐LTR retrotransposons. Cell 115, 135–138. Bushman, F. D., and Miller, M. D. (1997). Tethering human immunodeficiency virus type 1 preintegration complexes to target DNA promotes integration at nearby sites. J. Virol. 71, 458–464. Chandrasegaran, S., and Smith, J. (1999). Chimeric restriction enzymes: What is next? Biol. Chem. 380, 841–848. Chen, Y., Narendra, U., Iype, L. E., Cox, M. M., and Rice, P. A. (2000). Crystal structure of a Flp recombinase‐Holliday junction complex: Assembly of an active oligomer by helix swapping. Mol. Cell 6, 885–897. Collins, C. H., Yokobayashi, Y., Umeno, D., and Arnold, F. H. (2003). Engineering proteins that bind, move, make and break DNA. Curr. Opin. Biotechnol. 14, 371–378. Colosimo, A., Goncz, K. K., Holmes, A. R., Kunzelmann, K., Novelli, G., Malone, R. W., Bennett, M. J., and Gruenert, D. C. (2000). Transfer and expression of foreign genes in mammalian cells. Biotechniques 29, 314–324. Craig, N. L. (2002a). Mobile DNA: An introduction. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 3–11. ASM Press, Washington, DC. Craig, N. L. (2002b). Tn7. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 423–456. ASM Press, Washington, DC. Dorgai, L., Yagil, E., and Weisberg, R. A. (1995). Identifying determinants of recombination specificity: Construction and characterization of mutant bacteriophage integrases. J. Mol. Biol. 252, 178–188. Eickbush, T. H. (2002). R2 and related site‐specific non‐long terminal repeat retrotransposons. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 813–835. ASM Press, Washington, DC. Elrod‐Erickson, M., Rould, M. A., Nekludova, L., and Pabo, C. O. (1996). Zif268 protein‐DNA ˚ : A model system for understanding zinc finger‐DNA interactions. complex refined at 1.6 A Structure 4, 1171–1180. Epinat, J‐C., Arnould, S., Chames, P., Rochaix, P., Desfontaines, D., Puzin, C., Patin, A., Zanghellini, A., Paques, F., and Lacroix, E. (2003). A novel engineered meganuclease induces homologous recombination in yeast and mammalian cells. Nucl. Acids Res. 31, 2952–2962. Gorman, C., and Bullock, C. (2000). Site‐specific gene targeting for gene expression in eukaryotes. Curr. Opin. Biotechnol. 11, 455–460. Goryshin, I. Y., Jendrisak, J., Hoffman, L. M., Meis, R., and Reznikoff, W. S. (2000). Insertional transposon mutagenesis by electroporation of released Tn5 transposition complexes. Nat. Biotechnol. 18, 97–100. Goulaouic, H., and Chow, S. A. (1996). Directed integration of viral DNA mediated by fusion proteins consisting of human immunodeficiency virus type 1 integrase and Escherichia coli LexA protein. J. Virol. 70, 37–46. Grindley, N. D. F. (1993). Analysis of a nucleoprotein complex: The synaptosome of resolvase. Science 262, 738–740. Grindley, N. D. F. (2002). The movement of Tn3‐like elements: Transposition and cointegrate resolution. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 272–302. ASM Press, Washington, DC. Groth, A. C., and Calos, M. P. (2003). Phage integrases: Biology and applications. J. Mol. Biol. 335, 667–678. Guo, F., Gopaul, D. N., and Van Duyne, G. D. (1999). Asymmetric DNA bending in the Cre‐loxP site‐specific recombination synapse. Proc. Natl. Acad. Sci. USA 96, 7143–7148.
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
21
Guo, H., Karberg, M., Long, M., Jones, J. P., IIIrd, Sullenger, B., and Lambowitz, A. M. (2000). Group II introns designed to insert into therapeutically relevant DNA target sites in human cells. Science 289, 452–457. Hallet, B., and Sherratt, D. J. (1997). Transposition and site‐specific recombination: Adapting DNA cut‐and‐paste mechanisms to a variety of genetic rearrangements. FEMS Microbiol. Rev. 21, 157–178. Hartung, M., and Kisters‐Woike, B. (1998). Cre mutants with altered DNA binding properties. J. Biol. Chem. 273, 22884–22891. Hatfull, G. F., Salvo, J. J., Falvey, E. E., Rimphanitchayakit, V., and Grindley, N. D. F. (1988). Site‐ specific recombination by resolvase. Symp. Soc. Gen. Microbiol. 43, 149–181. Hoess, R. H., Wierzbicki, A., and Abremski, K. (1986). The role of the loxP spacer region in P1 site‐ specific recombination. Nucl. Acids Res. 14, 2287–2300. Holliday, R. (1964). A mechanism for gene conversion in fungi. Genet. Res. 5, 282–304. Hurt, J. A., Thibodeau, S. A., Hirsh, A. S., Pabo, C. O., and Joung, J. K. (2003). Highly specific zinc finger proteins obtained by directed domain shuffling and cell‐based selection. Proc. Natl. Acad. Sci. USA 100, 12271–12276. Jayaram, M., Grainge, I., and Tribble, G. (2002). Site‐specific recombination by the Flp protein of Saccharomyces cerevisiae. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 192–218. ASM Press, Washington, DC. Johnson, R. C. (2002). Bacterial site‐specific DNA inversion systems. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 230–271. ASM Press, Washington, DC. Katz, R. A., Merkel, G., and Skalka, A. M. (1996). Targeting of retroviral integrase by fusion to a heterologous DNA‐binding domain: In vitro activities and incorporation of a fusion protein into viral particles. Virology 217, 178–190. Kilby, N. J., Snaith, M. R., and Murray, J. A. (1993). Site‐specific recombinases: Tools for genome engineering. Trends Genet. 9, 413–421. Kohn, D. B., Sadelain, M., and Glorioso, J. C. (2003). Occurrence of leukaemia following gene therapy of X‐linked SCID. Nat. Rev. Cancer 3, 477–488. Kolb, A. F. (2002). Genome engineering using site‐specific recombinases. Cloning Stem Cells 4, 65–80. Lacy, E., Roberts, S., Evans, E. P., Burtenshaw, M. D., and Costantini, F. D. (1983). A foreign beta‐ globin gene in transgenic mice: Integration at abnormal chromosomal positions and expression in inappropriate tissues. Cell 34, 343–358. Lambowitz, A. M., and Zimmerly, S. (2004). Mobile group II introns. Annu. Rev. Genet. 38, 1–35. Loonstra, A., Vooijs, M., Beverloo, H. B., Allak, B. A., van Drunen, E., Kanaar, R., Berns, A., and Jonkers, J. (2001). Growth inhibition and DNA damage induced by Cre recombinase in mammalian cells. Proc. Natl. Acad. Sci. USA 98, 9209–9214. Monahan, P. E., and Samulski, R. J. (2000). AAV vectors: Is clinical success on the horizon? Gene Ther. 7, 24–30. Nagy, A. (2000). Cre recombinase: The universal reagent for genome tailoring. Genesis 26, 99–109. Nash, H. A. (1996). Site‐specific recombination: Integration, excision, resolution, and inversion of defined DNA segments. In “Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology” (F. C. Neidhart, R. Curtis, J. L. Ingraham, E. C. C. Lin, K. B. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger, eds.), 2nd edn. pp. 2363–2376. American Society for Microbiology, Washington, DC. Pabo, C. O., Peisach, E., and Grant, R. A. (2001). Design and selection of novel Cys2His2 zinc finger proteins. Ann. Rev. Biochem. 70, 313–340. Palmiter, R. D., and Brinster, R. L. (1986). Germ‐line transformation of mice. Ann. Rev. Genet. 20, 465–499.
22
Akopian and Stark
Pannell, D., and Ellis, J. (2001). Silencing of gene expression: Implications for design of retrovirus vectors. Rev. Med. Virol. 11, 205–217. Pavletich, N. P., and Pabo, C. O. (1991). Zinc finger‐DNA recognition: Crystal structure of a ˚ . Science 252, 809–817. Zif268‐DNA complex at 2.1 A Peng, W.‐J., Chang, C.‐M., and Lin, T.‐H. (2002). Target integration by a chimeric Sp1 zinc finger domain‐Moloney murine leukemia virus integrase in vivo. J. Biomed. Sci. 9, 171–184. Pfeifer, A., and Verma, I. M. (2001). Gene therapy: Promises and problems. Ann. Rev. Genomics Hum. Genet. 2, 177–211. Porteus, M. H., and Baltimore, D. (2003). Chimeric nucleases stimulate gene targeting in human cells. Science 300, 763. Portlock, J. L., and Calos, M. P. (2003). Site‐specific genomic strategies for gene therapy. Curr. Opin. Mol. Therapeut. 5, 376–382. Reynolds, L., Ullman, C., Moore, M., Isalan, M., West, M. J., Clapham, P., Klug, A., and Choo, Y. (2003). Repression of the HIV‐1 50 LTR promoter and inhibition of HIV‐1 replication by using engineered zinc‐finger transcription factors. Proc. Natl. Acad. Sci. USA 100, 1615–1620. Sandmeyer, S. (2003). Integration by design. Proc. Natl. Acad. Sci. USA 100, 5586–5588. Santoro, S. W., and Schultz, P. G. (2002). Directed evolution of the site specificity of Cre recombinase. Proc. Natl. Acad. Sci. USA 99, 4185–4190. Sauer, B. (2002). Chromosome manipulation by Cre‐lox recombination. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 38–58. ASM Press, Washington, DC. Schmidt, E. E., Taylor, D. S., Prigge, J. R., Barnett, S., and Capecchi, M. R. (2000). Illegitimate Cre‐ dependent chromosome rearrangements in transgenic mouse spermatids. Proc. Natl. Acad. Sci. USA 97, 13702–13707. Schneider, F., Schwikardi, M., Muskhelishvili, G., and Droge, P. (2000). A DNA‐binding domain swap converts the invertase Gin into a resolvase. J. Mol. Biol. 295, 767–775. Sclimenti, C. R., Thyagarajan, B., and Calos, M. P. (2001). Directed evolution of a recombinase for improved genomic integration at a native human sequence. Nucl. Acids Res. 29, 5044–5051. Shaikh, A. C., and Sadowski, P. D. (2000). Chimeras of the Flp and Cre recombinases: Tests of the mode of cleavage by Flp and Cre. J. Mol. Biol. 302, 27–48. Smith, K. (2001). Theoretical mechanisms in targeted and random integration of transgene DNA. Reprod. Nutr. Dev. 41, 465–485. Stark, W. M., Grindley, N. D. F. F., Hatfull, G. F., and Boocock, M. R. (1991). Resolvase‐catalysed reactions between res sites differing in the central dinucleotide of subsite I. EMBO J. 10, 3541–3548. Stemmer, W. P. C. (1994). DNA shuffling by random fragmentation and reassembly: In vitro recombination for molecular evolution. Proc. Natl. Acad. Sci. USA 91, 10747–10751. Tan, W., Zhu, K., Segal, D. J., Barbas, C. F., III, and Chow, S. A. (2004). Fusion proteins consisting of human immunodeficiency virus type I integrase and the designed polydactyl zinc finger protein E2C: Direct integration of viral DNA into specific sites. J. Virol. 78, 1301–1313. Thomson, J. G., Rucker, E. B., III, and Piedrahita, J. A. (2003). Mutational analysis of loxP sites for efficient Cre‐mediated insertion into genomic DNA. Genesis 36, 162–167. Thyagarajan, B., Guimaraes, M. J., Groth, A. C., and Calos, M. P. (2000). Mammalian genomes contain active recombinase recognition sites. Gene 244, 47–54. Thyagarajan, B., Olivares, E. C., Hollis, R. P., Ginsburg, D. S., and Calos, M. P. (2001). Site‐specific genomic integration in mammalian cells mediated by phage phiC31 integrase. Mol. Cell Biol. 21, 3926–3934. Uil, T. G., Haisma, H. J., and Rots, M. G. (2003). Therapeutic modulation of endogenous gene function by agents with designed DNA‐sequence specificities. Nucl. Acids Res. 31, 6064–6078.
1. Site-Specific DNA Recombinases as Instruments for Genomic Surgery
23
Van Duyne, G. D. (2002). A structural view of tyrosine recombinase site‐specific recombination. In “Mobile DNA II” (N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds.), pp. 93–117. ASM Press, Washington, DC. Vasquez, K. M., Marburger, K., Intody, Z., and Wilson, J. H. (2001). Manipulating the mammalian genome by homologous recombination. Proc. Natl. Acad. Sci. USA 98, 8403–8410. Voziyanov, Y., Stewart, A. F., and Jayaram, M. (2002). A dual reporter screening system identifies the amino acid at position 82 in Flp site‐specific recombinase as a determinant for target specificity. Nucl. Acids Res. 30, 1656–1663. Voziyanov, Y., Konieczka, J. H., Stewart, A. F., and Jayaram, M. (2003). Stepwise manipulation of DNA specificity in Flp recombinase: Progressively adapting Flp to individual and combinatorial mutations in its target site. J. Mol. Biol. 326, 65–76. Yagil, E., Dorgai, L., and Weisberg, R. A. (1995). Identifying determinants of recombination specificity: Construction and characterization of chimeric bacteriophage integrases. J. Mol. Biol. 252, 163–177. Yanez, R. J., and Porter, A. C. G. (1998). Therapeutic gene targeting. Gene Ther. 5, 149–159. Yang, W., and Steitz, T. A. (1995). Crystal structure of the site‐specific recombinase gamma delta resolvase complexed with a 34 bp cleavage site. Cell 82, 193–207. Zhu, Y., Dai, J., Fuerst, P. G., and Voytas, D. F. (2003). Controlling integration specificity of a yeast retrotransposon. Proc. Natl. Acad. Sci. USA 100, 5891–5895.
2
Human Clinical Trials of Plasmid DNA Vaccines Margaret A. Liu* and Jeffrey B. Ulmer{ *Transgene S.A., 11, rue de Molsheim, 67082 Strasbourg Cedex, France { Chiron Corporation, Emeryville, California 94608
I. Background A. Historical perspective on DNA vaccines B. Mechanism of action II. Development of DNA Vaccines for Clinical Trials A. Manufacturing of plasmid DNA B. Qualification of plasmid DNA product C. Safety testing of DNA vaccines III. Human Clinical Trials A. Vaccines for infectious diseases B. Vaccines for cancer IV. Limitations of DNA Vaccine Potency V. Prospects and Conclusion References
ABSTRACT This article gives an overview of DNA vaccines with specific emphasis on the development of DNA vaccines for clinical trials and an overview of those trials. It describes the preclinical research that demonstrated the efficacy of DNA vaccines as well as an explication of the immunologic mechanisms of action. These include the induction of cognate immune responses, such as the generation of cytolytic T lymphocytes (CTL) as well as the effect of the plasmid DNA
Advances in Genetics, Vol. 55 Copyright 2005, Elsevier Inc. All rights reserved.
0065-2660/05 $35.00 DOI: 10.1016/S0065-2660(05)55002-8
26
Liu and Ulmer
upon the innate immune system. Specific issues related to the development of DNA as a product candidate are then discussed, including the manufacture of plasmid, the qualification of the plasmid DNA product, and the safety testing necessary for initiating clinical trials. Various human clinical trials for infectious diseases and cancer have been initiated or completed, and an overview of these trials is given. Finally, because the early clinical trials have shown less than optimal immunogenicity, methods to increase the potency of the vaccines are described. ß 2005, Elsevier Inc.
I. BACKGROUND A. Historical perspective on DNA vaccines It is now well established that plasmid DNA encoding a foreign protein can be expressed in situ, induce immune responses directed against the expressed protein, and elicit protective immune responses against the pathogen from which the protein was derived (Donnelly et al., 1997; Gurunathan et al., 2000). However, the initial reports on DNA vaccines were met with skepticism followed by considerable enthusiasm, as the technology proved to be quite facile and robust in a variety of laboratories and in vivo systems. Because viruses have evolved quite complex and specific mechanisms to deliver their genetic content to cells and because DNA is easily degraded in vivo, it was thought unlikely that DNA or RNA alone would be capable of transfecting cells in vivo. Moreover, because of the complexity of priming cellular immune responses (specifically processing and presentation of antigens leading to major histocompatibility complex [MHC] class I‐restricted CTL), simple administration of plasmid DNA was thought unlikely to be capable of generating the desired CTL responses. Nevertheless, DNA vaccines were shown to be capable of inducing both humoral and cellular immunity (i.e., both antibodies and CTL, along with T helper cells needed for the generation of both of the former responses) (Davis et al., 1993; Robinson et al., 1993; Tang et al., 1992; Ulmer et al., 1993; Wang et al., 1993; Xiang et al., 1994). The CTL responses were capable of providing immunity, which was protective against challenge of animals with a strain of virus that was different from the strain from which the gene was taken (Ulmer et al., 1993). These observations opened the door for a number of studies in different disease models in which DNA vaccines were able to induce immune responses and protective or therapeutic effects. Another development was the use of a so‐called ‘gene gun’ to propel microscopic gold beads coated with plasmid DNA into the skin, where the encoded protein was likewise produced with the subsequent generation of antibody and T cell responses against the protein (Fynan et al., 1993; Tang et al., 1992).
2. Human Clinical Trials of Plasmid DNA Vaccines
27
The DNA vaccines used were simply bacterial plasmids, which utilized a promoter element active in mammalian cells rather than bacteria. Other common elements included a transcriptional terminator and a selectable marker that enabled the production of the plasmids in transformed bacteria. Thus, the concept of recombinant protein vaccines was taken one step earlier, by focusing on the production of the bacterial plasmid rather than on the encoded protein. One advantage of focusing on the plasmid as the product, is the generic nature of producing plasmids (i.e., the process for manufacturing the plasmid is relatively independent of the gene). The general approaches to produce plasmids for clinical usage are discussed in later section, but it is clear that having a generic manufacturing process irrespective of the vaccine product would greatly facilitate the development of many new vaccines. This is in contrast to the challenges of producing a live attenuated virus or expressing then purifying a recombinant protein; each of which would present formidable obstacles that must be otherwise uniquely overcome for each new product. Vaccines are considered to be the most successful medical intervention in the history of mankind. However, limitations with traditional vaccine approaches led scientists in search of new technologies for targets such as HIV, malaria, tuberculosis, and cancer. One key hurdle was to develop means of inducing robust cellular immunity, specifically CTL. These immune cells play an important role in the containment and clearance of tumors and cells infected with viruses or intracellular bacteria. Cytolytic T lymphocytes may be particularly important for protection against pathogens that induce chronic infections, such as HIV. The mechanism for the induction of CTL requires that the antigen be synthesized within cells of the vaccinated individual; delivery of an exogenous protein, such as the case with inactivated virus and recombinant protein vaccines, usually does not induce CTL responses. Live viruses effectively induce CTL, but the use of attenuated live viruses for certain pathogens, such as HIV, is considered too risky because of the possibility of reversion of the vaccine strain to the virulent and wild‐type strain. Thus, a technology, such as DNA vaccines, which could effectively induce protective CTLs and provide in vivo cross‐strain protection, in addition to inducing antibodies, provided a new approach for a variety of disease targets.
B. Mechanism of action The use of DNA vaccines has helped to elucidate two important aspects of the mechanism of action of immune priming (e.g., presentation of antigens by MHC class I molecules for priming CTL responses and the role of innate immunity in the generation of cognate immune responses against the antigen encoded by the DNA vaccine). First, as noted in earlier section, it is necessary for an antigen to be endogenously produced in a cell, preferably a professional antigen‐presenting
28
Liu and Ulmer
cell (APC), in order to make the protein enter the MHC Class I processing pathway for presentation to, and priming of, naı¨ve CD8þ lymphocytes. If the antigen is delivered exogenously, or produced in a cell that lacks certain features (such as co‐stimulatory molecules found on professional APCs), CTL responses are not effectively generated. Since muscle cells are not APCs, the generation of CTL responses after intramuscular (IM) injection of DNA vaccines was surprising. The explanation may lay in observations by several groups that proteins encoded by DNA vaccines could be transferred from their site of synthesis (i.e., myocytes or keratinocytes) to professional APCs for presentation by MHC molecules (Cho et al., 2001; Corr et al., 1996, 1999; Fu et al., 1997; Kumaraguru et al., 2000; Ulmer et al., 1996). This phenomenon of cross priming may also account for induction of CTL against antigens on tumor cells and after infection of nonAPCs by viruses (Huang et al., 1994; Le Bon et al., 2003). Second, a key factor in the immunogenicity of DNA vaccines proved to be the bacterial DNA of the plasmid itself. Initially, the plasmid was thought to be simply the blueprint for the encoded protein and was considered immunologically inert. However, because DNA plasmids are derived from bacteria they stimulate the innate immune system by interacting with Toll‐like receptor 9 (TLR9) (Hemmi et al., 2000) and the resultant nonspecific immunity can augment the antigen‐specific immune response directed against the encoded protein (Klinman et al., 1997; Sato et al., 1996).
II. DEVELOPMENT OF DNA VACCINES FOR CLINICAL TRIALS The early successes of DNA vaccines in animal models provided the impetus to test this vaccine technology in human clinical trials. However, since this type of vaccine had never before been administered to humans, several formidable hurdles needed to be overcome to produce them, as well as to assure their safety and quality: (1) bacterial plasmids needed to be manufactured at scales never before attempted, (2) assays were required to measure the quality of this novel vaccine, and (3) certain safety evaluations not normally tested for conventional vaccines were needed (e.g., integration into host chromosomal DNA). This section will summarize the key development activities leading up to human testing of DNA vaccines.
A. Manufacturing of plasmid DNA Most plasmid DNA vaccines used to date contain the following basic elements: (1) an origin of replication for efficient propagation in Escherichia coli, (2) an antibiotic resistance gene for growth selection, (3) a strong promoter to drive expression in eukaryotic cells, (4) a polyadenylation termination sequence, and
2. Human Clinical Trials of Plasmid DNA Vaccines
29
(5) a gene insert coding for the antigen of interest. In addition, immunostimulatory CpG motifs normally present in bacterial DNA contribute to vaccine potency via stimulating the innate immune system, as described in an earlier section. Two other types of plasmid vectors include those that do not contain antibiotic resistance genes, such as those utilizing repressor titration (Williams et al., 1998), and those encoding an alphavirus RNA replicon rather than a discrete gene (Berglund et al., 1998; Hariharan et al., 1998; Leitner et al., 2003). Both of these types of vectors have entered human clinical trials. For manufacturing purposes, it is desirable to have a system, which stably produces plasmids at high copy number. These characteristics facilitate the downstream processing steps and purification of the plasmid. The ColE1 origin of replication present in pUC‐based plasmids is commonly used, since yields of 30 mg of plasmid per liter of culture can be achieved. Other factors contributing to high plasmid yield include the host E. coli strain, growth medium and conditions of fermentation. For example, E. coli strains that produce relatively low levels of nucleases can result in greater plasmid stability during processing. The manufacture of plasmid DNA has a potentially distinct advantage over production of other types of vaccines. Unlike other processes, such as those for recombinant proteins where specific procedures and conditions are required for each product, all plasmid DNA vaccines are physically similar to one another. Hence, methods for their production and purification are essentially generic. The majority of plasmids produced during fermentation of E. coli have a supercoiled topology (i.e., a coiled coil). Other minor forms are produced as a consequence of nicking one of the chains resulting in relaxed circles, nicking of both chains to produce linear plasmids, and denaturation. It is generally believed that supercoiled plasmids are the most active of these forms, based on transcription in vivo (Kano et al., 1981; Sekiguchi and Kmiec, 1989). However, both relaxed circle and linear plasmids are immunogenic as DNA vaccines in animals (unpublished observations). Stability studies in vitro have shown that supercoiled plasmids are sequentially converted to relaxed circles, linear plasmids, fragments, and oligomers (Evans et al., 2000). Therefore, it is desirable to have supercoiled plasmids to increase the shelf life of the product. After fermentation, plasmid DNA at best accounts for only 1% of the total cell mass. Thus, procedures are required for the release of plasmids from the cells and their separation from other cellular constituents. In the laboratory, this is effectively accomplished using alkaline lysis of the cells, precipitation of nonplasmid material by SDS, and sedimentation of the precipitate to yield an aqueous preparation of plasmid. A high degree of purity can be achieved by a subsequent CsCl gradient centrifugation step. Some of these steps are neither practical nor acceptable for use at large scale for preparation of human clinical materials. For example, the CsCl gradient centrifugation step was eliminated
30
Liu and Ulmer
due to the impracticalities of centrifugation and the toxicity of the materials used. Instead, anionic exchange chromatography is often utilized to take advantage of the negative charge of the plasmid DNA. Prazeres et al. (2001) reviewed these steps involved in plasmid DNA purification. The demand for plasmid DNA has resulted in a fledgling industry for the contract manufacture of both research‐grade material for animal studies and cGMP‐grade product for human clinical trials.
B. Qualification of plasmid DNA product Purified plasmid DNA products should conform to the following specifications (Robertson and Griffiths, 2001; Smith and Klinman, 2001). The product should appear clear and colorless by visual inspection. The identity of the plasmid can be confirmed by nucleotide sequence, PCR for plasmid‐specific sequences and/or restriction enzyme digest patterns by gel electrophoresis. As mentioned in an earlier section, the form of the plasmid should be predominantly circular (preferably supercoiled), as judged by gel electrophoresis. Potency of the product can be quantified by expression of the transgene in transient transfection of cells in vitro and/or by induction of immune responses against the transgene in an appropriate animal model. Plasmid DNA concentration is measured by absorbance at 260 nm. Purity of the product is assessed by the presence of residual proteins, chromosomal DNA, RNA, and endotoxin. The levels of these impurities should fall below specified limits. Finally, sterility is measured by standard microbiological assays.
C. Safety testing of DNA vaccines Standard preclinical evaluation of vaccine safety includes systemic and local reactogenicity, histopathology, and toxicity in appropriate animal models. Due to the unique nature of plasmid DNA vaccines, the following additional safety tests are performed. First, although hypothetical at this point, integration of DNA plasmids into host chromosomal DNA is possible. Therefore, quantitative assays to assess such integration events have been developed. These are based on physical separation of high molecular weight chromosomal DNA from plasmids by gel electrophoresis, followed by PCR for plasmid‐specific sequences. This approach can detect a single integration event in 150,000 cellular nuclei (Martin et al., 1999; Nichols et al., 1995). So far, little or no integration has been observed in animal models (Kang et al., 2003; Ledwith et al., 2000b; Manam et al., 2000). Second, because certain types of anti‐DNA antibodies can be pathogenic and cause autoimmune disorders (e.g., systemic lupus erythematosus), the potential for induction of such antibodies by plasmid DNA vaccines must be evaluated. This possibility, though, is also hypothetical, as
2. Human Clinical Trials of Plasmid DNA Vaccines
31
humans are frequently exposed to bacterial DNA without apparent consequence. So far, there have been no reports of induction of pathogenic anti‐ DNA antibodies by DNA vaccines (Parker et al., 1999, 2001). Third, the biodistribution and persistence of plasmid DNA vaccines are evaluated in animal models. Unlike vaccines based on proteins or inactivated organisms, DNA vaccines result in the in situ production of antigen. Therefore, it is important to know the specific cell types that may produce the antigen and how long this production may persist. This is accomplished by PCR analyses of various tissues over time for the presence of plasmid‐specific sequences. After IM injection, typical plasmid DNA vaccines have been shown to be present in several tissues for a period of days but are quickly cleared (Hanke et al., 2002; Parker et al., 1999). Thereafter, plasmids persist in the injected muscle for many weeks, but slowly disappear over time. These results are consistent with a lack of integration and loss of plasmid by nuclease digestion and cellular turnover. Based on this type of safety profile, many DNA vaccines have entered human clinical trials.
III. HUMAN CLINICAL TRIALS Plasmid DNA has been evaluated in a variety of human clinical trials for potential use as prophylactic vaccines, as immunotherapy for cancer, as immunomodulators for asthma/allergy, and as gene therapy for chronic diseases. As with any new vaccine, the initial phase I trials are designed to test the safety of the candidates, particularly because this technology had been used never before for direct human administration. An overview of the types of clinical trials will be given here to highlight the issues for usage and efficacy of plasmid DNA in humans.
A. Vaccines for infectious diseases The ability of DNA vaccines to induce cytolytic T cell responses and the potential for the technology to be amenable for manufacture and distribution on a global scale and in developing countries, led to the rapid clinical testing of DNA vaccines for infectious diseases such as HIV, malaria, and hepatitis B. The earliest clinical trials were with HIV DNA vaccines conducted in individuals already infected with HIV, followed by extension to volunteers uninfected with HIV. These studies, as well as later studies for other indications, demonstrated that DNA vaccines are well tolerated and safe. There has been no evidence of integration of DNA into host chromosomes, autoimmunity, or immunologic tolerance (Comerota et al., 2002; Conry et al., 2002; Epstein et al., 2002; Klencke et al., 2002; MacGregor et al., 2000; Swain et al., 2000; Tacket et al., 1999; Weber et al., 2001).
32
Liu and Ulmer
While the overall clinical immune responses have been less potent than expected based upon the preclinical results in animal models, there have been notable exceptions, and more trials utilizing higher DNA doses, alternative protocols, and novel formulations have resulted in greater immunogenicity. Different trials have resulted in different results in terms of the types of immune responses generated, perhaps as a reflection of the different DNA constructs, immunization protocols, and patient populations. For example, several studies demonstrated the generation of CTL responses following DNA immunization (Calarota et al., 1998, 2001; Roy et al., 2000). In another study, although some HIV seronegative patients immunized with DNA encoding env/rev were able to generate CD4 T helper responses including the production of IFN‐gamma in an ELISPOT assay (MacGregor et al., 2002), no CD8 responses were observed. Interestingly, in HIV infected patients, a DNA vaccine was able to induce antibody and CTL responses against certain viral antigens against which no detectable antibodies or CTL had existed prior to vaccination despite a high preexisting viral load (Calarota et al., 1998, 1999). A slightly different approach utilizing a DNA vaccine encoding consensus HIV clade A Gag p24/p17 proteins fused to a linear sequence of CTL epitopes tested alone and in conjunction with a viral vector (modified vaccinia Ankara, MVA) in a DNA prime‐vector boost approach was shown capable of generating responses directed against many of the CTL epitopes, as measured by IFN‐gamma ELISPOT (Mwau et al., 2004). These HIV trials have demonstrated that immune responses (primarily T cells) can be induced or boosted in humans by DNA vaccines. Although most of these trials were not designed to address the effectiveness of vaccination against HIV disease, it is very likely that significant enhancements will be required to elicit protective immune responses by DNA vaccines. DNA vaccines encoding malarial antigens have been shown capable of generating CD8þ CTL, even against epitopes restricted by more than one HLA haplotype in the same person (Wang et al., 1998). A follow up study confirmed the generation of CTL by a malaria DNA vaccine (Le et al., 2000; Wang et al., 2001). However, no antigen‐specific antibody responses were detected in either trial. The strategy of employing a heterologous gene delivery system in combination with DNA vaccines is in clinical evaluation and has been found promising. For example, although DNA and MVA vectors by themselves are both capable of generating IFN‐gamma antigen‐specific immune responses, the responses are substantially greater if the DNA is given first, followed by MVA (McConkey et al., 2003; Moorthy et al., 2003). Another clinical trial utilizing the prime‐boost approach has demonstrated the capability of a recombinant protein to act as a booster vaccine for DNA (Epstein et al., 2004; Wang et al., 2004). Clinical trials of malaria vaccines offer the possibility of measuring protective efficacy by challenge with live parasites. Preliminary results with the DNA prime‐MVA boost approach have shown protection, as seen by delayed parasitemia following sporozoite challenge with a different strain of malaria (McConkey et al., 2003).
2. Human Clinical Trials of Plasmid DNA Vaccines
33
Studies utilizing a hepatitis B DNA vaccine highlight an alternative delivery technology, employing a device to propel gold beads coated with DNA into the epidermis of vaccinated individuals. Using this approach, a DNA vaccine was shown capable of inducing both antibodies and cellular responses in hepatitis‐ naı¨ve individuals (Roy et al., 2000). Although the magnitude of the antibodies induced by this DNA vaccine was not as high as that elicited by the licensed recombinant protein vaccine, a potential advantage of the DNA vaccine is the ability to induce T cell responses. Both CD4þ and CD8þ cells were elicited by the DNA vaccine, with the helper cells being of the Th1 phenotype (i.e., secreting IFN‐gamma). In a more challenging clinical situation, patients who had either not responded to a recombinant protein vaccine or whose response had not persisted, were immunized with plasmid DNA‐coated gold beads. The DNA was capable of inducing an antibody response in 12 of these individuals despite their previous failure to respond adequately to the licensed vaccine (Rottinghaus et al., 2003). Taken together these data hold promise for potential use of a hepatitis B DNA vaccine in nonresponder individuals and for immunotherapy to clear infection.
B. Vaccines for cancer Plasmid DNA has likewise been clinically evaluated as a therapeutic vaccine approach for cancer. These DNA vaccines have encoded tumor antigens, such as viral epitopes from transforming viruses (Klencke et al., 2002), self‐antigens that are expressed on tumors (Mincheff et al., 2000; Rosenberg et al., 2003; Tagawa et al., 2003), and tumor‐specific antigens (Timmerman et al., 2002). While most trials of prophylactic DNA vaccines are injected i.m. or intraepidermally (via particle bombardment), the cancer immunotherapeutic vaccines have also been delivered intranodally (Tagawa et al., 2003) and intradermally (Mincheff et al., 2000) as well. One study utilized a formulation of the DNA vaccine in an effort to facilitate the delivery of the DNA to APCs after IM injection (Klencke et al., 2002). Most of these cancer DNA vaccines have shown limited effectiveness as an immunotherapeutic intervention, possibly due to a relatively immunocompromised status of the cancer patients. These clinical trials of DNA vaccines for infectious diseases and cancer have proven the principle that immune responses can be generated in humans, but highlight the need for increased potency if this vaccine technology is to be fully effective.
IV. LIMITATIONS OF DNA VACCINE POTENCY Effective vaccines typically consist of three key components. First, the antigen is required to elicit a specific memory immune response, so that upon subsequent exposure to the pathogen a rapid and specific immune response is mounted to
34
Liu and Ulmer
prevent and/or clear infection. For DNA vaccines, the antigen is expressed by the plasmid in cells in situ. Based on reporter gene expression, only very small amounts of protein are produced (Wolff et al., 1990). Thus, it is important that the DNA vaccine vectors used are optimized for protein expression. Elements of the plasmid that contribute to the level of protein expression include the promoter, enhancer, and polyadenylation terminator. In addition, the nucleotide sequence and codon usage of the gene insert can play significant roles in transcription and translation (Haas et al., 1996; Qiu et al., 1999; zur Megede et al., 2000). Second, an immune stimulus or adjuvant is usually needed to maximize the immunogenicity of the vaccine. Insight into the nature of innate immune signaling has revealed that adjuvants generally interact with receptors (e.g., TLRs) present on immune cells. This interaction results in rapid activation of cells leading to cytokine production and an increase in antigen presentation capacity. Thus, when this occurs in the presence of an antigen, the result is an enhancement of antigen‐specific immune responses. DNA vaccines appear to have a built‐in adjuvant in the form of immunostimulatory CpG motifs, which are known to utilize TLR9 (Hemmi et al., 2000). The presence of these motifs is thought to play a role in the potency of DNA vaccines (Klinman et al., 1997; Sato et al., 1996), suggesting that the correct number and placement of CpG within the plasmid are important. Third, a vaccine delivery system can protect the vaccine from degradation, facilitate cellular uptake, target specific cells or tissues, and ensure that the antigen and adjuvant are delivered together. Conventional vaccines based on live attenuated organisms have all of these three elements contained within the vaccine. However, highly purified subunit vaccines, such as recombinant proteins and naked DNA vaccines, generally lack adjuvants and delivery systems and require that they be added back to the vaccine. Given that naked DNA vaccines have no inherent means of efficiently transfecting cells, delivery systems may offer the greatest opportunity for improvement. Intramuscular injection of naked plasmid DNA results in the transfection of very few cells in situ (Dupuis et al., 2000; Wolff et al., 1990). Based on PCR analysis, it was estimated that only a few thousand copies of plasmid are functionally retained within cells of the injected muscle (Ledwith et al., 2000a). Therefore, approximately only one in ten million plasmids are effectively delivered into the nuclei of cells. Limitations on the cellular delivery of plasmid DNA appear to be manifested in several areas, including distribution of the vaccine solution within the injected tissue, cellular uptake, and intracellular delivery to the nucleus (Dupuis et al., 2000). Transfection of APCs was found to be particularly inefficient after i.m. injection. These observations suggest two basic strategies to enhance the delivery, hence potency, of DNA vaccines. First, increased distribution of the DNA vaccine within the injected tissue should allow a greater number of cells to be available for transfection. Technologies to
2. Human Clinical Trials of Plasmid DNA Vaccines
35
physically deliver DNA plasmids to tissues, such as the gene gun, needle‐free devices, electroporation, and hydrostatic pressure, result in broader expression and enhanced vaccine potency (Wells, 2004). Second, expression of DNA vaccine antigens by APCs is known to be an effective means to induce immune responses (Manickan et al., 1997; Timares et al., 1998). Hence, targeting plasmid DNA specifically to APCs should increase vaccine potency. The effectiveness of particle‐based formulations of plasmid DNA (Little et al., 2004; Singh et al., 2000), which facilitate DNA uptake by and antigen expression within phagocytic APCs (Denis‐Mize et al., 2000, 2003), is consistent with this hypothesis. These delivery systems and formulations have in many cases shown marked increase in DNA vaccine potency in nonhuman primates (O’Hagan et al., 2001; Otten et al., 2004) and some of these technologies have entered human clinical trials (Klencke et al., 2002).
V. PROSPECTS AND CONCLUSION DNA vaccines have been widely utilized in immunological studies and as a reagent for a variety of laboratory studies. Early clinical studies have demonstrated that the most likely strategies just now entering the clinic, such as novel delivery systems or combinations with other gene‐delivery vectors will possibly, be the means to boost the potency of the DNA vaccines. Despite the added complications of such second generation approaches, the inherent capabilities and characteristics of DNA vaccines (i.e., the ability to induce CTL and Th1 type helper responses, the potential advantages of their manufacture, and the platform nature of the technology) continue to make this a technology with significant appeal.
References Berglund, P., Smerdou, C., Fleeton, M. N., Tubulekas, I., and Liljestrom, P. (1998). Enhancing immune responses using suicidal DNA vaccines. Nat. Biotechnol. 16, 562–565. Calarota, S., Bratt, G., Nordlund, S., Hinkula, J., Leandersson, A. C., Sandstrom, E., and Wahren, B. (1998). Cellular cytotoxic response induced by DNA vaccination in HIV‐1‐infected patients. Lancet 351(9112), 1320–1325. Calarota, S. A., Kjerrstrom, A., Islam, K. B., and Wahren, B. (2001). Gene combination raises broad human immunodeficiency virus‐specific cytotoxicity. Hum. Gene. Ther. 12(13), 1623–1637. Calarota, S. A., Leandersson, A. C., Bratt, G., Hinkula, J., Klinman, D. M., Weinhold, K. J., Sandstrom, E., and Wahren, B. (1999). Immune responses in asymptomatic HIV‐1‐infected patients after HIV‐DNA immunization followed by highly active antiretroviral treatment. J. Immunol. 163(4), 2330–2338. Cho, J. H., Youn, J. W., and Sung, Y. C. (2001). Cross‐priming as a predominant mechanism for inducing CD8(þ) T cell responses in gene gun DNA immunization. J. Immunol. 167(10), 5549–5557.
36
Liu and Ulmer
Comerota, A. J., Throm, R. C., Miller, K. A., Henry, T., Chronos, N., Laird, J., Sequeira, R., Kent, C. K., Bacchetta, M., Goldman, C., Salenius, J. P., Schmieder, F. A., and Pilsudski, R. (2002). Naked plasmid DNA encoding fibroblast growth factor type 1 for the treatment of end‐stage unreconstructible lower extremity ischemia: Preliminary results of a phase I trial. J. Vasc. Surg. 35 (5), 930–936. Conry, R. M., Curiel, D. T., Strong, T. V., Moore, S. E., Allen, K. O., Barlow, D. L., Shaw, D. R., and LoBuglio, A. F. (2002). Safety and immunogenicity of a DNA vaccine encoding carcinoembryonic antigen and hepatitis B surface antigen in colorectal carcinoma patients. Clin. Cancer Res. 8(9), 2782–2787. Corr, M., Lee, D. J., Carson, D. A., and Tighe, H. (1996). Gene vaccination with naked plasmid DNA: Mechanism of CTL priming. J. Exp. Med. 184(4), 1555–1560. Corr, M., von Damm, A., Lee, D. J., and Tighe, H. (1999). In vivo priming by DNA injection occurs predominantly by antigen transfer. J. Immunol. 163(9), 4721–4727. Davis, H. L., Michel, M. L., and Whalen, R. G. (1993). DNA‐based immunization induces continuous secretion of hepatitis B surface antigen and high levels of circulating antibody. Hum. Mol. Genet. 2(11), 1847–1851. Denis‐Mize, K. S., Dupuis, M., MacKichan, M. L., Singh, M., Doe, B., O’Hagan, D., Ulmer, J. B., Donnelly, J. J., McDonald, D. M., and Ott, G. (2000). Plasmid DNA adsorbed onto cationic microparticles mediates target gene expression and antigen presentation by dendritic cells. Gene Ther. 7(24), 2105–2112. Denis‐Mize, K., Dupuis, M., Singh, M., Woo, C., Ugozzoli, M., O’Hagan, D., Donnelly, J., Ott, G., and McDonald, D. M. (2003). Mechanisms of Increased Immunogenicity for DNA‐Based Vaccines Absorbed onto Cationic Microparticles. Cell. Immunol. 225(1), 12–20. Donnelly, J. J., Ulmer, J. B., Shiver, J. W., and Liu, M. A. (1997). DNA vaccines. Annu. Rev. Immunol. 15, 617–648. Dupuis, M., Denis‐Mize, K., Woo, C., Goldbeck, C., Selby, M. J., Chen, M., Otten, G. R., Ulmer, J. B., Donnelly, J. J., Ott, G., and McDonald, D. M. (2000). Distribution of DNA vaccines determines their immunogenicity after intramuscular injection in mice. J. Immunol. 165(5), 2850–2858. Epstein, J. E., Charoenvit, Y., Kester, K. E., Wang, R., Newcomer, R., Fitzpatrick, S., Richie, T. L., Tornieporth, N., Heppner, D. G., Ockenhouse, C., Majam, V., Holland, C., Abot, E., Ganeshan, H., Berzins, M., Jones, T., Freydberg, C. N., Ng, J., Norman, J., Carucci, D. J., Cohen, J., and Hoffman, S. L. (2004). Safety, tolerability, and antibody responses in humans after sequential immunization with a PfCSP DNA vaccine followed by the recombinant protein vaccine RTS,S/ AS02A. Vaccine 22(13–14), 1592–1603. Epstein, J. E., Gorak, E. J., Charoenvit, Y., Wang, R., Freydberg, N., Osinowo, O., Richie, T. L., Stoltz, E. L., Trespalacios, F., Nerges, J., Ng, J., Fallarme‐Majam, V., Abot, E., Goh, L., Parker, S., Kumar, S., Hedstrom, R. C., Norman, J., Stout, R., and Hoffman, S. L. (2002). Safety, tolerability, and lack of antibody responses after administration of a PfCSP DNA malaria vaccine via needle or needle‐free jet injection, and comparison of intramuscular and combination intramuscular/intradermal routes. Hum. Gene Ther. 13(13), 1551–1560. Evans, R. K., Xu, Z., Bohannon, K. E., Wang, B., Bruner, M. W., and Volkin, D. B. (2000). Evaluation of degradation pathways for plasmid DNA in pharmaceutical formulations via accelerated stability studies. J. Pharm. Sci. 89(1), 76–87. Fu, T. M., Ulmer, J. B., Caulfield, M. J., Deck, R. R., Friedman, A., Wang, S., Liu, X., Donnelly, J. J., and Liu, M. A. (1997). Priming of cytotoxic T lymphocytes by DNA vaccines: Requirement for professional antigen presenting cells and evidence for antigen transfer from myocytes. Mol. Med. 3, 362–371. Fynan, E. F., Webster, R. G., Fuller, D. H., Haynes, J. R., Santoro, J. C., and Robinson, H. L. (1993). DNA vaccines: Protective immunizations by parenteral, mucosal, and gene‐gun inoculations. Proc. Natl. Acad. Sci. USA 90(24), 11478–11482.
2. Human Clinical Trials of Plasmid DNA Vaccines
37
Gurunathan, S., Klinman, D. M., and Seder, R. A. (2000). DNA vaccines: Immunology, application, and optimization. Annu. Rev. Immunol. 18, 927–974. Haas, J., Park, E. C., and Seed, B. (1996). Codon usage limitation in the expression of HIV‐1 envelope glycoprotein. Curr. Biol. 6(3), 315–324. Hanke, T., McMichael, A. J., Samuel, R. V., Powell, L. A., McLoughlin, L., Crome, S. J., and Edlin, A. (2002). Lack of toxicity and persistence in the mouse associated with administration of candidate DNA‐ and modified vaccinia virus Ankara (MVA)‐based HIV vaccines for Kenya. Vaccine 21(1–2), 108–114. Hariharan, M. J., Driver, D. A., Townsend, K., Brumm, D., Polo, J. M., Belli, B. A., Catton, D. J., Hsu, D., Mittelstaedt, D., McCormack, J. E., Karavodin, L., Dubensky, T. W., Jr., Chang, S. M., and Banks, T. A. (1998). DNA immunization against herpes simplex virus: Enhanced efficacy using a Sindbis virus‐based vector. J. Virol. 72(2), 950–958. Hemmi, H., Takeuchi, O., Kawai, T., Kaisho, T., Sato, S., Sanjo, H., Matsumoto, M., Hoshino, K., Wagner, H., Takeda, K., and Akira, S. (2000). A Toll‐like receptor recognizes bacterial DNA. Nature 408(6813), 740–745. Huang, A. Y., Golumbek, P., Ahmadzadeh, M., Jaffee, E., Pardoll, D., and Levitsky, H. (1994). Role of bone marrow‐derived cells in presenting MHC class I‐restricted tumor antigens. Science 264, 961–965. Kang, K. K., Choi, S. M., Choi, J. H., Lee, D. S., Kim, C. Y., Ahn, B. O., Kim, B. M., and Kim, W. B. (2003). Safety evaluation of GX‐12, a new HIV therapeutic vaccine: Investigation of integration into the host genome and expression in the reproductive organs. Intervirology 46 (5), 270–276. Kano, Y., Miyashita, T., Nakamura, H., Kuroki, K., Nagata, A., and Imamoto, F. (1981). In vivo correlation between DNA supercoiling and transcription. Gene 13(2), 173–184. Klencke, B., Matijevic, M., Urban, R. G., Lathey, J. L., Hedley, M. L., Berry, M., Thatcher, J., Weinberg, V., Wilson, J., Darragh, T., Jay, N., Da Costa, M., and Palefsky, J. M. (2002). Encapsulated plasmid DNA treatment for human papillomavirus 16‐associated anal dysplasia: A Phase I study of ZYC101. Clin. Cancer Res. 8(5), 1028–1037. Klinman, D., Yamshchikov, G., and Ishigatsubo, Y. (1997). Contribution of CpG Motifs to the Immunogenicity of DNA Vaccines. J. Immunol. 158, 3635–3639. Kumaraguru, U., Rouse, R. J., Nair, S. K., Bruce, B. D., and Rouse, B. T. (2000). Involvement of an ATP‐dependent peptide chaperone in cross‐presentation after DNA immunization. J. Immunol. 165(2), 750–759. Le Bon, A., Etchart, N., Rossmann, C., Ashton, M., Hou, S., Gewert, D., Borrow, P., and Tough, D. F. (2003). Cross‐priming of CD8þ T cells stimulated by virus‐induced type I interferon. Nat. Immunol. 4(10), 1009–1015. Epub 2003 September 21. Le, T. P., Coonan, K. M., Hedstrom, R. C., Charoenvit, Y., Sedegah, M., Epstein, J. E., Kumar, S., Wang, R., Doolan, D. L., Maguire, J. D., Parker, S. E., Hobart, P., Norman, J., and Hoffman, S. L. (2000). Safety, tolerability and humoral immune responses after intramuscular administration of a malaria DNA vaccine to healthy adult volunteers. Vaccine 18(18), 1893–1901. Ledwith, B. J., Manam, S., Troilo, P. J., Barnum, A. B., Pauley, C. J., Griffiths, T. G., II, Harper, L. B., Beare, C. M., Bagdon, W. J., and Nichols, W. W. (2000a). Plasmid DNA vaccines: Investigation of integration into host cellular DNA following intramuscular injection in mice. Intervirology 43(4–6), 258–272. Ledwith, B. J., Manam, S., Troilo, P. J., Barnum, A. B., Pauley, C. J., Griffiths, T. G., II, Harper, L. B., Schock, H. B., Zhang, H., Faris, J. E., Way, P. A., Beare, C. M., Bagdon, W. J., and Nichols, W. W. (2000b). Plasmid DNA vaccines: Assay for integration into host genomic DNA. Dev. Biol. (Basel) 104, 33–43. Leitner, W. W., Hwang, L. N., deVeer, M. J., Zhou, A., Silverman, R. H., Williams, B. R., Dubensky, T. W., Ying, H., Restifo, N. P., Bergmann‐Leitner, E. S., Finkelstein, S. E., and Frank, S. (2003). Alphavirus‐based DNA vaccine breaks immunological tolerance by activating innate antiviral
38
Liu and Ulmer
pathways Apoptosis is essential for the increased efficacy of alphaviral replicase‐based DNA vaccines. Nat. Med. 9(1), 33–39. Little, S. R., Lynn, D. M., Ge, Q., Anderson, D. G., Puram, S. V., Chen, J., Eisen, H. N., and Langer, R. (2004). Poly‐beta amino ester‐containing microparticles enhance the activity of nonviral genetic vaccines. Proc. Natl. Acad. Sci. USA 101(26), 9534–9539. Epub 2004 Junuary 21. MacGregor, R. R., Boyer, J. D., Ciccarelli, R. B., Ginsberg, R. S., and Weiner, D. B. (2000). Safety and immune responses to a DNA‐based human immunodeficiency virus (HIV) type I Env/ Rev vaccine in HIV‐infected recipients: Follow‐up data [In Process Citation]. J. Infect. Dis. 181(1), 406. MacGregor, R. R., Ginsberg, R., Ugen, K. E., Baine, Y., Kang, C. U., Tu, X. M., Higgins, T., Weiner, D. B., and Boyer, J. D. (2002). T‐cell responses induced in normal volunteers immunized with a DNA‐based vaccine containing HIV‐1 env and rev. AIDS 16(16), 2137–2143. Manam, S., Ledwith, B. J., Barnum, A. B., Troilo, P. J., Pauley, C. J., Harper, L. B., Griffiths, T. G., II, Niu, Z., Denisova, L., Follmer, T. T., Pacchione, S. J., Wang, Z., Beare, C. M., Bagdon, W. J., and Nichols, W. W. (2000). Plasmid DNA vaccines: Tissue distribution and effects of DNA sequence, adjuvants and delivery method on integration into host DNA. Intervirology 43(4–6), 273–281. Manickan, E., Kanangat, S., Rouse, R. J., Yu, Z., and Rouse, B. T. (1997). Enhancement of immune response to naked DNA vaccine by immunization with transfected dendritic cells. J. Leukoc. Biol. 61, 125–132. Martin, T., Parker, S. E., Hedstrom, R., Le, T., Hoffman, S. L., Norman, J., Hobart, P., and Lew, D. (1999). Plasmid DNA malaria vaccine: The potential for genomic integration after intramuscular injection. Hum. Gene Ther. 10(5), 759–768. McConkey, S. J., Reece, W. H., Moorthy, V. S., Webster, D., Dunachie, S., Butcher, G., Vuola, J. M., Blanchard, T. J., Gothard, P., Watkins, K., Hannan, C. M., Everaere, S., Brown, K., Kester, K. E., Cummings, J., Williams, J., Heppner, D. G., Pathan, A., Flanagan, K., Arulanantham, N., Roberts, M. T., Roy, M., Smith, G. L., Schneider, J., Peto, T., Sinden, R. E., Gilbert, S. C., and Hill, A. V. (2003). Enhanced T‐cell immunogenicity of plasmid DNA vaccines boosted by recombinant modified vaccinia virus Ankara in humans. Nat. Med. 9(6), 729–735. Mincheff, M., Tchakarov, S., Zoubak, S., Loukinov, D., Botev, C., Altankova, I., Georgiev, G., Petrov, S., and Meryman, H. T. (2000). Naked DNA and adenoviral immunizations for immunotherapy of prostate cancer: A phase I/II clinical trial. Eur. Urol. 38(2), 208–217. Moorthy, V. S., Pinder, M., Reece, W. H., Watkins, K., Atabani, S., Hannan, C., Bojang, K., McAdam, K. P., Schneider, J., Gilbert, S., and Hill, A. V. (2003). Safety and immunogenicity of DNA/modified vaccinia virus ankara malaria vaccination in African adults. J. Infect. Dis. 188(8), 1239–1244. Epub 2003 September 30. Mwau, M., Cebere, I., Sutton, J., Chikoti, P., Winstone, N., Wee, E. G., Beattie, T., Chen, Y. H., Dorrell, L., McShane, H., Schmidt, C., Brooks, M., Patel, S., Roberts, J., Conlon, C., Rowland‐ Jones, S. L., Bwayo, J. J., McMichael, A. J., and Hanke, T. (2004). A human immunodeficiency virus 1 (HIV‐1) clade A vaccine in clinical trials: Stimulation of HIV‐specific T‐cell responses by DNA and recombinant modified vaccinia virus Ankara (MVA) vaccines in humans. J. Gen. Virol. 85(Pt 4), 911–919. Nichols, W. W., Ledwith, B. J., Manam, S. V., and Troilo, P. J. (1995). Potential DNA vaccine integration into host cell genome. Ann. N Y Acad. Sci. 772, 30–39. O’Hagan, D., Singh, M., Ugozzoli, M., Wild, C., Barnett, S., Chen, M., Otten, G. R., and Ulmer, J. B. (2001). Induction of potent immune responses by cationic microparticles with adsorbed HIV DNA vaccines. J. Virol. 75(19), 9037–9043. Otten, G., Schaefer, M., Doe, B., Liu, H., Srivastava, I., zur Megede, J., O’Hagan, D., Donnelly, J., Widera, G., Rabussay, D., Lewis, M. G., Barnett, S., and Ulmesr, J. B. (2004). Enhancement of DNA vaccine potency in rhesus macaques by electroporation. Vaccine 22(19), 2489–2493.
2. Human Clinical Trials of Plasmid DNA Vaccines
39
Parker, S. E., Borellini, F., Wenk, M. L., Hobart, P., Hoffman, S. L., Hedstrom, R., Le, T., and Norman, J. A. (1999). Plasmid DNA malaria vaccine: Tissue distribution and safety studies in mice and rabbits. Hum. Gene Ther. 10(5), 741–758. Parker, S. E., Monteith, D., Horton, H., Hof, R., Hernandez, P., Vilalta, A., Hartikka, J., Hobart, P., Bentley, C. E., Chang, A., Hedstrom, R., Rogers, W. O., Kumar, S., Hoffman, S. L., and Norman, J. A. (2001). Safety of a GM‐CSF adjuvant‐plasmid DNA malaria vaccine. Gene Ther. 8(13), 1011–1023. Prazeres, D. M., Monteiro, G. A., Ferreira, G. N., Diogo, M. M., Ribeiro, S. C., and Cabral, J. M. (2001). Purification of plasmids for gene therapy and DNA vaccination. Biotechnol. Annu. Rev. 7, 1–30. Qiu, J. T., Song, R., Dettenhofer, M., Tian, C., August, T., Felber, B. K., Pavlakis, G. N., and Yu, X. F. (1999). Evaluation of novel human immunodeficiency virus type 1 Gag DNA vaccines for protein expression in mammalian cells and induction of immune responses. J. Virol. 73(11), 9145–9152. Robertson, J. S., and Griffiths, E. (2001). Assuring the quality, safety, and efficacy of DNA vaccines. Mol. Biotechnol. 17(2), 143–149. Robinson, H. L., Hunt, L. A., and Webster, R. G. (1993). Protection against a lethal influenza virus challenge by immunization with a haemagglutinin‐expressing plasmid DNA. Vaccine 11(9), 957–960. Rosenberg, S. A., Yang, J. C., Sherry, R. M., Hwu, P., Topalian, S. L., Schwartzentruber, D. J., Restifo, N. P., Haworth, L. R., Seipp, C. A., Freezer, L. J., Morton, K. E., Mavroukakis, S. A., and White, D. E. (2003). Inability to immunize patients with metastatic melanoma using plasmid DNA encoding the gp100 melanoma‐melanocyte antigen. Hum. Gene Ther. 14(8), 709–714. Rottinghaus, S. T., Poland, G. A., Jacobson, R. M., Barr, L. J., and Roy, M. J. (2003). Hepatitis B DNA vaccine induces protective antibody responses in human non‐responders to conventional vaccination. Vaccine 21(31), 4604–4608. Roy, M. J., Wu, M. S., Barr, L. J., Fuller, J. T., Tussey, L. G., Speller, S., Culp, J., Burkholder, J. K., Swain, W. F., Dixon, R. M., Widera, G., Vessey, R., King, A., Ogg, G., Gallimore, A., Haynes, J. R., and Heydenburg Fuller, D. (2000). Induction of antigen‐specific CD8þ T cells, T helper cells, and protective levels of antibody in humans by particle‐mediated administration of a hepatitis B virus DNA vaccine. Vaccine 19(7–8), 764–778. Sato, Y., Roman, M., Tighe, H., Lee, D., Corr, M., Nguyen, M. D., Silverman, G. J., Lotz, M., Carson, D. A., and Raz, E. (1996). Immunostimulatory DNA sequences necessary for effective intradermal gene immunization. Science 273(5273), 352–354. Sekiguchi, J. M., and Kmiec, E. B. (1989). DNA superhelicity enhances the assembly of transcriptionally active chromatin in vitro. Mol. Gen. Genet. 220(1), 73–80. Singh, M., Briones, M., Ott, G., and O’Hagan, D. (2000). Cationic microparticles: A potent delivery system for DNA vaccines. Proc. Natl. Acad. Sci. USA 97(2), 811–816. Smith, H. A., and Klinman, D. M. (2001). The regulation of DNA vaccines. Curr. Opin. Biotechnol. 12, 299–303. Swain, W. E., Heydenburg Fuller, D., Wu, M. S., Barr, L. J., Fuller, J. T., Culp, J., Burkholder, J., Dixon, R. M., Widera, G., Vessey, R., and Roy, M. J. (2000). Tolerability and immune responses in humans to a PowderJect DNA vaccine for hepatitis B. Dev. Biol. (Basel) 104, 115–119. Tacket, C. O., Roy, M. J., Widera, G., Swain, W. F., Broome, S., and Edelman, R. (1999). Phase 1 safety and immune response studies of a DNA vaccine encoding hepatitis B surface antigen delivered by a gene delivery device. Vaccine 17(22), 2826–2829. Tagawa, S. T., Lee, P., Snively, J., Boswell, W., Ounpraseuth, S., Lee, S., Hickingbottom, B., Smith, J., Johnson, D., and Weber, J. S. (2003). Phase I study of intranodal delivery of a plasmid DNA vaccine for patients with Stage IV melanoma. Cancer 98(1), 144–154.
40
Liu and Ulmer
Tang, D. C., De Vit, M., and Johnston, S. A. (1992). Genetic immunization is a simple method for eliciting an immune response. Nature 356(6365), 152–154. Timares, L., Takashima, A., and Johnston, S. A. (1998). Quantitative analysis of the immunopotency of genetically transfected dendritic cells. Proc. Natl. Acad. Sci. USA 95(22), 13147–13152. Timmerman, J. M., Singh, G., Hermanson, G., Hobart, P., Czerwinski, D. K., Taidi, B., Rajapaksa, R., Caspar, C. B., Van Beckhoven, A., and Levy, R. (2002). Immunogenicity of a plasmid DNA vaccine encoding chimeric idiotype in patients with B‐cell lymphoma. Cancer Res. 62(20), 5845–5852. Ulmer, J. B., Donnelly, J. J., Parker, S. E., Rhodes, G. H., Felgner, P. L., Dwarki, V. J., Gromkowski, S. H., Deck, R. R., De Witt, C. M., Friedman, A., Hawe, L. A., Leander, K. R., Martinez, D., Perry, H. C., Shiver, J. W, Montgomery, D. L., and Liu, M. A. (1993). Heterologous protection against influenza by injection of DNA encoding a viral protein. Science 259(5102), 1745–1749. Ulmer, J., Deck, R., Dewitt, C., Donnelly, J., and Liu, M. (1996). Generation of MHC class I‐ restricted cytotoxic T lymphocytes by expression of a viral protein in muscle cells: Antigen presentation by non‐muscle cells. Immunology 89, 59–67. Wang, B., Ugen, K. E., Srikantan, V., Agadjanyan, M. G., Dang, K., Refaeli, Y., Sato, A. I., Boyer, J., Williams, W. V., and Weiner, D. B. (1993). Gene inoculation generates immune responses against human immunodeficiency virus type 1. Proc. Natl. Acad. Sci. USA 90(9), 4156–4160. Wang, R., Doolan, D. L., Le, T. P., Hedstrom, R. C., Coonan, K. M., Charoenvit, Y., Jones, T. R., Hobart, P., Margalith, M., Ng, J., Weiss, W. R., Sedegah, M., de Taisne, C., Norman, J. A., and Hoffman, S. L. (1998). Induction of antigen‐specific cytotoxic T lymphocytes in humans by a malaria DNA vaccine. Science 282(5388), 476–480. Wang, R., Epstein, J., Baraceros, F. M., Gorak, E. J., Charoenvit, Y., Carucci, D. J., Hedstrom, R. C., Rahardjo, N., Gay, T., Hobart, P., Stout, R., Jones, T. R., Richie, T. L., Parker, S. E., Doolan, D. L., Norman, J., and Hoffman, S. L. (2001). Induction of CD4(þ) T cell‐dependent CD8(þ) type 1 responses in humans by a malaria DNA vaccine. Proc. Natl. Acad. Sci. USA 98(19), 10817–10822. Wang, R., Epstein, J., Charoenvit, Y., Baraceros, F. M., Rahardjo, N., Gay, T., Banania, J. G., Chattopadhyay, R., de la Vega, P., Richie, T. L., Tornieporth, N., Doolan, D. L., Kester, K. E., Heppner, D. G., Norman, J., Carucci, D. J., Cohen, J. D., Hoffman, S. L., Epstein, J. E., Newcomer, R., Fitzpatrick, S., Ockenhouse, C., Majam, V., Holland, C., Abot, E., Ganeshan, H., Berzins, M., Jones, T., Freydberg, C. N., Ng, J., and Cohen, J. (2004). Induction in humans of CD8þ and CD4þ T cell and antibody responses by sequential immunization with malaria DNA and recombinant protein Safety, tolerability, and antibody responses in humans after sequential immunization with a PfCSP DNA vaccine followed by the recombinant protein vaccine RTS,S/ AS02A. J. Immunol. 172(9), 5561–5569. Weber, R., Bossart, W., Cone, R., Luethy, R., and Moelling, K. (2001). Phase I clinical trial with HIV‐1 gp160 plasmid vaccine in HIV‐1‐infected asymptomatic subjects. Eur. J. Clin. Microbiol. Infect. Dis. 20(11), 800–803. Wells, D. J. (2004). Gene Therapy Progress and Prospects: Electroporation and other physical methods. Gene Thererapy 11(18), 1363–1369. Williams, S. G., Cranenburgh, R. M., Weiss, A. M., Wrighton, C. J., Sherratt, D. J., and Hanak, J. A. (1998). Repressor titration: A novel system for selection and stable maintenance of recombinant plasmids. Nucleic Acids Res. 26(9), 2120–2124. Wolff, J. A., Malone, R. W., Williams, P., Chong, W., Acsadi, G., Jani, A., and Felgner, P. L. (1990). Direct gene transfer into mouse muscle in vivo. Science 247, 1465–1468. Xiang, Z. Q., Spitalnik, S., Tran, M., Wunner, W. H., Cheng, J., and Ertl, H. C. (1994). Vaccination with a plasmid vector carrying the rabies virus glycoprotein gene induces protective immunity against rabies virus. Virology 199(1), 132–140. zur Megede, J., Chen, M. C., Doe, B., Schaefer, M., Greer, C. E., Selby, M., Otten, G. R., and Barnett, S. W. (2000). Increased expression and immunogenicity of sequence‐modified human immunodeficiency virus type 1 gag gene. J. Virol. 74(6), 2628–2635.
3 Genetic and Environmental
Influences on Antisocial Behaviors: Evidence from Behavioral–Genetic Research Terrie E. Moffitt Institute of Psychiatry, King’s College London, SE5 8AF United Kingdom and University of Wisconsin, Madison, Wisconsin 53706
I. Appraising the Quantitative Behavioral–Genetic Evidence Base A. The number of studies has increased B. Techniques for statistical analysis are more sophisticated C. More studies use large sample sizes D. More is known about how twins and adoptees represent the population E. Behavioral–genetic studies have measured antisocial behaviors with different kinds of valid methods F. Data are now available from many different types of behavior–genetic designs G. Looking for a sturdy finding II. Estimating the Relative Influences of Genes and Environments A. Genes influence approximately 50% of the population variation in antisocial behaviors B. Environmental factors shared by family members influence about 20% of population variation in antisocial behaviors C. Environmental factors experienced uniquely by individuals influence about 20–30% of population variation in antisocial behaviors D. Summary of quantitative genetic findings
Advances in Genetics, Vol. 55 Copyright 2005, Elsevier Inc. All rights reserved.
0065-2660/05 $35.00 DOI: 10.1016/S0065-2660(05)55003-X
Terrie E. Moffitt
42
III. Do the Omnibus Estimates of Genetic and Environmental Influences Always Apply, or Do These Influences Vary Under Different Conditions? A. Is there a sex difference in the genetic influence on antisocial behaviors? B. Are there historical cohort differences in the genetic influence on antisocial behaviors? C. Is there a genetic influence on physical violence? D. Is there a genetic influence on the psychopath? E. Are genetic influences involved when antisocial behavior co‐occurs with other forms of psychopathology, such as hyperactivity? F. Can antisocial experience influence genes? IV. Testing Developmental Theory of Antisocial Behavior A. Stronger genetic liability may be associated with life‐course‐ persistent than adolescence‐limited antisocial behaviors V. Testing Hypotheses About Environmental Causation VI. Testing the Hypothesis of Interaction Between Genes and Environments A. Adoption studies of latent G E B. Adoption studies of latent G measured E C. A twin study of latent G measured E D. Studies of measured G measured E; testing a measured gene E. Research implications of the nil main effect of the MAOA polymorphism on behavior F. Strategy for future G E studies using measured genes VII. The Way Forward Acknowledgments References
ABSTRACT This article reviews behavioral–genetic research into human antisocial behavior. The focus is on studies of antisocial behavior that have been leading the way in investigating environmental and genetic influences on human behavior. The first generation of studies, which provided quantitative estimates attesting that genes and environments each influence about half of the population’s variation in antisocial behaviors is interpreted. Then how behavioral–genetic methods are being applied to test developmental theory and to detect environmental causes
3. Genetic and Environmental Influences on Antisocial Behaviors
43
of antisocial behavior is illustrated. Evidence for interactions between genes and the environment in the etiology of antisocial behavior is also examined. The article ends by envisioning future work on gene–environment interplay in the etiology of antisocial behavior. ß 2005, Elsevier Inc. Despite many years of assiduous efforts to eliminate it, antisocial behavior is still a problem. Approximately 20% of people in the developed world experience victimization by perpetrators of violent and nonviolent illegal behavior each year (U.S. Bureau of Justice Statistics, 2002). The World Report on Violence and Health (WHO, 2002) tallies the staggering burden of mortality, disease, disability, and compromised well‐being brought about by perpetrators of family violence and other violent crimes. Behavioral science needs to achieve a complete understanding of the causes of antisocial behavior to provide an evidence base for effectively controlling and preventing antisocial behavior. Research into genetic and environmental influences is making great strides toward uncovering the root causes of antisocial behavior. Studies are revealing which risk factors are causes, not just correlates. Studies are testing for effects of measured candidate genes. Studies are sorting out how our genotypes sway our susceptibility to environmental causes and how our environments rule the behavioral expression of our genotypes. Studies are refining the antisocial phenotype, uncovering a serious and persistently antisocial subgroup that appears to be more genetically influenced than ordinary antisocial behavior. Journalists have drawn public attention to certain families that seem to contain far more than their share of criminal family members across several generations (Butterfield, 1996, 2002). This familial concentration of crime has been confirmed as a characteristic of the general population (Farrington et al., 1996, 2001; Rowe and Farrington, 1997). In general, fewer than 10% of the families in any community account for more than 50% of that community’s criminal offenses. The family concentration of antisocial behavior could be explained by a genetic influence on antisocial behavior, but it could just as easily be explained by nongenetic social transmission of antisocial behavior within families. Behavioral–genetics research disentangles genetic from nongenetic aspects of familial transmission. Behavioral genetics also has methods to put genetic and nongenetic influences back together again in a systematic and controlled way, to work out how they jointly influence behavior. Behavioral genetics has been rapidly moving beyond the initial question of whether behavior is heritable (Dick and Rose, 2002; Kendler, 2001) to apply its methods to a broad array of causal questions about developmental processes influencing behavior. Given that virtually all behavior and certainly antisocial behavior must be the product of interplay between genes and environments, progress toward understanding cause–effect processes depends on studies not only to separate
44
Terrie E. Moffitt
genetic from nongenetic influences but also studies that can reunite them to observe their interplay. Here, interplay refers to processes in which genes and environments conferring risk for psychopathology co‐occur (gene–environment correlation) and jointly influence the probability that psychopathology will develop (gene environment interaction). Section I of this article critiques the methodological quality of the behavioral–genetic research that has apportioned genetic versus environmental influences on antisocial behavior. How good is the evidence base? Section II summarizes the quantitative estimates of genetic and environmental influences on antisocial behavior resulting from this evidence base, and explains what the findings mean. Section III queries genetic findings with respect to sex differences, cohort effects, violence, the psychopath, antisocial behavior that co‐ occurs with mental disorders, and assortative mating. Section IV illustrates ways that behavioral–genetic designs are being applied to test developmental theory about antisocial behavior. Section V explains how behavioral–genetic designs are newly being used to distinguish risk factors for antisocial behavior that are bona fide environmental causes. Section VI examines interactions between genetic and environmental causes of antisocial behaviors. Section VII puts forward directions for future research.
I. APPRAISING THE QUANTITATIVE BEHAVIORAL–GENETIC EVIDENCE BASE Tables 3.1 to 3.3 bring together all quantitative behavioral–genetic studies of antisocial behaviors, inclusively defined, that were available as at 2004. Reviews up to the mid‐1990s concluded that evidence was accumulating that genetic factors influence which individuals in the population exhibit antisocial behaviors (Carey, 1994; Carey and Goldman, 1997; Gottesman et al., 1997; McGuffin and Thapar, 1998; Miles and Carey, 1997; Raine, 1993; Walters, 1992). The literature of published behavioral–genetic studies of antisocial behaviors has expanded dramatically since those reviews appeared, and there have been six notable improvements in the quality of research into the genetic and environmental architecture of antisocial behaviors.
A. The number of studies has increased To date, more than 100 quantitative genetic studies of antisocial behaviors have been published from more than 60 different genetically informative samples, totaling more than 77,000 families. The last columns of Tables 3.1 to 3.3 show that more than one‐third of the studies have been published after 1994.
Table 3.1. Early Childhood: Estimates of Genetic and Environmental Influences on Population Variation in Antisocial Behavior, from Different Types of Behavioral–Genetic Studies of Young Children Environment estimates Heritability estimate
Common
Unique þ error
Measure of antisocial behavior
Data source
Number of families
Age of probands
Nation/ sample
Authors
Year of publication
Twins reared together design 34%
32%
34%
60%
20%
20%
49–75%
0–22%
25–29%
58%
0%
42%
82%
0%
18%
69%
0%
31%
76%
0%
24%
42%
0%
58%
61%
0%
39%
CBCL Externalizing CBCL Externalizing
Mother
260
2–3
USA, CO
Parents
1358
3
The Netherlands
CBCL Externalizing Physical aggression CBCL Externalizing/ opposition observations/ Berkeley Puppet Conduct Problems CBCL Externalizing CBCL Externalizing Berkeley Puppet Conduct Problems Oppositional behavior at home
Mother
3620
2–3
Mother
4562
19 months
Composite mother/teacher/ observer/self
1116
5
The Netherlands Canada, Quebec UK, E‐risk
Mother
1116
5
UK, E‐risk
Teacher
1116
5
UK, E‐risk
Self
1116
5
UK, E‐risk
Observer
1116
5
UK, E‐risk
Schmitz et al. van den Oord et al. van der Valk et al. Dionne et al.
1995
Arseneault et al.
2003
Arseneault et al. Arseneault et al. Arseneault et al.
2003
Arseneault et al.
1996
1998 2003
2003 2003
2003
(Continues)
Table 3.1. (Continued) Environment estimates Heritability estimate 23%
Common 10%
Unique þ error 67%
Measure of antisocial behavior Disruptive behaviors on videotape, blind observers
Data source Observation in a game of SNAP!
Number of families 1116
Age of probands 5
Nation/ sample UK, E‐risk
Authors
Year of publication
Hughes et al.
2002
Adoptees design No studies Twins reared apart design No studies
Note 1: Studies having fewer than 75 families are excluded from the table. We make an exception to this rule for studies having rare design features (twins reared apart or observational measurement). Note 2: When variance estimates are presented as ranges (e.g., 45–65%), this is usually because the original publication provided estimates separately for males and females. In a few cases, ranges are estimates provided separately for successive waves of a longitudinal study, or for different subscales of a measure. Note 3: Certain samples are represented on more than one row of the table, but each row presents data from a different cohort, a different age or a different measurement source or instrument.
Table 3.2. Middle Childhood to Adolescence: Estimates of Genetic and Environmental Influences on Population Variation in Antisocial Behavior, from Different Types of Behavioral–Genetic Studies of School‐Aged Children and Adolescents Environment estimates Heritability estimate
Common
Unique þ error
Measure of antisocial behavior
Data source
Number of families
Age
172 total
Adolescence
USA 2, Germany 1, Japan 3, UK 1
5–11 7
USA, CO USA, CO
Nation/sample
Authors
Year of publication
Twins reared together design Very low in 7 studies
Not estimated
Not estimated
Delinquency
Official
0% 60%
45% 12%
35% 28%
Hit Bobo doll Bullying
Observer Adult
0–42%
19–69%
31–39%
Rutter antisocial scale
Adult
205
13
UK, London children
57% 51%
22% 28%
21% 21%
CBCL Externalizing CBCL Externalizing
Adult Adult
399 181
4–18 7–15
0%
61%
39%
Adult
198
8–16
37–46%
46–50%
6–12%
Rutter antisocial scale CBCL Externalizing
USA, CO USA, Western Reserve UK, Cardiff
Adult
780
5–9
Norway
13–38% 25–66%
46–62% 4–42%
16–25% 29–33%
Adult Adult
1264 1197
8–16 8–16
USA, VA USA, Virgnia
25–74% 57–65%
0–44% 22–50%
25–77% 6–12%
CBCL Externalizing Rutter antisocial scale Conduct disorder CBCL Externalizing
Adult Adult
1412 1048
8–16 12–15
USA, VA Norway
60%
30%
10%
CBCL Externalizing
Adult
720
9–18
USA, NEAD
85 87
DiLalla and Gottesman, 1989, Table 1; Raine, 1993, Table 5 Plomin et al. O’Connor et al. Stevenson and Graham Schmitz et al. Edelbrock et al. Thapar and McGuffin Gjone and Stevenson Silberg et al. Silberg et al. Eaves et al. Gjone and Stevenson Deater‐ Deckard et al.
1934– 1977
1981 1980 1988
1995 1995 1996 1997 1994 1996 1997 1997b 1997
(Continues)
Table 3.2. (Continued) Environment estimates Heritability estimate
Common
Unique þ error
62–77% 50–76% 70–77% 29–69%
4–12% 0–18% 0% 0–27%
18–27% 24–32% 23–30% 31–44%
50%
18%
70%
Measure of antisocial behavior
Data source
Number of families
Age
Nation/sample
Authors
Year of publication
Adult Adult Adult Adult
1022 501 492 125
7–9 8–16 8–12 7–11
Sweden UK USA, Missouri UK
Eley et al. Eley et al. Hudziak et al. Kuntsi et al.
1999 1999 2000 2000
32%
CBCL Aggression CBCL Aggression CBCL Aggression Conners’ conduct probs CBCL Externalizing
Mother
1116
7
UK, E‐risk
2003
1%
28%
CBCL Externalizing
Teacher
1116
7
UK, E‐risk
7%
36%
57%
Self
1116
7
UK, E‐risk
63%
29%
8%
Dominic‐R DSM conduct problems Olweus scale
Unpublished data from the author Unpublished data from the author Arseneault et al.
1412
8–16
USA, VA
Simonoff et al.
1995
61%
14%
25%
BPI antisocial
405
10–18
USA, NEAD
14%
34%
1506
11
USA, MN
90%
0%
43%
524
17
USA, MN
Krueger et al.
2001
62%
0%
0%
Conduct disorder symptoms Broad externalizing disorder spectrum Conduct problems scale
O’Connor et al. Burt et al.
1998c
52%
Self/adult composite Self/adult composite Self/adult composite Self/adult composite Self/adult composite
5–17
Scourfield et al.
2004
80%
No report
No report
2004
6%
52%
Self/adult composite Self
Hicks et al.
42%
Broad externalizing disorder spectrum MMPI psychopathy
CaStANET, Wales and North England USA, MN USA,Boston
Gottesman
1966
1109
542 152
17 (plus parent) 14–18
2003
2005
2001
14–74%
0–41%
26–45%
70% 25–36% 81%
No report 0–42% No report
No report 52–77% No report
47–54% 18% 35%
0–13% 26%
40–46% 56%
61%
0%
39%
7%
31%
62%
23%
No report
No report
71%
0%
29%
34%
30%
37%
38%
0%
61%
31–36%
10–17%
52–54%
Sibhip design 55%
14%
32%
5%
Socialization aggression Delinquency Conduct disorder Conduct disorder items
Self
326
Self Self Self
99 1412 81
Socialization Delinquency Conduct disorder items Conduct disorder items
Self Self Self
18 13–18 8–16 Adolescence
USA, merit scholars USA, Ohio USA, VA UK, Cardiff
Loehlin and Nichols Rowe Eaves et al. McGuffin and Thapar Taylor et al. Taylor et al. Young et al.
1976
381 486 334
16–18 16–18 12–18
USA, MN USA, MN USA, CO
Self
740
13–21
Miles et al.
2002
Adults recall conduct symptoms Adults recall conduct symptoms Adults recall conduct symptoms Adults recall conduct symptoms Adults recall conduct symptoms Antisocial to parent
Self
3226
<15
USA, NLS Adolescent Health USA, Vietnam Era
Lyons et al.
1995
Self
3372
<15
Slutske et al.
2001
Self
2682
<15
USA, Vietnam Era Australia
Slutske et al.
1997
Self
1075
<18
USA, VA
Jacobson et al.
2000
Self
558
<15
USA, VA
Goldstein et al.
2001
Observer
675
10–18
USA, NEAD
O’Connor et al.
1995
31%
BPI scale
Self
822
5–12
USA, NLSY
1994
63%
4 physical violence items
Self
1515
Adolescence
USA, NLSY
Rodgers et al. Rowe et al.
1983 1997 1997
2000 2000a 2000
1999
(Continues)
Table 3.2. (Continued) Environment estimates Heritability estimate
Measure of antisocial behavior
Data source
Common
Unique þ error
36%
0%
64%
12 delinquent offences
Self
Adoptees design Low 0%
No report No report
No report No report
Antisocial symptoms Conflict
17–49%
0–27%
42–70%
60%
20%
20%
CBCL aggression, CBCL delinquency CBCL Externalizing
48–55%
17–19%
26–35%
Nil
No report
No report
No report
0%
No report
Number of families
Age
Nation/sample
Authors
Year of publication
2338
15–23
USA, NLSY
Rodgers et al.
2001
Adult Observer
513 124
Adol. 3–11
Cadoret et al. Rende et al.
1983 1992
Adult
172
7–12
332
10–15
CBCL Externalizing
Adult
1816
12–15
Conduct disorder diagnosis CBL Externalizing
Adult
162
10–17
Deater‐ Deckard and Plomin van den Oord et al. van der Valk et al. Cadoret
1999
Adult
USA, IA USA, CO Adoption Study USA, Colarado Adoption Study International, The Netherlands International, The Netherlands USA, IA
1978
Self
266
12–18
USA, Midwest
McGue et al.
1996
Self
32
<15
USA, UK
Grove et al.
1990
1994 1998
Twins reared apart design 41%
No report
No report
Adults recall conduct symptoms
Note 1: Studies having fewer than 75 families are excluded from the table. We make an exception to this rule for studies having rare design features (twins reared apart or observational measurement). Note 2: When variance estimates are presented as ranges (e.g., 45–65%), this is usually because the original publication provided estimates separately for males and females. In a few cases, ranges are estimates provided separately for successive waves of a longitudinal study, or for different subscales of a measure. Note 3: Certain samples are represented on more than one row of the table, but each row presents data from a different cohort, a different age or a different measurement source or instrument.
Table 3.3. Adulthood: Estimates of Genetic and Environmental Influences on Population Variation in Antisocial Behavior from Different types of Behavioral– Genetic studies of Adults Environment estimates Heritability estimate
Common
Unique þ error
Measure of antisocial behavior
Data source
Number of families
Age
Nation/sample
Authors
Carey and Goldman, 1997, Table 23.1; DiLalla and Gottesman, 1989, Table 2 Raine, 1993, Tables 1 and 2 Cloninger and Gottesman Centerwall et al.
Year of publication
Twins reared together design Approximately 50%, 10 studies
Not estimated
Not estimated
Crime
Official
607
Adult
Denmark, Norway, UK, USA, Holland, Japan, Finland, Germany
54%
20%
26%
Crime
Official
8350
Adult
Denmark
90%
No report
No report
Official
13,487
36
USA, NAS‐NRC
Nil 18% 34%
No report 7% 1%
No report 76% 65%
Military dishonorable discharge Crime Agression MMPI psychopathy
official Self Self
201 503 133
20–80 28–37 20–25
UK, Maudsley Finland USA, IN
70%
No report
No report
Self
98
40–45
USA, Midwest
58% 0% 28–47%
0% 53% Nil
42% 47% 53–72%
Self Self Self
331 175 300
19–41 16–71 36–54
43%
5%
52%
Self
3226
0%
37%
62%
Self
558
Buss‐Durkee, verbal, indirect, anger scales MPQ aggression Conduct problems Buss‐Durkee aggression scales Antisocial personality disorder Antisocial personality disorder
1929– 1977
1987 1989
Coid et al. Partanen et al. Pogue‐Geile and Rose Cates et al.
1993 1966 1985
USA, MN Canada, Vancouver USA, Vietnam Era Study
Tellegen et al. Livesley et al. Coccaro et al.
1988 1993 1997
36–55
USA, Vietnam Era Study
Lyons et al.
1995
38
USA, VA
Goldstein et al.
2001
1993
(Continues)
Table 3.3. (Continued) Environment estimates Heritability estimate 74% 55% 35–39% 52% 54%
Adoptees design 20–78% in 5 studies
50% 50%
Common
Unique þ error
Measure of antisocial behavior
Data source
Number of families
Age
Nation/sample
136 274 1257 397 247
19–64 19–64 27–64 33 Adult
UK, London UK, London USA, MN USA, MN Ohio and British Columbia
Rushton et al. Rushton Finkel and McGue Krueger et al. Vernon et al.
1986 1996 1997 2001 1999
16,500
Adult
Denmark, Sweden, USA
Carey and Goldman, 1997, Table 23.2; Raine, 1993, Table 4 Loehlin et al. Loehlin et al.
1974– 1989
Grove et al. Tellegen et al. Bouchard and McGue DiLalla et al. DiLalla et al.
1990 1988 1990
0% 0% 0% 0% 0%
26% 45% 61–65% 48% 46%
Aggression Violence MPQ aggression Offending Composite of 18 aggression questionnaires
Self Self Self Self Self
Not estimated
Not estimated
Offending
Official
No report No report
No report No report
Socialization MMPI psychopathy
Self Self
253 253
14–76 14–76
USA, TX USA, TX
Authors
Year of publication
1985 1987
Twins reared apart design 28% 80% 28%
No report 0% 25%
No report 20% 47%
ASPD symptoms MPQ agression Socialization
Self Self Self
32 71 71
16–68 19–68 19–68
USA–UK USA, MN USA, MN
60% 61%
40% 0%
0% 39%
MMPI psychopathy MMPI psychopathy
Self Self
76 119
19–68 18–77
USA, MN USA, MN
1996 1996
Note 1: Studies having few than 75 families are excluded from the table. We make an exception to this rule for studies having rare design features (twins reared apart or observational measurement). Note 2: When variance estimates are presented as ranges (e.g., 45–65%), this is usually because the original publication provided estimates separately for males and females. In a few cases, ranges are estimates provided separately for successive waves of a longitudinal study, or for different subscales of a measure. Note 3: Certain samples are represented on more than one row of the table, but each row presents data from a different cohort or a different age or a different measurement instrument.
3. Genetic and Environmental Influences on Antisocial Behaviors
53
New behavioral–genetic reports on antisocial behaviors appear steadily from large, representative family samples examined by research teams around the world (Boomsma et al., 2002; Martin, 2002).
B. Techniques for statistical analysis are more sophisticated Prior to the 1980s, researchers reported the percentages of MZ versus DZ twins concordant for a disorder, or chi‐square tests of association between adoptees’ and biological parents’ diagnoses. At that time the research aim was merely to test for any evidence of heritable influence on behavior. Quantitative model‐fitting approaches are now standard practice (Carey, 2003; Plomin et al., 2001). These statistical procedures offer the advantage of comparing which of several different theoretically derived models to find out which fits a data‐set best. In the course of estimating genetic influences, these data‐analysis procedures also yield estimates of environmental influences on population variation in antisocial behavior. This advance reflects the field’s growing interest in using genetically sensitive designs to study environmental effects. More information about statistical methods used in behavioral genetics can be found in Carey, 2003; http://psych.colorado.edu/hgss/; Neale and Cardon, 1992; Plomin et al., 2001; Purcell http://statgen.iop.kcl.ac.uk/bgim/.
C. More studies use large sample sizes With the exception of reports from the very large Scandinavian twin and adoption registers reported during the 1980s (Bohman et al., 1982; Cloninger and Gottesman, 1987; Mednick et al., 1984), most studies of antisocial behaviors designed prior to 1995 examined 300 or fewer families. In contrast, most studies reported since 1995 examine a minimum of 1000 families, yielding more reliable estimates of genetic and environmental effects (Hopper, 1999; Martin et al., 1978).
D. More is known about how twins and adoptees represent the population An important issue is whether the prevalence rates and distributions of antisocial behaviors among twins and adoptees represent antisocial behaviors among ordinary people (Rutter, 2002). This issue is relevant to the question of whether estimates of genetic and environmental influence from behavioral–genetic samples apply to the general population. Although this has been assumed more than it has been examined, the assumption is probably defensible for twin studies because twin‐versus‐singleton comparisons have not found differences in the prevalence rates of antisocial behavior or antisocial personality traits (Gjone and Novik, 1995; Johnson et al., 2002; Levy et al., 1996; Moilannen et al., 1999; Simonoff et al., 1997; van den Oord et al., 1995; Van der Valk et al.,
54
Terrie E. Moffitt
1998a). Adoptees, on the other hand, tend to show elevated rates of antisocial outcomes, although the skewed distribution of these outcomes has the same shape within adoptee samples as in the general population (Hutchings and Mednick, 1973; Sharma et al., 1998). Importantly, the effect sizes for associations between risk factors and psychopathology outcomes have been found to be similar across behavioral–genetic and nongenetic studies. For instance, in the Environmental Risk Longitudinal Twin Study (The E‐risk Study) (Moffitt et al., 2002), associations between children’s antisocial behavior and maternal depression, exposure to domestic violence, maternal warmth and negativity, maternal smoking during pregnancy, socio‐economic status, and neighborhood deprivation are all comparable to these associations in the wider literature.
E. Behavioral–genetic studies have measured antisocial behaviors with different kinds of valid methods Prior to 1980 most behavioral–genetic studies relied on official records of criminal conviction to measure antisocial behaviors. Since then, researchers have gathered self‐reports of antisocial behaviors from research participants, reports from other informants, such as parents and teachers, and even observational measures (see Tables 3.1 to 3.3). Behavioral–genetic studies have assessed participants’ antisocial behavior on a frequency continuum of acts, and also as formal psychiatric diagnostic categories of conduct disorder and antisocial personality disorder. This increased variety of measurements offers advantages for inference, because each measurement method’s weaknesses are offset by the strengths of other methods. For example, court conviction records attest that antisocial behavior was serious enough to warrant official sanction by a judge or jury, but research participants’ self‐reports allow access to the majority of antisocial behaviors, which never come to the attention of the courts, or even the police. Contemporary behavioral genetics is surprisingly multidisciplinary. Four disciplines (i.e., psychopathologists, criminologists, personality psychologists, and child psychologists) conceptualize and measure antisocial behaviors somewhat differently, but all share a defining assumption about the construct—antisocial behaviors are behaviors that violate the rights and safety of others. Research from all four disciplines is included in Tables 3.1 to 3.3. Despite disciplinary differences in conceptualization and data‐collection methods, research shows that genetic and environmental influences are more similar than different for clinical, legal, personality, and observational measures of antisocial behavior.
F. Data are now available from many different types of behavior–genetic designs Tables 3.1 to 3.3 show that antisocial behavior has been studied in twins reared together, adoptees, and twins reared apart. Because adoptions became unusual in the 1970s, most studies designed after 1990 feature twins. There are exceptions
3. Genetic and Environmental Influences on Antisocial Behaviors
55
(e.g., a Dutch study is examining international children adopted by Dutch parents) (Table 3.2.; van der Valk et al., 1998b), and a new adoption study will be launched in California (Ge, 2002, personal communication). Behavior– genetics research is not limited to exotic samples; researchers also examine ordinary families whose members vary in genetic relatedness (e.g., full siblings, half‐siblings, step‐siblings, cousins, and unrelated children reared in the same family) (Rowe et al., 1999). This variety of research designs offers a special advantage for inference because comparing their estimates tells us that the genetic and environmental effect sizes for antisocial behavior are robust across different designs; these estimates are not biased by the limitations and flaws peculiar to one design. A number of potential flaws are unique to adoption studies. First, adoption agencies may attempt to maximize similarity between the adoptee’s biological and adoptive families to increase the child’s chance of fitting in with the new family (this is called “selective placement”). Relatedly, biological mothers who intend to give their baby away may neglect prenatal care and continue to abuse substances during pregnancy, and many unwanted babies experience institutionalization before they are adopted (Mednick et al., 1986). If adoptive homes, prenatal care, and institutional care are selectively worse for the babies given up by antisocial biological mothers, this could bias estimates of heritability upward by adding the criminogenic influences of these three unmeasured nongenetic factors to any criminogenic influence of genes. Second, both adoptees and twins reared apart are likely to be reared in home environments that are unusually good for children because adoptive parents are carefully screened. The resulting restricted range of rearing environments could suppress estimates of environmental effects and thus bias heritability estimates upward (Fergusson et al., 1995; Stoolmiller, 1999). However, this flaw of adoption studies is offset by studies of national twin registers (e.g., Cloninger and Gottesman, 1987) or stratified high‐risk twin samples (e.g., Moffitt et al., 2002) because such sampling frames represent the complete population range of environmental and genetic backgrounds. Studies of twins avoid the potential flaws of adoption studies, but they suffer several potential flaws of their own. First, the logic of the twin design assumes that all of the greater similarity between MZ than DZ twins can safely be ascribed to MZ twins’ greater genetic similarity. This “equal environments assumption” requires that MZ twins are not treated more alike than DZ twins on the causes of antisocial behavior (Kendler et al., 1994). Because MZ twins look identical, in theory they might be treated more similarly than DZ twins in some way that promotes antisocial behavior, and as a result, estimates of heritability from studies of twins reared together could be biased upward relative to the correct population value (DiLalla, 2002). In fact, we found that young MZ twins receive somewhat more similar discipline than
56
Terrie E. Moffitt
DZ twins (Jaffee et al., 2004a). However, studies of adoptees do not suffer this flaw, and neither do studies of twins reared apart because MZ twins reared apart do not share environments (unless their genetically influenced behaviors evoke similar reactions from care‐givers in their separate rearing environments, which is a genetic effect). Second, in studies of twins, MZ twins differ more than DZ twins in prenatal factors affecting intrauterine growth (e.g., MZ twins sharing the same chorion appear to suffer more fetal competition for nutrients). These intrauterine factors also violate the assumption that environments are equal for MZ and DZ twins, but intrauterine differences tend to make MZ twins less alike than their genotypes and thus would bias heritability estimates downward (Rutter, 2002). Third, genomic factors that make some MZ twin pairs’ genotypes less than perfectly identical (e.g., random inactivation of genes on one of each girl’s two X chromosomes; Jorgensen et al., 1992) could in theory affect twin‐study heritability estimates, but no evidence shows that these processes influence antisocial behavior. Fourth, parental assortative mating can bias heritability estimates. Coupled partners are known to share similarly high or low levels of antisocial behaviors (Galbaud du Fort et al., 2002; Krueger et al., 1998). When parents of twins mate for similarity, this should increase the genetic similarity of DZ twins, but MZ twins’ genetic similarity cannot increase beyond its original 100%, and as a result, heritability estimates will be biased downward relative to the correct population value. The implication of biological‐parent assortative mating for adoption studies is the opposite, biological‐parent similarity for antisocial behaviors would bias adoptees’ heritability upward relative to the correct population value (because adoptee– biological‐parent correlations would represent a double‐dose of parental genes). Fifth, twin studies using adult reports to measure behavior sometimes suffer from rater artifacts (e.g., adults may mix up or conflate the behavior of MZ twins and they may exaggerate differences between DZ twins). Such a rater artifact does not afflict adoption studies (nor twin studies using the twins’ self‐reports, as twins do not confuse themselves). Skeptics about twin research often comment that adoption studies show lower heritability estimates than twin studies, casting doubt about the twin method. However, where this discrepancy exists for a behavioral outcome, it is easy to explain. Adoption studies may show lower heritability estimates than twin studies because adoption breaks up the association between genetic and environmental risks naturally occurring in ordinary families by removing genetically at‐risk children from damaging homes and placing them in salutary homes. As a result, interactions between environmental adversity and genetic vulnerability that enhance genetic influences on twins and ordinary children are suppressed among adoptees (Stoolmiller, 1999). Another reason for lower heritabilities in most adoption studies is that adoption studies must measure the behavioral phenotype for the family members being compared (parent versus
3. Genetic and Environmental Influences on Antisocial Behaviors
57
child) at different ages, in different cohorts, often using different instruments. In contrast, twin studies measure the phenotype for the family members being compared (two siblings) at the same age, in the same historical cohort, using the same instrument. As evidence that participants’ age matters in this way, parent–child correlations are virtually always lower than sibling–sibling correlations, despite the fact that the genetic relatedness of these two pairs of relatives is the same, 50%. In our E‐risk Longitudinal Twin Study, the father–son correlation for antisocial behavior is .28, whereas the brother–brother correlation for DZ twins is .45. Consistent with this analysis, a metaanalysis found that heritability estimates from traditional adoption studies of parent–child similarity were slightly different from heritability estimates from twin studies of twin–twin similarity, or from newer adoption studies comparing adopted versus natural siblings (Rhee and Waldman, 2002). In any case, an eyeball comparison between designs on Tables 3.1 to 3.3 does not suggest that studies of twins reared together yield heritability estimates that are systematically higher, or lower, than estimates from studies of twins reared apart or of adoptees. On the one hand, this is because any bias arising from factors, such as selective adoptee placement, violations of the equal‐ environment assumption, intrauterine twin differences, or assortative mating, is only very small (Miles and Carey, 1997; Rutter, 2002). On the other hand, these factors bias heritability estimates upward as often as they bias them downward, canceling each other out.
G. Looking for a sturdy finding A fundamental assumption guiding this review is that sturdy inferences ought to be drawn from a cumulative body of studies whose methods differ as much as possible, but provide convergent findings about the same construct. As we have seen, each of the primary designs used by behavioral geneticists has its own Achilles heel(s), but fortunately, each design’s idiosyncratic flaws are offset by compensatory strengths of the other designs. As a consequence, although particular studies and particular designs may be subject to critique, this does not invalidate inferences derived from the entire cumulative evidence base. The greatest confidence can be attained in science when studies deliberately employ different people, times, places, measurements, and research designs, but nonetheless converge on a similar finding (Robins, 1978). Tables 3.1 to 3.3 illustrate that behavioral–genetic studies of antisocial behaviors ranged in participants’ age from 19 months to 70 years, covered the period from the Great Depression to the turn of the 2000 millennium, took place in numerous western nations, used a wide variety of measurement instruments and reporting sources, and comprised twins‐together, adoption, twins‐apart, and sibling designs. The next section asks: Have these studies converged on similar findings?
58
Terrie E. Moffitt
II. ESTIMATING THE RELATIVE INFLUENCES OF GENES AND ENVIRONMENTS This article does not adopt a metaanalytic approach. Rhee and Waldman (2002) already reported a valuable metaanalysis of 50 studies that clarifies the basic effect sizes for latent genetic and environmental influences on antisocial behavior. Instead, this review queries what the estimates of genetic and environmental influence are likely to mean. Before going forward, it is worth acknowledging that coefficients estimating genetic and environmental influence are not immutable properties of nature; they are just statistical coefficients that indicate the balance between genetic and environmental sources of variation in a particular study sample (Hopper, 1998). Thus, it is true that every study’s estimates of heritable and environmental influence on antisocial behavior can apply only to the specific balance of genetic and environmental variation in the time and place in which its participants grew up. Yet if there were no prospect of ever generalizing findings, there would be little reason to conduct any research. Further, the question of interest here concerns what relative influences on the population’s antisocial behaviors can be expected under the balance of genome and environment during contemporary times and in places where we hope to reduce the burden of antisocial behaviors. After all, this is the setting to which all behavioral science hopes to generalize its findings. The field of behavioral genetics has accumulated enough data to address this question (Kendler, 2001).
A. Genes influence approximately 50% of the population variation in antisocial behaviors The basic logic used to make inferences about genetic influences is straightforward. In adoption studies, the correlation between adoptee and biological parent represents genetic transmission, whereas the correlation between adoptee and adoptive parent represents social (i.e., environmental) transmission. In twin studies, genetic influence is inferred if DZ twins’ behavior is less similar than MZ twins’ because DZ twins share on average only half of the genes on which humans can vary, whereas MZ twins share all their genes. In other words, if there were no genetic influence on a phenotype, zygosity should not matter. The more than 100 heritability estimates in Tables 3.1 to 3.3 form a distribution approximating a bell‐shaped normal curve, as shown in Fig. 3.1, with its peak at 50%, and small tails to the left (0% heritability) and right (80% heritability). According to psychometric test theory, this distribution is to be expected from a sample of more than 100 imperfect estimates of a true effect that equals 50% in nature. Estimates below 20% or above 70% tend to emerge from studies with exceptional design features (e.g., observational measures, small sample sizes, very
3. Genetic and Environmental Influences on Antisocial Behaviors
59
Figure 3.1. Heritability estimates for antisocial behavior.
wide age ranges, adults asked to retrospectively report childhood symptoms, and very young children). The most reliable estimates come from the contemporary studies in Australia, The Netherlands, Norway, Sweden, United Kingdom, and the United States, because these studies examined large, population‐representative samples using quantitative modeling techniques. These studies’ estimates tend to converge quite tightly around 50%. Our summary of these studies is buttressed by complementary metaanalyses of twin and adoption studies. One metaanalysis reported a heritability estimate of 50% for 24 studies of measures of the personality trait, aggression (Miles and Carey, 1997). Another metaanalysis reported that the best fitting model to data from 51 studies of antisocial behaviors yielded a heritability estimate of 41% (Rhee and Waldman, 2002). Two differences between the studies in Tables 3.1 to 3.3 and those in Rhee and Waldman’s metaanalysis may explain why our estimate is slightly higher than theirs. First, the metaanalysis allowed one heritability estimate per sample, whereas Tables 3.1 to 3.3 include more than one estimate per sample, if each estimate was for an independent measure. Because studies having better samples also tend to collect and report more measures, the metaanalysis ruled out much information from the most informative samples, which tend to generate heritability estimates near 50%. Second, the metaanalysis lacked information about three of the five early childhood studies shown on Table 3.1, and these studies of very young children tend to estimate higher heritabilities than studies of older children and adolescents.
60
Terrie E. Moffitt
1. Now that a heritable liability toward antisocial behaviors is known, what of it? To answer this question, it is useful to revisit what heritability does not imply. First, evidence of genetic influence for antisocial behavior neither implies immutability nor resistance to intervention. Hair color, that most heritable of traits, is easily changed. The mean height of the population increased notably this century, whereas the amount of variation between individuals’ heights attributable to genes remained the same. Second, evidence of genetic influence does imply that biological processes are involved in the etiology of antisocial behaviors, but biological etiology does not imply that change can only be brought about through biological intervention. A genetic liability to alcoholism is best treated through abstinence from the environmental agent, alcohol. Adoption studies have repeatedly illustrated that adoption into a good home can be an effective treatment for a genetic liability to antisocial behavior (Cadoret et al., 1995; Mednick et al., 1984). A third idea that the omnibus 50% estimate of genetic influence does not imply is that the influence of genes is the same for everyone. It could conceal important heterogeneity in genetic effects within subgroups, such as males versus females, older versus younger generations, or children versus adults (Slutske, 2001). Sections III and IV of this article will address this issue. Fourth, a high heritability estimate does not imply confidence that research at the molecular level will easily find genes for antisocial behavior. This is amply illustrated by the chequered history of molecular research into highly heritable complex phenotypes, such as schizophrenia, autism, and intelligence (Risch, 2000). Nonetheless, if heritability were low, there would be less justification for pursuing molecular genetic research avenues in the first place (Martin et al., 1997; McGuffin et al., 2001), and a high heritability coefficient should be greeted with “healthy skepticism ... but not unhealthy cynicism” about finding genes functional in psychopathology (Insel and Collins, 2003, p. 618). Fifth, evidence of high heritability does not imply that the causal role of nongenetic factors is trivial. To the contrary, it is now recognized that the heritability coefficient indexes not only the direct effects of genes but also the effects of interactions between genes and family‐wide environments (Purcell, 2002; Rutter and Silberg, 2002). In such interactions the effect of an environmental risk may be even larger than previously reported, among the subgroup of individuals having a vulnerable genotype. This is likely to be the case for antisocial behaviors. Section VI of this article will address gene environment interactions. Perhaps the most pragmatic implication from evidence of genetic influence for antisocial behavior is that much of what we thought we knew about environmental causation could be wrong. It tells us that we can no longer
3. Genetic and Environmental Influences on Antisocial Behaviors
61
blindly assume environmental causation by risk factors that are known to predict antisocial outcomes. Rather, we must reexamine each putative environmental risk factor for antisocial behavior, one by one, while using research methods that are capable of applying explicit controls for genetic explanations of the connection between risk factor and antisocial outcome (Rutter et al., 2001). Section V of this article presents research on controls for genetic influence when testing for environmental causation.
B. Environmental factors shared by family members influence about 20% of population variation in antisocial behaviors One useful feature of behavioral–genetic research designs is that they offer two powerful methods for documenting the importance of environmental effects (Plomin et al., 2001). One of these methods of detecting environmental influence tests whether any of the family members in a study sample are more similar than can be explained by the proportion of genes they share. For instance, MZ twins’ genetic similarity is twice that of DZ twins, and therefore, if nothing but genes influenced antisocial behavior, then MZ twins ought to be at least twice as similar as DZ twins. If not, then something environmental has influenced the twins and enhanced their similarity. For almost all human behavioral traits studied so far, environmental factors shared by family members (labeled as “family‐wide,” “shared,” or “common” environment) have not been found to make family members similar. In other words, the estimated influence of common environment has been found to be almost nil for most human behavioral traits (Rowe, 1994). Antisocial behavior is a marked exception. A comparison of common environment effects across 10 psychiatric disorders revealed that such effects were stronger for antisocial personality and conduct disorder than for affective, anxiety, or substance disorders (Kendler et al., 2003). The estimates of common environment effects in the second columns on Tables 3.1 to 3.3 are highly variable across the studies. However, in general it is fair to say that these estimates cluster around 20%. This conclusion is consistent with the estimate from the aforementioned metaanalyses. One metaanalysis reported that a model omitting common environment effects made a poor fit to the data (Miles and Carey, 1997) and the other metaanalysis reported an estimate of 16% for the influence of common environments on antisocial behaviors (Rhee and Waldman, 2002). The small size of this common environment estimate should not be too surprising, because the twin‐study coefficient indexing the common environment does not include environmental effects involved in gene–environment interactions. We can think of the common environment coefficient as the “residual” effects of common environments that remain, after controlling for the influence of gene–environment interactions on the phenotype. Most human behavior involves nature–nurture interplay,
62
Terrie E. Moffitt
and gene–environment interactions have been shown to play an important role in individual’s differences in antisocial behavior in particular (see Section VI of this article). Therefore, it is remarkable that as much as 20% of the population variation in antisocial behavior can be attributed to direct environmental effects not conditional on genetic vulnerability.
1. Caveats Three caveats qualify the conclusion that common environments account for 20% of population variation in antisocial behaviors. The first caveat is that a lot of statistical power is needed to estimate the common environment’s effect. Although some studies are big enough, many have been underpowered to detect statistical significance for the quantitative effects of common environments (Hopper, 1999). Because it is customary to report coefficients from the most parsimonious model, after trimming away any nonsignificant parameters, many studies that obtained modest but non-significant common environment effects have trimmed them away and reported them as zero (see Tables 3.1 to 3.3). However, confidence intervals around non-significant common environment coefficients sometimes imply that the effect size of these coefficients could be small to moderate. The strategy of selecting the most parsimonious model with fewest significant parameters may have led smaller studies to conclude prematurely that the common environment makes no contribution to antisocial behavior. In keeping with this notion that limited power impedes finding common environment effects, Rhee and Waldman (2002) were able to detect a significant common environment effect for antisocial behavior because they reanalyzed prior studies’ twin correlations to generate their own model parameters, with more than adequate power. A second caveat is about the effects of sample composition on estimates of the common environment. Because selective enrolment and selective attrition are common occurrences in twin research, many twin studies have inadvertently restricted their range of participating families to mainly middle class families who are happy to volunteer for research. One consequence of failing to represent wider environmental variation may be underestimating the effects of common environment. A simulation study revealed that restricted range (i.e., censoring that results from selective sample recruitment–attrition) can lead to such biased parameter estimates in quantitative genetic models, and subsequent extension of the simulation exercise to real data (subsamples drawn with deliberate censoring from a twin registry) confirmed that censored samples underestimate common environment influences on behavioral phenotypes (Taylor, 2004). A third caveat is that if the multiple genes influencing antisocial behavior interact (instead of summing), then common environment effects
3. Genetic and Environmental Influences on Antisocial Behaviors
63
will be underestimated in twin studies (but not adoption studies) (Miles and Carey, 1997). Rhee and Waldman (2002) were able to metaanalyze twin and adoption studies’ data together, which allowed them to test for such nonadditive genetic effects. Their best‐fitting model divided the genetic effect into additive (32%) and nonadditive (9%) parts, suggesting that some genes do interact in the etiology of antisocial behavior. Nonetheless, the common environment effect on antisocial behavior (16%) remained unchanged. In sum, limited statistical power, restricted sample composition, and gene–gene interactions may have lead to underestimates of the common environment component in behavioral–genetic studies of psychopathology.
C. Environmental factors experienced uniquely by individuals influence about 20–30% of population variation in antisocial behaviors The second method of detecting environmental influence is to test whether any family members are less similar than expected from the proportion of genes they share (Plomin and Daniels, 1987). For instance, if a pair of MZ twins is not perfectly identical in antisocial behavior, despite sharing all their genes, this indicates that experience has reduced their behavioral similarity. After estimates of the influences of heritability (50%) and common family environment (20%) on antisocial behavior are calculated, the remainder of population variation, 30%, is assumed to reflect environmental influences not shared by family members (labeled as “nonshared,” “unique,” or “person‐specific” experiences). These experiences might include criminogenic experiences unique to the individual and not shared with his sibling, such as a head injury, being the unique target of sexual abuse, living with an antisocial spouse, or serving a prison sentence. Estimates of this effect are shown in the third column of Tables 3.1 to 3.3.
1. Caveats Four important caveats to the conclusion that unique experiences account for 30% of population variation in antisocial behaviors should be considered. First, we should not underestimate the part of the so‐called “unique environment” estimate that may arise from simple measurement error. Measurement error comes into the calculation because random mistakes in measuring behavior will result in scores that look different for twins in an MZ pair, and it is not easy to differentiate such faux MZ differences from true MZ differences caused by the twins’ unique experiences. To ascertain a crude estimate of how much of the variance estimate might reflect real experience, as opposed to measurement error, we might focus on the estimates from studies that use structural factor modeling approaches to derive latent measures of the antisocial behavior
64
Terrie E. Moffitt
construct that are virtually free from measurement error. A handful of studies have constructed latent composites of twins’ antisocial behaviors from multiple sources of data, using reports from the self, mothers, fathers, and teachers (Arseneault et al., 2003; Burt et al., 2001; O’Connor et al., 1998a; Scourfield et al., 2004, 1995). Their estimates of the unique environment effect were 18, 34, 25, 0, and 8%, respectively, averaging about 20%. This exercise suggests that as much as one‐third of the 30% unique environment estimate from quantitative behavioral–genetic models may reflect error, and as such, the unique environment component may have been overemphasized somewhat. The second caveat is that the coefficient for unique environmental effects indexes not only the direct effects of unique experiences but also the effects of interactions between unique environments and genes (Purcell, 2002; Rutter and Silberg, 2002). To the extent that an experience influences twin one more than co‐twin two, and this greater influence depends on twin one’s vulnerable genotype, this effect will be subsumed in the estimate of unique environment effects. The third caveat is that a proposal has been made that this variance component reflecting unexpected dissimilarity among family members might not represent the influence of objective environmental experiences at all (Turkheimer and Waldron, 2000). The alternative put forward is that this variance component may arise from stochastic developmental events and idiosyncratic processes not accessible to empirical study. Debate about this is underway, resolution has not been achieved, and an in‐depth treatment of the debate is beyond the scope of this article, but a couple of points are relevant to the study of antisocial behavior. First, Turkheimer and Waldron (2000) invoked stochastic causation in large part because they observed that studies measuring nonshared experiences so far (e.g., different child–peer interactions, different child–teacher interactions, birth order) report that such experiences have only small effects on behavioral differences among family members. However, it is not known whether these small effect sizes will also apply to antisocial behavior, because its main risk factors (e.g., child abuse and neglect, bonding with an antisocial partner, and drug addiction) have not yet been tested in studies of differential experience. Second, the reason for debate about interpretation of the unique‐environment component in the first place was that, after common environment influences on most phenotypes were found to be nil, unique experience was put forward as the best if not only hope for finding environmental causes of psychopathology (Plomin and Daniels, 1987). However, with respect to antisocial behaviors, common and unique environment influences each explain about 20% of population variation in antisocial behavior and new studies of gene environment interactions suggest the effects of individual experience can be large among subgroups at genetic risk. Thus, students of
3. Genetic and Environmental Influences on Antisocial Behaviors
65
antisocial behavior should be aware of the hypothesis that nonshared variance is stochastic variance but not overly discouraged by it. The fourth caveat is that some longitudinal twin studies have suggested that whatever the actual experiences are that produce this unique component of variance; their effects on behavior are not lasting. For example, in a sample of twins assessed longitudinally, unique effects explained 24% of sample variation in antisocial behavior at age 7 and 24% again at age 14, but these unique effects explained only 6% of the stability of antisocial behavior across age. This means that although unique influences were important in many twins’ lives at both ages, this influence was being shown by different twins at each age (Eley et al., 2003). Common environment influences on antisocial behavior, in contrast, were very stable over time. This pattern has emerged in other studies, but it is not yet clear whether it will become a rule of thumb.
D. Summary of quantitative genetic findings Quantitative behavioral–genetic research reviewed in this section has revealed that genetic causal processes account for only about half of the population variation in antisocial behavior in any given research sample, thereby unequivocally proving that environmental influences account for the other half. This fact constitutes a remarkable contribution to the understanding of causation (Plomin, 1994). We need to know the sizes of omnibus genetic and environmental contributions for two reasons. First, if the genetic contribution is not zero, then it must be controlled in all further studies of alleged nongenetic causes of antisocial behavior. Without control for genetic variation, further risk‐factor research is ambiguous if not uninformative. Second, if the environmental contribution is not zero, this affords more optimism about intervention. Of course, even if a disorder’s origins were 100% genetic, this would not preclude intervention. This reassuring truism is often illustrated using the genetic disease phenylketonuria (PKU), which can be controlled by adhering lifelong to a strict diet. Hair color, like PKU, is 100% genetically determined. But it is easy to change brunette to blonde, again by using a targeted biological intervention. However, note that in the cases of hair color and PKU because knowledge of genetic influences has helped us to understand the causal processes, we know what environmental interventions to apply. Moreover, because we understand the causal processes, we know that intervention must be stringently maintained over the long term, lest the brunette color or the degenerative brain disease resurface. It would be highly unlikely that any behavior disorder is wholly determined by genes, but it is important to begin any program of research into causal processes by ascertaining what effect sizes we can expect for both genetic and environmental influences under natural conditions, in the absence of
66
Terrie E. Moffitt
intervention. For individual differences in antisocial behavior in the overall population, these effects are approximately 50:50 (keeping in mind all the previously‐noted caveats).
III. DO THE OMNIBUS ESTIMATES OF GENETIC AND ENVIRONMENTAL INFLUENCES ALWAYS APPLY, OR DO THESE INFLUENCES VARY UNDER DIFFERENT CONDITIONS? Research reviewed here and elsewhere has suggested that genes account for about 50% of the population variation in antisocial behaviors. However, this omnibus estimate might conceal important heterogeneity in the genetic effects across subgroups. This section addresses genetic influence with regard to sex differences, cohort effects, violence, psychopathy, antisocial behavior that is comorbid with other disorders, and assortative mating between antisocial men and women.
A. Is there a sex difference in the genetic influence on antisocial behaviors? Sex differences in heritability of antisocial behaviors have been the target of extensive research. A sex difference may exist, but if it does it is very small (Eley et al., 1999; Jacobson et al., 2002; Silberg et al., 1994; van der Valk et al., 1998a). The estimates are not tabled by sex in this article because the evidence goes against any importance for sex differences. There are several sources of confusion about sex differences in genetic influences on behavior. Rhee and Waldman (2002) point out that some of the unclarity arises because researchers have confused behavioral–genetic models (which ascertain the magnitude of genetic influences on individual differences within each sex) versus threshold models (which ascertain differences between the sexes in the degree of combined genetic and environmental liability required to express a disorder; Slutske et al., 1997). The threshold model argues that because females experience gender‐role socialization against antisocial acts, any female who becomes antisocial must be under very strong genetic influence. Elsewhere the empirical evidence is reviewed, the bulk of which suggests this model is not correct; the relatively few females who express antisocial behavior experience the same risk factors to the same extent as antisocial males (Moffitt et al., 2001). Further confusion arises from studies that have observed that more antisocial children are produced by antisocial mothers versus antisocial fathers, and deduced from this that antisocial women must be transmitting
3. Genetic and Environmental Influences on Antisocial Behaviors
67
stronger genetic liability to their offspring (Baker et al., 1989). However, as a natural consequence of the relative rarity of antisocial women, only one‐third of antisocial men can mate with an antisocial woman, whereas fully two‐thirds of antisocial women mate with an antisocial man (Cloninger et al., 1978; Smith and Farrington, 2003). It follows that the children of an antisocial mother are twice as likely as children of an antisocial father to have two antisocial parents instead of one, and the children’s greater likelihood of becoming antisocial probably stems from this two‐parent risk, rather than from their mothers’ female sex. Finally, a little‐appreciated feature of sex‐related chromosomal variation could produce an artifactual sex difference in heritability estimated from twins. Females have two X chromosomes, one of them is randomly subjected to inactivation, and this random inactivation could make female MZ twins less identical than male MZ twins on phenotypes associated with X‐linked genes (Jorgensen et al., 1992; Loat et al., 2004). If antisocial behavior were associated with X‐linked genes, we might expect somewhat lower heritability for female than male antisocial behavior. On balance, the results of model tests in large samples indicate that heritability estimates sometimes appear to be slightly lower among females than males. However, formal tests of sex‐specific models of heritability concur that sex‐specific models are not justified (Eaves et al., 1997; Finkle and McGue, 1997; Gjone and Stevenson, 1997a; Kendler et al., 2003; Taylor et al., 2000). The metaanalyses agree, and they had ample statistical power to detect a sex difference, if it had been there (Miles and Carey, 1997; Rhee and Waldman, 2002). Nonetheless, even if the genetic processes leading to antisocial behavior were exactly the same processes in males and females, this would not preclude the possibility that genetic differences between the sexes can explain the sex difference in antisocial behavior. This is because sex differences can be produced if the sexes experience the same cause but in different amounts (Moffitt et al., 2001; Rowe et al., 1995). This truism can apply to genes, raising the interesting possibility that further research into X‐linked (and perhaps Y‐linked) genotypes may help to explain part of the sex difference in antisocial behavior. To date, molecular genetic research has not been harnessed to the question of sex differences in antisocial behavior, but it may prove informative in the future (Rutter et al., 2003).
B. Are there historical cohort differences in the genetic influence on antisocial behaviors? On the one hand, a historical increase in the amount of environmental variation relative to genetic variation might produce a downward shift in heritability estimates (e.g., if the genotype remains stable, but economic inequalities widen). On the other hand, a historic shift toward liberal social conditions that allow
68
Terrie E. Moffitt
people more freedom to express their genetic liability might produce an upward shift in heritability estimates (Dunne et al., 1997). Two studies have compared birth cohorts from roughly 1900 to the 1970s (Jacobson et al., 2000; Slutske, 2001). In both studies, the elder cohorts’ recall of antisocial behavior was more remote, less reliable, and subject to more underreporting than the younger cohort’s recall, which could have produced spurious cohort differences (Simon and VonKorff, 1995), but no differences were found. This null finding can be confirmed by thought experiment using Tables 3.2 and 3.3, to compare the 20 studies of adults published before 1989 or earlier (cohorts born between 1920 and 1960s) versus the 11 studies of young people published 1998 or later (cohorts born after the 1970s). The older‐cohort studies used mainly official measures, whereas the younger‐cohort studies used mainly parent reports, which again should bias toward spurious cohort differences. Nonetheless, no systematic pattern of cohort differences suggests itself.
C. Is there a genetic influence on physical violence? Public debates about the implications of heritability for criminal responsibility often focus on violent crime (Buchanan et al., 2001; Glover, 1996; Rose, 2000; Ross and Shestowsky, 2003). To date, findings about genetic effects on violence are rare and inconsistent. Three studies report evidence of nil heritability for violence (Bohman et al., 1982; Mednick et al., 1984; Sigvardsson et al., 1982), whereas three other studies report evidence that the genetic influence on violence is about the same as that for nonviolent antisocial behavior, 50% (50% in Cloninger and Gottesman, 1987; 32% in Rowe et al., 1999; 55% in Rushton, 1996). The studies finding nil heritability used Scandinavian registers of conviction and low base rates compromised their power to detect effects (Carey, 1994). An area overlooked by behavioral–genetic research to date is partner violence; only one abstract has been reported, in which 1711 men in the Vietnam Era Twin Study were asked whether they ever hit or threw things at their female partners, yielding a heritability estimate of 38% for this question (Toomey et al., 1997). Three behavioral–genetic studies have found moderate genetic influence on the MPQ Aggression scale (Table 3.3), which measures attitudes approving physical violence and has been shown to predict future conviction for violent crime (Moffitt et al., 2000). A heritable liability toward violence remains a reasonable hypothesis.
D. Is there a genetic influence on the psychopath? Psychopathy is a clinical term applied to a subset of individuals considered to be qualitatively distinct from other offenders because their clinical presentation combines persistent and severe antisocial acts with a distinctive personality style
3. Genetic and Environmental Influences on Antisocial Behaviors
69
(callousness, lack of remorse, egocentricity, manipulativeness, and superficial charm; Hart and Hare, 1997) and distinguishing cognitive features (Blair, 1995). Behavioral–genetic studies report moderate heritability (22–67%) for scales assessing psychopathy‐related traits in samples of children (Viding et al., 2005), adolescents (Taylor et al., 2003), and adults (Blonigen et al., 2003; Lyons et al., 1995). Raine (1993, pp. 76–78) concluded that no existing twin or adoption study had used a compelling diagnosis of true psychopathy, and this is still the case. The evidence base for psychopathy is not sufficient to support conclusions, but it seems reasonable to expect psychopathy to be heritable.
E. Are genetic influences involved when antisocial behavior co‐occurs with other forms of psychopathology, such as hyperactivity? Epidemiological studies concur that more than 90% of individuals having conduct or antisocial personality disorder also meet diagnostic criteria for other disorders (Newman et al., 1996; Robins and Regier, 1991), and conduct disorder has been shown to feature prominently in the developmental history of virtually every adult psychiatric disorder, including schizophrenia and eating disorders (Kim‐Cohen et al., 2003). A key hypothesis is that the co‐occurrence of two disorders in one person arises from shared genetic causation. This hypothesis is tested in a “bivariate behavioral–genetic analysis” in which one twin’s symptoms of conduct disorder are correlated with his co‐twin’s symptoms of the other disorder. If the resulting cross‐trait cross‐twin correlations are stronger for MZ than DZ pairs, this implies that twin A’s genetic liability to conduct disorder influences twin B’s genetic liability to the other disorder (i.e., the same genetic factors contribute to the two disorders) (Neale and Cardon, 1992; Waldman and Slutske, 2000). If two disorders are frequently comorbid and have substantial genetic influence and if most of their comorbidity arises from shared genetic influence, these circumstances stimulate important questions (Slutske et al., in press). This directs (1) nosologists to question the two disorders’ putative independence, (2) developmentalists to ask which disorder emerges earliest in ontology, (3) neuroscientists to investigate the two disorders’ shared brain correlates, and (4) gene hunters to check whether genes associated with one disorder are also associated with the other. There is frequent comorbid overlap between antisocial behaviors and hyperactive–impulsive–inattentive behaviors in the young population, and this comorbid presentation predicts poor prognosis into adulthood (Lynam, 1996; Waschbusch, 2002). Six studies have reported quantitative behavioral–genetic analyses of overlap between hyperactive–impulsive–inattentive and antisocial behaviors (Burt et al., 2001; Nadder et al., 2002; Silberg et al., 1996; Thapar et al., 2001; Waldman et al., 2001; Young et al., 2000). Taken together these
70
Terrie E. Moffitt
studies indicate that most if not all of the considerable overlap between hyperactive–impulsive–inattentive and antisocial behaviors can be ascribed to genetic influences they share, with several estimates reaching 90–100%. Other comorbid combinations are being investigated as well. There are reports about antisocial behavior and depression symptoms (e.g., Jaffee et al., 2002; O’Connor et al., 1998b), antisocial behavior and pathological gambling (e.g., Slutske et al., 2001), and antisocial behavior and substance abuse (e.g., Hicks et al., 2004; Jang et al., 2000; Kendler et al., 2003; Krueger et al., 2002; Miles et al., 2002; Slutske et al., 1998; True et al., 1999; Waldman and Slutske, 2000). In each pair of comorbid problem behaviors, so far genes explain roughly 40–90% of the overlap. The promiscuous way in which antisocial behavior co‐occurs beyond chance with virtually all major mental disorders (Kim‐Cohen et al., 2003; Lambert et al., 2001), coupled with initial evidence that much of this co‐ occurrence has genetic origins, suggests a hypothesis—the presence of marked antisocial symptoms might signal greater severity in an individuals’ genetic liability toward any and all psychopathologies.
F. Can antisocial experience influence genes? This review has emphasized the way by which genes influence the distribution of antisocial behaviors in the population. A less‐appreciated finding is that antisocial behaviors can influence how genes are distributed in the population. Men and women mate on the basis of similarity between the two partners’ antisocial behavior (Farrington et al., 1996, 2001; Galbaud du Fort et al., 2002; Krueger et al., 1998; Rowe and Farrington, 1997). Reported correlations between couple members are moderate to large in size (approximately .50). Although many dual‐ antisocial relationships do not last, the individuals in them tend to have more children than the norm (Krueger et al., 1998). If parents mate assortatively for successive generations, genes relevant to antisocial behavior will become concentrated within families and entire families will increasingly differ from each other as a result of genetic and environmental modes of intergenerational transmission. Height provides a corollary; after generations of mild assortative mating for height, families are made up of people who are similar on height, and whole families tend to differ from other families on height. It is possible that assortative mating may in part account for a puzzling and unexplained fact about antisocial behavior. The prevalence of conduct problems among young people has been rising slightly during the 20th century (Collishaw et al., 2004; Rutter and Smith, 1995). This increase is thought to be too rapid to be explained by a shift in the population genome, so it is presumed to be the result of worsening trends in the environmental causes of antisocial behaviors (Rutter and Smith, 1995). Nevertheless, if assortative mating for antisocial behavior increased over several successive generations and if antisocial individuals who mate assorta-
3. Genetic and Environmental Influences on Antisocial Behaviors
71
tively produced more children than average, then in combination these two processes could increase the population prevalence of antisocial behavior. It is likely that assortative mating for antisocial behavior has increased as a side effect of assortative mating for education. An enormous historical rise in assortative mating across the 20th century has been quantified for educational attainment (Mare, 1991). People tend to select a mate when they complete schooling. In the 1930s, when almost all Americans finished schooling in their mid‐teens (e.g., 90% of women completed only 12 years of schooling), a variety of partners was available to choose from. But as demands for more education grew, people leaving school began to find their pool of prospective partners limited to those leaving school at the same age (in 1987, 50% of women had 12 years or fewer of schooling, 25% had 1–3 years of college, and 25% had 16 years or more of education). As women came to be breadwinners across the 20th century, men with good economic prospects increasingly competed for women with similarly good economic prospects. Antisocial behavior is a strong correlate of educational attainment (Harlow, 2003; Miech et al., 1999; Moffitt et al., 2002). Young people who engage in antisocial activities have high rates of school drop‐out and school expulsion and restricted opportunities for higher education (Cairns and Cairns, 1994; Nagin et al., 2003). As a result, their pool of potential mates may be limited to others also having little education; mates who are also statistically likely to engage in antisocial behavior. In this way, increasing assortative mating for education may inadvertently bring about an increase in the number of large families afflicted with coinciding genetic and environmental causes of antisocial behavior. Coinciding genetic and environmental risks often potentiate each other (Rutter and Silberg, 2002) and one consequence could be an increase in the prevalence of young people with conduct problems. This scenario remains speculative. It is included here to make the point that behavior can influence the gene pool and to provoke research. Uncovering reasons behind historic rises in prevalence can inform efforts to reduce prevalence.
IV. TESTING DEVELOPMENTAL THEORY OF ANTISOCIAL BEHAVIOR It is often said disparagingly that behavioral–genetics research is “a–developmental” and “a–theoretical”. However, this is far from true, and behavioral–genetic methods are ideal for testing developmental theories about age‐related changes in etiology (DiLalla, 2002). Think of the heritability statistic as an outcome variable. When the heritability coefficient changes with age, this suggests that the balance of genetic versus environmental causal processes in the population leading to antisocial behavior differs at successive developmental stages in the life course. For example, the heritability coefficient for IQ increases
72
Terrie E. Moffitt
from childhood to adulthood, and conversely the common environment coefficient for IQ decreases from childhood to adulthood (Cherney et al., 1997; McGue et al., 1993). A similar age‐shift has been reported for bone mineral density (Hopper, 1999). These findings have been interpreted developmentally, suggesting that the effects of environmental factors shared by siblings dissipate once the siblings leave home and live apart. The study of antisocial behavior provides another compelling example of the application of behavioral genetics to testing developmental theory, but the pattern of developmental findings is rather different from that of IQ and bone mineral density.
A. Stronger genetic liability may be associated with life‐course‐ persistent than adolescence‐limited antisocial behaviors Research has provided support for a developmental taxonomy of antisocial behavior (Moffitt, 1993), which proposed two primary prototypes (i.e., life‐course‐ persistent versus adolescence‐limited antisocial individuals). Life‐course‐persistent offenders’ antisocial behavior is thought to have its origins in neuro‐developmental processes, and it begins in childhood and continues persistently thereafter. In contrast, adolescence‐limited offenders’ antisocial behavior is thought to have its origins in processes of social influence; it begins in adolescence and desists in young adulthood. Research has shown that adolescence‐limited antisocials are common, relatively transient, and near normative, whereas life‐course persistent antisocials are few, persistent, and pathological (for a review of research into this taxonomy since 1993, see Moffitt, 2003). If genetic etiological processes contribute more to life‐course‐persistent than adolescence‐limited antisocial development, we would expect to find that heritability estimates are larger for antisocial behaviors committed by young children and adults than for antisocial behaviors committed by adolescents (Moffitt, 1993, p. 694). DiLalla and Gottesman (1989) were the first to observe that adult crime seemed to be more heritable than adolescent juvenile delinquency. As it turns out, the lack of heritability among juveniles in their review probably resulted from low power and insensitive measurement; in 1989 the entire literature of behavioral–genetic studies of juvenile delinquency consisted of fewer than 200 twin pairs, and the measure of antisocial behaviors was conviction, a rare outcome for juveniles. Since then, a large number of better‐designed behavioral–genetic studies have proven that juvenile antisocial behavior is at least somewhat heritable (see Tables 3.1 and 3.2). Nonetheless, among these studies, four groups of studies suggest that life‐course‐persistent antisocial behavior may have stronger heritable origins than its adolescence‐limited counterpart. The first group comprises four studies of large representative samples of very young twins (Table 3.1). Because life course persistent antisocial behavior
3. Genetic and Environmental Influences on Antisocial Behaviors
73
onsets early in life, if it is genetically influenced we would expect high heritability coefficients from studies of very young children. Dionne et al. (2003) report 58% heritability for aggression among 19‐month olds. Van den Oord et al. (1996) report 69% heritability for aggression among 3‐year olds. Van der Valk et al. (1998a) report 50% heritability for externalizing behaviors among 2– 3‐year‐old boys and 75% for girls. Arseneault et al. (2003) report heritabilities of 61, 69, and 76 among 5‐year‐olds, for ratings of antisocial behaviors made by observers, mothers, and teachers, respectively. These high estimates for very young twins contrast against the lower estimate of 41% heritability from a metaanalysis of older samples (Rhee and Waldman, 2002). A second group of studies has identified the two sub‐types on the basis of the heterogeneity in the phenotype, often using the “Aggression” and “Delinquency” scales from the Child Behavior Checklist (CBCL) (Achenbach, 1991). The Aggression scale is thought to be associated with the life‐course‐persistent prototype because it measures antisocial personality and physical violence and its scores are stable across development, whereas the Delinquency scale is associated with the adolescence‐limited prototype because it measures rule‐breaking and its mean scores rise steeply during adolescence (Stanger et al., 1997). In fact, both life‐course‐persistent and adolescence‐limited young people engage in the behaviors on the Delinquency scale, but adolescence‐limited young people are relatively more numerous and if they have less genetic risk, we would expect the Delinquency scale to yield lower heritability estimates than the Aggression scale. Twin and adoption studies of these scales report higher heritability for Aggression (around 60%), than Delinquency (around 30–40%), while the shared environment is significant only for the Delinquency scale (also around 30–40%) (e.g., Deater‐Deckard and Plomin, 1999; Edelbrock et al., 1995; Eley et al., 1999; Schmitz et al., 1995). The approach of contrasting two phenotypes within antisocial behavior was also taken by Viding et al. (2005), who reported that genes influenced 81% of the variation in antisocial behavior among callous, unemotional children, but only 30% of variation among the other children who engaged in ordinary antisocial behaviors. A different approach to contrasting two heterogeneous phenotypes was followed by Arseneault et al. (2003), who found that antisocial behavior that was pervasive across settings, was more heritable than antisocial behavior that was situational; heritability was 82% if a child’s antisocial behavior was agreed on by four different reporters across settings at home and at school, but lower (28–51%) for antisocial behavior limited to one setting or one reporter. This finding has been replicated (Scourfield et al., 2004). A third group of studies has defined life‐course‐persistent antisocial behavior in terms of preadolescent onset, contrasting it against antisocial behavior that begins during the adolescent period. One study found early onset to be strongly familial and substantially heritable in contrast to adolescent onset,
74
Terrie E. Moffitt
which was less familial and largely influenced by environment (Taylor et al., 2000b). In a Swedish twin study, 5‐year continuity from childhood to adolescence in the CBCL Aggression scale was largely mediated by genetic influences, whereas continuity in the Delinquency scale was mediated both by the shared environment and genetic influences (Eley et al., 2003). A fourth group of studies has taken a developmental approach to the other end of the lifespan, defining life‐course persistent antisocial behavior in terms of presence in adolescence combined with subsequent persistence to adulthood antisocial personality disorder. Two studies demonstrated that such persistent antisocial behavior was significantly more heritable than that limited to adolescence (Jacobson et al., 2001; Lyons et al., 1995). These longitudinal studies are supported by a metaanalysis containing adolescent and adult samples assessed with similar measures of aggression, in which adult samples generated significantly higher heritability estimates, on average, than adolescent samples (Miles and Carey, 1997). Rhee and Waldman’s (2002) metaanalysis did not find higher heritability for adults than adolescents, because in the pool of studies they examined, age was wholly confounded with reporting source; adolescent studies used rating scales, whereas adult studies used official crime records. Taken together, the four groups of existing studies suggest that the pattern of antisocial behavior that (1) begins early in life, (2) is pervasive across settings, (3) is characterized by aggressive personality traits, (4) includes physical aggression, and (5) persists into adulthood is associated with relatively more genetic influence than is the pattern of late onset, situational, and transient delinquency. What is missing from the evidence base are prospective‐longitudinal twin studies that ascertain individual differences in trajectories derived from repeated measures of antisocial behaviors over meaningful developmental periods (Nagin et al., 1995, 1999). The taxonomic theory would predict stronger MZ than DZ twin similarity for membership in a childhood‐onset persistent trajectory and less MZ versus DZ difference in twin similarity for membership in an adolescent‐onset transient trajectory.
V. TESTING HYPOTHESES ABOUT ENVIRONMENTAL CAUSATION During the 1990s, the assumption that “nurture” influences behavior came under fire. Traditional socialization studies of antisocial behavior that could not separate environmental influences from their correlated genes were challenged by four important empirical discoveries: (1) ostensible environmental measures are influenced by genetic factors (Plomin and Bergeman, 1991); (2) parents’ heritable traits influence the environments they provide for their children (Kendler, 1996; Plomin, 1994); (3) people’s genes influence the environments they encounter (Kendler, 1996; Plomin et al., 1977); and (4) environmental
3. Genetic and Environmental Influences on Antisocial Behaviors
75
influences did not seem to account for the similarity among persons growing up in the same family (Rowe, 1994). It was said that although nonbehavioral– genetic studies might show that certain rearing experiences predict young people’s antisocial outcomes, theories of causation based on findings from such designs were guilty of a fundamental logical error (i.e., mistaking correlation for causation) (Scarr, 1992). These challenges culminated in admonishments that so far the evidence for genetic influences outweighed the evidence for environmental influences within the family (Harris, 1998; Pinker, 2002; Rowe, 1994). Many social scientists responded to this claim, reasserting evidence for family environmental influences (Collins et al., 2000; Reid et al., 2002; Vandell, 2000). The best way forward to resolve the debate is to use genetically‐sensitive designs that can provide leverage to test environmental causation. To our knowledge, a study of antisocial behavior was the first ever in the behavioral sciences to apply behavioral–genetic methods to control for genetic confounds while testing an environmental hypothesis (VanDusen et al., 1983). It was well established that low socioeconomic status is a risk factor for offending, but Mednick et al. were concerned that some dysfunctional genetic susceptibilities transmitted within families might account for the coincidence of fathers’ low‐status occupations with sons’ antisocial activities. As such, they used the Danish Adoption Study data to disentangle the socioeconomic status that adoptees were conceived in (their biological father’s occupational status) from the socioeconomic status in which they were reared (adoptive father’s status). Results demonstrated that biological inheritance could not explain the majority of the class–crime connection; the social class in which people grow up had a direct causal environmental effect on their probability of criminal offending (VanDusen et al., 1983). A central barrier to interpreting an association between an alleged environmental risk factor and antisocial outcome as a cause–effect association is, of course, the old bugbear that correlation is not causation. Some unknown third variable may account for the association, and that third variable may well be heritable. For example, does the cycle of violence from abusive parent to aggressive child arise from environmental transmission, or genetic transmission (DiLalla and Gottesman, 1991)? This question is fundamental because studies of adoptions have documented the dispiriting fact that aggression emerges in adoptive children despite the fact that they were separated from their at‐risk biological parents at birth and reared by skilled and loving adoptive parents. (For this result in a cross‐fostering study of Rhesus macaques, see Maestripieri, 2003.) Because much research on the intergenerational transmission of antisocial behavior continues without genetic controls, this point cannot be made too often (Serbin and Karp, 2003). A variable is called a risk factor if it has a documented capacity to statistically predict antisocial outcome. The causal status of most risk factors is
76
Terrie E. Moffitt
unknown, we know what predicts antisocial outcomes, but not how or why (Kraemer, 2003). Genetic influences can confound an environmental interpretation of the association between a risk factor and antisocial outcomes in two different ways. The first confound is termed a passive correlation between genotype and environment (rGE). A passive rGE confound occurs when a child’s behavior and the environment his parents provide are correlated because they have the same origins in his parents’ genotype (i.e., not because the bad family environment itself causes children’s aggression). Parents may transmit to their child a genetic liability for aggression, and simultaneously provide an environment of violent and abusive maltreatment, which is symptomatic of the parent’s own genetic liability for aggression. To the extent that parent‐provided environments are under genetic influence, then the observed association between family environment and young peoples’ antisocial outcomes could be a spurious artifact of a third variable causing both (i.e., genetic transmission). The second confound is termed an “active” correlation between genotype and an environmental measure, and it is also abbreviated as “rGE” (Plomin et al., 1977). Active rGE occurs when a person’s behavior and the environment he receives are correlated because they have the same origins in his own genotype (i.e., not because the bad environment itself causes aggression). Active rGE occurs when people’s genetically influenced behavior leads them to “(1) create, (2) seek, or (3) otherwise end up in environments that match their genotypes” (Rutter and Silberg, 2002, p. 473). Antisocial behavior “creates” social reactions from others when aggressive toddlers evoke harsh discipline, when bullies are rejected by peers and expelled from schools, when young shoplifters are convicted by courts, or when abusive husbands are divorced by their wives. Antisocial individuals “seek” environmental settings consonant with their proclivities when antisocial children gravitate toward a delinquent peer group, when antisocial young men and women mate assortatively, or when pedophiles seek volunteer work with children. People who have behaved antisocially selectively “end up in” criminogenic environments when misbehaving children are tracked into special classes for disturbed pupils, when troubled teenagers are recruited by gangs, when violent young men are imprisoned, or when parolees find nothing but unskilled jobs available to them. Like passive rGE, active rGE confounds interpretation of the association between risk factors and antisocial behavior. To the extent that criminogenic environments are elicited by a young person’s genetically influenced behavior, then the observed association between environment and antisocial outcome could be a spurious artifact of a third variable that causes both (i.e., the young person’s genotype). Ordinary studies cannot test whether a risk factor is causal, and it would be unethical to assign children to experimental conditions expected to induce aggression. To rule out rGE to test causal hypotheses, researchers have three options. First, researchers can use the longitudinal method to show that an
3. Genetic and Environmental Influences on Antisocial Behaviors
77
environmental event coincides with the timing of within‐individual behavior change (a natural experiment that rules out rGE by using participants as their own controls) (Cicchetti, 2003; Costello et al., 2003). Second, researchers can test causation by implementing a randomized trial to show that altering the environment can reduce disordered behavior (a treatment experiment that rules out rGE by random assignment) (Howe et al., 2002). Third, behavioral–genetic designs are a useful addition to a toolkit for testing environmental causation. It is somewhat counterintuitive to think about using behavioral–genetic designs to control for and rule out genetic influences while highlighting environmental influences in bas relief, but paradoxically, this is one of their strongest applications. The studies reviewed in a later section are not intended to be exhaustive, but are intended to illustrate the kinds of studies being done. The new generation of behavioral–genetic research designs that can evaluate whether a risk factor has an environmentally mediated effect on children’s aggression has three key features. First, the studies must employ a genetically sensitive design to control for the confounding effects of parents’ genes (passive rGE) or children’s genes (active rGE) on putative environmental measures. The second key feature is that designs must employ an observed measure of the construct alleged to have environmental effects on antisocial behavior. Traditional quantitative behavioral–genetic studies have reported latent environmental variance components, but not observed measures. The third key feature is that genetically informative samples must accurately represent the full range of families’ environmental circumstances. Many behavioral– genetic samples suffer substantial biases in recruitment and attrition, inadvertently restricting their range of participating families to primarily the middle class. Contemporary theories of psychopathology implicate experiences outside the normal range, such as exposure to domestic violence or child maltreatment, which are generally concentrated in the poorest segment of the population, the segment not sampled by most behavioral–genetic studies. (Scandinavian national twin registers of psychiatric hospital and court records accurately represent variation in the population, but such register studies have been unable to measure children’s environments directly.) It is possible to incorporate a measured “environmental” risk factor into a study of twins to test if a shared experience makes twins more similar on antisocial behavior than could be predicted, based on their degree of genetic relationship. A basic approach is to conduct ordinary behavior–genetic modeling, which apportions genetic versus environmental effects on twins’ behavior (ACE), and then add a measured putative environmental risk factor (M‐ACE) to test if the twins’ shared experience of that risk factor can account for any of the shared environmental variation in their behavioral phenotype. The first twin study to apply this approach to problem behavior reported that living in a deprived neighborhood explained a significant 5% of the shared environmental
78
Terrie E. Moffitt
variation in 2‐year olds’ behavior problems (Caspi et al., 2000). Another study applied this approach to examine 5‐year‐old’s exposure to their mothers’ experience of domestic violence (Jaffee et al., 2002). Exposure to domestic violence over the first 5 years of their lives was particularly relevant for children who developed both externalizing and internalizing problems simultaneously; such co‐occurring problems are associated with poor prognosis. Domestic violence exposure explained a significant 13.5% of the shared environment variance in children’s comorbid outcome. A caveat about this approach is in order. Inference of environmental causation is compromised if parent and child share genes that simultaneously influence both the measure of environment and the measure of child antisocial behavior. Use of the basic twin design to test “environmental” risk factors has been improved upon by adding indicators of mothers’ and fathers’ behavioral phenotype to the usual indicators of twin behavior. This approach, called the “extended twin‐family design” (Kendler, 1993), estimates the effect of the putative environmental risk factor on child behavior when controlling for genetic effects on both parents and children. An assumption of the design is that the parental phenotype measures carry genetic information parallel to that in the child phenotype measures. (Although this assumption is seldom fulfilled perfectly it seems not unreasonable for antisocial behavior, which has strong child‐to‐adult continuity.) The first twin study to apply this approach to parenting was reported from the Virginia Twin Study of Adolescent Behavioral Development (Meyer et al., 2000). Antisocial conduct problems were assessed for adolescent twins and their parents in 1350 families. The measured parenting variables were called “marital discord” and “family adaptability”. No effect was found for marital discord, but measured family adaptability accounted for 4% of the variance in adolescents’ conduct problems. Another approach is to use twin‐specific measures of the “environmental” risk factor, which allows researchers to test whether an active rGE (i.e., a child effect evoking risk) accounts for the association between the risk factor and antisocial outcome. The Environmental Risk Longitudinal Twin Study used this approach to examine the effects of physical maltreatment on young children’s aggression (Jaffee et al., 2004b), using twin‐specific reports of maltreatment. This study satisfied six conditions that together supported the hypothesis that physical maltreatment has an environmentally mediated causal influence on children’s aggression: (1) children’s maltreatment history prospectively predicted aggression, (2) the severity of maltreatment bore a dose–response relation to aggression, (3) the experience of maltreatment was followed by increases in aggression from prior levels, within individual children, (4) there was no child effect evoking maltreatment, (5) maltreatment predicted aggression while mothers and fathers’ antisocial behavior were statistically controlled; and (6) modest but significant effects of maltreatment on aggression remained present
3. Genetic and Environmental Influences on Antisocial Behaviors
79
after controlling for genetic transmission of liability to aggression in the family. A similar analytic approach using twin‐specific measures of risk was taken by the Minnesota Twin Family Study (Burt et al., 2003), which studied 808 cases with 11‐year‐old twin pairs. Models revealed that measured parent–child conflict accounted for 12% of the variance in the externalizing syndrome of oppositional, conduct, and attention‐deficit‐hyperactivity disorders (23% of the common environment variation in this syndrome). The Environmental Risk Longitudinal Twin Study also focused on MZ twins only to test if differences between siblings’ exposure to differential parental treatment makes them different on antisocial behavior. The fact that MZ twins are not perfectly concordant for antisocial behavior opens a window of opportunity to uncover if a nongenetic cause specific to one twin has produced the behavioral difference. Comparing the experiences of discordant MZ twins allows the least ambiguous interpretation of results. Three studies have reported that MZ twin differences in parental treatment are correlated with MZ twin differences in antisocial behavior (Asbury et al., 2003; Caspi et al., 2004; Pike et al., 1996). The Environmental Risk Longitudinal Twin Study reported that within 600 MZ twin pairs, the twin who received relatively more maternal negativity and less maternal warmth developed more antisocial behavior problems (Caspi et al., 2004). Negativity and warmth were measured by coding voice tone and speech content in mothers’ audio taped speech about each of their twins separately, according to the “expressed emotion” paradigm. This study provided evidence that the effect of mothers’ emotional treatment of children causes antisocial behavior, by ruling out five alternative explanations of the finding: (1) using MZ twin pairs ruled out the possibility that a genetically transmitted liability explained both the mother’s emotion and her child’s antisocial behavior; (2) using MZ twins also ruled out the possibility that a genetic child effect provoking maternal emotion accounted for the finding; (3) the study included a longitudinal natural experiment to rule out the possibility that any nongenetic child effect accounted for the finding, by controlling for prior child behavior that could have provoked maternal negative emotion; individual children whose mothers were negative toward them at age 5 evidenced a subsequent increase of antisocial behavior between age 5 and 7; (4) the study controlled for twin differences in birth weight in an effort to rule out the possibility that twins with neuro‐developmental difficulties had more behavior problems that elicited more negative emotion from mothers; (5) the study measured the children’s behavior using teacher reports to rule out the possibility that a mother’s negativity toward a child led her to exaggerate her report of the child’s behavior problems. These studies testing measured environmental variables were conducted very recently, illustrating that such testing is a new direction in behavior–genetic research (Dick and Rose, 2002; Kendler, 2001). So far, behavior–genetic designs
80
Terrie E. Moffitt
have been applied to test causation for only a handful of risk factors for antisocial behavior. We expect the list of risk factors studied with behavioral–genetic controls to grow to encompass environmental factors ranging from prenatal teratogens to prison sanctioning of adult offenders.
VI. TESTING THE HYPOTHESIS OF INTERACTION BETWEEN GENES AND ENVIRONMENTS The study of gene–environment interaction entails substantial methodological challenges. It requires measured environments that are truly environmental, measured genetic influence, some means of separating them from each other, and enough statistical power for a sensitive test of interaction (Rutter and Silberg, 2002). Despite the challenges, theory driven hypotheses of G E interaction are well worth testing because where measured G E are found to influence behavior disorders, both specific genes and specific environmental risks can conceivably have moderate‐to‐large effects. Specific genes revealed to be stronger in the presence of environmental risk would guide strategic research into those genes’ expression, possibly leading to genetic diagnostics and improved pharmacological interventions (Evans and Relling, 1999). Specific environmental effects revealed to be stronger in the presence of genetic risk would prompt a new impetus for specific environmental prevention efforts, and would help to identify who needs the prevention programs most. The study of G E is especially exciting in antisocial behavior research, where investigations have pioneered the way for all behavioral disorders. Studies of antisocial behavior were first to report evidence of interaction between latent genetic and latent environmental risks ascertained in adoption studies, and also first to report evidence of an interaction between a measured genetic polymorphism and a measured environmental risk.
A. Adoption studies of latent G E The first evidence that genetic and environmental risks influence antisocial behavior together in a synergistic way came from adoption studies. Among the 6000 families of male adoptees in the Danish Adoption Study, 14% of adoptees were convicted of crime though neither their biological nor adoptive parents had been convicted, whereas 15% were convicted if their adoptive parent alone was convicted, 20% were convicted if their biological parent alone was convicted, and 25% were convicted if both biological and adoptive parents were convicted, although there were only 143 such cases (Mednick and Christiansen, 1977). This pattern of percentages did not represent a statistically significant cross‐over interaction term, but it did illustrate clearly that the effects of genetic
3. Genetic and Environmental Influences on Antisocial Behaviors
81
and environmental risk acting together were greater than the effects of either factor acting alone. The finding was buttressed by two studies from American and Swedish adoption registers completed about the same time (Cadoret et al., 1983; Cloninger et al., 1982).
B. Adoption studies of latent G measured E In a pool of 500 adoptees from the Iowa and Missouri adoption studies, adoptees had the most elevated antisocial behaviors when they experienced “adverse circumstances” in their adoptive homes as well as having birth mothers with antisocial personality problems or alcoholism (Cadoret et al., 1983). This landmark study documented that the interaction was statistically significant, and replicated across two independent samples. This finding was replicated and extended in another Iowa adoption cohort of 200 families (Cadoret et al., 1995). Adoptive parents’ adversity was defined according to the presence of marital problems, legal problems, substance abuse, or mental disorder and it interacted significantly with biological parents’ antisocial personality disorder to predict elevated rates of childhood aggression, adolescent aggression, and diagnosed conduct disorder in the adoptees. This Iowa adoption study was creatively analyzed to demonstrate that adversity in the adoptive home can moderate the genetic child‐effect in which children’s aggression provokes bad parenting (Riggins‐Caspers et al., 2003). Adoptees’ genetic liability for antisocial behavior (defined as biological parents’ psychopathology) provoked more harsh discipline from the adoptive parents in homes where the adoptive parents suffered adversity (marital, legal, substance, or psychopathology problems). The one problem in studying G E in adoption designs is the adoption breaks up the naturally occurring processes of rGE that characterize the nonadopted majority population, thereby precluding the possibility of G E (Stoolmiller, 1999). This separation allows the empirical study of G E, but paradoxically, it probably results in an underestimate of the influence of G E on antisocial outcomes. For this reason, adoption G E studies should be complemented with twin studies.
C. A twin study of latent G measured E Our E‐risk twin study also yielded evidence that genetic and environmental risks interact (Jaffee et al., 2005). Because we already knew that conduct problems were highly heritable in the E‐risk twin sample at age 5 years (Arseneault et al., 2003), we were able to estimate each child’s personal genetic risk for conduct problems by considering whether his or her co‐twin had already been diagnosed with conduct disorder, and whether he or she shared 100% versus 50% of genes with that diagnosed co‐twin. This method’s usefulness had been demonstrated previously in a landmark G E study showing that the risk of depression
82
Terrie E. Moffitt
following life‐event stress depends on genetic vulnerability (Kendler et al., 1995). For example, an individual’s genetic risk is highest if his or her co‐twin sibling already has a diagnosis of disorder and the pair is monozygotic. Likewise, an individual’s genetic risk is lowest if his or her co‐twin has been free from disorder and the pair is monozygotic. Individuals in dizygotic twin pairs fall between the high and low genetic risk groups. In this study, an interaction was obtained so that the effect of maltreatment on conduct problem symptoms was significantly stronger among children at high genetic risk than among children at low genetic risk. (Because there was no genetic child effect provoking maltreatment, the genetic risk groups did not differ on concordance for maltreatment or the severity of maltreatment). In addition, the experience of maltreatment was associated with an increase of 24% in the probability of diagnosable conduct disorder among children at high genetic risk, but an increase of only 2% among children at low risk.
D. Studies of measured G measured E; testing a measured gene The aforementioned adoption and twin studies established that genotype does interact with bad parenting in the etiological processes leading to antisocial behavior. However, the studies did not implicate any particular genes. One study was conducted to test the hypothesis of gene environment interaction using a measured environmental risk, child maltreatment, and an identified gene, the MAOA polymorphism (Caspi et al., 2002). We selected the MAOA gene as the candidate gene for our study for four reasons (Caspi et al., 2002). First, the gene encodes the MAOA enzyme, which metabolizes the neurotransmitters linked to maltreatment victimization and aggressive behavior by previous research. Second, drugs inhibiting the action of the MAO enzyme have been shown to prevent animals from habituating to chronic stressors analogous to maltreatment, and to dispose animals toward hyperreactivity to threat. Third, in studies of mice having the MAOA gene deleted, increased levels of neurotransmitters and aggressive behavior were observed, and aggression was normalized by restoring MAOA gene expression. Fourth, an extremely rare mutation causing a null allele at the MAOA locus was associated with aggressive psychopathology among some men in a Dutch family pedigree, although no relation between MAOA genotype and aggression had been detected for people in the general population. We selected maltreatment for this study for four reasons (Caspi et al., 2002). First, childhood maltreatment is a known predictor of antisocial outcomes. Second, not all maltreated children become antisocial, suggesting that vulnerability to maltreatment is influenced by heretofore unstudied individual characteristics. Third, our abovementioned twin research had established that
3. Genetic and Environmental Influences on Antisocial Behaviors
83
maltreatment’s effect on children’s aggression is environmentally mediated (i.e., the association is neither an artifact of a genetic child‐effect provoking maltreatment nor of transmission of aggression‐prone genes from parents). As such, maltreatment can serve as the environmental variable in a test of gene environment interaction. Fourth, animal and human studies suggest that maltreatment in early life alters neurotransmitter systems in ways that can persist into adulthood and can influence aggressive behavior. Based on this logic to support our hypothesis of G E, we measured childhood maltreatment history (8% severe, 28% probable, and 64% not maltreated) and MAOA genotype (37% low‐activity risk allele and 63% high‐ activity allele) in the 442 Caucasian males of the longitudinal Dunedin Multidisciplinary Health and Development Study. We found that maltreatment history and genotype interacted to predict four different measures of antisocial outcome; an adolescent diagnosis of conduct disorder, an age‐26 personality assessment of aggression, symptoms of adult antisocial personality disorder reported by informants who knew the study members well, and court conviction for violent crime up to age 26, the latest age of follow‐up. Among boys having the combination of the low‐MAOA‐activity allele and severe maltreatment, 85% developed some form of antisocial outcome. Males having the combination of the low‐activity allele and severe‐to‐probable maltreatment were only 12% of the male birth cohort, but they accounted for 44% of the cohort’s violent convictions because they offended at a higher rate on average than other violent offenders in the cohort. Replication of this study was of utmost importance, because the study reported the first instance of interaction between a measured gene and a measured environment in the behavioral sciences, and because reports of connections between measured genes and disorders are notorious for their poor replication record (Hamer, 2002). One initial positive replication, and extension, has emerged from the Virginia Twin Study for Adolescent Behavioral Development (Foley et al., 2004). This team studied 514 Caucasian male twins and measured environmental risk using an adversity index comprised of parental neglect; interparental violence, and inconsistent discipline. MAOA genotype and adversity interacted significantly so that 15% of boys having adversity, but the high‐MAOA‐activity allele developed conduct disorder, in comparison to 35% of boys having adversity plus the low‐ activity allele. This study went a step further, controlling for maternal antisocial personality disorder to rule out the possibility that passive rGE might have resulted in the co‐occurrence of environmental and genetic risk. This study thus replicated the original G E between the MAOA polymorphism and maltreatment, extended it to other forms of parental treatment, and showed that it is not an artifact of passive rGE.
84
Terrie E. Moffitt
E. Research implications of the nil main effect of the MAOA polymorphism on behavior One important finding from these two studies of a measured gene was that, in contrast to the G E interaction’s marked effects on antisocial outcomes, the unique effects of MAOA genotype apart from its role in the G E interaction were virtually nil. MAOA genotype was statistically unrelated to antisocial outcomes in the full cohorts, its effects were only revealed in the presence of maltreatment or adversity. Moreover, this pattern of a significant gene environment interaction in the presence of an initial nil main effect of the measured gene has now emerged from a number of other G E studies (Moffitt et al., in press). This pattern of nil main effects for measured genes appears to be widespread, and if so, it has an implication for gene hunters. Gene‐to‐disorder connections may be diluted across all the individuals in a sample if the connection is apparent only among individuals exposed to specific environmental risks. The expectation that simple direct paths will be found from gene to disease has not proven fruitful for complex psychiatric disorders—few linkage studies detect genes, many candidate gene association studies fail consistent replication, and genes that replicate account for little variation in the phenotype (Hamer, 2002). The MAOA G E finding suggests four guiding hypotheses for future genetics research. First, a major source of error in linkage pedigrees, incomplete gene penetrance, could occur if a gene’s effects are expressed only among family members exposed to environmental risk. Linkage studies should ascertain pedigree members’ environmental risk exposure. Second, candidate gene studies will not replicate each other if G E is operating and there are differences between research samples on risk exposure. Where possible candidate‐gene association studies should measure and take into account subjects’ environmental risk exposure. Third, most psychiatric genetics research, including genome‐wide scans, aims to identify genes having main effects (i.e., to find genes that show associations with behavior irrespective of the environment), but this main‐effects approach will not be efficient for detecting genes whose effects are conditional on environmental risk. (Interactions are independent of main effects, so main effects of risk factors are not a prerequisite for interactions between them.) Genome‐wide scans might be more powerful if gene hunters deliberately recruit samples selected for known exposure to environmental risks for the disorder they wish to study. Finally, the two studies of the MAOA polymorphism showed that when G E operates and risk exposure differs among participants within a sample, genes will account for little phenotypic variation and their effect sizes will be small to nil. To detect such small effects researchers have called for extremely large samples. Quantitative models of complex disorders having continuously distributed phenotypes have been interpreted to implicate many genes of small effect, but it is conceivable
3. Genetic and Environmental Influences on Antisocial Behaviors
85
that some of those “genes” might be environmental causes. In the Dunedin cohort, the G E accounted for a sufficient proportion of psychiatric outcome to suggest a provocative hypothesis, that some multifactorial disorders, instead of resulting from many genes of small effect, might result from relatively fewer genes whose effect sizes are conditional on exposure to environmental risks. For revealing the effects of such conditional‐effect genes, researchers will need to use strategic G E research.
F. Strategy for future G E studies using measured genes One can hope for careful, deliberate, and theory‐guided G E hypothesis testing of plausible triads of a candidate genetic polymorphism, a candidate environmental risk, and a behavioral phenotype. Elsewhere, we described in detail a research strategy to guide studies of measured gene–environment interaction (Moffitt et al., 2005). What follows is a brief summary. Step 1 is to consult quantitative genetic models of the behavior in question derived from twin and adoption research. The estimate of genetic influence (i.e., the A term) in part represents gene–environment interplay, as does the estimate of unique environmental influence (i.e., the E term) (Purcell, 2002). When these quantitative estimates are moderate‐to‐large, this encourages constructing hypotheses about potential G E interaction effects. Step 2 is to identify candidate environmental risks for the behavior in question. It is necessary to glean from the literature the candidate environmental risk factors having the strongest predictive efficacy for each disorder. Fortunately, for the study of antisocial behavior, a large pool of candidate environmental risk factors is available (Loeber and Farrington, 1998). The best candidate environmental risks are those having evidence of a plausible effect on biological systems involved in psychopathology (e.g., maltreatment, DeBellis, 2001). Once candidate risks have been identified, it is important to go a step further to test whether each candidate risk factor has effects that are actually environmentally mediated. Why must G E researchers prove environmental mediation? If an alleged environmental risk factor’s association with psychopathology is wholly genetically mediated, then a putative G E is really only an interaction between one specific gene and other unidentified genes. This article has described the several ways to test for environmental mediation. Step 3 is to measure the environmental risk as well as possible. Measuring environmental risk exposure precisely and reliably can be costly, but simulations show that reliable risk measurement can hugely enhance power to detect G E, thus reducing the need for large samples (Luan et al., 2001; Wong et al., 2003). Step 4 is to identify candidate susceptibility genes for a G E hypothesis. The temptation to name candidate genes associated with antisocial behavior was resisted in this article because gene detection advances so rapidly that any
86
Terrie E. Moffitt
list made now will be shortly outdated; by the time the article comes to press a list would feature disappointing replication failures and omit newly found hot possibilities (Insel and Collins, 2003). However, one can propose the following guidelines for choosing candidate genes are best for a G E hypothesis, as they emerge. First, seek empirical evidence that a gene has functional physiological significance in the brain (Tabor et al., 2002). Second, good candidate genes for G E will be those whose polymorphic variants are relatively common in the population. If a potentially damaging variant is maintained at a high prevalence rate, this might imply (but certainly does not guarantee) that natural selection has not eliminated the variant because it is only expressed under particular environmental conditions, or perhaps even because it confers advantage under particular conditions (Hill, 1999; Searle and Blackwell, 1999). From a more pragmatic point of view, common variants confer advantages of statistical power when testing interaction effects. As a third guideline for gene selection, if the gene has already been shown to have a replicated main‐effect association with the psychiatric disorder, it will be an easy candidate choice. However, it is very important to appreciate that the endeavor cannot rely on such rare replicated main‐effect associations because of the following paradox. Logically, if a gene’s effects are conditional on the environment; this will have the natural consequence of diminishing researchers’ capacity to detect a main effect! As a final guideline for step 4, the most sound logical basis for selecting a candidate gene for G E is evidence that the gene is related not to a disorder but rather to organisms’ responses to environmental risk. This evidence is necessary to frame a biologically plausible hypothesis that the gene moderates responses to an environmental risk (i.e., G E). As one example, the serotonin transporter polymorphism (5‐HTTLPR) is a good candidate for G E research into psychopathology because its two variants have been shown to affect physiological responsiveness to stressful environmental conditions in three experimental paradigms, including knockout mice (Murphy et al., 2001), stress‐reared R. macaques (Bennett et al., 2002), and a human functional neuro‐imaging paradigm (Hariri et al., 2002). Most evidence of connections between genes and risk‐responsiveness will emerge from studies of rodents and nonhuman primates having known human‐relevant genotypes, because nonhuman animals can be subjected to environmental risk manipulations under experimental control (Crabbe, 2003; Flint, 2003; Francis et al., 2003; Maxson, 2000; Meaney, 2001; Suomi, in press). A new wave of experimental investigations will ask if genotype influences human participants’ responsiveness to emotion‐eliciting stimuli or laboratory stress paradigms. These human studies will use psychophysiological phenotypes in G E experimental designs, such as electrodermal reactivity, or reactivity of the brain as measured by the electroencephalograph and by functional neuro‐imaging tools (Hariri et al., 2002). The results of such studies will provide an evidence base to nominate gene candidates in G E
3. Genetic and Environmental Influences on Antisocial Behaviors
87
hypotheses. Step 5 is to test for an interaction between the candidate gene and the environmental risk factor. The minimum design will begin with a pool of individuals exposed to environmental risk and hypothesize that genotype‐risk individuals develop more psychopathology than genotype‐controls (or the complement, beginning with individuals thought to be at genotypic risk and ascertaining whether individuals exposed to environmental risk develop more psychopathology than unexposed controls). The more informative design begins with a representative population‐based cohort. For example, in the case of dichotomous genotypic and environmental variables, groups would include (1) low genotypic risk and environmental risk to establish the baseline level of psychopathology associated with factors apart from our hypothesis, (2) high genotypic but low environmental risk to ascertain any effect of the gene in isolation, (3) high environmental but low genotypic risk to ascertain any effect of risk environment in isolation, and (4) high genotypic and environmental risk to ascertain whether their joint association with psychopathology is additive or interactive (for more discussion of design issues and statistical approaches see Moffitt et al., in press; Ottman, 1990; van Os and Sham, 2003; Yang and Khoury, 1997). Cohort designs allow us to report not only statistical significance but also to characterize the size of the G E effect in the population, which is prerequisite for evaluating the potential clinical validity and utility of a finding. Step 6 ensues if and only if the hypothesized G E interaction is obtained. Step 6 is to evaluate whether the resulting associations show specificity to the initially hypothesized triad of gene, environmental risk factor, and disorder or extend beyond that triad. Work at this step systematically ascertains whether the interaction holds when the gene is replaced with other relevant candidate genes, when the environmental risk is replaced with the disorders’ other risk factors, and when the disorder is replaced with other related disorder phenotypes. This step is exploratory. Whereas it is vital to frame a specific hypothesis of G E prior to analyzing any data, once an initial hypothesis has been tested in the affirmative, it is also responsible scientific practice to ascertain how far beyond the original hypothesis the G E may extend (Licinio, 2003). Step 7 is replication, which is particularly vital because of the known difficulty of detecting interaction terms between any two factors, including genes. Other kinds of research might also be stimulated by G E findings. Knowledge that an environmental risk factor has stronger connections with a disorder ought to likewise kick‐start new research into what brain mechanisms convert environmental experiences into the symptoms of psychopathology. Applied research might address the relevance of the G E for clinical diagnostics and therapeutics. The 2002 report that maltreatment and the MAOA polymorphism interacted to predict antisocial outcomes stimulated
88
Terrie E. Moffitt
investigations in ethics, legal scholarship, and even theology (e.g., Nuffield Council on Bioethics, 2002; Peters, 2003; Ross and Shestowsky, 2003; Sankar, 2003; Stone, 2003).
VII. THE WAY FORWARD Behavioral–genetic research should be a priority because it has real world authenticity. Genetic and environmental risks for antisocial behavior often coincide in the same families, and these risks are concentrated together in the same small segment of the population. Because of this fundamental fact, developmental processes that originate where genetic and environmental risks coincide are the most relevant causal processes to study. “Bio‐social” models of antisocial behavior made a good start at studying this co‐occurrence (Raine et al., 1997), but they did not tell the whole story because they did not disentangle environments from genes. Perhaps the most important point is that research that does not attack the co‐occurrence of genetic and environmental risks will have only limited relevance for prevention. In contrast, findings from studies of the coincidence of genetic and environmental risks can be generalized to the real world circumstances where interventionists usually find their clients. As such, research into gene–environment interplay will continue to prove critical in the future of research into antisocial behavior. Some recommendations follow. Behavior–genetic research into antisocial behavior should expand its focus beyond the current focus on young samples. If the aim is to explain the root etiology of serious and persistent antisocial behavior, then a research focus on childhood and the family environment is appropriate because that is when and where serious life‐course‐persistent antisocial behavior begins. However, there is far more to antisocial behavior that needs explaining. Gene–environment interplay research should embrace other risk factors, in other age periods, for example, the effects of peers on adolescents’ delinquent offending and the effects of substance abuse on adults’ domestic violence against partners. Behavioral–genetic research into the role of peers in antisocial behavior is well underway (e.g., Carey, 1992; Iervolino et al., 2002; Plomin, 1994; Rose et al., 2003; Rowe, 1985; Rowe and Osgood, 1984; Rowe et al., 1992). To my knowledge there is no behavioral–genetic research into domestic violence outcomes. Behavioral–genetic research into antisocial behavior should examine “endophenotypes.” These are phenotypic traits or markers thought to represent biologic systems underlying a behavioral disorder, and therefore assumed to be under greater genetic influence than the disorder itself (Gottesman and Gould, 2003). For some disorders, such as schizophrenia, attention is shifting from the search for connections between genes and the disorder to the search for
3. Genetic and Environmental Influences on Antisocial Behaviors
89
connections between genes and endophenotypes, such as eye tracking or working memory. This shift offers advantages in statistical power because endophenotypes are generally better distributed than severe disorders, and they can be studied in nonpatients. However, the promise of endophenotypes must be tempered by cautions that each “underlying biological” variable is as likely to be a consequence as cause and may well be subject to the same gene–environment interplay processes as are disorders themselves (as opposed to representing a purer genetic etiology). That said, endophenotype studies will be very useful for explaining how genes increase the probability that people will commit antisocial acts. One edited volume suggests a starting list of endophenoytpes for antisocial behavior–sensation‐seeking, over‐activity, fearlessness, low self‐ control, negative emotionality, callous‐unemotional style, weak verbal ability, poor memory, executive dysfunction, frontal lobe hypoarousal, serotonergic dysfunction, testosterone imbalance, and even large toddler body mass index (Lahey et al., 2003). Bringing these traits into research in gene–environment interplay involves several steps. First, they can be examined in quantitative twin studies to ascertain if they are under genetic influence. Second, an endophenotype can be entered with antisocial behavior into a quantitative bivariate model, to ascertain how much of the correlation between endophenotype and disorder arises from genes predisposing to both. Third, traditional mediation models can ascertain whether the endophenotype mediates the pathway between measured genes and antisocial outcomes. Behavioral–genetic research into antisocial behavior should be informed by findings from epigenetic studies. Although the DNA sequence is not itself altered by the environment, the science of epigenetic processes is revealing how environments can affect genes’ capacity to influence phenotypes (Pray, 2004; Varmuza, 2003). Theorists are putting forward conceptualizations of genes as dependent variables that can be “switched on or off” by nongenetic influences (Johnston and Edwards, 2002; Ridley, 2003), or genes as mediating variables that carry out developmental processes initiated at the level of the environment (Belsky, 1997; Gottlieb, 2003). For example, compelling experiments show how variation in quality of parental care can alter gene expression and behavior in rat offspring (Cameron et al., 2005; Meaney, 2001). This work has yet to be integrated into the study of antisocial behavior, but this is only a matter of time. Behavioral–genetic research into antisocial behavior should benefit more from animal models. Nonhuman animal models of behavioral disorders offer undeniable scientific advantages, but the world of animal research has remained somewhat apart from the world of psychological research into human antisocial behavior, primarily as a result of skepticism about the validity of animal models for human behavior. The most often‐used animal model in behavioral–genetic studies of aggression is the mouse intruder paradigm, in which the researcher introduces an intruder to the cage of the mouse subject
90
Terrie E. Moffitt
and measures the subject’s latency to attack. The validity question applies to such animal models of aggression because the aggression is appropriate in the wild, biologically adaptive, and highly stylized (Sluyter et al., 2003). Such animal behavior lacks strong parallels to the illegal behavior carried out by humans, which is considered inappropriate, is often maladaptive for the actor, and takes amazingly varied forms, involving force, but also fraud and theft committed by the same individual as part of an antisocial lifestyle. In contrast to this variation in the forms of human antisocial behavior, animals’ aggression to defend territory has been so strongly selected that there is little variation within a species. Moreover, animal researchers must exert great efforts to artificially create individual differences by enforcing assortative mating to breed strains that differ on aggressive behavior (Cairns et al., 1983). This stands in sharp contrast to the marked individual differences in aggression within the human population. Luckily, animal models of disorders are not necessary for making a contribution to future GE research. Instead, there is huge potential for developing new animal models of environmental risk mechanisms, to clarify the nature of G E interactions. Once an interaction between a gene and environmental risk is discovered in humans, uncovering the mechanisms behind it requires experimental inducement of environmental risk exposure, studies of the consequences for gene expression in brain tissue, and experimental manipulation of the genome. Such manipulations cannot be accomplished with human participants, but analogue methods are available, particularly in mice (Tecott and Wehner, 2001). The primary outcome measures in such research will be indicators of the animal’s physiological and neurological reactivity to environmental risks, and indicators of gene expression in tissues of animals exposed to environmental risks. Animal models of environmental risk will prove to be invaluable tools for unpacking many elements of gene–environment interplay. Behavioral–genetic research into antisocial behavior should incorporate treatment trials. Interventions are environments, and true randomized intervention trials are environments disentangled from any control by genetic influence. As such, harnessing interventions as environmental variables brings the power of experimental manipulation to the study of G E interaction with human subjects. Note that this research is well under way in the field of pharmacogenetics, which explores genetic individuality in drug response to improve the efficacy and safety of prescribing (Evans and Relling, 1999; Wolf et al., 2000). Given the known genetic influence on antisocial behavior, how far can interventions go to prevent the expression of genetic risk; just how powerful can the environment be when it is under our control? Integrating prevention research and behavioral–genetic research offers unprecedented opportunities to test etiological theories (Howe et al., 2002).
3. Genetic and Environmental Influences on Antisocial Behaviors
91
Research into genetic and environmental influences is making great strides toward uncovering the root causes of antisocial behavior. This article has reviewed the large group of quantitative behavioral–genetic studies leading to the conclusion that environmental and genetic causes are equally important for antisocial outcomes. Newer findings are also emerging. Studies are revealing which risk factors are environmental causes not just correlates. Studies are testing for effects of measured candidate genes. Studies are sorting out how our genotypes sway our susceptibility to environmental causes, and how our environments rule the behavioral expression of our genotypes. The future of this work is exciting.
Acknowledgments Work on this review was supported by grants from the US National Institute of Mental Health (MH45070 and MH49414), the UK Medical Research Council (G9806489 and G0100527), and a Royal Society‐Wolfson Research Merit Award. Helpful reviews of the article were provided by Louise Arseneault, Jay Belsky, Avshalom Caspi, Honalee Harrington, Julia Kim‐Cohen, Robert Plomin, Michael Rutter, Kali Trzesniewski, Alan Taylor, and Essi Viding. Technical assistance was provided by Helena Kiernan.
References Achenbach, T. M. (1991). “Manual for the Child Behavior Checklist.” University of Vermont Department of Psychiatry, Burlington, VT. Arseneault, L., Kim‐Cohen, J., Taylor, A., Caspi, A., and Moffitt, T. E. (2005). Psychometric evaluation of 5‐ and 7‐year‐old children’s self‐reports of conduct problems. J. Abnorm. Child Psychol. 33, 537–550. Arseneault, L., Moffitt, T. E., Caspi, A., Taylor, A., Rijsdijk, F. V., Jaffee, S., Ablow, J. C., and Measelle, J. R. (2003). Strong genetic effects on cross‐situational antisocial behavior among 5‐year‐old children according to mothers, teachers, examiner‐observers, and twins’self‐reports. J. Child Psychol. Psychiatry 44, 832–848. Asbury, K., Dunn, J., Pike, A., and Plomin, R. (2003). Nonshared environmental influences on individual differences in early behavioral development: A monozygotic twin differences study. Child Dev. 74, 933–943. Baker, L. A., Mack, W., Moffitt, T. E., and Mednick, S. A. (1989). Etiology of sex differences in criminal convictions in a Danish adoption cohort. Behav. Genet. 19, 355–370. Belsky, J. (1997). Variation in susceptibility to environmental influence: An evolutionary argument. Psychol. Inquiry 8, 230–235. Bennett, A. J., Lesch, K. P., Heils, A., Long, J. C., Lorenz, J. G., Shoaf, S. E., Champoux, M., Suomi, S. J., Linnoila, M. V., and Higley, J. D. (2002). Early experience and serotonin transporter gene variation interact to influence primate CNS function. Mol. Psychiatry 7, 188–122. Blair, R. J. R. (1995). A cognitive developmental approach to morality : Investigating the psychopath. Cognition 57, 1–29. Blonigen, D. M., Carlson, S. R., Krueger, R. F., and Patrick, C. J. (2003). A twin study of self‐ reported psychopathic personality traits. Pers. Individual Differences 35, 179–197.
92
Terrie E. Moffitt
Bohman, M., Cloninger, R., Sigvardsson, S., and von Knorring, A. L. (1982). Predisposition to petty criminality in Swedish adpotees. I. Genetic and environmental heterogeneity. Arch. Gen. Psychiatry 39, 1233–1241. Boomsma, D., Busjahn, A., and Peltonen, L. (2002). Classical twin studies and beyond. Nat. Rev. Genet. 3, 872–882. Bouchard, T. J., and McGue, M. (1990). Genetic and rearing environmental influences on adult personality: An analysis of adopted twins reared apart. J. Pers. 58, 263–292. Buchanan, A., Brock, D., Daniels, N., and Wikler, D. (2001). “From Chance to Choice: Genetics and Justice.” Cambridge University Press, New York. Burt, A. S., Krueger, R. F., McGue, M., and Iacono, W. (2003). Parent‐child conflict and the comorbidity among childhood externalizing disorders. Arch. Gen. Psychiatry 60, 505–513. Burt, A. S., Krueger, R. F., McGue, M., and Iacono, W. G. (2001). Sources of covariation among attention deficit‐hyperactivity disorder, oppositional defiant disorder, and conduct disorder: The importance of shared environment. J. Abnorm. Psychol. 110, 516–525. Butterfield, F. (1996). “All God’s Children: The Bosket Family and the American Tradition of Violence.” Avon, NY. Butterfield, F. (2002). Father steals best: Crime in an American family. August 21. The New York Times, from http://www.nytimes.com/2002/08/21/national/21FAMI.html. Retrieved 1 September 2002. Cadoret, R. J. (1978). Psychopathology in adopted‐away offspring of biological parents with antisocial behavior. Arch. Gen. Psychiatry 35, 176–184. Cadoret, R. J., Cain, C. A., and Crowe, R. R. (1983). Evidence for gene‐environment interaction in the development of adolescent antisocial behavior. Behav. Genet. 13, 301–310. Cadoret, R. J., Yates, W. R., Troughton, E., Woodworth, G, and Stewart, M. A. S. (1995). Genetic‐ environmental interaction in the genesis of aggressivity and conduct disorders. Arch. Gen. Psychiatry 52, 916–924. Cairns, R. B., and Cairns, B. D. (1994). “Lifelines and Risks: Pathways of Youth in Our Time.” Cambridge University Press, Cambridge. Cairns, R. B., MacCombie, D. J., and Hood, K. E. (1983). A developmental‐genetic analysis of aggressive behavior in mice. I. Behavioral outcomes. J. Comp. Psychol. 97, 69–89. Cameron, N. M., Parent, C., Champagne, F. A. Fish, E. W., Ozaki‐Kuroda, K., and Meaney, M. J. (2005). The programming of individual differences in defensive responses and reproductive strategies in the rat through variations in maternal care. Neurosci. Biobehav. Rev. 29, 843–865. Carey, G. (1992). Twin imitation for antisocial behavior. J. Abnorm. Psychol. 101, 18–25. Carey, G. (1994). Genetics and violence. In “Understanding and Preventing Violence: Vol. 2, Biobehavioral Influences” (A. J. Reiss Jr., K. A. Miczek, and J. A. Roth, eds.), pp. 21–58. National Academy Press, Washington, DC. Carey, G. (2003). “Human Genetics for the Social Sciences.” Sage Publications, Thousand Oaks, CA. Carey, G., and Goldman, D. (1997). The genetics of antisocial behavior. In “Handbook of Antisocial Behavior” (D. M. Stoff, J. Breiling, and J. D. Maser, eds.), pp. 243–254. Wiley, New York. Caspi, A, Taylor, A, Moffitt, T. E., and Plomin, R. (2000). Neighborhood deprivation affects children’s mental health: Environmental risks identified using a genetic design. Psychol. Sci. 11, 338–342. Caspi, A., McClay, J, Moffitt, T. E., Mill, J., Martin, J., Craig, I., Taylor, A., and Poulton, R. (2002). Evidence that the cycle of violence in maltreated children depends on genotype. Science 297, 851–854. Caspi, A., Moffitt, T. E., Morgan, J., Rutter, M., Taylor, A., Arseneault, L., Tully, L., Jacobs, C., Kim‐Cohen, J., and Polo‐Tomas, M. (2004). Maternal expressed emotion predicts children’s
3. Genetic and Environmental Influences on Antisocial Behaviors
93
antisocial behavior problems: Using MZ‐twin differences to identify environmental effects on behavioral development. Dev. Psychol. 40, 149–161. Cates, D. S., Houston, B. K., Vavak, C. R., Crawford, M. H., and Uttley, M. (1993). Heritability of hostility‐related emotions, attitudes, and behaviors. J. Behav. Med. 16, 237–256. Centerwall, B. S., and Robinette, D. (1989). Twin concordance for dishonourable discharge from the military: With a review of the genetics of antisocial behavior. Comp. Psychiatry 30, 442–446. Cherney, S. S., Fulker, D. W., and Hewitt, J. K. (1997). Cognitive development from infancy to middle childhood. In “Intelligence, Heredity, and Environment” (R. J. Sternberg and E. L. Grigorenko, eds.), pp. 463–482. Cambridge University Press, Cambridge. Cicchetti, D. (2003). Experiments of nature: Contributions to developmental theory. Dev. Psychopathol. 15, 833–835. Cloninger, C. R., and Gottesman, I. I. (1987). Genetic and environmental factors in antisocial behavior disorders. In “The Causes of Crime: New biological Approaches” (S. A. Mednick, T. E. Moffitt, and S. A. Stack, eds.), pp. 92–109. Cambridge University Press, New York. Cloninger, C. R., Christiansen, K. O., Reich, T., and Gottesman, I. I. (1978). Implications of sex differences in the prevalences of antisocial personality, alcoholism, and criminality for familial transmission. Arch. Gen. Psychiatry 35, 941–951. Cloninger, C. R., Sigvardsson, S., Bohman, M., and von Knorring, A. L. (1982). Predisposition to petty criminality in Swedish adoptees. II. Cross‐fostering analysis of gene‐environment interaction. Arch. Gen. Psychiatry 39, 1242–1247. Coccaro, E. F., Bergeman, C., Kavoussi, R. J., and Seroczynski, A. D. (1997). Heritability of aggression and irritability: A twin study of the Buss‐Durkee Aggression Scales in adult male subjects. Biol. Psychiatry 41, 273–284. Coid, B., Lewis, S. W., and Reveley, A. M. (1993). A twin study of psychosis and criminality. Br. J. Psychiatry 162, 87–92. Collins, W. A., Maccoby, E. E., Steinberg, L., Hetherington, E. M., and Bornstein, M. H. (2000). Contemporary research on parenting. Am. Psychol. 55, 218–232. Collishaw, S., Maughan, B., Goodman, R., and Pickles, A. (2004). Time trends in adolescent mental health. J. Child Psychol. Psychiatry 45, 1350–1362. Costello, E. J., Compton, S. N., Keeler, G., and Angold, A. (2003). Relationships between poverty and psychopathology: A natural experiment. JAMA 290, 2023–2029. Crabbe, J. C. (2003). Finding genes for complex behaviors: Progress in mouse models of the addictions. In “Behavioral Genetics in the Postgenomic Era” (R. Plomin, J. C. DeFries, I. W. Craig, and P. McGuffin, eds.), pp. 291–308. American Psychological Association, Washington, DC. Deater‐Deckard, K., and Plomin, R. (1999). An adoption study of the etiology of teacher and parent reports of externalizing behavior problems in middle childhood. Child Dev. 70, 144–154. Deater‐Deckard, K., Reiss, D., Heatherington, E. M., and Plomin, R. (1997). Dimensions and disorders of adolescent adjustment: A quantitative genetic analysis of unselected samples and selected extremes. J. Child Psychiatry Psychol. 38, 515–525. DeBellis, M. D. (2001). Developmental traumatology: The psychobiological development of maltreated children and its implications for research, treatment, and policy. Dev. Psychopathol. 13, 539–564. Dick, D. M., and Rose, R. J. (2002). Behavior Genetics. What’s new? What’s next? Curr. Dir. Psychol. Sci. 11, 70–74. Di Lalla, D.L, Gottesman, I. I., Carey, G., and Bouchard, T. J., Jr. (1996). Heritability of MMPI personality indicators of psychopathology in twins reared apart. J. Abnorm. Psychol. 105, 491–499. DiLalla, L. F. (2002). Behavior Genetics of aggression in children: Review and future directions. Dev. Rev. 22, 593–622.
94
Terrie E. Moffitt
DiLalla, L. F., and Gottesman, I. I. (1989). Heterogeneity of causes for delinquency and criminality: Lifespan perspectives. Dev. Psychopathol. 1, 339–349. DiLalla, L. F., and Gottesman, I. I. (1991). Biological and genetic contributions to violence: Widom’s untold tale. Psychol. Bull. 109, 125–129. Dionne, G., Tremblay, R., Boivin, M., Laplante, D., and Perusse, D. (2003). Physical aggression and expressive vocabulary in 19‐month‐old twins. Dev. Psychol. 39, 261–273. Dunne, M. P., Statham, D. J., Slutske, W. S., Dinwiddie, S. H., Bucholz, K. K., Madden, P. A. F, and Heath, A. C. (1997). Genetic and environmental contributions to variance in age at first sexual intercourse. Psychol. Sci. 8, 211–216. Eaves, L. J., Silberg, J. L., Meyer, J. M., Maes, H. H., Simonoff, E., Pickles, A., Rutter, M., Neale, M. C., Reynolds, C., Erickson, M. T., Heath, A. C., Loeber, R., Truett, K. R., and Hewitt, J. K. (1997). Genetics and developmental psychopathology: 2. The main effects of genes and environment on behavioral problems in the Virginia study of adolescent behavioral development. J. Child Psychol. Psychiatry 38, 965–980. Edelbrock, C., Rende, R., Plomin, R., and Thompson, L. A. (1995). A twin study of competence and problem behavior in childhood and early adolescence. J. Child Psychol. Psychiatry 36, 775–785. Eley, T. C., Lichtenstein, P., and Moffitt, T. E. (2003). A longitudinal analysis of the etiology of aggressive and non‐aggressive antisocial behavior. Dev. Psychopathol. 15, 383–402. Eley, T. C., Lichtenstein, P., and Stevenson, J. (1999). Sex differences in the etiology of aggressive and non‐aggressive antisocial behavior: Results from two twin studies. Child Dev. 70, 155–168. Evans, W. E., and Relling, M. V. (1999). Pharmacogenetics: Translating functional genomics into rational therapeutics. Science 286, 487–491. Farrington, D. P., Barnes, G. C., and Lambert, S. (1996). The concentration of offending in families. Legal Criminol. Psychol. 1, 47–63. Farrington, D. P., Jolliffe, D., Loeber, R., Stouthamer‐Loeber, M., and Kalb, L. (2001). The concentration of offenders in families, and family criminality in the prediction of boys’ delinquency. J. Adolesc. 24, 579–596. Fergusson, D. M., Lynskey, M., and Horwood, L. J. (1995). The adolescent outcomes of adoption: A 16‐year longitudinal study. J. Child Psychol. Psychiatry, 36, 597–615. Finkel, D., and McGue, M. (1997). Sex differences and nonadditivity in the heritability on the Multidimensional Personality Questionnaire scales. J. Pers. Soc. Psychol. 72, 929–938. Flint, J. (2003). Animal models of anxiety. In “Behavioral Genetics in the Postgenomic Era” (R. Plomin, J. C. DeFries, I. W. Craig, and P. McGuffin, eds.), pp. 425–442. American Psychological Association, Washington, DC. Foley, D, Wormley, B., Silberg, J., Maes, H., Hewitt, J., Eaves, L., and Riley, B. (2004). Childhood adversity, MAOA genotype, and risk for conduct disorder. Arch. Gen. Psychiatry 61, 738–744. Francis, D., Insel, T., Szegda, K., Campbell, G, and Martin, W. D. (2003). Epigenetic sources of behavioral differences in mice. Nat. Neurosci. 6, 445–446. Galbaud du Fort, G., Boothroyd, L. J., Bland, R. C., Newman, S. C., and Kakuma, R. (2002). Spouse similarity for antisocial behavior in the general population. Psychol. Med. 32, 1407–1416. Ge, X. (2002). Personal communication April. Gjone, H., and Novik, T. S. (1995). Parental ratings of behavior problems: A twin and general population comparison. J. Child Psychol. Psychiatry 36, 1213–1224. Gjone, H., and Stevenson, J. (1997a). The association between internalizing and externalizing behavior in childhood and early adolescence: Genetic or environmental common influences? J. Abnorm. Child Psychol. 25, 277–286. Gjone, H., and Stevenson, J. (1997b). A longitudinal twin study of temperament and behavior problems: Common genetic or environmental influences? J. Am. Acad. Child Adolesc. Psychiatry 36, 1448–1456.
3. Genetic and Environmental Influences on Antisocial Behaviors
95
Glover, J. (1996). The implications for responsibility of possible genetic factors in the explanation of violence. In “Symposium on Genetics of Criminal and Antisocial Behavior, Ciba Foundation 194” (G. R. Bock and J. A. Goode, eds.), pp. 237–247. Wiley, Chichester, UK. Goldstein, R. B., Prescott, C. A., and Kendler, K. S. (2001). Genetic and environmental factors in conduct problems and adult antisocial behavior among adult female twins. J. Nervous Mental Dis. 189, 201–209. Gottesman, I. I. (1966). Genetic variance in adaptive personality traits. J. Child Psychol. Psychiatry 7, 199–208. Gottesman, I. I., and Gould, T. D. (2003). The endophenotype concept in psychiatry: Etymology and strategic intentions. Am. J. Psychiatry 160, 636–645. Gottesman, I. I., Goldsmith, H. H., and Carey, G. (1997). A developmental and a genetic perspective on aggression. In “Uniting psychology and Biology: Integrative Perspectives on Human Development” (N. L. Segal and G. E. Weisfeld, eds.), pp. 107–130. American Psychological Association, Washington, DC. Gottlieb, G. (2003). On making behavioral genetics truly developmental. Hum. Develop. 46, 337–355. Grove, W. M., Eckert, E. D., Heston, L., and Bouchard, T. J. (1990). Heritability of substance abuse and antisocial behavior: A study of monozygotic twins reared apart. Biol. Psychiatry 27, 1293–1304. Hamer, D. (2002). Rethinking behavior genetics. Science 298, 71–72. Hariri, A. R., Mattay, V. S., Tessitore, A., Kolachana, B., Fera, F., Goldman, D., Egan, M. F., and Weinberger, D. R. (2002). Serotonin transporter genetic variation and the response of the human amygdala. Science 297, 400–404. Harlow, C. W. (2003). Education and corrections populations.. Bureau of Justice Statistics Special Report (NCJ 195670). U. S. Dept. of Justice, Washington, DC. Available at: http://www.ojp. usdoj.gov/bjs/ Harris, J. R. (1998). “The Nurture Assumption.” Free Press, New York. Hart, S. D., and Hare, R. D. (1997). Psychopathy: Assessment and association with criminal conduct. In “Handbook of Antisocial Behavior” (D. M. Stoff, J. Breiling, and J. D. Maser, eds.), pp. 22–35. Wiley, New York. Hicks, B. M., Krueger, R. F., Iacono, W. G., McGue, M., and Patrick, C. J. (2004). Family transmission and heritability of externalizing disorders. Arch. Gen. Psychiatry 61, 922–928. Hill, A. V. S. (1999). Genetics and genomics of infectious disease susceptibility. Br. Med. Bull. 55, 401–413. Hopper, J. L. (1998). Heritability. In “The Encyclopedia of Biostatistics,” Vol. 3, pp. 1905–1906. Wiley, New York. Hopper, J. L. (1999). Why “common environmental affects” are so uncommon in the literature. In “Advances in Twin and Sib Pair Analysis” (T. Spector, H. Sneider, and A. MacGregor, eds.), pp. 152–165. Greenwich Medical Media, London. Howe, G. W., Reiss, D., and Yuh, J. (2002). Can prevention trials test theories of etiology? Dev. Psychopathol. 14, 673–694. Hudziak, J. J., Rudiger, L. P., Neale, M. C., Heath, A. C., and Todd, R. D. (2000). A twin study of inattentive, aggressive, and anxious/depressed behaviors. J. Am. Acad. Child Adolesc. Psychiatry 39, 469–476. Hughes, C., Oksanen, H., Taylor, A., Jackson, J., Murray, L., Caspi, A., and Moffitt, T. E. (2002). “I’m gonna beat you, SNAP!” An observational paradigm for assessing disruptive behavior in young children. J. Child Psychol. Psychiatry 43, 507–516. Hutchings, B., and Mednick, S. A. (1973). Genetic and environmental influences on criminality in Denmark. In “International Collaboration in Mental Health” (B. S. Brown and E. F. Torrey, eds.), pp. 113–123. National Institute of Mental Health, Rockville, MD.
96
Terrie E. Moffitt
Iervolino, A. C., Pike, A., Manke, B., Reiss, D., Hetherington, E. M., and Plomin, R. (2002). Genetic and environmental influences in adolescent peer socialization: Evidence from two genetically sensitive designs. Child Dev. 73, 162–174. Insel, T. R., and Collins, F. S. (2003). Psychiatry in the genomics era. Am. J. Psychiatry 160, 616–620. Jacobson, K. C., Prescott, C. A., and Kendler, K. S. (2000). Genetic and environmental influences on juvenile antisocial behavior assessed on two occasions. Psychol. Med. 30, 1315–1325. Jacobson, K. C., Prescott, C. A., and Kendler, K. S. (2002). Sex differences in the genetic and environmental influences on the development of antisocial behavior. Dev. Psychopathol. 14, 395–416. Jacobson, K. C., Prescott, C. A., Neale, M. C., and Kendler, K. S. (2000). Cohort differences in genetic and environmental influences on retrospective reports of conduct disorder among adult male twins. Psychol. Med. 30, 775–787. Jacobson, K. C., Neale, M. C., Prescott, C. A., and Kendler, K. S. (submitted for publication). Behavioral genetic confirmation of a life‐course perspective on antisocial behavior. Presentation at the 2001 Behavior Genetics Association, Cambridge. Available from
[email protected]. vcu.edu Jaffee, S. R., Caspi, A., Moffitt, T. E., Polo‐Tomas, M., Price, T. S., and Taylor, A. (2004a). The limits of child effects: Evidence for genetically‐mediated child effects on corporal punishment, but not on maltreatment. J. Dev. Psychol. 40(6), 1047–1058. Jaffee, S. R., Caspi, A., Moffitt, T. E., and Taylor, A. (2004b). Physical maltreatment victim to antisocial child: Evidence of an environmentally‐mediated process. J. Abnorm. Psychol. 113, 44–55. Jaffee, S. R., Moffitt, T. E., Caspi, A., Taylor, A., and Arseneault, L. (2002). The influence of adult domestic violence on children’s internalizing and externalizing problems: An environmentally‐ informative twin study. J. Am. Acad. Child Adolesc. Psychiatry 41, 1095–1103. Jaffee, S. R., Caspi, A., Moffitt, T. E., Dodge, K., Rutter, M., and Taylor, A., Tully, L. (2005). Nature x Nurture: Genetic vulnerabilities interact with physical maltreatment to promote behavior problems. Dev. Psychopathol. 17, 67–84. Jang, K. L., Vernon, P. A., and Livesley, W. J. (2000). Personality disorder traits, family environment, and alcohol misuse: A multivariate behavioral genetic analysis. Addiction 95, 873–888. Johnson, W., Krueger, R. F., Bouchard, T. J., and McGue, M. (2002). The personalities of twins: Just ordinary folks. Twin Res. 5, 125–131. Johnston, T. D., and Edwards, L. (2002). Genes, interactions, and the development of behavior. Psychol. Rev. 109, 26–34. Jorgensen, A. L., Phillip, J., Raskind, W. H., Matsushita, M., Christensen, B., Dreyer, V., and Motulsky, A. A. (1992). Different patterns of X inactivation in MZ twins discordant for red‐ green color vision deficiency. Am. J. Hum. Genet. 51, 291–298. Kendler, K. S. (1993). Twin studies of psychiatric illness. Arch. Gen. Psychiatry 50, 905–915. Kendler, K. S. (1996). Parenting: A genetic epidemiologic perspective. Am. J. Psychiatry 153, 11–20. Kendler, K. S. (2001). Twin studies of psychiatric illness: An update. Arch. Gen. Psychiatry 58, 1005–1014. Kendler, K. S., Kessler, R. C., Walters, E. E., MacLean, C., Neale, M. C., Heath, A. C., and Eaves, L. J. (1995). Stressful life events, genetic liability and onset of an episode of major depression in women. Am. J. Psychiatry 152, 833–842. Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., and Eaves, L. J. (1994). Parental treatment and the equal environments assumption in twin studies of psychiatric illness. Psychol. Med. 24, 579–590.
3. Genetic and Environmental Influences on Antisocial Behaviors
97
Kendler, K. S., Prescott, C. A., Myers, J., and Neale, M. C. (2003). The structure of genetic and environmental risk factors for common psychiatric and substance use disorders in men and women. Arch. Gen. Psychiatry 60, 929–937. Kim‐Cohen, J., Caspi, A., Moffitt, T. E., Harrington, H. L., Milne, B., and Poulton, R. (2003). Prior juvenile diagnoses in adults with mental disorder: Developmental follow‐back of a prospective‐ longitudinal cohort. Arch. Gen. Psychiatry 60, 709–717. Kraemer, H. C. (2003). Current concepts of risk in psychiatric disorders. Curr. Opin. Psychiatry 16, 421–430. Krueger, R. F., Hicks, B. M., and McGue, M. (2001). Altruism and antisocial behavior: Independent tendencies, unique personality correlates, distinct etiologies. Psychol. Sci. 12, 397–402. Krueger, R. F., Hicks, B. M., Patrick, C. J., Carlson, S. R., Iacono, W. G., and McGue, M. (2002). Etiologic connections among substance dependence, antisocial behavior, and personality: Modeling the externalizing spectrum. J. Abnorm. Psychol. 111, 411–424. Krueger, R. F., Moffitt, T. E., Caspi, A., Bleske, A., and Silva, P. A. (1998). Assortative mating for antisocial behavior: Developmental and methodological implications. Behav. Genet. 28, 173–186. Kuntsi, J., Gayan, J., and Stevenson, J. (2000). Parents’ and teachers’ ratings of problem behaviors in children: Genetic and contrast effects. Twin Res. 3, 251–258. Lahey, B., Moffitt, T. E., and Caspi, A. (eds.) (2003). “Causes of Conduct Disorder and Juvenile Delinquency.” Guilford Press, New York. Lambert, E. W., Wahler, R. G., Andrade, A. R., and Bickman, L. (2001). Looking for the disorder in conduct disorder. J. Abnorm. Psychol. 110, 110–123. Levy, F., Hay, D., McLaughlin, M., Wood, C., and Waldman, I. (1996). Twin‐sibling differences in parental reports of ADHD, speech, reading, and behavioral problems. J. Child Psychol. Psychiatry 37, 569–578. Licinio, J. (2003). Editorial. Molecular psychiatry: Moving away from one‐polymorphism genetic association studies. Mol. Psychiatry 8, 247. Livesley, W. J., Jang, K. L., Jackson, D. N., and Vernon, P. A. (1993). Genetic and environmental contributions to dimensions of personality disorder. Am. J. Psychiatry 150, 1826–1831. Loat, C. S., Asbury, K., Galsworthy, M. J., Plomin, R., and Craig, I. W. (2004). X inactivation as a source of behavioral differences in monozygotic female twins. Twin Res. 7, 54–61. Loeber, R., and Farrington, D. (1998). “Serious and Violent Juvenile Offenders: Risk factors and successful Interventions.” Sage Publications, Thousand Oaks, CA. Loehlin, J. C., and Nichols, R. C. (1976). “Heredity, Environment, and Personality: A Study of 850 Sets of Twins.” University of Texas Press, Austin. Loehlin, J. C., Willerman, L., and Horn, J. (1985). Personality resemblances in adoptive families when the children are late‐adolescent or adult. J. Pers. Soc. Psychol. 48, 376–392. Loehlin, J. C., Willerman, L., and Horn, J. (1987). Personality resemblance in adoptive families: A 10‐year follow‐up. J. Pers. Soc. Psychol. 53, 961–969. Luan, J. A., Wong, M. Y., Day, N. E., and Wareham, N. J. (2001). Sample size determination for studies of gene‐environment interaction. Int. J. Epidemiol. 30, 1035–1040. Lynam, D. R. (1996). Early identification of chronic offenders: Who is the fledgling psychopath? Psychol. Bull. 120, 209–234. Lyons, M. J., True, W. R., Eisen, S. A., Goldberg, J., Meyer, J. M., Faraone, S. V., Eaves, L. J., and Tsuang, M. T. (1995). Differential heritability of adult and juvenile antisocial traits. Arch. Gen. Psychiatry 52, 906–915. Maestripieri, D. (2003). Similarities in affiliation and aggression between cross‐fostered rhesus macaque females and their biological mothers. Dev. Psychobiol. 43, 321–327. Mare, R. D. (1991). Five decades of educational assortative mating. Am. Sociol. Rev. 56, 15–32. G.M.Martin (ed.) (2002). Twin registers across the globe [Special issue]. Twin Res. 5(5).
98
Terrie E. Moffitt
Martin, N., Boomsma, D., and Machin, G. (1997). A twin‐pronged attacked on complex traits. Nat. Genet. 17, 387–392. Martin, N. G., Eaves, L. J., Kearsey, M. J., and Davies, P. (1978). The power of the classical twin study. Heredity 40, 97–116. Maxson, S. C. (2000). Genetic influences on aggressive behavior. In “Genetic Influences on Neural Behavioral Functions” (D. W. Pfaff, W. H. Berrettini, H. J. Toh, and S. C. Maxson, eds.), pp. 405–415. CRC Press, New York. McGue, M., Bouchard, T. J., Jr., Iacono, W. G., and Lykken, D. T. (1993). Behavioral genetics of cognitive ability: A life‐span perspective. In “Nature, Nurture and Psychology” (R. Plomin and G. E. McClearn, eds.), pp. 59–76. American Psychological Association, Washington, DC. McGue, M., Sharma, A., and Benson, P. (1996). The effect of common rearing on adolescent adjustment: Evidence from a U. S. adoption cohort. Dev. Psychol. 32, 604–613. McGuffin, P., and Thapar, A. (1997). The genetic basis of bad behavior in adolescents. Lancet 350, 411–412. McGuffin, P., and Thapar, A. (1998). Genetics and antisocial personality disorder. In “Psychopathy: Antisocial, Criminal and Violent Behavior” (T. Millon and E. Simonsen, eds.), pp. 215–230. Guilford Press, New York. McGuffin, P., Riley, B., and Plomin, R. (2001). Toward behavioral genomics. Science 291, 1232–1234. Meaney, M. J. (2001). Maternal care, gene expression, and the transmission of individual differences in stress reactivity across generations. Ann. Rev. Neurosci. 24, 1161–1192. Mednick, S. A., and Christiansen, K. O. (1977). “Biosocial Bases of Criminal Behavior.” Gardner Press, New York. Mednick, S. A., Gabrielli, W. F., and Hutchings, B. (1984). Genetic factors in criminal behavior: Evidence from an adoption cohort. Science 224, 891–893. Mednick, S. A., Moffitt, T. E., Gabrielli, W. F., and Hutchings, B. (1986). Genetic factors in criminal behavior: A review. In “The Development of Antisocial and Prosocial Behavior” ( J. Block, D. Olweus, and M. R. Yarrow, eds.), pp. 33–50. Academic Press Inc., New York, NY. Meyer, J. M., Rutter, M., Silberg, J. L., Maes, H., Simonoff, E., Shillady, L. L., Pickles, A., Hewitt, J. K., and Eaves, L. J. (2000). Familial aggregation for conduct disorder symptomatology: The role of genes, marital discord, and family adaptability. Psychol. Med. 30, 759–774. Miech, R. A., Caspi, A., Moffitt, T. E., Wright, B. E., and Silva, P. A. (1999). Low socio‐economic status and mental disorders: A longitudinal study of causation and selection. Am. J. Sociol. 104, 1096–1131. Miles, D. R., and Carey, G. (1997). Genetic and environmental architecture of human aggression. J. Pers. Soc. Psychol. 72, 207–217. Miles, D. R., van den Bree, M. B. M., and Pickens, R. W. (2002). Sex differences in shared genetic and environmental influences between conduct disorder symptoms and marijuana use in adolescents. Neuropsychiatric Genet. 114, 159–168. Moffitt, T. E. (1993). “Life‐course‐persistent” and “adolescence‐limited” antisocial behavior: A developmental taxonomy. Psychol. Rev. 100, 674–701. Moffitt, T. E. (2003). Life‐course persistent and adolescence‐limited antisocial behavior: A 10‐year research review and a research agenda. In “The Causes of Conduct Disorder and Serious Juvenile Delinquency” (B. Lahey, T. E. Moffitt, and A. Caspi, eds.), pp. 49–75. Guilford, New York. Moffitt, T. E., and the E‐risk Study, Team (2002). Teen‐aged mothers in contemporary Britain. J. Child Psychol. Psychiatry 43, 1–16. Moffitt, T. E., Caspi, A., and Rutter, M. (2005). Strategy for investigating interaction between measured genes and measured environments. Arch. Gen. Psychiatry 62, 473–481.
3. Genetic and Environmental Influences on Antisocial Behaviors
99
Moffitt, T. E., Caspi, A., and Rutter, M. (in Press). Interaction between measured genes and measured environments in psychopathology: Concepts, methods, and implications for intervention, research, and public understanding of genetics. Perspectives of Psychological Science, in press. Moffitt, T. E., Caspi, A., Harrington, H., and Milne, B. (2002). Males on the life‐course persistent and adolescence‐limited antisocial pathways: Follow‐up at age 26. Dev. Psychopathol. 14, 179–206. Moffitt, T. E., Caspi, A., Rutter, M., and Silva, P. A. (2001). “Sex Differences in Antisocial Behavior: Conduct Disorder, Delinquency, and Violence in the Dunedin Longitudinal Study.” Cambridge University Press, Cambridge, UK. Moffitt, T. E., Krueger, R. F., Caspi, A., and Fagan, J. (2000). Partner abuse and general crime: How are they the same? How are they different? Criminology 38, 201–235. Moilannen, I., Linna, S. L., Ebeling, H., Kampulainen, K., Tamminen, T., Piha, J., and Almqvist, F. (1999). Are twins’ behavioral/emotional problems different from singletons’? Eur. Child Adolesc. Psychiatry 8, 62–67. Murphy, D. L., Li, Q, Engel, S., Wichems, C., Andrews, A., Lesch, K. P., and Uhl, G. (2001). Genetic perspectives on the serotonin transporter. Brain Res. Bull. 56, 487–494. Nadder, T. S., Rutter, M., Silberg, J. L., Maes, H. H., and Eaves, L. J. (2002). Genetic effects on the variation and covariation of attention‐deficit hyperactivity disorder (ADHD) and oppositional‐ defiant disorder/conduct disorder (ODD/CD) symptomatologies across informant and occasion of measurement. Psychol. Med. 32, 39–53. Nagin, D. S. (1999). Analyzing developmental trajectories: A semiparametric, group‐based approach. Psychol. Methods 4, 139–157. Nagin, D. S., Farrington, D. P., and Moffitt, T. E. (1995). Life‐course trajectories of different types of offenders. Criminology 33, 111–139. Nagin, D. S., Pagani, L., Tremblay, R. E., and Vitaro, F. (2003). Life course turning points: The effect of grade retention on physical aggression. Dev. Psychopathol. 15, 343–361. Neale, M. C., and Cardon, L. R. (1992). “Methodology for Genetic Studies of Twins and Families.” Kluwer Academic Publishers, Dordrecht, The Netherlands. Newman, D. L., Moffitt, T. E., Caspi, A., Magdol, L., and Silva, P. A. (1996). Psychiatric disorder in a birth cohort of young adults: Prevalence, co‐morbidity, clinical significance, and new case incidence from age 11 to 21. J. Consult. Clin. Psychol. 64, 552–562. Nuffield, Council on Bioethics (2002). “Genetics and Human Behavior: The Ethical Context.“ Author, London. Report also available from www.nuffieldbioethics.org O’Connor, M., Foch, T., Sherry, T., and Plomin, R. (1980). A twin study of specific behavioral problems of socialization as viewed by parents. J. Abnorm. Child Psychol. 8, 189–199. O’Connor, T. G., Deater‐Deckard, K., Fulker, D., Rutter, M., and Plomin, R. (1998a). Genotype‐ environment correlations in late childhood and early adolescence: Antisocial behavioral problems in coercive parenting. Dev. Psychol. 34, 970–981. O’Connor, T. G., McGuire, S., Reiss, D., Hetherington, E. M., and Plomin, R. (1998b). Co‐ occurrence of depressive symptoms and antisocial behavior in adolescence: A common genetic liability. J. Abnorm. Psychol. 107, 27–37. O’Connor, T. G., Neiderhiser, J. M., Reiss, D., Hetherington, E. M., and Plomin, R. (1998c). Genetic contributions to continuity, change, and co‐occurrence of antisocial and depressive symptoms in adolescence. J. Child Psychol. Psychiatry 39, 323–336. O’Connor, T. G., Heatherington, E. M., Reiss, D., and Plomin, R. (1995). A twin‐sibling study of observed parent‐adolescent relations. Child Dev. 66, 812–829. Ottman, R. (1990). An epidemiologic approach to gene‐environment interaction. Genet. Epidemiol. 7, 177–185.
100
Terrie E. Moffitt
Partanen, J., Bruun, K., and Markannen, T. (1966). “Inheritance of Drinking Behavior: A Study on Intelligence, Personality, and Use of Alcohol of Adult Twins.” Keskuskirjapaino, Helsinki, Finland. Peters, T. (2003). Holy Therapy: Can a drug do the work of the spirit? Christian Century 9, August, 23–26. Pike, A., Reiss, D., Hethington, E. M., and Plomin, R. (1996). Using MZ differences in the search for nonshared environmental effects. J. Child Psychol. Psychiatry 37, 695–704. Pinker, S. (2002). “The Blank Slate.” Penguin, NY. Plomin, R. (1994). “Genetics and Experience: The Interplay Between Nature and Nurture.” Sage Publications, Thousand Oaks, CA. Plomin, R., and Bergeman, C. S. (1991). The nature of nurture: Genetic influences on ‘environmental’ measures. Behav. Brain Sci. 14, 373–427. Plomin, R., and Daniels, D. (1987). Why are children in the same family so different from each other? Behavioral and Brain Sci. 10, 1–16. Plomin, R., DeFries, J. C., and Loehlin, J. C. (1977). Genotype‐environment interaction and correlation in the analysis of human behavior. Psychol. Bull. 84, 309–322. Plomin, R., DeFries, J. C., McClearn, G. E., and McGuffin, P. (2001). “Behavioral Genetics,” 4th ed. W.H. Freeman, New York. Plomin, R., Foch, T. T., and Rowe, D. C. (1981). Bobo clown aggression in childhood: Environment not genes. J. Res. Pers. 15, 331–342. Pogue‐Geile, M. F., and Rose, R. J. (1985). Developmental genetic studies of adult personality. Dev. Psychol. 21, 547–557. Pray, L. A. (2004). Epigenetics: Genome, meet your environment. The Scientist 18, 5 July 2004. Purcell, S. (2002). Variance components models for gene‐environment interaction in twin analysis. Twin Res. 5, 554–571. Purcell, S. (n. d). Introduction to Quantitative Genetics. From Kings College London, Institute of Psychiatry web site: http://www.statgen.iop.kcl.ac.uk/bgim/ Raine, A. (1993). “The Psychopathology of Crime.” Academic Press, New York. Raine, A., Brennan, P. A., Farrington, D. P., and Mednick, S. A. (eds.) (1997). “Biosocial Bases of Violence.” Plenum Press, NY. Reid, J., Patterson, G. R., and Snyder, J. (2002). “Antisocial Behavior in Children and Adolescents.” American Psychological Association, Washington DC. Rende, R. D., Slomkowski, C. L., Stocker, C., Fulker, D. W., and Plomin, R. (1992). Genetic and environmental influences on maternal and sibling interaction in middle childhood: A sibling adoption study. Dev. Psychol. 28, 484–490. Rhee, S. H., and Waldman, I. D. (2002). Genetic and environmental influences on antisocial behavior: A meta‐analysis of twin and adoption studies. Psychol. Bull. 128, 490–529. Ridley, M. (2003). “Nature via Nurture.” Fourth Estate, London. Riggins‐Caspers, K. M., Cadoret, R. J., Knutson, J. F., and Langbehn, D. (2003). Biology‐environment interaction and evocative biology‐environment correlation: Contributions of harsh discipline and parental psychopathology to problem adolescent behaviors. Behav. Genet. 33, 205–220. Risch, N. (2000). Searching for genetic determinants in the new millennium. Nature 405, 847–856. Robins, L. N. (1978). Sturdy childhood predictors of antisocial behavior: Replications from longitudinal studies. Psychol. Med. 8, 611–622. Robins, L. N., and Regier, D. A. (1991). “Psychiatric Disorders in America: The Epidemiological Catchment Area Study.” Free Press, NY. Rodgers, J. L., Buster, M., and Rowe, D. C. (2001). Genetic and environmental influences on delinquency: DF analysis of NLSY kinship data. J. Quantitative Criminol. 17, 145–168. Rodgers, J. L., Rowe, D. C., and Li, C. (1994). Beyond nature versus nurture: DF analysis of nonshared influences on problem behaviors. Dev. Psychol. 30, 374–384.
3. Genetic and Environmental Influences on Antisocial Behaviors
101
Rose, N. (2000). The biology of culpability: Pathological identity and crime control in a biological culture. Theor. Criminol. 4, 5–34. Rose, R. J., Viken, R. J., Dick, D. M., Bates, J. E., Pulkinnen, L., and Kaprio, J. (2003). It does take a village: Nonfamilial environments and children’s behavior. Psychol. Sci. 14, 273–277. Ross, L., and Shestowsky, D. (2003). Contemporary psychology’s challenges to legal theory and practice. Northwestern University Law Review 97, 1081–1114. Rowe, D. C. (1983). Biometrical genetic models of self‐reported delinquent behavior: A twin study. Behav. Genet. 13, 473–489. Rowe, D. C. (1985). Sibling interaction and self‐reported delinquent behavior: A study of 265 twin pairs. Criminology 23, 223–240. Rowe, D. C. (1994). “The Limits of Family Influence: Genes, Experience, and Behavior.” Guilford Press, NY. Rowe, D. C., and Farrington, D. P. (1997). The familial transmission of criminal convictions. Criminology 35, 177–201. Rowe, D. C., and Osgood, D. W. (1984). Heredity and sociological theories of delinquency: A reconsideration. Am. Sociol. Rev. 49, 526–540. Rowe, D. C., Almeida, D. M., and Jacobson, K. C. (1999). School context and genetic influences on aggression in adolescence. Psychol. Sci. 10, 277–280. Rowe, D. C., Rodgers, J. L., and Meseck‐Bushey, S. (1992). Sibling delinquency and the family environment: Shared and unshared influences. Child Dev. 63, 59–67. Rowe, D. C., Vazsonyi, A. T., and Flannery, D. J. (1995). Sex difference in crime: Do means and within‐sex variation have similar causes? J. Res. Crime Delinquency 32, 84–101. Rushton, J. P. (1996). Self‐report delinquency and violence in adult twins. Psychiatric Genet. 6, 87–89. Rushton, J. P., Fulker, D. W., Neale, M. C., Nias, D. K. B., and Eysenck, H. J. (1986). Altruism and aggression: The heritability of individual differences. J. Pers. Social Psychol. 50, 1192–1198. Rutter, M. (2002). Nature, nurture, and development: From evangelism through science toward policy and practice. Child Dev. 73, 1–21. Rutter, M., and Silberg, J. (2002). Gene‐environment interplay in relation to emotional and behavioral disturbance. Ann. Rev. Psychol. 53, 463–490. Rutter, M., and Smith, D. J. (1995). “Psychosocial Disorders in Young People: Time Trends and Their causes.” Wiley, Chichester, UK. Rutter, M., Caspi, A., and Moffitt, T. E. (2003). Using sex differences in psychopathology to study causal mechanisms: Unifying issues and research strategies. J. Child Psychol. Psychiatry 44, 1092–1115. Rutter, M., Pickles, A., Murray, R., and Eaves, L. (2001). Testing hypotheses on specific environmental causal effects on behavior. Psychol. Bull. 127, 291–324. Sankar, P. (2003). Genetic privacy. Ann. Rev. Med. 54, 393–407. Scarr, S. (1992). Developmental theories for the 1990’s: Development and individual differences. Child Dev. 63, 1–19. Schmitz, S., Fulker, D. W., and Mrazek, D. A. (1995). Problem behavior in early and middle childhood: An initial behavior genetic analysis. J. Child Psychol. Psychiatry 36, 1443–1458. Scourfield, J., Van den Bree, M., Martin, N., and Mcguffin, P. (2004). Conduct problems in children and adolescents: A twin study. Arch. Gen. Psychiatry 61, 489–496. Searle, S., and Blackwell, J. M. (1999). Evidence for a functional repeat polymorphism in the promoter of the human NRAMP1 gene that correlates with autoimmune versus infectious disease susceptibility. J. Med. Genet. 36, 295–299. Serbin, L., and Karp, J. (2003). Intergenerational studies of parenting and the transfer of risk from parent to child. Curr. Dir. Psychol. Sci. 12, 138–142. Sharma, A. R., McGue, M. K., and Benson, P. L. (1998). The psychological adjustment of United States adopted adolescents and their nonadopted siblings. Child Dev. 69, 791–802.
102
Terrie E. Moffitt
Sigvardsson, S., Cloninger, C. R., Bohman, M., and von Korring, A. (1982). Predisposition to petty criminality in Swedish adoptees. III. Sex differences and validation of the male typology. Arch. Gen. Psychiatry 39, 1248–1253. Silberg, J., Rutter, M., Meyer, J., Maes, H. H., Hewitt, J., Simonoff, E., Pickles, A., Loeber, R., and Eaves, L. (1996). Genetic and environmental influences on the covariation between hyperactivity and conduct disturbance in juvenile twins. J. Child Psychol. Psychiatry 37, 803–816. Silberg, J. L., Erickson, M. T., Meyer, J. M., Eaves, L. J., Rutter, M., and Hewitt, J. K. (1994). The application of structural equation modelling to maternal ratings of twins’ behavioral and emotional problems. J. Consult. Clin. Psychol. 62, 510–521. Simon, G. E., and VonKorff, M. (1995). Recall of psychiatric history in cross‐sectional surveys: Implications for epidemiologic research. Epidemiologic Rev. 17, 221–227. Simonoff, E., Pickles, A., Hewitt, J., Silberg, J., Rutter, M., Loeber, R., Meyer, J., Neale, M., and Eaves, L. (1995). Multiple raters of disruptive child behavior: Using a genetic strategy to examine shared views and bias. Behav. Genet. 25, 311–326. Simonoff, E., Pickles, A., Meyer, J. M., Silberg, J., Maes, H., Loeber, R., Rutter, M., Hewitt, J. K., and Eaves, L. J. (1997). The Virginia Twin Study of adolescent behavioral development. Arch. Gen. Psychiatry 54, 801–808. Slutske, W. S. (2001). The genetics of antisocial behavior. Curr. Psychiatry Rep. 3, 158–162. Slutske, W. S., Cronk, N. J., and Nabors‐Oberg, R. E. (in press). Familial and genetic factors. In “Conduct and Oppositional Defiant Disorders: Epidemiology, Risk Factors, and Treatment” (C. Essau, ed.). Lawrence Erlbaum Assoc., New Jersey. Available from
[email protected] Slutske, W. S., Eisen, S. A., Xian, H., True, W. R., Lyons, M. J., Goldberg, J., and Tsuang, M. T. (2001). A twin study of the association between pathological gambling and antisocial personality disorder. J. Abnorm. Psychology 110, 297–308. Slutske, W. S., Heath, A. C., Dinwiddie, S. H., Madden, P. A. F., Bucholz, K. K., Dunne, M. P., Statham, D. J., and Martin, N. G. (1997). Modelling genetic and environmental influences in the etiology of conduct disorder: A study of 2682 adult twin pairs. J. Abnorm. Psychol. 106, 266–279. Slutske, W. S., Heath, A. C., Dinwiddie, S. H., Madden, P. A. F., Bucholz, K. K., Dunne, M. P., Statham, D. J., and Martin, N. G. (1998). Common genetic risk factors for conduct disorder and alcohol dependence. J. Abnormal Psychol. 107, 363–374. Sluyter, F., Arseneault, L., Moffitt, T. E., Veenema, A. H., de Boer, S., and Koolhas, J. (2003). Toward an animal model for antisocial behavior: Parallels between mice and humans. Behav. Genet. 33, 563–574. Smith, C. A., and Farrington, D. P. (2003). Continuities in antisocial behavior and parenting across three generations. J. Child Psychol. Psychiatry 44, 1–18. Stanger, C., Achenbach, T., and Verhulst, F. C. (1997). Accelerated longitudinal comparisons of aggressive versus delinquent syndromes. Dev. Psychopathol. 9, 43–58. Stevenson, J., and Graham, P. (1988). Behavioral deviance in 13‐year twins: An item analysis. J. Child Adolesc. Stud. 27, 791–797. Stone, R. D. (2003). The cloudy crystal ball: Genetics, child abuse, and the perils of predicting behavior. Vanderbilt Law Rev. 56, 1558–1588. Stoolmiller, M. (1999). Implications of the restricted range of family environments for estimates of heritability and nonshared environment in behavior‐genetic adoption studies. Psychol. Bull. 125, 392–409. Suomi, S. J. (in press). How gene‐environment interactions can shape the development of socioemotional regulation in rhesus monkeys. In “Socioemotional Regulation: Dimensions, Developmental Trends, and Influences” (B. S. Zuckerman and A. F. Lieberman, eds.), Skillman, NJ. Tabor, H. K., Risch, N. J., and Meyers, R. M. (2002). Candidate‐gene approaches for studying complex genetic traits: Practical considerations. Nat. Genet. Rev. 3, 1–7.
3. Genetic and Environmental Influences on Antisocial Behaviors
103
Taylor, A. (2004). The consequences of selective participation on behavioral‐genetic findings: Evidence from simulated and real data. Twin Res. 7(5), 485–504. Taylor, J., McGue, M., and Iacono, W. G. (2000a). Sex differences, assortative mating, and cultural transmission effects on adolescent delinquency: A twin family study. J. Child Psychiatry Psychol. 41, 433–440. Taylor, J., Iacono, W. G., and McGue, M. (2000b). Evidence for a genetic etiology of early‐onset delinquency. J. Abnorm. Psychol. 109, 634–643. Taylor, J., Loney, B. R., Bobadilla, L., Iacono, W., and McGue, M. (2003). Genetic and environmental influences on psychopathy trait dimensions in a community sample of male twins. J. Abnorm. Child Psychol. 31, 633–645. Taylor, J., McGue, M., Iacono, W. G., and Lykken, D. T. (2000). A behavioral genetic analysis of the relationship between the socialization scale and self‐reported delinquency. J. Pers. 68, 29–50. Tecott, L. H., and Wehner, J. M. (2001). Mouse molecular genetic technologies: Promise for psychiatric research. Arch. Gen. Psychiatry 58, 995–1004. Tellegen, A., Lykken, D. T., Bouchard, T. J., Wilcox, K. J., Segal, N. L., and Rich, S. (1988). Personality similarity in twins reared apart and together. J. Pers. Soc. Psychol. 6, 1031–1039. Thapar, A., and McGuffin, P. (1996). A twin study of antisocial and neurotic symptoms in childhood. Psychol. Med. 26, 1111–1118. Thapar, A., Harrington, R., and McGuffin, P. (2001). Examining the comorbidity of ADHD‐related behaviors and conduct problems using a twin study design. Br. J. Psychiatry 179, 224–229. Toomey, R., Lyons, M. J., Eisen, S. A., True, W., Goldberg, J., and Tsuang, M. T. (1997). Abstract: A twin study of domestic violence. Am. J. Med. Genet. (Neuropsychiatric Genet.) 74. True, W. R., Heath, A. C., Scherrer, J. F., Xian, H., Lin, N., Eisen, S. A., Lyons, M. J., Goldberg, J., and Tsuang, M. T. (1999). Interrelationship of genetic and environmental influences on conduct disorder and alcohol and marijuana dependence symptoms. Am. J. Medical Genet.(Neuropsychiatric Genet.) 88, 391–397. Turkheimer, E., and Waldron, M. (2000). Nonshared environment: A theoretical, methodological, and quantitative review. Psychol. Bull. 126b, 78–108. U. S. Bureau of Justice Statistics (2002). Criminal Victimization 2001. (NCJ Report 194610). U. S. Department of Justice, Washington, DC. Available at http://www.ojp.usdof.gov/bjs van den Oord, E. J. C. G., Boomsma, D. I., and Verhulst, F. C. (1994). A study of problem behavior in 10‐ to 15‐year‐old biologically related and unrelated international adoptees. Behav. Genet. 24, 193–205. van den Oord, E. J. C. G., Koot, H. M., Boomsma, D. I., Verhulst, F. C., and Orlebeke, J. F. (1995). A twin‐singleton comparison of problem behavior in 2–3 year olds. J. Child Psychol. Psychiatry 36, 449–458. van den Oord, E. J. C. G., Verhulst, F. C., and Boomsma, D. I. (1996). A genetic study of maternal and paternal ratings of problem behaviors in 3‐year‐old twins. J. Abnorm. Psychol. 105, 349–357. van der Valk, J. C., Verhulst, F. C., Stroet, T. M., and Boomsma, D. I. (1998a). Quantitative genetic analysis of internalising and externalizing problems in a large sample of 3‐year‐old twins. Twin Res. 1b, 25–33. van der Valk, J. C., Verhulst, F. C., Neale, M. C., and Boomsma, D. I. (1998b). Longitudinal genetic analysis of problem behaviors in biologically related and unrelated adoptees. Behav. Genet. 28, 365–380. van Os, J., and Sham, P. (2003). Gene‐environment interaction and correlation in schizophrenia. In “The Epidemiology of Schizophrenia” (R. Murray, P. B. Jones, E. Susser, J. van Os, and M. Cannon, eds.), pp. 235–253. Cambridge University Press, Cambridge, UK. Vandell, D. L. (2000). Parents, peer groups, and other socializing influences. Dev. Psychol. 36, 699–710.
104
Terrie E. Moffitt
VanDusen, K., Mednick, S. A., Gabrielli, W. F., and Hutchings, B. (1983). Social class and crime in an adoption cohort. J. Criminal Law Criminol. 74, 249–269. Varmuza, S. (2003). Epigenetics and the renaissance of heresy. Genome 46, 963–967. Vernon, P. A., McCarthy, J. M., Johnson, A. M., Jang, K. L., and Harris, J. A. (1999). Individual differences in multiple dimensions of aggression: A univariate and multivariate genetic analysis. Twin Res. 2, 16–21. Viding, E., Blair, J. R., Moffitt, T. E., and Plomin, R. (2005). Psychopathic syndrome indexes strong genetic risk for antisocial behavior in 7‐year‐olds. J. Child Psychol. Psychiatry 46, 592–597. Waldman, I. D., and Slutske, W. S. (2000). Antisocial behavior and alcoholism: A behavioral genetic perspective on comorbidity. Clin. Psychol. Rev. 20, 255–287. Waldman, I. D., Rhee, S. H., Levy, F., and Hay, D. A. (2001). Causes of the overlap among symptoms of attention deficit hyperactivity disorder, oppositional defiant disorder, and conduct disorder. In “Attention, genes, and ADHD” (F. Levy and D. A. Hay, eds.), pp. 115–138. Taylor &Francis, Philadelphia, PA. Walters, G. D. (1992). A meta‐analysis of the gene‐crime relationship. Criminology 30, 595–613. Waschbusch, D. A. (2002). A meta‐analytic examination of comorbid hyperactive‐impulsive‐ inattention problems and conduct problems. Psychol. Bull. 128, 118–150. Wolf, C. R., Smith, G., and Smith, R. L. (2000). Science, medicine and the future: Pharmacogenetics. BMJ 320, 987–990. Wong, M. Y., Day, N. E., Luan, J. A., and Wareham, N. J. (2003). The detection of gene‐ environment interaction for continuous traits: Should we deal with measurement error by bigger studies or better measurement? Int. J. Epidemiol. 32, 51–57. World Health Organization (2002). “World Report on Violence and health.” Author, Geneva. Yang, Q., and Khoury, M. J. (1997). Evolving methods in genetic epidemiology. III. Gene‐environment interaction in epidemiological research. Epidemiol. Rev. 19, 33–43. Young, S. E., Stallings, M. C., Corley, R. P., Krauter, K. S., and Hewitt, J. K. (2000). Genetic and environmental influences on behavioral disinhibition. Am. J. Med. Genet. 96, 684–695.
4
Genetics of Graviperception in Animals Kathleen M. Beckingham,* Michael J. Texada, * Dean A. Baker,{ Ravi Munjaal,* J. Douglas Armstrong{ *Department of Biochemistry and Cell Biology, Rice University Houston, Texas 77005 { School of Informatics, University of Edinburgh, Edinburgh EH8 9LE, United Kingdom
I. Introduction II. Vertebrate Genetic Models (M. Musculus and D. Rerio) A. The gravity‐sensing organs B. Genetic studies of the development of the gravity‐sensing organs C. Genetic analysis of signaling pathways involved in graviperception III. The Arthropod Genetic Model (D. Melanogaster) A. The gravity‐sensing organs B. Genetic studies of mechanosense organ development C. Genetics of mechanosensory signaling pathways in Drosophila D. Genetic insights from behavioral studies of gravitaxis IV. The Nematode Model System (C. Elegans) A. Gravity‐sensing organs? B. Genetic control of mechanosense organ development C. Mechanosensory signaling pathways V. Conclusion Acknowledgments References
Advances in Genetics, Vol. 55 Copyright 2005, Elsevier Inc. All rights reserved.
0065-2660/05 $35.00 DOI: 10.1016/S0065-2660(05)55004-1
106
Beckingham et al.
ABSTRACT Gravity is a constant stimulus for life on Earth and most organisms have evolved structures to sense gravitational force and incorporate its influence into their behavioral repertoire. Here we focus attention on animals and their diverse structures for perceiving and responding to the gravitational vector–one of the few static reference stimuli for any mobile organism. We discuss vertebrate, arthropod, and nematode models from the perspective of the role that genetics is playing in our understanding of graviperception in each system. We describe the key sensory structures in each class of organism and present what is known about the genetic control of development of these structures and the molecular signaling pathways operating in the mature organs. We also discuss the role of large genetic screens in identifying specific genes with roles in mechanosensation and graviperception. ß 2005, Elsevier Inc.
I. INTRODUCTION Gravity is an all‐pervasive force on Earth. The bodies of most life forms have been shaped over evolutionary time by the constant presence of the Earth’s gravitational force. Further, with the exception of perhaps the smallest unicellular organisms, the evolution of systems for detecting and responding to gravity in order to move or grow in functionally advantageous ways, has been essential for all forms of life. In plants, most of the responses to gravity are trophic (i.e., growth‐ related), and a substantial body of research has uncovered much about the mechanisms whereby gravity influences the orientation of growth in various parts of the plant (Haswell, 2003; Hoson and Soga, 2003). In contrast, in animals, the effects of gravity are more immediate, with information on the gravitational vector being incorporated into many aspects of animal behavior. These fundamental differences in the role of gravity in the lives of plants and animals have resulted in very different mechanisms for sensing and responding to gravity, with slow‐acting mechanisms predominating in plants and rapidly transducing systems predominating in animals. Here we will limit ourselves to a discussion of the role of gravity in animals, with the goal of asking what molecular genetics has contributed thus far to our understanding of graviperception in these species. Four multicellular animals offer genetics as an effective approach to the study of graviperception— the mouse Mus musculus, the zebrafish Danio rerio, the fruit fly Drosophila melanogaster, and the nematode Caenorhabditis elegans. Genetic studies with these species have provided insight into both: (1) the molecular mechanisms underlying specification and differentiation of gravity‐sensing organs and
4. Genetics of Graviperception in Animals
107
(2) the molecular components of the graviperception signal transduction processes. We will focus on both these aspects of graviperception for these model systems, providing information on the structure of the relevant sense organs and behavioral responses as necessary. In much of the older literature, responses to gravity have been described using the prefix “geo‐” (e.g., geotropisms or geotaxes). This nomenclature narrowly attributes the growth or movement in question to the influence of the Earth rather than more appropriately to the influence of gravitational force in general. In essence, it is equivalent to naming responses to light after the Sun (i.e., heliotropism or heliotaxis) as opposed to the electromagnetic radiation the Sun emits (phototropism, phototaxis). Throughout this review, nomenclature that refers to gravitational force in general (gravitaxis, graviperception, and so on) will be used.
II. VERTEBRATE GENETIC MODELS (M. MUSCULUS AND D. RERIO) A. The gravity‐sensing organs Across most of evolution, the sense organs used to detect the gravitational vector are based on a very similar design. This design is centered on displacement of the cilia on a field of sensory cells (hair cells) by a mass (typically a dense calcareous stone or statolith) falling freely under the influence of gravity (Fig. 4.1). These statocyst‐type organs are found in many marine invertebrates and in all vertebrates. However, the statocyst organs of these species are not evolutionarily homologous, since they share no complex ancestor, but their development and signaling processes (e.g., cilial development and molecular pathways) presumably rely on conserved mechanisms. Each mechanosensitive cilium (Fig. 4.2), or kinocilium, of a hair cell is polarized (i.e., it has a preferred plane of movement; this plane passes between the two central microtubules of the cilium). Movement in one direction within this plane causes activation of the hair cell, whereas bending in the opposite direction inhibits the hair cell. A hair cell may bear many kinocilia, however, and they may not all be polarized in the same direction. Vertebrate hair cells bear only one kinocilium. In the place of the missing kinocilia, vertebrate hair cells bear stereocilia, which are more properly named stereovilli, since they do not contain the 92þ2‐microtubule arrangement found in true cilia. These stereocilia are linked to the kinocilium by a linkage at each tip. In contrast, in invertebrates the multiple kinocilia are linked along their lengths by tight junctions. Whereas the statocyst organs of invertebrates are discrete and separate structures, the equivalent organs of vertebrates represent one of three
108
Beckingham et al.
Figure 4.1. Schematic of sensory hair cells of the vertebrate vestibular apparatus. Two types of hairs are often present—differing mainly in their neural connections. Type I cells show only afferent (AFF) innervation, whereas type II show both afferent and efferent (EFF) connections. KC—kinocilium. From “The vestibular system: Function and morphology” (Torquato Gualtierotti, ed.), p. 23, Fig. 1–22. Reproduced by kind permission of Springer‐Verlag, New York and Heidelberg.
mechanosensory subsystems of the inner ear, the cochlear and semicircular canals being the other two. Most vertebrates have three such organs, the utricle, saccule, and lagena, each being an endolymph‐filled cavity with a macula, or field of sensory hair cells. In fish, a single calcareous stone or otolith is present in each organ, partially embedded in a gelatinous otolith membrane apposed to the macular surface. In mammals, multiple tiny stones or otoconia are used (Fig. 4.3). The macula of the utricle lies in a roughly horizontal plane when the head is held at normal posture and detects front–back and left–right motion; the maculae of the saccule and lagena are within the medial plane to detect front–back and vertical motion. The mouse, like all placental mammals, lacks a lagena. In some fish, but not D. rerio, a fourth otolith organ, the macula neglecta, is present. The otolith organs together with the semicircular canals are termed as vestibular apparatus, with the semicircular canals detecting angular acceleration
4. Genetics of Graviperception in Animals
109
Figure 4.2. Transverse EM section of the single kinocilium and field of stereocilia of a macular hair cell. The kinocilium has the (9þ2) arrangement of microtubules along its length. The stereocilia are filled with numerous longitudinal actin fibers. From “The vestibular system: Function and morphology” (Torquato Gualtierotti, ed.), p. 21, Fig. 1–19. Reproduced by kind permission of Springer‐Verlag, New York and Heidelberg.
and the otolith organs linear acceleration. The three canals lie in the three planes of space with each canal connected at both ends to the utricle. The sensory cells of the canals lie in bulbous structures termed ampullae composed of a cup of gelatinous matrix, the cupula, which lies above the sensory epithelium (the crista ampullaris) and surrounds the cilia. As for the gravity‐sensing otolith organs, the sensory cells are activated by deflection of the cilia but in this case it is movement of the endolymph within the canals, which creates ciliary bending as opposed to movement of the otoliths. Mammalian hearing is accomplished in a similar manner; the cilia of the cochlea’s inner and outer hair cells are embedded in the jelly‐like tectorial membrane, and when a pressure wave moves down the cochlea, a shearing motion is generated between the tectorial membrane and the basilar membrane, to which the hair cell bodies are attached. This shearing bends the stereocilia, leading to depolarization of the hair cells. In fish, such as D. rerio, the cochlea is absent, and an alternative sensory system contributing to sound perception,
110
Beckingham et al.
Figure 4.3. Scanning EM images of the otoconia in the saccule of a 2‐year old human. (A) The entire toconial field of the saccular macula. (B–D) Increasingly higher magnification to show the individual otoconia. From “The vestibular system :Function and morphology” (Torquato Gualtierotti, ed.), p. 90, Fig. 5–2. Reproduced by kind permission of Springer‐Verlag, New York and Heidelberg.
termed as lateral line, is distributed in a stereotypic manner along the longitudinal axis of the body. Interestingly, in zebrafish, two of the three otolith organs of the adult inner ear (those in the saccule and the lagena) probably also largely respond to “sound,” (i.e., water vibrations, with the balance and gravity‐sensing functions primarily performed by the otolith organ of the utricle) (Riley and Moorman, 2000).
B. Genetic studies of the development of the gravity‐sensing organs As discussed above, a common sensory cell type—the mechanosensory hair cell—is used in all the subsystems of the inner ear. In each subsystem, an array of such cells is interspersed among support cells and the field of cells is covered by an extracellular matrix in which the stereocilia are embedded. However, the overall architecture of each subsystem differs; the final differentiated form of
4. Genetics of Graviperception in Animals
111
the hair cells, the spatial arrangement of the hair cells relative to one another, the extracellular matrix and neighboring support cells—these features are all unique to the individual subsystems. Moreover, a unique feature, the ear stones (otoliths or otoconia) is present in the gravity‐sensing maculae. It seems likely then, that after specification of the location and overall structure of the mechanosensory subsystems, the later steps in inner ear development involve mechanisms to generate the differentiation and cell‐type‐specific gene expression necessary for each subsystem. Understanding of the developmental program leading to formation of the inner ear mechanoreceptors is most advanced in the zebrafish (Riley, 2003; Whitfield, 2002; Whitfield et al., 2002). Members of the fibroblast growth factor family, in particular Fgf3 and Fgf8, appear to be the primary factors required to initiate otic induction, with members of the Pax2/5/8 transcription factor family being potential downstream effectors. As described in later section, intense genetic analysis in Drosophila has led to the identification of a hierarchy of genes that execute progressively finer specification of cell fates within mechanosensory systems. Although far from complete, studies to date indicate that zebrafish homologs of several classes of these genes act in related ways to specify the mechanosensory cells of the inner ear. In Drosophila, genes of the achaete‐scute and atonal classes perform the so‐called “proneural” function of initiating the sensory program. Their expression patterns define broad equivalence patches in which any cell has the potential to become a sense organ precursor (SOP) cell. Further, they select the category of SOP cell, which will emerge from this equivalence patch (see Section III.B.). In a second or “neurogenic” phase the SOP gives rise to the various cell types of the final sense organ. The expression pattern of the zebrafish gene zath1, an atonal homolog, suggests a proneural role comparable to that of atonal, in the developing maculae of the saccule and utricle (Whitfield et al., 2002). However, loss of zath1 expression does not result in loss of the equivalence group, but rather only in loss of the hair cells, one of several cell types of the final sense organ. Thus, zath1 may only play a “neurogenic” role, with other genes acting at the level of equivalence group specification. Another atonal homolog, neurogenin 1 that overlaps zath1 in expression, is a possible candidate (Andermann et al., 2002; Riley, 2003). In Drosophila, a lateral inhibition system involving interactions between Notch and its ligands Delta and Serrate leads to restriction of neural fates within the equivalence groups. In zebrafish, the expression of Delta and Notch homologs within the developing maculae is consonant with a lateral inhibition model, and two mutant phenotypes, those of mind bomb (mib) and a dominant negative mutation of the delta A homolog, provide strong evidence that Notch–Delta interactions underlie the choice between the sensory hair cell and support cell fates (Haddon et al., 1999; Riley et al., 1999). Thus, both
112
Beckingham et al.
mutations disrupt Notch–Delta signaling and both produce more hair cells at the expense of support cells. Unexpectedly, mib has proved to encode an E3 ubiquitin ligase that somehow facilitates the Notch–Delta interactions by targeting Delta for ubiquitination (Itoh et al., 2003). In the mouse, genetic and molecular studies implicate similar proteins and regulatory networks at comparable stages of ear mechanosense organ development, although the detailed actions of particular genes probably differs somewhat (Bryant et al., 2002). A cell–cell adhesion molecule of the Ig superfamily, termed BEN, that is expressed early in the development of the single initial prosensory patch has also been identified in mouse and may act to keep this field of cells separate and distinct from other components (Pourquie´ et al., 1992). The mouse atonal homolog Math1 has been more extensively studied than its zebrafish counterpart and has been definitively shown to be necessary and sufficient for hair cell specification (Bermingham et al., 1999; Chen et al., 2002). Like zath1 it does not appear to be critical for equivalence group specification (Hassan and Bellen, 2000). A further aspect in which the mouse system appears to be better understood concerns the roles of Hes1 and Hes5, mammalian homologs of the Drosophila genes hairy and Enhancer of split, which act to enhance Notch activity and thus repress neurogenesis (Xheng et al., 2000; Zine et al., 2001). Mutants of each gene produce more hair cells in the inner ear, demonstrating that they perform a comparable role in this developmental system. A complication of genetic studies in all vertebrates as compared to Drosophila is that there are multiple versions of many of these regulatory genes in the genome. In many developmental contexts, including the development of the inner ear, more than one version of a given gene is expressed. For example, several isoforms of the Notch ligands Serrate and Delta are expressed during inner ear development. Presumably, more complex regulatory interactions are produced by this route. Further, in Drosophila, beyond the initial determination of the SOP cells, invariant cell divisions give rise to the final specialized cells of the mechanosensory organs (see later section). This rigid progression of a developmental program may be absent in vertebrates. For the chick, a boundary model has been proposed to explain the generation of the three types of sense organ in the inner ear from the initial single prosensory patch present in this species (Fekete, 1996). The model proposes that the boundaries formed at the intersection points for expression of key regulatory genes are the sites at which particular sense organ precursors are specified. Transcription factors, such as Pax2, Dlx5, Otx1, and Hmx3 are candidates for the key regulators in this process. Available data suggest that this mechanism may also be tenable for the mouse (Brigande et al., 2000), but data for the zebrafish do not so readily fit this model (Whitfield et al., 2002), perhaps reflecting the fact that the prosensory patches for the various sense organs in this species are distinct and separate right from their inception.
4. Genetics of Graviperception in Animals
113
Studies of Drosophila eye development have identified another network of regulatory genes that operates in sense organ specification. Homologs of the four classes of genes in question (the Pax family of transcription factors, and the genes eyes absent (eya), Sine oculis (So), and Dachsund (Dach), which encode nuclear proteins) are found in vertebrates, often in multiple variant copies, suggesting developmental roles for individual isoforms of these proteins. One of the first mutations identified as affecting inner ear development in the zebrafish, dog‐eared (dog), has proved to be a mutation in an eya homolog (Riley, 2003), and other members of the other gene families involved in this network are known to be expressed in the developing inner ear (Whitfield et al., 2002). Available evidence, particularly from the dog mutant phenotype, suggests the network plays a role in overall pattern formation and cell survival in the inner ear. A knockout mutation of the Six 1 gene in mouse (an So homolog) has provided direct evidence that the Pax–Six–Eya–Dach protein network plays a role in otic vesicle formation (Ozaki et al., 2004). At some point within inner ear development, the steps most relevant to this review are initiated (i.e., the specification of the sensory structures of each of the three [two in fish] sense organ subtypes takes place and gene action specifically directing formation of the gravity‐sensing organs occurs). The macular hair cell fields of the otolith organs then develop separately from those of the cristae ampullaris (the hair cell fields of the semicircular canals) and in mammals from the hair cells of the cochlea. In all vertebrates examined, expression of bone morphogenetic protein 4 (BMP4) prefigures the positions of the cristae, indicating a role in determining the position of these structures (Bryant et al., 2002; Whitfield et al., 2002). In zebrafish, the homeobox proteins MsxC and MsxD appear to be the downstream effectors for BMP4 in determination of the cristae (Whitfield et al., 2002). Given that in this species the utricular and saccular maculae develop before the cristae, a critical distinction would seem to be that macular sensory cell development can proceed without MsxC and MsxD function. Little else is known about the genes specifying the various sense organ types. In zebrafish, as discussed in earlier section, further functional distinctions are thought to exist between the individual maculae of the vestibular system, which raises the issue of how these differences are specified. To date, no mutations that distinguish the development of utricular and saccular sensory patches have been identified. It is possible that the distinction between these very similar structures mainly lies in their central neural connections rather than any differences in structural components. Nevertheless, the connection of these sense organs to different processing centers in the brain still requires some distinction between them in terms of the routing of afferent axons. Such a difference should ultimately be tractable genetically.
114
Beckingham et al.
C. Genetic analysis of signaling pathways involved in graviperception
1. Otoconia–Otolith formation In vertebrates, as discussed in earlier section, the sense organs associated with perception of sound are structurally quite similar to those for perception of linear acceleration and gravity. It is not surprising then that most of the vertebrate mutations affecting balance and coordination (phenotypes suggesting problems with gravity perception) also cause deafness. However, a number of mutations have been isolated in mice and zebrafish that specifically affect the saccular and utricular maculae without affecting auditory responses from the cochlea (in mice) or the lateral line system (in zebrafish). To date the best characterized of these mutants have proved to affect formation of the otoliths and otoconia. In zebrafish, the first step in otolith formation is the capture of small otolith‐seeding granules by the earliest hair cells of the maculae (termed tether cells). Surprisingly, in zebrafish, glycogen particles secreted by the vestibular epithelial cells are the founding component of the otolith (Pisam et al., 2002). These coalesce into globules to which proteins secreted from the epithelial layers are added. These proteins are then thought to promote trapping of the major otolith component, CaCO3, in a defined crystalline lattice. The critical role of secreted proteins in tethering the preotolith particles is demonstrated by experiments to block expression of an endoplasmic reticulum protein (GP96), which acts as a chaperonin for these proteins (Sumanas et al., 2003). A specific defect in the tethering of the seeding particles to the sensory cells is produced, resulting in formation of large detached otolith clumps. A number of mutants that affect otolith synthesis have been isolated from zebrafish genetic screens (Riley and Grunwald, 1995; Whitfield et al., 1996). Studies for one such mutant, monolith (Riley and Grunwald, 1996), suggest that the support cells of the maculae play a major role in secreting products required for otolith formation and tethering. One of the otolith‐deficient mutants from the screen of Whitfield et al. has been cloned and shown to encode a conserved vestibular protein termed Otopetrin‐1 (Otop‐1) (see later section). However, the first protein component of the otolith characterized in zebrafish was identified by a more directed approach (Sollner et al., 2003). Recognizing that a human protein involved in tooth mineralization (dentin sialophosphoprotein, DSPP) is also expressed in the ear, Nicolson’s group identified the most related zebrafish gene and then designed modified antisense oligonucleotides (morpholinos) to abolish its function. Strikingly, the overall structure of the otolith was dramatically altered; animals with intermediate loss of function had star‐shaped otoliths (Fig. 4.4), whereas severe loss of function resulted in lumpy otoliths, which appeared to be almost entirely crystalline CaCO3. Further, the crystalline lattice in these
4. Genetics of Graviperception in Animals
115
Figure 4.4. Changes to zebrafish otolith morphology produced by decreased expression of Starmaker. (a) Wild‐type otolith. (b) Star‐like otolith produced with intermediate loss of Starmaker function. The CaCO3 lattice is still aragonite. (c) With severe loss of Starmaker, otoliths are almost entirely crystalline CaCO3 in a calcite‐type lattice. Scale bar 20 M. SEM photos: Juergen Berger, MPI, Tubingen, Germany, provided courtesy of Dr. T. Nicolson, Oregon Hearing Research Center and Vollum Institute, Portland, OR.
malformed masses of CaCO3 is that of calcite, as opposed to aragonite, the CaCO3 polymorph normally found in the zebrafish otolith. The zebrafish gene has been named starmaker, and its encoded protein is extremely rich in acidic residues, which could act to nucleate calcium binding. In wild‐type animals, the Starmaker protein is concentrated at the surface of the otolith that is juxtaposed to the sensory epithelium. It has been proposed that the protein may act to control crystal growth at this surface and thus regulate the final form of the otolith. In mice, three mutations that cause failure of the development of the otoconia have been well characterized: head tilt (Bergstrom et al., 1998), tilted head (Lim et al., 1978), and tilted (Ornitz et al., 1998). All have very similar behavioral phenotypes indicative of failure to sense gravity, including abnormal head posture, failure to right themselves during a vertical drop, and an inability to swim due to poor orientation under water. Mutation tilted was cloned (Hurle et al., 2003) and found to encode a novel protein with 10 transmembrane domains, which was named Otop‐1. Other Otopetrin genes are present in the mouse and zebrafish genomes, and two regions of Otop‐1 show conservation in Drosophila and C. elegans. Otop‐1 mRNA is synthesized in the sensory epithelia of the saccular and utricular maculae, and the protein is a secreted component of the gelatinous matrix overlying the epithelia (Hurle et al., 2003). The otoconia are embedded in this matrix and are composed of a crystalline lattice of CaCO3 stabilized by acidic glycoproteins. Otoconin‐90 (Oc‐90) is the major representative of this glycoprotein class in mammals (Wang et al., 1998). In addition to the otoconia,
116
Beckingham et al.
this matrix contains numerous small vesicles secreted from the sensory cells. However, Oc‐90 is not synthesized by the sensory cells immediately below the matrix but rather by adjacent nonsensory epithelial cells. Hurle et al. hypothesize therefore that Otop‐1 is present in the membranes of these small vesicles and acts to trap Oc‐90 in the matrix above the sensory cells, leading to formation of the otoconial matrix and subsequent nucleation of CaCO3 crystals. The phenotype of the backstroke zebrafish mutant isolated in the screen of Whitfield et al. (1996) closely parallels that of the tilted mouse—the otoliths are missing and aberrant behavior, including swimming upside down and circling, is seen. This lead (Sollner et al., 2004) to investigate and ultimately establish that backstroke represents a mutation of the homologous Otopetrin gene in zebrafish. Sollner and collaborators have examined the fate of Starmaker in the backstroke mutant and in Otop‐1 loss‐of‐function animals generated with morpholinos. Starmaker is synthesized by both the hair cells and support cells of the sensory epithelium, and although Starmaker influences the morphology of the otoliths, it is not the critical initiating component in otolith formation (Sollner et al., 2003). In the absence of Otop‐1, no otoliths are formed and Starmaker accumulates in the gelatinous membrane above the sensory epithelium, a finding that is in general agreement with the proposed role for Otop‐1 (Hurle et al., 2003). Intriguingly, evidence from mouse mutations suggests that the melanocytes of the inner ear labyrinth play some role in trace metal function that is essential for formation of the otoconia. Correlation of loss of pigment‐producing cells with loss of the otoconia is seen for two mouse mutations, pallid (Lyon, 1952) and muted (Lyon and Meredith, 1969), and in pallid, otoconia can be partially restored by a dietary supplement of manganese (Erway et al., 1971).
2. Mechanosignaling by hair cells Physiological studies have led to a model for the critical mechanosignaling event in all categories of hair cells of the inner ear. The stereocilia are connected at their tips by fragile filaments called tip links. Bending of the stereocilia is postulated to put tension on the tip links causing them to literally pull open mechanosensory ion channels in the stereociliary membrane. The inward ion flux leads to membrane depolarization and ultimately neurotransmitter release at the synapses. Calcium is one of the ions that enters upon stereocilial bending and is thought to mediate the adaptive changes, which occur during prolonged mechanostimulation. Given the tiny number of hair cells present in the animal, genetics has been a major route for dissecting this process. Various mouse mutants, congenital human diseases, and zebrafish mutants have proved to affect genes encoding molecular entities of the mechanotransduction pathway.
4. Genetics of Graviperception in Animals
117
Components identified include: (1) proteins, such as otogelin (Simmler et al., 2000), that are part of the gelatinous membrane overlying the sensory hairs; (2) plasma membrane ion channels and transporters, such as the Ca2þ ATPase isoform 2 (Kozel et al., 1998), which affect the production of the Kþ‐rich endolymph or the recycling of calcium; (3) cell–cell adhesion and junctional molecules, such as cadherin23 (Di Palma et al., 2001) and protocadherin15 (Alagramam et al., 2001), which provide correct orientation–interaction of the hair cells; (4) cytoskeletal elements, such as espin (Zheng et al., 2000), which orient the actin‐rich cytoskeletons of the stereocilia, and (5) motors that move ciliary components relative to their actin cable cores. Four myosins, myosin 1c (Holt et al., 2002), VI (Kappler et al., 2004; Seiler et al., 2004), VIIA (Avraham et al., 1995), and XV (Gibson et al., 1995), are known to have roles in hair cell signal transduction. Several reviews provide greater detail on these processes (Ernstrom and Chalfie, 2002; Frolenkov et al., 2004). Clearly events at the tip links are key to understanding the primary mechanotransduction event. An exciting development is the discovery that cadherin23 (CDH23) is a major component of the tip links. The cell–cell adhesion function of cadherins involves dimerization of the extracellular domains of the cadherins on one cell and interdigitation of these dimers with identical cadherin dimers on the partner cell. Cadherin23 has an unusually large number of cadherin domain repeats in its extracellular region (27 total) and modeling studies suggest that the tip links may primarily represent a coiled coil of CDH23 homodimers formed between adjacent stereocilia pairs. This recognition on CDH23 location has come from studies of the zebrafish CDH23 mutant sputnik (see later section) and from improved immunolocalization studies in wild‐type and waltzer mice, which are mutant for CDH23 (Siemens et al., 2004; Sollner et al., 2004). Further, the cytoplasmic domain of CDH23 has been shown to physically interact with the motor force (myosin 1c) that underlies the adaptation response of the mechanotransduction machinery (Siemens et al., 2004). Given that this response is thought to reflect movement of the transduction channels relative to the ciliary tips this direct CDH23–myosin 1c interaction has intriguing implications for the underlying mechanism. Identification of the actual mechanotransduction channels themselves is of great interest. Kurima et al. (2002) have suggested that the gene affected in the mouse mutations deafness and Beethoven and in some dominant forms of human deafness could be a hair cell mechanosensory channel. The strongest mutations in the gene termed transmembrane cochlear‐expressed gene 1 (TMC1) show complete absence of an electrophysiological response to sound with no initial hair cell structural defects, suggesting a purely functional role. Further, TMC1 has a six transmembrane domain topology similar to that of the superfamily of multimeric ion channels. However immunolocalization and electrophysiological studies of TMC1 have yet to be performed.
118
Beckingham et al.
As described in Section III.C., the nompC gene of Drosophila encodes a mechanosensory channel, which is a member of the TRP family of ion channels (TRPN subfamily). The founding member of the TRP family (trp) was identified through its role in visual signaling in Drosophila; mutations at the locus produced a transient receptor potential. TRP family ion channels are now known to play key roles in almost all forms of sensation, including olfaction, taste, pain, and heat. The Nicolson laboratory cloned a highly conserved NompC homolog from zebrafish. Using morpholinos to knock out gene function, they have provided good evidence that this NompC gene plays a role in inner ear cell mechanotransduction (Sidi et al., 2003). The electrical responses associated with mechanosensation in the inner ear are defective and the extensive endocytotic activity at the apical surfaces of the hair cells, an activity that correlates strongly with transductional competence (Seiler and Nicolson, 1999) is lost. Although a clear NompC homolog was also identified in C. elegans by this group, surprisingly none have been detected in the mouse or human genomes. Thus, NompC cannot be a universal vertebrate mechanotransduction channel. However (Corey et al., 2004) have provided convincing evidence that another TRP channel family member, TRPA1, is a conserved component of the vertebrate hair cell transduction channel. TRPA1 was shown to co‐localize with CDH23 and myosin 1c to the tips of the stereocilia and more importantly, inhibition of TRPA1 gene expression in the inner ear of both mouse and zebrafish reduced mechanotransduction associated currents. TRPA1, like NompC (see later section) is unusual in that it has a large number of ankyrin repeats (17 repeats for TRPA1) at its N‐terminus. An intriguing suggestion concerning the function of these repeats has been made (Corey and Sotomayer, 2004). Although the tip links themselves are usually considered to represent the mechanical “spring” that gates the mechanosensory channels, Corey and Sotomayer argue that CDH23, the major component of the links (see earlier section), does not have the requisite elasticity and that perhaps the ankyrin repeats of the channels themselves might perform this function (Fig. 4.5). For the most part, these molecular components of hair cell transduction have been shown to act in both the cochlear and the vestibular hair cells indicating that the mechanosignaling apparatus in the various hair cell subtypes is fundamentally similar. However, distinctions are emerging that suggest the graviperceptory hair cells do have unique features. Although some genes are expressed in both the cochlear and vestibular sensory epithelia, the known mutations in these genes affect cochlear and vestibular structure and function differentially. This is true for mutations in two genes affecting cell–cell interactions (Celsr1, see later section and claudin, encoding a tight junction protein). Most strikingly mutations in the putative transduction channel TMC1, although causing complete deafness, cause no disruptions in vestibular function. A second gene with high homology to TMC1 is present in the human and
4. Genetics of Graviperception in Animals
119
Figure 4.5. Contrasting models of gating mechanosensory channels. (A) Deflection of the stereocilia in response to sound or movement. (B) Traditional model where the tip links are elastic. (C) Alternative model exploiting an Ankyrin spring. From Corey and Sotomayor (2004). Reproduced by kind permission of DP Corey.
mouse genomes (TMC2) and is also expressed in the inner ear (Kurima et al., 2002). It is possible that this is a major functional channel in the macular hair cells.
3. Behavioral screening as a possible route to mutants defective in graviperception The normal hearing and clearly defined behavioral phenotypes associated with tilted, head tilt and tilted head mutations suggest a genetic approach to identifying further mutations in genes uniquely acting in gravity perception as opposed to other functions of the inner ear. But given the expense and effort associated with mouse genetic screening, it is unlikely that such a directed search will ever be undertaken. However, an open‐ended, large‐scale mouse mutagenesis project has been executed in Germany (Hrabe de Angelis et al., 2000) and behavioral phenotypes identified in some mutants (head weaving and circling) suggests that they will yield further components of the vestibular signaling pathway (Nolan et al., 2000). One such behavioral mutant analyzed in detail to date has proved to affect Celsr1, a homolog of the Drosophila gene encoding the protocadherin Flamingo–Starry night, which plays a role in planar cell polarity (Curtin et al., 2003). But surprisingly, despite the clear vestibular‐linked behavioral defects and relatively good hearing shown by this mutant, the morphological effects detected are purely limited to the cochlear hair cells. In zebrafish, as in the mouse, study of various mutants without otoliths has led to a description of the behavioral defects that reflect a failure to sense gravity. These include inability to maintain a balanced dorsal‐up posture and
120
Beckingham et al.
aberrant swimming with circling, zigzagging, and rolling. In addition, responses that are known to require vestibular input can provide quantitative information on functioning of the graviperception machinery. In particular, the dorsal light reflex (Rahmann and Anken, 2000) whereby fish use vestibular and visual information to tilt their dorsal surfaces toward a light source and compensatory eye rotation upon tilting, are valuable assays. Analysis of some of these behaviors in fish lacking either utricular or saccular otoliths led to the demonstration that it is the utricular otoliths, and not the saccular otoliths, that provide gravity sensing (Riley and Moorman, 2000). Although behavioral screens based specifically on these assays have not been conducted, a general screen for mutants producing behavioral defects was performed as part of the large genetic screen in Tubingen (Granato et al., 1996) and from this screen, a series of mutants producing “circler” behavior (i.e., abnormal swimming indicative of aberrant vestibular function) was identified (Nicolson et al., 1998). These mutants proved to represent eight complementation groups, which have been named for space‐related activities—mariner, skylab, and so on. Most of the mutations affect both swimming behavior and hearing (water vibration), and when the dorsal light reflex response was tested as a specific criterion for gravity sensing, no single mutant locus emerged as showing defects specifically in this system. It seems likely, then, that these mutations affect components of mechanosensation, which are common to all classes of hair cells. Three loci produce structural defects in the hair cells, and strikingly, two of these upon cloning have proved to encode proteins shown from mouse work and/or human disease studies (see Section II.C.2.) to represent components of the mechanosensory cells. Thus, mariner encodes myosin VIIA (Ernest et al., 2000), and sputnik encodes CDH23 (Sollner et al., 2004). Mutations at three further loci produce no structural defects and from electrophysiological studies appear to have defects in the actual transduction process. One of these, gemini, has proved to encode an L‐type calcium channel, which localizes to the synaptic regions (so‐called ribbon synapses) at the apical membrane of the hair cell (Sidi et al., 2004). This location implies a role in neural transmission as opposed to the initial mechanosensory event.
III. THE ARTHROPOD GENETIC MODEL (D. MELANOGASTER) A. The gravity‐sensing organs
1. Specialized gravity‐sensing organs Unlike other phyla, the species of the vast phylum Arthropoda do not possess statocysts organs of the type discussed in an earlier section. Within the subphylum Crustacea, however, convergent evolution has resulted in the production of
4. Genetics of Graviperception in Animals
121
Figure 4.6. Statocysts of the crayfish. (a) Dorsal view of the animal’s head. Bristles covering the opening to the statocyst have been removed on its left side. The outline of the cyst sac is indicated by a dotted line. (b) Anterior view of the preparation cut through at the level of the arrows in (a). The statolith mass is shown in the statocyst on the left but is removed to show the field of hairs on the right. From “Gravity and the organism” (S. A Gordon and M. J. Cohen, eds.)., Fig. 1, p. 224. Reproduced by kind permission of University of Chicago Press, Chicago and London.
a related structure, also termed a statocyst (Fig. 4.6), which operates in a functionally analogous manner. An inorganic mass of particles (the statolith), cemented together by secretions from cells of the statocyst, rests on a bed of feathery hairs. Movements relative to the gravity vector shift the statolith, altering the pattern of hair bending and, as a result, changing the pattern of firing in associated afferent neurons. As for the statocysts in higher vertebrates, these statocysts are paired head structures, which in the decapod Crustacea, lie ventral to the eye sockets, at the base of the first two antennae (Schone, 1971). However, in contrast to the classic statocysts of other phyla, these structures are open to the environment and basically represent cavities in the
122
Beckingham et al.
hard outer exoskeleton of the animal. As a result, the statolith is discarded and must be reformed after each molt. This aspect of the statocysts was the basis of the highly original experiment performed by Kreidl in 1893 (Schone, 1971) that demonstrated their role in gravity perception, rather than in audition. After molting, the statoliths of shrimp were reformed from iron filings as opposed to sand particles, and the animals now oriented their bodies with respect to a magnetic, as opposed to the gravitational, field. A further major difference in the crustacean statocysts is the mechanism of neuronal stimulation. The key sensory event in the archetypal statocyst is the deformation of ciliary projections on the surface of the sensory neurons. In contrast, the feathery hairs of these statocysts are not cilia but are sclerotized cuticular structures, related to the mechanosensory hairs found throughout the phylum (see in later section). The stimulus is transmitted to neural tissue through a fine cuticular thread (the chorda), which travels inwards from the hair and forms contacts with the ciliary projections of several afferent neurons. The complex structure seen at these connections resembles somewhat the structure of a type of stretch receptor (chordotonal organ) found throughout this phylum (see in later section). In some aquatic insects, organs that have the appearance of statocysts have been reported and in some Hemiptera, specialized structures termed buoyancy organs use trapped air to sense the gravity vector under water (Horn, 1985). The only other known type of specialized gravity‐sensing organ within this phylum are the large pointed conical structures termed cerci, which protrude from the dorsal caudal tip of the abdomen in crickets and cockroaches. The cerci are decorated with four fields of club‐shaped hairs arranged such that hairs within each field will all deflect in the same direction. Each hair has a heavy bulbous tip on a slender stem, leading to the name club‐shaped sensillae. Electrophysiological recordings, behavioral responses, and ablation of the sensillae confirm that they respond to changes in the gravitational field and induce reflexive responses (Horn and Bischof, 1983; Horn and Foller, 1985).
2. Dispersed mechanosensory sense organs For the vast majority of arthropod species, including D. melanogaster, no specialized graviperception organs have been identified. Surprisingly, in fact, no physiological experimentation to address this question has been performed for Drosophila. Since perception of the gravity vector is always mediated via its directed mechanical effects on sensory structures, it seems likely that mechanosensory organs of some type are used to detect gravitational force. Most investigations of gravity perception in arthropods without statocysts have been performed on insect species. Many of the initial studies were conducted
4. Genetics of Graviperception in Animals
123
in German laboratories 30–40 years ago and the original work is published in German in journals that are difficult to retrieve. Some of the references given here therefore are reviews in English that discuss these studies rather than the original references themselves. Most studies of insect gravitaxis have focused on identifying the body parts in which mechanosensory structures contributing to gravity detection are localized. These studies relied upon identifying behavioral responses to gravity and then performing experimental manipulations of various body parts to address their effects upon these responses. By this route, roles for receptors in the legs and the antennae of insects, such as crickets (Horn and Foller, 1985), dipteran flies (Horn, 1982; Horn and Kessler, 1975), stick insects (Wendler, 1971), bees (Horn, 1975), and ants (Markl, 1971) have been identified, and for proprioceptive bristle fields on the necks, bodies and leg joints of dragon flies, bees, and ants (Markl, 1971; Wilson, 1971). In any one insect species, receptors that influence gravity‐based responses are found in several different body parts. Thus, a relevant question is whether all of these structures contain true gravity receptors, or rather whether some, or all, of these appendages provide proprioceptive information that contributes to a gravitational response. A related question is whether any of these mechanoreceptors, like the statocysts of other organisms, are purely gravity receptors, with no other mechanosensory functions. One criterion for identifying receptors that only function to provide responses to gravity is suggested by the work on crustacean (and higher vertebrate) responses; there are some responses, such as the righting responses of the swimmerets and uropods in lobsters that are completely controlled by the statocysts with no input from proprioceptive type information (Davis, 1971). A variety of responses have been used to examine gravity perception in insects. One simple assay uses the negatively gravitaxic walking response seen on an inclined plane. In addition, compensatory head or antennal movements induced by body tilting have been examined. An interesting finding is that, in those insects examined (crickets and blow flies), these compensatory movements are only induced if the insect is walking or simulating walking (Horn, 1982; Horn and Foller, 1985). To date, no response that is dependent upon a single body part has been identified and the level of redundancy in terms of sensory input is high for most responses. For example, in ants, receptors in the antennae, neck joint, thorax–petiole joint, petiole–gaster joint and one leg joint all contribute to gravitational responses and yet, faced with inactivation of most of these systems, the animal can still mount a directed gravitational response (Markl, 1971). Are any of the receptors providing information about the gravity vector per se or are they all providing proprioceptor information? This is a difficult question to answer since it involves use of experimental paradigms that provide one kind of stimulus but not the other. Wendler (1971)
124
Beckingham et al.
showed conclusively that, in the absence of antennal input, gravity‐directed walking in the stick insect is controlled by a receptor system (probably in the legs) that is purely dependent on the direction of the gravity vector as detected by body weight. Further he showed that receptors capable of detecting proprioceptive information from torque‐induced movement of body parts were not used to provide gravity vector information in this situation. These findings suggest that there are sensory structures that can respond to the gravity vector alone, but whether they also respond to other body movements and provide proprioceptive information in other situations is unknown. Leaving aside the question of how information from the various mechanosensory organs contributes to gravity perception, what progress has been made towards identifying the responding sensory structures themselves in the various body parts implicated in gravity responses? Four classes of insect mechanoreceptors have been identified, all representing modifications of a basic developmental unit, which generates a ciliate sensory neuron and three support cells (Keil, 1997). Three of these classes of receptor—the mechanoreceptive bristles, trichobothria, and campaniform sensillae have associated external structures on the cuticle and respond, respectively, to touch, air currents, and deformation of the cuticle. The fourth class, the scolopidia have no external structures and respond to stretch. There is good evidence for the contribution of the bristle type mechanoreceptors to gravity responses (Horn, 1985). The receptors in the neck and body joints of bees and ants are of this class and particular fields of bristles have been identified as producing gravity related responses from experiments in which they were shaved off, or immobilized with wax (Markl, 1971). Horn and Kessler (1975) also provided evidence that the mechanobristles on the first antennal segment of Calliphora erythrocephala contribute to gravity perception. Similarly, scolopidial type mechanoreceptors may be part of the graviresponse system; the strain receptors identified in Wendler’s experiment probably correspond to the scolopidial type femoral chordotonal (CH) organ identified as playing a role in graviperception in the stick insect Carausius morosus (Horn, 1985). To date there is no evidence to implicate the campaniform class of receptors as playing a role in gravity perception. Two highly specialized mechanosensory structures are found in many insect species and it is worth considering whether these play a role in gravity sensing. An antennal organ termed “Johnston’s organ” in the Diptera and “Janet’s organ” in ants is composed of highly specialized arrays of the scolopidial type CH stretch receptors. In the Diptera, it is clear that a major function of this organ is sound reception in the 1–500 Hz range (Caldwell and Eberl, 2002), whereas in ants, it can play a role in fast antennal movements associated with tactile probing of the environment (Ehmer and Gronenberg, 1997). Nevertheless, in dipterans, such as D. melanogaster the geometry of the organ is such
4. Genetics of Graviperception in Animals
125
that it clearly could respond to passive movements of the last antennal segment under gravitational force and therefore might also play a role in detection of the gravity vector. In the Diptera there are highly specialized campaniform sensillar arrays on the halteres (the modified second wings). The halteres function as gyroscopes during flight, vibrating in synchrony with the wings in a vertical plane. The campaniform sensillary arrays are believed to register strain on the cuticle as a result of torque forces during flight, information, which is used to generate corrective movements in flight. Whether in walking insects these arrays might also respond tonically to the gravity vector is a relatively unexplored question, although Horn could not detect a role for these structures in the blow fly C. erythrocephala (Horn and Kessler, 1975).
B. Genetic studies of mechanosense organ development As discussed above, two types of insect mechanosense organs are implicated in gravity perception—mechanosensory bristles and CH organs. Thanks to the intense genetic study of neural development in Drosophila significant progress has been made in identifying the determinative genes that act in the development of these two classes of sense organs. Many of these regulatory genes originally characterized in Drosophila are now known to play related roles in vertebrate inner ear development (see Section II.B.). Both the mechanosensory bristles and the CH organs belong to the so‐ called “type 1” class of insect mechanosense organs (Fig. 4.7). This class is characterized by the presence of cilia, which are the initial site of mechanical
Figure 4.7. Contrasting the mechanosensory bristle (A) and the chordotonal organ (B) sense organs of Drosophila. From Fig. 1 of Chung et al. (2001). Reproduced by kind permission of Elsevier.
126
Beckingham et al.
action. Mechanosensory bristles are external sense organs (ES) with an obvious external cuticular structure. In contrast, CH organs have no external structure and respond to internal stretch as a result of movement of one body part relative to another. These receptors provide proprioceptive information as well as, potentially, gravity perception. The initial step in mechanosense organ development is the specification, by prepatterning proteins, of fields of equipotent precursor cells. Competence to assume a neural fate within these regions is then conferred through expression of one of several “proneural” genes—atonal confers CH competence (Gupta and Rodrigues, 1997; Jarman et al., 1993, 1994, 1995); genes of the achaete‐scute complex (AS‐C) confer ES precursor competence (Cabrera et al., 1987; Roberts, 1961; Villares and Cabrera, 1987; Young and Lewontin, 1966). The multidendritic, or “type 2” sensory neurons of the larva and most adult chemoreceptors are designated by absent multidendritic neurons and olfactory sensillae (amos) (Goulding et al., 2000; Huang et al., 2000). All of these proneural genes encode transcription factors of the basic helix–loop–helix (bHLH) category. Although the initial positions of precursor fields are dictated by the combinations of patterning proteins acting to induce expression of particular proneural genes, continued proneural gene expression is probably maintained in an autoregulatory manner (Sun et al., 1998). Once a cluster of neural‐competent cells has been specified by patterning and proneural genes, the Notch–Delta–Serrate lateral inhibition system comes into operation to produce a single neural SOP cell from this cluster (Cabrera, 1990; Campos‐Ortega and Knust, 1990). As discussed in an earlier section, this lateral inhibition system is now recognized as playing a comparable role in the choice of cell fate in many developing sensory systems including the mechanosensors of the mammalian inner ear. All cells of an initial neural cluster express both the transmembrane receptor Notch and its transmembrane ligands Delta and Serrate. However, activation of Notch in a given cell by binding of Delta or Serrate from neighboring cells leads to the suppression of neural cell fate in that cell and loss of the ability to express Delta and Serrate (Baker et al., 1996). Thus, although all cells initially express both Notch and Delta–Serrate and have the capacity to block neural fate in one another, small initial differences in the relative expression of Notch versus Delta–Serrate by individual cells are amplified by this feedback loop and ultimately a single cell expressing high levels of Delta–Serrate emerges stochastically. This is the only cell still expressing the proneural gene and thus capable of pursuing a neural fate. For a review of Drosophila Notch functions, see (Portin, 2002). The Notch lateral inhibition system is modulated by other signaling molecules, allowing the “winnowing” process to act across relatively large cell clusters; among these are Scabrous, which binds and stabilizes surface Notch receptors (Lee et al., 1996; Mlodzik et al., 1990; Powell et al., 2001), and
4. Genetics of Graviperception in Animals
127
epidermal growth factor receptor (EGFR) and its ligands (Baonza et al., 2001; Dominguez et al., 1998). Further, in structures where clusters of sensory precursor cells are needed (e.g., where multiple CH precursor cells are required to form a specialized organ), each initially designated CH precursor delaminates from the field of competent cells, thus removing the Delta–Serrate‐based activation of Notch in neighboring cells. Further, the delaminating cell activates expression of the transmembrane protease Rhomboid, resulting in cleavage and diffusion of a ligand (Spitz) to activate EGFR in the remaining cells of the layer. Epidermal growth factor receptor activation blocks Notch signaling and thus restores the potential to form more CH precursors. Beyond selection of a precursor cell for sense organ formation, specification of the sense organ type is necessary. Although as discussed above, specific proneural genes are associated with the specification of particular mechanosense organs, surprisingly the proneural genes are somewhat interchangeable. For example, ectopic amos expression in thoracic regions leads to clustered bristle formation (Lai, 2003; Villa‐Cuesta et al., 2003). Possibly the reason for the existence of multiple proneural genes is that they are regulated differently, allowing more complex spatial patterning of sense organs (Pistillo et al., 2002; Skaer et al., 2002; Vervoort and Ledent, 2001; Wheeler et al., 2003). For example, the atonal promoter is somewhat insensitive to lateral inhibition leading to the formation of clusters of sense organs, such as that found in Johnston’s organ (the hearing organ of the fly). Downstream of the proneural genes, the homeobox transcription factor Cut is the key switch factor in the choice between ES and CH organ formation. Cut activity is essential to produce the ES lineages, with cut mutants producing CH organs in the place of ES organs (Blochlinger et al., 1988, 1990; Bodmer et al., 1987). The proneural protein Scute upregulates cut expression to promote ES formation, whereas Atonal blocks cut expression to produce CH precursors. Further, the action of Atonal on cut expression is dominant over that of Scute since ectopic Atonal expression in ES precursors blocks cut expression and converts them into CH organs (Jarman and Ahmed, 1998). This repression of cut by Atonal is the essential step in production of CH organs; expression of cut in CH precursors under an alternative, nonrepressible promoter causes CH‐to‐ES transformation (Blochlinger et al., 1991). These proneural activities generate a SOP cell competent to assume a neural fate. The steps that transform this cell into a mature mechanosense organ are known as “neurogenic” and for the ES‐type sense organs, these steps are well characterized. A defined and invariant series of divisions by the SOP gives rise to four final cell types—the mechanosensory neuron itself, the bristle‐ forming cell, the socket cell, which surrounds the bristle, and a supporting sheath cell. The ES SOP divides into two daughters, IIa and IIb; the IIa cell divides into the socket cell and the bristle cell, and the IIb divides into the
128
Beckingham et al.
neuron and the sheath cell. The Notch lateral inhibition system is again active in this pathway and cell fate largely rests on asymmetric distribution between daughter cells of critical determinants that will affect Notch activity. For example, asymmetric distribution of the protein Numb, which suppresses Notch activity, to the IIb cell allows this cell to resist lateral inhibition by the IIa cell and to assume a neural fate (Frise et al., 1996; Guo et al., 1996). numb knockout mutants show IIb‐to‐IIa lineage transformations, and ectopic Numb expression in IIa leads to the inverse result (Rhyu et al., 1994; Uemura et al., 1989). In the IIa cell, Notch activates the expression of tramtrack, a transcriptional repressor of neural fate (Guo et al., 1995, 1996). In contrast, less is known of the neurogenic steps in the formation of the CH class of sense organs. Like ES organs, the individual units (or scolopidia) of a CH organ often comprise four cells—a cap cell and a ligament cell, which provide mechanical linkage to the cuticle, a neuron, and an ensheathing scolopale cell. In the most studied example of this class, the larval body wall CH organ, the divisions by which these cells are derived have been characterized, but less is known about the genetic determinants acting in these divisions than for the ES lineages. It is known, however, that some of the same components, such as Numb, are used. The development of the CH organs of the adult has been little studied. Clearly additional regulatory steps must come into play in the development of complex structures, such as Johnston’s organ.
C. Genetics of mechanosensory signaling pathways in Drosophila Although uncertainty remains as to which mechanosensory organs mediate graviperception in Drosophila, the ability to perform directed genetic screening in this organism has meant that mutants specifically defective in mechanosensation have been identified. Thus, insight into the signaling pathways underlying mechanosensation in general, and presumably graviperception in particular, has been made. The first mechanosensory screen performed in Drosophila was designed to identify genes with roles in perception of light touch during the larval stages (Kernan et al., 1994). It yielded a number of mutations, of which at least two (uncoordinated [unc] and uncoordinated‐like [uncl]) appear to act in the ES class of mechanoreceptors. The electrophysiology of these organs is severely aberrant. The genes underlying these mutations have not been cloned as yet. Based on the discovery that adults defective in larval touch responses are severely uncoordinated, a second screen was executed, using this uncoordinated phenotype as the primary screening criterion (Kernan et al., 1994). An assay of the electrophysiological properties of the mechanosensory bristles was then used for second selection in the screening. Three loci at which mutations produced mechanosensory bristles with no transepithelial membrane potential (an indication that
4. Genetics of Graviperception in Animals
129
these sense organs are in essence physiologically “dead”) were identified and termed nomp (for no membrane potential) A, B and C. All three nomp genes have been cloned. nompA encodes a transmembrane protein with a large extracellular zona pellucida (ZP)‐domain‐containing region. Its closest mammalian homologs are alpha and beta tectorin, proteins found in the tectorial membrane that overlies the sensory cells of the mammalian cochlea (Chung et al., 2001). In Drosophila, NompA is a component of the dendritic cap—a structure that overlies the sensory cilium in CH organs, ES organs, and the mechanosensory bristles of the multimodal sensillae of the wing margin. In the nompA mutant the connection between the cilium and the dendritic cap, which presumably plays a role in mechanotranduction, is lost. NompB is actually a component (IFT88) of the intraflagellar transport (IFT) complex, a particle that plays a dynamic role in ciliary extension in many species (Han et al., 2003). The NompB protein is present in the cilia of all type I mechanosense organs in flies and also in cilia of olfactory neurons. The mutant phenotype is a loss of these cilia and hence of mechanotransduction. NompC proved to be a transmembrane ion channel, one of the more divergent members of the TRP family of channels (Walker et al., 2000). As discussed in Section II.C.2, there is good evidence that the zebrafish homolog of NompC is a mechanotransduction channel of the inner ear. It thus seems highly likely that NompC is a component of the initiating response in mechanosensation in Drosophila ES organs. However, small residual current in the nompC null mutant suggests that another channel also has a role. Like its zebrafish counterpart, Drosophila NompC has a long stretch of 29 ankyrin‐type repeats in its cytoplasmic domain. The novel hypothesis that these repeats could represent a gating spring for the channel was discussed above (Corey and Sotomayer, 2004). In vertebrates we have seen that many elements of signaling components for gravity overlap with those for audition and similar findings have emerged when mutants that affect mechanosensation and hearing in flies are compared. The unc, uncl, nompA and nompB mutations that affect function in the ES class of mechanoreceptors all affect the auditory response detected from Johnston’s organ—the complex array of CH organs present in the second antennal segment and responsible for detection of sound vibrations (Eberl et al., 2000). Two mutations that uniquely affect hearing (a CH organ function) and not ES‐type mechanoperception have also been identified (Eberl et al., 2000). One of these, tilB, was isolated in the larval touch screen described above and the other, beethoven (btv), by a specific screen for deafness (Eberl et al., 1997). Although both CH and ES organs involve cilia‐mediated mechanoperception, different mechanisms appear to be involved, with actual ciliary movement probably only occurring in CH organs. The btv and tilB gene products may have roles relating to this movement. In btv, features of the cilia structure are detectably altered.
130
Beckingham et al.
A further distinction between ES and CH type sense organs concerns the mechanosensory ion channels present. The observation that nompC mutant flies respond relatively well to sound initially suggested that the major mechanosensory channel in the CH chordotonal organs differs from that in ES organs and studies have confirmed this interpretation. The critical ion channel subunit is encoded by the gene Nanchung (Nan) and is a vanilloid‐receptor–related (TRPV) channel of the TRP superfamily (Kim et al., 2003). The founding member of this subfamily was OSM‐9, a protein expressed in ciliated neurons in C. elegans and involved in the “nose touch” response (see later section). The organization of the Nan protein is typical for the family, with five ankyrin repeats in the cytoplasmic tail of the protein and six transmembrane domains as the pore forming unit. Nan is expressed exclusively in the CH organs in the outer segments of the cilia, and nan mutants have morphologically normal CH cilia but are completely unresponsive to sound. The most studied mechanosensory channels in C. elegans are of the DEG/ENaC class (see Section IV.C.), a channel family that is widely distributed across evolution with roles in mechanoperception and fluid–electrolyte homeostasis. One Drosophila gene (called pickpocket1) encoding a channel subunit of this class is implicated in larval mechanoperception (Adams et al., 1998), but it functions in a class of sensory neurons (the multiple dendritic neurons) that seem unlikely to contribute to graviperception at this point.
D. Genetic insights from behavioral studies of gravitaxis
1. Population studies of gravitaxic behavior One advantage of the Drosophila system over that of the vertebrate models is that an assay for gravitaxis, involving use of a gravitaxic maze, was developed and refined many years ago. Although this assay was mainly used initially in population studies, it has been used in genetic screening as a direct route to identifying genes with roles in graviperception. During the late 1950s and early 1960s, a growing interest in the hereditary component of behavior provided the impetus for work by Hirsch et al. into the heritability of gravitaxic responses in Drosophila—work that continued well into the late 1990s. Hirsch (1959) designed a gravitaxic maze that tests flies for a preference between walking up or down at sequentially arranged choice points. Thus, the final exit point for any given fly from the maze represents the sum effect of its sequential choices. By screening large fly populations, reliable group behavioral scores could be achieved for relatively small differences in behavior. Starting in 1957 with a wild‐type population of Drosophila generated by mixing one lab strain and two wild strains, Hirsch et al. began selecting for flies
4. Genetics of Graviperception in Animals
131
that emerged at high or low exits from the mazes. They were able to show that “high” and “low” were heritable traits and after 48 generations a maximal separation of the high and low strains was achieved (Hirsch and Erlenmeyer‐ Kimling, 1961). These high and low lines, which had reached generation 770 by 1994 (Stoltenberg et al., 1994), have been maintained for over 40 years, and presently two versions of the high line (termed Hi1 and Hi5) and one of the low line (Lo1) are available. By generation 560, the lines were fixed so that their high or low maze behavior was maintained without continuous selection (Ricker and Hirsch, 1985). Nevertheless, some genetic variability was present, since reverse selection to get more positive gravitaxic behavior from the high lines, or vice versa, was still possible. However, by generation 770/771 reverse selection for the opposite behavior was not possible (Stoltenberg et al., 1994). Much of the work on these lines has been aimed at examining the contributions of individual chromosomes to the gravitaxic phenotype (Erlenmeyer‐ Kimling and Hirsch, 1961; Hirsch and Ksander, 1969; Ricker and Hirsch, 1988a,b). Initially, individual chromosome contributions were assessed by comparing (1) lines homozygous for a chromosome from a high or a low line with (2) lines heterozygous for such a chromosome and a dominantly marked balancer chromosome. To avoid the problems associated with this approach, such as assaying flies that have morphological abnormalities, more complex crossing schemes that yield stocks homozygous for a given chromosome from a high line or a low line or for a homozygous wild‐type chromosome (so‐called isochromosomal stocks) were used (e.g., Fig. 4.1, Ricker and Hirsch, 1988b). The general conclusion from these studies was that genes on all three of the major chromosomes of Drosophila (X, 2, and 3) contribute to the gravitaxic phenotype. In fact a very small effect of the Y chromosome was indicated in some lines by similar sorts of experiments (Stoltenberg and Hirsch, 1997). McGuire (1992) undertook an improved biometrical genetic reanalysis of the isochromosomal data from two prior studies and determined that most of the individual differences between lines resulted from additive, dominance, or epistatic effects between the three major chromosomes. He concluded that probably a small number of genes (with a lower limit of three genes) actually contributed to the gravitaxic phenotype in the lines examined. Allozyme analysis for the high and low lines led to the discovery that one of the determinants for “highness” maps within 1 centiMorgan of the alcohol dehydrogenase (ADH) gene. In examining the high and low lines for variants of 22 enzymes it was discovered that they were fixed for different versions of the ADH gene, with the high line being homozygous for the ADH‐S version and the low line homozygous for ADH‐F (Stoltenberg et al., 1995). F2 flies from two crosses between the high and low lines (high female low male [HL cross] and low female high male [LH cross], were allowed to breed for at least 66 generations. For each of the crosses, the maze behavior and
132
Beckingham et al.
ADH allozyme present were determined for individual flies (Stoltenberg and Hirsch, 1996). For the HL cross, the correlation between ADH‐S and high gravitaxic score was maintained, whereas for LH it was not. This allows the conclusion that the ADH gene itself does not influence gravitaxic behavior but that it is closely linked to a gene that does. Further, a rough approximation of the recombination frequency and hence of the genetic distance between ADH and this gene could be determined. Provocative insights have come from these population experiments on gravitaxic behavior. For example, as pointed out by Hirsch and Erlenmeyer‐ Kimling (1961), the notion of an instinctive invariant response in a wild‐type population is made less tenable by this work in that it proved possible to isolate “wild‐type” strains stably showing opposing responses to gravity. Beyond generation 515, the low line in fact stably reverted to a low gravitaxic score after reverse selection for higher scores (Ricker and Hirsch, 1985), suggesting that the “norm” for this population had been reset to a different response. However, this phenomenon might relate more to the severe bottlenecking reported for both the high and low strains over time (Ricker and Hirsch, 1988b). The alternative approach of directly seeking single genes with roles in Drosophila graviperception has received limited application to date. Markow and Merriam (1977) established the validity of this approach, however, by showing that several mutant lines, initially selected for altered behavior in a counter current selection paradigm, had altered behavior in the gravitaxic maze when compared to carefully prepared controls. McMillan and McGuire (1992) also tested individual mutations of four genes that affect antennal structure, based on the indications that the antenna plays a role in graviperception (see earlier section). All the mutations tested produced gravitaxic scores, which were significantly lower than the Canton‐S control strain. However the most dramatic effect was produced by the spineless‐aristapedia mutant (ssa), a mutant that results in homeotic transformation of antennal segments to leg structures; ssa flies have extremely low gravitaxic scores. Appropriate controls were performed to eliminate possible trivial explanations for this finding, leading McGuire to propose that the phenotype reflects a neuronal wiring problem, with the ectopic legs making inappropriate connections in the CNS that alter the overall perception of gravity.
2. A forward genetic screen for mutants defective in gravitaxis It is clear from the population genetics described in the previous section that the behavioral response to the gravity vector, like other behaviors, is very much a polygenic trait. Moreover, it seems likely that many genes affecting gravity responses will have roles in other behaviors. Dissecting this pleiotropy is difficult
4. Genetics of Graviperception in Animals
133
since behavioral assays, such as the gravitaxic maze require many aspects of locomotor behavior to be otherwise unaffected. With full cognizance of these limitations, we completed a large scale genetic screen to look for mutations that alter behavior in the gravitaxic mazes (Armstrong et al., 2005). We focused on mutants derived from a collection of GAL4 enhancer trap lines generated previously (Yang et al., 1995). Two features made this collection attractive. First, these lines were all known to express GAL4 in the nervous system, effectively offering a preselection towards mutations likely to affect perceptual activity; and, second, the GAL4 transposon in question permitted rapid cloning of its insertion site, and thus access to candidate affected genes. Analysis of performance in other behavioral paradigms (e.g., general locomotor activity, courtship, and flight) allowed us to select mutations with specific effects on the gravitaxic response. We have cloned out the insertion sites for 23 of these GAL4 transposon‐ induced mutations and in all but four cases, can tentatively identify the affected gene (Armstrong et al., 2005). Intriguingly several of the genes in question are previously studied genes with clear roles in neuronal signaling or axonal modeling throughout the organism. Our success in identifying mutations that selectively affect gravitaxic responses in these genes appears to derive from the kind of mutagen we have studied. On inspection, the insertion site of many of the transposons in our mutants can be seen to affect only particular transcripts or particular regulatory regions of the genes in question. Thus, it seems likely that these mutations affect only a subset of functions of the genes, in some cases perhaps in a small number of tissues that are directly involved in graviperception or graviresponses. Several novel genes have also been isolated, and these will be the focus of our continued work. To acknowledge our connection with NASA and our plans for microgravity experiments with these mutants, the genes in question are being named after astronauts. Intriguingly, one of these novel genes, named yuri gagarin, maps very close to ADH. As discussed in the previous section, one of the determinants of “highness” in the Hirsch high line was also shown to map close to ADH, and thus we have investigated the possibility that our GAL4 insertion mutation might affect the same gene. Interestingly, in both of Hirsch’s high lines, Hi1 and Hi5, there is a single amino acid change in the protein sequence of yuri; a proline residue that is part of a run of five such residues is mutated to a leucine in these lines. We have examined the Yuri protein sequence in 30 other lines, including the Hirsch low line, and do not see this amino acid change. It would seem therefore to be a rare polymorphism within Drosophila populations. At the population level, 1 out of every 10 individuals screened from three African populations (Zimbawe, Kenya, and Gabbon) and one American population (Philadelphia) recorded the amino acid polymorphism confirming its rare natural occurrence. However, further experiments do not support the hypothesis
134
Beckingham et al.
that this polymorphism is a determinant of “highness “in the Hirsch lines (Armstrong et al., 2005).
3. Genomics approaches to identifying genes with roles in gravitaxis With the advent of the genomic era, and complete sequencing of the Drosophila genome, microarray technology is being used to identify genes with roles in gravity responses. Toma et al. (2002) used the first generation of Drosophila cDNA microarray chips to identify genes showing differing expression in the Hirsch high and low strains described in the previous section. These array chips contained the Berkeley cDNA collection covering approximately one‐third of the Drosophila genome. They identified approximately 250 genes that showed a two‐fold or greater change in mRNA expression levels between the Hirsch high and low strains used. Several genes were then pursued using qPCR, which confirmed the differential expression levels suggested by the microarray analysis. Aberrant gravitaxic maze behavior was also uncovered for mutations in three (cryptochrome, Pendulin, Pigment‐dispersing factor) of the four genes for which mutants were available (a further gene, prospero, did not display a significant maze phenotype). It should be noted that all of the existing mutants uncovered by the functional genomics approaches employed by Toma et al. have significant additional behavioral or neurological defects. It is very likely that these defects would have complicated interpretation of the altered gravitaxic behavior if other, more traditional, approaches to behavioral genetics had been used. Bhattacharya’s group at NASA Ames (personal communication) has extended this functional genomics approach to the analysis of genes, which are affected by exposure to hypergravity conditions. They used Affymetrix gene chip technology covering the entire genome and report a similar number of genes to those reported by Toma et al. (2002) that respond to an increase in G‐force as simulated in a centrifuge system (experiments performed in wild‐type flies). They report that the significantly affected genes mostly differ from those reported by Toma (2002), and that they include (not unexpectedly) many stress response genes as well as many neuronal genes, which may mediate behavioral responses. Interestingly they also demonstrate a distinctive behavioral profile exhibited by flies on exposure to prolonged hypergravity conditions (3g). The locomotor activity levels of flies remain at normal levels immediately after the change to hypergravity. After prolonged exposure (30 min), locomotor activity ceases and the animals become inactive for a period that lasts a further 100 min. They term this change hypergravity‐induced quiescence (HiQ). Hypergravity‐ induced quiescence decays after 100 min or so with the activity levels returning
4. Genetics of Graviperception in Animals
135
Table 4.1. Known Drosophila genes with Roles in Gravity Responses Pathway/process Circadian rhythm
cAMP signaling
Neural development transcription factors Cell adhesion molecules Cell signaling/kinases
Gene cryptochrome period timeless Pigment‐dispersing factor dunce rutabaga forager broad escargot starry night Echinoid shaggy Connector of kinase to AP1 off‐track
Phenotype Array Array Array Array Array Array Array Maze Maze Array Array Array Maze Maze
(i), Maze (ii), HiQ (ii), HiQ (i) (ii) (ii) (ii)
(i) (ii) (i)
Example of Drosophila genes uncovered by forward genetics or functional genomics approaches. Microarray: (Array) (i) results from Toma et al. (2002), Array (ii) results from Bhattacharya, S. (personal communication). Behavioral phenotype information of mutants: maze results from Armstrong et al. (2005) except cryptochrome (Toma et al., 2002). Hypergravity induced quiescence (HiQ) from Bhattacharya, S. Note: This table summarizes selected, well‐known examples. The three studies cited list over 200 candidate genes with remarkably little overlap.
to normal levels despite the continued exposure to a hypergravity environment. A combination of mutant analysis and targeted gene knock‐outs was then used to link HiQ to the mushroom bodies of the central brain and to the circadian rhythm and cAMP molecular pathways. Examples of genes found by the forward genetic screens and the two functional genomics studies are listed in Table 4.1.
IV. THE NEMATODE MODEL SYSTEM (C. ELEGANS) A. Gravity‐sensing organs? To date, gravity perception in C. elegans has received limited attention, and thus the possible roles of its various classes of mechanosensory neurons in gravisensing have not been investigated. However, studies have demonstrated acute behavioral responses to hypergravity in this organism with microarray data indicating altered patterns of gene expression upon prolonged exposure to hypergravity (Hoffman, D., Udranszky, I., Kim, S., and Conley, C., unpublished observations). The existence of graviperception mechanisms in C. elegans is thus
136
Beckingham et al.
indicated. These findings offer the prospect of genetic screening for mutants with altered gravity‐induced behavior and thus for detailed analysis of gravity‐ based responses. In terms of identifying the relevant sensory structures mediating graviperception, an important advantage offered by C. elegans is the invariant cell lineages underlying cell fate determination (Sulston and Horvitz, 1977; Sulston et al., 1983). This highly programed development has allowed every sensory neuron in the adult body to be identified and named. Further, prior laser ablation studies (reviewed in Herman, 1996) to identify neurons involved in touch perception provide a perfect blueprint for equivalent studies aimed at identifying sensory neurons involved in graviperception. Given that gravity perception is always mediated via mechanical effects, it is possible that some of the neurons already shown to be involved in touch perception in C. elegans also transmit information on gravitational force. Mechanosensation in C. elegans has been reviewed by several authors (Ernstrom and Chalfie, 2002; Goodman and Schwarz, 2003; Herman, 1996; Syntichaki and Tavernarakis, 2003) and their work can be consulted for detailed information on this topic. Here we summarize findings relevant to the prospect of identifying a graviperception pathway in this species. Four touch responses have been investigated in C. elegans (“gentle touch,” “hard touch,” “nose touch” and “texture sensing” of surfaces). Several different sensory neuron types mediate these responses using differing sensory structures. For example, the three “nose touch” response classes of neurons all have ciliary extensions, whereas the single “gentle touch” class of neurons has a unique “mantle” attached by a fibrous organelle to the outer cuticle. Interestingly, one of the six cells of the “gentle touch” type, termed PVM, appears not to play a role in the touch response per se (Goodman and Schwarz, 2003) suggesting a distinct and separate function (gravity sensing??). Other cells that could also contribute to gravity sensing are present in this organism. In particular, several classes of motor neurons that also function as stretch receptors (Goodman and Schwarz, 2003) could play a role. Proprioceptive input from such cells could contribute to graviperception.
B. Genetic control of mechanosense organ development The C. elegans “gentle touch” response has been the subject of exhaustive mutant screening for genes with specific roles in the pathway. Four hundred and fifty mutants have been isolated that define seventeen complementation groups. Five of these genes have roles in determining the cell fate and differentiation of the six sensory neurons that mediate the response. MEC‐3, a homeodomain‐class transcription factor, acts early and is thought to be required for the formation of the six “gentle touch cells” and two further classes of mechanosensory neurons. Uncoordinated‐86 (UNC‐86), a POU‐class homeodomain protein, also functions early in the specification of these cells and
4. Genetics of Graviperception in Animals
137
can heterodimerize with MEC‐3, presumably acting with it to regulate transcription. The remaining three genes act through more indirect routes. For example, MEC‐8, has a role in regulating RNA splicing.
C. Mechanosensory signaling pathways The remaining 12 genes with roles in the “gentle touch” response are components of the sensory transduction mechanism itself. Given the exhaustive genetic screening performed, it seems likely that these 12 gene products represent all the specialized components necessary for this signaling pathway. Further, from their genetic interactions it has been possible to generate a detailed model for how these components generate the mechanotransduction event. The mechanotransduction ion channels themselves are composed of two subunit types (MEC‐4 and MEC‐10) that are members of the DEG/ENaC superfamily of amiloride‐sensitive channels. Two further proteins, MEC‐2 and MEC‐6, appear to be additional components of the channels. Three proteins encode specialized components of the extracellular matrix and channel gating is postulated to involve interaction of the channels with these extracellular proteins. The mechanotransduction event also involves cytoskeletal components within the neurons since two required genes encode unique tubulin variants that form a special 15‐element microtubule structure, which may directly interact with the channel complex. The roles of the other genes that are required for the response are less clear. The signal transduction pathways underlying the other “touch” responses in C. elegans are less well‐defined. Both the “nose touch” and “texture” response involve ciliated neurons and in both cases mechanotransduction channels of the TRP family are implicated; OSM‐9 is expressed in the ciliated head neurons that are activated on nose touching and the C. elegans homolog of NompC is expressed in neurons involved in texture perception. As for “gentle touch,” the sensory neurons underlying the “harsh touch” and “nose touch” responses have been delineated. Laser ablation studies to identify neurons that mediate gravity responses would therefore address the issue of potential overlap in function for any of these neuron types.
V. CONCLUSION The response of animals to gravity is a prime example of a naturally occurring complex behavior that is difficult to study using any single genetic approach. The underlying sensory mechanics alone present significant challenges for dissection using classical genetic techniques. Presenting an even larger challenge is the elucidation of gravity specific behavioral processes and the underlying neural
138
Beckingham et al.
substrates. This is largely due to such responses being directly affected by many other factors including locomotor capability, general neurophysiological processes and the involvement of other linked sensory modalities. However, as we have outlined here, combining different strategies into an integrative approach facilitates significant progress. Notably the emergence of functional genomic techniques when combined with the more mature forward genetics and population genetics approaches has facilitated very rapid advances in previous years, particularly in model organisms. Although we remain far from a full understanding of actual “geotaxis” it is already clear that we are soon to be faced with the additional complexity of long‐term changes to different gravity regimes in nonterrestrial environments. In the immediate future, long‐term culture of model organisms on the International Space Station will be possible, with survival on Mars as a real, although more distant goal.
Acknowledgments Studies from the authors’ laboratories were initiated under funding from a NASA Specialized Center of Research and Training (NSCORT) grant in Gravitational Biology at Rice University (to K.M.B.) and continued under NIH grant DC05164 (K.M.B.) with additional support from the Robert A. Welch Foundation (grant C‐1119 to K.M.B.). J. D.A and D.A.B are supported in the UK by the BBSRC (grant S18944 to J.D.A.).
References Adams, C. M., Anderson, M. G., Motto, D. G., Price, M. P., Johnson, W. A., and Welsh, M. J. (1998). Ripped Pocket and Pickpocket, novel Drosophila DEG/ENaC subunits expressed in early development and in mechanosensory neurons. J. Cell Biol. 140, 143–152. Alagramam, K. N., Murcia, C. L., Kwon, H. Y., Pawlowski, K. S., Wright, C. G., and Woychik, R. P. (2001). The mouse Ames waltzer hearing‐loss is caused by mutation of Pcdh15, a novel protocadherin. Nat. Genet. 27, 99–102. Andermann, P., Ungos, J., and Raible, D. W. (2002). Neurogenin 1 defines zebrafish cranial sensory ganglia precursors. Dev. Biol. 251, 45–58. Armstrong, J. D., Texada, M. J., Munjal, R., Baker, D. A., and Beckingham, K. M. (2005). Gravitaxis in Drosophila melanogaster: A forward genetic screen. Genes Brain Behav., in press. Avraham, K. B., Hasson, T., Steel, K. P., Kingsley, D. M., Russell, L. B., Mooseker, M. S., Copeland, N. G., and Jenkins, N. A. (1995). The mouse Snell’s waltzer deafness gene encodes an unconventional myosin required for structural integrity of inner ear hair cells. Nat. Genet. 11, 369–375. Baker, N. E., Yu, S., and Han, D. (1996). Evolution of proneural atonal expression during distinct regulatory phases in the developing Drosophila eye. Curr. Biol. 6, 1290–1301. Baonza, A., Casci, T., and Freeman, M. (2001). A primary role for the epidermal growth factor receptor in ommatidial spacing in the Drosophila eye. Curr. Biol. 11, 396–404. Bergstrom, R. A., You, Y., Erway, L. C., Lyon, M. F., and Schimenti, J. C. (1998). Deletion mapping of the head tilt (het) gene in mice: A vestibular mutation causing specific absence of otoliths. Genetics 150, 815–822.
4. Genetics of Graviperception in Animals
139
Bermingham, N. A., Hassan, B. A., Price, S. D., Vollrath, M. A., Nissim, B.‐A., Eatock, R. A., Bellen, H. J., Lysakowski, A., and Zoghbi, H. Y. (1999). Math1: An essential gene for the generation of inner ear hair cells. Science 284, 1837–1841. Blochlinger, K., Bodmer, R., Jack, J., Jan, L. Y., and Jan, Y. N. (1988). Primary structure and expression of a product from cut, a locus involved in specifying sensory organ identity in Drosophila. Nature 333, 629–635. Blochlinger, K., Bodmer, R., Jan, L. Y., and Jan, Y. N. (1990). Patterns of expression of cut, a protein required for external sensory organ development in wild‐type and cut mutant Drosophila embryos. Genes Dev. 4, 1322–1331. Blochlinger, K., Jan, L. Y., and Jan, Y. N. (1991). Transformation of sensory organ identity by ectopic expression of Cut in Drosophila. Genes Dev. 5, 1124–1135. Bodmer, R., Barbel, S., Sheperd, S., Jack, J. W., Jan, L. Y., and Jan, Y. N. (1987). Transformation of sensory organs by mutations of the cut locus of D. melanogaster. Cell 51, 293–307. Brigande, J. V., Kiernan, A. E., Gao, X., Iten, L. E., and Fekete, D. M. (2000). Molecular genetics of pattern formation in the inner ear: Do compartment boundaries play a role? Proc. Natl. Acad. Sci. USA 97, 11700–11706. Bryant, J., Goodyear, R. J., and Richardson, G. P. (2002). Sensory organ development in the inner ear: Molecular and cellular mehcanisms. Br. Med. Bull. 63, 39–57. Cabrera, C. V. (1990). Lateral inhibition and cell fate during neurogenesis in Drosophila: The interactions between scute, Notch and Delta. Development 110, 733–742. Cabrera, C. V., Martinez‐Arias, A., and Bate, M. (1987). The expression of three members of the achaete‐scute gene complex correlates with neuroblast segregation in Drosophila. Cell 50, 425–433. Caldwell, J. C., and Eberl, D. F. (2002). Towards a molecular understanding of Drosophila hearing. J. Neurobiol. 53, 172–189. Campos‐Ortega, J. A., and Knust, E. (1990). Molecular analysis of a cellular decision during embryonic development of Drosophila melanogaster: Epidermogenesis or neurogenesis. Eur. J. Biochem. 190, 1–10. Chen, P., Johnson, J. E., Zoghbi, H. Y., and Segil, N. (2002). The role of Math1 in inner ear development: Uncoupling the establishment of the sensoryprimordium from hair cell fate determination. Development 129, 2495–2505. Chung, Y. D., Zhu, J., Han, Y.‐G., and Kernan, M. J. (2001). nompA encodes a PNS‐specific, ZP domain protein required to connect mechanosensory dendrites to sensory structures. Neuron 29, 415–428. Corey, D. P., Garcia‐Anoveros, J., Holt, J. R., Kwan, K. Y., Lin, S.‐Y., Vollrath, M. A., Amalfitano, A., Cheung, E. L.‐M., Derfler, B. H., Duggan, A., Geloc, G. S., Gray, P. A., Hoffman, M. P., Rehm, H. L., Tamasauskas, D., and Zhang, D. S. (2004). TRPA1 is a candidate for the mechanosensitive transduction channel of vertebrate hair cells. Nature 432, 723–730. Corey, D. P., and Sotomayer, M. (2004). Tightrope act. Nature 428, 901–903. Curtin, J. A., Quint, E., Tsipouri, V., Arkell, R. M., Cattanach, B., Copp, A. J., Henderson, D. J., Spurr, N., Stanier, P., Fisher, E. M., Nolan, P. M., Steel, K. P., Brown, S. D., Gray, I. C., and Murdoch, J. N. (2003). Mutation of Celsr1 disrupts planar cell polarity of inner ear hair cells and causes severe neural tube defects in the mouse. Curr. Biol. 13, 1129–1133. Davis, W. J. (1971). The integrative action of the nervous system in crustacean equilibrium reactions. In “Gravity and the Organism” (S. A. Gordon and M. J. Cohen, eds.), pp. 237–250. University of Chicago, Chicago. Di Palma, F., Holme, R. H., Bryda, E. C., Belyantseva, I. A., Pellegrino, R., Kachar, B., Steel, K. P., and Noben‐Trauth, K. (2001). Mutations in Cdh23, encoding a new type of cadherin, cause stereocilia disorganization in waltzer, the mouse model for Usher syndrome type 1D. Nat. Genet. 27, 103–107.
140
Beckingham et al.
Dominguez, M., Wasserman, J. D., and Freeman, M. (1998). Multiple functions of the EGF receptor in Drosophila eye development. Curr. Biol. 8, 1039–1048. Eberl, D. F., Duyk, G. M., and Perrimon, N. (1997). A genetic screen for mutations that disrupt an auditory response in Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 94, 14837–14842. Eberl, D. F., Hardy, R. W., and Kernan, M. J. (2000). Genetically similar transduction mechanisms for touch and hearing in Drosophila. J. Neurosci. 20, 5981–5988. Ehmer, B., and Gronenberg, W. (1997). Proprioceptors and fast antennal reflexes in the ant Odontomachus (Formicidae, Ponerinae). Cell Tissue Res. 290, 153–165. Erlenmeyer‐Kimling, L., and Hirsch, J. (1961). Measurement of the relations between chromosomes and behavior. Science 134, 1068–1069. Ernest, S., Rauch, G. J., Haffter, P., Geisler, R., Petit, C., and Nicolson, T. (2000). Mariner is defective in myosin VIIA: A zebrafish model for human hereditary deafness. Hum. Mol. Genet. 9, 2189–2196. Ernstrom, G. G., and Chalfie, M. (2002). Genetics of sensory mechanotransduction. Ann. Rev. Genet. 36, 411–453. Erway, L. C., Fraser, A. S., and Hurley, L. S. (1971). Prevention of congenital otolith defect in pallid mutant mice by manganese supplementation. Genetics 67, 97–108. Fekete, D. M. (1996). Cell fate specification in the inner ear. Curr. Opin. Neurobiol. 6, 533–541. Frise, E., Knoblich, J. A., Younger‐Shepherd, S., Jan, L. Y., and Jan, Y. N. (1996). The Drosophila Numb protein inhibits signaling of the Notch receptor during cell‐cell interaction in sensory organ lineage. Proc. Natl. Acad. Sci. USA 93, 11925–11932. Frolenkov, G. I., Belyantseva, I. A., Friedman, T. B., and Griffith, A. J. (2004). Genetic insights into the morphogenesis of inner ear hair cells. Nat. Rev. Genet. 5, 489–498. Gibson, F., Walsh, J., Mburu, P., Varela, A., Brown, K. A., Antonio, M., Beisel, K. W., Steel, K. P., and Brown, S. D. (1995). A type VII myosin encoded by the mouse deafness gene shaker‐1. Nature 374, 62–64. Goodman, M. B., and Schwarz, E. M. (2003). Transducing touch in Caenorhabditis elegans. Ann. Rev. Physiol. 65, 429–452. Goulding, S. E., zur Lage, P., and Jarman, A. P. (2000). amos, a proneural gene for Drosophila olfactory sense organs that is regulated by lozenge. Neuron 25, 69–78. Granato, M., van Eeden, F. J. M., Schach, U., Trowe, T., Brand, M., Furutani‐Seiki, M., Haffter, P., Hammerschmidt, M., Heisenberg, C.‐P., Jiang, Y.‐J., Kane, D. A., Kelsh, R. N., Mullins, M. C., Odenthal, J., and Nusslein‐Volhard, C. (1996). Genes controlling and mediating locomotion behavior of the zebrafish embryo and larva. Development 123, 399–413. Guo, M., Bier, E., Jan, L. Y., and Jan, Y. N. (1995). tramtrack acts downstream of numb to specify distinct daughter cell fates during asymmetric cell divisions in the Drosophila PNS. Neuron 14, 913–925. Guo, M., Jan, L. Y., and Jan, Y. N. (1996). Control of daughter cell fates during asymmetric division: Interaction of Numb and Notch. Neuron 17, 27–41. Gupta, B. P., and Rodrigues, V. (1997). Atonal is a proneural gene for a subset of olfactory sense organs in Drosophila. Genes Cells 2, 225–233. Haddon, C., Mowbray, C., Whitfield, T. T., Jones, D., Gschmeissner, S., and Lewis, J. (1999). Hair cells without supporting cells: Further studies in the ear of the zebrafish mind bomb mutant. J. Neurocytol. 28, 837–850. Han, Y.‐G., Kwok, B. H., and Kernan, M. J. (2003). Intraflagellar transport is required in Drosophila to differentiate sensory cilia but not sperm. Curr. Biol. 13, 1679–1686. Hassan, B. A., and Bellen, H. J. (2000). Doing the MATH: Is the mouse a good model for fly development? Genes Dev. 14, 1852–1865. Haswell, E. S. (2003). Gravity perception: How plants stand up for themselves. Curr. Biol. 13, R761–R763.
4. Genetics of Graviperception in Animals
141
Herman, R. K. (1996). Touch sensation in Caenorhabditis elegans. BioEssays 18, 199–206. Hirsch, J. (1959). Studies in experimental behavior genetics: II. Individual differences in geotaxis as a function of chromosome variations in synthesized Drosophila populations. J. Comp. Physiol. Psychol. 52, 304–308. Hirsch, J., and Erlenmeyer‐Kimling, L. (1961). Sign of taxis as a property of the genotype. Science 134, 835–836. Hirsch, J., and Ksander, G. (1969). Studies in experimental behavior genetics: V. Negative geotaxis and further chromosome analyses in Drosophila melanogaster. J. Comp. Physiol. Psychol. 67, 118–122. Holt, J. R., Gillespie, S. K. H., Provance, D. W., Jr., Shah, K., Shokat, K. M., Corey, D. P., Mercer, J. A., and gillespie, P. G. (2002). A chemical‐genetic strategy implicates myosin‐1c in adaptation by hair cells. Cell 108, 371–381. Horn, E. (1975). Mechanisms of gravity processing by leg and abdominal gravity receptors in bees. J. Insect Physiol. 21, 673–679. Horn, E. (1982). Gravity reception in the walking fly, Calliphora erythrocephala: Tonic and modulatory influences of leg afferents on the head position. J. Insect Physiol. 28, 713–721. Horn, E. (1985). Gravity. In “Nervous System: Sensory” (G. A. Kerkut and L. I. Gilbert, eds.), pp. 557–576. Pergamon Press, New York. Horn, E., and Bischof, H.‐J. (1983). Gravity reception in crickets; the influence of cercal and antennal afferences on the head position. J. Comp. Physiol. 150, 93–98. Horn, E., and Foller, W. (1985). Tonic and modulatory subsystems of the complex gravity receptor system in crickets, Gryllus bimaculatus. J. Insect Physiol. 31, 937–946. Horn, E., and Kessler, W. (1975). The control of antennae lift movements and its importance on the gravity reception in the walking blowfly Calliphora erythrocephala. J. Comp. Physiol. 97, 189–203. Hoson, T., and Soga, K. (2003). New aspects of gravity responses in plant cells. Int. Rev. Cytol. 229, 209–244. Hrabe de Angelis, M., Flaswinkel, H., Fuchs, H., Rathkolb, B., Soewarto, D., Marschall, S., Heffner, S., Pargent, W., Wuensch, K., Jung, M., Reis, A., Richter, T., Alessandrini, F., Jakob, T., Fuchs, E., Kolb, H., Kremmer, E., Schaeble, K., Rollinski, B., Roscher, A., Peters, C., Meitinger, T., Strom, T., Steckler, T., Holsboer, F., Klopstock, T., Gekeler, F., Schindewolf, C., Jung, T., Avraham, K., Behrendt, H., Ring, J., Zimmer, A., Schughart, K., Pfeffer, K., Wolf, E., and Balling, R. (2000). Genome‐wide, large‐scale production of mutant mice by ENU mutagenesis. Nat. Genet. 25, 444–447. Huang, M. L., Hsu, C. H., and Chien, C. T. (2000). The proneural gene amos promotes multiple dendritic neuron formation in the Drosophila peripheral nervous system. Neuron 25, 57–67. Hurle, B., Ignatova, E., Massironi, S. M., Mashimo, T., Rios, X., Thalmann, I., Thalmann, R., and Ornitz, D. M. (2003). Non‐syndromic vestibular disorder with otoconial agenesis in tilted/mergulhador mice caused by mutations in otopetrin 1. Hum. Mol. Genet. 12, 777–789. Itoh, M., Kim, C. H., Palardy, G., Oda, T., Jiang, Y. J., Maust, D., Yeo, S. Y., Lorick, K., Wright, G. J., Ariza‐McNaughton, L., Weissman, A. M., Lewis, J., Chandrasekharappa, S. C., and Chitnis, A. B. (2003). Mind bomb is a ubiquitin ligase that is essential for efficient activation of Notch signaling by Delta. Dev. Cell 4, 67–82. Jarman, A. P., and Ahmed, I. (1998). The specificity of proneural genes in determining Drosophila sense organ identity. Mech. Dev. 76, 117–125. Jarman, A. P., Grau, Y., Jan, L. Y., and Jan, Y. N. (1993). Atonal is a proneural gene that directs chordotonal organ formation in the Drosophila peripheral nervous system. Cell 73, 1307–1321. Jarman, A. P., Grell, E. H., Ackerman, L., Jan, L. Y., and Jan, Y. N. (1994). Atonal is the proneural gene for Drosophila photoreceptors. Nature 369, 398–400. Jarman, A. P., Sun, Y., Jan, L. Y., and Jan, Y. N. (1995). Role of the proneural gene, atonal, in formation of Drosophila chordotonal organs and photoreceptors. Development 121, 2019–2030.
142
Beckingham et al.
Kappler, J. A., Starr, C. J., Chan, D. K., Kollmar, R., and Hudspeth, A. J. (2004). A nonsense mutaton in the gene encoding a zebrafish myosin VI isoform causes defects in hair‐cell mechanotransduction. Proc. Natl. Acad. Sci. USA 101, 13056–13061. Keil, T. A. (1997). Functional morphology of insect mechanoreceptors. Microsc. Res. Tech. 39, 506–531. Kernan, M. J., Cowan, D., and Zuker, C. S. (1994). Genetic dissection of mechanosensory transduction: Mechanoreception‐defective mutations of Drosophila. Neuron 12, 1195–1206. Kim, J., Chung, Y. D., Park, D.‐Y., Choi, S., Shin, D. W., Soh, H., Lee, H. W., Son, W., Yim, J., Park, C.‐S., et al. (2003). A TRPV family ion channel required for hearing in Drosophila. Nature 424, 81–84. Kozel, P. J., Friedman, R. A., Erway, L. C., Yamoah, E. N., Liu, L. H., T., R., Duffy, J. J., Doetschman, T., Miller, M. L., Cardell, E. L., et al. (1998). Balance and hearing deficits in mice with a null mutation in the gene encoding plasma membrane Ca2þ ‐ATPase isoform 2. J. Biol. Chem. 273, 18693–18696. Kurima, K., Peters, L. M., Yang, Y., Riazuddin, S., Ahmed, Z. M., Naz, S., Arnaud, D., Drury, S., Mo, J., Makishima, T., Ghosh, M., Menon, P. S., Deshmukh, D., Oddoux, C., Ostrer, H., Khan, S., Riazuddin, S., Deininger, P. L., Hampton, L. L., Sullivan, S. L., Battey, J. F., Jr., Keats, B. J., Wilcox, E. R., Friedman, T. B., and Griffith, A. J. (2002). Dominant and recessive deafness caused by mutations of a novel gene, TMC1, required for cochlear hair‐cell function. Nat. Genet. 30, 277–284. Lai, E. C. (2003). Drosophila tufted is a gain‐of‐function allele of the proneural gene amos. Genetics 163, 1413–1425. Lee, E. C., Hu, X., Yu, S. Y., and Baker, N. E. (1996). The scabrous gene encodes a secreted glycoprotein dimer and regulates proneural development in Drosophila eyes. Mol. Cell. Biol. 16, 1179–1188. Lim, D. J., Erway, L. C., and Clark, D. L. (1978). Tilted‐head mice with genetic otoconial anomaly. Behavioural and morphological correlates. In “Vestibular Mechanisms in Health and Disease” (J. D. Hood, ed.), pp. 195–206. Academic press, London. Lyon, M. F. (1952). Absence of otoliths in the mouse: An effect of the pallid mutant. J. Genet. 51, 638–650. Lyon, M. F., and Meredith, R. (1969). Muted, a new mutant affecting coat colour and otoliths of the mouse, and its position in linkage group XIV. Genet. Res. 14, 163–166. Markl, H. (1971). Proprioceptive gravity perception in Hymenoptera. In “Gravity and the Organisms” (S. A. Gordon and M. J. Cohen, eds.), pp. 185–194. The University of Chicago Press, Chicago. Markow, T. A., and Merriam, J. (1977). Phototactic and geotactic behavior of countercurrent defective mutants of Drosophila melanogaster. Behav. Genet. 7, 447–455. McGuire, T. R. (1992). A biometrical genetic approach to chromosome analysis in Drosophila: detection fo epistatic interactions in geotaxis. Behav. Genet 22, 453–467. McMillan, P. A., and McGuire, T. R. (1992). The homeotic gene spineless‐aristapedia affects geotaxis in Drosophila melanogaster. Behav. Genet. 22, 557–573. Mlodzik, M., Baker, N. E., and Rubin, G. M. (1990). Isolation and expression of scabrous, a gene regulating neurogenesis in Drosophila. Genes Dev. 4, 1848–1861. Nicolson, T., Rusch, A., Friedrich, R. W., Granato, M., Ruppersburg, J. P., and Nusslein‐Volhard, C. (1998). Genetic analysis of vertebrate sensory hair cell mechonsensation: The zebra fish circler mutants. Neuron 20, 271–283. Nolan, P. M., Peters, J., Strivens, M., Rogers, D., Hagan, J., Spurr, N., Gray, I. C., Vizor, L., Brooker, D., Whitehill, E., Washbourne, R., Hough, T., Greenaway, S., Hewitt, M., Liu, X., McCormack, S., Pickford, K., Selley, R., Wells, C., Tymowska‐Lalanne, Z., Roby, P., Glenister, P., Thornton, C., Thaung, C., Stevenson, J. A., Arkell, R., Mburu, P., Hardisty, R., Kiernan, A., Erven, A.,
4. Genetics of Graviperception in Animals
143
Steel, K. P., Voegeling, S., Guenet, J. L., Nickols, C., Sadri, R., Nasse, M., Isaacs, A., Davies, K., Browne, M., Fisher, E. M., Martin, J., Rastan, S., Brown, S. D., and Hunter, J. (2000). A systematic, genome‐wide, phenotype‐driven mutagenesis programme for gene function studies in the mouse. Nat. Genet. 25, 440–443. Ornitz, D. M., Bohne, B. A., Thalmann, I., Harding, G. W., and Thalmann, R. (1998). Otoconial agenesis in tilted mutant mice. Hearing Res. 122, 60–70. Ozaki, H., Nakamura, K., Funahashi, J., Ikeda, K., Yamada, G., Tokano, H., Okamura, H. O., Kitamura, K., Muto, S., Kotaki, H., Sudo, K., Horai, R., Iwakura, Y., and Kawakami, K. (2004). Six1 controls patterning of the mouse otic vesicle. Development 131, 551–562. Pisam, M., Jammet, C., and Laurent, D. (2002). First steps in otolith formation of the zebrafish: Role of glycogen? Cell Tissue Res. 310, 163–168. Pistillo, D., Skaer, N., and Simpson, P. (2002). scute expression in Calliphora vicina reveals an ancestral pattern of longitudinal stripes on the thorax of higher Diptera. Development 129, 563–572. Portin, P. (2002). General outlines of the molecular genetics of the Notch signalling pathway in Drosophila melanogaster: A review. Hereditas 136, 89–96. Pourquie´ , O., Corbel, C., Le Caer, J.‐P., Rossier, J., and Le Douarin, N. M. (1992). BEN, a surface glycoprotein of the immunoglobulin superfamily, is expressed in a variety of developing systems. Proc. Natl. Acad. Sci. USA 89, 5261–5265. Powell, P. A., Wesley, C., Spencer, S., and Cagan, R. L. (2001). Scabrous complexes with Notch to mediate boundary formation. Nature 409, 626–630. Rahmann, H., and Anken, R. H. (2000). Gravitational neurobiology of fish. Adv. Space Res. 25, 1985–1995. Rhyu, M. S., Jan, L. Y., and Jan, Y. N. (1994). Asymmetric distribution of numb protein during division of the sensory organ precursor cell confers distinct fates to daughter cells. Cell 76, 477–491. Ricker, J. P., and Hirsch, J. (1985). Evolution of an instinct under long‐term divergent selection for geotaxis in domesticated populations of Drosophila melanogaster. J. Comp. Psychol. 1985, 380–390. Ricker, J. P., and Hirsch, J. (1988a). Genetic changes occurring over 500 generations of lines of Drosophila melanogaster selected divergently for geotaxis. Behav. Genet. 18, 13–25. Ricker, J. P., and Hirsch, J. (1988b). Reversal of genetic homeostasis in laboratory populations of Drosophila melanogaster under long‐term selection for geotaxis and estimates of gene correlates: evolution of behavior‐genetic systems. J. Comp. Psychology 102, 203–214. Riley, B. B. (2003). Genes controlling the development of the zebrafish inner ear and hair cells. Curr. Topics Dev. Biol. 57, 357–388. Riley, B. B., Chiang, M.‐Y., Farmer, L., and Heck, R. (1999). The delta A gene of zebrafish mediates lateral inhibition of hair cells in the inner ear and is regulated by pax 2. 1. Development 126, 5669–5678. Riley, B. B., and Grunwald, D. J. (1995). Efficient induction of point mutations allowing recovery of specific locus mutations in zebra fish. Proc. Natl. Acad. Sci. USA 92, 5997–6001. Riley, B. B., and Grunwald, D. J. (1996). A mutation in zebrafish affecting a localized cellular function is required for normal ear development. Dev. Biol. 179, 427–435. Riley, B. B., and Moorman, S. J. (2000). Development of utricular otoliths, but not saccular otoliths, is necessary for vestibular function and survival in zebrafish. J. Neurobiol. 43, 329–337. Roberts, P. (1961). Bristle formation controlled by the achaete locus in genetic mosaics of Drosophila melanogaster. Genetics 46, 1241–1243. Schone, H. (1971). Gravity Receptors and Gravity Orientation in Crustacea. In “Gravity and the Organism” (S. A. Gordon and M. J. Cohen, eds.), pp. 223–235. The University of Chicago Press, Chicago. Seiler, C., Ben‐David, O., Sidi, S., O., H., Rusch, A., B., B., Avraham, K. B., and Nicolson, T. (2004). Myosin VI is required for structural integrity of the apical surface of sensory hair cells in zebrafish. Dev. Biology 272, 328–338.
144
Beckingham et al.
Seiler, C., and Nicolson, T. (1999). Defective calmodulin‐dependent rapid apical endocytosis in zebrafish sensory hair cell mutants. J. Neurobiol. 41, 424–434. Sidi, S., Busch‐Nentwich, E., Friedrich, R., Schoenberger, U., and Nicolson, T. (2004). gemini encodes a zebrafish L‐type calcium channel that localizes at sensory hair cell ribbon synapses. J. Neurosci. 24, 4213–4223. Sidi, S., Friedrich, R. W., and Nicolson, T. (2003). NompC TRP channel required for vertebrate sensory hair cell mechanotransduction. Science 301, 96–99. Siemens, J., Lillo, C., Dumont, R. A., A., R., Williams, D. S., Gillespie, P. G., and Muller, U. (2004). Cadherin 23 is a component of the tip link in hair‐cell stereocilia. Nature 428, 950–954. Simmler, M. C., Cohen‐Salmon, M., El‐Amraoui, A., Guillaud, L., Benichou, J. C., Petit, C., and Panthier, J. J. (2000). Targeted disruption of Otog results in deafness and severe imbalance. Nat. Genet. 24, 139–143. Skaer, N., Pistillo, D., Gibert, J. M., Lio, P., Wulbeck, C., and Simpson, P. (2002). Gene duplication at the achaete‐scute complex and morphological complexity of the peripheral nervous system in Diptera. Trends Genet. 18, 399–405. Sollner, C., Burghammer, M., Busch‐Nentwich, E., Berger, J., Schwarz, H., Riekel, C., and Nicolson, T. (2003). Control of crystal size and lattice formation by Starmaker in otolith biomineralization. Science 302, 282–286. Sollner, C., Rauch, G.‐J., Siemens, J., Geisler, R., Schuster, S. C., Consortium, t.T. S., Muller, U., and Nicolson, T. (2004). Mutations in cadherin 23 affect tip links in zebrafish sensory hair cells. Nature 428, 955–959. Sollner, C., Schwarz, H., Geisler, R., and Nicolson, T. (2004). Mutated otopetrin 1 affects the genesis of otoliths and the localization of Starmaker in zebrafish. Dev. Genes Evolution 214, 582–590. Stoltenberg, S. F., Chopra, P., Wang, J., and Hirsch, J. (1994). Replicating reverse selection in lines of D. melanogaster that have evolved stable, extreme geotaxis performance. Drosophila Information Service 75, 162–163. Stoltenberg, S. F., and Hirsch, J. (1996). A gene correlate of geotaxis near Adh (2‐50. 1) in Drosophila melanogaster. J. Comp. Psychol. 110, 252–259. Stoltenberg, S. F., and Hirsch, J. (1997). Y‐chromosome effects on Drosophila geotaxis interact with genetic or cytoplasmic background. Anim. Behav. 53, 853–864. Stoltenberg, S. F., Hirsch, J., and Berlocher, S. H. (1995). Analyzing correlations of three types in selected lines of Drosophila melanogaster that have evolved stable extreme geotactic performance. J. Comp. Psychol. 109, 85–94. Sulston, J. E., and Horvitz, H. R. (1977). Post embryonic cell lineages of the nematode Caenorhabditis elegans. Dev. Biol. 56, 110–156. Sulston, J. E., Schierenberg, E., White, J. G., and Thomson, J. N. (1983). The embryonic cell lineage of the nematode Caenorhabditis elegans. Dev. Biol. 100, 64–119. Sumanas, S., Larson, J. D., and Bever, M. M. (2003). Zebrafish chaperone protein GP96 is required for otolith formation during ear development. Dev. Bio. 261, 443–455. Sun, Y., Jan, L. Y., and Jan, Y. N. (1998). Transcriptional regulation of atonal during development of the Drosophila peripheral nervous system. Development 125, 3731–3740. Syntichaki, P., and Tavernarakis, N. (2003). Genetic models of mechanotransduction: The nematode Caenorhabditis elegans. Physiol. Rev. 84, 1097–1153. Toma, D. P., White, K. P., Hirsch, J., and Greenspan, R. J. (2002). Identification of genes involved in Drosophila melanogaster geotaxis, a complex behavioral trait. Nat. Genet. 31, 349–353. Uemura, T., Shepherd, S., Ackerman, L., Jan, L. Y., and Jan, Y. N. (1989). numb, a gene required in determination of cell fate during sensory organ formation in Drosophila embryos. Cell 58, 349–360.
4. Genetics of Graviperception in Animals
145
Vervoort, M., and Ledent, V. (2001). The evolution of the neural basic helix‐loop‐helix proteins. Scientific World J. 1, 396–426. Villa‐Cuesta, E., de Navascues, J., Ruiz‐Gomez, M., Diez del Corral, R., Dominguez, M., de Celis, J. F., and Modolell, J. (2003). Tufted is a gain‐of‐function allele that promotes ectopic expression of the proneural gene amos in Drosophila. Genetics 163, 1403–1412. Villares, R., and Cabrera, C. V. (1987). The achaete‐scute gene complex of D. melanogaster: Conserved domains in a subset of genes required for neurogenesis and their homology to myc. Cell 50, 415–424. Walker, R. G., Willingham, A. T., and Zuker, C. S. (2000). A Drosophila mechanosensory transduction channel. Science 287, 2229–2234. Wang, Y., Kowalski, P. E., Thalmann, I., Ornitz, D. M., Mager, D. L., and Thalmann, R. (1998). Otoconin‐90, the mammalian otoconial matrix protein, contains two domains of homology to secretory phospholipase A2. Proc. Natl. Acad. Sci. USA. 95, 15345–15350. Wendler, G. (1971). Gravity orientation in insects: The role of different mechanoreceptors. In “Gravity and the Organism” (S. A. Gordon and M. J. Cohen, eds.), pp. 195–201. The University of Chicago Press, Chicago. Wheeler, S. R., Carrico, M. L., Wilson, B. A., Brown, S. J., and Skeath, J. B. (2003). The expression and function of the achaete‐scute genes in Tribolium castaneum reveals conservation and variation in neural pattern formation and cell fate specification. Development 130, 4373–4381. Whitfield, T. T. (2002). Zebrafish as a model for hearing and deafness. J. Neurob. 53, 157–171. Whitfield, T. T., Granato, M., van Eeden, F. J. M., Schach, U., Brand, M., Furutani‐Seiki, M., Haffter, P., Hammerschmidt, M., Heisenberg, C.‐P., Jiang, Y.‐J., Kane, D. A., Kelsh, R. N., Mullins, M. C., Odenthal, J., and Nusslein‐Volhard, C. (1996). Mutations affecting development of the zebrafish inner ear and lateral line. Development 123, 241–254. Whitfield, T. T., Riley, B. B., Chiang, M.‐Y., and Phillips, B. (2002). Development of the zebrafish inner ear. Dev. Dyn. 223, 427–458. Wilson, D. M. (1971). Stabilizing mechanisms in insect flight. In “Gravity and the Organism” (S. A. Gordon and M. J. Cohen, eds.), pp. 169–176. University of Chicago, Chicago. Xheng, J. L., Shou, J., Guillemot, F., Kageyama, F., and Gao, W.‐Q. (2000). Hes1 is a negative regulator of inner ear hair cell differentiation. Development 127, 4551–4560. Yang, M. Y., Armstrong, J. D., Vilinsky, I., Strausfeld, N. J., and Kaiser, K. (1995). Subdivision of the Drosophila mushroom bodies by enhancer‐trap expression patterns. Neuron 15, 45–54. Young, S. S., and Lewontin, R. C. (1966). Differences in bristle‐making abilities in scute and wild‐ type Drosophila melanogaster. Genet. Res. 7, 295–301. Zheng, L., Sekerkova, G., Vranich, K., Tilney, L. G., Mugnaini, E., and Bartles, J. R. (2000). The deaf jerker mouse, has a mutation in the gene encoding the espin actin‐bundling proteins of hair cell stereocilia and lacks espins. Cell 102, 377–385. Zine, A., Aubert, A., Qiu, J., Therianos, S., Guillemot, F., Kageyama, R., and Ribaupierre, D. F. (2001). Hes1 and Hes5 activities are required for the normal development of the hair cells in the mammalian inner ear. J. Neurosci. 21, 4712–4720.
5
Retroviral DNA Integration— Mechanism and Consequences Mary K. Lewinski* and Frederic D. Bushman{ *Infectious Disease Laboratory, The Salk Institute for Biological Studies La Jolla, California 92186 { Department of Microbiology, University of Pennsylvania School of Medicine Philadelphia, Pennsylvania 19104
I. Introduction II. Retroviral Life Cycle A. Integrase is essential B. Phenotypes of integrase mutants III. Mechanism of Integration A. DNA breaking and joining reactions catalyzed by integrase B. Host factors involved late in the integration reaction IV. Integrase Structure A. N-terminal domain B. Catalytic core C. C-terminal domain D. Structures containing two domains of integrase E. Multimerization V. Composition of Integrase Complexes In Vivo A. Viral proteins B. Host proteins VI. Retroviral Integration Targeting A. Preferred DNA sequence and structure B. Genome-wide studies C. Tethering factors VII. Consequences of Integration into Host Chromosomes A. Insertional mutagenesis B. Latency Advances in Genetics, Vol. 55 Copyright 2005, Elsevier Inc. All rights reserved.
0065-2660/05 $35.00 DOI: 10.1016/S0065-2660(05)55005-3
Lewinski and Bushman
148 VIII. Conclusions References
ABSTRACT Integration of retroviral cDNA into the host cell chromosome is an essential step in its replication. This process is catalyzed by the retroviral integrase protein, which is conserved among retroviruses and retrotransposons. Integrase binds viral and host DNA in a complex, called the preintegration complex (PIC), with other viral and cellular proteins. While the PIC is capable of directing integration of the viral DNA into any chromosomal location, different retroviruses have clear preferences for integration in or near particular chromosomal features. The determinants of integration site selection are under investigation but may include retrovirus-specific interactions between integrase and tethering factors bound to the host cell chromosomes. Research into the mechanisms of retroviral integration site selection has shed light on the phenomena of insertional mutagenesis and viral latency. ß 2005, Elsevier Inc.
I. INTRODUCTION The retroviruses are of utmost significance in biology and medicine. The propensity of some retroviruses to cause cancer in animals has led to the discovery of numerous oncogenes. Human immunodeficiency virus, a member of the retroviral genus Lentivirus, is responsible for one of the most destructive pandemics in human history. Here we discuss how a process central to the retroviral replication cycle (i.e., integration of viral DNA into the host cell chromosome) mediates some of the consequences of retroviral infection.
II. RETROVIRAL LIFE CYCLE Following binding of the retroviral envelope glycoprotein to its cellular receptor(s), the viral membrane fuses with the cell membrane, releasing the viral core into the host cell cytoplasm. The viral genomic RNA is then reverse transcribed to form double-stranded DNA. The viral DNA, in a complex with integrase and other viral and cellular proteins, enters the nucleus. There, the viral integrase protein covalently joins the viral DNA to the host cell DNA. Once integrated, the viral DNA, called the “provirus,” acts as a transcription template for efficient synthesis of viral mRNA and genomic RNA. Viral proteins are translated and
5. Retroviral DNA Integration—Mechanism and Consequences
149
assemble with the viral genomic RNA. These new virions then bud from the host cell membrane, mature, and infect new host cells. The provirus persists indefinitely in the host cell chromosome and is inherited by daughter cells like any other cellular gene during cell division.
A. Integrase is essential For retroviral DNA to efficiently direct production of progeny virions, it must become covalently integrated into the host cell chromosome (Coffin et al., 1997; Hansen et al., 1998). Some expression from unintegrated viral DNA can be detected (Panganiban and Temin, 1983), but this is not sufficient to sustain a spreading infection (Engelman et al., 1995; Englund et al., 1995). Analyses of mutants have identified the viral integrase coding region (part of the retroviral pol gene) as essential for the integration process (Donehower, 1988; Donehower and Varmus, 1984; Panganiban and Temin, 1984; Quinn and Grandgenett, 1988; Schwartzberg et al., 1984). Also essential are regions at the ends of the viral long terminal repeats (LTRs) that serve as recognition sites for integrase protein (Colicelli and Goff, 1985, 1988; Panganiban and Temin, 1983).
B. Phenotypes of integrase mutants Extensive work has shown that integrase mutants can have a variety of effects on viral replication. Integrase mutants containing substitutions in the enzyme active site (considered in a later section) generally have effects only at the integration step in the viral life cycle. However, integrase may play additional roles in viral replication, perhaps as a structural component of replication intermediates. Integrase is present as part of the retroviral gag-pol polyprotein during assembly and budding and is present in reverse transcription complexes after infection of new host cells (Fassati and Goff, 1999; Nermut and Fassati, 2003). Many mutants of integrase, including deletion mutants, can have pleiotropic effects on the viral life cycle, including effects on particle budding, infectivity, and reverse transcription (Engelman et al., 1995).
III. MECHANISM OF INTEGRATION Integration of viral DNA into the host cell chromosome involves several coordinated steps (i.e., processing of the viral DNA ends, joining of those ends to target DNA, and repairing the gaps). The first two reactions are catalyzed by the viral integrase protein, whereas the last is mediated by as-yet-undefined factors.
150
Lewinski and Bushman
A. DNA breaking and joining reactions catalyzed by integrase The viral genomic RNA is reverse transcribed to form a linear double-stranded DNA molecule, the precursor to the integrated provirus (Brown et al., 1987, 1989; Fujiwara and Mizuuchi, 1988). The provirus is co-linear with unintegrated linear viral DNA (Dhar et al., 1980; Hughes et al., 1978) but differs from the reverse transcription product in that it is missing two (or for some retroviruses, three) bases from each end (Hughes et al., 1981). Flanking the integrated provirus are direct repeats of the cellular DNA that are usually 4–6 base pairs in length, depending on the virus (Hughes et al., 1981; Vincent et al., 1990). This duplication of cellular sequences flanking the viral DNA is generated as a consequence of the integration mechanism (Coffin et al., 1997). Linear viral DNA is found in a complex with proteins in the cytoplasm of infected cells. These complexes (termed “PICs”) can be isolated and have been shown to mediate integration of viral DNA into target DNA in vitro (Bowerman et al., 1989; Brown et al., 1987; Ellison et al., 1990; Farnet and Haseltine, 1990, 1991). The development of in vitro assays with purified integrase has allowed its enzymatic functions to be elucidated. The provirus is the result of two reactions catalyzed by the viral integrase (i.e., terminal cleavage and strand transfer). Studies with purified integrase have shown that it is sufficient for both 30 end cleavage (Bushman and Craigie, 1991; Craigie et al., 1990; Katzman et al., 1989; Sherman and Fyfe, 1990) and joining of the viral DNA to the cellular chromosome or naked target DNA (Bushman et al., 1990; Craigie et al., 1990; Katz et al., 1990). Most integrase proteins catalyze the removal of two bases from the 30 end of each viral DNA strand, leaving recessed 30 hydroxyl groups (Brown et al., 1989; Fujiwara and Mizuuchi, 1988; Roth et al., 1989; Sherman and Fyfe, 1990). This terminal cleavage reaction is required for proper integration. It may allow the virus to create a standard end from viral DNA termini that can be heterogeneous due to the terminal transferase activity of reverse transcriptase (Miller et al., 1997; Patel and Preston, 1994). In addition, the terminal cleavage step is coupled to the formation of a stable integrase–DNA complex (Ellison and Brown, 1994; Vink et al., 1994). Following terminal cleavage, a recessed hydroxyl is exposed that immediately follows a CA dinucleotide. This CA is conserved among retroviruses and many related transposons. Evidence suggests that more internal LTR sites are also important for integration (Balakrishnan and Jonsson, 1997; Bushman and Craigie, 1990; Leavitt et al., 1992). After end processing, integrase catalyzes the covalent attachment of hydroxyl groups at the viral DNA termini to protruding 50 phosphoryl ends of the host cell DNA (Brown et al., 1987, 1989; Fujiwara and Mizuuchi, 1988). The DNA cleavage and joining reactions involved in integration are shown in Fig. 5.1A. Both the viral DNA 30 end cleavage and strand transfer reactions are mediated by
5. Retroviral DNA Integration—Mechanism and Consequences
151
single-step transesterification chemistry as shown by stereochemical analysis of the reaction course (Engelman et al., 1991). Purified integrase can also catalyze the “reverse” of the strand transfer reaction, termed disintegration (Chow et al., 1992) (see Fig. 5.1B). Assays for disintegration activity have been useful in the analysis of defective integrase mutants because the requirements for disintegration seem to be more lenient than those for integration. Biochemical analysis of purified integrase revealed that it requires a divalent metal, either Mg2þ or Mn2þ, to carry out reactions with model substrates (Chow et al., 1992). As discussed in later section, several structures of integrase show a divalent metal bound at the active site. Modeling suggests that two cations at the active site are important, the second of which is likely carried to the active site by the DNA substrate (Bujacz et al., 1997; Lins et al., 2000). A report detailing Cys substitutions at human immunodeficiency virus type 1 (HIV-1) integrase active site residues D64 and D116 suggested that these residues act by binding divalent metal (Gao et al., 2004). Divalent metal is also involved in assembly and stabilization of integrase–DNA complexes (Bujacz et al., 1997; Gao et al., 2004; Hazuda et al., 1997; Lee et al., 1995; Yi et al., 1999).
B. Host factors involved late in the integration reaction Integrase carries out the terminal cleavage and strand transfer steps that initiate viral DNA integration. Integration of both ends of the viral DNA, followed by melting of the target DNA segments between the points of joining, yields singlestranded gaps at each host–virus DNA junction and a two base overhang derived from the viral DNA (Fig. 5.1A). The manner by which this intermediate is subsequently repaired to yield the fully integrated provirus is unclear. For many parasitic DNA replication reactions, the parasite carries out reaction steps only up to a point that the host cannot easily reverse, forcing the host to complete the job (Bushman, 2001; Craig et al., 2002). For retroviral integration, it is reasonable to infer that host DNA repair enzymes complete provirus formation. DNA gap repair enzymes are known to be involved in a variety of DNA repair pathways, so their recruitment to gaps at host–virus DNA junctions is readily envisioned. Consistent with this, known gap repair enzymes have been shown to act on model host–virus DNA junctions in vitro (Yoder and Bushman, 2000). A number of studies have attempted to investigate this issue using cell death as a surrogate marker for DNA repair (Daniel et al., 1999, 2001a,b, 2003, 2004, 2005). In this work, cells with mutations in DNA repair pathways were infected with viral stocks and cell death was monitored. An increased rate of cell death in mutant cells compared to wild-type controls was interpreted as evidence that the mutated repair pathway is involved in normal infection. In some cases, control infections with a virus mutant for integrase were shown to be less
152
Lewinski and Bushman
Figure 5.1. Reactions catalyzed by integrase. Gray ovals represent integrase monomers, thick lines are viral DNA, thin lines are target DNA, and black dots are 50 ends. (A) DNA cleavage and joining reactions involved in integration. (1) The linear viral cDNA is
5. Retroviral DNA Integration—Mechanism and Consequences
153
toxic, bolstering the idea that the integration step was involved. Such data has led to the proposal that the nonhomologous DNA end joining (NHEJ) proteins DNA-PK, Ku, and XRCC4 are involved, and that signaling by ATR is also important. However, this is surprising because the integration reaction yields DNA gaps—not double-strand breaks. There have been a variety of reports questioning this conclusion that the NHEJ pathway is involved in gap repair at integration intermediates. Four papers from other laboratories describe their inability to reproduce some or all of this data (Ariumi et al., 2005; Baekelandt et al., 2000; Dehart et al., 2005; Kilzer et al., 2003). Infection of cells with retroviral stocks can be toxic to cells for many reasons, including trivial technical ones, leaving the interpretation problematic. Another study suggested that unintegrated DNA might be the proximal signal for cell death; in these studies viral DNA was clearly shown to become circularized in vivo in a reaction requiring the NHEJ pathway (i.e., Ku, XRCC4, and ligase 4), documenting involvement of NHEJ at this step (Li et al., 2001). At present the idea that the NHEJ pathway is important in the late steps of retroviral integration (e.g., gap repair) is not widely popular in the field.
IV. INTEGRASE STRUCTURE The integrase protein is composed of three separate domains—the N-terminal zinc-binding domain, catalytic core, and the C-terminal DNA-binding domain. The domain structure was initially suggested by partial proteolysis studies (Engelman et al., 1993). Later the domain structures were solved by NMR and x-ray crystallography. The crystal and NMR structures of each domain indicate that each dimerizes (Cai et al., 1997; Chen et al., 2000a,b; Eijkelenboom et al., 1999; Goldgur et al., 1998, 1999; Lodi et al., 1995; Maignan et al., 1998; Wang et al., 2001; Yang et al., 2000), but the relevance of these structures to integrase function in vivo remains under investigation. It is known that all three domains
bound by integrase as part of the preintegration complex (PIC). (2) Integrase removes two nucleotides from the 30 ends of the viral DNA (“terminal cleavage”). (3) Integrase catalyzes the joining of the recessed 30 ends of viral DNA to the target DNA (“strand transfer”). (4) The integration intermediate unfolds, yielding gaps at the viral–target DNA junction. (5) Gap repair is possibly mediated by host DNA repair enzymes. (6) The provirus is flanked by repeated segments of the target DNA (“target site duplication”). The length of these repeats is retrovirus-specific (five base pairs shown here, as with HIV-1) and is a consequence of the integration mechanism. (B) The disintegration reaction. Disintegration is the reverse of the strand transfer reaction. Integrase catalyzes the disjoining of a viral DNA 30 end and the target DNA to yield two DNA molecules from a Y-shaped intermediate.
154
Lewinski and Bushman
are essential for the full catalytic activity of integrase (Drelich et al., 1992; Schauer and Billich, 1992; Vink et al., 1993). The structure and function of each domain, along with what is known about how they are assembled in the full-length protein and in integrase oligomers, are discussed in later sections.
A. N-terminal domain The N-terminal domain (approximately the first 50 amino acids) of integrase is thought to promote DNA binding and multimerization. It has a conserved HHCC zinc-binding motif with an overall fold resembling that of the helixturn-helix bacterial repressors (Cai et al., 1997; Eijkelenboom et al., 1997) that is conserved in all retroviral and retrotransposon integrases. Evidence indicates that this domain must bind Zn2þ to function (Bushman et al., 1993; Coffin et al., 1997; Eijkelenboom et al., 1997). Integrase mutants with the N-terminal domain deleted or with substitutions in the conserved His or Cys residues are significantly impaired in their ability to catalyze 30 end cleavage and strand transfer reactions but still maintain disintegration activity (Bushman and Wang, 1994; Bushman et al., 1993; Engelman and Craigie, 1992; Vincent et al., 1993). Other mutants of less highly conserved amino acids in the N-terminal domain have weak end cleavage and strand transfer activities (Vincent et al., 1993). Adding Zn2þ in vitro was found to enhance the Mg2þ-dependent terminal cleavage reaction by HIV-1 integrase (Lee and Han, 1996). This suggests that the N-terminal domain, while having no direct role in catalysis, might play some role in viral DNA recognition. Another possible role for the N-terminal domain of integrase is in multimerization (Heuer and Brown, 1998; Lee et al., 1997; Zheng et al., 1996) (discussed in later section). Studies of the zinc-binding properties of integrase found that the Zn2þ-bound N-termini dimerized (Yang et al., 1999) and Zn2þ-bound integrase tetramerized more easily than integrase without zinc or with mutations in the HHCC motif (Zheng et al., 1996). Binding of zinc to the N-terminal domain of integrase likely stabilizes the enzyme, allowing for proper multimerization and efficient enzymatic activity. Cross-linking studies have also implicated the N-domain in binding of target DNA (Heuer and Brown, 1997).
B. Catalytic core The central domain of integrase (e.g., residues 50–212 of HIV-1 integrase) functions primarily in catalysis and DNA binding. The catalytic core is composed of mixed alpha helix and beta sheets folded so that three acidic residues of the D,DX35E motif are in proximity. This three-dimensional structure is an
5. Retroviral DNA Integration—Mechanism and Consequences
155
RNaseH-type fold that is conserved among members of the D,DX35E phosphotransferase enzyme family, which includes retroviral and retrotransposon integrases and bacterial transposases (Dyda et al., 1994; Kulkosky et al., 1992; Rowland and Dyke, 1990; Yang and Steitz, 1995). Site-directed mutagenesis of conserved amino acids in this catalytic core resulted in integrase proteins inactive in 30 end cleavage, DNA strand transfer, and disintegration assays, suggesting that this domain is essential for catalysis (Engelman and Craigie, 1992; Hazuda et al., 1994; Leavitt et al., 1996). In fact, the catalytic domain alone is sufficient to catalyze disintegration (Bushman and Wang, 1994; Bushman et al., 1993; Kulkosky et al., 1995; Vink et al., 1993), although efficient 30 end cleavage and strand transfer also require the N-terminal and C-terminal domains (Bushman and Wang, 1994; Drelich et al., 1992; Schauer and Billich, 1992; Vink et al., 1993). Each residue of the D,DX35E motif catalytic triad is required for catalysis of integration (Engelman and Craigie, 1992; Kulkosky et al., 1992; van Gent et al., 1993a). D,DX35E motif residues D64 and D116 of HIV-1 integrase are thought to act by coordinating at least one divalent metal ion and probably two (Gao et al., 2004). While initial crystal structures of the catalytic domain did not include a bound cation (Bujacz et al., 1995, 1996a; Dyda et al., 1994), later structures (Bujacz et al., 1996b; Goldgur et al., 1998; Maignan et al., 1998) and models (Lins et al., 1999) showed that the aspartic acid residues of the catalytic triad can coordinate Mn2þ and/or Mg2þ. One structure of avian sarcoma virus (ASV) integrase catalytic domain has been visualized with two bound metal atoms (although the Zn2þ and Cd2þ atoms bound are not biological ligands) (Bujacz et al., 1997) and the catalytic domain of HIV-1 integrase with two bound cations at the active site was subsequently modeled (Lins et al., 2000). Although integrase bound to two metal atoms has not yet been proven capable of catalyzing integration in vitro, these crystal and model structures suggest that two metal mechanisms are likely. In addition to catalysis of terminal cleavage and strand transfer reactions, the core domain functions in binding to viral DNA. Studies with chimeric integrases have shown that the core domain is responsible for recognition of the viral DNA substrate (Katzman and Sudol, 1995, 1998; Pahl and Flugel, 1995) and cross-linking studies with HIV-1 integrase found that residues Q148 and Y143 bind to the viral DNA ends (Esposito and Craigie, 1998). Cross-linking data suggest that the conserved residues K156 and K159 (near the active site in the catalytic core domain) of HIV-1 integrase are essential for the interaction between integrase and viral DNA, specifically the conserved deoxyadenosine (Jenkins et al., 1997). Further, the core domain is thought to be responsible for target site selection in vitro (Appa et al., 2001; Harper et al., 2003; Shibagaki and Chow, 1997).
156
Lewinski and Bushman
C. C-terminal domain The final 75–100 amino acids of integrase are composed of the C-terminal domain—the least conserved of the three domains. Structural analysis has found that it is has an SH3-type fold, and may form dimers (Eijkelenboom et al., 1995, 1999; Lodi et al., 1995). The C-terminal domain has strong but nonspecific DNA-binding activity and thus has been called the DNA-binding domain (Engelman et al., 1994; Khan et al., 1991; Lutzke and Plasterk, 1998; Lutzke et al., 1994; Mumm and Grandgenett, 1991; Vink et al., 1993; Woerner and Marcus-Sekura, 1993). Its ability to dimerize in solution has led some to suggest that the C-terminal domain plays a role in multimerization (Andrake and Skalka, 1995; Lutzke and Plasterk, 1998). Mutagenesis data support a role for the C-terminal domain in proper folding of the integrase protein (Moreau et al., 2003).
D. Structures containing two domains of integrase Although there are NMR and crystal structures for the individual domains of integrase, these are not sufficient to determine the structural arrangement of domains in full-length integrase protein. Full-length integrase has not been crystallized. However, several structures of two-domain integrase fragments have been solved. These two-domain structures provide insight into the mechanism of host and viral DNA binding and multimerization of integrase. Two-domain structures with the catalytic core and C-terminal domains have been solved for Rous sarcoma virus (RSV) (Yang et al., 2000), HIV-1 (Chen et al., 2000a), and simian immunodeficiency virus (SIV) (Chen et al., 2000b) integrases. Additionally, the structure of a two-domain HIV-1 integrase fragment with the catalytic and N-terminal domains has been determined (Wang et al., 2001). In each of these structures the catalytic core domains are associated as dimers, as they are in structures of the catalytic domain alone (Bujacz et al., 1995; Goldgur et al., 1999, 1998; Lubkowski et al., 1999; Maignan et al., 1998). However, the position of the C-terminal domain varies considerably among these two-domain structures. The two-domain structure of RSV integrase shows that the C-terminal domains associated as a dimer in a canted conformation so that one C-terminal domain contacts its catalytic domain (Yang et al., 2000). In the catalytic/C-terminal two-domain structure for HIV1 integrase, the catalytic cores exist as dimers, but the C-terminal domains are monomeric and at the ends of extended alpha-helical linkers so that the structure is in a Y conformation (Chen et al., 2000a). In the two-domain structure of SIV integrase, only one of the four C-terminal domains associated with two dimers of the catalytic domain can be visualized (Chen et al., 2000b).
5. Retroviral DNA Integration—Mechanism and Consequences
157
The C-terminal domain is poorly conserved among different retroviral integrases so it is not unexpected that its conformations differ in these two-domain structures. However, it is unclear whether any of these structures is similar to the actual conformation of these domains in vivo. The variation in C-terminal domain position relative to the catalytic domain can be attributed to the flexibility of the linker and/or the lack of the stabilizing N-terminal domain or DNA. An HIV-1 integrase fragment that includes the catalytic core and Nterminal domains also crystallized as a dimer (Wang et al., 2001). In this structure, the N-terminal domains are arranged differently than seen in dimers of the individual N-terminal domain (Cai et al., 1997). This two-domain structure can accommodate the C-terminal domain in the same orientation observed in the catalytic/C-terminal two-domain structure of HIV-1 integrase (Chen et al., 2000a). This suggests that the N-terminal domain could stabilize the structure of the C-terminal and catalytic domains of HIV-1 integrase. The two-domain structures of integrase allow for modeling of integrase bound to viral and target DNA. Using time-resolved fluorescence anisotropy (TFA) (Deprez et al., 2000; Leh et al., 2000), protein footprinting (Dirac and Kjems, 2001), and cross-linking data (Esposito and Craigie, 1998; Gao et al., 2001; Heuer and Brown, 1997, 1998; Jenkins et al., 1997) in addition to the structural data reviewed in earlier section, Podtelezhnikov et al. (2003) modeled HIV-1 integrase dimers bound to DNA. Their model differs from the full-length integrase structure suggested by Wang et al. (2001) in that the domains are tightly compacted together. This conflicts with the catalytic core/C-terminal two-domain HIV-1 integrase structure (Chen et al., 2000a), which has the two domains linked by an extended alpha-helix. Such a structure was not compatible with the TFA data. In their model of this compacted integrase dimer bound to DNA, the terminal three bases of viral DNA interacts only with the catalytic core domain while host target DNA binds to all three domains (Podtelezhnikov et al., 2003). This model is able to accommodate both structural and experimental data. The C-terminal and catalytic core domains are known to bind DNA nonspecifically (Engelman et al., 1994). Also, in this model the zinc finger of the N-terminal domain contacts host DNA as seen with cross-linking data (Heuer and Brown, 1997). The structures of these two-domain integrases and the subsequent models of integrase-DNA complexes further support the idea that integrase acts as a tetramer. Dimers in the two-domain structures have the catalytic core active sites on opposite sides of the complex—too far to account for the spacing between sites of integration of the viral DNA ends. This suggests that integration in vivo proceeds with each viral DNA end associated with an integrase dimer assembled as a tetramer.
158
Lewinski and Bushman
E. Multimerization As mentioned in earlier sections, structural analysis of integrase and its domains has determined that integrase can self-associate to form dimers and tetramers in vitro. Studies have shown that pairs of integrase mutants that are inactive alone can complement each other and function to near-wild-type levels in vitro (Engelman et al., 1993; Fletcher et al., 1997; van Gent et al., 1993b). This suggests that integrase acts as a multimer. Other studies have found that multimerization is required for integrase end cleavage and joining reactions (Jones et al., 1992), with the smallest functional integrase unit being a dimer (Bao et al., 2003; Jones et al., 1992). Podtelezhnikov et al. modeled the structure of an HIV-1 integrase tetramer bound to viral and host DNA using TFA data (Deprez et al., 2000; Leh et al., 2000) and computer simulations of the hydrodynamic properties of integrase oligomers. They also incorporated data from crystal structures, crosslinking and other biochemical data on integrase–DNA interactions. They reasoned that their model dimer of HIV-1 integrase (discussed in earlier section) is not sufficient to catalyze concerted integration because the active sites are too far apart to account for the five base pairs that separate integration sites of HIV-1 DNA ends. Thus, a tetramer, with a dimer catalyzing integration of each viral DNA end is the possible functional oligomer in vivo. The model tetramer is composed of monomers with the same structure. One monomer from each dimer catalyzes the integration of one end of the viral DNA while the other monomer serves a structural role. The viral DNA is bound to the catalytic core domain of the active monomer as described in earlier section and also contacts the C-terminal domain of the monomer that catalyzes integration of the other viral DNA end, as suggested by experimental data (Gao et al., 2001). The host DNA binds the catalytic core domain near the site of integration and contacts both C-terminal domains of the catalytically active monomers at the five base pairs between the sites of integration. The N-terminal domains bind host DNA outside of this region according to the model. The dimer–dimer interface involves N-terminal domains of the structural, catalytically inactive monomers, explaining why zinc-binding facilitates tetramerization (Deprez et al., 2000; Zheng et al., 1996). This model tetramer is structurally similar to the Tn5 transposase–DNA complex (Rice and Baker, 2001). A consequence of higher-order assembly of nucleoprotein complexes containing integrase is coupled joining. Coupled joining is the integration of both viral DNA ends into opposite strands of the target DNA. Correct integration in vivo requires joining of both ends of viral DNA with two points in target DNA that are a specific number of base pairs apart (five for HIV-1), depending on the retrovirus. Such coupled joining reactions can be reproduced under
5. Retroviral DNA Integration—Mechanism and Consequences
159
carefully controlled conditions in vitro (Aiyar et al., 1996; Carteau et al., 1999; Goodarzi et al., 1995; Sinha et al., 2002), though complex assembly is inefficient as yet. Coupled joining can be detected as a DNA product of a distinctive length in gels and by sequencing of viral–host DNA junctions to ensure that target site duplication of the correct length is formed after gap repair. The DNA forms detected as a result of the reactions in vitro are frequently a mixture of coupled and uncoupled products. Although progress has been made, efficient reconstitution of integration complexes from purified components has not been fully achieved. One possibility is that additional proteins have a role in assembly of fully functional integrase complexes.
V. COMPOSITION OF INTEGRASE COMPLEXES IN VIVO Integration in vivo is carried out by a nucleoprotein complex that includes the viral DNA and integrase (Bowerman et al., 1989; Brown et al., 1987; Ellison et al., 1990; Farnet and Haseltine, 1990, 1991; Li et al., 2001; Miller et al., 1997). With the development of assays involving PICs purified from virally infected cells (Brown et al., 1987; Ellison et al., 1990; Farnet and Haseltine, 1990), it has become possible to study the organization and function of authentic replication intermediates. Preintegration complex preparations have been generated for cells infected with HIV-1 and murine leukemia virus (MLV). For avian sarcoma-leukosis virus (ASLV), complexes did not efficiently complete reverse transcription, suggesting a late block in replication (Lee and Coffin, 1991). A limitation on studies of PICs has been the difficulty of obtaining large amounts of material. Even if cells are infected at high multiplicity, only a few PICs per cell can be purified. Therefore, only small quantities of PICs can be studied in the background of a complex mixture of cellular proteins. So far PICs have not been purified to homogeneity. Nevertheless, it has been possible to infer a number of their features using sensitive biochemical approaches. Preintegration complexes can be shown to have proteins tightly bound at the viral DNA ends. The ends are protected from attack by exonucleases (Miller et al., 1997) or recombination complexes (Wei et al., 1997, 1998). In addition, it can be shown that the ends of the viral DNA are held together by a protein–DNA complex because the viral DNA can be cut internally with restriction endonucleases and integration can still occur with the viral DNA ends (Miller et al., 1997). As judged by sedimentation and gel filtration, PICs are fairly large particles. For example, with gel filtration using calibrated columns, PICs form a broad peak centered on a 54 nm radius (assuming a spherical shape) (Miller et al., 1997).
160
Lewinski and Bushman
A. Viral proteins Given the difficulty in purifying PICs, it has been challenging to get precise information on their protein composition. Preintegration complexes of HIV-1 have been shown to contain the viral integrase, matrix, and reverse transcriptase (Miller et al., 1997) but very little capsid (Farnet and Haseltine, 1991). The HIV Vpr and nucleocapsid protein (NC) are detectable in early fractions and probably remain associated with the PIC but this has been difficult to demonstrate with more purified preparations (Miller et al., 1997). For MLV, integrase and capsid are readily detected in the PIC, suggesting that more capsid remains associated with MLV than HIV-1 PICs (Bowerman et al., 1989; Li et al., 2001). Several viral proteins have been shown to stimulate reactions with purified integrase in vitro, notably the NC (Carteau et al., 1999; Gao et al., 2003). Under specific conditions in vitro, the magnitude of the stimulation by NC can be 1000-fold or more (Carteau et al., 1999). The effects of nucleocapsid mutants in vivo have been difficult to study because NC is required for multiple steps in the viral life cycle, including RNA dimerization, packaging, and reverse transcription. Studies from Gorelick et al. using carefully selected nucleocapsid mutants have provided some support for the idea that nucleocapsid is important for integration as well (Buckman et al., 2003; Carteau et al., 1999). They found that viral DNAs captured in junctions between 2-LTR circles tended to be predominantly uncleaved by integrase in the presence of the zinc-finger residue substitution CCCC/CCHC nucleocapsid mutant, suggesting a requirement for NC in this integrase-catalyzed step. This readout is indirect but does support the notion that NC is involved in integrase function.
B. Host proteins Several host cell proteins have been suggested to be important for retroviral DNA integration. None have yet been shown to be strictly required for integration in vivo, however, leaving the importance of each proposed protein uncertain. The functions of many DNA-binding proteins and DNA-modifying enzymes are assisted by architectural DNA-binding proteins. These proteins act by changing the direction of the long axis of the DNA helix and/or neutralizing negative charges in the DNA phosphate backbone, assisting in the formation of precise three-dimensional nucleoprotein structures. Many examples have been reported (Bushman, 2001; Craig et al., 2002) to the point where it would be surprising if architectural DNA-binding proteins were not involved in integration. A complication, however, is that in many cases multiple small basic proteins can satisfy the requirement for architectural DNA-binding proteins,
5. Retroviral DNA Integration—Mechanism and Consequences
161
so that redundancy complicates the assessment of in vivo importance of any single protein. In one experimental paradigm, PICs were subjected to gel filtration in the presence of high salt, resulting in a loss of integrase activity (Chen and Engelman, 1998; Farnet and Bushman, 1997; Harris and Engelman, 2000; Lee and Craigie, 1994; Li et al., 1998). Adding back extracts from uninfected cells restored activity. Fractionation of these extracts has led to the identification of several cellular proteins, which can support reconstitution, two of which have been identified as HMGA and barrier-to-autointegration factor (BAF). HMGA was identified through studies of HIV-1 PICs (Farnet and Bushman, 1997), whereas BAF was identified with studies of MLV (Cai et al., 1998; Lee and Craigie, 1994). Murine leukemia virus PICs exposed to high salt tend to use their own DNA as an integration target by a process called autointegration and BAF succeeded in blocking autointegration, hence the name—barrierto-autointegration factor. The importance of both of these proteins in vivo is uncertain. Cells knocked-out for the two HMGA family proteins nevertheless supported wild-type levels of integration (Beitzel and Bushman, 2003), indicating that either HMGA is not important in vivo or it is redundant with other factors. Mutation of BAF is lethal to cells and so cells lacking this factor cannot be studied. However, MLV autointegration is very efficient in vitro, suggesting that there may be a mechanism, such as BAF binding to viral DNA, which blocks this in vivo (Lee and Coffin, 1990; Lee and Craigie, 1994). The viral NC shows some activity in reconstitution after salt-stripping (Farnet and Bushman, 1997), raising the possibility that this viral protein is a contributor during normal infection. Assays in vitro using purified integrase can also be used to assess the function of candidate cofactors. Both NC and HMGA can stimulate reactions with purified HIV-1 integrase (Carteau et al., 1997; Gao et al., 2003; Hindmarsh et al., 1999), while BAF inhibits integrase (unpublished results). The relationship of these results to integration in vivo has not been clarified. Another route to identify candidate cellular proteins has involved searching for proteins that bind tightly to HIV-1 integrase. The first protein to be identified using the yeast two-hybrid assay was Ini1 (Kalpana et al., 1994), a cellular protein that is a member of the SWI/SNF chromatin remodeling complex. Purified Ini1 is able to stimulate integration in vitro under certain conditions. Data suggestive of its in vivo importance comes from overexpression of Ini1 fragments. Overexpression of Ini1 fragments that contain the integraseinteraction domain can show very strong dominant-negative effects, though unexpectedly, these inhibited HIV late in the viral replication cycle (i.e., after integration) (Yung et al., 2001). Though these data are suggestive, it is still uncertain what role, if any, Ini1 plays in normal HIV replication.
162
Lewinski and Bushman
Yet another cellular protein identified by binding to HIV-1 integrase is lens epithelium-derived growth factor (LEDGF)/p75 (Cherepanov et al., 2003). Lens epithelium-derived growth factor was first identified as a transcriptional mediator protein using biochemical assays (Ge et al., 1998) and as a stress-responsive transcription factor in ocular tissues, thus given the name—lens epithelium-derived growth factor. The name notwithstanding, LEDGF appears to be expressed in most tissues assayed. Lens epitheliumderived growth factor can affect the location of integrase inside cells (Llano et al., 2004b). In the presence of LEDGF, HIV-1 integrase can be detected bound to cellular chromatin, suggesting that LEDGF may help bring integrase to target DNA (Maertens et al., 2003). Binding to LEDGF also protects integrase from proteolysis (Llano et al., 2004a). In spite of the data supporting its potential importance in vivo, functional studies indicate that knock-down of LEDGF does not diminish viral replication. Lens epithelium-derived growth factor can stimulate the function of purified integrase in vitro (Cherepanov et al., 2003), but this is a somewhat permissive assay. These data on LEDGF are provocative but as yet definitive evidence of its importance in vivo has not been adduced.
VI. RETROVIRAL INTEGRATION TARGETING While most sequences tested in vitro can serve as targets for integration (Bor et al., 1996; Brown et al., 1987; Craigie et al., 1990), all retroviruses tested exhibit nonrandom selection of integration target sites in cells (Mitchell et al., 2004; Schroder et al., 2002; Wu et al., 2003). Explanations for integration target site specificity include the variable accessibility of certain regions of chromosomal DNA or tethering of the PIC to genomic sites through its interaction with specific cellular DNA-binding proteins.
A. Preferred DNA sequence and structure In vitro studies of integration target site selection with naked DNA found that retroviruses exhibit weak primary sequence preferences (Bor et al., 1996; Carteau et al., 1998; Fitzgerald and Grandgenett, 1994; Goodarzi et al., 1997; Pryciak and Varmus, 1992). Genome-wide studies (see later section) have shown that different retroviruses have weak but distinguishable primary sequence preferences in vivo (Carteau et al., 1998; Holman and Coffin, 2005; Stevens and Griffith, 1996; Wu et al., 2005), but they probably play only a minor role in integration site selection.
5. Retroviral DNA Integration—Mechanism and Consequences
163
Proteins bound to target DNA can influence integration positively or negatively. Steric hindrance prevents integration from occurring in chromosomal areas occupied by DNA-binding proteins as assayed in vitro (Bushman, 1994; Pryciak and Varmus, 1992) and observed in vivo (Maxfield et al., 2005; Weidhaas et al., 2000). However, not all protein-bound target DNA is unfavorable for integration. DNA assembled into nucleosomes, for instance, has been shown to be more favorable for integration than naked DNA (Pruss et al., 1994a,b; Pryciak and Varmus, 1992). Close examination of sites preferred in nucleosomal DNA indicates that the most severely bent regions of DNA on the nucleosomes are hotspots for integration (Pruss et al., 1994a; Pryciak et al., 1992), suggesting that distortion of DNA itself facilitates integration. In fact, distortion of DNA in nonnucleosomal protein complexes has been shown to favor integration (Bor et al., 1995; Muller and Varmus, 1994). Distortion of viral and target DNA is probably involved in the process of integration (Bushman and Craigie, 1992; Scottoline et al., 1997) so targeting of DNA that is already distorted could facilitate the reaction.
B. Genome-wide studies Studies of integration site selection across the human genome have been performed for three retroviruses (i.e., HIV-1, MLV, and ASLV). The patterns of target site selection differ among these retroviruses. Transcription units were strongly favored targets of HIV-1 integration (Schroder et al., 2002) regardless of host cell type studied (Mitchell et al., 2004; Wu et al., 2003). Murine leukemia virus integrase favored transcription units to a lesser extent but exhibited a strong bias for areas within five kilobases of transcription start sites, with about 20% of integration sites found in these regions (Wu et al., 2003). No bias was found in the location of HIV integration sites within transcription units (i.e., the frequency of integration was the same across the length of the transcription unit) (Mitchell et al., 2004; Wu et al., 2003). Avian sarcoma-leukosis virus shows the most random distribution of integration sites with only a weak preference for genes (Mitchell et al., 2004; Narezkina et al., 2004). To study the influence of host cell gene expression on integration targeting, microarrays were used for transcriptional profiling of the target cells. The median expression level of genes targeted for integration by HIV was found to be much higher than the median expression of all genes assayed, indicating that HIV has a preference for integration into active genes (Schroder et al., 2002). Studies of HIV integration site selection and gene transcription in two other human cell types revealed that in these cell types as well, transcription units were favored integration targets (Mitchell et al., 2004). The bias for active genes was tissue-specific in that genes targeted for integration in a specific host tissue were more likely to be highly expressed in that cell type than in the others
164
Lewinski and Bushman
tested. Murine leukemia virus and ASLV were shown to have a weak preference for active genes (Mitchell et al., 2004; Narezkina et al., 2004). Surprisingly, studies of ASLV integration into two genes in quail cells suggested that high level transcription disfavored integration (Maxfield et al., 2005; Weidhaas et al., 2000). The reason for the difference in the results of these studies and genome-wide studies is unknown. Several chromosomal features were found to influence integration site selection in the genome-wide studies. HIV integration is biased toward GC-rich regions and cytogenetic R bands (Elleder et al., 2002; Mitchell et al., 2004; Schroder et al., 2002). This might be explained by HIV’s preference for generich regions of chromosomes, which are correlated with high gene expression, high GC content and R banding. CpG islands are chromosomal regions enriched in the CpG dinucleotide corresponding to gene regulatory regions. Studies have found that CpG islands are favored integration targets of MLV (Wu et al., 2003) and MLV-based vectors, such as those used in gene therapy (Laufs et al., 2004). However, although CpG islands are found in regions of high gene density, regions that are favored for HIV-1 integration, CpG islands themselves are disfavored for HIV-1 integration (Mitchell et al., 2004). Early studies of integration targeting suggested that MLV integration may be biased toward DNase hypersensitive sites (Rohdewohld et al., 1987; Vijaya et al., 1986). This bias may be a consequence of favored integration into areas of open chromatin (Panet and Cedar, 1977). Accessibility may also be the explanation for the opposite trend seen with heterochromatin. Human immunodeficiency virus type 1 disfavors integration into alpha satellite DNA, a marker for centromeric heterochromatin (Carteau et al., 1998; Schroder et al., 2002). Centromeric heterochromatin is tightly packed and presumably less accessible to the retroviral PIC. A bias appears in the selection of whole chromosomes for HIV integration that cannot be entirely accounted for the variations in gene density among the chromosomes (Laufs et al., 2003; Mitchell et al., 2004). If found to be reproducible, this may point to additional factors involved in integration targeting, such as the intranuclear positions of chromosomes. The genome-wide studies of integration targeting mentioned in earlier sections were all done with human cells. Human cells were chosen because of their relevance to medicine and the feasibility of such studies following the completion of the human genome sequence. However, the biological relevance of studies in cell types that are not the natural hosts of the viruses studied is unclear. It is possible that cellular factors responsible for integration site selection are not well conserved among different species. One study that considered integration targeting in nonhuman cells (Hematti et al., 2004) surveyed integration site selection by SIV- and MLV-based vectors in Rhesus macaque
5. Retroviral DNA Integration—Mechanism and Consequences
165
hematopoetic stem cells. Hematti et al. found that SIV integration preferences are similar to those of HIV, with a strong bias towards transcription units. Murine leukemia virus targeting was similar as that in human cells (i.e., with a preference for regions near the 50 ends of genes). This suggests that among primates, at least, there is conservation of the cellular determinants of integration site selection. The most striking result from these genome-wide studies of integration targeting is that different retroviruses have distinct integration target site preferences. This suggests that virus-specific factors, not simply the accessibility of genomic targets, determine integration site selection.
C. Tethering factors All retroviruses studied thus far have at least a weak preference for integration into transcription units, as discussed in an earlier section. This could be explained, in part, by the accessibility of open chromatin to the PIC. However, accessibility alone cannot account for the distinct target site preferences of HIV1, MLV, and ASLV. Thus virus specific factors likely play a role. An attractive model based on studies of the yeast retrotransposons is that the retroviral PICs interact with tethering factors bound to specific regions of host cell chromosomes that direct integration to nearby sites. The yeast retrotransposons, the Ty elements, are very similar to retroviruses in genome organization and replication. The major difference is that, unlike retroviruses, retrotransposons lack env genes and thus do not have an extracellular stage in their replication cycle. Because they cannot produce progeny that leave the host cell, retrotransposons must avoid killing their host cell during replication. Replication without disruption of host cell transcription is particularly difficult for Ty elements because their host, Saccharomyces cerevisiae, has a very gene-dense genome. Ty1, Ty3, and Ty5 have developed their own strategy for targeting their integration to benign regions of the yeast genome (Boeke and Devine, 1998; Bushman, 2003; Sandmeyer, 2003). Both Ty1 and Ty3 integrate upstream of Pol III-transcribed genes. Ty3 does this through integrase binding to the Pol III transcription complex and directing insertion of its DNA nearby (Kirchner et al., 1995). Ty5 integrase targets telomeres or the silent mating loci by interacting with the heterochromatin protein Sir4p (Zhu et al., 1999, 2003). Evidence suggests that proper integration targeting by retroviruses plays a role in their replication as well. For instance, HIV-1 has a small window of time to replicate because productively infected cells are quickly eliminated by cytotoxic T lymphocytes and the cytopathic effects of the virus. In order to maximize progeny production, HIV may have evolved to target integration to regions of the host genome most conducive to high proviral expression
166
Lewinski and Bushman
(Lewinski et al., 2005). As with retrotransposons, tethering of the retroviral PIC to target DNA might play a role in retroviral integration site selection. In vitro studies with artificial tethering of retroviral integrases have confirmed the feasibility of such a mechanism for integration targeting. In these studies, integrase fusions to sequence-specific DNA-binding domains were able to direct site-specific integration in vitro (Bushman, 1994; Bushman and Miller, 1997; Goulaouic and Chow, 1996; Holmes-Son and Chow, 2000; Katz et al., 1996; Tan et al., 2004). Several cellular factors are known to bind PICs and/or facilitate integration in vitro, suggesting they might influence targeting of retroviral integration to cellular chromosomes. They are BAF, HMGA1, Ini-1, Ku, and LEDGF/ p75, among others (Bushman, 2001, 2003; Coffin et al., 1997; Engelman, 2005; Sandmeyer, 2003). Of these, an attractive candidate tethering factor for HIV-1 integrase is LEDGF/p75. Lens epithelium-derived growth factor binds to HIV integrase and is found in HIV PICs but does not bind MLV integrase (Cherepanov et al., 2003; Llano et al., 2004b; Maertens et al., 2003). If LEDGF is found to be involved in HIV-1 integration targeting, it might account for the different integration site selection preferences of these viruses.
VII. CONSEQUENCES OF INTEGRATION INTO HOST CHROMOSOMES The fates of the provirus and its host cell are intimately intertwined. The provirus can influence transcription of host genes in its vicinity and the chromosomal environment exerts its effects on proviral transcription. This reciprocal relationship is at least in part responsible for two fascinating phenomena associated with retroviral integration (i.e., insertional mutagenesis and viral latency).
A. Insertional mutagenesis Defining the determinants of integration, targeting has become topical due to setbacks faced in gene therapy trials using retroviral vectors. In these trials, two of nine children successfully treated for X-linked severe combined immunodeficiency with an MLV-based vector delivering the IL2RG gene developed T cell leukemia (Hacein-Bey-Abina et al., 2003a,b). In these children, the leukemic cells harbored a vector integration site in or near the LMO-2 gene, a candidate protooncogene, which resulted in an increase in LMO-2 expression. A third case of insertional mutagenesis has also been reported. Retroviruses have long been implicated in tumorigenesis in animals (Coffin et al., 1997). They can exert oncogenic effects by transducing an oncogene (v-onc) that they encode, as in oncogenesis by acute transforming
5. Retroviral DNA Integration—Mechanism and Consequences
167
viruses. As another means of oncogenesis, the retrovirus can integrate near and affect the expression of a cellular protooncogene or tumor suppressor. This phenomenon is called proviral insertional mutagenesis. In the case of promoter insertion, the provirus integrates in the same orientation upstream of a protooncogene, thus increasing its expression via the promoter and enhancer elements of an LTR. If inserted upstream in the opposite orientation of the gene or downstream in either orientation, the proviral enhancer sequences can boost expression of the protooncogene. A provirus can be inserted within a gene where it disrupts the formation of normal transcripts. This can result in proteins that lack regulatory domains, abolishing negative regulation, or it can increase mRNA stability. A retroviral insertion can also contribute to oncogenesis by inactivating one copy of a tumor suppressor gene. In order for cancer to result in this case, however, there must be inactivation of the other genomic copy of the tumor suppressor as well. The propensity of retroviruses to cause cancer in model vertebrates by the mechanisms discussed in earlier section has made them important for the discovery of protooncogenes. Numerous protooncogenes have been identified through the sequencing of integration sites in endogenous tumors (Coffin et al., 1997). Also, high-throughput methods for proviral tagging of protooncogenes have been developed (Li et al., 1999; Suzuki et al., 2002). For the adverse events in the X-linked SCID gene therapy trial, the IL2RG-transducing MLV-based vector integrated near the 50 end of the LMO-2 gene. In one case the vector was in the 50 promoter region in the same orientation as LMO-2 and in the other case it was in the first intron in the opposite orientation (Hacein-Bey-Abina et al., 2003a). Insertions in Lmo-2 and Il2rg have been associated with leukemia in mice by retroviral tagging (Dave et al., 2004). It is likely that the growth-promoting effects of the IL2RG transgene and the increased expression of LMO-2 acted synergistically in the development of leukemia in these children. The retroviral vector employed in the X-linked SCID gene therapy trial was based on MLV. Murine leukemia virus-based vectors are widely used for gene transduction in animals and in human gene therapy. As discussed in earlier section, studies of MLV integration targeting have determined that MLV exhibits a preference for integration near the 50 ends of genes (Wu et al., 2003). Considering the potential for activation of the host gene via promoter insertion or proviral enhancer activity underscores the potential dangers of using MLV as a gene therapy vector. Its weak bias for integration in genes (Mitchell et al., 2004) might make ASLV a preferable vector for gene therapy applications. Further studies of integration targeting are necessary in order to elucidate the determinants of site selection so that less toxic gene therapy vectors can be engineered.
168
Lewinski and Bushman
B. Latency The primary obstacle to the eradication of HIV from the body, and thus a cure for AIDS, is the existence of latent viral reservoirs. These are cells that harbor replication competent but unexpressed virus that is invulnerable to antiretroviral therapy and immune surveillance. Upon cessation of drug therapy and activation of the host cells, the latent virus can re-seed the body with virus. Several factors probably contribute to latency, many of which are consequences of the fact that the preferred host cells of HIV-CD4þ are T lymphocytes. CD4þ T lymphocytes vary widely in how permissive they are for HIV replication depending on their activation status. For instance, HIV-1 infection of resting CD4þ T cells does not result in efficient formation of integrated proviruses due to a number of blocks in replication—a phenomenon called preintegration latency (Pierson et al., 2002). Productive infection of these cells generally occurs only if the resting host cell is activated before viral PIC is degraded. Stable latent reservoirs are due to primarily postintegration latency (Lassen et al., 2004). Rarely, an activated CD4þ T cell is infected with HIV and then reverts to a quiescent memory phenotype. The integrated provirus is thought to remain unexpressed because the quiescent T cell lacks the necessary transcription factors and its chromatin is condensed and inactive (Brown et al., 1999; Lassen et al., 2004; Setterfield et al., 1983). Studies have considered additional factors determining postintegration latency in vivo (Finzi et al., 1997; Wong et al., 1997), including the proviral integration site (Han et al., 2004). The site of integration in the genome has long been known to influence the expression of genes within proviruses. Nevertheless, the influence of integration site selection on HIV gene expression has been hard to study in vivo. This is due to the scarcity of latently infected cells in infected individuals. Han et al. characterized 74 integration sites from T cells of patients on prolonged antiretroviral therapy and found that the distribution of sites was similar to that of HIV in cell culture-active genes were favored targets (Han et al., 2004). However, because these proviruses were not sequenced or otherwise tested, it is impossible to know whether they were truly latent (i.e., replication competent but silenced) or inactivated by mutation. In fact, only 1% of inactive HIV proviruses are thought to be authentically latent (Chun et al., 1997a,b). Because of the challenges of studying HIV latency in vivo, cell culture models of this phenomenon have been developed and studied. Jordan et al. (2001) developed a model system in which the human CD4þ T cell line, Jurkat, was infected with an HIV-based vector expressing green fluorescent protein (GFP) from the viral promoter. Following infection, cells were sorted into GFP-expressing and nonexpressing populations. Cells from the GFP-negative population were stimulated with a cytokine or mitogen to activate expression of
5. Retroviral DNA Integration—Mechanism and Consequences
169
the silenced proviruses. Those cells harboring inducible proviruses were analyzed. Studies with this model found that inducible proviruses were frequently found integrated into heterochromatic regions (Jordan et al., 2003). A genome-wide study using this model comparing integration sites of productively infected and latent proviruses found three chromosomal features, which are correlated with poor expression of an HIV provirus (Lewinski et al., 2005). As with the previous study, heterochromatic regions, specifically centromeric alpha satellite repeats, were associated with silencing of proviruses. Heterochromatin has a condensed conformation that blocks access of transcriptional machinery to the provirus. The second genomic feature overrepresented in the latent population was highly expressed host genes. A high level of transcription of the host gene might interfere with transcription from the HIV promoter. Several studies support the idea that transcriptional interference can silence proviral transcription (Callen et al., 2004; Cullen et al., 1984; Greger et al., 1998, 2000; Hausler and Somerville, 1979). The third chromosomal feature enriched in the latent population was long intergenic regions or gene deserts. These long chromosomal stretches devoid of genes may be a marker for transcription-silencing heterochromatin or may have an intranuclear position unfavorable for gene transcription. All of these features are plausible contributors to HIV postintegration latency, blocking proviral transcription until overcome by host cell activation. Thus, latency appears to be, in part, a byproduct of HIV infrequently integrating into areas of the genome unfavorable for its expression.
VIII. CONCLUSIONS The retroviral integrase protein plays a central role in the retroviral replication cycle, influencing the processes of reverse transcription, assembly, and budding in addition to mediating the covalent attachment of the viral DNA to the host cell chromosome. Integration of retroviral cDNA involves catalysis of terminal cleavage and strand transfer reactions by integrase followed by repair of the gaps at the host–virus DNA junction, probably by cellular DNA repair enzymes. The joining of the two ends of viral DNA to the host cell DNA is concerted. Evidence for coupled joining along with structural and cross-linking data have contributed to the favored model of integration in vivo where an integrase tetramer, with each dimer bound to one viral DNA end, directs integration of both viral DNA ends into opposite strands of the target DNA. In vivo, this integrase tetramer, bound to viral and cellular DNA, exists in a PIC with other viral and cellular proteins. The variable composition of these PICs among different retroviruses and their interactions with particular tethering factors
170
Lewinski and Bushman
bound to host chromosomes are likely determinants of their characteristic integration site preferences. Insertional mutagenesis is one consequence of retroviral integration into particular genomic sites. Data suggests that, due to their integration target site preferences, some retroviruses may be prone to insertional activation of protooncogenes. This has important implications for the development of retroviral vectors for use in gene therapy. As the viral and cellular determinants of integration site selection are elucidated, safer and more effective gene therapy vectors may be developed. Studies of HIV integration targeting and its effects on proviral transcription have identified integration site as a potential contributor to the molecular mechanisms of HIV latency, which is the primary barrier to a cure for AIDS. The more we understand about retroviral integrases (i.e., their structure and determinants of their function in vivo) the closer we come to the development of new therapies for the fight against AIDS and genetic diseases.
References Aiyar, A., Hindmarsh, P., Skalka, A. M., and Leis, J. (1996). Concerted integration of linear retroviral DNA by the avian sarcoma virus integrase in vitro: Dependence on both long terminal repeat termini. J. Virol. 70, 3571–3580. Andrake, M. D., and Skalka, A. M. (1995). Multimerization determinants reside in both the catalytic core and C terminus of avian sarcoma virus integrase. J. Biol. Chem. 270, 29299–29306. Appa, R. S., Shin, C. G., Lee, P., and Chow, S. A. (2001). Role of the nonspecific DNA-binding region and alpha helices within the core domain of retroviral integrase in selecting target DNA sites for integration. J. Biol. Chem. 276, 45848–45855. Ariumi, Y., Turelli, P., Masutani, M., and Trono, D. (2005). DNA damage sensors ATM, ATR, DNA-PKcs, and PARP-1 are dispensable for human immunodeficiency virus type 1 integration. J. Virol. 79, 2973–2978. Baekelandt, V., Claeys, A., Cherepanov, P., De Clercq, E., De Strooper, B., Nuttin, B., and Debyser, Z. (2000). DNA-dependent protein kinase is not required for efficient lentivirus integration. J. Virol. 74, 11278–11285. Balakrishnan, M., and Jonsson, C. B. (1997). Functional identification of nucleotides conferring substrate specificity to retroviral integrase reactions. J. Virol. 71, 1025–1035. Bao, K. K., Wang, H., Miller, J. K., Erie, D. A., Skalka, A. M., and Wong, I. (2003). Functional oligomeric state of avian sarcoma virus integrase. J. Biol. Chem. 278, 1323–1327. Beitzel, B., and Bushman, F. (2003). Construction and analysis of cells lacking the HMGA gene family. Nucleic Acids Res. 31, 5025–5032. Boeke, J. D., and Devine, S. E. (1998). Yeast retrotransposons: Finding a nice quiet neighborhood. Cell 93, 1087–1089. Bor, Y. C., Bushman, F. D., and Orgel, L. E. (1995). In vitro integration of human immunodeficiency virus type 1 cDNA into targets containing protein-induced bends. Proc. Natl. Acad. Sci. USA 92, 10334–10338. Bor, Y. C., Miller, M. D., Bushman, F. D., and Orgel, L. E. (1996). Target-sequence preferences of HIV-1 integration complexes in vitro. Virology 222, 283–288.
5. Retroviral DNA Integration—Mechanism and Consequences
171
Bowerman, B., Brown, P. O., Bishop, J. M., and Varmus, H. E. (1989). A nucleoprotein complex mediates the integration of retroviral DNA. Genes Dev. 3, 469–478. Brown, K. E., Baxter, J., Graf, D., Merkenschlager, M., and Fisher, A. G. (1999). Dynamic repositioning of genes in the nucleus of lymphocytes preparing for cell division. Mol. Cell. 3, 207–217. Brown, P. O., Bowerman, B., Varmus, H. E., and Bishop, J. M. (1987). Correct integration of retroviral DNA in vitro. Cell 49, 347–356. Brown, P. O., Bowerman, B., Varmus, H. E., and Bishop, J. M. (1989). Retroviral integration: Structure of the initial covalent product and its precursor, and a role for the viral IN protein. Proc. Natl. Acad. Sci. USA 86, 2525–2529. Buckman, J. S., Bosche, W. J., and Gorelick, R. J. (2003). Human immunodeficiency virus type 1 nucleocapsid Zn(2þ) fingers are required for efficient reverse transcription, initial integration processes, and protection of newly synthesized viral DNA. J. Virol. 77, 1469–1480. Bujacz, G., Alexandratos, J., Qing, Z. L., Clement-Mella, C., and Wlodawer, A. (1996a). The catalytic domain of human immunodeficiency virus integrase: Ordered active site in the F185H mutant. FEBS Lett. 398, 175–178. Bujacz, G., Alexandratos, J., Wlodawer, A., Merkel, G., Andrake, M., Katz, R. A., and Skalka, A. M. (1997). Binding of different divalent cations to the active site of avian sarcoma virus integrase and their effects on enzymatic activity. J. Biol. Chem. 272, 18161–18168. Bujacz, G., Jaskolski, M., Alexandratos, J., Wlodawer, A., Merkel, G., Katz, R. A., and Skalka, A. M. (1995). High-resolution structure of the catalytic domain of avian sarcoma virus integrase. J. Mol. Biol. 253, 333–346. Bujacz, G., Jaskolski, M., Alexandratos, J., Wlodawer, A., Merkel, G., Katz, R. A., and Skalka, A. M. (1996b). The catalytic domain of avian sarcoma virus integrase: Conformation of the active-site residues in the presence of divalent cations. Structure 4, 89–96. Bushman, F. D. (1994). Tethering human immunodeficiency virus 1 integrase to a DNA site directs integration to nearby sequences. Proc. Natl. Acad. Sci. USA 91, 9233–9237. Bushman, F. D. (2001). “Lateral DNA Transfer: Mechanisms and Consequences.” Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Bushman, F. D. (2003). Targeting survival: Integration site selection by retroviruses and LTRretrotransposons. Cell 115, 135–138. Bushman, F. D., and Craigie, R. (1990). Sequence requirements for integration of Moloney murine leukemia virus DNA in vitro. J. Virol. 64, 5645–5648. Bushman, F. D., and Craigie, R. (1991). Activities of human immunodeficiency virus (HIV) integration protein in vitro: Specific cleavage and integration of HIV DNA. Proc. Natl. Acad. Sci. USA 88, 1339–1343. Bushman, F. D., and Craigie, R. (1992). Integration of human immunodeficiency virus DNA: Adduct interference analysis of required DNA sites. Proc. Natl. Acad. Sci. USA 89, 3458–3462. Bushman, F. D., Engelman, A., Palmer, I., Wingfield, P., and Craigie, R. (1993). Domains of the integrase protein of human immunodeficiency virus type 1 responsible for polynucleotidyl transfer and zinc binding. Proc. Natl. Acad. Sci. USA 90, 3428–3432. Bushman, F. D., Fujiwara, T., and Craigie, R. (1990). Retroviral DNA integration directed by HIV integration protein in vitro. Science 249, 1555–1558. Bushman, F. D., and Miller, M. D. (1997). Tethering human immunodeficiency virus type 1 preintegration complexes to target DNA promotes integration at nearby sites. J. Virol. 71, 458–464. Bushman, F. D., and Wang, B. (1994). Rous sarcoma virus integrase protein: Mapping functions for catalysis and substrate binding. J. Virol. 68, 2215–2223.
172
Lewinski and Bushman
Cai, M., Huang, Y., Zheng, R., Wei, S. Q., Ghirlando, R., Lee, M. S., Craigie, R., Gronenborn, A. M., and Clore, G. M. (1998). Solution structure of the cellular factor BAF responsible for protecting retroviral DNA from autointegration. Nat. Struct. Biol. 5, 903–909. Cai, M., Zheng, R., Caffrey, M., Craigie, R., Clore, G. M., and Gronenborn, A. M. (1997). Solution structure of the N-terminal zinc binding domain of HIV-1 integrase. Nat. Struct. Biol. 4, 567–577. Callen, B. P., Shearwin, K. E., and Egan, J. B. (2004). Transcriptional interference between convergent promoters caused by elongation over the promoter. Mol. Cell 14, 647–656. Carteau, S., Batson, S. C., Poljak, L., Mouscadet, J. F., de Rocquigny, H., Darlix, J. L., Roques, B. P., Kas, E., and Auclair, C. (1997). Human immunodeficiency virus type 1 nucleocapsid protein specifically stimulates Mg2þ-dependent DNA integration in vitro. J. Virol. 71, 6225–6229. Carteau, S., Gorelick, R. J., and Bushman, F. D. (1999). Coupled integration of human immunodeficiency virus type 1 cDNA ends by purified integrase in vitro: Stimulation by the viral nucleocapsid protein. J. Virol. 73, 6670–6679. Carteau, S., Hoffmann, C., and Bushman, F. (1998). Chromosome structure and human immunodeficiency virus type 1 cDNA integration: Centromeric alphoid repeats are a disfavored target. J. Virol. 72, 4005–4014. Chen, H., and Engelman, A. (1998). The barrier-to-autointegration protein is a host factor for HIV type 1 integration. Proc. Natl. Acad. Sci. USA 95, 15270–15274. Chen, J. C., Krucinski, J., Miercke, L. J., Finer-Moore, J. S., Tang, A. H., Leavitt, A. D., and Stroud, R. M. (2000a). Crystal structure of the HIV-1 integrase catalytic core and C-terminal domains: A model for viral DNA binding. Proc. Natl. Acad. Sci. USA 97, 8233–8238. Chen, Z., Yan, Y., Munshi, S., Li, Y., Zugay-Murphy, J., Xu, B., Witmer, M., Felock, P., Wolfe, A., Sardana, V., Emini, E. A., Hazuda, D., and Kuo, L. C. (2000b). X-ray structure of simian immunodeficiency virus integrase containing the core and C-terminal domain (residues 50–293)—an initial glance of the viral DNA binding platform. J. Mol. Biol. 296, 521–533. Cherepanov, P., Maertens, G., Proost, P., Devreese, B., Van Beeumen, J., Engelborghs, Y., De Clercq, E., and Debyser, Z. (2003). HIV-1 integrase forms stable tetramers and associates with LEDGF/p75 protein in human cells. J. Biol. Chem. 278, 372–381. Chow, S. A., Vincent, K. A., Ellison, V., and Brown, P. O. (1992). Reversal of integration and DNA splicing mediated by integrase of human immunodeficiency virus. Science 255, 723–726. Chun, T. W., Carruth, L., Finzi, D., Shen, X., DiGiuseppe, J. A., Taylor, H., Hermankova, M., Chadwick, K., Margolick, J., Quinn, T. C., Kuo, Y. H., Brookmeyer, R., Zeiger, M. A., BarditchCrovo, P., and Siliciano, R. F. (1997a). Quantification of latent tissue reservoirs and total body viral load in HIV-1 infection. Nature 387, 183–188. Chun, T. W., Stuyver, L., Mizell, S. B., Ehler, L. A., Mican, J. A., Baseler, M., Lloyd, A. L., Nowak, M. A., and Fauci, A. S. (1997b). Presence of an inducible HIV-1 latent reservoir during highly active antiretroviral therapy. Proc. Natl. Acad. Sci. USA 94, 13193–13197. Coffin, J. M., Hughes, S. H., and Varmus, H. E. (1997). “Retroviruses.” Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Colicelli, J., and Goff, S. P. (1985). Mutants and pseudorevertants of Moloney murine leukemia virus with alterations at the integration site. Cell 42, 573–580. Colicelli, J., and Goff, S. P. (1988). Sequence and spacing requirements of a retrovirus integration site. J. Mol. Biol. 199, 47–59. Craig, N. L., Craigie, R., Gellert, M., and Lambowitz, A. (eds.) (2002). “Mobile DNA II.” American Society Microbiology. Craigie, R., Fujiwara, T., and Bushman, F. (1990). The IN protein of Moloney murine leukemia virus processes the viral DNA ends and accomplishes their integration in vitro. Cell 62, 829–837. Cullen, B. R., Lomedico, P. T., and Ju, G. (1984). Transcriptional interference in avian retroviruses– implications for the promoter insertion model of leukaemogenesis. Nature 307, 241–245.
5. Retroviral DNA Integration—Mechanism and Consequences
173
Daniel, R., Greger, J. G., Katz, R. A., Taganov, K. D., Wu, X., Kappes, J. C., and Skalka, A. M. (2004). Evidence that stable retroviral transduction and cell survival following DNA integration depend on components of the nonhomologous end joining repair pathway. J. Virol. 78, 8573–8581. Daniel, R., Kao, G., Taganov, K., Greger, J. G., Favorova, O., Merkel, G., Yen, T. J., Katz, R. A., and Skalka, A. M. (2003). Evidence that the retroviral DNA integration process triggers an ATRdependent DNA damage response. Proc. Natl. Acad. Sci. USA 100, 4778–4783. Daniel, R., Katz, R. A., Merkel, G., Hittle, J. C., Yen, T. J., and Skalka, A. M. (2001a). Wortmannin potentiates integrase-mediated killing of lymphocytes and reduces the efficiency of stable transduction by retroviruses. Mol. Cell. Biol. 21, 1164–1172. Daniel, R., Katz, R. A., and Skalka, A. M. (1999). A role for DNA-PK in retroviral DNA integration. Science 284, 644–647. Daniel, R., Litwin, S., Katz, R. A., and Skalka, A. M. (2001b). Computational analysis of retrovirusinduced scid cell death. J. Virol. 75, 3121–3128. Daniel, R., Marusich, E., Argyris, E., Zhao, R. Y., Skalka, A. M., and Pomerantz, R. J. (2005). Caffeine inhibits human immunodeficiency virus type 1 transduction of nondividing cells. J. Virol. 79, 2058–2065. Dave, U. P., Jenkins, N. A., and Copeland, N. G. (2004). Gene therapy insertional mutagenesis insights. Science 303, 333. Dehart, J. L., Andersen, J. L., Zimmerman, E. S., Ardon, O., An, D. S., Blackett, J., Kim, B., and Planelles, V. (2005). The ataxia telangiectasia-mutated and rad3-related protein is dispensable for retroviral integration. J. Virol. 79, 1389–1396. Deprez, E., Tauc, P., Leh, H., Mouscadet, J. F., Auclair, C., and Brochon, J. C. (2000). Oligomeric states of the HIV-1 integrase as measured by time-resolved fluorescence anisotropy. Biochemistry 39, 9275–9284. Dhar, R., McClements, W. L., Enquist, L. W., and Vande Woude, G. F. (1980). Nucleotide sequences of integrated Moloney sarcoma provirus long terminal repeats and their host and viral junctions. Proc. Natl. Acad. Sci. USA 77, 3937–3941. Dirac, A. M., and Kjems, J. (2001). Mapping DNA-binding sites of HIV-1 integrase by protein footprinting. Eur. J. Biochem. 268, 743–751. Donehower, L. A. (1988). Analysis of mutant Moloney murine leukemia viruses containing linker insertion mutations in the 30 region of pol. J. Virol. 62, 3958–3964. Donehower, L. A., and Varmus, H. E. (1984). A mutant murine leukemia virus with a single missense codon in pol is defective in a function affecting integration. Proc. Natl. Acad. Sci. USA 81, 6461–6465. Drelich, M., Wilhelm, R., and Mous, J. (1992). Identification of amino acid residues critical for endonuclease and integration activities of HIV-1 IN protein in vitro. Virology 188, 459–468. Dyda, F., Hickman, A. B., Jenkins, T. M., Engelman, A., Craigie, R., and Davies, D. R. (1994). Crystal structure of the catalytic domain of HIV-1 integrase: Similarity to other polynucleotidyl transferases. Science 266, 1981–1986. Eijkelenboom, A. P., Lutzke, R. A., Boelens, R., Plasterk, R. H., Kaptein, R., and Hard, K. (1995). The DNA-binding domain of HIV-1 integrase has an SH3-like fold. Nat. Struct. Biol. 2, 807–810. Eijkelenboom, A. P., Sprangers, R., Hard, K., Puras Lutzke, R. A., Plasterk, R. H., Boelens, R., and Kaptein, R. (1999). Refined solution structure of the C-terminal DNA-binding domain of human immunovirus-1 integrase. Proteins 36, 556–564. Eijkelenboom, A. P., van den Ent, F. M., Vos, A., Doreleijers, J. F., Hard, K., Tullius, T. D., Plasterk, R. H., Kaptein, R., and Boelens, R. (1997). The solution structure of the amino-terminal HHCC domain of HIV-2 integrase: A three-helix bundle stabilized by zinc. Curr. Biol. 7, 739–746.
174
Lewinski and Bushman
Elleder, D., Pavlicek, A., Paces, J., and Hejnar, J. (2002). Preferential integration of human immunodeficiency virus type 1 into genes, cytogenetic R bands and GC-rich DNA regions: Insight from the human genome sequence. FEBS Lett. 517, 285–286. Ellison, V., Abrams, H., Roe, T., Lifson, J., and Brown, P. (1990). Human immunodeficiency virus integration in a cell-free system. J. Virol. 64, 2711–2715. Ellison, V., and Brown, P. O. (1994). A stable complex between integrase and viral DNA ends mediates human immunodeficiency virus integration in vitro. Proc. Natl. Acad. Sci. USA 91, 7316–7320. Engelman, A. (2005). The ups and downs of gene expression and retroviral DNA integration. Proc. Natl. Acad. Sci. USA 102, 1275–1276. Engelman, A., Bushman, F. D., and Craigie, R. (1993). Identification of discrete functional domains of HIV-1 integrase and their organization within an active multimeric complex. EMBO J. 12, 3269–3275. Engelman, A., and Craigie, R. (1992). Identification of conserved amino acid residues critical for human immunodeficiency virus type 1 integrase function in vitro. J. Virol. 66, 6361–6369. Engelman, A., Englund, G., Orenstein, J. M., Martin, M. A., and Craigie, R. (1995). Multiple effects of mutations in human immunodeficiency virus type 1 integrase on viral replication. J. Virol. 69, 2729–2736. Engelman, A., Hickman, A. B., and Craigie, R. (1994). The core and carboxyl-terminal domains of the integrase protein of human immunodeficiency virus type 1 each contribute to nonspecific DNA binding. J. Virol. 68, 5911–5917. Engelman, A., Mizuuchi, K., and Craigie, R. (1991). HIV-1 DNA integration: Mechanism of viral DNA cleavage and DNA strand transfer. Cell 67, 1211–1221. Englund, G., Theodore, T. S., Freed, E. O., Engelman, A., and Martin, M. A. (1995). Integration is required for productive infection of monocyte-derived macrophages by human immunodeficiency virus type 1. J. Virol. 69, 3216–3219. Esposito, D., and Craigie, R. (1998). Sequence specificity of viral end DNA binding by HIV-1 integrase reveals critical regions for protein-DNA interaction. EMBO J. 17, 5832–5843. Farnet, C. M., and Bushman, F. D. (1997). HIV-1 cDNA integration: Requirement of HMG I(Y) protein for function of preintegration complexes in vitro. Cell 88, 483–492. Farnet, C. M., and Haseltine, W. A. (1990). Integration of human immunodeficiency virus type 1 DNA in vitro. Proc. Natl. Acad. Sci. USA 87, 4164–4168. Farnet, C. M., and Haseltine, W. A. (1991). Determination of viral proteins present in the human immunodeficiency virus type 1 preintegration complex. J. Virol. 65, 1910–1915. Fassati, A., and Goff, S. P. (1999). Characterization of intracellular reverse transcription complexes of Moloney murine leukemia virus. J. Virol. 73, 8919–8925. Finzi, D., Hermankova, M., Pierson, T., Carruth, L. M., Buck, C., Chaisson, R. E., Quinn, T. C., Chadwick, K., Margolick, J., Brookmeyer, R., Gallant, J., Markowitz, M., Ho, D. D., Richman, D. D., and Siliciano, R. F. (1997). Identification of a reservoir for HIV-1 in patients on highly active antiretroviral therapy. Science 278, 1295–1300. Fitzgerald, M. L., and Grandgenett, D. P. (1994). Retroviral integration: In vitro host site selection by avian integrase. J. Virol. 68, 4314–4321. Fletcher, T. M., IIIrd, Soares, M. A., McPhearson, S., Hui, H., Wiskerchen, M., Muesing, M. A., Shaw, G. M., Leavitt, A. D., Boeke, J. D., and Hahn, B. H. (1997). Complementation of integrase function in HIV-1 virions. EMBO J. 16, 5123–5138. Fujiwara, T., and Mizuuchi, K. (1988). Retroviral DNA integration: Structure of an integration intermediate. Cell 54, 497–504. Gao, K., Butler, S. L., and Bushman, F. (2001). Human immunodeficiency virus type 1 integrase: Arrangement of protein domains in active cDNA complexes. EMBO J. 20, 3565–3576.
5. Retroviral DNA Integration—Mechanism and Consequences
175
Gao, K., Gorelick, R. J., Johnson, D. G., and Bushman, F. (2003). Cofactors for human immunodeficiency virus type 1 cDNA integration in vitro. J. Virol. 77, 1598–1603. Gao, K., Wong, S., and Bushman, F. (2004). Metal binding by the D,DX35E motif of human immunodeficiency virus type 1 integrase: Selective rescue of Cys substitutions by Mn2þ in vitro. J. Virol. 78, 6715–6722. Ge, H., Si, Y., and Roeder, R. G. (1998). Isolation of cDNAs encoding novel transcription coactivators p52 and p75 reveals an alternate regulatory mechanism of transcriptional activation. EMBO J. 17, 6723–6729. Goldgur, Y., Craigie, R., Cohen, G. H., Fujiwara, T., Yoshinaga, T., Fujishita, T., Sugimoto, H., Endo, T., Murai, H., and Davies, D. R. (1999). Structure of the HIV-1 integrase catalytic domain complexed with an inhibitor: A platform for antiviral drug design. Proc. Natl. Acad. Sci. USA 96, 13040–13043. Goldgur, Y., Dyda, F., Hickman, A. B., Jenkins, T. M., Craigie, R., and Davies, D. R. (1998). Three new structures of the core domain of HIV-1 integrase: An active site that binds magnesium. Proc. Natl. Acad. Sci. USA 95, 9150–9154. Goodarzi, G., Chiu, R., Brackmann, K., Kohn, K., Pommier, Y., and Grandgenett, D. P. (1997). Host site selection for concerted integration by human immunodeficiency virus type-1 virions in vitro. Virology 231, 210–217. Goodarzi, G., Im, G. J., Brackmann, K., and Grandgenett, D. (1995). Concerted integration of retrovirus-like DNA by human immunodeficiency virus type 1 integrase. J. Virol. 69, 6090–6097. Goulaouic, H., and Chow, S. A. (1996). Directed integration of viral DNA mediated by fusion proteins consisting of human immunodeficiency virus type 1 integrase and Escherichia coli LexA protein. J. Virol. 70, 37–46. Greger, I. H., Aranda, A., and Proudfoot, N. (2000). Balancing transcriptional interference and initiation on the GAL7 promoter of Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 97, 8415–8420. Greger, I. H., Demarchi, F., Giacca, M., and Proudfoot, N. J. (1998). Transcriptional interference perturbs the binding of Sp1 to the HIV-1 promoter. Nucleic Acids Res. 26, 1294–1301. Hacein-Bey-Abina, S., von Kalle, C., Schmidt, M., Le Deist, F., Wulffraat, N., McIntyre, E., Radford, I., Villeval, J. L., Fraser, C. C., Cavazzana-Calvo, M., and Fischer, A. (2003a). A serious adverse event after successful gene therapy for X-linked severe combined immunodeficiency. N. Engl. J. Med. 348, 255–256. Hacein-Bey-Abina, S., Von Kalle, C., Schmidt, M., McCormack, M. P., Wulffraat, N., Leboulch, P., Lim, A., Osborne, C. S., Pawliuk, R., Morillon, E., Sorensen, R., Forster, A., Fraser, P., Cohen, J. I., de Saint Basile, G., Alexander, I., Wintergerst, U., Frebourg, T., Aurias, A., StoppaLyonnet, D., Romana, S., Radford-Weiss, I., Gross, F., Valensi, F., Delabesse, E., Macintyre, E., Sigaux, F., Soulier, J., Leiva, L. E., Wissler, M., Prinz, C., Rabbitts, T. H., Le Deist, F., Fischer, A., and Cavazzana-Calvo, M. (2003b). LMO2-associated clonal T cell proliferation in two patients after gene therapy for SCID-X1. Science 302, 415–419. Han, Y., Lassen, K., Monie, D., Sedaghat, A. R., Shimoji, S., Liu, X., Pierson, T. C., Margolick, J. B., Siliciano, R. F., and Siliciano, J. D. (2004). Resting CD4þ T cells from human immunodeficiency virus type 1 (HIV-1)-infected individuals carry integrated HIV-1 genomes within actively transcribed host genes. J. Virol. 78, 6122–6133. Hansen, M. S., Carteau, S., Hoffmann, C., Li, L., and Bushman, F. (1998). Retroviral cDNA integration: Mechanism, applications and inhibition. Genet. Eng. (NY) 20, 41–61. Harper, A. L., Sudol, M., and Katzman, M. (2003). An amino acid in the central catalytic domain of three retroviral integrases that affects target site selection in nonviral DNA. J. Virol. 77, 3838–3845.
176
Lewinski and Bushman
Harris, D., and Engelman, A. (2000). Both the structure and DNA binding function of the barrierto-autointegration factor contribute to reconstitution of HIV type 1 integration in vitro. J. Biol. Chem. 275, 39671–39677. Hausler, B., and Somerville, R. L. (1979). Interaction in vivo between strong closely spaced constitutive promoters. J. Mol. Biol. 127, 353–356. Hazuda, D. J., Felock, P. J., Hastings, J. C., Pramanik, B., and Wolfe, A. L. (1997). Differential divalent cation requirements uncouple the assembly and catalytic reactions of human immunodeficiency virus type 1 integrase. J. Virol. 71, 7005–7011. Hazuda, D. J., Wolfe, A. L., Hastings, J. C., Robbins, H. L., Graham, P. L., La Femina, R. L., and Emini, E. A. (1994). Viral long terminal repeat substrate binding characteristics of the human immunodeficiency virus type 1 integrase. J. Biol. Chem. 269, 3999–4004. Hematti, P., Hong, B. K., Ferguson, C., Adler, R., Hanawa, H., Sellers, S., Holt, I. E., Eckfeldt, C. E., Sharma, Y., Schmidt, M., von Kalle, C., Persons, D. A., Billings, E. M., Verfaillie, C. M., Nienhuis, A. W., Wolfsberg, T. G., Dunbar, C. E., and Calmels, B. (2004). Distinct genomic integration of MLV and SIV vectors in primate hematopoietic stem and progenitor cells. PLoS Biol. 2, E423. Heuer, T. S., and Brown, P. O. (1997). Mapping features of HIV-1 integrase near selected sites on viral and target DNA molecules in an active enzyme-DNA complex by photo-cross-linking. Biochemistry 36, 10655–10665. Heuer, T. S., and Brown, P. O. (1998). Photo-cross-linking studies suggest a model for the architecture of an active human immunodeficiency virus type 1 integrase-DNA complex. Biochemistry 37, 6667–6678. Hindmarsh, P., Ridky, T., Reeves, R., Andrake, M., Skalka, A. M., and Leis, J. (1999). HMG protein family members stimulate human immunodeficiency virus type 1 and avian sarcoma virus concerted DNA integration in vitro. J. Virol. 73, 2994–3003. Holman, A. G., and Coffin, J. M. (2005). Symmetrical base preferences surrounding HIV-1, avian sarcoma/leukosis virus, and murine leukemia virus integration sites. Proc. Natl. Acad. Sci. USA 102, 6103–6107. Holmes-Son, M. L., and Chow, S. A. (2000). Integrase-lexA fusion proteins incorporated into human immunodeficiency virus type 1 that contains a catalytically inactive integrase gene are functional to mediate integration. J. Virol. 74, 11548–11556. Hughes, S. H., Mutschler, A., Bishop, J. M., and Varmus, H. E. (1981). A Rous sarcoma virus provirus is flanked by short direct repeats of a cellular DNA sequence present in only one copy prior to integration. Proc. Natl. Acad. Sci. USA 78, 4299–4303. Hughes, S. H., Shank, P. R., Spector, D. H., Kung, H. J., Bishop, J. M., Varmus, H. E., Vogt, P. K., and Breitman, M. L. (1978). Proviruses of avian sarcoma virus are terminally redundant, coextensive with unintegrated linear DNA and integrated at many sites. Cell 15, 1397–1410. Jenkins, T. M., Esposito, D., Engelman, A., and Craigie, R. (1997). Critical contacts between HIV-1 integrase and viral DNA identified by structure-based analysis and photo-crosslinking. EMBO J. 16, 6849–6859. Jones, K. S., Coleman, J., Merkel, G. W., Laue, T. M., and Skalka, A. M. (1992). Retroviral integrase functions as a multimer and can turn over catalytically. J. Biol. Chem. 267, 16037–16040. Jordan, A., Bisgrove, D., and Verdin, E. (2003). HIV reproducibly establishes a latent infection after acute infection of T cells in vitro. EMBO J. 22, 1868–1877. Jordan, A., Defechereux, P., and Verdin, E. (2001). The site of HIV-1 integration in the human genome determines basal transcriptional activity and response to Tat transactivation. EMBO J. 20, 1726–1738. Kalpana, G. V., Marmon, S., Wang, W., Crabtree, G. R., and Goff, S. P. (1994). Binding and stimulation of HIV-1 integrase by a human homolog of yeast transcription factor SNF5. Science 266, 2002–2006.
5. Retroviral DNA Integration—Mechanism and Consequences
177
Katz, R. A., Merkel, G., Kulkosky, J., Leis, J., and Skalka, A. M. (1990). The avian retroviral IN protein is both necessary and sufficient for integrative recombination in vitro. Cell 63, 87–95. Katz, R. A., Merkel, G., and Skalka, A. M. (1996). Targeting of retroviral integrase by fusion to a heterologous DNA binding domain: In vitro activities and incorporation of a fusion protein into viral particles. Virology 217, 178–190. Katzman, M., Katz, R. A., Skalka, A. M., and Leis, J. (1989). The avian retroviral integration protein cleaves the terminal sequences of linear viral DNA at the in vivo sites of integration. J. Virol. 63, 5319–5327. Katzman, M., and Sudol, M. (1995). Mapping domains of retroviral integrase responsible for viral DNA specificity and target site selection by analysis of chimeras between human immunodeficiency virus type 1 and visna virus integrases. J. Virol. 69, 5687–5696. Katzman, M., and Sudol, M. (1998). Mapping viral DNA specificity to the central region of integrase by using functional human immunodeficiency virus type 1/visna virus chimeric proteins. J. Virol. 72, 1744–1753. Khan, E., Mack, J. P., Katz, R. A., Kulkosky, J., and Skalka, A. M. (1991). Retroviral integrase domains: DNA binding and the recognition of LTR sequences. Nucleic Acids Res. 19, 851–860. Kilzer, J. M., Stracker, T., Beitzel, B., Meek, K., Weitzman, M., and Bushman, F. D. (2003). Roles of host cell factors in circularization of retroviral DNA. Virology 314, 460–467. Kirchner, J., Connolly, C. M., and Sandmeyer, S. B. (1995). Requirement of RNA polymerase III transcription factors for in vitro position-specific integration of a retroviruslike element. Science 267, 1488–1491. Kitamura, Y., Lee, Y. M., and Coffin, J. M. (1992). Nonrandom integration of retroviral DNA in vitro: Effect of CpG methylation. Proc. Natl. Acad. Sci. USA 89, 5532–5536. Kulkosky, J., Jones, K. S., Katz, R. A., Mack, J. P., and Skalka, A. M. (1992). Residues critical for retroviral integrative recombination in a region that is highly conserved among retroviral/retrotransposon integrases and bacterial insertion sequence transposases. Mol. Cell. Biol. 12, 2331–2338. Kulkosky, J., Katz, R. A., Merkel, G., and Skalka, A. M. (1995). Activities and substrate specificity of the evolutionarily conserved central domain of retroviral integrase. Virology 206, 448–456. Lassen, K., Han, Y., Zhou, Y., Siliciano, J., and Siliciano, R. F. (2004). The multifactorial nature of HIV-1 latency. Trends Mol. Med. 10, 525–531. Laufs, S., Gentner, B., Nagy, K. Z., Jauch, A., Benner, A., Naundorf, S., Kuehlcke, K., Schiedlmeier, B., Ho, A. D., Zeller, W. J., and Fruehauf, S. (2003). Retroviral vector integration occurs in preferred genomic targets of human bone marrow-repopulating cells. Blood 101, 2191–2198. Laufs, S., Nagy, K. Z., Giordano, F. A., Hotz-Wagenblatt, A., Zeller, W. J., and Fruehauf, S. (2004). Insertion of retroviral vectors in NOD/SCID repopulating human peripheral blood progenitor cells occurs preferentially in the vicinity of transcription start regions and in introns. Mol. Ther. 10, 874–881. Leavitt, A. D., Robles, G., Alesandro, N., and Varmus, H. E. (1996). Human immunodeficiency virus type 1 integrase mutants retain in vitro integrase activity yet fail to integrate viral DNA efficiently during infection. J. Virol. 70, 721–728. Leavitt, A. D., Rose, R. B., and Varmus, H. E. (1992). Both substrate and target oligonucleotide sequences affect in vitro integration mediated by human immunodeficiency virus type 1 integrase protein produced in Saccharomyces cerevisiae. J. Virol. 66, 2359–2368. Lee, M. S., and Craigie, R. (1994). Protection of retroviral DNA from autointegration: Involvement of a cellular factor. Proc. Natl. Acad. Sci. USA 91, 9823–9827. Lee, S. P., and Han, M. K. (1996). Zinc stimulates Mg2þ-dependent 30 -processing activity of human immunodeficiency virus type 1 integrase in vitro. Biochemistry 35, 3837–3844. Lee, S. P., Kim, H. G., Censullo, M. L., and Han, M. K. (1995). Characterization of Mg(2þ)-dependent 30 -processing activity for human immunodeficiency virus type 1 integrase in vitro: Real-time kinetic studies using fluorescence resonance energy transfer. Biochemistry 34, 10205–10214.
178
Lewinski and Bushman
Lee, S. P., Xiao, J., Knutson, J. R., Lewis, M. S., and Han, M. K. (1997). Zn2þ promotes the selfassociation of human immunodeficiency virus type-1 integrase in vitro. Biochemistry 36, 173–180. Lee, Y. M., and Coffin, J. M. (1990). Efficient autointegration of avian retrovirus DNA in vitro. J. Virol. 64, 5958–5965. Lee, Y. M., and Coffin, J. M. (1991). Relationship of avian retrovirus DNA synthesis to integration in vitro. Mol. Cell. Biol. 11, 1419–1430. Leh, H., Brodin, P., Bischerour, J., Deprez, E., Tauc, P., Brochon, J. C., Le Cam, E., Coulaud, D., Auclair, C., and Mouscadet, J. F. (2000). Determinants of Mg2þ-dependent activities of recombinant human immunodeficiency virus type 1 integrase. Biochemistry 39, 9285–9294. Lewinski, M. K., Bisgrove, D., Shinn, P., Chen, H., Hoffmann, C., Hannenhalli, S., Verdin, E., Berry, C. C., Ecker, J. R., and Bushman, F. D. (2005). Genome-wide analysis of chromosomal features repressing HIV transcription. J. Virol. 79, 6610–6619. Li, J., Shen, H., Himmel, K. L., Dupuy, A. J., Largaespada, D. A., Nakamura, T., Shaughnessy, J. D., Jr., Jenkins, N. A., and Copeland, N. G. (1999). Leukaemia disease genes: Large-scale cloning and pathway predictions. Nat. Genet. 23, 348–353. Li, L., Farnet, C. M., Anderson, W. F., and Bushman, F. D. (1998). Modulation of activity of Moloney murine leukemia virus preintegration complexes by host factors in vitro. J. Virol. 72, 2125–2131. Li, L., Olvera, J. M., Yoder, K. E., Mitchell, R. S., Butler, S. L., Lieber, M., Martin, S. L., and Bushman, F. D. (2001). Role of the non-homologous DNA end joining pathway in the early steps of retroviral infection. EMBO J. 20, 3272–3281. Lins, R. D., Briggs, J. M., Straatsma, T. P., Carlson, H. A., Greenwald, J., Choe, S., and McCammon, J. A. (1999). Molecular dynamics studies on the HIV-1 integrase catalytic domain. Biophys. J. 76, 2999–3011. Lins, R. D., Straatsma, T. P., and Briggs, J. M. (2000). Similarities in the HIV-1 and ASV integrase active sites upon metal cofactor binding. Biopolymers 53, 308–315. Llano, M., Delgado, S., Vanegas, M., and Poeschla, E. M. (2004a). Lens epithelium-derived growth factor/p75 prevents proteasomal degradation of HIV-1 integrase. J. Biol. Chem. 279, 55570–55577. Llano, M., Vanegas, M., Fregoso, O., Saenz, D., Chung, S., Peretz, M., and Poeschla, E. M. (2004b). LEDGF/p75 determines cellular trafficking of diverse lentiviral but not murine oncoretroviral integrase proteins and is a component of functional lentiviral preintegration complexes. J. Virol. 78, 9524–9537. Lodi, P. J., Ernst, J. A., Kuszewski, J., Hickman, A. B., Engelman, A., Craigie, R., Clore, G. M., and Gronenborn, A. M. (1995). Solution structure of the DNA binding domain of HIV-1 integrase. Biochemistry 34, 9826–9833. Lubkowski, J., Dauter, Z., Yang, F., Alexandratos, J., Merkel, G., Skalka, A. M., and Wlodawer, A. (1999). Atomic resolution structures of the core domain of avian sarcoma virus integrase and its D64N mutant. Biochemistry 38, 13512–13522. Lutzke, R. A., and Plasterk, R. H. (1998). Structure-based mutational analysis of the C-terminal DNA-binding domain of human immunodeficiency virus type 1 integrase: Critical residues for protein oligomerization and DNA binding. J. Virol. 72, 4841–4848. Lutzke, R. A., Vink, C., and Plasterk, R. H. (1994). Characterization of the minimal DNA-binding domain of the HIV integrase protein. Nucleic Acids Res. 22, 4125–4131. Maertens, G., Cherepanov, P., Pluymers, W., Busschots, K., De Clercq, E., Debyser, Z., and Engelborghs, Y. (2003). LEDGF/p75 is essential for nuclear and chromosomal targeting of HIV-1 integrase in human cells. J. Biol. Chem. 278, 33528–33539. Maignan, S., Guilloteau, J. P., Zhou-Liu, Q., Clement-Mella, C., and Mikol, V. (1998). Crystal structures of the catalytic domain of HIV-1 integrase free and complexed with its metal cofactor: High level of similarity of the active site with other viral integrases. J. Mol. Biol. 282, 359–368.
5. Retroviral DNA Integration—Mechanism and Consequences
179
Maxfield, L. F., Fraize, C. D., and Coffin, J. M. (2005). Relationship between retroviral DNAintegration-site selection and host cell transcription. Proc. Natl. Acad. Sci. USA 2, 1436–1441. Miller, M. D., Farnet, C. M., and Bushman, F. D. (1997). Human immunodeficiency virus type 1 preintegration complexes: Studies of organization and composition. J. Virol. 71, 5382–5390. Mitchell, R. S., Beitzel, B. F., Schroder, A. R., Shinn, P., Chen, H., Berry, C. C., Ecker, J. R., and Bushman, F. D. (2004). Retroviral DNA integration: ASLV, HIV, and MLV show distinct target site preferences. PLoS Biol. 2, E234. Moreau, K., Faure, C., Violot, S., Verdier, G., and Ronfort, C. (2003). Mutations in the C-terminal domain of ALSV (Avian Leukemia and Sarcoma Viruses) integrase alter the concerted DNA integration process in vitro. Eur. J. Biochem. 270, 4426–4438. Muller, H. P., and Varmus, H. E. (1994). DNA bending creates favored sites for retroviral integration: An explanation for preferred insertion sites in nucleosomes. EMBO J. 13, 4704–4714. Mumm, S. R., and Grandgenett, D. P. (1991). Defining nucleic acid-binding properties of avian retrovirus integrase by deletion analysis. J. Virol. 65, 1160–1167. Narezkina, A., Taganov, K. D., Litwin, S., Stoyanova, R., Hayashi, J., Seeger, C., Skalka, A. M., and Katz, R. A. (2004). Genome-wide analyses of avian sarcoma virus integration sites. J. Virol. 78, 11656–11663. Nermut, M. V., and Fassati, A. (2003). Structural analyses of purified human immunodeficiency virus type 1 intracellular reverse transcription complexes. J. Virol. 77, 8196–8206. Pahl, A., and Flugel, R. M. (1995). Characterization of the human spuma retrovirus integrase by sitedirected mutagenesis, by complementation analysis, and by swapping the zinc finger domain of HIV-1. J. Biol. Chem. 270, 2957–2966. Panet, A., and Cedar, H. (1977). Selective degradation of integrated murine leukemia proviral DNA by deoxyribonucleases. Cell 11, 933–940. Panganiban, A. T., and Temin, H. M. (1983). The terminal nucleotides of retrovirus DNA are required for integration but not virus production. Nature 306, 155–160. Panganiban, A. T., and Temin, H. M. (1984). The retrovirus pol gene encodes a product required for DNA integration: Identification of a retrovirus int locus. Proc. Natl. Acad. Sci. USA 81, 7885–7889. Patel, P. H., and Preston, B. D. (1994). Marked infidelity of human immunodeficiency virus type 1 reverse transcriptase at RNA and DNA template ends. Proc. Natl. Acad. Sci. USA 91, 549–553. Pierson, T. C., Zhou, Y., Kieffer, T. L., Ruff, C. T., Buck, C., and Siliciano, R. F. (2002). Molecular characterization of preintegration latency in human immunodeficiency virus type 1 infection. J. Virol. 76, 8518–8531. Podtelezhnikov, A. A., Gao, K., Bushman, F. D., and McCammon, J. A. (2003). Modeling HIV-1 integrase complexes based on their hydrodynamic properties. Biopolymers 68, 110–120. Pruss, D., Bushman, F. D., and Wolffe, A. P. (1994a). Human immunodeficiency virus integrase directs integration to sites of severe DNA distortion within the nucleosome core. Proc. Natl. Acad. Sci. USA 91, 5913–5917. Pruss, D., Reeves, R., Bushman, F. D., and Wolffe, A. P. (1994b). The influence of DNA and nucleosome structure on integration events directed by HIV integrase. J. Biol. Chem. 269, 25031–25041. Pryciak, P. M., Sil, A., and Varmus, H. E. (1992). Retroviral integration into minichromosomes in vitro. EMBO J. 11, 291–303. Pryciak, P. M., and Varmus, H. E. (1992). Nucleosomes, DNA-binding proteins, and DNA sequence modulate retroviral integration target site selection. Cell 69, 769–780. Quinn, T. P., and Grandgenett, D. P. (1988). Genetic evidence that the avian retrovirus DNA endonuclease domain of pol is necessary for viral integration. J. Virol. 62, 2307–2312. Rice, P. A., and Baker, T. A. (2001). Comparative architecture of transposase and integrase complexes. Nat. Struct. Biol. 8, 302–307.
180
Lewinski and Bushman
Rohdewohld, H., Weiher, H., Reik, W., Jaenisch, R., and Breindl, M. (1987). Retrovirus integration and chromatin structure: Moloney murine leukemia proviral integration sites map near DNase Ihypersensitive sites. J. Virol. 61, 336–343. Roth, M. J., Schwartzberg, P. L., and Goff, S. P. (1989). Structure of the termini of DNA intermediates in the integration of retroviral DNA: Dependence on IN function and terminal DNA sequence. Cell 58, 47–54. Rowland, S. J., and Dyke, K. G. (1990). Tn552, a novel transposable element from Staphylococcus aureus. Mol. Microbiol. 4, 961–975. Sandmeyer, S. (2003). Integration by design. Proc. Natl. Acad. Sci. USA 100, 5586–5588. Schauer, M., and Billich, A. (1992). The N-terminal region of HIV-1 integrase is required for integration activity, but not for DNA-binding. Biochem. Biophys. Res. Commun. 185, 874–880. Schroder, A. R., Shinn, P., Chen, H., Berry, C., Ecker, J. R., and Bushman, F. (2002). HIV-1 integration in the human genome favors active genes and local hotspots. Cell 110, 521–529. Schwartzberg, P., Colicelli, J., and Goff, S. P. (1984). Construction and analysis of deletion mutations in the pol gene of Moloney murine leukemia virus: A new viral function required for productive infection. Cell 37, 1043–1052. Scottoline, B. P., Chow, S., Ellison, V., and Brown, P. O. (1997). Disruption of the terminal base pairs of retroviral DNA during integration. Genes Dev. 11, 371–382. Setterfield, G., Hall, R., Bladon, T., Little, J., and Kaplan, J. G. (1983). Changes in structure and composition of lymphocyte nuclei during mitogenic stimulation. J. Ultrastruct. Res. 82, 264–282. Sherman, P. A., and Fyfe, J. A. (1990). Human immunodeficiency virus integration protein expressed in Escherichia coli possesses selective DNA cleaving activity. Proc. Natl. Acad. Sci. USA 87, 5119–5123. Shibagaki, Y., and Chow, S. A. (1997). Central core domain of retroviral integrase is responsible for target site selection. J. Biol. Chem. 272, 8361–8369. Sinha, S., Pursley, M. H., and Grandgenett, D. P. (2002). Efficient concerted integration by recombinant human immunodeficiency virus type 1 integrase without cellular or viral cofactors. J. Virol. 76, 3105–3113. Stevens, S. W., and Griffith, J. D. (1996). Sequence analysis of the human DNA flanking sites of human immunodeficiency virus type 1 integration. J. Virol. 70, 6459–6462. Suzuki, T., Shen, H., Akagi, K., Morse, H. C., Malley, J. D., Naiman, D. Q., Jenkins, N. A., and Copeland, N. G. (2002). New genes involved in cancer identified by retroviral tagging. Nat. Genet. 32, 166–174. Tan, W., Zhu, K., Segal, D. J., Barbas, C. F., 3rd, and Chow, S. A. (2004). Fusion proteins consisting of human immunodeficiency virus type 1 integrase and the designed polydactyl zinc finger protein E2C direct integration of viral DNA into specific sites. J. Virol. 78, 1301–1313. van Gent, D. C., Oude Groeneger, A. A., and Plasterk, R. H. (1993a). Identification of amino acids in HIV-2 integrase involved in site-specific hydrolysis and alcoholysis of viral DNA termini. Nucleic Acids Res. 21, 3373–3377. van Gent, D. C., Vink, C., Groeneger, A. A., and Plasterk, R. H. (1993b). Complementation between HIV integrase proteins mutated in different domains. EMBO J. 12, 3261–3267. Vijaya, S., Steffen, D. L., and Robinson, H. L. (1986). Acceptor sites for retroviral integrations map near DNase I-hypersensitive sites in chromatin. J. Virol. 60, 683–692. Vincent, K. A., Ellison, V., Chow, S. A., and Brown, P. O. (1993). Characterization of human immunodeficiency virus type 1 integrase expressed in Escherichia coli and analysis of variants with amino-terminal mutations. J. Virol. 67, 425–437. Vincent, K. A., York-Higgins, D., Quiroga, M., and Brown, P. O. (1990). Host sequences flanking the HIV provirus. Nucleic Acids Res. 18, 6045–6047.
5. Retroviral DNA Integration—Mechanism and Consequences
181
Vink, C., Lutzke, R. A., and Plasterk, R. H. (1994). Formation of a stable complex between the human immunodeficiency virus integrase protein and viral DNA. Nucleic Acids Res. 22, 4103–4110. Vink, C., Oude Groeneger, A. M., and Plasterk, R. H. (1993). Identification of the catalytic and DNA-binding region of the human immunodeficiency virus type I integrase protein. Nucleic Acids Res. 21, 1419–1425. Wang, J. Y., Ling, H., Yang, W., and Craigie, R. (2001). Structure of a two-domain fragment of HIV1 integrase: Implications for domain organization in the intact protein. EMBO J. 20, 7333–7343. Wei, S. Q., Mizuuchi, K., and Craigie, R. (1997). A large nucleoprotein assembly at the ends of the viral DNA mediates retroviral DNA integration. EMBO J. 16, 7511–7520. Wei, S. Q., Mizuuchi, K., and Craigie, R. (1998). Footprints on the viral DNA ends in moloney murine leukemia virus preintegration complexes reflect a specific association with integrase. Proc. Natl. Acad. Sci. USA 95, 10535–10540. Weidhaas, J. B., Angelichio, E. L., Fenner, S., and Coffin, J. M. (2000). Relationship between retroviral DNA integration and gene expression. J. Virol. 74, 8382–8389. Woerner, A. M., and Marcus-Sekura, C. J. (1993). Characterization of a DNA binding domain in the C-terminus of HIV-1 integrase by deletion mutagenesis. Nucleic Acids Res. 21, 3507–3511. Wong, J. K., Hezareh, M., Gunthard, H. F., Havlir, D. V., Ignacio, C. C., Spina, C. A., and Richman, D. D. (1997). Recovery of replication-competent HIV despite prolonged suppression of plasma viremia. Science 278, 1291–1295. Wu, X., Li, Y., Crise, B., and Burgess, S. M. (2003). Transcription start regions in the human genome are favored targets for MLV integration. Science 300, 1749–1751. Wu, X., Li, Y., Crise, B., Burgess, S. M., and Munroe, D. J. (2005). Weak palindromic consensus sequences are a common feature found at the integration target sites of many retroviruses. J. Virol. 79, 5211–5214. Yang, F., Leon, O., Greenfield, N. J., and Roth, M. J. (1999). Functional interactions of the HHCC domain of moloney murine leukemia virus integrase revealed by nonoverlapping complementation and zinc-dependent dimerization. J. Virol. 73, 1809–1817. Yang, W., and Steitz, T. A. (1995). Recombining the structures of HIV integrase, RuvC and RNase H. Structure 3, 131–134. Yang, Z. N., Mueser, T. C., Bushman, F. D., and Hyde, C. C. (2000). Crystal structure of an active two-domain derivative of Rous sarcoma virus integrase. J. Mol. Biol. 296, 535–548. Yi, J., Asante-Appiah, E., and Skalka, A. M. (1999). Divalent cations stimulate preferential recognition of a viral DNA end by HIV-1 integrase. Biochemistry 38, 8458–8468. Yoder, K. E., and Bushman, F. D. (2000). Repair of gaps in retroviral DNA integration intermediates. J. Virol. 74, 11191–11200. Yung, E., Sorin, M., Pal, A., Craig, E., Morozov, A., Delattre, O., Kappes, J., Ott, D., and Kalpana, G. V. (2001). Inhibition of HIV-1 virion production by a transdominant mutant of integrase interactor 1. Nat. Med. 7, 920–926. Zheng, R., Jenkins, T. M., and Craigie, R. (1996). Zinc folds the N-terminal domain of HIV-1 integrase, promotes multimerization, and enhances catalytic activity. Proc. Natl. Acad. Sci. USA 93, 13659–13664. Zhu, Y., Dai, J., Fuerst, P. G., and Voytas, D. F. (2003). Controlling integration specificity of a yeast retrotransposon. Proc. Natl. Acad. Sci. USA 100, 5891–5895. Zhu, Y., Zou, S., Wright, D. A., and Voytas, D. F. (1999). Tagging chromatin with retrotransposons: Target specificity of the Saccharomyces Ty5 retrotransposon changes with the chromosomal localization of Sir3p and Sir4p. Genes Dev. 13, 2738–2749.
Index A Active rGE confound, 76 Adjuvart, 34 Adolescent-limited antisocial behaviors, 72 Adoption studies latent G x E, 80 latent G x measured E, 81 Adverse circumstances, 81 Alcohol dehydrogenase (ADH) gene, 131 Alcoholism, 81 Alternated live viruses, 27 Aniomic exchange chromatography, 30 Anthropod genetic models gravity-sensing organs, 120 Antigen-presenting cell (APC), 27 Antigens, 26, 27, 33, 34 Antisocial behavior, 82, 85 environmental influences experienced by individual influences, 63 environmental influences shared by family members, 61 genetic influences, 66 gene distribution, 70 Antisocial personality disorder, 69 Artificial alterations, 2 Avian sarcoma-leukosis virus (ASLV), 159 B Bad parenting, 81 Basic helix–loop–helix (bHLH), 126 Behavioral screening of defective mutants, 119 Behavioral-genetic studies, 45–52 BEN molecule, 112 Biological inheritance, 75 Bivariate behavioral-genetic analysis, 69 Bone morphogenetic protein 4 (BMP4), 113
C CaCO3, 114, 115 Cadherin23 (CDH23), 117 Calcareous stone, 108 Campaniform sensillary arrays, 125 CBCL aggression scale, 74 Child behavior checklist, 73 Childhood maltreatment, 82 Chimeric recombinases, 13 Chromatin structure, 17 Clinical trials, 25 Cognate immune responses gene gun, 26, 35 Collateral damage, 17 Consequences of integration insertional mutagenesis, 166 viral latency, 168 CpG islands, 164 C-terminal domains, 6, 12. See also Integrase structures Cytolytic T lymphocytes (CTL), 26–27 D DDE transposases, 8–9, 17 Dentin sialophospho protein (DSPP). See Tooth mineralization Designer site specific recombinases, 15, 18 Dizygotic twin pairs, 82 DNA binding domain C-terminal, 12–13, 15 Zif268, 13–14 DNA damage, 18 DNA fragment, transfected, 3 DNA gap repair enzymes, 151 DNA modification, 3 DNA shuffling, 12 DNA transposases, 8, DNA vaccine potency limitations, 33
183
Index
184 DNA-binding protein, 16, 18 E2C, 16 LexA, 16 Zif268, 16 Drosophila, 111, 113, 118–119, 122, 125, 130, 133 E Ectopis Sir4p binding sites, 16 Endophenotypes, 88 Endoplasmic reticulum protein (GP96), 114 Epidermal growth factor receptor (EGFR), 127 Eschetichia coli, 28–29 Expressed emotion paradigm, 79 Extended twin-family design, 78 F Family adaptability, 78 G -Gal, 11 Gene difference on antisocial behavior, 66 Gene distribution. See Antisocial behavior Gene expression, 18, 80, 111 MAOA, 82 Gene identification using genomics, 134 Gene mutation Dog-eared (Dog), 113 Gene therapy, 8 of X-SCID, 18 Gene–environment interactions, 62, 80. See also Hypothesis testing Genetic influence on psychopathy, 69 Genome sequences, 2 DNA sequence, 2, 4, 18 Genomic sequences, 16 Genomic Surgery, 2–3
Holliday junctions, 5 Homologous recombination (HR), 3, 13 Host proteins, 160 Hot spots formation, 12 Human anti social behavior in adoptees, 53, 55 in twins, 42, 53, 55–56 Human chemical trials, 29, 31 Hyperactive resolve variants, 13 Hypergravity-induced quiescence (HIQ), 134 Hypothesis testing environment causation, 74 gene–environment interaction, 80 strategies for future, 85 I IFN-gamma antigen-specific immune responses, 32 Immune stimulus, 34 In site production, 31 In vitro assays, 150 In vivo integrase composition complexes, 159 In vivo test substrate, 11 Integrase structures, 153 catalytic core, 154, 158 C-terminal domain, 155–156 multimerization, 158 N-terminal domain, 154–155, 158 two domain structures, 156 Integration mechanism, 149 factors involved, 151 DNA breaking and joining reactions, 150 IQ, 72 J Janet’s organ, 124 Johnson’s organ, 124 K
H Hair cells, 107–111, 113–114 mechanosignaling, 116 Hair color, 60 Heritability(ies), 60, 73 Historical cohort differences on antisocial behavior, 68 HIV-1 virus, 159
Kinociliun, 107 Knock out phenomenon, 3 L Lens epithelium-derived growth factor (LEDGF), 162 LMO-2 gene expression, 166
Index Long terminal repeat (LTR) retrotransposous, 9 L-type calcium channel, 120 M Major histocombatibility, complex (MHC) class I molecules, 27 Male adoptees, 80 Mammalian hearing, 109 MAOA genotype(s), 83–84 MAOA polymorphism, 82–84 effects on behavior, 84 Marital discord, 78 Mechanosense organ development, 125 Mechanosensory hairs, 122 Mechanosensory signaling pathways in Drosophila, 128 behavioral studies of gravitaxis, 130 MHC class I processing pathology, 28 Moloney nutrine Leukemia virus, 16 Monozygotic twin pairs, 82 MPQ aggression scale, 68 Murine leukemia virus (MLV), 159 Mutagenesis selection strategies, 10 N Natural Library of sites, 9, 11 Nematode model systems gravity-sensing organs, 135 mechano sense organ development, 136 mechano sensory signaling pathways, 137 Neuronal stimulation, 122 Neurotransmitter systems, 83 Nomp genes, 129 NompC homolog, 118 Nonhomologous DNA end joining (NHEJ), 153 Nonspecific reactions, 17 Notch lateral inhibition system, 126 Notch–Delta signaling, 112 Novel chimeric nucleases, 13 N-terminal domains, 6, 12 O Octoconia–otolith formation, 114
185 P
Passive rGE confound, 76 Pax family of transcription factors, 113 Eyes absent (eya), 113 Sine oculis (So), 113 Dachsund (Dach), 113 Pax–Six–Eya–Dach protein, 113 Personality trials, see human antisocial behavior Phage display technique, 13 Phenotypes of integrase mutants, 149 Physical violence, 73. See also Genetic influence on physical violence Plasmid DNA vaccines action mechanisms, 27 manufacturing, 28 quality standards, 30 safety testing, 30 Plasmid DNA vaccines, 26, 28, 33 Programmed genetic rearrangements, 5 Protein engineering, 10 Protein expression, 34 Protein–protein interactions, 16 Proviral insertional mutagenesis, 167 R Recombinase and transgene delivery, 16 Residual effects, 62 Retrotransposition, 8 Retroviral integration targeting, 162 integration site selection, 162–163 preferred DNA sequence, 162 preferred DNA structure, 162 tethering factors, 165 Retroviral life cycle, 148 essential factors, 149 Reversibility, 17 Risk factors, 76, 84 environment, 87 Rous Sarcoma virus (RSV), 156 S Sachharomyces Cerevisial, 9 Sense organ precursor (SOP) cell, 111 Serene recombinases, 4 Serotonin transporter polymorphism (5-HTTLPR), 86 Site specific recombinases and transposases application considerations, 16
Index
186 Site-specific recombination, 4, 13 applied, 8 SOP cell, 127 Starmaker gene, 115 Starmaker protein, 115, 116 Statocyst-type organs, 107 Stereocilia, 116 Structure based strategies, 12 T T cell, 168 T helper cells, 26 TATA-centred Z-site, 14 Taxonomic theory, 74 Therapeutic transgene, 18 Tidying-up operations, 8 Time-resolved fluorescence anisotrophy (TFA), 157 Toll-like receptor 9 (TLR9), 28 Tooth mineralization, 114 Touch response gentle, 136, 137 hard, 136, 137 nose, 136, 137 texture sensing, 136, 137 Trangene organisms, 18 Transgene expression, 3 Transmembrane Cohlear-expressed gene 1 (TMC1), 117 Transposition mechanism, 8 systems, 9 targeting, 15 Transposon, 8 TRP family, 118 Twin study of latent G x measured E, 81 Ty elements, 165 Type I mechanosense oragans, 125
TYrocine recombinases, 4, 5 Crc, 11–13, 17 FLP, 11, 13 integrase, 4 U Unique environment, 63 V Vaccines for infectious diseases cancer, 33 HIV DNA vaccines, 31 lepatitis B DNA vaccines, 33 malaris DNA vaccines, 32 Vertabrate genetic models, 107 gravity-sensing organs, 107 gravity-sensing organs development, 110 signaling pathways of graviperceptions, 114 Viral DNA, 148 Viral genomic RNA, 150 Viral proteins, 148, 160 W Winnowing process, 126 X X-linked genes, 67 X-linked SCID gene therapy, 167 Z Zebrafish, 110, 112, 114, 119 Zif268. See DNA binding Z-resolvases, 15. See also Chimeric recombinases domain Z-site, 14
