Advances in Virus Research Vol33 (1987)

Advances in

VIRUS RESEARCH VOLUME 33

ADVISORY BOARD

DAVIDBALTIMORE

PAULKAESBERG

ROBERTM. CHANOCK

BERNARD Moss

PETERC. DOHERTY

ERLING NORRBY

BERNARD N. FIELDS

AKIRAOYA

H. J. GROSS

J. J. SKEHEL

B. D. HARRISON

R. H. SYMONS

M. H. V. VAN REGENMORTEL

Advances in

VIRUS RESEARCH Edited by

KARL MARAMOROSCH FREDERICK A. MURPHY Department of Entomology Rutgers University Cook Campus New Brunswick, New Jersey

Division of Viral Diseases Centers for Disease Control Atlanta, Georgia

AARON J. SHATKIN New Jersey Center for Advanced Biotechnology and Medicine Rutgers-UMDNJ Piscataway, New Jersey

VOLUME 33

1987

ACADEMIC PRESS, INC. Harcourt Brace Jovanovich, Publishers

Orlando San Diego New York Austin Boston London Sydney Tokyo Toronto

BY ACADEMIC PRESS,INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM OR BY ANY MEANS. ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY. RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.

COPYRIGHT 0 1987

ACADEMIC PRESS, INC. Orlando. Florida 32887

United Kingdom Edition published by

ACADEMIC PRESS INC.

(LONDON) 24-28 Oval Road. London NWI 7DX

LTD

LIBRARY OF CONGRESS CATALOG C A R D NUMBER 53-11559 ISBN 0-12-039833-8

(alk. paper)

PRINTED IN THE UNITED STATES OF AMERICA 87 88 89 90

9 8 7 6 5 4 3 2 I

CONTENTS Domains of Virus Glycoproteins

MILTONJ . SCHLESINGER AND SONDRA SCHLESINGER I. I1. 111. IV . V. VI . VII . VIII .

Introduction ....................................................... Influenza Virus Hemagglutinin ..................................... Alphavirus Glycoproteins .......................................... Vesicular Stomatitis and Rabies Virus Glycoproteins ................. Retrovirus Glycoproteins ........................................... Other Virus Glycoproteins .......................................... Virus Glycoproteins and Polarized Cells ............................. Summary and Perspectives ......................................... References ........................................................

1 2 8 13 22 28 32 33 35

Flavivirus Replication Strategy

E . G. WESTAWAY I. I1. 111. IV .

Introduction ....................................................... Structure of the Virion ............................................. Replication ........................................................ Summary and Conclusions ......................................... References ........................................................

45 46

50 82 84

The Autonomously Replicating Parvoviruses of Vertebrates

SUSANF. COTMOREAND PETERTATTERSALL I . Introduction ....................................................... I1. Viral Structure and Organization ................................... 111. Viral Life Cycle ................................................... References ........................................................

91 96 137 169

Regulation of Translation by Poliovirus

NAHUMSONENBERG I . Introduction ....................................................... I1. Cap Binding Proteins of Eukaryotic mRNAs ......................... 111. Involvement of the CBP Complex in the Shutoff of Host mRNA Translation after Poliovirus Infection ............................... IV . Alternative Models to Explain Poliovirus Inhibition of Host Protein Synthesis ......................................................... V . Translational Inhibition by Other Picomaviruses .................... VI . Concluding Remarks and Perspectives .............................. References ........................................................ V

175 176 184 191 194 199 200

vi

CONTENTS

Disease Induction by Plant Viruses I . Introduction

L . C . VAN LOON .......................................................

I1. The Interactions of Viruses with Their Hosts ........................ 111. The Genetics of Host-Virus Interactions ............................

IV . Alterations in Host Plant Metabolism ............................... V . Concluding Remarks ............................................... References ........................................................

205 206 224 234 247 249

The Dianthoviruses: A Distinct Group of Isometric Plant Viruses with Bipartite Genome

C. HIRUKI I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Dianthoviruses ........ I11. Diseases Caused in Plan ........................................ IV . Physical. Chemical. and Biochemical Properties . . . . . v . Serological Studies ................................................. VI . Host Range and Symptomatology . VII . Replication ........................................................ VIII . Genetic Reassortment Studies .......... IX . Cytopathology ..................................................... X . Transmission by Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI . Ecological Studies ................................................. XI1. Concluding Remarks ............................................... References ........................................................

257 257 258 260 269 276 278 281 286 289 291 294 297

Banana Bunchy Top: An Economically Important Tropical Plant Virus Disease

JAMESL . DALE I . Introduction ....................................................... I1. The Distribution of Bananas and the Geographical History of BBTV . . I11. The Characteristics of the Virus .................................... IV . Control: The Australian Experience ................................. V. The Future: Possible Directions ..................................... References ........................................................

301 302 306 314 322 324

Approaches to the Study of Vector Specificity for Arboviruses-Model Systems Using Cultured Mosquito Cells

VICTORSTOLLAR ..............................................

I . Introductory Remarks I1. Sindbis Virus Replication in Vertebrate and in Mosquito Cells-A Model System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

327 331

CONTENTS 111. Role of the Host Cell ............................................... IV . Role ofthe Virus .................................................. V . Concluding Remarks ............................................... References ........................................................

INDEX ...................................................................

vii 332 343 362 363 367

This Page Intentionally Left Blank

ADVANCES IN VIRUS RESEARCH,VOL. 33

DOMAINS OF VIRUS GLYCOPROTEINS Milton J. Schlesinger and Sondra Schlesinger Deportment of Microbiology and Immunology Washington University School of Medicine St. Louis, Missouri 63110

I. INTRODUCTION Replication of a virus in a susceptible host cell begins with its attachment to the cell surface and ends with the assembly of newly formed virus components into organized structures which ultimately are released from the cell. For enveloped viruses, attachment is mediated by glycoproteins which form spikes projecting outward from the virion’s surface. The polypeptide backbone of the spike glycoprotein is encoded by virus-specific genes and these proteins are extensively modified after synthesis by host-cell-specific enzymes. Most of the protein mass lies outside the virion’s lipid bilayer but a short sequence of hydrophobic amino acids within the protein spans the bilayer, thereby anchoring the spikes to the membrane. In addition to their role as major determinants of cell tropism these glycoproteins have two other functions in the replication process. They possess a membrane fusion activity which enables the viral nucleoprotein to enter the cell cytoplasm, and they participate in the assembly and budding of new virions. Viral glycoproteins are also important in another context. They are the major determinants to which the immune system responds when an organism is infected with an enveloped virus. The interactions between the components of the immune system and viral glycoproteins are complex and most likely are the determining factor in the outcome of an infection. In this article, we review current information about the structure and function of virus glycoproteins. We do not intend this to be a comprehensive accounting of the virus glycoprotein literature, and have selected a few virus glycoproteins which we feel provide prototypes for illustrating important relationships between the functions cited above and glycoprotein structure. One of the major advances in our knowledge about virus glycoproteins has come from the application of recombinant DNA technology, which led to a determination of the primary structure of many virus glycoproteins as well as to information about the role of various amino acid sequences in glycoprotein 1

Copyright 8 1987 by Academic Press, Inc. All rights of reproduction in any form reserved.

2

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER Virion Surface or lntracellular Vesicles

Virion Interior or Lipid Bilayer Cell Cytoplasm - - 1

I

t

t

f

DOMAINS:

Ecto

Tronsmembranal

Cytoplasmic

FUNCTIONS:

Receptor ( R ) Furlon ( F ) Subunit interacttons Sites for neutralizing and enhancing antibodies

Membrane anchor

Nucleocapsid recognition

FIG.1. A typical transmembranal virus glycoprotein. Some proteins have a reversed orientation with their amino terminus inside and the carboxy terminus outside the virion. The distribution of polypeptide sequences between the ecto- and cytoplasmic domain can also vary greatly; however, the transmembranal domain usually consists of about 20 to 25 hydrophobic amino acids. Most proteins have oligosaccharide groups (0) bound only to asparagine but others have glycosyl residues attached to serine and threonine as well. A posttranslational proteolytic cleavage (noted by A) converts many virus glycoproteins into disulfide-linked dimers. In addition these polypeptides often exist in their native state as oligomers (dimers, trimers, tetramers), held together by strong nonconvalent protein-protein interactions.

localization and function. The amino acid sequences for those glycoproteins discussed here will not be presented; instead we consider these proteins in terms of their structural domains. Figure 1 illustrates these domains and their assigned functions. One set of domains we shall not discuss in depth are those regions of glycoproteins mapped as antigenic determinants or epitopes. An article in a recent volume of this series describes in detail influenza virus glycoprotein antigenic sites and their variability (Air and Laver, 1986).

11. INFLUENZAVIRUSHEMAGGLUTININ The influenza A and B viruses contain two genes encoding membrane glycoproteins-one produces a neuraminidase (NA) and the other a hemagglutinin (HA). There is more information about influenza virus hemagglutinin than any other virus membrane protein and its structure-function relationships could well serve as a paradigm for a large number of similarly structured virus and cell transmembrane glycoproteins. Three factors account for the wealth of data about HA:

DOMAINS OF VIRUS GLYCOPROTEINS

3

(1) a determination of the three-dimensional structure of the crystallized, bromelain-cleaved soluble form of HA by Wilson et al. (19811, (2) the sequencing of a large number of HA polypeptides with distinct serological types (reviewed in Palese and Kingsbury, 19831, and (3) the cloning of a complete cDNA copy of the HA-RNA gene and the ability to form HA in cultured cells carrying this cDNA (Gething et al., 1980; Gething and Sambrook, 1981). We draw on data based on each of these to describe the domains of the HA protein (see Fig. 2). The typical HA protein is initially synthesized as a molecule of 566 amino acids; however, the 16 amino acids at the amino terminus constitute a signal peptide and are cleaved shortly after the complete polypeptide is made. Oligosaccharides are added to asparagine residues at six to seven sites along the chain during synthesis; most of these are attached to the amino-terminal portion of the protein and are on the stalk of the HA trimer. They are also in other regions and can influence the receptor-binding and fusion activities. Processing of these glycosyl residues occurs shortly after their attachment and such modifications can profoundly influence the conformation of the protein (see Section IV,A,l). A model for the folding of this large protein is found in the analysis of the X-ray crystal structure (Wilson et al., 1981). Six disulfide bridges form during folding and prior to a critical, single proteolytic cleavage in a highly conserved region of HA at a site Receptors

FIG.2. The trimer structure of influenza virus HA. (Adapted in part from Wilson et al., 1981.)

4

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

221 amino acids from the carboxy terminus. A protease in the Golgi stacks converts HA to a disulfide-linked dimer of HA1 with 328 and HA2 with 221 residues. A few additional amino acids are removed from the carboxy terminus of HA1. An important event during processing is the formation of HA1-HA2 trimers which are stabilized through a-helical-coiled coils and salt bridges in the stalk region of the protein. Changes in antigenic properties of HA have been detected during these processing events (Bachi et al., 1985; Nestorowicz et al., 1985). A . Ectodomain 1 . Fusion Activity HA has a cryptic fusogenic activity but two changes must occur before this protein can fuse membranes. First, there must be a proteolytic cleavage at a lysine (arginine) at position 328, which produces the disulfide-linked HA1 and HA2 chains (Klenk et al., 1975; Lazarowitz and Choppin, 1975). This cleavage is accompanied by conformational changes in the protein’s structure since the amino terminus of HA2 and the carboxy terminus of HA1 move some 21 A apart. In addition, the HA2 amino terminus folds into the interior of the HA trimer. The importance of this cleavage and subsequent fusogenic activity to influenza virulence has recently been noted as a result of examining the HAS from virulent forms of a chicken influenza virus. Although cells infected with the virulent A/chick/Penn/83 (H5N2) could cleave HA t o HA1 and HA2, the avirulent form of this virus has an HA that required exogenously added trypsin for cleavage (Kawaoka et al., 1984).At least one epitope is different between the virulent and avirulent types (Robertson et al., 1985). Amino acid sequences are not different between HAS in the region of the connecting HA1-HA2 peptide, but the molecular weight of the virulent HA1 appears lower than the avirulent strain except when glycosylation was blocked by tunicamycin, a result indicating that there is one less oligosaccharide in the virulent HA. Four amino acids are changed between avirulent and virulent types and one of these, at position 13 of HA1 located near the HA1-HA2 connecting peptide, is postulated to alter a glycosylation site such that the HA1-HA2 cleavage site now becomes accessible. The actual cleavage sequence contains dibasic amino acids, which are sites for limited proteolysis in a wide variety of proteins including other enveloped virus glycoproteins, peptide hormones, and neuropeptide precursors. The proteolytic activity is localized to a trans-Golgi vesicle (Morrison et al., 1985) and is possibly the calcium-activated, thiol-type protease described by Steiner et al. (1984).

DOMAINS OF VIRUS GLYCOPROTEINS

5

The sequences at the amino terminus of HA2 are the most highly conserved among HA variants. Furthermore, an analogous sequence of very hydrophobic amino acids is found in the fusion proteins of all paramyxoviruses (Spear, 1986), indicating that this site is essential for fusion. This hypothesis is now strongly supported by two separate studies of HA variants that fuse at pH values significantly above that of the wild type. In one series of analyses, variants were selected by growth in chorioantoic membranes treated with amantadine hydrochloride to raise the endosomal pH (Daniels et al., 1985). After two passages and plaquing on chicken embryo fibroblasts in the presence of the drug, a number of isolates hemolyzed red blood cells at pH values 0.2 to 0.8 greater than the wild-type HA. Changes in amino acid sequences of 41 variants were analyzed with regard to their sites in the quaternary structure of the trimer. All changes could be interpreted to affect either the binding of the HA2 amino-terminal sequence to internal regions of the protein or the stability of subunit interactions in the trimer. The other set of experiments consisted of performing in uitro sitedirected mutagenesis in regions of the cDNA encoding the HA2 amino terminus (Gething et al., 1986). Three substitutions were made: glutamic acid for glycine at positions 1and 4 and glycine to glutamic acid at position 11. A glutamic acid at the amino terminus blocked all fusogenic activity, although the HA bound to lipid at lower pH and became protease sensitive. This finding indicates that this HA could still be structurally altered, possibly as a result of dissociation to a monomeric form. Substituting a charged residue at position 4 destabilized HA and led to an increased threshold pH (5.6 compared to 5.3 for normal HA) but a 50% loss in “efficiency” of fusion. The glycine replacement at position 11 extended the hydrophobic character of the HA2 another seven amino acids. This had little effect on HA stability but interfered with the ability of this HA to cause cell-cell fusion. All of these results suggest a dual role for the fusion site. One is coincident with an exposure of the hydrophobic sequence as the trimers “relax” and possibly dissociate; a second is an interaction of the polypeptide with lipid, producing a destabilization and coalescence of bilayers. A two-step kinetic mechanism of HA-mediated fusion has been proposed by Van Meer et al. (1985) in which a rapid association of the two membranes is followed by the actual fusion at a rate varying with lipid composition. It is clear from studies using expressed cDNAs in transfected cultured cells that HA fusion does not require the neuraminidase (White et al., 1982). In other systems employing liposomes containing orthomyxovirus glycoproteins, however, the HA-promulgated fusion is

6

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

detected only when an active neuraminadase is present although it does not have to be inserted into the liposome (Huang et al., 1980). 2 . Receptor Binding Site It has long been known that sialic acid is the host cell surface receptor for influenza virus, but the receptor site on HA has only recently been characterized. A tentative identification of this site is described for the three-dimensional structure (Wilson et al., 1981) and consists of a pocket on the surface of HA at the distal end of the spike in the HA1 subunit (see Fig. 2). Assignment of this region is based on the observation that a variety of HA sequences show a strong conservation for five distinctive amino acids located in the same topological position in the HA structure. More recent analyses of HA variants which differ in their recognition of two different sialic acid structures confirm the assignment of the receptor site (reviewed in Air and Laver, 1986). The variants differ in amino acid 226 of HA1, which is in the postulated surface pocket. Other sites at positions adjacent to the pocket also influence receptor structure since variants with changes in these positions show differences in agglutinating red blood cells containing three distinct sialic acids: N-acetyl-, N-glycolyl-, and N-O-diacetylneuraminic acids (Higa et al., 1985). HA receptor binding is affected by glycosylation at sites on HA1 close to the pocket. For example, influenza B strains harvested after growth on Madin-Darby canine kidney (MDCK) cells differ in their HAS from those strains isolated after adaptation to growth in the chicken embryo allantoic membrane (Robertson et al., 1985). One major difference, a decrease in the oligosaccharide content of the egggrown virus, arose as a result of an amino acid substitution that removed an HA1 glycosylation site adjacent to the receptor site. A glycosylation site is also lost from HA1 in a variant of influenza A WSN strain that is selected during passage of virus on Madin-Darby bovine kidney (MDBK) cells (Deom et al., 1986). Compared to the initial isolate from chicken embryo fibroblasts, the selected variant shows a stronger binding to MDBK cells and erythrocytes (Crecelius et al., 1984).

B . Transmembrane Domain The amino acid sequences of a large number of influenza virus HAS show a hydrophobic region of about 25 amino acids close to the carboxy terminus. The evidence that this region of the glycoprotein tethers the molecule to the lipid bilayer comes from in vitro constructed deletion mutants that remove the carboxy-terminal hydrophobic sequences

DOMAINS OF VIRUS GLYCOPROTEINS

7

(Gething and Sambrook, 1982). A cDNA containing a partial copy of the RNA gene for HA from influenza A/Japan/305/57 was inserted into an SV40 vector and monolayers of CV-1 cells infected with virus stocks prepared so they contained helper and recombinant viruses. The particular cDNA used for the analysis had altered sequences near the 3' end of the gene which replaced in the HA the last 38 amino acids at the carboxy terminus with 11 amino acids that were predominantly hydrophilic. When expressed in the monkey cells, the truncated HA was secreted into the culture medium. HA cDNAs have also been inserted into bovine papilloma virus plasmids and these vectors used for isolation of cell lines which constitutively make the HA proteins. The 3' truncated cDNA encodes a protein that was secreted, but that coding for the normal HA made a glycoprotein that was retained at the surface of the cells (Sambrook et al., 1985). Additional mutants of HA with deletions and insertions in the region containing the anchor sequences have been constructed from cDNAs of HA (Sveda et al., 1982) and SV40 vectors carrying these cDNAs expressed in African monkey kidney cells. Normal HAS were completely cell associated but varying amounts (one-fourth to two-thirds) of mutant HAS were secreted from the cells, and mutant and wild-type HAS were differently glycosylated. Another HA mutant that had a totally different carboxy-terminal sequence beginning near the membrane domain of HA2 was retained inside the cell (Sveda et al., 1984). Clearly, it is possible to affect glycoprotein transport and sorting (see below) by modifying large portions of the membrane domain. C . Cytoplasmic Domain The carboxy-terminal 10 amino acids of HA protrude from the cytoplasmic face of the lipid bilayer. The last five amino acids (Lys. Arg-Ile-Cys-Ile) are highly conserved among HAS of different strains of influenza, suggesting a critical role in HA function. This sequence is not essential, however, for synthesis and transport of the glycoprotein from the endoplasmic reticulum (ER) to the cell surface and rather extensive changes can occur in the cytoplasmic domain without seriously altering intracellular transport (Doyle et al., 1985). These results emerged from in vitro constructions of the HA cDNA in which the carboxy-terminal sequences were mutated to highly diverse structures. The HA cDNA was inserted into the SV40 genome such that high levels of expression occurred and sufficient HA was produced to measure cellular location, state of processing, and fusogenesis as well as hemagglutinating properties of the surface glycoprotein. The most subtle change of three amino acid substitutions (Arg to Ser, Cys t o

8

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

Glu, and Ile to Arg) in the carboxy terminus had no effect on synthesis and transport. A more drastic change in sequence and an elongation of six amino acids, as a result of substituting the cytoplasmic tail of Rous sarcoma virus (RSV) glycoprotein cytoplasmic sequence, also had no effect on transport. Some partial effects were noted in those cytoplasmic tails truncated to three amino acids or extended by 22 amino acids with RSV cytoplasmic sequences. However, extension with 16 amino acids from a bacterial plasmid sequence did interfere with transport of HA from the ER. Another random 16-amino acid sequence also blocked transport, but only after the protein reached the Golgi network. The cytoplasmic domain appears also to be the site for fatty acylation. The human and avian viruses of HO, H1, H2, H3, H7, and H10 serotypes have palmitate linked to the HA2 subunit and the fatty acid is retained in the membrane-bound fragment after bromelain releases HA (Schmidt, 1982). In addition a carboxy-terminal 6-kDa cyanogen bromide peptide isolated from fowl plague virus contains palmitate. Removal of the bound fatty acid from fowl plague virus o r isolated HAS by hydroxylamine inhibits HA-induced hemolysis under conditions that do not affect hemagglutination titers or cause gross morphological change in virus structure or HA conformation (Schmidt and Lambrect, 1986). Neuraminidase activity of virions is unaffected by the hydroxylaminolysis which blocks fusogenic function; however, until the deacylated HA can be restored to function by reacylation, the conclusion that acyl groups are required for fusion must be a tentative one.

111. ALPHAVIRUS GLYCOPROTEINS Glycoproteins of the two closely related alphaviruses, Sindbis and Semliki Forest (SFV), have been studied in considerable detail (reviewed by Schlesinger and Schlesinger, 1986). The two major glycoproteins ( E l and E2) encoded by these viruses are initially expressed as a polyprotein (Strauss and Strauss, 1986). Nascent proteolytic cleavages produce a precursor of E2 (P62) and E l . The two glycoprotein genes are separated by a sequence encoding a 55-amino acid peptide. The E l protein has 439 amino acids with two extended regions of hydrophobic amino acids. The P62 has 487 amino acids and 64 of these are removed during the proteolytic conversion to E2. E l and E2 form a relatively tight, noncovalent heterodimer, and three of these heterodimers are arranged in triangular clusters on the virion’s surface (Harrison, 1986). Figure 3 illustrates the El-E2 spike with four distinct domains.

DOMAINS OF VIRUS GLYCOPROTEINS

9

FIG.3. A model for the El -E2 heterodimer of Sindbis virus. (From Schlesinger and Schlesinger, 1986.)Domains noted are hydrophilicglobular portion (I),membrane-spanning region (2),cytoplasmic fragment (3), and hydrophobic area not in the membrane (4); ) oligosaccharides; (4)acetylated amino terminus; (-4 covalent fatty acids; hydrophobic regions.

(v

The P62 contains a hydrophobic sequence of 19 amino acids at the amino terminus of the nascent polypeptide, which functions as a signal sequence for insertion and transfer of this protein into the ER lumen. Unlike most signal sequences, however, the P62 sequence is glycosylated at an asparagine residue in the middle of the sequence and there is no release of this sequence by signal peptidase. In addition, a significant fraction of the P62 molecules are acetylated at the amino-terminal residue (Bell et al., 1982). For insertion of the E l protein through the ER membrane, there is a separate signal sequence which consists of the carboxy-terminal one-third part of a 6-kDa peptide located between the P62 and E l genes (Melancon and Garoff, 1986). The 6-kDa peptide does not appear in the virion (Welch and Sefton, 1979). Early (immature) forms of P62 and E l can be distinguished from the E2 and E l structures on the virion by appropriate monoclonal antibodies (Roehrig et al., 1982; Schmaljohn et al., 1983; Burke et al., 1983). In addition, the weak protein-protein interactions

10

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

between P62 and E l become much stronger when P62 is converted to E2 (Rice and Strauss, 1982; Ziemiecki and Garoff, 1978). Transport of E l from the ER t o the cell surface is facilitated by the presence of P62 (Hashimoto et al., 19811, but P62 can traverse the secretory pathway in the absence of E l (Garoff et al., 1983). A . Ectodomain 1 . Fusion Activity The heterodimer spikes of alphaviruses contain a cryptic fusion site which is activated by treating virions or infected cells at pH values of 5.5-6.0 (White and Helenius, 1980). The regions of the glycoprotein assigned to this function are postulated to consist of a conserved, hydrophobic amino acid sequence between positions 45 and 58 of the E l subunit (Rice and Strauss, 1981). With SFV, in addition to low pH, fusion requires cholesterol or related P-hydroxysterols embedded in the target membrane (Kielian and Helenius, 1985; White and Helenius, 1980). Dependence on cholesterol was noted earlier for the binding of low-pH-treated Sindbis virus to liposomes (Mooney et al., 1975). At low pH, irreversible changes occur in the conformation of both E l and E2 of SFV (Kielian and Helenius, 1985) and the E2 of Sindbis virus (Edwards et al., 1983). Additional information comes from analysis of water-soluble forms of the El-E2 complex released from their membrane anchor domains by brief protease treetment (Kielian and Helenius, 1985). The soluble complex has much weaker heterodimer binding and monomeric E2 could be studied. E l retained a resistance to protease after low pH but only in the presence of membrane cholesterol, suggesting a binding of cholesterol to the E l hydrophobic sequence near its amino terminus. At low pH, the water-soluble form of E2 becomes somewhat hydrophobic but is still unable to bind to liposomes. Fusion activity as measured by syncytial formation is detectable at the surface of mammalian cells expressing cDNA copies of SFV genome sequences for E l and E2 (Kondor-Koch et al., 19831, but low pH is required. In a cDNA with the sequences for the carboxy terminus of E l altered, no E l was at the surface and fusion did not occur, a result confirming a role for this protein in fusion. However, absence of E l also interferes with P62 processing. In Sindbis virus-infected cells, the P62-El complex (Mann et al., 1983) has fusion activity. Isolated preparations of E l from Western equine encephalitis virus reconstituted into lipid bilayers are able to lyse erythrocytes (Yamamoto et al., 1981). Antibodies specific to E l can inhibit fusion (Chanas et al.,

DOMAINS OF VIRUS GLYCOPROTEINS

11

19821, and monoclonal antibodies have been found that recognize an E l epitope appearing after low pH (Schmaljohn et al., 1983). At least one SFV fusion mutant has been isolated based on its ability to have a lower pH threshold for fusion, but location of the mutation has not yet been determined (Kielian et al., 1984). 2 . Host Range Determinants Alphaviruses have a broad host range; their natural hosts are arthropods and avian species. No specific host cell surface structure has been identified as a receptor and no unique region of the spike has been assigned as a site for receptor. Neutralizing antibodies are mainly directed against the E2 glycoprotein (Roehrig, 19861, but some monoclonal antibodies directed against epitopes on E l have neutralizing activity (Schmaljohn et al., 1983; Boere et al., 1984). A change in the ionic charge of the surface of Sindbis virus can have a profound effect on the host range of the virus (Symington and Schlesinger, 1975). The mutations that led to changes in the host range and surface charge of the variant were located in both E l and E2, but the precise place in the primary sequence was not determined (Symington and Schlesinger, 1978). Adaptation of an avirulent Sindbis virus strain t o neurovirulence led also to changes in both E l and E2 (Stanley et al., 1985); however, a selection for Sindbis virus variants based on rapid growth in baby hamster kidney (BHK) cells and avirulence in suckling mice led to a strain with an alteration only in E2 (Olmsted et al., 1984).

B . Transmembrane Domain In common with most virus transmembranal glycoproteins, the membrane domains for the alphavirus consist of 20-30 hydrophobic amino acids close to the carboxy terminus of the polypeptide and have basic amino acids at the cytoplasmic face of the bilayer. For SFV E2, the membrane domain could be replaced by an analogous region of the VSV G or the fowl plague virus hemagglutinin, yielding chimeric transmembranal proteins which are transported to the cell surface and retain fusogenic capacity (Riedel, 1985). Other site-directed mutations in the P62/E2 cDNA have been constructed to determine effects of inserting a single charged amino acid (i.e., glutamic acid) in the middle of the domain and of removing the positively charged amino acids at the cytoplasmic face of the bilayer (Cutler and Garoff, 1986; Cutler et al., 1986). For the latter, the wild-type sequence of Arg-Ser-Lys was changed to Gly-Ser-Glu or Gly-Ser-Met. When expressed in mammalian cells in constructs that also contained E l sequences or lacked them, the three mutants were translocated and glycosylated, and

12

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

reached the cell surface at efficiencies similar to the wild-type P62/E2, regardless of whether E l was present or absent. At the surface, they retained fusogenic activity after low pH treatment, like the wild-type protein. They were nondefective in P62 proteolytic processing except in the absence of E l when, like the wild type, cleavage did not occur but transport of P62 alone to the cell surface was observed. One major difference was detected between the mutants and wild type when membranes were treated at pH 11.5: at this pH, only 10% of the wildtype glycoprotein was released but 4 0 4 0 % of the mutant proteins were lost from the membranes. These results show that a hydrophobic sequence alone, in the absence of cytoplasmic charged groups, can stop the translocation process and that the basic amino acids contribute to anchorage stability. The lack of effect of the glutamic acid in the middle of the membrane domain is puzzling and raises the possibility that a nonhelical conformation with fewer amino acids can span the lipid bilayer. The alterations tested in these experiments apparently do not affect those P62-El interactions that are needed for proteolytic cleavage and E l transport.

C . Cytoplasmic Domain Only one of the alphavirus glycoproteins (P62/E2) has a substantial number of amino acids extending into the cytoplasm. The E l protein has two basic amino acids at this position, whereas E2 has 31 amino acids which include three cysteines and three prolines at sites conserved among three different alphavirus strains (Strauss and Strauss, 1985). Fatty acids are postulated to be bound to some of the cysteines (Magee et al., 1983). The E2 of Sindbis virus contains three palmitic acid groups which are acylated to the glycoprotein after synthesis but before transport to the Golgi (Berger and Schmidt, 1985; Schmidt and Schlesinger, 1980). Sindbis virus E l has one fatty acyl group postulated to be bound to a hydroxyamino acid (Magee et al., 1983). The cytoplasmic portion of P62 may loop back through the bilayer during polypeptide synthesis since P62 is believed to be released from the nascent polyprotein by a signalase activity located on the lumenal side of the membrane. It has been proposed that the E2 cytoplasmic domain binds to nucleocapsid during viral assembly at the plasma membrane (Ziemiecki and Garoff, 1978). What prevents this interaction from occurring at intracellular membranes is unclear and, in fact, initiation of the assembly process may well take place intracellularly (Johnson et al., 1981).An inhibition in glycoprotein intracellular transport imposed by the ionophore monensin leads to an accumulation of P62 and E l in

DOMAINS OF VIRUS GLYCOPROTEINS

13

Golgi membranes and nucleocapsids are found on these membranes (Johnson and Schlesinger, 1980; Kaariainen et al., 1980). The cytoplasmic domain is not essential for intracellular transport, since cDNAs with deletions in this domain express a P62 which moves to the cell surface (Garoff et al., 1983). The largest deletion left only three amino acids beyond the membrane domain, but deleting both the cytoplasmic and membrane sequences produced a protein which remained in the endoplasmic reticulum, presumably because of misfolding and aggregation.

IV. VESICULARSTOMATITIS AND RABIESVIRUS GLYCOPROTEINS The spikes covering the surface of the vesicular stomatitis and rabies virions are composed of a single species of glycoprotein, the G protein. The amino acid sequences of these G proteins have been deduced from cDNA sequences, rabies G by Anilionis et al. (1981) and VSV G by Rose and Gallione (1981). Both proteins are similar in size; the rabies (ERA) G has 523 amino acids and the VSV (San Juan) has 511 amino acids. A comparison of the two sequences shows only a 20% identity with the introduction of seven gaps. There are, however, several regions including those of the carboxy-terminal glycosylation site that show a stronger homology (Rose et al., 1982). These two proteins contain amino-terminal signal sequences that are removed during synthesis of the polypeptide and they are oriented so their carboxy termini form the cytoplasmic domain. There is little information about the tertiary structure of these glycoproteins but studies using cross-linking reagents indicate that the G on the surface of VSV is a trimer (Dubovi and Wagner, 1977; Mudd and Swanson, 1978). More recent results obtained with VSV-infected cells show the presence of G oligomers (Kreis and Lodish, 1986). A soluble form of G that is monomeric is produced by treatment of VS virions with cathepsin D at pH 5 (Crimmins et al., 1983). Although exposure to low pH could be responsible for dissociation of an oligomer, G protein solubilized from virions with octyl-6-glucosidealso appears to be monomeric (Crimmins et al., 1983). The finding that these treatments of VSV produce a monomeric form of G can be reconciled with the data indicating that G is a n oligomer if interactions between monomers are weak. Such interactions may be analogous to those that occur between the trimers on the surface of alphavirus virions which are disrupted by nonionic detergents (Harrison, 1986). The subunits of the trimer consist of E1-E2 heterodimers that are stable to nonionic detergents. VSV has a broad host range, and infection leads to a rapid and

14

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

synchronous formation of virus-specific proteins accompanied by inhibition of host cell protein synthesis. As a result, the G protein of VSV has provided a valuable model for many studies on the synthesis and localization of membrane glycoproteins. This section is devoted almost entirely t o studies of the G protein of VSV, but there are two areas of research with the rabies G protein that deserve special mention. First, considerable effort has gone into mapping the antigenic domains of this protein. That work has recently been reviewed (Wunner et al., 1985) and will not be discussed further. Second, the nicotinic acetylcholine receptor is thought to be a receptor for rabies virus and the evidence for this proposal is reviewed below.

A . Ectodomain Most of the mass of the G protein extends outward from the surface of the virion. This domain includes two sites for attachment of oligosaccharides, the region that binds cellular receptors, and the sequences involved in the fusion of the virus membrane with cellular membranes. 1 . Oligosaccharide Sites

There are two potential asparagine-linked oligosaccharide sites at residues 178 and 335 in the VSV G protein, based on the deduced amino acid sequence which contains two Asn-X-Ser/Thr stretches (Rose and Gallione, 1981). This is the amino acid configuration that specifies an asparagine-linked glycosylation site (Marshall, 1972, 1974). Both sites contain complex oligosaccharides (Etchison and Holland, 1974; Reading et al., 1978). Glycosylation normally begins with the transfer of the precursor oligosaccharide, Glc,Man,GlcNAc, from the carrier lipid, dolichol phosphate, to nascent polypeptides. The extent to which this precursor is processed determines the final structure of the asparagine-linked oligosaccharides on mature glycoproteins (Kornfeld and Kornfeld, 1985). The first step in processing, the removal of the three glucose residues, occurs shortly after completion of the polypeptide chain while the protein is still in the ER. The synthesis of complex oligosaccharides involves the removal of all but three of the mannose residues and the addition of N-acetylglucosamine, galactose, sialic acid, and fucose. These reactions take place in the Golgi vesicles. Initial studies on the role of glycosylation in the synthesis and localization of the G protein made extensive use of the drug tunicamycin, an antibiotic which inhibits the synthesis of N-acetylglucosaminylpyrophosphorylpolyisoprenol and prevents the addition

DOMAINS OF VIRUS GLYCOPROTEINS

15

of any carbohydrate to asparagine residues of potential glycoproteins (Struck and Lennarz, 1980).The effects of this drug on VSV depend on which strain of the Indiana serotype is analyzed (Gibson et al., 1979; Chatis and Morrison, 1981).The two strains, VSV (San Juan) and VSV (Orsay), have related but distinct polypeptides. In the presence of tunicamycin, virus yields for both strains are severely inhibited at 38°C. At 30°C the yield of VSV (San Juan) is still inhibited by tunicamycin but that of VSV (Orsay), containing nonglycosylated G protein, is nearly equal to that in the absence of drug. The retention of VSV (Orsay) production at 30°C correlates with the ability of the nonglycosylated G to fold correctly at the lower temperature (Gibson et ul., 1979). Thus, in the absence of carbohydrate, the folding of G protein becomes temperature sensitive, but the sensitivity depends on the amino acid sequence of the protein. A comparison of the deduced amino acid sequences of G (San Juan) and G (Orsay) reveals a difference of nine amino acids in the ectodomain (Gallione and Rose, 1985). Four of these represent an increase in hydrophilic amino acids in G (Orsay). Rose and Gallione pointed out that at position 179, which is six amino acids from the glycosylation site, there is a tyrosine in G (San Juan) and an aspartic acid in G (Orsay). The presence of more hydrophilic amino acids may explain why the oligosaccharide requirement for G (Orsay) is less stringent than that of G (San Juan). To determine the role of each of the two glycosylation sites on the intracellular transport of G, Machamer et ul. (1985) site-mutagenized the cloned cDNA of the G protein. Their results show that retention of either oligosaccharide site permits the G protein to reach the cell surface at 37"C, but when both sites are removed, a condition analogous to treatment with tunicamycin, the protein appears in a Golgi-like compartment of the cell and does not reach the cell surface. The role of oligosaccharides in determining the proper localization and function of the G protein has also been analyzed using cell mutants and drugs that affect the processing pathways. The importance of the structure of the oligosaccharides on G to the formation of VSV was first seen in a mutant cell line which transfers Glc,Man,GlcNAc, instead of the normal precursor to nascent polypeptides (Gibson et al., 1981). The yield of VSV is temperature sensitive in this mutant and VSV (San Juan) is more temperature sensitive than VSV (Orsay). Recently, the drugs l-deoxynojirimycin and castanospermine were found to inhibit the glucosidases which act at the initial stages of oligosaccharide processing (Saunier et al., 1982; Saul et al., 1983; Pan et al., 1983).To determine if these early processing events and some of the following steps in oligosaccharide processing are critical for certain proteins, the yields of VSV (San Juan) and VSV (Orsay) in the

16

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

presence of these drugs and in cell variants altered in the processing pathway were measured (Schlesinger et al., 1984).These studies show that a block in the removal of the glucose residues by l-deoxynojirimycin, by castanospermine, or by growth in a cell mutant lacking glucosidase I1 inhibits the yields of VSV (San Juan), but not of VSV (Orsay), at 40°C. Inhibition of later oligosaccharide processing steps has no effect on virus yield. In these experiments, the G protein reaches the cell surface, indicating that the alteration in the G protein is insufficient to prevent migration to the cell surface, but is significant enough to prevent virion formation. The finding that it is only the initial processing of the oligosaccharide that affects the formation of VSV suggests that the structure of the oligosaccharideplays an important role in the folding of the polypeptide chain. By the time the glucose residues have been removed, the polypeptide has achieved a conformation that is no longer influenced by the oligosaccharide structure. The effect of l-deoxynojirimycin on a number of other glycoproteins is consistent with this conclusion, since some glycoproteins such as the hemagglutinin of influenza virus and IgM are not affected by this drug (Burke et al., 1984; Peyrieras et al., 19831, while other glycoproteins, for example IgD (Peyrieras et al., 19831, acetylcholine receptor (Smith et al., 19861, and al-proteinase (Gross et al., 19831, are affected.

2. Receptor Binding and Fusogenic Activity The oligosaccharide chains play little or no role in binding of virus t o cells since virions with no carbohydrate on the G protein are infectious (Gibson et al., 1978). Potential sites on G required for attachment to cells have not been identified, although Schlegel et al. (1983) suggest that phosphatidylserine is a receptor for VSV. The G protein itself is a fusogen under appropriate conditions. The cell surface expression of G protein from cloned cDNA is sufficient to cause cell-cell fusion if the cells are subjected to a brief treatment at acid pH (Riedel et al., 1984; Florkiewicz and Rose, 1984).The effect of low pH on the conformation of G has not been examined in detail, but a reversible conformation change in the cathepsin D-treated, solubilized form of G is observed when the protein is acidified to pH 5 (Crimmins et al., 1983). Schlegel and Wade (1984) found that a 25-amino acid peptide corresponding to the amino terminus of the G protein is a pHdependent hemolysin. Antibodies prepared against this peptide are nonneutralizing, but react with denatured protein. This result suggests that the amino terminus is buried in the native protein and may become exposed upon acidification. Further studies by Schlegel and Wade (1985) with smaller peptides identified the six amino-terminal

DOMAINS OF VIRUS GLYCOPROTEINS

17

amino acids as the hemolytic domain, a result supported by the observation that a single amino acid change of a lysine to glutamic acid at the amino terminus abolished the peptide’s hemolytic activity. However, when the amino terminus of the G is changed by site-directed mutagenesis of the G cDNA to give the same amino terminus as the inactive peptide, the intact G protein retains its pH-dependent fusion activity (Woodgett and Rose, 1986).Thus, the domain in intact G responsible for fusion has yet t o be identified. 3. Receptor Binding Activity of Rabies Virus Rabies virus invades neuronal cells as a result of retrograde axonal transport along peripheral nerves to the spinal chord and eventually to the brain. The first suggestions that acetylcholine receptors might act as receptors for rabies virus came from the observation that rabies virus was distributed on mouse diaphragms and cultured chick myotubes coincident with the receptor (Lentz et al., 1982).Furthermore, binding of virus to these tissues was prevented by a-bungarotoxin, a n irreversible inhibitor of the nicotinic acetylcholine receptor. Lentz et al. (1984)found a significant degree of homology between the sequences of neurotoxins and the rabies glycoprotein. There is a 50% identity between residues 189 to 214 of the glycoprotein and alignment positions 30 to 56 of the neurotoxins. These findings raise the possibility that the neuronal cell tropism and resulting pathogenesis of rabies virus may be due to the affinity of the viral glycoprotein for the acetylcholine receptor. Lentz (1985)pointed out, however, that the acetylcholine receptor is not the only receptor for rabies virus since cells lacking this receptor are susceptible to the virus. The similarity between a domain of the rabies glycoprotein and that of neurotoxins may reflect an evolutionary relatedness between this viral protein and a cellular protein.

B . Transmembrane Domain Most transmembrane glycoproteins contain a stretch of about 20 hydrophobic amino acids which span the bilayer in the form of an ahelix. The 20 amino acids of the G protein that span the membrane can be identified by inspection of the carboxy-terminal sequence (Kyte and Doolittle, 1982).They have also been defined by several types of experiments. Protease digestion of microsomes prepared from VSV-infected cells removes only 20-30 amino acids from the carboxy terminus, presumably because the rest of the protein is buried inside the microsomes (Katz et al., 1977;Chatis and Morrison, 1979).In contrast, protease treatment of intact virions protects a carboxy-terminal frag-

18

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

ment which includes the cytoplasmic tail and the adjacent hydrophobic sequences (Rose et al., 1980). Furthermore, there is a naturally occurring form of G (Gs) lacking both the membrane-spanning and carboxyterminal regions, which is secreted from infected cells (Kang and Previc, 1970; Little and Huang, 1978). A similar form of G has been constructed by making a specific deletion in the cDNA clone expressing the G protein, and this form of G is also secreted from cells (Rose and Bergmann, 1982). To determine the actual requirements for spanning the lipid bilayer, Adams and Rose (1985a,b) analyzed two types of alterations in this region of G protein. They generated specific deletions and also changed a specific amino acid. In those examples in which the transmembrane region is shortened, G proteins containing 18,16, or 14 amino acids are still able to be transported to the cell surface. When this region contains only 12 or 8 amino acids the G protein still spans the bilayer but it is transported only to Golgi-like regions. A mutant in which the transmembrane sequence is deleted is only detected in the cell in the ER and behaves much the same as the form of G protein lacking both the transmembrane and cytoplasmic domains. Secretion of the former, however, has a half-time of about 12 hours which is much slower than that of the latter which is about 2-4 hours (Rose and Bergmann, 1982). Adams and Rose (198513) also altered the membrane-spanning domain by replacing an isoleucine with either glutamine or arginine. The substitution of glutamine for isoleucine has no effect on membrane anchoring or localization to the cell surface. When arginine is the substituted amino acid, however, the protein still spans the membrane but is transported poorly to the cell surface. Another approach to examining domains of a protein is to produce chimeric proteins, either exchanging a segment of one protein for another or adding a segment from one protein to another (Riedel, 1985). The exchange of domains between VSV G protein and influenza HA produced polypeptides that are not transported to the cell surface (McQueen et al., 1984), but a hybrid between the HA and a retrovirus glycoprotein is transported (see Section VII). Guan and Rose (1984) fused the membrane-spanning and cytoplasmic domains of the G protein cDNA to the cDNA encoding rat growth hormone, a protein that is normally secreted from cells. The fused protein becomes membrane bound, does not reach the cell surface membrane, and appears to remain in the Golgi. Guan et al. (1985) then created, by site-directed mutagenesis, glycosylation sites in this protein, and showed that a single site in either of two positions or glycosylation at both positions allows the protein to be transported to the cell surface. The evidence

DOMAINS OF VIRUS GLYCOPROTEINS

19

that the fused protein reaches the cell surface came from cell surface immunofluorescence and lactoperoxidase-catalyzed cell surface iodination. Although it was not possible to calculate the percentage of the protein that reached the cell surface, the authors did find that iodination of the singly glycosylated protein was at least 10-fold higher and that of the doubly glycosylated protein 34-fold greater than the iodination of the original, nonglycosylated protein. One explanation for this result is that carbohydrate acts as a signal for transport to the cell surface. This interpretation, however, does not take into account the variety of results obtained by treating cells with tunicamycin to prevent the addition of carbohydrate. Although tunicamycin treatment does affect the transport of many glycoproteins to the cell surface, there are a significant number of membrane proteins that do reach the cell surface in the absence of glycosylation (Gibson et al., 1980). If carbohydrate acts as a specific recognition signal, it can do so only for a subset of glycoproteins. The result described above in which two different VSV G proteins behave differently when they are not glycosylated (Gibson et al., 1979) complicates any attempt to divide membrane glycoproteins into such subsets. The finding that an oligosaccharide chain at different sites on a polypeptide permits cell surface transport is difficult to interpret in the absence of a three-dimensional structure of the protein. The explanation that an oligosaccharide chain can have effects on the folding, conformation, or stability of some proteins provides an interpretation that encompasses all of the data so far described.

C . Cytoplasmic Domain The cytoplasmic domain of the G protein consists of the carboxyterminal 29 amino acids (Fig. 4). Rose and Bergmann (1983) altered this domain by creating deletion mutants and in some cases constructing plasmids such that a stretch of amino acids derived from SV40 was added to the carboxy-terminal tail. These altered sequences are shown in Fig. 4. Two criteria were used to analyze the effect that these changes had on the ability of the G protein to migrate to the cell surface. The first was the rate at which the protein oligosaccharides become resistant to endoglycosidase H (Endo H). Endo H cleaves highmannose oligosaccharide chains but not complex oligosaccharides from the polypeptide chain. The acquisition of the complex sugars (Nacetylglucosamine and galactose) converts the oligosaccharide to an Endo H-resistant form and this event occurs in the Golgi apparatus; thus, the loss of Endo H sensitivity can be correlated with the move-

20

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

G PROTEIN

END0 H RESISTANCE (HALF-TIME )

PERCENT P.M. POSlTIVE

~0.3HR

100

K S S I A S F F F I l G L l IGLFLVLRVGIHLCIKLKHTKKRQIYTDIEMNRLGK

1428

______

1429

______ L E G S l Q T

SKDRSRHCK I H

3.5 HRS

0

3.5 HRS

0

CC TCGHGGg atcGRTCCffiACATGATAAGATACATTGATGAGTTTGGACAAACCACAACTAGAATGC AGTGA P R G l D P D M l R V l D E F G q T T T R M ~ * L E G S l Q T * S R U R S R H D K I H ’

Hpal 1473

______

GTTCcctcgaGTTAAC V P S S *

Xba I

1514

_______

ACCcctcgaCTCTAGAG T P R L *

Hpal

1554

______

CGACcctcgaGTTAAC R P S S *

FIG.4. The amino acid sequences of the carboxy terminus of wild-type G protein and mutationally altered derivatives. The predicted amino acid sequence of the carboxy terminus of the VSV G protein is shown on the top line with the hydrophobic transmembrane segment indicated by the first 21 amino acids. Below this are the predicted amino acid sequences of the G proteins specified by the deletion mutants, in which the dotted line indicates the presence of normal G protein sequence followed by the new amino acid residues encoded by SV40 nucleotide sequences. Deletions are numbered by the nucleotide residue in the G mRNA (Rose and Gallione, 1981)to which the deletion extends. The nucleotide sequence shown at the bottom is the vector sequence to which each of the deletions is joined via the indicated XhoI linker sequence. The sequence GATC at the junction is derived from filling in of the BamHI site. The origin of the “extra” sequences in the deleted G proteins is illustrated by the translation of this SV40 sequence in the three possible reading frames. The predicted sequences of the carboxy termini of the deleted G proteins, which should be synthesized after insertion of HpaI and XbaI linkers, are shown with the sequences specified by each of the parent plasmids. The nucleotide sequences a t the junction of the VSV G sequence with the linker are indicated along with the predicted protein sequences for each junction. The sequence shown in lowercase resulted from filling in of the BamHI site in pSVGL2. The asterisk indicates translation termination sites. The approximate half-times required for the oligosaccharides on each of the proteins to become resistant to endoglycosidase H digestion is given. The percentage of the transfected cells which showed internal G protein labeling and cell surface labeling is indicated as “percent P.M. (plasma membrane) positive.” This figure is reproduced with permission of Cell (Rose and Bergmann, 1983).

DOMAINS OF VIRUS GLYCOPROTEINS

21

ment of a polypeptide to the Golgi membranes. The second criterion was the percentage of transfected cells showing both internal and cell surface immunofluorescence. All of the mutants require longer times to become Endo H resistant and move from the ER to the Golgi at a much slower rate than the wild-type G protein. The proteins fall into three catagories illustrated in Fig. 4. The first group (numbered 1428, 1429, and 14731,which either lack the carboxy-terminal cytoplasmic domain or have a completely different cytoplasmic domain, acquire Endo H resistance with a half-time about 10-fold longer than the wild-type G protein, and are not detected on the plasma membrane. The second, designated 1514,in which 13 amino acids following the membrane-spanning domain are retained followed by a stretch of amino acids coming from the SV40 vector, is slow to become Endo H resistant, but does reach the plasma membrane. The third category, a protein (1554)with almost all of the correct cytoplasmic domain but with an added 12 amino acids, was a surprise because in most cells it does not become Endo H resistant and appears not to reach the Golgi membranes. A derivative of this mutated protein (1554H1)was constructed by the introduction of a translation termination codon between the G gene sequences and the SV40 sequences (Fig. 4). This mutated protein now behaves in a manner indistinguishable from the wild-type protein. Similar alterations of the proteins in the first group do not permit recovery of movement. These studies demonstrate an important role for the cytoplasmic domain of the G protein in the proper localization of this protein. One possibility is that the cytoplasmic domain can affect the oligimerization of the protein and thus the conformation of the G polypeptide. Alternatively, there is increasing evidence that integral membrane proteins are transported between organelles via vesicles. Rose and Bergmann suggested that the cytoplasmic domain could influence the transport of G to sites in ER where membrane vesicles form or that the formation of membrane vesicles is affected by the structure of the cytoplasmic tail. VSV G, in common with many membrane glycoproteins, contains fatty acids covalently bound to the cytoplasmic domain (see Schlesinger, 1985). Acylation occurs posttranslationally in the ER with transfer from palmitoyl-CoA (Berger and Schmidt, 1985) to a cysteine. The identification of this amino acid as the acceptor site on the polypeptide comes from a series of studies based on both the stability of the acyl-protein’s linkage and site-directed mutagenesis experiments. Magee et al. (1983)noted that the fatty acids are removed from the protein by treatment with neutral hydroxylamine, a reaction indicative of a labile thioester bond, and the deacylated G forms disul-

22

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

fide-linked dimers. Disulfide-linked oligomers of Sindbis virus E1-E2 proteins are detected under similar deacylation conditions. Rose et al. (1984) used the cDNA of G to change the cysteine codon in the cytoplasmic domain of VSV (Indiana) to serine and showed that this mutated form of G was not acylated. The nonacylated G is transported to the cell surface membrane; thus, acylation is not essential for intracellular localization. Schlesinger and Malfer (1982) showed that blocking the acylation in VSV-infected cells with the antibiotic cerulenin does not inhibit G movement to the cell surface but does prevent virus budding. These data indicate that acylation has an important function but it is not clear yet what that function is. There are strains of VSV, in particular those of the New Jersey serotype, that replicate perfectly well and their G proteins are not acylated (Kotwal and Gosh, 1984). Among the differences in the sequence between the New Jersey and the Indiana serotypes of VSV G is a substitution of serine for cysteine in the cytoplasmic domain (Gallione and Rose, 1983; Rose et al., 1984). Thus, fatty acids are not essential for the function of some VSV G proteins. A possible role for this modification is to block free -SH groups in the cytoplasmic domain so that aberrant covalent oligomer formation cannot occur.

V. RETROVIRUSGLYCOPROTEINS Retroviruses have been isolated from a wide variety of species and show a diverse spectrum of disease potentials. The most detailed information about these viruses and their glycoproteins has been obtained for the avian, murine, and, in recent years, human retroviruses. The glycoproteins of retroviruses share many of the characteristics of those viral glycoproteins in which the amino-terminal domain is exposed on the surface (Fig. 1).The synthesis and overall structure of different retrovirus glycoproteins appear similar; the major difference is in the size of the polypeptide. Retrovirus glycoproteins are translated from a spliced subgenomic RNA (Hayward, 1977; Mellon and Duesberg, 1977; Weiss et al., 1977). The primary product, synthesized on membranebound polyribosomes, is a glycosylated precursor of large molecular weight, 92,000 for Rous sarcoma virus (RSV) and 90,000 for murine leukemia virus (MuLV) (see Dickson et al., 1982, for review). This precursor is cleaved to produce gp85 and gp37, and gp70 and p15E from the avian and murine protein, respectively. The cleaved polypeptides are covalently bound by disulfide bridges (Leamnson and Halpern, 1976). The larger amino-terminal fragments (gp85 or gp70) are heavily glycosylated; there are 14 potential glycosylation sites on gp85

DOMAINS OF VIRUS GLYCOPROTEINS

23

of Rous sarcoma virus, all of which are thought to be glycosylated (Schwartz et al., 1983; Hunter et al., 1983). The smaller polypeptides (gp37 or p15E) are derived from the carboxy terminus of the precursor and comprise the membrane spanning and cytoplasmic domains of the glycoproteins. Some sequences in p15E are highly conserved among retroviruses, and the isolated p15E is reported to be immunosuppressive (Cianciolo et al., 1984,1985). It is of considerable interest that a synthetic peptide of 17 amino acids, corresponding to a highly homologous region, is able to mimic this immunosuppressive activity (Cianciolo et al., 1985). The retrovirus glycoproteins play a role in the life cycle of the virus similar to that of other viral glycoproteins discussed here. They are essential for the adsorption to and penetration of the virus into the host cell, for neutralization of infectious virus, and for the interference specificities of viruses. The major focus of this section is the glycoprotein gp85 of Rous sarcoma virus. We have also included a description of the glycoprotein of the murine spleen focus-forming virus. This glycoprotein is implicated in the pathogenicity of the virus, and the studies identifying the domain associated with the disease potential are relevant to the theme of this article.

A. Ectodomain of the Rous Sarcoma Virus Glycoprotein A specific region in a protein may be recognized by a biological parameter before it is identified as a specific stretch of amino acids. In this case, to make the correlation between a biological activity and an amino acid sequence it is essential to determine if the “region” is a contiguous stretch of amino acids. An example of a biological property of RSV that is associated with specific regions of amino acids in gp85 is the host range specificity of this virus. Rous sarcoma viruses have been classified into five subgroups based on their ability to infect genetically defined chicken cells (Weiss, 1982). There are at least three autosomal loci in chickens that encode susceptibility determinants for the three RSV subgroups A, B, and C. These loci are thought to encode specific virus receptors. The conclusion that gp85 determines the host range specificity of RSV stems from the finding that viruses with a defect in the enu gene assume the host range of the helper virus (Weiss, 1982). TI-resistant oligonucleotide mapping of the enu gene initially defined a region of the gene that segregated with a particular subgroup phenotype (Coffin et al., 1978; Joho et al., 1975). Based on this initial finding Doerner et al. (1985) determined the sequence of the enu gene encoding amino acids 8 through 280 of gp85 from a RSV of subgroup B and Rous-

24

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

associated virus (subgroup El, and compared the deduced amino acid sequence with those of RSV, subgroup C. Two variable regions termed h r l and hr2 were defined based on the decreased sequence homology in these regions. The region h r l consists of 32 amino acids beginning with amino acid 137 and corresponding to nucleotide 5654 in the Prague-WV-C genome. The hr2 region begins at amino acid 207, corresponding to nucleotide 5846, and extends for a total of 27 amino acids. A further test of the significance of these variable regions was made by sequencing the relevant region of the enu gene of NTRE-4, a recombinant virus between Prague-RSV, subgroup B and RAV-0. This recombinant recognizes both the subgroup B receptor on chicken cells and the subgroup E receptor found on turkey cells. The sequence analysis shows that the region h r l in NTRE-4 comes from the subgroup B genome, but the hr2 region comes from subgroup E. Thus, both regions appear to be involved in the determination of host range. A further analysis of the sites in gp85 involved in host range determinants was made by Bova et al. (1986). They sequenced molecular clones of the enu gene derived from a subgroup A and a subgroup B virus. Of the four variable regions they describe, two of them, VR-2 and VR-3, correspond to h r l and hr2, respectively. To establish the role of these variable regions in host range determination Bova et al. produced recombinant viruses by substituting a fragment of the gp85 sequence from either RAV-2 (subgroup B) or RAV-0 (subgroup E) for the equivalent fragment in the cDNA clone for the subgroup A genome. These hybrid cDNAs were transfected into susceptible cells to produce virus stocks. These molecularly cloned viruses display the host range expected for the particular cDNA fragment inserted.

B . Cytoplasmic domain of the Rous Sarcoma Virus Glycoprotein Hunter and colleagues have analyzed the effects of deletions and substitutions in the carboxy-terminal domain on the transport and subcellular localization of the RSV glycoprotein (Wills et al., 1984). The enu gene of RSV was inserted into an SV40 expression vector and the effects of mutations on the viral glycoprotein were analyzed in CV-1 cells. The rate of transport of the viral glycoprotein to the Golgi cisternae and to the cell surface were not affected by alterations in the five amino acids at the carboxy terminus. Changing the composition of these amino acids and lengthening the tail had no effect. Removal of 15 amino acids from the carboxy terminus and addition of 4 unrelated amino acids did slow the rate of movement to the Golgi apparatus but did not inhibit the ultimate transport to the cell surface. Finally, re-

DOMAINS OF VIRUS GLYCOPROTEINS

25

moval of both the cytoplasmic and transmembrane domains blocked transport and the truncated protein was not secreted. It may be relevant to these findings that the 22 amino acids of the cytoplasmic domain can be subdivided into an 18-amino acid segment that is highly conserved among strains of RSV and the most carboxyterminal amino acids which show wide divergence (Hughes, 1982; Hunter et al., 1983). These data with the RSV glycoprotein show a pattern similar to that found for the other viral glycoproteins, namely that only some of the amino acid changes made in this region prevent the glycoprotein from reaching the plasma membrane.

C . Domains of the Glycoprotein of Spleen Focus-Forming Virus Spleen focus-forming virus (SFFV)is a complex of a competent helper murine leukemia virus and a defective virus. The complex is responsible for causing an erythroleukemia in mice and it is the defective genome that is the causal component of the disease (see Ruscetti and Wolff, 1984, for a review). There have been several independent isolations of the SFFV complex; the first, F-SFFV, was described by Friend in 1957 and the second, R-SFFV, by Rauscher in 1962. The defective component of the SFFV complexes has been biologically cloned free of helper virus, making it possible to analyze the defective genomes in the absence of helper virus. The defective genomes of F-SFFV and R-SFFV contain different amounts of the retrovirus genome, and code, to different extents, for viral-specific proteins. The critical part of the genome, however, is that region coding for the envelope glycoprotein. The glycoproteins coded by F-SFFV and RSFFV have apparent M, values of 54,000 (gp54) and 55,000 (gp55), respectively. These proteins are associated with the disease potential of SFFV. The enu gene of the defective SFFV is a recombinant containing sequences from murine leukemia virus and from mink-cell focus-inducing (MCF) virus (Troxler et al., 1977a,b). MCF viruses are able to grow in both mouse and heterologous cells (Teich, 1982). They are, themselves, recombinants between ecotropic and xenotropic murine leukemia virus and can be distinguished by several criteria including the presence of specific antigens (Cloyd et al., 1979). Studies from several laboratories (reviewed in Ruscetti and Wolff, 1984) on both the defective genomes of SFFV and the glycoprotein coded by the genomes demonstrate that the 3’ terminus (the carboxyterminal domain of the protein) is derived from the murine leukemia virus genome, whereas the 5’ sequences (the amino-terminal domain) are derived from the MCF genome.

26

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

A more detailed and exact comparison between the defective enu gene of SFFV and the enu gene of murine leukemia virus followed the cloning and sequencing of these genes (Amanuma et al., 1983; Wolff et al., 1983; Clark and Mak, 1983). The defective genomes have a large deletion spanning the gp70-pl5E junction site (Fig. 5). In addition there is a single base insertion in the p15E domain of both gp54 and gp55 that shifts the reading frame leading not only to a different set of terminal amino acids, but also to a shorter carboxy terminus on p15E. Although the exact position of the insertion and the particular base inserted are not always the same, all of the SFFV enu genes so far examined have this insertion, implying that it may be crucial to the expression of the SFFV phenotype. The premature termination leads to the lack of any significant cytoplasmic domain. How this affects the transport of this particular glycoprotein is not known, but as discussed earlier, alterations in the cytoplasmic tail can affect the movement of a glycoprotein through the intracellular membrane compartments. Furthermore, only a small polyprollne hinae

p 15E cleavage site

single b a s e insertion

d t gp70

9P54-55

FIG. 5. A model showing the origin and domains of the gp54-55 of SFFV. The dualtropic (dt) gp70 from which the SFFV glycoprotein is derived is shown above. V1 represents xenotropic sequences; V2 is derived from ecotropic sequences. This figure is reproduced by permission of Dr.David Kabat.

DOMAINS OF VIRUS GLYCOPROTEINS

27

fraction of the SFFV defective glycoprotein does reach the cell surface, indicating that there is a defect in intracellular transport (Srinivas and Compans, 1983). Recent studies of Pinter and Honnen (1985) show that a highly processed form of the glycoprotein of F-SFFV, termed gp65, is secreted from cells. The structure of the glycoprotein of the SFFV proposed by Kabat and colleagues is shown in Fig. 5. Machida et al. (1985) presented evidence that there are two independent domains, V-1 and V-2, joined by a proline-rich stretch. Strong support for this model comes from their finding that Staphylococcus aureus V8 protease cleaves the SFFV glycoprotein to yield the amino-terminal V-1 fragment and the carboxy-terminal V-2 fragment. The cleavage occurs at Glu-238, located within the proline-rich region. This region also marks the division between the ecotropic sequences (V-2) and the xenotropic sequences (V-1). The position at which the amino acid sequences of the glycoprotein become strongly homologous to murine leukemia viral glycoprotein sequences occurs exactly at the end of the proline-rich stretch. There are several other results supporting the model that V-1 and V-2 are independent domains. Thus, disulfide bonds are only detected within the V-1 domain and are not found between domains; mutations in one domain appear to affect the protease sensitivity of that domain but not of the other domain. The analyses of nonleukemogenic mutants of SFFV and revertants of these mutants have provided an important tool in developing a model for the structure of the glycoprotein and for establishing the role of the defective glycoprotein in the pathogenesis of SFFV. The enu genes of two nonleukemogenic mutants were molecularly cloned and sequenced, as was one leukemogenic revertant (Li et al., 1986). One mutant contains three noncontiguous point mutations that cause nonconservative amino acid changes in the V-1 domain. It is not clear if all three changes are required for the loss in pathogenicity. A second mutant contains a point mutation leading to an early termination codon at amino acid residue 304. This mutation occurs immediately after the proposed xenotropic-ecotropic recombination site and therefore eliminates the V-2 domain including the membrane anchor. The leukemogenic revertant regains the wild-type sequence at this site. These results establish that a point mutation in the env gene can lead to the loss in leukemogenic potential of SFFV. enu gene mutants were also constructed by the insertion of small inphase HpaI or XhoI linkers into different restriction sites in the cloned F-SFFV proviral DNA (J.-P. Li and D. Kabat, personal communication). Three mutants with insertions in V-1, the xenotropic region, are no longer pathogenic, although one mutant with an insertion in this

28

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

region retains activity. Two mutants with insertions in V-2, the ecotropic region, and one mutant with an insertion in the proline-rich region retain their pathogenicity. Li and Kabat (personal communication) found that glycoproteins coded by pathogenic F-SFFV are expressed on the cell surface but the glycoprotein of nonpathogenic mutants remains intracellular. They also reported a correlation between F-SFFV pathogenicity and the ability to cause a weak interference to superinfection by dualtropic murine leukemia viruses. The latter result suggests that the SFFV glycoprotein which can cause leukemia is also able to interact with viral receptors. These authors suggested that the xenotropic domain of the SFFV membrane glycoprotein binds to these receptors and causes erythroblastosis. They proposed that these receptors normally function as receptors for a hematopoietic growth factor and the SFFV glycoprotein causes its constitutive activation. VI. OTHERVIRUSGLYCOPROTEINS There are of course many more virus glycoproteins than those described above. Most of these have structural and functional domains which are similar to those we have described; however, there are others which are quite distinct in structure and membrane orientation. Some of these are noted below.

A. Influenza Virus Neuraminidase Four distinct domains can be identified in the subunits of the enzymatically active tetramer of influenza virus neuraminidase (NA). One of these, the cytoplasmic domain, is unusual among transmembrane glycoproteins in that it is the precise amino terminus of the polypeptide chain and consists of only six amino acids (Blok and Air, 1982;Blok et al., 1982). These residues are highly conserved and invariant among nine serologically distinct subtypes. It has been postulated that this domain, in common with other glycoprotein cytoplasmic sequences, interacts with viral components associated with the virus core or with a virus matrix protein. A second domain consists of a nonconserved sequence of -28 hydrophobic amino acids which probably serves as both a signal peptide and a membrane anchor. Most of the NA structure extends outward from the virion’s surface and is composed of two domains: a short helical stalk near the membrane and a large globular hydrophilic head distal to the surface of the virus. Multiple disulfide bridging occurs within

DOMAINS OF VIRUS GLYCOPROTEINS

29

NA subunits and the typical NA has four sites to which N-linked oligosaccharides are bound. Two glycosylation sites are in the stalk region and are possibly involved in interchain contacts of the tetramer; another site is found near the surface at a region close to a subunit interface. No proteolytic modifications are known for NA. The catalytic site of this protein is near the surface in the globular head domain. X-Ray analysis of crystals of NA released from membranes by pronase cleavage at amino acid residues 74-77 in the stalk permitted assignments of specific amino acids in the catalytic site (Varghese et al., 1983;Colman et al., 1983).Sialic acid is the product of NA catalysis and binds in a large pocket on the surface. Nine acidic, six basic, and three hydrophobic amino acids surround this pocket and all are conserved in sequence among all NA serotypes. Amino acid substitutions at positions adjacent to several of the invariant residues, however, are found in serologically distinct subtypes arising from antigenic drift. For example, Asp-151 and -152are invariant but residue 153 varies among subtypes. From the structure, the conserved amino acids face inward toward the pocket and the variable site is oriented outward toward the surface, a geometry in accord with the experimental data. Thus, NA antigenicity, which is known to vary among the virus isolates from different flu epidemics, can be modified in the absence of effects on catalysis. Studies with cDNAs encoding NA have yielded additional information about the NA domains. Substitution of leucine for tryptophan at position 178 in the catalytic pocket destroyed enzymatic activity (M. R. Lentz and Air, 1986).Deletions in the membrane anchor domain significantly affected the translocation and glycosylation of nascent polypeptide (Markoff et al., 1984). It is not completely clear what role NA plays in virus replication but its enzymatic activity will remove host cell receptors, thereby allowing elution of progeny virus from infected cells and, as well, preventing self-aggregation of virions. NA allows also for enhanced mobility of virus through mucin encountered in normal routes of infection in nature.

B . Paramyxoviruses Another interesting variation in virus glycoprotein function is the presence of both a neuraminidase and hemagglutination activity in a single glycoprotein (HN) of the paramyxoviridae family with the fusion activity in a separate glycoprotein (F). In contrast, as noted earlier, the influenza virus hemagglutinin contains the virus fusion activity while the neuraminidase is in a separate glycoprotein. In

30

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

common with influenza virus neuraminidase, the Simian virus 5 (Hiebert et al., 1985) and Sendai virus (Blumberg et al., 1985) HNs have amino termini in the cytoplasmic part of the bilayer. The paramyxovirus F glycoprotein has a fusion region closely resembling that of the influenza HA (see above), and amino acid sequences share high degrees of homology to the hydrophobic amino terminal sequences of the HA2 subunit. The F proteins, however, can act as fusogens in the absence of a low pH environment. All of these F proteins are similar t o influenza HA in that they must be proteolytically cleaved in order to activate the fusion site. These F proteins are oriented with their amino terminus on the outside of the virus membrane and carboxy terminus inside. In contrast, the nonfusogenic glycoproteins of these paramyxoviruses (for example, the G of respiratory syncytial virus) have the amino terminus inside and the carboxy terminus outside (Wertz et al., 1985). This distinction in orientation between the fusion protein and the G/HN proteins has been proposed to explain differences in transport rates from the ER to the cell surface, with F proteins (carboxy-terminal anchors) moving much more rapidly than HN/G (amino-terminal anchor) (Blumberg et al., 1985). The sequence of the major glycoprotein (G) of the respiratory syncytial virus has recently been derived from a cDNA clone (Wertz et al., 1985) and shows a very high content of serine and threonine (30.6%of the total amino acid composition). This is a characteristic of glycoproteins that have carbohydrate linked via O-glycosidic bonds, and indirect evidence based on studies with tunicamycin indicates that this protein is extensively glycosylated on hydroxyamino acids. No aminoterminal signal sequence exists nor is there a hydrophobic membrane anchor domain near the carboxy terminus. These results have led to the suggestion that this G protein has its amino terminus, consisting of about 38 amino acids, in the cytoplasm. A hydrophobic sequence from residues 38 to 66 would serve as a signal sequence and membrane-spanning domain, and the balance of the 232 amino acids would constitute an ectodomain. This portion of the polypeptide has 77 of the 91 hydroxyamino acids which are believed to be sites for glycosylation. The protein also has an unusually high content of prolines (10.1%). C . Coronaviruses There are two membrane-associated glycoproteins (El and E2) in the virions of the Coronaviridae family. One of these, the E l of mouse hepatitis virus, resembles the G protein of respiratory syncytial virus, discussed above, in that there are a number of O-linked sugars on the

DOMAINS OF VIRUS GLYCOPROTEINS

31

protein (Sturman and Holmes, 1983). Coronavirus E l appears to be tightly membrane associated and its derived amino acid sequence shows a very hydrophobic region near the amino terminus. A postulated topological distribution of this protein in the membrane shows the amino terminus in the vesicle lumen (outside of the virion), a looping of sequences across the membrane several times, and a long hydrophilic carboxy-terminal region at the cytoplasmic face affixed to the polar groups of the bilayer (Rottier et al., 1986). The other coronavirus membrane glycoprotein, noted E2 or S, is more conventional in its structure; there is an amino-terminal signal sequence and a carboxy-terminal hydrophobic membrane anchor domain. The protein is posttranslationally cleaved from a molecule of 180 kDa to two equivalent-sized subunits of 90 kDa. The carboxy-terminal subunit (90A) has a fusogenic activity that is enhanced after the proteolytic cleavage. Unlike the ortho- and paramyxoviruses, the fusogenic region appears not to be localized to the region of the polypeptide cleavage since the amino acids around this site are not hydrophobic (L.S. Sturman, personal communication). The most likely region for a fusion site is some 200 amino acids from the amino-terminal side of the cleavage site, where there is an extended sequence of hydrophobic amino acids. Another unusual feature of the coronavirus E2 is a clustering of cysteines in the cytoplasmic domain. This sequence contains a tricysteine and two dicysteines; a somewhat similar arrangement is found also in the cytoplasmic tails of the E2 glycoprotein of alphaviruses.

D . Herpesviruses Relatively little information is currently available about domains of glycoproteins encoded by the herpesviruses, but the genetic analyses of these proteins is now in progress and we can expect considerable more data in the near future. At least four glycoproteins (gB, gC, gD, gE) are encoded in the herpesvirus genome (reviewed by Spear, 1985). One of these, gB, is essential for virion growth and mutations in this protein affect rate of virus entry. This protein appears to have a fusion activity, defined by sites in the genome encoding gB (Bzik et al., 1984). The gC and gE are nonessential for virus growth in tissue culture and do not appear in virions. Mutants which are truncated at the carboxy terminus of gC have been obtained and the altered polypeptide is secreted into the medium, indicating a membrane anchor domain exists at the carboxy terminus of gC (Homa et al., 1986). The herpes gC may have a receptor for the C3b component of complement (Friedman et al., 19841, and gE has a site which binds the Fc portion of the immunoglobulin (Baucke and Spear, 1979; Para et al., 1982). These

32

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

proteins are probably made by the virus to thwart the host immune system.

E . Nonenveloped Viruses The presence of nonessential virus genes coding for glycoproteins is not limited to the herpesviruses. Adenovirus type 2 carries within its genome sequences for a 19-kDa glycoprotein which never appears in the virion and is nonessential for virus growth in tissue culture. This glycoprotein is believed to assist the virus in growth in animals since it is found in a complex with the HLA and &-microglobulin in a manner that inhibits transport of the HLA to the cell surface (Burgert and Kvist, 1985). Without sufficient HLA molecules on an infected cell surface, cytotoxic T cells are unable to bind and destroy the infected cell. Another role for a virus-coded glycoprotein is found in the Reoviridae family whose members do not have lipid bilayers or spikes in the virions. One member of this family, the human rotavirus, produces a glycoprotein which participates in the assembly and secretion of virus into intracellular organelles. Later, the lipid suface is removed from the virion. Expression of the cDNA of this virus glycoprotein has been studied, and deletions of one of its two putative hydrophobic domains altered the location of this glycoprotein in the cell membrane (Poruchynsky et al., 1985). The normal protein remains in the ER where the virus buds and transiently contains a lipid envelope. The protein carrying the deletions was transported from ER to Golgi and secreted. The authors also tested glycoproteins with deletions in the other hydrophobic domain and found no change in location. Thus, one of the hydrophobic domains of the glycoprotein is essential for specifying localization to the ER membrane.

VII. VIRUSGLYCOPROTEINS AND POLARIZED CELLS An important feature of enveloped virus glycoproteins is their apparent ability to determine which cellular membrane is used for virus assembly. We have alluded to this property indirectly in the earlier discussions of the structure and posttranslational processing of these proteins, but there is more direct evidence for this role of the glycoprok i n in the virus-infected polarized epithelial cell. Rodriguez-Boulan and Sabatini (1978) first noted a specific distinction in the localization of virus budding between the apical and basal lateral membranes of this kind of cell. They found that influenza virus buds only from the

DOMAINS OF VIRUS GLYCOPROTEINS

33

apical surface whereas VSV is secreted from the basal lateral membrane. More recent experiments confirm this specific sorting of other enveloped viruses: paramyxoviruses are found at the apical surface and retroviruses and herpesviruses move to the basal lateral surface (Roth et al., 1983; Srinivas et al., 1986). It is the glycoproteins of these viruses that determine which membrane of the cell is utilized, and there have been intensive efforts to determine what properties of these proteins specify sorting (reviewed by Simons and Fuller, 1986). Sorting is not dependent on glycosylation pattern or sialic acid, and chimeric recombinant molecules have been constructed and expressed from cDNAs to determine which domains of the protein control sorting. A chimeric molecule composed of an influenza HA as the luminal portion and a VSV G as the transmembrane and cytoplasmic portion localizes according to that part determined by the luminal domain. That is, the chimeric protein moves to the apical surface (McQueen et al., 1986). This result implicates the “ectodomain” as the determinant for sorting. However, a conflicting set of data were found with “recombinant” molecules containing the ectodomain of the Friend mink cell focus-inducingretrovirus but lacking the normal transmembranal and cytoplasmic domains. In this case the protein sorts to both the basal lateral and apical surfaces of the cell where it is secreted (Stephens and Compans, 1986). Thus, the signal for sorting in the polarized cell remains unknown. AND PERSPECTIVES VIII. SUMMARY

The primary sequences of many viral membrane glycoproteins are now known. Based on inspection of their sequence most of these proteins can be divided into the three major domains described in Fig. 1. These domains have been defined with respect to their orientation in the lipid bilayer, but a complete description of a domain should also include its quaternary structure and function. This is possible, however, only for the HA and NA of influenza virus. X-Ray crystallographic studies provide us with a picture of what the ectodomain of these proteins looks like and permit a specific function to be correlated with a definitive structure. The ectodomains of viral glycoproteins are responsible for several important functions; receptor binding, fusogenic activity, and disease potential are among those discussed here. Although in some cases these functions can be associated with specific sequences of a protein, the crystallographic data will be essential to complete the picture. One focus of this article has been the studies involving directed

34

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

mutagenesis and construction of chimeric proteins. The effects of altering specific amino acid sequences, of swapping domains, and of adding a new domain to a protein serve to define the functions of a domain and to show that a domain can be independently associated with a specific function. The experiments described have been carried out by inserting the genes of particular viral glycoproteins, as cDNAs, into expression vectors and transcribing the cDNAs from the promoter provided by the expression vector. This approach established that localization and functions such as the fusogenic activity are properties of the viral glycoprotein per se and do not require other viral-coded components. The altered proteins have been analyzed for their ability to reach the appropriate location in the cell and to undergo the expected posttranslational modifications. Viral glycoproteins must also be able to participate in the assembly of infectious virions, and some of the changes, particularly those in the transmembrane and cytoplasmic domains that do not affect localization, may not permit critical interactions needed for the budding of particles. Now that we have accumulated some details about the requirements for transport, future work should be directed toward the requirements for virion assembly. We have concentrated most of our discussion on those viral glycoproteins that (1)span the lipid bilayer once, (2)are oriented such that the carboxy terminus comprises the cytoplasmic domain and (3) contain asparagine-linked oligosaccharides. An increasing number of viral glycoproteins that don’t conform to this description are now undergoing scrutiny. They include proteins such as the E l of coronavirus that may span the membrane more than once, and those oriented with the amino terminus in the cytoplasmic domain. There are also viral glycoproteins with extensive 0-linked glycosylation, some of which have been noted here. Viral glycoproteins have served as important models for cellular membrane glycoproteins that localize to the outer surface of the plasma membrane. Not all viral glycoproteins move to the cell surface and some remain in internal membranes at sites of virion assembly. These distinctions among viral glycoproteins may reflect the diversity of cellular membrane glycoproteins; therefore, more detailed knowledge of different viral glycoproteins should provide valuable models for the spectrum of cellular glycoproteins. Further analyses of these viral glycoproteins will also surely contribute to our understanding of virion assembly and pathogenesis.

ACKNOWLEDGMENTS We wish to thank our colleagues who sent us their reprints and material prior to publication. We particularly want to thank Dr. David Kabat for his help and for providing the model for Fig. 5.

DOMAINS OF VIRUS GLYCOPROTEINS

35

REFERENCES Adams, G. A,, and Rose, J. K. (1985a).Incorporation of II charged amino acid into the membrane-spanning domain blocks cell surface transport but not membrane anchoring of a viral glycoprotein. Mol. Cell. Biol. 5, 1442-1448. Adams, G. A., and Rose, J. K. (1985b).Structural requirements of a membrane-spanning domain for protein anchoring and cell surface transport. Cell 41, 1007-1015. Air, G. M., and Laver, W. G. (1986).The molecular basis of antigenic variation in influenza virus. Adu. Virus Res., in press. Amanuma, H., Katori, A., Obata, M., Sagata, N., and Ikawa, Y. (1983).Complete nucleotide sequence of the gene for the specific glycoprotein (gp55) of the Friend spleen focus-forming virus. Proc. Natl. Acad. Sci. U S A . 80, 3913-3917. Anilionis, A., Wunner, W. H., and Curtis, P. J. (1981).Structure of the glycoprotein gene in rabies virus. Nature (London) 294, 275-277. Bachi, T., Gerhard, W., and Yewdell, J. W. (1985).Monoclonal antibodies detect different forms of influenza virus hemagglutinin during viral penetration and biosynthesis. J . Virol. 55,307-313. Bauke, R. B., and Spear, P. G. (1979).Membrane proteins specified by herpes simplex viruses. V. Identification of an Fc-binding glycoprotein. J. Virol. 32, 779-789. Bell, J. R.,Kinney, R. M., Trent, D. W., Strauss, E. C., and Strauss, J. H. (1982).The Nterminus of PE2 in Sindbis virus-infected cells. Virology 119,255-267. Berger, M., and Schmidt, M. F. G. (1985).Protein fatty acyltransferase is located in the rough endoplasmic reticulum. FEBS Lett. 187, 289-294. Blok, J., and Air, G. M. (1982).Variation in the membrane insertion and “stalk” sequences in eight subtypes of influenza type A virus neuraminidase. Biochemistry 21, 4001-4007. Blok, J., Air, G. M., Laver, W. G., Ward, C. W., Lilly, G. G., Woods, E. F., Roxburgh, C. M., and Inglis, A. S. (1982).Studies on the size, chemical composition and partial sequence of the neuraminidase (NA) from type A influenza viruses show that the Nterminal regions of the NA is not processed and serves to anchor the NA in the viral membrane. Virology 119, 109-121. Blumberg, B., Giogri, C., Roux, L., Rqju, R., Dowling, P., Chollet, A., and Kolakofsky, D. (1985).Sequence determination of the Sendai virus HN gene and its comparison to the influenza virus glycoproteins. Cell 41, 269-278. Boere, W. A. M., Harmsen, T., Vinje, J., Benaissa-Troun, B. J., Kraaijeveld, C. A., and Snippe, H. (1984).Identification of distinct antigenic determinants on Semliki Forest virus by using monoclonal antibodies with different antiviral activities. J. Virol. 52, 575-582. Bova, C. A., Manfredi, J. P., and Swanstrom, R. (1986).Enu genes of avian retroviruses: Nucleotide sequence and molecular recombinants define host range determinants. Virology 152,343-354. Burgert, H.-G., and Kvist, S. (1985).An adenovirus type 2 glycoprotein blocks cell surface expression of human histocompatibility class I antigens. Cell 41, 987-997. Burke, B., Walter, C., Griffiths, G., and Warren, G. (1983).Viral glycoproteins at different stages of intracellular transport can be distinguished using monoclonal antibodies. Eur. J. Cell Biol. 31, 315-324. Burke, B., Matlin, K., Bause, E., Legler, G., Peyrieras, N., and Ploegh, H. (1984).Inhibition of N-linked oligosaccharide trimming does not interfere with surface expression of certain integral membrane proteins. EMBO J. 3, 551-556. Bzik, D. J., Fox, B. A., DeLuca, N. A., and Person, S. (1984).Nucleotide sequence of a region of the herpes simplex virus type 1 gB glycoprotein gene: Mutations affecting rate of virus entry and cell fusion. Virology 137, 185-190. Chanas, A. C., Gould, E. A., Clegg, J. C. S., and Varma, M. G. R. (1982).Monoclonal

36

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

antibodies to Sindbis virus glycoprotein E l can neutralize, enhance infectivity, and independently inhibit haemagglutination of haemolysis. J. Gen. Virol. 58, 37-46. Chatis, P. A., and Morrison, T. G. (1979). Vesicular stomatitis virus glycoprotein is anchored to intracellular membranes near its carboxyl end and is proteolytically cleaved at its amino terminus. J . Virol. 29,957-963. Chatis, P. A., and Morrison, T. G. (1981). Mutational changes in the vesicular stomatitis virus glycoprotein affect the requirement of carbohydrate in morphogenesis. J. Virol. 37,307-316. Cianciolo, G. J., Kipnis, R. J., and Snyderman, R. (1984). Similarity between p15E of murine and feline leukemia viruses and p21 of HTLV. Nature (London) 311, 515. Cianciolo, G. J., Copeland, T. D., Oroszlan, S., and Snyderman, R. (1985). Inhibition of lymphocyte proliferation by a synthetic peptide homologous to retroviral envelope proteins. Science 230, 453-455. Clark, S. P., and Mak, T. W. (1983). Complete nucleotide sequence of an infectious clone of Friend spleen focus-forming provirus: gp55 is an envelope fusion glycoprotein. Proc. Natl. Acad. Sci. U.S.A. 80, 5037-5041. Clegg, J. C. S., Chanas, A. C., and Gould, E. A. (1983). Conformational changes in Sindbis virus E l glycoprotein induced by monoclonal antibody binding. J . Gen. Virol. 64,1121-1126. Cloyd, M. W., Hartley, J. W., and Rowe, W. P. (1979). Cell-surface antigens associated with recombinant mink cell focus-inducing murine leukemia viruses. J.Exp. Med. 149, 702-712. Coffin, J. M., Champion, M., and Chabot, J. M. (1978). Nucleotide sequence relationships between the genomes of an endogenous and exogenous avian tumor virus. J. Virol. 28, 972-991. Collins, P. L., Huang, Y.T., and Wertz, G. W. (1984). Nucleotide sequence of the gene encoding the fusion (0 glycoprotein of human respiratory syncytial virus. Proc. Natl. Acad. Sci. U S A . 81, 7683-7687. Colman, P. M., Varghese, J. N., and Laver, W. G. (1983). Structure of the catalytic and antigenic sites in influenza virus neuraminidase. Nature (London) 303, 41-44. Crecelius, D. M., Deom, C. M., and Schulze, I. T. (1984). Biological properties of a hemagglutinin mutant of influenza virus selected by host cells. Virology 139, 164177. Crimmins, D. L., Mehard, W. B., and Schlesinger, S. (1983). Physical properties of a soluble form of the glycoprotein of vesicular stomatitis virus a t neutral and acidic pH. Biochemistry 22, 5790-5796. Cutler, D. F., and Garoff, H. (1986). Mutants of the membrane-binding region of Semliki Forest virus. I. Cell surface transport and fusogenic activity. J. Cell Biol. 102, 889901. Cutler, D. F., Melancon, P., and Garoff, H. (1986). Mutants of the membrane-binding region of Semliki Forest virus E2 protein. 11. Topology and membrane binding. J . Cell B i d . 102, 902-910. Daniels, R. S., Downie, J. C., Hay, A. J., Knossow, M., Skehel, J. J., Wang, M. L., and Wiley, D. C. (1985). Fusion mutants of the influenza virus hemagglutinin glycoprokin. Cell 40,431-439. Deom, C. M., Caton, A. J., and Schulze, I. T. (1986). Removal of a complex oligosaccharide from the tip of a n influenza A virus hemagglutinin potentiates host cellmediated virus selection. Proc. Natl. Acad. Sci. U.S.A. 83, 3171-3775. Dickson, C., Eisenman, R., Fan, H., Hunter, E., and Teich, N. (1982). Protein biosynthesis and assembly. In “RNA Tumor Viruses” (R. Weiss, N. Teich, H. Varmus, and J. Coffin, eds.), pp. 513-648. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York.

DOMAINS OF VIRUS GLYCOPROTEINS

37

Dorner, A. J., Stoye, J. P., and Coffin, J. M. (1985). Molecular basis of host range variation in avian retroviruses. J. Virol. 53, 32-39. Doyle, C., Roth, M. G., Sambrook, J., and Gething, M J . (1985). Mutations in the cytoplasmic domain of the influenza virus hemagglutinin afect different stages of intracellular transport. J. Cell Biol. 100, 704-714. Dubovi, E. J.,and Wagner, R. R. (1977). Spatial relationships of the proteins of vesicular stomatitis virus: Induction of reversible oligomers by cleavable protein cross-linkers and oxidation. J. Virol. 22, 500-509. Edwards, J., Mann, E., and Brown, D. T. (1983). Conformational changes in Sindbis virus envelope proteins accompanying exposure to low pH. J . Virol. 46, 1090-1097. Etchison, J. R., and Holland, J. J. (1974). Carbohydrate composition of the membrane glycoprotein of vesicular stomatitis virus grown in four mammalian cell lines. Proc. Natl. Acad. Sci. U S A . 71, 4011-4014. Florkiewicz, R. Z., and Rose, J. K. (1984). A cell line expressing vesicular stomatitis virus glycoprotein fuses a t low pH. Science 226, 721-723. Friedman, H. M., Cohen, G. H., Eisenberg, R. J., Seidel, C. A., and Cines, D. B. (1984). Glycoprotein C of HSV-1acts as a receptor for the C3b complement component on infected cells. Nature (London) 309, 633-635. Gallione, C. J.,and Rose, J. K. (1983).Nucleotide sequence of a cDNA clone encoding the entire glycoprotein from the New Jersey serotype of vesicular stomatitis virus. J. Virol. 46, 162-169. Gallione, C. J., and Rose, J. K. (1985). A single amino acid substitution in a hydrophobic domain causes temperature-sensitive cell-surface transport of a mutant viral glycoprotein. J. Virol. 64,374-382. Garoff, H., Kondor-Koch, C., Pettersson, R., and Burke, B. (1983). Expression of Semliki Forest virus proteins from cloned complementary DNA. 11. The membrane-spanning glycoprotein E2 is transported to the cell surface without its normal cytoplasmic domain. J. Cell Biol. 97, 652-658. Gething, M.J., and Sambrook, J. (1981). Cell-surface expression of influenza haemagglutinin from a cloned DNA copy of the RNA gene. Nature (London) 293, 620-625. Gething, M.J.,and Sambrook, J. (1982). Construction of influenza haemagglutinin genes that code for intracellular and secreted forms of the protein. Nature (London) 300, 598-603. Gething, M.J., Bye, J., Skehel, J. J., and Waterfield, M. D. (1980). Cloning and DNA sequence of double-stranded copies of haemagglutinin genes from H2 and H3 strains elucidates antigenic shift and drift in human influenza virus. Nature (London) 287, 301-306. Gething, MA., Doms, R. W., York, D., and White, J. (1986). Studies on the mechanism of membrane fusion: Site-specific mutagenesis of the hemagglutinin of influenza virus. J. Cell Biol. 102, 11-23. Gibson, R., Leavitt, R., Kornfeld, S., and Schlesinger, S. (1978). Synthesis and infectivity of vesicular stomatitis virus containing nonglycosylated G protein. Cell 13, 671-679. Gibson, R., Schlesinger, S.,and Kornfeld, S. (1979). The nonglycosylated glycoprotein of vesicular stomatitis virus is temperature-sensitive and undergoes intracellular aggregation at elevated temperatures. J. Biol.Chem. 254, 3600-3607. Gibson, R., Kornfeld, S., and Schlesinger, S. (1980). A role for oligosaccharides in glycoprotein biosynthesis. Trends Biochem. Sci. 5, 290-293. Gibson, R., Kornfeld, S., and Schlesinger, S. (1981). The effect of oligosaccharide chains of different sizes on the maturation and physical properties of the G protein of vesicular stomatitis virus. J. Biol. Chem. 266, 456-462. Gross, V., Andus, T., Tran-Thi, T.-A., Schwarz, R. T., Decker, K., and Heinrich, P. C.

38

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

(1983). 1-Deoxynojirimycin impairs oligosaccharide processing of 1-proteinase inhibitor and inhibits its secretion in primary cultures of rat hepatocytes. J. Biol. Chem. 258, 12203-12209. Guan, J.-L., and Rose, J. K., (1984). Conversion of a secretory protein into a transmembrane protein results in its transport to the Golgi complex but not to the cell surface. Cell 37, 779-787. Guan, J.-L., Machamer, C. E., and Rose, J. K. (1985). Glycosylation allows cell-surface transport of an anchored secretory protein. Cell 42, 489-496. Harrison, S. C. (1986). Alphavirus structure. In “The Togaviridae and Flaviviridae” (S. Schlesinger and M. J. Schlesinger, eds.), pp. 21-34. Plenum, New York. Hashimoto, K., Erdel, S., Deranen, S., Saraste, J., and Kaariainen, L. (1981). Evidence for a separate signal sequence for the carboxy-terminal protein E l of Semliki Forest virus. J. Virol. 38, 34-40. Hayward, W. S. (1977). Size and genetic content of viral RNAs in avian oncovirusinfected cells. J. Virol. 24, 47-63. Hiebert, S. W., Paterson, R. G., and Lamb, R. A. (1985). Hemagglutinin-neuraminidase protein of the paramyxovirus Simian virus 5 : Nucleotide sequence of the mRNA predicts an N-terminal membrane anchor. J. Virol. 54, 1-6. Higa, H. H., Rogers, G. N., and Paulson, J . C. (1985). Influenza virus hemagglutinins differentiate between receptor determinants bearing N-acetyl, N-glycollyl, and N,Odiacetylneuraminic acids. Virology 144, 279-282. Homa, F. L., Purifoy, D. J. M., Glorioso, J. C., and Levine, M. (1986). Molecular basis of the glycoprotein C-negative phenotypes of herpes simplex virus type 1 mutants selected with a virus-neutralizing monoclonal antibody. J. Virol. 58 281-289. Huang, R. T. C., Rott, R., Wahn, K., Klenk, H.-D. and Kohama, T. (1980). The function of neuraminidase in membrane fusion induced by myxoviruses. Virology 107, 313319. Hughes, S. (1982). Sequence of the long terminal repeat and adjacent segments of the endogenous avian virus Rous associated virus. J. Virol. 43, 191-200. Hunter, E., Hill, E., Hardwick, M., Brown, A., Schwartz, D. E., and Tizard, R. (1983). Complete sequence of the Rous sarcoma virus enu gene: Identification of structural and functional regions of its product. J. Virol. 46, 920-936. Johnson, D. C., and Schlesinger, M. J. (1980). Vesicular stomatitis virus and Sindhis virus glycoprotein transport to the cell surface is inhibited by ionophores. Virology 103, 407-424. Johnson, D. C., Schlesinger, M. J., and Elson, E. L. (1981). Fluorescence photobleaching recovery measurements reveal differences in envelopment of Sindbis and vesicular stomatitis viruses. Cell 23, 423-431. Joho, R. H., Billeter, M. A., and Weissman, C. (1975). Mapping of biological functions on RNA of avian tumor viruses: Location of regions required for transformation and determination of host range. Proc. Natl. Acad. Sci. U S A . 72, 4772-4776. Kaariainen, K., Hashimoto, K., Saraste, J., Virtanen, I., and Penttinen, K. (1980). Monensin and FCCP inhibit the intracellular transport of alphavirus membrane glycoproteins. J. Cell Biol. 87, 783-791. Kang, C. Y., and Prevec, L. (1970).Proteins of vesicular stomatitis virus. 11. Immunological comparisons of viral antigens. J. Virol. 6, 20-27. Katz, F. N., Rothman, J. E., Lingappa, V. R., Blobel, G., and Lodish, H. F. (1977). Membrane assembly in uitro: Synthesis, glycosylation, and asymmetric insertion of a transmembrane protein. Proc. Natl. Acad. Sci. U.S.A. 74, 3278-3282. Kawaoka, Y., Naeve, C., and Webster, R. G. (1984). Is virulence of H5N2 influenza viruses in chickens associated with loss of carbohydrate from the hemagglutinin? Virology 139, 303-316.

DOMAINS OF VIRUS GLYCOPROTEINS

39

Kielian, M. C., and Helenius, A. (1985).pH-induced alterations in the fusogenic spike protein of Semliki Forest virus. J. Cell Biol. 101, 2284-2291. Kielian, M. C., Keranen, S., Kaariainen, L., and Helenius, A. (1984).Membrane fusion mutants of Semliki Forest virus. J. Cell Biol. 98, 139-145. Klenk, H.-D., Rott, R., Orlich, M., and Blodom, J. (1975).Activation of influenza viruses by trypsin treatment. Virology 68, 426-439. Kondor-Koch, C., Burke, B., and Garoff, H. (1983).Expression of Semliki Forest virus proteins from clonal complementary DNA. I. The fusion activity of the spike glycoprotein. J. Cell Biol. 97, 644-651. Kornfeld, R.,and Kornfeld, S. (1985).Assembly of asparagine-linked oligosaccharides. Annu. Rev. Biochem. 54,631-664. Kotwal, G. J.,and Gosh, H. P. (1984).Role of fatty acid acylation of membrane glycoproteins-absence of palmitic acid in glycoproteins of two serotypes of vesicular stomatitis virus. J. Biol. Chem. 259, 4699-4701. Kries, T. E., and Lodish, H. F. (1986).Oligomerization is essential for transport of vesicular stomatitis viral glycoprotein to the cell surface. Cell 46, 929-937. Kyte, J., and Doolittle, R. F. (1982).A simple method for displaying the hydropathic character of a protein. J. Mol. Bwl. 157, 105-132. Lazarowitz, S. G., and Choppin, P. W. (1975).Enhancement of the infectivity of influenza A and B viruses by proteolytic cleavages of the hemagglutinin polypeptide. Virology 68, 440-454. Leamnson, R. N., and Halpern, M. S. (1976).Subunit structure of the glycoprotein complex of avian tumor virus. J. Virol. 18, 956-968. Lentz, M. R., and Air., G. M. (1986).Loss of enzyme activity in a site-directed mutant of influenza neuraminidase compared to expressed wild-type protein. Virology 148, 7483. Lentz, T. L. (1985).Rabies virus receptors. Trends Neurol. Sci. 8, 360-364. Lentz, T. L., Burrage, T. G., Smith, A. L., Crick, J., and Tignor, G. H. (1982).Is the acetylcholine receptor a rabies virus receptor? Science 215, 182-184. Lentz, T. L., Hawrot, E., and Speicher, D. W. (1984).Amino acid sequence similarity between rabies virus glycoprotein and snake venom curaremimetic neurotoxins. Science 226, 847-848. Li, J.-P., Bestwick, R. K., Machida, C., and Kabat, D. (1986).Role of a membrane glycoprotein in Friend virus erythroleukemia: Nucleotide sequences of nonleukemogenic mutant and spontaneous revertant viruses. J. Virol. 57, 534-538. Little, S. P.,and Huang, A. S. (1978).Shedding of the glycoprotein from vesicular stomatitis virus-infected cells. J. Virol. 27, 330-339. Machamer, C. E.,Florkiewicz, R. Z., and Rose, J. K. (1985).A single N-linked oligosaccharide at either of the two normal sites is sufficient for transport of vesicular stomatitis virus G protein to the cell surface. Mol. Cell. Biol. 5, 3074-3083. Machida, C. A., Bestwick, R. K., Boswell, B. A., and Kabat, D. (1985).Role of a membrane glycoprotein in Friend virus-induced erythroleukemia: Studies of mutant and revertant viruses. Virology 144, 158-172. McQueen, N. L., Nayak, D. P., Jones, L. V., and Compans, R. W. (1984).Chimeric influenza virus hemagglutinin containing either the NH2 terminus or the COOH terminus of G protein of vesicular stomatitus virus is defective in transport to the cell surface. Proc. Natl. Acad. Sci. U S A . 81,395-399. McQueen, N. L., Nayak, D. P., Stephens, E. B., and Compans, R. W. (1986).Polarized expression of a chimeric protein in which the transmembrane and cytoplasmic domains of the influenza virus hemagglutinin have been replaced by those of the vesicular stomatitis virus G protein. Proc. Natl. Acad. Sci. U S A . , in press. Magee, A. I., Koyama, A. H., Malfer, C.,Wen, D., and Schlesinger, M. J. (1983).Release

40

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

of fatty acids from virus glycoproteins by hydroxylamine. Biochim. Biophys. Actu 798, 156-166. Mann, E., Edwards, J., and Brown, D. T. (1983).Polycaryocyte formation mediated by Sindbis virus glycoproteins. J. Virol. 45, 1083-1089. Markoff, L., Lin, B.-C., Sveda, M. M., and Lai, C-J. (1984).Glycosylation and surface expression of the influenza virus neuraminidase requires the N-terminal hydrophobic region. Mol. Cell. Biol. 4, 8-16. Marshall, R. D. (1972).Glycoproteins. Annu. Reu. Biochem. 41, 673-702. Marshall, R. D.(1974).The nature and metabolism of the carbohydrate-peptide linkages of glycoproteins. Biochem. Soc. Symp. 40, 17-26. Melancon, P., and Garoff, H. (1986).Reinitiation of translocation in the Semliki Forest virus structural polyprotein: Identification of the signal for E l glycoprotein. EMBO J . 5, 155-156. Mellon, P., and Duesberg, P. H. (1977).Subgenomic, cellular Rous sarcoma virus RNAs contain oligonucleotides from the 3’half and the 5’terminus of virion RNA. Nature (London) 270,631-634. Mooney, J. J., Dalrymple, J. M., Alving, C. R., and Russell, P. K. (1975).Interaction of Sindbis virus with liposomal model membranes. J. Virol. 15,225-231. Morrison, T., Ward, L. J., and Semerjian, A. (1985).Intracellular processing of the Newcastle disease virus fusion glycoprotein. J. Virol. 53,851-857. Mudd, J. A., and Swanson, R. E. (1978).In situ cross-linking of vesicular stomatitus virus proteins with reversible agents. Virology 88, 263-280. Nesterowicz, A., Laver, G., and Jackson, D. C. (1985).Antigenic determinants of influenza virus hemagglutinin. x.A comparison of the physical and antigenic properties of monomeric and trimeric forms. J. Gen. Virol. 68, 1687-1695. Olmsted, R. A., Baric, R. S., Sawyer, B. A., and Johnston, R. E. (1984).Sindbis virus mutants selected for rapid growth in cell culture display attenuated virulence in animals. Science 225, 424-426. Palese, P., and Kingsbury, D. W., eds. (1983).“Genetics of Influenza Viruses.” SpringerVerlag, Berlin and New York. Pan, Y. T., Hori, H., Saul, T., Sanford, B. A., Molyneux, R. J., and Elbein, A. D. (1983). Castanospermine inhibits the processing of the oligosaccharide portion of the influenza viral hemagglutinin. Biochemistry 22, 3975-3984. Para, M. F., Goldstein, L., and Spear, P. G. (1982).Similarities and differences in the Fcbinding glycoprotein (gE) of herpes simplex virus types 1 and 2 and tentative mapping of the viral gene for this glycoprotein. J. Virol. 41, 137-144. Peyrieras, N., Bause, E., Legler, G., Vasilov, R., Claesson, L., Peterson, P., and Ploegh, H. (1983).Effects of the glucosidase inhibitors nojirimycin and deoxynojirimycin on the biosynthesis of membrane and secreted proteins. EMBO J. 2,823-832. Pinter, A., and Honnen, W. J. (1985).The mature form of the Friend spleen focusforming virus envelope protein, gp65,is efficiently secreted from cells. Virology 143, 646-650. Poruchynsky, M. S., Tyndall, C., Both, G. W., Sato, F., Bellamy, A. R., and Atkinson, P. H. (1985).Deletions into a n NH2-terminal hydrophobic domain result in secretion of rotavirus vp7, a resident endoplasmic reticulum membrane glycoprotein. J.Cell Biol. 101, 2199-2209. Reading, C. L., Penhoet, E. E., and Ballou, C. E. (1978).Carbohydrate structure of vesicular stomatitis virus glycoprotein. J. Biol. Chem. 253,5600-5612. Rice, C. M., and Strauss, J. H. (1981).Nucleotide sequence of the 26s mRNA of Sindbis virus and deduced sequence of the encoded virus structural proteins. Proc. Nutl. Acud. Sci. U S A . 78.2062-2066.

DOMAINS OF VIRUS GLYCOPROTEINS

41

Rice, C. M., and Strauss, J. H. (1982).Association of Sindbis virion glycoproteins and their precursors. J. Mol. Biol. 164, 325-348. Riedel, H. (1985).Different membrane anchors allow the Semliki Forest virus spike subunit E2 to reach the cell surface. J. Virol. 54, 224-228. Riedel, H., Konder-Koch, C., and Garoff, H. (1984).Cell surface expression of fusogenic vesicular stomatitis virus G protein from cloned DNA. EMBO J. 3,1477-1483. Robertson, J. S.,Naeve, C. W., Webster, R. G., Bootman, J. S., Newman, R., and Schild, G. C. (1985).Alterations in the hemagglutinin associated with adaptation of influenza B virus to growth in eggs. Virology 143, 166-174. Rodriguez-Boulan, E. J., and Sabatini, D. D. (1978).Asymmetric budding of viruses in epithelial monolayers: A model system for the study of epithelial cell polarity. Proc. Natl. Acad. Sci. U S A . 75, 5071-5075. Roehrig, J. T. (1986).The use of monoclonal antibodies in studies of the structural proteins of togaviruses and flaviviruses. In “The Togaviridae and Flaviviridae” (S. Schlesinger and M. J. Schlesinger, eds.), pp. 251-278. Plenum, New York. Roehrig, J. T., Gorski, D., and Schlesinger, M. J. (1982).Properties of monoclonal antibodies directed against the glycoproteins of Sindbis virus. J. Gen. Virol. 59,421-425. Rose, J. K.,and Bergmann, J. E. (1982).Expression from cloned cDNA of cell-surface secreted forms of the glycoprotein of vesicular stomatitis virus in eucaryotic cells. Cell 30, 753-762. Rose, J. K., and Bergmann, J. E. (1983).Altered cytoplasmic domains affect intracellular transport of the vesicular stomatitis virus glycoprotein. Cell 34, 513-524. Rose, J.K.,and Gallione, C. J. (1981).Nucleotide sequences of the mRNA’s encoding the vesicular stomatitis virus G and M proteins determined from the cDNA clones containing the complete coding regions. J. Virol. 39,519-528. Rose, J. K., Welch, W. J., Sefton, B, M., Esch, F. S., and Lund, N. C. (1980).Vesicular stomatitis virus glycoprotein is anchored in the viral membrane by a hydrophobic domain near the COOH terminus. Proc. Natl. Acad. Sci. U S A . 77,3884-3888. Rose, J. K., Doolittle, R. F., Anilionis, A., Curtis, P. J., and Wunner, W. H. (1982). Homology between the glycoproteins of vesicular stomatitis virus and rabies virus. J. Virol. 43, 361-364. Rose, J. K.,Adams, G. A., and Gallione, C. J. (1984).The presence of cysteine in the cytoplasmic domain of the vesicular stomatitis virus glycoprotein is required for palmitate addition. Proc. Natl. Acad. Sci. U S A . 81,2050-2054. Roth, M. G.,Srinivas, R. V., and Compans, R. W. (1983). Basolateral maturation of retroviruses in polarized epithelial cells. J . Virol. 45, 1065-1073. Rott, R., Orlich, M., Klenk, H.-D., Wang, M. L., Skehel, J. J., and Wiley, D. C. (1984). Studies on the adaptation of influenza viruses to MDCK cells. EMBO J. 3, 33293332. Rottier, P.J. M., Welling, G. W., Welling-Wester, S., Niesters, H. G. M., Lenstra, J. A., and Van der Zdeijst, B. A. M. (1986).Predicted membrane topology of the coronavirus protein El.Biochemistry 25, 1335-1339. Ruscetti, S.,and Wolff, L. (1984).Spleen focus-forming virus: Relationship of an altered envelope gene to the development of a rapid erythroleukemia. Cum. Top. Mlcrobiol. Immunol. 112, 21-44. Sambrook, J.,Rogers, L. White, J., and Gething, M J . (1985).Lines of BPV-transformed murine cells that constitutively express influenza virus hemagglutinin. EMBO J. 4, 91-103. Saul, R., Chambers, J. P., Molyneux, R. J.,and Elbein, A. D. (1983).Castanospermine, a tetrahydroxylated alkaloid that inhibits p-glucosidase and p-glucocerebmsidase. Arch. Biochem. Biophys. 221,593-597.

42

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

Saunier, B., Kilker, R. D., Tkacz, J . S., Quaroni, A., and Herscovics, A. (1982). Inhibition of N-linked complex oligosaccharide formation by 1-deoxynojirimycin, an inhibitor of processing glucosidases. J. Biol. Chem. 257, 14155-14161. Schlegel, R., and Wade, M. (1984). A synthetic peptide corresponding to the NH2 terminus of vesicular stomatitis virus glycoprotein is a pH-dependent hemolysin. J . Biol. Chem. 259,4691-4694. Schlegel, R., and Wade, M. (1985). Biologically active peptides of the vesicular stomatitis virus glycoprotein. J. Virol. 53, 319-323. Schlegel, R., Tralka, T. S., Willingham, M. C., and Pastan, I. (1983). Inhibition of VSV binding and infectivity by phosphatidylserine: Is phosphatidylserine a VSV-binding site? Cell 32, 639-646. Schlesinger, M. J . (1985). Fatty acylation of eucaryotic cell and virus membrane proteins. In “Microbiology-1985’’ (L. Leive, ed.), pp. 316-321. American Society for Microbiology, Washington, D.C. Schlesinger, M. J., and Malfer, C. (1982). Cerulenin blocks fatty acid acylation of glycoproteins and inhibits vesicular stomatitis and Sindbis virus particle formation. J. Biol. Chem. 257,9887-9890. Schlesinger, M. J., and Schlesinger, S. (1986). Formation and assembly of alphavirus glycoproteins. In “The Togaviridae and Flaviviridae” (S. Schlesinger and M. J. Schlesinger, eds.), pp. 121-148. Plenum, New York Schlesinger, S., Malfer, C., and Schlesinger, M. J . (1984). The formation of vesicular stomatitis virus (San Juan strain) becomes temperature-sensitive when glucose residues are retained on the oligosaccharides of the glycoprotein. J. Biol. Chem. 259, 7597-7601. Schmaljohn, A. L., Kokubun, K. M., and Cole, G. A. (1983). Protective monoclonal antibodies define maturational and pH-dependent antigenic changes in Sindbis virus E l glycoprotein. Virology 130, 144-154. Schmidt, M.F. G. (1982). Acylation of viral spike glycoproteins: A feature of envelope RNA viruses. Virology 116, 327-338. Schmidt, M.F. G., and Lambrecht, B. (1985). On the structure of the acyl linkage and the function of fatty acyl chains in the influenza virus haemagglutinin and the glycoproteins of Semliki Forest virus. J . Gen. Virol. 66, 2635-2647. Schmidt, M. F. G., and Schlesinger, M. J. (1980). Relation of fatty acid attachment to the translation and maturation of vesicular stomatitis and Sindbis virus membrane glycoproteins. J. Biol. Chem. 255, 3334-3339. Schwartz, D. E., Tizard, R., and Gilbert, W. (1983). Nucleotide sequence of Rous sarcoma virus. Cell 32, 853-869. Simons, K., and Fuller, S. D. (1986). Cell surface polarity in epithelia. Annu. Reu. Cell Biol. 1, 243-288. Smith, M. M., Schlesinger, S., Lindstrom, J., and Merle, J. P. (1986). The effects of inhibiting oligosaccharide trimming by 1-deoxynojirimycin on the nicotinic acetylcholine receptor. J . Biol. Chem. 261, 14825-14832. Spear, P. G. (1985). Glycoproteins specified by herpes simplex viruses. In “The Herpes Viruses” (B. Roizman, ed.), Vol. 3, pp. 315-356. Plenum, New York. Spear, P. G . (1986). Virus-induced cell fusion. In “Cell Fusion” (A. E. Sowers, ed.). Plenum, New York, in press. Srinivas, R. V., and Compans, R. W. (1983). Membrane association and defective transport of spleen focus-forming virus glycoprotein. J. Biol. Chem. 258, 14718-14724. Srinivas, R. V., Balachandran, N., Alonso-Caplen, F. V., and Compans, R. W. (1986). Expression of herpes simplex virus glycoproteins in polarized epithelial cells. J. Virol. 58,689-693.

DOMAINS OF VIRUS GLYCOPROTEINS

43

Stanley, J., Cooper, S. J., and Griffin, D. E. (1985). Alphavirus neurovirulence: Monoclonal antibodies discriminating wild-type from neuroadapted Sindbis virus. J. Virol. 66,110-119. Steiner, D. F., Docherty, K., and Carroll, R. (1984). Golgi/granule processing of peptide hormone and neuropeptide precursors: A minireview. J. Cell. Biochem. 24, 121-130. Stephens, E. B., and Compans, R. W. (1986). Nonpolarized expression of a secreted murine leukemia virus glycoprotein in polarized epithelial cells. Cell 47, in press. Strauss, E. C., and Strauss, J. H. (1986). Structure and replication of the alphavirus genome. In “The Togaviridae and Flaviviridae” (S.Schlesinger and M. J. Schlesinger, eds.), pp. 35-90. Plenum, New York. Strauss, J. H., and Strauss, E. G. (1985). Antigenic structure of togaviruses. In “Immunochemistry of Viruses” (M. H. V. Van Regenmortel and A. R. Neurath, eds.), Chap. 22, pp. 407-424. Elsevier, Amsterdam. Struck, D. K., and Lennan, W. J. (1980).The function of saccharide-lipids in synthesis of glycoproteins. In “The Biochemistry of Glycoproteins and Proteoglycans” (W. J. Lennarz, ed.), pp. 35-83. Plenum, New York. Sturman, L. S., and Holmes, K. V. (1983). The molecular biology of coronaviruses. Adu. Virus Res. 28, 36-112. Sveda, M. M., Markoff, L. J., and Lai, C J . (1982). Cell surface expression of the influenza virus hemagglutinin requires the hydrophobic carboxy-terminal sequences. Cel2 30,649-656. Sveda, M. M., Markoff, L. J., and Lai, C-J. (1984). Influenza virus hemagglutinin containing an altered hydrophobic carboxy firminus accumulates intracellularly. J. V i rol. 49, 223-228. Symington, J., and Schlesinger, M. J. (1975). Isolation of a Sindbis virus variant by passage on mouse plasmacytoma cells. J. Virol. 15, 1037-1041. Symington, J.,and Schlesinger, M. J. (1978).Characterization of a Sindbis virus variant with altered host range. Arch. Virol. 58, 127-136. Teich, N. (1982). Taxonomy of retroviruses. In “RNA Tumor Viruses” (R.Weiss, N. Teich, H. Varmus, and J. Coffin, eds.), pp. 25-207. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Torrisi, M. R., and Bonatti, S. (1985). Immunocytochemical study of the partition and distribution of Sindbis virus glycoproteins in freeze-fractured membranes of infected baby hamster kidney cell. J . Cell Biol. 101, 1300-1306. Troxler, D. H., Lowy, D., Howk, R., Young, H., and Scolnick, E. M. (1977a).Friend strain of spleen focus-forming virus is a recombinant between ecotropic murine type C virus and the enu gene region of xenotropic type C virus. Proc. Natl. Acad. Sci. U S A . 14, 4671-4675. Troxler, D. H., Boyars, J. K., Parks, W. P., and Scolnick, E. M. (1977b). Friend strain of spleen focus-forming virus: A recombinant between mouse type C ecotropic viral sequences and sequences related to xenotropic virus. J. Virol. 22, 361-372. Van Meer, G., Davoust, J., and Simons, K. (1985). Parameters affecting low-pH-mediated fusion of liposomes with the plasma membrane of cells infected with influenza virus. Biochemistry 24,3593-3602. Varghese, J. N., Laver, W. G., and Colman, P. M. (1983). Structure of the influenza virus glycoprotein antigen neuraminidase at 2.9 A resolution. Nature (London) 303, 3540. Weiss, R. (1982). Experimental biology and assay of RNA tumor viruses. In “RNA Tumor Viruses” (R. Weiss, N. Teich, H. Varmus, and J. Coffin, eds.), pp. 209-260. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Weiss, S. R., Varmus, H. E., and Bishop, J. M. (1977). The size and genetic composition of

44

MILTON J. SCHLESINGER AND SONDRA SCHLESINGER

virus-specific RNAs in the cytoplasm of cells producing avian sarcoma-leukosis viruses. Cell 12, 983-992. Welch, W. J., and Sefton, B. M. (1979). Two small virus-specific polypeptides are produced during infection with Sindbis virus. J . Virol. 38, 1186-1195. Wertz, G. W., Collins, P. L., Huang, Y., Gruber, C., Levine, S., and Ball, L. A. (1985). Nucleotide sequence of the G protein gene of human respiratory syncytial virus reveals a n unusual type of viral membrane protein. Proc. Natl. Acad. Sci. U S A . 82, 4075-4079. White, J., and Helenius, A. (1980). pH-dependent fusion between the Semliki Forest virus membrance and liposomes. Proc. Natl. Acad. Sci. U S A . 77,3273-3277. White, J., Helenius, A., and Gething, M J . (1982). Haemagglutinin of influenza virus expressed from a cloned gene promotes membrane fusion. Nature (London) 300,658659. Wills, J. W., Srinivas, R. V., and Hunter, E. (1984). Mutations of the Rous sarcoma virus enu gene that affect the transport and subcellular location of the glycoprotein products. J . Cell Biol. 99, 2011-2023. Wilson, I. A., Skehel, J. J., and Wiley, D. C. (1981). Structure of the haemagglutinin membrane glycoprotein of influenza virus a t 3 A resolution. Nature (London) 289, 366-373. Wolff, L., Scolnick, E., and Ruscetti, S. (1983). Envelope gene of the Friend spleen focusforming virus: Deletion and insertions in 3’ gp70/pl5E-encoding region have resulted in unique features in the primary structure of its protein product. Proc. Natl. Acad. Sci. U S A . 80,4718-4722. Woodgett, C., and Rose, J. K. (1986). Amino-terminal mutation of the vesicular stomatitis virus glycoprotein does not affect its fusion activity. J . Virol. 59, 486-489. Wunner, W. H., Dietzschold, B., and Wiktor, T. J. (1985). Antigenic structure of rhabdoviruses. In “Immunochemistry of Viruses. The Basis for Serodiagnosis and Vaccines” (M.H. V. van Regenmortel and A. R. Neurath, eds.), pp. 367-388. Elsevier, Amsterdam. Yamamoto, K., Suzuki, K., and Simizu, B. (1981). Hemolytic activity of the envelope glycoproteins of western equine encephalitis virus in reconstitution experiments. Virology 109, 452-454. Ziemiecki, A., and Garoff, H. (1978). Subunit composition of the membrane glycoprotein complex of Semliki Forest virus. J . Mol. Biol. 122, 259-269.

ADVANCES IN VIRUS RESEARCH, VOL. 33

FLAVlVlRUS REPLICATION STRATEGY E.

G.Westaway

Department of Microbiology Monash University Claflon, Melbourne 3168, Australia

I. INTRODUCTION Flaviviruses have been studied experimentally since early this century, when the disease agent of yellow fever virus was shown to be filterable and transmitted by hematophagous mosquitoes (Theiler and Downs, 1973).Despite the relatively early success in development of a safe attenuated virus vaccine for yellow fever about 50 years ago, and extensive epidemiological studies, flaviviruses remained until quite recently the most poorly characterized of the RNA viruses infecting man. They are the smallest of the enveloped RNA viruses (diameter of the spherical virions, 45 nm) but comprise one of the largest groups (65 species) including many species pathogenic for man and domestic and wild animals (Westaway et al., 1985). Man is usually an incidental host in the arthropod-vertebrate cycle which maintains the virus in nature, and the virus replicates in both hosts. Many species besides yellow fever virus cause severe, often fatal, infections in man with distinct tissue trophisms, e.g., dengue, dengue hemorrhagic fever, and Japanese, St. Louis, Murray Valley, and tick-borne encephalitis. Until recently, the Flauzuirus genus was classified in the family Togaviridae, but accumulating evidence indicated that flaviviruses are distinct from the well-characterized togaviruses in regard to size, structure, gene sequence, and replication strategy. Accordingly, the genus is now classified in the Flaviviridae, a new family, with type species yellow fever (YF)virus, approved by the International Committee on Taxonomy of Viruses (Brown, 1986;Westaway et al., 1985). Flaviviruses are readily distinguished from other positive-strand RNA viruses of vertebrates (see Strauss and Strauss, 1983). Advances in the molecular biology of the family progressed slowly, and involved a surprising number of species, e.g., dengue 2 (DEN-2), Kunjin (KUN), Japanese encephalitis (JE), St. Louis encephalitis (SLE), West Nile (WN), Murray Valley encephalitis (MVE), and tickborne encephalitis (TBE) viruses (Westaway, 1980).This diversity re45

Copyright 8 1987 by Academic Reen, Inc. All rights of reproduction in any form reserved.

46

E. G . WESTAWAY

flects the inherent difficulty in selecting an appropriate strain for molecular studies as well as parochial or medical interest. The impediments are relatively low yields, long latent periods, lack of switch-off of cell macromolecular synthesis, and fragility of viruses. The publication by Rice et al. (1985) of the complete nucleotide sequence of the RNA of the 17D vaccine strain of YF virus is a n outstanding achievement; it has opened new vistas and changed or modified existing concepts in the molecular biology of flaviviruses. This has been followed closely or concurrently by extensive nucleotide sequence data on regions of RNA representing several genes for structural or nonstructural proteins of WN (Castle et al., 1985, 1986; Wengler et al., 19851, MVE (Dalgarno et al., 19861, and DEN-2 (from Padmanabhan and colleagues, cited by Rice et al., 1986). Two reviews relevant to flavivirus replication have also recently appeared (Brinton, 1986; Rice et al., 1986).

11. STRUCTURE OF THE VIRION A . Morphology and Composition In negatively stained preparations (see inset in Fig. 61, flaviviruses appear as uniformly spherical particles, 45 nm in diameter, with a lucent envelope enclosing a densely staining core, 30 nm in diameter. Spikes or clublike projections form a fringe about 5 nm thick around the envelope. The elements of icosahedral symmetry have not yet been demonstrated in the virus. Murphy (1980) commented that the nature of the interior structure of flaviviruses is “one of the major unanswered questions” for arthropod-transmitted viruses. Doughnutshaped fragments 7 nm in diameter were produced during storage of DEN-2 virions at 4”C, and these were similar t o 7-nm particles visible on the surface of the intact virion (Smith et al., 1970). However, the true native state of the virion surface remains uncertain (Murphy, 1980). Virions are physically fragile and great care must be taken during purification to avoid loss or degradation of particles, likely to be caused by association of particles with cell debris, by changes in pH or tonicity, by freeze-thaw cycles, or by storage for more than a day at 4°C. For biochemical purposes, the best preparations are obtained from infected cell culture fluids containing some protein (e.g., 0.1% bovine serum albumin); the virions may be rapidly concentrated therefrom and partially purified by precipitation with 6-7% polyethylene glycol

FLAVIVIRUS REPLICATION

47

6000 (Westaway and Reedman, 1969; Della-Porta and Westaway, 1972). The virion consists of three proteins, E and M, associated with the envelope, and C, associated with the single piece of single-stranded RNA. The chemical composition of SLE virus is 6%RNA, 66%protein, 9%carbohydrate, and 17%lipid (Trent and Naeve, 1980). Most of E is located externally and may be digested with proteases (Heinz and Kunz, 1979). E, or E and M, may be removed to leave a core particle after treatment of virions with appropriate concentrations of selected detergents; the association of E and M in aggregates after such removal from TBE virus was assumed to be caused by hydrophobic interaction of M with the hydrophobic tail of E (Heinz and Kunz, 1980). The recently published amino acid sequences of M and E for YF, MVE, and WN viruses show that the regions near their carboxy termini are indeed hydrophobic (Rice et al., 1985; Wengler et al., 1985; Dalgarno et al., 1986).

B . RNA RNA extracted from purified virions is single stranded and infectious, and the sedimentation coefficient is 44 S relative to 49 S RNA of the alphavirus Sindbis (Westaway, 1980). The complete nucleotide sequence of 10,862 nucleotides for the RNA of yellow fever virus has been obtained by sequencing cloned cDNA and is discussed in detail by Rice and colleagues (Rice et al., 1985, 1986). The base composition of YF virus RNA is 27.3%A, 23.0%U, 28.4%G, and 21.3%C, and the M, calculated as the Na+ form is 3.75 x lo6 (Rice et al., 19851, confirming the value obtained by electrophoresis under denaturing conditions of 3.8 x lo6 (Deubel et al., 1983). Sequences of large portions of the RNA have been characterized via cloned cDNA also for WN virus (Castle et al., 1985; Wengler et al., 1989, DEN-2 virus (Rice et al., 1986), and MVE virus (Dalgarno et al., 1986).The RNA has a type 1 cap structure, m7GpppAmpNp,at the 5’ end, and no detectable poly(A) tract at the 3’ end of several species (Wengler et al., 1978; Cleaves and Dubin, 1979; Vezza et al., 1980; Trent and Naeve, 1980; Rice et al., 1985).The terminal sequences of WN and YF RNA are identical: 5’ capAGUA. . .CU,, 3’ (Rice et al., 1985; Wengler and Wengler, 1981). Stable secondary structures at the 3’ terminus of YF RNA appear to ensure that the last four or five nucleotides are hydrogen bonded (Rice et al., 1985; Grange et al., 1985) and hence protect the 3’ terminus from enzymatic modification as noted for WN and YF RNA (Wengler and Wengler, 1981; Deubel et al., 1983).

48

E. G . WESTAWAY

Oligonucleotide maps of TI ribonuclease digests readily distinguished the RNAs of different species (Wengler et al., 1978; Vezza et al., 19801, or of strains of DEN-2 virus (Walker et al., 1982; Trent et al., 1983),DEN-1 virus (Repik et al., 19831, and SLE virus (Trent et al., 1981). Nevertheless, the flavivirus genome has remarkable strain stability; oligonucleotide changes probably representing less than 1% change in RNA sequence occurred during over 200 passages of the 17D strain of YF virus from the original Asibi strain (Monath et al., 1983). The extent of homology among the four dengue species was shown by Blok and colleagues using cDNA-RNA hybridization followed by S, nuclease digestions; DEN-1, -3, and -4 shared 53 to 73% of their genomes, but DEN-2 shared only 23-36% homology (Blok et al., 1984; Blok, 1985). Surprisingly, the latter RNA was 71% homologous with Edge Hill virus, an Australian virus which tends to be serologically distinct from the dengue complex (Westaway, 1966). The 65 flavivirus species were designated almost entirely on neutralization tests involving antibody reactions with only one gene product, i.e., protein E (Westaway et al., 1985).Although comparisons of flavivirus RNAs have been made with only a limited number of species, the extent of variation in homology indicates that each species is truly distinct. Nevertheless, the variation is obviously subject to evolutionary constraints throughout the genus; the hydrophobicity plots of the amino acid sequences translated from the available nucleotide sequences of YF, MVE, and DEN-2 RNA show striking similarity (Rice et al., 1986).

C . Proteins The three structural proteins E, C, and M were first identified in 1969 for DEN-2 virus (Stollar, 1969) and for KUN virus (Westaway and Reedman, 1969). Many other flaviviruses have been shown to have similar proteins (Westaway, 1980); M, ranges are 51,000-60,000 for E, 13,000-15,000 for C, and about 8000 for M. The nomenclature of E, C, and M replaced an earlier system of V3, V2, and V1, respectively (Westaway et al., 1980). In virions extracted from infected cells, protein M is replaced by a glycosylated larger protein (M,about 20,000) which migrates similarly to a nonstructural (ns) protein during electrophoresis; the original designation NV2 for J E virus did not distinguish between these comigrating proteins from cytoplasm (Shapiro et al., 1971,1973a).The glycoprotein NV2 is sometimes found in extracellular virions (Fig. 1).Shapiro et al. (1972a) proposed that NV2 is converted to M by cleavage during virus maturation. This relationship is discussed in Section 111.

FLAVIVIRUS REPLICATION

49

The proteins E and M, or sometimes E, M, and NV2, are found also in a “slow sedimenting hemagglutinin” (SHA) which is released from infected vertebrate cells (Stollar, 1969; Shapiro et al., 1971; Westaway, 1975; Wright et al., 1977). The SHA particle for DEN-2 virus has the appearance of a doughnut 14 nm in diameter (Smith et al., 1970). 1 . Protein E E represents the spikes of the virion; hemagglutination by virions is blocked and virus infectivity is neutralized by specific polyclonal antibody to E (Della-Porta and Westaway, 1977; Heinz et al., 1984) and by monoclonal antibodies to some epitopes of E (Peiris et al., 1982; Roehrig et al., 1983). In conformity with the structure of other enveloped viruses, E of most but not all flaviviruses analyzed is glycosylated, the notable exceptions being KUN virus (Wright, 1982) and a strain of WN virus (Wengler et al., 1985). Tunicamycin is able to block glycosylation of E of DEN-2 virus, indicating that the oligosaccharides are attached to the polypeptide backbone in N-linkage (Smith and Wright, 1985). The amino acid sequence of close to 500 residues terminates with a long hydrophobic sequence, containing centrally one or more basic amino acids which probably anchors E in the lipid envelope (Rice et al., 1985; Wengler et al., 1985). As expected from the RNA homologies, there are similarities in amino acid sequences of E from different species, which reflect the extent of serological relatedness. The overall homology of sequence for E of MVE and WN viruses from the same antigenic complex is 78%, but only 44%for MVE and YF viruses (from the data of Wengler et al., 1985; Dalgarno et al., 1986). Chromatography of a-chymotryptic peptides of E of MVE, WN, and KUN viruses also showed significant homology for the three members of the subgroup; when expressed as an overlap index after analyses involving 45 peptides, the homology for MVE-WN was 0.60, and 0.80 for KUN-WN (Wright et al., 1983). KUN virus is serologically related more closely to WN than to MVE virus (Westaway, 1966). The epitopes of E of severaI flaviviruses have been analyzed using monoclonal antibodies (Gentry et al., 1982; Peiris et al., 1982; Heinz et al., 1983; Kimura-Kuroda and Yasui, 1983; Roehrig et al., 1983; Schlesinger et al., 1984). Thus antibodies to E exhibit patterns of (1) type or subtype/strain specificity, (2) subgroup or complex/subcomplex specificity, and (3) group specificity. Topographical analyses of epitopes in E by competition binding assays with the monoclonal antibodies alluded to above defined a continuum of overlapping domains, nearly all in one or two clusters. Identification of the coding regions for these epitopes within the nucleotide sequence for E protein should

50

E. G . WESTAWAY

facilitate the design of genetically engineered flavivirus vaccines, or of appropriate synthetic peptides as vaccines.

2. Protein C The core particle containing C and the genomic RNA is equivalent to the capsid in which the RNA is sensitive to RNase (Stollar, 1969; Boulton and Westaway, 1972). The basic amino acid content of C is very high, approximately 20% of the 123-125 residues (Rice et al., 1985; Castle et al., 1985; Dalgarno et al., 1986). Rice et al. (1985) suggested that the function of these positive charges is to partially neutralize the negative charges of the RNA. The homology of C for YF and MVE viruses is relatively low, about 27%, but rises to 63% for C of MVE and WN viruses (from the data of Rice et al., 1985; Castle et al., 1985; Dalgarno et al., 1986). Thus amino acid residues 42-62 and 7591 in C of the latter pair differ by one or no residues. The conservation within the serological subgroup is probably uniform, e.g., the first seven or eight residues in C of MVE, SLE, and WN viruses are identical. 3. Protein M Of the 75 amino acids in M, nearly half comprise the hydrophobic carboxy tail containing a single arginine residue and function as a membrane spanning domain. The homology in amino acid sequence is 83%for M of MVE and WN viruses, compared to only 36% for MVE and YF viruses (from the data of Rice et al., 1985; Castle et al., 1985; Dalgarno et al., 1986). The predominant feature is the structural similarity of the carboxy end, probably dictated by the role of M during virus assembly (Rice et al., 1986).

111. REPLICATION A. Preliminary Overview Flaviviruses replicate in a wide range of vertebrate and invertebrate hosts. The vector species of mosquitoes or ticks tend to be restricted for individual viruses, and for some species no arthropod vector has been detected (Brown, 1986). The incubation period in vertebrates is several days, and involves a relatively brief viremia. Tissue trophism and pathogenesis vary widely, from silent infections, fever with malaise, through exanthems to jaundice, hemorrhage, and encephalitis (Theiler and Downs, 1973). Susceptibility to infection varies with the host and with individual flaviviruses. For example, cells from

FLAVIVIRUS REPLICATION

51

the C3H/He and C3H/RV congenic strains of mice were susceptible or genetically resistant, respectively, to WN virus infection (Darnel1 and Koprowski, 1974), but cells of both strains were equally sensitive to infection by Banzi virus (Bhatt et al., 1981). Despite the diversity of hosts and the large number of virus species, the growth curves are remarkably constant. The latent period of about 12-16 hours is invariably much slower than that of RNA viruses of similar complexity, and yields are often low (Westaway, 1980). Replication appears to be confined to the cytoplasm, although the location of immunofluorescent foci in infected cells and biochemical studies indicate that the perinuclear region plays an important role (Westaway, 1980; Trent and Naeve, 1980). Thin sections of infected cells show gross changes in ultrastructure, especially a proliferation of smooth and rough endoplasmic reticulum, most evident after the latent period (Westaway, 1980; Murphy, 1980). Virus particles accumulate for many hours in the medium of infected cells, reaching a maximum at about 24 hours (Schlesinger, 1977). The yield of infectious virus particles from cells disrupted by physical or mechanical means is disappointingly low, in view of the accumulation of apparently mature particles within cells (see discussion by Schlesinger, 1977). These particles probably represent immature virions containing the proteins E, C, and NV2, as noted earlier. It is possible that the infectious titer of intracellular virus could be increased dramatically in the presence of small amounts of enhancing antibody by assay in cells bearing Fc receptors (Gollins and Porterfield, 1984).

B . Early Events Because some flavivirus species infect mosquito, avian, and mammalian cells, the cell receptor for the single envelope protein must be relatively ubiquitous. However, the host range of some species appears to be relatively restricted, and the tissue trophisms within the one host are also indicative of selectivity. Initiation of infection had not been characterized in any detail (Schlesinger, 1977) until the recent report of Gollins and Porterfield (1985) which convincingly established the mode of entry of WN virus into the macrophage-like cell line P388D1. Single virus particles were taken up in coated pits within 30 seconds at 37"C,and appeared within 2 minutes in fully or partially coated vesicles. After 3-5 minutes, viral particles appeared in large, uncoated prelysosomal endocytic vacuoles more than 700 nm in diameter. After 10 minutes, degradation commenced in the lumen of lysosomal vacuoles, which by 30-45 minutes contained granular and membranous material. Aggregates of viruses were taken up more slowly by phago-

52

E. G. WESTAWAY

cytosis. In the presence of enhancing antibody, uptake of single virions also occurred within coated pits but more efficiently. No uptake occurred a t 0°C. After uncoating, the viral RNA is presumably released from lysosomes and translation commences, probably in the rough endoplasmic reticulum (Boulton and Westaway, 1976; Westaway and Ng, 1980). Envelope protein of KUN virus has been detected in cells by immunofluorescence at 4 hours postinfection (Ng et al., 1983). An early phase of viral RNA synthesis begins at about 6 hours (Trent and Naeve, 1980; Westaway, 1980) but the initial RNA products have not been characterized. Viral RNA synthesis and virus yield are enhanced by short pulses (2 or 3 hours) of actinomycin D during the latent period, but continuous actinomycin D treatment decreases both, presumably due to adverse effects on the cell (Leary and Blair, 1983; Chu and Westaway, 1986).RNA polymerase activity during the latent period was detected in cytoplasm both in uiuo (JE virus; Leary and Blair, 1983) and in uitro (KUN virus; Chu and Westaway, 1985, 1986); the nucleus was not involved (JE virus) but late in infection the nuclear extract was more active than the cytoplasm (KUN virus). A major problem in detection of the events of replication during the long latent period is the very low production of virus-specified RNA and protein against the high host background of continuing macromolecular synthesis. The ultrastructural changes are also not prominent early in infection.

C . Viral Messenger RNA The virion RNA and genomic-length RNA associated with polysomes in infected cells appear identical in size, type 1 cap, and terminal sequences (Wengler et al., 1978; Wengler and Wengler, 1981; Cleaves and Dubin, 1979). RNA from both sources is infectious and may be translated to give similar products in cell-free translation systems (Wengler et al., 1979; Svitkin et al., 1981; Monckton and Westaway, 1982). Gel electrophoresis of flavivirus RNA radioactively labeled in infected cells in the presence of actinomycin D revealed no subgenomic mRNA comparable to that of the alphaviruses or coronaviruses (Boulton and Westaway, 1977; Naeve and Trent, 1978). Single-stranded RNA of low molecular weight (less than lo5) from WN virus-infected cells was of positive polarity but had no mRNA activity in cell-free translation systems (Wengler et al., 1978). The initiating methionine codon is at nucleotide 119 in Y F virus RNA and an open reading frame continues until an opal termination

FLAVIVIRUS REPLICATION

53

codon (UGA) ending at nucleotide 10,356; this could encode a polyprotein of 3411 amino acids, equal to 380,763 Da (Rice et al., 1985). For MVE virus RNA, the 5’ untranslated region is 96-97 nucleotides (Dalgarno et al., 1986). The total length of the YF RNA is 10,862 nucleotides; flavivirus RNA is thus the largest mRNA which is completely translated in eukaryotic cells (see Section 111,G). The lack of any poly(A) sequence at the 3’ end also sets the flavivirus RNA apart from most other eukaryotic messengers.

D . Identification of Virus-Specified Proteins In the normal absence of cell switch-off during infection, labeled cell proteins often obscure viral proteins or cause ambiguities in analysis by gel electrophoresis (see Figs. 2 and 3).This problem has been partly overcome by radioimmunoprecipitation (Wright et al., 1981; Schlesinger et al., 1983; Smith and Wright, 1985) and by difference analysis (Westaway, 1973; Westaway et al., 1984; Trent and Naeve, 1980). Since the original report of nonstructural (ns) proteins in flavivirusinfected cells (Westaway and Reedman, 19691, improvements in electrophoretic resolution of radioactively labeled proteins have revealed increasing complexity. In addition to intracellular equivalents of E and C, at least seven ns proteins are found (Shapiro et al., 1971; Westaway, 1973; Wengler et al., 1979; Trent and Naeve, 1980; Heinz and Kunz, 1982). These were designated from smallest to largest as NV1, NV2, NV2’/2, NVX, NV3, NV4, and NV5, and subsequently were defined in terms of their size in kilodaltons, e.g., for KUN virus P10, GP19, P21, p32, gp44, P71, and P98, respectively (Table I; Westaway et al., 1980).All the KUN proteins except p32(NVX) and gp44(NV3)were shown to be unique and distinct from C and E by tryptic peptide mapping; the map of the polypeptide with M, 19,000 was at that time assumed incorrectly to be that of the glycoprotein NV2 (Wright et al., 1977; Wright and Westaway, 1977). Similar results by peptide mapping were obtained for WN virus-specified proteins by Wengler et al., (1979). In some SDS-discontinuous gels, three bands of similar molecular weight can be resolved for KUN proteins which are now designated P19, GP20, and P21 (Fig. 1). In earlier reports with KUN and JE proteins, heterogeneity was noted in migration of NV2 or “GP19” and the presence of additional ns protein(s) was not suspected (Shapiro et al., l971,1972b, 1973b; Westaway and Shew, 1977). Henceforth, KUN glycosylated NV2 will be equated only with the glycoprotein GP20 or prM (putative precursor to M, see below), and P19 with the distinct ns protein which migrates slightly faster than P21 and GP20 in SDS-

54

E. G . WESTAWAY TABLE I

NOMENCLATURE AND COMPARISONS OF THE MOLECULAR WEIGHTSOF FLAVIVIRUS-SPECIFIED PROTEINS Original nomenclature0 NV5 NV4 v3 NV3 NVX NV21/2 NV2g NV2.r V2/NV11/2* NV1

M, x 10-3 KUN

JE

P98C P7 1 E/P51(E) GP44 p30-32 P2 1 GP19-20 P19 C/P14(C) P10

97” 69

MVE SLE

53

55 47

52

98 71 54/52

31 21

29 22

32 21

27 21

19

19

19 20.5

19 14 10

98 67

18 15 10

98 71

WN

44 47 19 14 10

TBE 91 67 53 47 25 *f

Proposed nomenclature*

DEN-2

YF

98 67 60 46 28/30

98 (104P 69 (69) 59/54 (53) 48/45 (46) 29 (31)

*

NS5 NS3 E NS1 nsla? ns2a/2b?

23 (19) 19 (18) 13 (13) 10 (12)

prM ns2a/2b? C ns4b?

*

* 20 * 18 19 15114.5 15114.5 15/14 * 10 9

a From Shapiro et al. (1971) and Westaway (1973).The prefix NV designates nonstructural proteins, and V the structural proteins. b Proposed by Rice et al. (19851, based on the gene order shown in Fig. 4. NS or ns indicates nonstructural protein, and queries indicate uncertainty where the N-terminal amino acid sequence is unavailable. The glycoprotein prM is the precursor of the structural protein M (M,about 8 x E and C are as defined in footnote c. c Nomenclature for flavivirus proteins from Westaway et al. (1980), as applied to KUN virus. E and C refer to envelope and core proteins, respectively. P98, P71, etc. indicate M, x of nonstructural proteins measured in SDS-phosphate gels. G P indicates glycoprotein; lowercase p or gp is used for flavivirus products not yet shown to be unique end products. d M, x 10-3 based on data in the following publications. Where possible weights were chosen from experiments including equivalent products of other flavivirus species, compared under similar conditions of electrophoresis. Sizes of established glycoproteins are underlined. From Heinz and Kunz (1982), Rice et al. (1985).Schlesinger et al. (1983), Shapiro et al. (1971),Smith and Wright (1985),Stollar (19691, Wengler et al. (1979, 1985), Westaway (1973, 1975), Westaway et al. (1977), Wright (1982), Wright and Warr (1985), Wright et al. (1983). e Value in parentheses is the theoretical M, x of the polypeptide predicted by translation of the nucleotide sequence for each product coded by YF virus RNA (Rice et al., 1986; Dalgarno et al., 1986). fAsterisk denotes that the product is not identified by gel electrophoresis. NV3 and NVX are also often difficult to detect for several flaviviruses, or they cornigrate with cell proteins. g NV2 was apparently heterogeneous in electrophoretic migration and was often incorrectly identified in gels as a single product with variable glycosylation (see text). The glycosylated species is almost certainly prM, the precursor of M (Castle et al., 1985; Rice et al., 1985; Dalgarno et al., 1986). The unrelated nonstructural proteins P18-19 and P21 cornigrate or overlap with prM in some virus-gel systems. h NV11/2 is a n intracellular equivalent of core protein C that is deficient in at least one tryptic peptide.

discontinuous gels; a similar distinction should be made for other flaviviruses. Thus GP20 was the only product precipitated from DEN-2 virus-infected cytoplasm by monospecific antibody to GP20 purified by binding to concavalin A-Sepharose (Smith and Wright, 1985). Some ns proteins smaller than E are often absent or not readily identifiable in gels, e.g., NV3 and NVX, M, 44,000-47,000 and

FLAVIVIRUS REPLICATION

55

25,000-32,000, respectively (Westaway, 1973; see Fig. 11, and the smaller ns proteins for TBE virus (Heinz and Kunz, 1982; Svitkin et al., 1981).When present, KUN NVX migrates heterogeneously in gels (Westaway, 1973)or as a doublet for DENS virus, e.g., p28/30 (Smith and Wright, 1985; Fig. 2), p32/33 (Ozden and Poirier, 1985),or p29/31

1

2

3

4

FIG.1. Analyses in a 13% SDS-discontinuous polyacrylamide gel of KUN virusspecified proteins and glycoproteins labeled at 27 hours postinfection in Vero cells. Cells were labeled for 60 minutes with [35S]methionine (lane 1)and with [aHImannose (lane 4). Purified virus had been labeled with [35S]methionine (lane 2) and [3Hlglucosamine (lane 3). GP20 is the glycoprotein NV2 discussed in the text. Three nonstructural proteins listed in Table I were not resolved in lane 1: GP44(NV3), p30-32(NVX), and PlO(NV1). Note that (1)previously P19 and GP20 were incorrectly aasumed to be identical or related, and both were designated NV2 or GP19 (Westaway et al., 1980) and (2) envelope protein E is not glycosylated. The alternative forms of nomenclature are shown in Table I. (Figure adapted from Wright, 1982, with permission.)

56

E. G . WESTAWAY

FIG.2. Analyses in 12% polyacrylamide gels of proteins and glycoproteins from Vero cells a t 75 hours. (a) Lanes 1and 2, DEN-2 virus-infected (I)and mock-infected (M) cells labeled for 30 minutes with [36S]methionine and then chased with excess methionine for 150 minutes; lane 3, 1%-labeled molecular weight markers (Amersham); lanes 4 and 5, cells labeled for 3 hours with FJHImannose. p15 and/or p14 are probably related to core protein C. (b) Effects of glycosylation inhibitor tunicamycin (TUN). Cells were labeled with [36S]methionine for 3 hours in the presence of TUN a t the concentrations (in micrograms per milliliter) indicated (lanes 2 to 7). For the sample in lane 8, cells were labeled with [3H]mannose for 3 hours. (Adapted from Smith and Wright, 1985, with permission.)

(Cleaves, 19851, and the larger DEN-2 product appears to be chased into the smaller of the pair. For DEN-2 virus NV3 is equivalent to GP46 (Fig. 21, whose unique identity was established by peptide mapping and specific radioimmunoprecipitation (Smith and Wright, 1985). Protein C in cells suffers loss of at least one tryptic peptide and then

FLAVIVIRUS REPLICATION

57

FIG.2b.

migrates slightly faster (Fig. 1) (Wright and Westaway, 1977;Wengler et al., 1979).The sequence data of Castle et al. (1985)with WN virus suggest that this apparent loss could be due to initiation of translation at a second initiation codon, equivalent to 15 amino acid residues further downstream from the normal NH, terminus. M is not found in infected cells (Shapiro et al., 1971;Westaway, 1973);however, three recent reports have shown that the amino acid sequence of M com-

58

E. G. WESTAWAY

prises the carboxyl end of a glycoprotein, with a polypeptide backbone of M, about 18,000. This was established from the nucleotide sequence of WN virus RNA and the amino acid sequence of NV2 associated with intracellular virions (Castle et al., 1985). For YF and MVE viruses, the coding content of M was also located within the nucleotide sequence for the precursor glycoprotein prM, which migrates in gels at the expected position of NV2 from cytoplasm (Rice et al., 1985, 1986; Dalgarno et al., 1986).These results confirm the original proposal that NV2 is cleaved to yield M during virus maturation (Shapiro et al., 1972a). Thus all the flavivirus proteins listed in Table I except for p25-32(NVX)have been shown to be unique end products. Their total molecular weights for (say) KUN virus amount to 380,000 (including about 6000 for carbohydrate), close to the total theoretical coding content of 380,763 within the open reading frame of the YF genome (Rice et al., 1985). The longest other open reading frame which is possible for YF virus is in the complementary strand, 804 nucleotides in length (Rice et al., 1985, 1986). In addition to the unique products in Table I, other flavivirus-specified proteins observed in gels include (1)carbohydrate-labeled intermediates of E for KUN and WN viruses (Wright et al., 1980, 1981; Wright and Wan-, 1985), (2) p80-82 or “NV4’/2” for TBE virus (Svitkin et al., 1981), KUN virus (Westaway et al., 1984), and DEN-2 virus (Smith and Wright, 1985; Ozden and Poirier, 1985; Cleaves, 19851,and (3) YF virus gp17 (Schlesinger et al., 1983) and DEN-2 virus gp22, gp16, and gp13 (Smith and Wright, 1985). DEN-2 gp22 is chased into GP20, and gp13 may be the residue of prM (putatively GP20) after cleavage to produce M. TBE p80 contains the amino acid sequences of P71(NV4) (Svitkin et al., 1981). Several transient additional products, including some much larger than 100,000 Da, have been observed for DEN-2 virus (Cleaves, 1985; Ozden and Poirier, 1985).Application of a nomenclature scheme for ns proteins based on the 5’ to 3‘ coding sequence in the genome (i.e., NS1 to NS5) proposed by Rice et al. (1985) to identification of proteins in gels poses some difficulties at present (see Table I), and must await further N-terminal amino acid sequence data.

E . Glycosylation of Virus-Specified Proteins Glycosylation occurs within the lumen of the endoplasmic reticulum and involves recognition during translation of a preceding hydrophobic sequence of 15-25 amino acids. This is inserted into the lumen and cleaved at a specific site at its carboxy terminus by a cell signalase (Perlman and Halvorson, 1983; von Heijne, 1984).The coding informa-

FLAVIVIRUS REPLICATION

59

tion in the nucleotide sequence preceding that of the amino terminus of each of the glycoproteins of YF, MVE, and WN viruses specifies the appropriate signal sequence, which was confirmed where N-terminal amino acid sequences were obtained (Riceetal., 1985;Castle etal., 1985; Wengler et al., 1985;Dalgarno et al., 1986).The process of glycosylation, and analyses of the glycans attached to E, GP44-47(NV3), and GP19-20(NV2),have been investigated only for KUN, WN, and DEN-2 viruses. As noted earlier, E in virions is not invariably glycosylated. The glycosylation pathway for E involves no prominent intermediates for DEN-2 virus (Smith and Wright, 1985) but several for KUN and WN viruses (Wright et al., 1980, 1981; Wright and Warr, 1985). These are gp66(E),which after pulse labeling with L3HImannoseis chased rapidly into the stable end product gp53/54(E), and gp58/59(E),which is stable after pulse labeling (Wright et al., 1981; Wright and Warr, 1985). Another KUN virus-specified product, gp56(E),is resolved in gels when cells are labeled with [3Hl-fucose. The relationship of gp66(E), gp58/59(E), and gp53/54(E) to E were established by radioimmunoprecipitation using antiserum to E (KUN) and by peptide mapping (KUN and WN) (Wright et al., 1981; Wright and Warr, 1985). These three KUN glycoproteins incorporated three glycans, M, 2900, 2400, and 1600). Of these the largest glycan appears to be complex (incorporating mannose, glucosamine, galactose, and fucose), the glycan of M,2400 is similar but lacks fucose, and the minor glycan of M, 1600 is simple (incorporating mannose and glucosamine only). In addition to the glycan of M, 2900, a large heterogeneous glycan, M, 36004200, was found reproducibly in gp56(E); it was unusually rich in galactose and fucose and contained terminal sialic acid, which was absent from the three smaller glycans of gp59(E) and gp53(E) (Wright et al., 1980, 1981). Only the two smallest glycans, M, 2400 and 1600, were found in KUN GP20(NV2). The three major glycoproteins specified by DEN-2 virus in Vero cells, GP6O(E), GP46(NV3), and GP20(NV2), were also found in infected C6/36 mosquito cells; tunicamycin inhibited glycosylation of all of these in both cell systems and hence all the glycans must be Nlinked to asparagine (Fig. 2; Smith and Wright, 1985). The three glycans of the DEN-2 glycoproteins are similar to those of KUN virus, viz. M, 2800 [in GP6O(E) only], M, 2300, and (in minor amounts) M, 1500. Both the smaller glycans were found in GPGO(E), GP46(NV3), and GP20(NV2). All the DEN-2 glycans were sensitive to digestion by endo-P-N-acetylglucosaminidaseand therefore are not of the complex type, consistent with the observation that none of them was sensitive to mild acid hydrolysis and therefore presumably they are not sialylated. The glycans of DEN-2 GP46 (NV3) appear similar in size to

60

E. G. WESTAWAY

those of GP20 (NV2), but are heterogeneous in charge when analyzed by Tris-borate electrophoresis. The sizes of the unglycosylated polypeptides of DEN-2 virus indicated that the M, of the total amount of glycan for each molecule was approximately 4000 for GP6O(E) and GP46(NV3), and 2000 for GP20(NV2) (Smith and Wright, 1985). The reported difference in M, for the unglycosylated and glycosylated protein E in YF virions was 5000 (Schlesinger et al., 1983). Reference to the sizes of the four glycans noted above and of the relevant polypeptides shows that all the glycans could not be attached to each polypeptide molecule; a single glycan of M, 2300-2400 would appear to be the predominant linkage to GP20(NV2), and various combinations of three or four glycans are probably attached at one or more Asn-X-Ser/Thr sites of the larger glycoproteins. The possible locations of these glycosylation sites (Asn-X-Ser/Thr) within the glycoproteins were recently defined by translation of the nucleotides comprising the open reading frames of YF, MVE, and WN viruses (Rice et aZ., 1985; Castle et al., 1985; Wengler et al., 1985; Dalgarno et al., 1986). In the NV2 product (equivalent to prM), the possible glycosylation sites are in the N-terminal half, one each for WN and MVE viruses (at Asn-15 for both), and three for YF virus (at Asn residues 12, 28, and 50). The preferred site for all is probably the common first one. In E for YF and MVE viruses, there are two, and one possible, glycosylation sites, respectively; their locations are Asn-309 and Asn-470 for YF virus, and Asn-154 for MVE virus. The second potential site for E of YF virus is in the hydrophobic C-terminal domain where it probably is not available for glycosylation (Rice et al., 1985). The strain of WN virus employed by Wengler and colleagues has no carbohydrate in E and no glycosylation site in the deduced amino acid sequence (Wengler et al., 1985). From a survey of available data, Pollack and Atkinson (1983) concluded that glycosylation sites in the first 100 amino acid residues of glycoproteins are enriched in complex glycans, and that high mannose (simple) glycans predominate when the site location is at 200 amino acid residues or greater. The glycosylation pattern for the E l and E2 glycoproteins of the alphavirus Sindbis conforms with this sequence (Mayne et al., 1985). If flavivirus glycosylation is also orthodox the glycans of E for YF virus should be of the high mannose type. In regard to the absence of apparent glycosylation of some of the E protein in virions of YF virus, Rice et d . (1986) suggested that this may result from the fact that Pro is the bridging amino acid in the Asn-ProThr glycosylation site. It should be noted also that the host cell may influence glycosylation. E of DEN-2 virus grown in C6/36 mosquito

FLAVIVIRUS REPLICATION

61

cells migrates more rapidly than E from virions grown in vertebrate (Vero) cells (Smith and Wright, 19851, and the difference was not simply the consequence of the absence of the anzyme sialyltransferase in cultured mosquito cells (Stollar et al., 1976). Glycoprotein NV3 (or NS1 in the terminology of Rice et al., 1985) of YF virus has two glycosylation sites at Asn-130 and Asn-208, measured from the amino terminus of NS1, and MVE virus NS1 has correspondingly conserved sites plus a third potential site at Asn-135. In view of the sizes of glycans for GP46(NV3) of DEN-2 virus (M,2300) and the reduction in M, of NV3 by about 4000 in cells treated with tunicamycin (Smith and Wright, 1985; Fig. 2), it seems likely that both the conserved sites for YF (and probably MVE) virus are glycosylated. Loss of glycan(s) or cleavage of the polypeptide backbones must occur when DEN-2 gp22 is chased into GP20(NV2), and when KUN gp66(E) is chased into gp53(E), as noted earlier. If processing of KUN gp66(E) is necessary for synthesis of virion E, the absence of carbohydrate in KUN E in virions is puzzling (Wright, 1982). However, glycosylation of E in YF virions is variable (Schlesinger et al., 1983), and a strain of WN virus has no glycosylation sites in E (Wengler et al., 1985). Such variation appears to have no effect on virus yield. Until the process of maturation of flaviviruses is elucidated (see Section 111, I), the unusual features of glycosylation of E remain unresolved.

F. Intracellular Location of Proteins and Their Possible Functions The flavivirus-specified proteins listed in Table I including the “soluble complement fixing antigen” or GP44-48(NV3) (Cardiff et al., 1970) remain bound during sedimentation to disrupted membranes of infected cells. Treatment of J E virus-infected membranes with dilute detergents or neutral or alkaline salts failed to completely release the virus-specified proteins; however, P97(NV5) and P69(NV4) were completely degraded by 0.1% trypsin (Shapiro et al., 1972b). In similar experiments with KUN virus (P. J. Wright, unpublished results), P21(NV2’/2) was largely removed from membranes by dilute detergent alone; the order of degradation by 0.1% trypsin was as follows: P98(NV5) and P71(NV4) (susceptible), gp66(E), gp53(E), P51(E), p32(NVX), GP20(NV2), and P19 (partially resistant); P14(C) and PlO(NV1) were completely resistant. The susceptibility to trypsin of P98(NV5)and P71(NV4) suggests they are only loosely bound to membranes, in agreement with the report that corresponding ns proteins of YF, MVE, and DEN-2 viruses are the least hydrophobic of the flavivirus products (Rice et al., 1986).

62

E. G. WESTAWAY

A number of attempts have been made to obtain membrane fractions enriched in proteins specified by JE, KUN, or DEN-2 viruses, by sedimentation in sucrose density gradients (Shapiro et al., 1972b; Kos et al., 1975; Boulton and Westaway, 1976; Stohlman et al., 1978). All these attempts employed loading of the cell extracts centrally within a discontinuous gradient. Electron microscopy of the material recovered in visible bands showed that in general smooth membranes sedimented in the less dense regions of gradients, whereas the denser bands contained increasing amounts of rough membranes (Kos et al., 1975; Boulton and Westaway, 1976; Leary and Blair, 1980). Location of plasma membranes varied, probably influenced by the preparation method (Westaway, 1980), being recovered either in the most dense band (Boulton and Westaway, 1976) or in several regions of the gradient (Kos et al., 1975; Leary and Blair, 1980). The distribution of flavivirus-specified proteins labeled in amino acids was similar in all membrane fractions; the variation in relative proportions of each protein was no greater than twofold, or slightly greater for P95-98(NV5), which is often subject to proteolytic digestion in cell extracts (Westaway, 1980). Furthermore, all membrane fractions contained a background smear of labeled cell proteins in electropherograms. In further experiments with KUN virus, cell extracts were loaded on top of 25-60% (w/w) discontinuous sucrose density gradients, rather than centrally, prior to rapid sedimentation (Westaway and Ng, 1980). The striking result was that after labeling in amino acids for 10 minutes or 3 hours, the labeled cell proteins were located predominantly in the least dense membranes at the interface of 2 5 3 0 %sucrose, whereas P98(NV5),P71(NV4), P51(E),and the small ns proteins were incorporated in the denser (“heavy”) membranes at the interface of 30-44% sucrose (Fig. 3). In a pulse-chase experiment using L3H-1mannose, gp66(E), gp59(E), gp53(E), and GP20(NV2) were labeled only in the heavy membranes. Significant amounts of the intracellular forms of KUN E protein, whether pulse-labeled in amino acid or carbohydrate migrated during the chase period to nuclear-associated membranes which were removable only by detergent. While the top-loading technique for membrane sedimentation has made it possible to follow more easily the intracellular movement of flavivirus proteins, specific enrichment of individual ns proteins in different membrane fractions has still not been achieved. However, in uitro assays of KUN virus RNA polymerase activity have shown that most activity is concentrated in the heavy membrane fractions, and details will be reported elsewhere (Chu and Westaway, 1987). An alternative method for probing the intracellular location of virus proteins or biosynthetic processes is immunofluorescence using specif-

FLAVIVIRUS REPLICATION

63

FIG. 3. Distribution of KUN virus-specified proteins and glycoproteins in subcellular and membrane fractions of infected Vero cells. Cells were labeled at 24 hours postinfection for 10 minutes with W-labeled mixed amino acids (a) or for 3 hours with [35S]methionine (b). Cells were disrupted and separated into cytoplasmic and nuclear components. The cytoplasmic extracts were layered on discontinuous sucrose gradients and sedimented at 50,000 rpm for 3 hours in a SW56 rotor; visible bands of membranes were harvested and equal volumes were analyzed in polyacrylamide gels. In (a),the 10minute pulse (P) and 30-minute chase (C) samples of the cytoplasmic extract were analyzed; the membrane fractions of increasing density are designated “light” (L),“intermediate” (I), “heavy” (H), and “polysome-like” (Polys), and lane U represents an uninfected cytoplasmic extract. Visible H bands were also strongly labeled in GP19 and in E-related glycoproteins gp66, gp59, and gp53 (refer to Fig. 1) during a 10-minute pulse with [3H]mannose in a parallel experiment. In (b), comparisons are made of infected and uninfected cytoplasmic extracts. Cy, N, and NP refer to the subcellular components comprising cytoplasmic extract (Cy), and the nuclear supernate (N) of washed and pelleted nuclei (NP); fractions 1 to 5 represent visible membrane bands of increasing density in gradients modified from (a) to obtain additional bands from the cytoplasmic extract such that fractions 1 + 2 = L, 3 = I, 4+5 = H, and 6 = Polys. (“a” is adapted from Westaway and Ng, 1980, with permission, and ‘8” is from P. W. G. Chu, P. Siatskis, and E. G. Westaway, unpublished data.)

64

E. G. WESTAWAY

FIG.3b. See legend on p. 63.

ic antibodies. KUN E was localized by anti-E antibodies as early as 4 hours postinfection in a rim around the nucleus, which subsequently thickened and extended peripherally as a branching fringe, often from one side of the nucleus (Ng et al., 1983). Antibodies to the “soluble complement fixing antigen” of DEN-2 virus, which was recently identified definitively as GP46(NV3) by Smith and Wright (1985), reacted with viral antigens on plasma membranes (Cardiff and Lund, 1976) and produced perinuclear fluorescence (Cardiff et al., 1973) in DEN-2 virus-infected cells. The sequence data of Rice et al. (1985, 1986) indicate that overall NS1 of YF virus (equivalent to NV3) is hydrophilic, and Rice et al. suggested that it may occur in a membrane-bound form and in a soluble form in which a putative hydrophobic C-terminal segment is absent. Monoclonal antibodies t o P69(NV4) of J E virus reacted by immunofluoresence in a characteristic, very thin rim around the nucleus (Gould et al., 1983). Because of the involvement of double-stranded RNA in flavivirus replication (Westaway, 19801, we looked for possible sites of flavivirus

FLAVIVIRUS REPLICATION

65

RNA synthesis by reacting antibody to double-stranded RNA [antipoly(A):poly(U)] with KUN-infected cells during the period of maximum RNA synthesis (20-28 hours). Staining was intense in discrete foci in the perinuclear region and extended peripherally in a fine network (Ng et al., 1983). Subsequently, the monoclonal antibody of Gould et al. (1983) which cross-reacts with KUN virus P7UNV4) was compared directly by immunofluorescence with anti-double-stranded RNA, but only rarely did the two fluorescent patterns in KUN-infected cells correspond (E. G. Westaway and M. Goodman, unpublished results). However by modifying the cell fixation method, absolute correspondence of immunofluorescence with the two antibody preparations was obtained; furthermore, the same dual pattern of fluorescence occurred using an antiserum t o P98(NV5). These results will be published in detail elsewhere, but a reasonable conclusion is that both the ns proteins P98(NV5) and P71(NV4) are associated with polymerase activity. The conservation and similarity in sizes of these flavivirus proteins to two virus-specified components of the purified alphavirus polymerase, M, 90,000 and 63,000 (Clewley and Kennedy, 1976), were noted earlier (Westaway, 1977) when it was first suggested that they may be flavivirus polymerase components. Further light on the possible polymerase role of NV5 has been shed by Rice et al. (1986) in an analysis of evolutionary relationships. The deduced amino acid sequence of NS5(NV5) between residues 3037 and 3181 was compared with the sequences of RNA-dependent polymerases of 10 positive-stranded RNA viruses infecting animals and plants. The canonical sequence Gly-Asp-Asp, thought to be essential for RNA-dependent RNA polymerases (Kamer and Argos, 19841, was completely conserved in all cases. Apart from the role of NV2 or prM as precursor to M, and the apparent association of the largest stable ns proteins NV4 and NV5 with polymerase activity, the functions of the other ns proteins remain completely unknown. One or more may be involved in providing the type 1 cap for viral RNA, and it has been suggested that prM(NV2) and NSl(NV3) may have autoproteolytic activity (Rice et al., 1986; Dalgarno et al., 1986).

G . Translation Strategy The control of translation of the flavivirus genome has been dificult to analyze for a number of reasons (Westaway, 1980). First, translation of cell messengers continues largely unabated in infected cells and their labeled products often obscure viral proteins in analytical gels. Second, with several flavivirus species some ns proteins are pro-

66

E. G . WESTAWAY

duced in very small amounts or are not detectable. Third, conventional pulse-chase experiments have until recently failed t o show any precursor polyproteins, and translation mapping experiments failed to provide a definitive or consistent gene sequence. Finally, amino acid sequences of ns proteins have not been identified in cell-free translation products using genomic RNA. The recent publication of the complete nucleotide sequence of YF virus showing one long, open reading reading frame is a major advance (Rice et al., 1985). These results have been confirmed in part by the RNA sequences coding for the structural proteins of MVE virus (Dalgarno et al., 1986) and of WN virus (Castle et al., 1985; Wengler et al., 19851, and by some sequence data for the ns regions for MVE virus (Dalgarno et al., 1986) and DEN-2 virus (unpublished results of V. Vakaria, T. Yaegashi, R. Feigny, S. Kohlekar, and R. Padmanabhan, cited by Rice et al., 1986).Recently the complete nucleotide sequence of the genes coding for all the ns proteins of WN virus became available (Castle et al., 1986). The logical interpretation of these sequences is that translation is initiated at the first methionine codon near the 5’ terminus and a polyprotein is translated which is rapidly posttranslationally cleaved to yield the final products. The site of translation appears t o be in heavy membranes (Westaway and Ng, 1980). The proposed gene sequence is shown in Fig. 4 (Rice et al., 1985). The flanking sequences for the initiation codon are not uniform: GAACAAUGU for YF RNA (Rice et al., 19851, U U C A A E G U for MVE RNxDalgarno et al., 1986), and U C U C G E G U for WN RNA (Castle et al., 1985). In all cases, a second in-phase AUG codon is located 14, 15, or 15 codons downstream, respectively; this lies within the common sequence UCAAUCGC which still does not agree very well with the consensus sequence for the initiation site of eukaryotic translation, viz. CC&X&G(G) (Kozak, 1984). Heterogeneity in the N-terminal amino acid sequence of C for WN virus indicates that the second initiation codon may also be a site for initiation (Castle et al., 1985).N-terminal amino acid sequence data established unequivocally the exact location within the known nucleotide sequence of the coding sequence C-prM-E-NS1, NS3, and NS5, and the predicted size of the gene products is in reasonable agreement with known sizes except possibly for NS5 (apparently 6000 Da in excess) where the N-terminal amino acid sequence data are less precise (Rice et al., 1985; Dalgarno et al., 1986; Castle et al., 1985; Wengler et al., 1985). The translation products intervening between NS1, NS3, and NS5 are indeterminate because their N-terminal amino acid sequences are not available. These intervening regions have been subdivided, as ns2a and ns2b, and ns4a and ns4b, based on putative cleavage sites similar to those

FLAVIVIRUS REPLICATION

67

YELLOW FEVER 17D GENOME (10. 862nt)

FIG.4. Organization and processing of proteins encoded by the yellow fever virus genome. Untranslated regions of the genome are shown as single lines and the translated region as an open box. The open triangle is the initiation codon (AUG); the solid diamond the termination codon (UGA). The protein nomenclature is described in Table I. The single-letter amino acid code is used for sequences flanking assigned cleavage sites (solid lines). Two other potential cleavage sites are shown as dotted lines. Structural proteins, identified nonstructural proteins, and hypothesized nonstructural proteins are indicated by solid, open, and hatched boxes, respectively. Cleavage sites shown are predicted by NH2-terminal protein sequence data (see Table II), or are based on homology with confirmed cleavage sites and the sizes of yellow fever-specific polypeptides observed in infected cells. Less homologous alternative cleavage sites in the nonstructural region occur after residue 1946 (Gln-Arg-Arg 1Gly), residue 2548 (Ala-Arg-Arg .1His), residue 2707 (Gln-Arg-Arg 4 Phe), and residue 3104 (Ser-Arg-Arg 1Asp). (Reproduced from Rice et al., 1985, with permission. Copyright 1985 by the American Association for the Advancement of Science.)

used for generating M, NS3, and NS5, and on observed sizes of small ns proteins. Rice et al. (1986) suggested that heterogeneity among flavivirus polypeptides may result from variable exopeptidase digestion of the C-terminal residues, or from alternative internal cleavages. The cleavage enzymes involved in proteolytic processing appear to be common to Y F and WN virus infections (Rice et al., 1985; Castle et al., 1985; Wengler et al., 1985). They produce cleavages (1)immediately after the initiating methionine, (2) after double basic amino acid residues and (3) after short side-chain amino acids preceded by the signal sequences of the three glycoproteins (Table 11). For WN virus, the initiating methionine is not always cleaved from C, and it seems likely that cellular proteases or signalases are involved in all cleavages, although the possibility of viral-encoded proteases exists (see detailed discussions by Castle et al., 1985; Wengler et al., 1985; and Rice et al., 1986). Proteolytic cleavage after a pair of basic amino acids is well documented for secretory cell proproteins or prohormones, and for other viral membrane proteins (see Garoff et al., 1982). The late cleavage of NV2(prM) to produce M may be catalyzed by a cellular protease associated with the Golgi (Rice et al., 1985;’Dalgarno et al., 1986). However, as noted earlier, M has not been detected in infected

68

E. G. WESTAWAY TABLE I1 SIMILARITY OF CLEAVAGE SITESAT AMINOTERMINIFOR WHICHAMINOACIDSEQUENCES ARE AVAILABLEQ Protein C prMorNV2 M E NSlorNV3 NS3orNV4

Virus

Cleavage site at amino terminusb

YF MVE WN YF MVE WN YF MVE WN YF MVE WN YF MVE WN YF MVE

M .1 SGR M 4 SKK M .1 SKK .1 VTLV -TGG 4 LKLS -AAA J VTLS -AGA SSRSRR J AID SKRSRR .1 SIT SRRSRR .1 SLT -VGPAYS .1 AHCIG -VAPAYS FNCLG .1 FNCLG -VAPAYS -VGA .1 DQGCAI --HA .1 DTGCAI .1 DTGCAI -VHA ARR .1 SGDVLWD TKR .1 GGVFWD

a Data from Rice et al. (1985, 1986), Dalgarno et al. (1986), Castle et al. (1985), and Wengler et al. (1985), which can be related unambiguously to the nucleotide sequence. The single letter code for amino acids has been used. b Line represents a preceding signal sequence.

cells. The initiating methionine may be removed by a cellular methionine amino peptidase, and a possible autoproteolytic role of prM(NV2) or NSl(NV3) is discussed by Rice et al. (1985, 1986) and Dalgarno et al. (1986). The gene sequence of Y F virus accords well with the apparent translation sequence of KUN virus obtained by following the incorporation of amino acid label in infected cells after pactamycin treatment or in the presence of hypertonic salt, during which translation of all KUN proteins was completed in less than 20 minutes in the sequence 5' CE...NV4-NV5 3', where NV4 or P71 corresponds to NS3 and NV5 or P98 corresponds to NS5 (Westaway, 1977). Note that this order corresponds with a regular increase in molecular weights of the products. However, when KUN-infected Vero cells were synchronously reinitiated in translation by reversal of a 20-minute hypertonic inhibition block, under conditions which yielded correct gene sequences for poliovirus and Sindbis virus, NV5 was labeled within l minute and com-

FLAVIVIRUS REPLICATION

69

pleted in translation in 6-7 minutes. Because some other recognizable products also incorporated label within 2 minutes, the conclusion was drawn that multiple reinitiation occurred (Westaway, 1977). In a n alternative interpretation, Rice et al. (1986) suggested that the labeling observed after synchronous reinitiation was associated with long ribosome transit times (greater than the period of 20-minute salt treatment), and hence was due to preinitiated translation products. Another explanation is based on the observation of Koch et al. (1982) that distal internal initiation may occur preferentially on poliovirus RNA in HeLa cells after removal of hypertonic inhibition imposed for 25 minutes (it was shown previously that all poliovirus proteins were completed in translation in 12 minutes under hypertonic conditions in HeLa cells by Saborio et al., 1974; and in Vero cells by Westaway, 1977). Furthermore, McClure and Perrault (1985) found that 28 S ribosomal RNA bound to the same internal region of poliovirus RNA (bases 5075-5250) as did ribosomes in uitro (position 5300) (McClain et al., 19811, and during cell-free translation when internal initiation occurred in apparently the same (P3)region of poliovirus RNA (Dorner et al., 1984).The results of Koch et al. (1982) therefore suggest that the 20-minute pretreatment of KUN virus-infected Vero cells with hypertonic salt may have induced internal initiation (Westaway, 1977). Although at that time 40-minute pretreatment proved impractical (reinitiation was too slow and too weak), recently in one experiment we have been successful in obtaining rapid reinitiation after reversal of a 40-minute hypertonic salt block as used by Saborio et al. (1974) with poliovirus; under these conditions, the largest ns protein P98 or NV5 could not be labeled during the first 5 minutes of reinitiated translation whereas C and E were (E. G. Westaway and A. Schrader, unpublished result). It is of interest that the possibly artifactual internal distal initiation of both poliovirus and KUN virus occurred about twothirds of the way along the genome, in the region of commencement of putative RNA polymerase genes (P3 region and NV5, respectively). The sequence 5' C-E...3' is supported by cell-free translation experiments with genomic RNA of WN, TBE, and KUN viruses (Wengler et al., 1979; Svitkin et al., 1981; Monckton and Westaway, 1982). Several in uitro products were in the M, range 120,000-160,000 and all contained tryptic peptides of E and C; the other amino acid sequences were not identifiable. Subsequently, Svitkin et al. (1984) showed that inclusion of cell membranes during in uitro translation of TBE virus RNA was essential to obtain good yields of p13 and p53, equivalent to C and E, respectively. Presumably signal sequences were then recognized in the nascent polyprotein, which was subsequently cleaved by cell signalase in the membranes.

70

E. G. WESTAWAY

The largest in uitro product was approximately equal in size to a polyprotein probably comprising C, NV2(prM), E, and NV3 (M,about 130,000); it is of interest that C, NV2, E, and NV3 of J E virus and C, NV2 and E of KUN virus were the only products labeled in uitro in the presence of relatively high concentrations of translation inhibitors, e.g., puromycin and cycloheximide (Shapiro et al., 1973b; Westaway and Shew, 1977, and unpublished results), or emetine (E. G. Westaway, unpublished results). All the inhibitors thus appear to block elongation beyond the gene for NV3, i.e., beyond NS1 in the sequence shown in Fig. 4. Puromycin substitutes for aminoacyl-tRNAs at the acceptor site of the peptidyltransferase center on the 60 S ribosomal subunit. Emetine inhibits translocation by acting on the 40 S ribosomal subunit and irreversibly stabilizes ribosomes on polysomes. However, cycloheximide blocks translocation by reversibly binding t o the 60 S ribosomal subunit (Vazquez et al., 1982). The common reason why all these inhibitors exert their greatest inhibitory effect on translation of genes downstream from NV3 (designated NS1 by Rice et al., 1985) remains unclear. It may be relevant that NV3 is the last of the glycoproteins in the translation sequence. As no more signal sequences are thenceforth translated, the accessibility of the messenger to ribosomes, membranes, and possibly translation inhibitors may change in this region. Also relevant may be the comment by Rice et al. (1986) that a potential secondary structure in YF virus RNA can be predicted in the junction region between the coding regions for the structural and ns proteins. Ultraviolet irradiation of cells infected with the alphavirus Sindbis enabled translation mapping by measurement of rates of inactivation of individual genes on the viral messengers (Fuller and Marcus, 1980). We applied this method to KUN virus (Fig. 9 ,after verifying that it yielded the correct gene sequence and approximate gene target sizes for the products of the 26 S messenger RNA of the alphavirus Semliki Forest virus (Westaway et al., 1984). Apart from the results with P98, the order of inactivation obtained for the identifiable products accords reasonably well with that expected by reference to the YF virus genomic sequence (see Fig. 41, e.g., 5' C-E-P19/P21-P71-P103', bearing in mind that analyses for KUN GP46(NV3 or NS1) and p32(NVX or ns4a?) were not possible, as neither was detectably labeled. Furthermore, if p20 (Fig. 5 ) is assumed to be glycosylated and equivalent to prM, its target size measured over the first 50 seconds of irradiation placed its gene adjacent to that of E. However, the target size for the P98(NV5) gene placed it completely overlapping with the gene for E; the similar resistance t o inactivation for the two genes is obvious merely by inspection in Fig. 5 of their products labeled during the 60

FLAVIVIRUS REPLICATION

71

FIG.5. Measurement of the order of inactivation of genes of KUN virus by ultraviolet irradiation, analyzed by the effects on translation. Petri dish cultures of Vero cells were irradiated and then labeled for 30 minutes with [35Slmethionine a t 24 hours postinfection followed by a 1-hour cold chase. (a) Equal volumes of infected cytoplasm were electrophoresed in a 10% discontinuous gel. The first lane (M) contains a mockinfected sample, and the period of exposure to ultraviolet light in seconds is marked below each lane of the infected samples. (b) Inactivation curves of translation products. The relative incorporation of [35S]methionine in proteins translated after irradiation of infected cells was measured by densitometer scans of autoradiograms as in (a), using a difference analysis method. Also shown is the inactivation curve (- - -) for plaqueforming particles (PFP) under the same conditions of irradiation. The regression lines shown were calculated from plots of the fractions of surviving activity, measured over inactivation periods of 10-60 seconds or greater; those for P14(C) and P10 are plotted from the results of analyses in a 8% phosphate gel, which resolves them. A regression line for p20 (not shown) was linear for only 50 seconds of irradiation and was similar in slope to that for P51(E). ( 0 )P51(E); (A)P98; (H)P71; (0) P19; (A)P21; (0) P14(C); ( 0 ) P10. (Adapted from Westaway et al.,1984, with permission.)

72

E. G. WESTAWAY 1.

b

0,

--*>

0.

U

C

0 0

p

->2

0,

I

a

PD r

0

-c 0

U

L

O

O

IL

\

0 I

10

P2 1 1

I

20

30

40

50

60

lrradlatlon time (seconds)

FIG.5b. See legend on p. 71.

minutes after irradiation. Because of this anomaly, we proposed that translation of P98 was initiated independently much further downstream, near a second ribosomal attachment site. It could be argued of course that irradiation possibly produced artifactual internal initiation similar to that discussed above in relation to hypertonic salt treatment; however, no precedent exists for this. Although no high molecular weight precursors were detected when KUN virus-infected Vero cells were pulse-labeled in amino acids for only 2 or 3 minutes (Westaway, 1977), proteins as large as 130,000 or 220,000 Da were labeled during 5-, 7-, 11-,or 15-minute pulses in DEN-2 virus-infected BHK2l cells (Cleaves, 1985; Ozden and Poirier, 1985). In the DEN-2 experiments, labeled ns proteins smaller than 30,000 Da were identified within 15 minutes of commencement of labeling, at the expense of label chased from p55 and the high molecular weight proteins. Precursor-product relationships were not examined by peptide mapping or by immunoprecipitation. The largest ns protein P92/98(NV5) was always strongly labeled during short pulses, and P71(NV4) less strongly. In view of the proposal that posttranslational cleavages of the ns proteins occur after a pair of basic amino

FLAVIVIRUS REPLICATION

73

acids (Rice et al., 1985, 19861, it is surprising that amino acid analogs of arginine and lysine did not block appearance of the DEN-2 small ns proteins (Ozden and Poirier, 1985), in agreement with earlier results for KUN virus (Westaway, 1973). However, Cleaves (1985) found that a combination of proteolytic inhibitors TPCK and Zn2+ caused an increase in the amounts and size (to 250,000 Da) of the high molecular weight proteins, and a decrease in the intensity of all viral bands in the 10,000-to 86,000-Da regions of the analytical gel, but notably no change for P98(NV5). In the pulse-chase experiments with DEN-2 virus, no relationship was shown or could be inferred between the largest ns protein NV5 and the transient larger products. In summary, the flavivirus structural proteins are translated in the sequence 5'-C-prM-E, and translation continues into the ns proteins, commencing with the glycoprotein NV3 designated NS1 by Rice et al. (1985). High molecular weight proteins as large as 220,000 Da are pulse-labeled in DEN-2 virus-infected BHK2l cells, and these appear to include the sequences of the ns proteins smaller than 30,000 Da, which like the structural proteins arise by posttranslational cleavage. Where characterized, the cleavage sites correspond to those recognized by cell signalases (for prM, E, and NS1 or NV3) or are preceded by a pair of basic amino acids (for M, NS3 or NV4, and possibly NS5 or NV5), as found for the cleavage sites in cell proproteins or in prohormones. The rate of cleavage must be very rapid in KUN virus-infected cells, but slow enough to be detected for the smaller DEN-2 ns proteins. Glycosylation at Asn-X-Thr/Ser occurs in membranes after recognition of signal sequences in nascent polyprotein, and because of hydrophobic sequences virtually all translation products are bound to membranes. However, the distribution of these products among the various membranes and the strength of their attachment appear to vary. The reasons for the apparent variability among flaviviruses in production of the ns proteins smaller than the two largest (see Section II1,D) remain unexplained. Rice et al. (1985) suggested that regulation could occur by premature termination as well as by nonuniform rates of translation. However, the most distal product NV5(NS5) is always labeled at least as strongly as NV4(NS3) and other upstream products. It seems unlikely that the variable products all suffer rapid proteolytic degradation or are exported rapidly from the cell, and hence the possibility remains that translation may not always occur in an orthodox manner. The discussion above indicates that major anomalies appear to exist in regard to translation of the gene for the largest product NV5(NS5), located at the 3' terminus. These are (1)NV5 was strongly and rapidly labeled and processed without detectable delay in short pulse experi-

74

E. G . WESTAWAY

ments whereas all other ns proteins were poorly labeled or suffered delays in posttranslational processing, (2) its appearance in gels was consistently unaffected when translated in the presence of inhibitors of proteolytic cleavage including analogs of basic amino acids, and (3) the target size of its gene during ultraviolet irradiation was smaller than those of all the other ns proteins, rather than being the largest as might be predicted (see Fig. 4). Anomaly 1 would imply that cleavage of other gene products nearer the NH, terminus of a polyprotein occurs much more slowly than that of NV5.

H . Strategy of Viral RNA Synthesis The original observation with DEN-2 virus of synthesis of 44 S RNA, double-stranded 20 S RNA, and a poorly resolved and partially RNase-resistant 26 S RNA (Stollar et al., 1967) was subsequently confirmed with several other flaviviruses (see Westaway, 1980). The 20 S RNA was shown to contain full-length strands of 44 S RNA of plus and of minus polarity and was labeled rapidly with r3H1uridine, but the roles of 20 S and 26 S RNA remained obscure. Cleaves et al. (1981) separated pulse-labeled DEN-:! virus 26 S and 20 S RNA by LiCl precipitation and established that the former had properties compatible with those of a replicative intermediate (RI) and that the latter was equivalent to a replicative form (RF). After the latent period, about 10%of incorporated [3Hluridine in RI was in RNA(- 1 strands, indicating that a very small amount of RNA(-) strand production continues throughout the replication cycles. Rice et al. (1986) proposed that binding of protein C to a stable 3’ terminus structure on the RNA(+) strand could prevent replication of minus strand RNA and initiate encapsidation; a similar role for C was proposed previously (Westaway, 1980). The RNase resistance of the flavivirus RI is high, about 60-70% (Cleaves et al., 1981; Chu and Westaway, 1985) and leads to a calculation of only 1.0 nascent RNA strand per template during asymmetric copying, based on the formula of Baltimore (1968). Separation of flavivirus RI from RF by LiCl precipitation was an important advance because the two RNA species cannot be satisfactorily resolved by gel electrophoresis or sedimentation. However, the role of RF in replication remained undefined. By comparing the total incorporation of [3Hluridine in KUN virus 44 S RNA, RF, and RI separated after pulse-chase experiments, it was deduced that RF functions as a recycling template for 44 S RNA synthesis; from these results it was concluded that initiation of semiconservative replication converts the RF to an RI on which only one nascent strand of RNA is

FLAVIVIRUS REPLICATION

75

asymmetrically replicated per cycle (Chu and Westaway, 1985). This model is in accord with the observed kinetics of labeling, the RNase resistances of RF and RI, and the products released when they are fully denatured. In recent in uitro polymerase assays for KUN virus RNA we have shown by kinetic studies that the sequence of incorporation of L3H1uridine is into RI, RF, and 44 S RNA (Chu and Westaway, 1986) in accord with the labeling sequence obtained in uiuo. By determination of the terminal sequence in the RF of WN virus, Wengler and Wengler (1981) showed that both RNA molecules are exact complements of each other. The RNA plus strand in the RF apparently lacked a cap but otherwise appeared identical in sequence to the viral genome. The presence at the 3’ ends of both plus and minus strands of the terminal dinucleotide CUoH, together with the common heptanucleotide ACACAGG terminating a few bases (5 or 10) upstream, suggested that the sequences were part of the recognition site for the WN RNA polymerase. Similar sequences were noted by Rice et al. (1985, 1986) in YF RNA; the terminal trinucleotide ACUoH and a pentanucleotide ACACA terminating a few bases (9 or 10) upstream are common to the 3’ ends of plus and minus strand RNA. The conservation reported for these hypothetical recognition sequences argues strongly that a similar polymerase complex initiates copying on both strands. Rice et al. (1986) observed that the conserved pentanucleotide ACACA forms a loop in which only two of the five nucleotides are hydrogen bonded within a stable hairpin loop of possible secondary structure involving the terminal 91 nucleotides, and suggested that the pentanucleotide sequence could be recognized and bound by a viral replicase even in the presence of a stable secondary structure at the 3’ terminus. The 3’ region of YF virus RNA has also been sequenced via cloned cDNA by Grange et al. (1985);they reported that many stable secondary structures are possible in the terminal 120 nucleotides, and in the most stable the pentanucleotide ACACA is again not hydrogen bonded. A further feature of the untranslated sequence at the 3’ terminus of YF virus RNA is a set of three closely spaced repeated sequences between nucleotides 10,374 and 10,520; each is about 40 nucleotides long, with only four to six changes between them in pairwise comparisons (Rice et al., 1985; Grange et al., 1985). Their significance in regard to replication is unknown. The composition, form, and structure of the polymerase complex remain obscure. As noted in Section III,F, the two largest ns proteins of KUN virus were located by immunofluorescence in the perinuclear region, coincident with the sites of double-stranded RNA, the presumptive template (E. G. Westaway and M. Goodman, unpublished

76

E. G. WESTAWAY

results). For the smaller ns proteins, no search for such association has been made. The membranous structures at the apparent sites of KUN RNA synthesis were visible by phase-contrast microscopy, and underwent rearrangement after treatment of cells with agents which disrupt microtubules, into relatively fewer but larger immunofluorescent foci (Ng et al., 1983). However, no reorganization of the cytoskeleton in untreated infected cells was observed by immunofluorescence using antibodies to tubulin, actin, or intermediate filaments. RNA polymerase complexes have been isolated in active form from flavivirus-infected cells on several occasions, but the in uitro products were poorly characterized (see Westaway, 1980). A 250 S detergentsensitive replication complex was isolated using sedimentation analysis from SLE virus-infected cells which retained both RNA polymerase activity and pulse-labeled 44 S RNA and partially RNase-sensitive 20 S and 26 S RNA (Qureshi and Trent, 1972). Crude cytoplasmic extracts from KUN virus-infected cells produced mainly RF, RI, and small amounts of 44 S RNA in polymerase assays; the activity started as early as 8 hours and reached a maximum at 24 hours (Chu and Westaway, 1985). In membrane fractions prepared from KUN virus-infected cells, RNA polymerase activity was concentrated mainly in the heavy membrane fractions of cytoplasm, but late in infection (32 hours) was associated most strongly with nuclear pellets (Chu and Westaway, 1985, 1987). In thin sections of flavivirus-infected cells, several virus-induced membranous structures are present in the perinuclear region, including clusters of spherical, smooth membrane vesicles, mean diameter of about 100 nm with reticular electron-dense centers (Fig. 6; Murphy, 1980; Westaway, 1980). These were described as smooth membranous structures (SMS)and sedimented to 45% sucrose with heavy membranes in membrane fractionation gradients (Leary and Blair, 1980). The appearance of these vesicles is superficially similar to the cytopathic vesicle type I described in alphavirus infections, which is claimed to be the site of viral RNA synthesis (Friedman et al., 1972). Small vesicles also accumulate in the perinuclear region of poliovirus infected cells and are reported to be the site of viral RNA synthesis; it is noteworthy that these poliovirus-induced vesicles are enriched in the 25% sucrose fraction of membrane gradients (Bienz et al., 1983). By analogy, the small vesicles or SMS in flavivirus-infected cells may represent the polymerase complex which cosediments with them (Chu and Westaway, 1987) but are of a much greater density (equivalent to 45% sucrose) than the poliovirus polymerase complex (equivalent to 25%sucrose).

FLAVIVIRUS REPLICATION

77

FIG.6. Thin section of a KUN virus-infected Vero cell at 24 hours postinfection, showing flavivirus-specific membrane structures. Cells in monolayer cultures were fixed in situ before embedding, sectioning, and staining. The bar represents 500 nm. An array of convoluted membranes (CM) near the nucleus (N)adjoins aggregates of spherical smooth membrane structures (SMS). The aggregates of SMS are partially bounded by endoplasmic reticulum. Virions (arrowed) are interspersed among the SMS and have commenced collecting within distended cisternae of endoplasmic reticulum which is an extension of parallel membranes or lamellae of the rough endoplasmic reticulum (RER). The inset shows two negatively stained 45-nm purified virus particles, with projections surrounding the envelope. (From P. W. G. Chu and E. G. Westaway, unpublished.)

I . Virus Maturation and Release

A large number of electron micrographs of thin sections of flavivirus-infected cells were published prior to 1980, and most of these observations were reviewed previously (Westaway, 1980). A more detailed review and many excellent electron micrographs of SLE virusinfected cells were published by Murphy (1980). At that time firm evidence of budding of flavivirus particles appeared to be lacking and the situation remains unchanged. All reports refer to extensive pro-

78

E. G . WESTAWAY

liferation of both smooth and rough endoplasmic reticulum and accumulation of virions within their cisternae. A typical thin section of a KUN virus-infected cell is shown in Fig. 6. The hypertrophied membranes appear to originate in the juxtanuclear region and form the following organelles: (1)Aggregates of the SMS noted previously appear as spherical structures ranging from 50 to 200 nm in diameter, each with a thickened smooth membrane and enclosing fibrils often visible as a diameter or with a denser center. The aggregates are sometimes contained within large cisternae of smooth or rough endoplasmic reticulum (Murphy, 1980), or are adjacent to the convoluted smooth membranes described below (see Fig. 6). Several reports presented evidence that the SMS are produced by budding from the bounding endoplasmic reticulum (Calberg-Bacq et al., 1975; Murphy, 1980; Leary and Blair, 1980). These reports also show accumulation of apparently mature virions among the aggregates of SMS. (2) Convoluted ordered masses of smooth membranes often adjacent to the clusters of SMS described in (11, becoming very prominent late in infection (Murphy, 1980). (3) Lamellae of rough endoplasmic reticulum enclosing rows of virions within the lumina. The lamellae are often adjacent to and may be connected to the organelles in (1)and (2) (Murphy, 1980),but also extend to the plasma membrane (Boulton and Westaway, 1976). (4) Distended cisternae bounded by smooth or rough membranes in which virions accumulate; this probably facilitates exocytosis by fusion of the bounding membrane with the plasma membrane, or by release of the “package” of virions as infected cells lyse (Murphy, 1980). No naked nucleocapsids or cores are released from disrupted cells, and the 25-nm particles attached to the outer surface of rough endoplasmic reticulum cannot be distinguished from ribosomes (Westaway, 1980; Murphy, 1980). Hence, it is impossible at present to define the events leading to morphogenesis. The possibilities discussed by Schlesinger (1977) and Murphy (1980) include rapid budding and release a t the plasma membrane (unlikely in view of accumulation of virions in juxtanuclear organelles), and some form of de novo synthesis in a cytoplasmic matrix. Leary and Blair (1980) propose that the SMS is formed by budding of envelope protein containing endoplasmic reticulum, thereby enclosing capsid protein and the viral RNA genome from the cytoplasm; the immature virion so formed begins condensation and is transported to the Golgi apparatus where glycosylation of E occurs and viral maturation takes place. Virions are found within the lumen of the Golgi membranes, but do not appear to accumulate as aggregates within them. A definitive account of flavivirus morphogenesis must describe how

FLAVIVIRUS REPLICATION

79

the genomic RNA becomes associated with the C protein, and how or where membrane sites containing E and prM or M are separated from other domains containing the hydrophobic ns proteins during the process of assembly. If assembly or condensation does occur rapidly, it may be necessary to delay or arrest this process by use of appropriate inhibitors so that the stages of morphogenesis can be captured in electron micrographs.

J . Persistent Infection, ts Mutants, and Defective Interfering Particles In general, acute flavivirus infections in vertebrate cells are progressively and rapidly cytocidal, whereas similar infections in mosquito cells are often persistent and cytopathic effects are variable. The outcome in cell cultures is similar to that in the respective natural hosts, but infections in mosquito cell cultures have some unique features. Thus a single passage of flaviviruses in Aedes albopictus cells is sufficient to produce phenotypic changes. For example, no slow-sedimenting hemagglutinin was released from DEN-2 virus-infected cells and the antigenic characteristics of the virion were altered (Sinarachatanant and Olson, 1973). Similarly in KUN virus-infected A . albopictus cells no significant hemagglutinin activity could be detected in virions or subviral particles, even though the yields were lo8 PFU/ml or greater, whereas normal ratios of hemagglutinating units to PFU were produced in J E and DEN-2 virus acute infections (Ng and Westaway, 1983). The phenotypic changes in KUN virions including increased fragility during sedimentation were not produced in virions grown in the C6/36 clone of A . albopictus cells, and were reversed by a single subsequent passage in Vero cells. The phenotypic changes induced during flavivirus replication in A. albopictus cells vary and hence are selective but appear to be restricted to processing of the envelope protein. For example, E from virions of DEN-2 virus grown in C6/36 cells had increased electrophoretic mobility relative to E of virions from Vero cells (Smith and Wright, 1985), as had E of KUN virus grown in the parental A . albopictus cell line (Ng and Westaway, 1983). The changes are subtle because in both these cases they were not detected in infected cytoplasm, and their elucidation awaits a comparison of virus maturation in vertebrate versus mosquito cells. When cytopathic effects are observed in flavivirus-infected mosquito cells, the most common and prominent symptom is formation of syncytia, especially in the A . albopictus cell line (Paul et al., 1969). Thus cycles of syncytia formation, disintegration, and cell recovery were observed in persistent infections at 28°C with JE, KUN, and MVE viruses; the virus yields varied but were as high as those in acute

80

E. G. WESTAWAY

infections (107 to lo8 PFUlml) as late as 60-80 days (Ng and Westaway, 1980). KUN ts mutants, identified initially as small plaque formers, appeared within 16 days and by 2 months represented 50%of the yield from persistently infected A . albopictus cells. Fluctuating yields were associated also with persistent infections of C6/36 cells by the four dengue viruses, and over a period of some months the persistent virus population became temperature sensitive (ts);these spontaneous ts mutants interfered with replication in C6/36 cells of the four wild-type dengue viruses, and except for DEN-1 were not able to kill suckling mice (Igarashi, 1979). Interference also occurred in C6/36 cells between DEN-1 and DEN-3 viruses in simultaneous (acute) or persistent infections, or in superinfection with either virus of cultures infected 2 weeks previously with any heterotypic dengue virus (Dittmar et al., 1982). Immunospecific staining of cells simultaneously infected with DEN-1 and DEN-3 viruses at low multiplicity showed that only one or other virus replicated in 99% of infected cells, and this interference obviously did not involve ts mutants or competition for receptor sites on cells. An inhibitor (probably the structural protein M) released from A. albopictus cells persistently infected with B a n i virus reduced the yield from pretreated cells in acute infections by 10,000-fold but had no effect on J E virus yield (Lee and Schloemer, 1981a,b). However, no antiviral activity was detected in culture fluids of A. albopictus cells persistently infected with KUN virus (Newton and Dalgarno, 1983). Because high infectious titers were often maintained in persistent infections with J E , KUN, and some other flaviviruses, it seems unlikely that defective interfering (DI) flavivirus particles are generated in mosquito cells. In some situations ts mutants are the obvious agents of interference, but the mechanism in these and other instances remains unknown. Despite the earlier general comment that flavivirus infections of vertebrate cells are cytocidal, persistent infections in mammalian cells have been established in several laboratories (summarized by Brinton et at., 1985). In J E virus persistent infections of rabbit kidney and Vero cell lines, Schmaljohn and Blair (1977, 1979) found only a low percentage of fluorescent antibody-positive cells and infectious centers. Maintenance of persistent infection did not appear to be caused by induced interferon production or by generation of ts virus mutants. Although the ratio of noninfectious to infectious J E virus particles was greater in persistent than in acute infections, the sedimentation profiles of intracellular virus-specified RNA labeled with L3H1uridine appeared unchanged, so no direct evidence of defective interfering particles containing RNA deletions was obtained. Brinton (1982) re-

FLAVIVIRUS REPLICATION

81

ported that persistent WN virus infections in genetically resistant and susceptible transformed mouse cells generated defective nonplaquing virus particles with low interfering activity by 16 weeks. Enhanced amplification of DI particles was favored in resistant cultures, and occurred even in acute infections at multiplicities greater than one; the DI particles were denser than wild-type WN virus and contained heterogeneous RNA smaller than the normal genome (Brinton, 1983). In contrast, plaquing WN virus continued to be produced and with increased yields in mouse cells transferred from 37 to 32°C after the sixth weekly subculture; the majority of progeny virus at 32°C from resistant cells became ts with continued passages, whereas ts mutants represented only a small proportion of the virus population from susceptible cells (Brinton et al., 1985). Similarly, only three ts plaques were found among over 300 plaques derived from culture fluids of J E virus persistent infections (Schmaljohn and Blair, 1977). Genetic analyses of flaviviruses have been hindered by the limited availability and properties of suites of ts mutants. From DEN-2 virus stocks passaged 10 times at 28"C, five stable ts mutants with reduced mouse lethality were induced by replication in hamster kidney cells treated with 5-azacytidine, and four of these were from three separate complementation groups (two RNA+ and one RNA-); however, characterization was difficult because of very low titers at the permissive temperature of 33.5"C, which reduced input multiplicity to less than one infectious particle per cell (Tarr and Lubiniecki, 1976). Spontaneous ts mutants of KUN virus from persistently infected mosquito cells had titers close to lo6 PFU/ml with low reversion rates but were all RNA - and complementation could not be established, possibly because of multiple lesions (Ng and Westaway, 1980, and unpublished results). Of nine ts mutants of SLE virus induced in infected PS-2 cells treated with 5-azacytidine or 5-fluorouracil, six were placed in four complementation groups (two RNA+ and two RNA-1, but some inconsistencies were apparent (Hollingshead et al., 1983). The same chemical mutagens induced ts mutants of J E virus in PK(15) cells and seven complementation groups were defined, three with RNA - phenotype and four RNA+ (Eastman and Blair, 1985). Three RNA+ groups produced normal amounts of J E viral proteins at the nonpermissive temperature and may represent the genes for the three structural proteins. Although several of the complementation indices were low and no infectious virus was produced in a mixed infection with mutants of one RNA+ and one RNA- group, the J E virus mutants represent a promising start in exploring the genetics of this difficult family of RNA viruses. Virus interference was not an uncommon observation in many of the

E. G. WESTAWAY

experiments involving persistent infections, ts mutants, or complementation. The cause or agent was not always identified, but a number of possibilities exist because of the size of the single-stranded RNA genome and the length of the single open reading frame greater than lo4 nucleotides. Indeed, it is surprising that so few errors in copying apparently do occur, in view of the estimated error rate for to (Reanney, 1982). Even in persistent RNA replicases of infections of mosquito cells in which spontaneous ts mutants are most likely to arise, such infections do not appear to be involved in excessive production of DI particles. These ts mutants are probably better adapted to growth at lower temperatures of incubation, at which their interference effect with wild-type virus is not readily evident. AND CONCLUSIONS IV. SUMMARY

The Flaviviridae comprise 65 species which globally represent the most widespread and medically important of the arthropod-borne viruses. Knowledge of flavivirus replication strategy has accumulated slowly; current information indicates that the strategy is apparently simple, involving the genome as the single capped messenger with one open reading frame of 10,233 nucleotides (for YF virus), and no involvement of the cell nucleus. About 380,000 Da of protein are encoded, of which the structural proteins represent close to 22%. However, the detailed analysis of the strategy still poses many challenges and problems, compounded by the absence of switch-off for cell macromolecular synthesis, by the long latent periods and relatively low yields of virions, RNA, and several of the gene products, and by the restricted expression of the genome in cell-free translation experiments. The broad outlines of replication may be summarized as follows. A wide variety of cells of vertebrates and some arthropods serve as hosts. Virions are taken up in coated pits and uncoating occurs within a few minutes. Then follows a latent period of 12-16 hours during which viral RNA and protein synthesis commences but in barely detectable amounts, and the membranes of the endoplasmic reticulum undergo the beginning of profound rearrangements. RNA is synthesized asymmetrically in semiconservative fashion with a fully double-stranded replicative form serving as a recycling template on which only one nascent strand is synthesized and released after a delay totaling about 20 minutes. The RNA polymerase probably comprises at least the two largest ns proteins NV4 and NV5. After the latent period mainly RNA of positive polarity is synthesized and both RNA and protein synthesis

FLAVIVIRUS REPLICATION

83

increase to readily detectable amounts if radioactive labels are used in substrates. Nucleotide sequence data show that translation is initiated close to the 5' end of the genome and the sequence of translation is NH2-C-prM-E-NV3...NV4 ...NV5-COOH. The three ns proteins shown are the largest and have been designated NS1, NS3, and NS5, respectively, by Rice et al. (1985);the precise location and identification of the intervening genes for four(?) smaller ns proteins are indeterminate and their function is unknown. The amino termini of the glycoproteins prM, E, and NV3 are preceded by signal sequences. Translation appears to be associated with heavy membranes of the rough endoplasmic reticulum, and the mainly hydrophobic products remain bound to these membranes. Some cleavage sites have been identified but evidence of production of precursor polyproteins is scanty. Glycosylation occurs by N-linkage using normal cell processes, although no information is available on involvement of the Golgi. The sites of RNA synthesis appear to be membrane associated in foci mainly in the perinuclear region but have not been characterized. Virions accumulate predominantly within cisternae of virus-induced membranes. The process of assembly and maturation is not understood; no nucleocapsids have been identified in cells, and morphogenesis may occur by a condensation process rather than by budding, which has not been unequivocally demonstrated. Protein prM is incorporated in intracellular virions and is cleaved into M and an unidentified residue as virions leave the cell, mainly by exocytosis. Production of DI particles occurs rarely, and a subviral hemagglutinating particle lacking RNA and core protein is released from vertebrate cells. Replication in mosquito cells is associated with changes in hemagglutinating particles, and persistent infections are readily established during which ts mutants arise spontaneously. Among the unusual features of flavivirus replication awaiting elucidation are the long latent period, the variable production and the function of small ns proteins, the anomalous translation of the most 3' distal and largest product NV5 in translation mapping experiments, the inconsistencies in carbohydrate content of envelope protein, and the apparent absence of nucleocapsids and budding in virus morphogenesis. Some of these problems may be solved as more amino acid and nucleotide sequence data are acquired, and this information may lead to the development of genetically engineered and safe flavivirus vaccines, or may provide a rationale for antiviral chemotherapy.. Meanwhile, the events programmed by the flavivirus genome, the largest messenger fully expressed in eukaryotic cells, and the responses of host cells will continue to intrigue and challenge the growing band of flavivirus researchers.

84

E. G. WESTAWAY

ACKNOWLEDGMENTS I wish to thank the several investigators who supplied copies of papers in press, and my colleague Peter Wright for his valuable contributions over many years. The work of myself and colleagues was supported by grants from the National Health and Medical Research Council of Australia.

REFERENCES Baltimore, D. (1968). Structure of the poliovirus replicative intermediate RNA. J. Mol. Biol. 32, 359-368. Bhatt, P. N., Johnson, E. A., Smith, A. L., and Jacoby, R. 0.(1981). Genetic resistance to lethal flaviviral encephalitis. 111. Replication of Banzi virus in uiuo and in uitro in tissues of congenic susceptible and resistant mice. Arch. Virol. 69, 273-286. Bienz, K., Egger, D., Rasser. Y., and Bossart, W. (1983). Intracellular distribution of poliovirus proteins and the induction of virus-specific cytoplasmic structures. Virology 131, 39-48. Blok, J. (1985). Genetic relationships of the dengue virus serotypes. J . Gen. Virol. 66, 1323-1325. Blok, J., Henchal, E. A., and Gorman, B. M. (1984).Comparison of dengue viruses and some other flaviviruses by cDNA-RNA hybridization analysis and detection of a close relationship between dengue virus serotype 2 and Edge Hill virus. J . Gen. Virol. 65,2171-2181. Boulton, R. W., and Westaway, E. G. (1972). Comparisons of togaviruses: Sindbis virus (Group A) and Kunjin virus (Group B). Virology 49, 283-289. Boulton, R. W., and Westaway, E. G. (1976). Replication of the flavivirus Kunjin: Proteins, glycoproteins and maturation associated with cell membranes. Virology 69, 416-430. Boulton, R. W., and Westaway, E. G. (1977). Togavirus RNA: Reversible effect of urea on genomes and absence of subgenomic RNA in Kunjin virus-infected cells. Arch. Virol. 55, 201-208. Brinton, M. A. (1982). Characterization of West Nile virus persistent infections in genetically resistant and susceptible mouse cells. I. Generation of defective, non-plaquing virus particles. Virology 116, 84-98. Brinton, M. A. (1983). Analysis of extracellular West Nile virus particles produced by cell cultures from genetically resistant and susceptible mice indicates enhanced amplification of defective interfering particles by resistant cultures. J. Virol. 46, 860870. Brinton, M. A. (1986). Replication of flaviviruses. In “The Togaviridae and the Flaviviridae” (S. Schlesinger and M. J. Schlesinger, eds.), pp. 327-374. Plenum, New York. Brinton, M. A., Davis, J., and Schaeffer, D. (1985).Characterization of West Nile virus persistent infections in genetically resistant and susceptible mouse cells. Virology 140, 152-158. Brown, F. (1986). Classification and nomenclature of viruses. Fifth Report of the International Committee on Taxonomy of Viruses. Karger, Base1 (in press). Calberg-Bacq, C.-M., Rentier-Delrue, F., Osterreith, P. M., and Duchesne, P. Y. (1975). Electron microscopy studies on Banzi virus particle and its development in the suckling mice brains. J . Ultrustruct. Res. 53, 193-203. Cardiff, R. D., and Lund, J. K. (1976). Distribution of dengue-2 antigens by electron microscopy. Infect. Zmmun.13, 1699-1709. Cardiff, R. D., McCloud, T. G., Brandt, W. E., and Russell, P. K. (1970). Molecular size

FLAVIVIRUS REPLICATION

85

and charge relationships of the soluble complement-fixing antigens of dengue viruses. Virology 41, 569-572. Cardiff, R. D., Russ, S. B., Brandt, W. E., and Russell, P. K. (1973). Cytological localization of dengue-2 antigens: An immunological study with ultrastructural correlation. Infect. Immun. 7 , 809-816. Castle, E., Nowak, T., Leidner, U., Wengler, G., and Wengler, G. (1985). Sequence analysis of the viral core protein and the membrane-associated proteins V1 and NV2 of the flavivirus West Nile virus and of the genome sequence for these proteins. Virology 145, 227-236. Castle, E., Leidner, V., Nowak, T., Wengler, G., and Wengler, G. (1986). Primary structure of the West Nile flavivirus genome region coding for all nonstructural proteins. Virology 149, 10-26. Chu, P. W. G., and Westaway, E. G. (1985). Replication strategy of Kunjin virus: Evidence for recycling role of replicative form RNA as template in semiconservative and asymmetric replication. Virology 140, 68-79. Chu, P. W. G., and Westaway, E. G. (1986). Characterization of Kunjin virus RNAdependent RNA polymerase: Reinitiation of synthesis in uitro. Virology 155, in press. Chu, P. W. G., and Westaway, E. G. (1987). In preparation. Cleaves, G. R. (1985). Identification of type 2 dengue virus-specific high molecular weight proteins in virus-infected BHK cells. J . Gen. Virol. 66, 2767-2771. Cleaves, G. R., and Dubin, D. T. (1979). Methylation status of intracellular dengue type 2 40s RNA. Virology 96, 159-165. Cleaves, G. R., Ryan, T. E., and Schlesinger, R. W. (1981). Identification and characterization of type 2 dengue virus replicative intermediate and replicative form RNAs. Virology 111, 73-83. Clewley, J. P., and Kennedy, S. I. T. (1976). Purification and polypeptide composition of Semliki Forest virus RNA polymerase. J . Gen. Virol. 32,395-411. Dalgarno, L., Trent, D. W., Strauss, J. H., and Rice, C. M. (1986). Partial nucleotide sequence of the Murray Valley encephalitis virus genome: Comparison of the encoded polypeptides with yellow fever virus structural and nonstructural proteins. J . Mol. Biol. 187, 309-323. Darnell, M. B., and Koprowski, H. (1974). Genetically determined resistance to infection with group B arboviruses. 11. Increased production of interfering particles in cell cultures from resistant mice. J. Infect. Dis. 129, 248-256. Della-Porta, A. J., and Westaway, E. G. (1972).Rapid preparation of haemagglutinins of togaviruses from infected cell culture fluids. Appl. Microbiol. 23, 158-160. Della-Porta, A. J., and Westaway, E. G. (1977).Immune response in rabbits to virion and nonvirion antigens of the flavivirus Kunjin. Infect. Immun. 15, 874-882. Deubel, V., Crouset, J., Benichou, D., Digoutte, J.-P., Bouloy, M., and Girard, M. (1983). Preliminary characterization of the ribonucleic acid of yellow fever virus. Ann. Virol. 134E, 581-588. Dittmar, D., Castro, A., and Haines, H. (1982). Demonstration of interference between dengue virus types in cultured mosquito cells using monoclonal antibody probes. J. Gen. Virol. 59, 273-282. Dorner, A. J., Semler, B. L., Jackson, R. L., Hanecak, R., Duprey, E., and Wimmer, E. (1984).In uitro translation of poliovirus RNA: Utilization of internal initiation sites in reticulocyte lysate. J . Virol. 50, 507-514. Eastmann, P. S., and Blair, C. D. (1985). Temperature-sensitive mutants of Japanese encephalitis virus. J . Virol. 55, 611-616. Friedman, R. M., Levin, J. G., Grimley, P. M., and Berezesky, I. K. (1972). Membraneassociated replication complex in arborvirus infection. J. Virol. 10, 504-515.

86

E. G. WESTAWAY

Fuller, F. J., and Marcus, P. I. (1980).Sindbis virus. I. Gene order of translation in uiuo. Virology 107, 441-451. Garoff, H., Kondor-Koch, C., and Riedel, H. (1982). Structure and assembly of alphaviruses. Curr. Top. Microbiol. Zmmunol. 99, 1-50. Gentry, M. K., Henchal, E. A., McCown, J. M., Brandt, W. E., and Dalrymple, J . M. (1982). Identification of distinct antigenic determinants on dengue-2 virus using tranoclonal antibodies. Am. J. Trop. Med. Hyg. 31, 548-555. Gollins, S. W., and Porterfield, J. S. (1984).Flavivirus infection enchancement in macrophages: Radioactive and biological studies on the effect of antibody on viral fate. J . Gen. Virol. 65, 1261-1272. Gollins, S. W., and Porterfield, J. S. (1985). Flavivirus infection enchancement in macrophages: An electron microscopic study of viral cellular entry. J . Gen. Virol. 66, 1969-1982. Gould, E. A., Chanas, A. C., Buckley, A., and Clegg, C. S. (1983). Monoclonal immunoglobulin M antibody to Japanese encephalitis virus that can react with a nuclear antigen in mammalian cells. Infect. Zmmun. 41, 744-779. Grange, T., Bouloy, M., and Girard, M. (1985). Stable secondary structure at the 3’ end of the genome of yellow fever virus (17D) vaccine strain). FEBS Lett. 188, 159-163. Heinz, F. X., and Kunz, Ch. (1979). Protease treatment and chemical cross linking of a flavivirus: Tick-borne encephalitis virus. Arch. Virol. 60, 207-216. Heinz, F. X., and Kunz. Ch. (1980).Formation of polymeric glycoprotein complexes from a flavivirus: Tick-borne encephalitis virus. J . Gen. Virol. 49, 125-132. Heinz, F. X., and Kunz, Ch. (1982). Molecular epidemiology of tick-borne encephalitis virus: Peptide mapping of large nonstructural proteins of European isolates and comparisons with other flaviviruses. J. Gen. Virol. 62, 271-285. Heinz, F. X., Berger, R., Tuma, W., and Kunz, Ch. (1983).A topological and functional model of epitopes on the structural glycoprotein of tick-borne encephalitis virus defined by monoclonal antibodies. Virology 126, 525-537. Heinz, F. X., Tuma, W., Guirakhoo, F., Berger, R., and Kunz, Ch. (1984). Immunogenicity of tick-borne encephalitis virus glycoprotein fragments: Epitope-specific analysis of the antibody response. J . Gen. Virol. 65, 1921-1929. Hollingshead, Ph. G., Jr., Brawner, Th.A., and Fleming, T. P. (1983).St. Louis encephalitis virus temperature-sensitive mutants. Arch. Virol. 75, 171-179. Igarashi, A. (1979). Characteristics of Aedes albopictus cells persistently infected with dengue viruses. Nature (London) 280, 690-691. Kamer, G., and Argos, P. (1984). Primary structural comparison of RNA-dependent polymerases from plant, animal and bacterial viruses. Nucleic Acids Res. 12, 72697282. Kimura-Kuroda, J., and Yasui, K. (1983).Topographical analysis of antigenic determinants on envelope glycoprotein V3(E) of Japanese encephalitis virus, using monoclonal antibodies. J. Virol. 45, 124-132. Koch, G., Koch, F., Bilello, J. A., Hiller, E., Scharli, C., Warnecke, G., and Weber, C. (1982). Biosynthesis, modification and processing of viral polyproteins. Zn “Protein Biosynthesis in Eukaryotes” (R. PBrez-Bercoff, ed.), pp. 275-309. Plenum, New York. Kos, K. A., Shapiro, D., Vaituzis, Z., and Russell, P. K. (1975).Viral polypeptide composition of Japanese encephalitis virus-infected cell membranes. Arch. Virol. 47, 217-224. Kozak, M. (1984). Compilation and analysis of sequences upstream from the translational start site in eukaryotic mRNAs. Nucleic Acids Res. 12, 857-872. Leary, K., and Blair, C. D. (1980). Sequential events in the morphogenesis of Japanese encephalitis virus. J. Ultrastruct.Res. 72, 123-129.

FLAVIVIRUS REPLICATION

87

Leary, K., and Blair, C. D. (1983).Japanese encephalitis virus replication: Studies on host cell nuclear involvement. Exp. Mol. Pathol. 38, 264-270. Lee, C.-H., and Schloemer, R. H. (1981a).Mosquito cells infected with Banzi virus secrete an antiviral activity which is of viral origin. Virology 110, 402-410. Lee, C.-H., and Schloemer, R. H. (1981b).Identification of the antiviral factor in culture medium of mosquito cells persistently infected with Banzi virus. Virology 110,445454. McClain, K., Stewart, M., Sullivan, M., and Maizel, J. V., Jnr. (1981).Ribosomal binding sites on poliovirus RNA. Virology 113, 150-167. McClure, M. A., and Perrault, J. (1985).Poliovirus genome RNA hybridizes specifically to higher eukaryotic rRNAs. Nucleic Acids Res. 13,6797-6816. Mayne, J. T., Bell, J. R., Strauss, E. G., and Strauss, J. H. (1985).Pattern of glycosylation of Sindbis virus envelope proteins synthesized in hamster and chicken cells. Virology 142, 121-133. Monath, T.P., Kinney, R. M., Schlesinger, J. J., Brandriss, M. W., and Brhs, P. (1983). Ontogeny of yellow fever 17D vaccine: RNA oligonucleotide fingerprint and monoclonal antibody analyses of vaccines produced worldwide. J . Gen. Virol. 64,627-637. Monckton, R. P., and Westaway, E. G. (1982).Restricted translation of the genome of the flavivirus Kunjin in uitro. J . Gen. Virol. 63, 227-232. Murphy, F. A. (1980).Morphology and morphogenesis. In “St. Louis Encephalitis” (T. Monath, ed.), pp. 65-193. Am. Public Health Assoc., Washington, D.C. Naeve, C. W., and Trent, D. W. (1978).Identification of Saint Louis encephalitis virus mRNA. J. Virol. 25,535-545. Newton, S. E., and Dalgarno, L. (1983).Antiviral activity released from Aedes albopictus cells persistently infected with Semliki Forest virus. J . Virol. 47, 652-655. Ng, M.L., and Westaway, E. G. (1980).Establishment of persistent infections by flaviviruses in Aedes albopictus cells. In “Invertebrate Systems in Vitro” (E. Kurstak, K. Maramorosch, and D. Durbendorfer, eds.), pp. 389-402. Elsevier, Amsterdam. Ng, M. L., and Westaway, E. G. (1983).Phenotypic changes in the flavivirus Kunjin after a single cycle of growth in an Aedes albopictus cell line. J . Gen. Virol.64,17151723. Ng, - M.L., Pedersen, J. S., Toh, B. H., and Westaway, E. G. (1983).Immunofluorescent sites in Vero cells infected with the flavivirus Kunjin. Arch. Virol. 78, 177-190. Ozden, S.,and Poirier, B. (1985).Dengue virus induced polypeptide synthesis. Arch. Virol. 85, 129-137. Paul, S. D., Singh, K. R. P., and Bhat, U. K. M. (1969).A study on the cytopathic effect of arboviruses on cultures from Aedes albopictus cell line. Indian J.Med. Res. 57,339348. Peiris, J. S. M., Porterfield, J. S., and Roehrig, J. T. (1982).Monoclonal antibodies against the flavivirus West Nile. J . Gen. Virol. 58, 283-289. Perlman, D.,and Halvorson, H. 0. (1983).A putative signal peptidase recognition site and sequence in eukaryotic and prokaryotic signal peptides. J. Mol. Biol. 167,391409. Pollack, L., and Atkinson, P. H. (1983).Correlation of glycosylation forms with position in amino acid sequence. J. Cell Biol. 97,293-300. Qureshi, A. A,, and Trent, D. W. (1972).Saint Louis encephalitis viral ribonucleic acid replication complex. J. Virol. 9,565-573. Reanney, D. C. (1982).The evolution of RNA viruses. Annu. Rev.Microbiol. 36,47-73. Repik, P. M., Dalrymple, J. M., Brandt, W. E., McCown, J. M., and Russell, P. K. (1983). RNA fingerprinting as a method for distinguishing dengue 1 virus strains. A m . J . Trop. Med. Hyg. 32, 577-589.

E. G. WESTAWAY Rice, C. M., Lenches, E. M., Eddy, S. R., Shin, S. J., Sheets, R. L., and Strauss, J. H. (1985). Nucleotide sequence of yellow fever virus: Implications for flavivirus gene expression and evolution. Science 229, 726-733. Rice, C. M., Strauss, E. G., and Strauss, J . H. (1986). Structure of the flavivirus genome. In “The Togaviridae and the Flaviviridae” (S. Schlesinger and M. J. Schlesinger, eds.), pp. 279-326. Plenum, New York. Roehrig, J. T., Mathews, J. H., and Trent, D. W. (1983). Identification of epitopes on the E glycoprotein of Saint Louis encephalitis virus using monoclonal antibodies. Virology 128, 118-126. Saborio, J. L., Pong, S.-S., and Koch, G. (1974). Selective and reversible inhibition of initiation of protein synthesis in mammalian cells. J. Mol. Biol. 86, 195-211. Schlesinger, R. W. (1977). Dengue viruses. Virol. Monogr. 16. Schlesinger, J. J., Brandriss, M. W., and Monath, T. P. (1983). Monoclonal antibodies distinguish between viral and vaccine strains of yellow fever virus by neutralization, hemagglutination inhibition, and immune precipitation of the virus envelope protein. Virology 125, 8-17. Schlesinger, J. J., Walsh, E. E., and Brandriss, M. W. (1984). Analysis of 17D yellow fever virus envelope protein epitopes using monoclonal antibodies. J. Gen. Virol. 65, 1637-1644. Schmaljohn, C., and Blair, C. D. (1977). Persistent infection of cultured mammalian cells by Japanese encephalitis virus. J . Virol. 24, 580-589. Schmaljohn, C., and Blair, C. D. (1979). Clonal analysis of mammalian cell cultures persistently infected with Japanese encephalitis virus. J . Virol. 31, 816-822. Shapiro, D., Brandt, W. E., Cardiff, R. D., and Russell, P. K. (1971). The proteins of Japanese encephalitis virus. Virology 44, 108-124. Shapiro, D., Brandt, W. E., and Russell, P. K. (1972a). Change involving a viral membrane glycoprotein during morphogenesis of group B arboviruses. Virology 50,906911. Shapiro, D., Kos, K., Brandt, W. E., and Russell, P. K. (1972b). Membrane-bound proteins of Japanese encephalitis virus-infected chick embryo cells. Virology 48, 360372. Shapiro, D., Kos, K. A., and Russell, P. K. (1973a).Japanese encephalitis virus glycoproteins. Virology 56, 88-94. Shapiro, D., Kos, K. A., and Russell, P. K. (1973b). Protein synthesis in Japanese encephalitis virus-infected cells. Virology 56, 95-109. Sinarachatanant, P., and Olson, L. C. (1973). Replication of dengue virus type 2 in Aedes albopictus cell culture. J . Virol. 12, 275-283. Smith, G. W., and Wright, P. J. (1985). Synthesis of proteins and glycoproteins in dengue type 2 virus-infected Vero and Aedes albopictus cells. J. Gen. Virol. 66, 559-571. Smith, T. J., Brandt, W. E., Swanson, J. L., McCown, J. M., and Buescher, E. L. (1970). Physical and biological properties of dengue-2 virus and associated antigens. J. Virol. 5, 524-532. Stohlman, S. A., Wisseman, C. L., Jr., and Eylar, 0. R. (1978). Dengue viral antigens in host cell membranes. Actu Virol. 22, 31-36. Stollar, V. (1969). Studies on the nature of dengue viruses. IV. The structural proteins of type 2 dengue virus. Virology 39, 426-438. Stollar, V., Schlesinger, R. W., and Stevens, T. M. (1967). Studies on the nature of dengue viruses. 111. RNA synthesis in cells infected with type 2 dengue virus. Virology 33,650-658. Stollar, V., Stollar, B. D., Koo, R., Harrap, K. A., and Schlesinger, R. W. (1976). Sialic acid contents of Sindbis virus from vertebrate and mosquito cells. Virology 69, 104115.

FLAVIVIRUS REPLICATION

89

Strauss, E. G., and Strauss, J. H. (1983). Replication strategies of the single-stranded RNA viruses of eukaryotes. Cum. Top. Microbiol. Immunol. 105, 1-98. Svitkin, Y. V., Ugarova, T. Y., Chernovskaya, T. V., Lyapustin, V. M., Lashkevich, V. A., and Agol, V. I. (1981). Translation of tick-borne encephalitis virus (flavivirus) genome in uitro: Synthesis of two structural polypeptides. Virology 110, 26-34. Svitkin, Y. V., Lyapustin, V. N., Lashkevich, V. A., and Agol, V. I. (1984). Differences between translation products of tick-borne encephalitis virus RNA in cell-free systems from Krebs-2 cells and rabbit reticulocytes: Involvement of membranes in the processing of nascent precursors of flavivirus structural proteins. Virology 135,536541. Tarr, G. C., and Lubiniecki, A. S. (1976). Chemically-induced temperature sensitive mutants of dengue virus type 2. Arch. Virol. 50, 223-235. Theiler, M., and Downs, W. G . (1973). “The Arthropod-Borne Viruses of Vertebrates.” Yale Univ. Press, New Haven. Trent, D. W., and Naeve, C. W. (1980). Biochemistry and replication. In “St. Louis encephalitis” (T. Monath, ed.), pp. 159-199. Am. Public Health Assoc., Washington, D.C. Trent, D. W., Grant, J . A., Vorndam, V., and Monath, T. P. (1981). Genetic heterogeneity among St. Louis encephalitis virus isolates of different geographic origin. Virology 114, 319-322. Trent, D. W., Grant, J. A., Rosen, L., and Monath, T. P. (1983). Genetic variation among dengue-2 viruses of different geographic origin. Virology 128, 271-284. Vazquez, D., Zaera, E., Dolz, H., and Jimbnez, A. (1982). Action of inhibitors of protein biosynthesis. In “Protein Biosynthesis in Eukaryotes” (R.Perez-Bercoff, ed.), pp. 311-337. Plenum, New York. Vezza, A. C., Rosen, L., Repik, P., Dalrymple, J., and Bishop, D. H. L. (1980). Characterization of the viral RNA species of prototype dengue viruses. Am. J . Trop. Med. Hyg. 29,643-652. Von Heijne, G. (1984). How signal sequences maintain cleavage specificity. J . Mol. Biol. 173, 243-251. Walker, P. J., Garrett, S.T., Gorman, B. M., and Burke, D. S.(1982). Genetic analysis of strain variation in dengue type 2 viruses. In “Viral diseases in South East Asia and the Western Pacific” (J. S. McKenzie, ed.), pp. 513-516. Academic Press, Sydney. Wengler, G., and Wengler, G. (1981).Terminal sequences of the genome and replicative form RNA of the flavivirus West Nile virus: Absence of poly(A) and possible role in RNA replication. Virology 113, 544-555. Wengler, G., Wengler, G., and Gross, H. J. (1978). Studies on virus-specific nucleic acids synthesized in vertebrate and mosquito cells infected with flaviviruses. Virology 89, 423-437. Wengler, G., Beato, M., and Wengler, G. (1979).In uitro translation of 42s virus-specific RNA from cells infected with the flavivirus West Nile virus. Virology 96, 516-529. Wengler, G., Castle, E., Leidner, U., Nowak, T., and Wengler, G. (1985). Sequence analysis of th? membrane protein V3 of the flavivirus West Nile virus and of its gene. Virology 147, 264-274. Westaway, E. G. (1966). Assessment and application of a cell line from pig kidney for plaque assay and neutralization tests with twelve group B arboviruses. Am. J . Epidemiol. 84, 439-456. Westaway, E. G. (1973). Proteins specified by group B togaviruses in mammalian cells during productive infections. Virology 51, 454-465. Westaway, E. G. (1975).The proteins of Murray Valley encephalitis virus. J . Gen. Virol. 27,283-292. Westaway, E. G. (1977). Strategy of the flavivirus genome: Evidence for multiple inter-

90

E. G. WESTAWAY

nal initiation of translation of proteins specified by Kunjin virus in mammalian cells. Virology 80, 320-335. Westaway, E. G. (1980).Replication of flaviviruses. In “The Togaviruses” (R. W. Schlesinger, ed.), pp. 531-581. Academic Press, New York., Westaway, E. G., and Ng, M, L. (1980).Replication of flaviviruses: Separation of translation sites of Kunjin virus proteins and of cell proteins. Virology 106, 107-122. Westaway, E.G., and Reedman, B. M. (1969).Proteins of the group B arbovirus Kunjin. J . Virol. 4,688-693. Westaway, E. G., and Shew, M. (1977).Proteins and glycoproteins specified by the flavivirus Kunjin. Virology 80, 309-319. Westaway, E. G.,McKimm, J. L., and McLeod, L. G. (1977).Heterogeneity among flavivirus proteins separated in slab gels. Arch. Virol. 53, 305-312. Westaway, E. G.,Schlesinger, R. W., Dalrymple, J. M., and Trent, D. W. (1980).Nomenclature of flavivirus-specified proteins. Intervirology 14, 114- 117. Westaway, E. G., Speight, G., and Endo, L. (1984).Gene order of translation of the flavivirus Kunjin: Further evidence of internal initiation in vivo. Virus Res. 1, 333350. Westaway, E. G.,Brinton, M. A., Gaidamovich, S.Ya., Horzinek, M. C., Igarashi, A., Kaiiriainen, L., Lvov, D. K., Porterfield, J. S., Russell, P. K., and Trent, D. W. (1985). Flaviviridae. Intervirology 24, 183-192. Wright, P. J. (1982).Envelope protein of the flavivirus Kunjin is apparently not glycosylated. J. Gen. Virol. 59,29-38. Wright, P. J., and Warr, H. M. (1985).Peptide mapping of envelope-related glycoproteins specified by the flaviviruses Kunjin and West Nile. J. Gen. Virol. 66, 597-601. Wright, P.J.,and Westaway, E. G. (1977).Comparisons of the peptide maps of Kunjin virus proteins smaller than the envelope protein. J. Virol. 24, 662-672. Wright, P. J., Bowden, D. S., and Westaway, E.G. (1977).lJnique peptide maps of the three largest proteins specified by the flavivirus Kunjin. J. Virol. 24,651-661. Wright, P.J.,Warr, H. M., and Westaway, E. G. (1980).Preliminary characterization of glycopeptides derived from glycoproteins specified by the flavivirus Kunjin. Virology 104,482-486. Wright, P. J., Warr, H. M., and Westaway, E. G. (1981).Synthesis of glycoproteins in cells infected by the flavivirus Kunjin. Virology 109, 418-427. Wright, P. J., Warr, H. M., and Westaway, E. G. (1983).Comparisons by peptide mapping of proteins specified by Kunjin, West Nile and Murray Valley encephalitis viruses. Aust. J. Exp. Biol. Med. Sci. 61,641-653.

ADVANCES IN VIRUS RESEARCH, VOL. 33

THE AUTONOMOUSLY REPLICATING PARVOWRUSES OF VERTEBRATES Susan F. Cotmore’ and Peter Taitersall*nt Departments of ‘Laboratory Medicine and tHurnan Genetics Yale University School of Medicine New Haven, Connecticut 06510

I. INTRODUCTION The parvoviruses are a large family of physically similar viruses which infect animals as diverse as man and moth (Berns, 1984; Siegl et al., 1985). Those which naturally infect vertebrates are divided into two genera on the basis of their requirement for helper viruses. Members of the adeno-associated virus (AAV) subgroup are defective and depend entirely upon adenovirus or herpesvirus coinfection for their replication (Berns, 1984; Siegl et al., 1985). In contrast, members of the autonomous parvovirus subgroup, which are listed in Table I, are capable of productive replication without the aid of a helper virus in the majority of host cells. Numerous studies over the past 25 years on the pathogenicity of. autonomous parvoviruses have shown that they are predominantly teratogenic agents. In general they cause fetal and neonatal abnormalities by destroying specific cell populations which are rapidly proliferating during the normal course of development (reviewed in Siegl, 1984a).These same tissues are usually resistant in the mature animal and, consequently, few of the viruses cause clinical disease in the adult. Animals are particularly sensitive to parvovirus infection in the first few days of life, and intracerebral inoculation of neonatal animals, especially hamsters, with these agents can cause runting and a characteristic “mongoloid-like” deformity (Toolan, 1960; Kilham, 1961). The craniofacial and periodontal lesions which generate the deformity appear to be due to selective viral attack on developing skeletal and dental tissues (Ferm and Kilham, 1965). Unlike Down’s syndrome in man, which it resembles, this condition is not associated with chromosomal abnormalities (Galton and Kilham, 1966) nor is it hereditary, since with careful husbandry such animals can be raised to breeding age and produce normal offspring (Toolan, 1978). Perhaps the most characteristic result of neonatal intracerebral parvovirus in91

Copyright 0 1987 by Academic Press, Inc. All rights of repduction in any form reserved.

92

SUSAN F. COTMORE AND PETER TATTERSALL TABLE I

PARVOVLRUSES AUTONOMOUS OF VERTEBRATESQ Virus Accepted members Rat virus H-1 virus (rat) RT virus (rat) TVX (unknown) Minute virus of mice LuIII virus (unknown) B19 (human) Porcine parvovirus Bovine parvovirus Feline parvovirus Species host range variants Mink enteritis virus Canine parvovirus Raccoon parvovirus Lapine parvovirus Aleutian disease virus (mink) Goose parvovirus Possible members Minute virus of canines HB virus (human?) RA-1 (human)

Acronym RV H- 1 RT TVX MVM LUIII B19 PPV BPV FPV MEV CPV RPV LPV ADV GPV MVC

HB RA-1

Where the virus name does not include it, the primary host species of each virus is indicated in parentheses. Assignments to the family Parvoviridae, genus Parvovirus, are taken from Siegl et al. (1985). Each member of the group is serotypically distinct, but often a number of variants with different pathogenic properties occur within the same host species. The species host range variants of FPV can be distinguished with monoclonal antibodies, but cross-neutralize with polyclonal sera (Parrish and Carmichael, 1983, 1986). (1

fection is cerebellar hypoplasia, often leading to chronic ataxia (Kilham and Margolis, 1975). This was also shown to be due to viral depletion of a rapidly proliferating cell population, in this case the cells of the cerebellar granular cortex (Margolis and Kilham, 1975). Many parvoviruses cross the placenta and establish infections of the

AUTONOMOUSLY REPLICATING PARVOVIRUSES

93

fetus. A number of factors affect the outcome of fetal infections, such as route of inoculation, virus strain and dose, species of pregnant host, and the time of infection during gestation. In some cases inoculation with a large dose of virus early in gestation give rise to a generalized and devastating lethal infection involving much of the mesodermal tissue of the embryo and resulting in resorption or mummification (Margolis and Kilham, 1975; Siegl, 1984a). Infection at late times in gestation may yield viable offspring which, in addition to the runting and cerebellar hypoplasia mentioned earlier, may also exhibit hemorrhagic encephalopathy or neonatal hepatitis and enteritis (Kilham and Margolis, 1975; Margolis and Kilham, 1975; Siegl, 1984a). Again, the involvement of cell populations with high mitotic activity at or subsequent to the time of infection is a consistent feature of these diseases. Factors such as host species, virus strain, and route of inoculation affect the course of disease mainly by determining whether the virus ever reaches the developing fetus, whereas the variability resulting from infection at different times during gestation suggests that different cell types in different embryonic tissues pass through a state of sensitivity and are sequentially, but transiently, at risk during fetal development. This risk period corresponds well with the time at which that particular cell type is undergoing rapid proliferation during the process of organogenesis. An absolute requirement for S-phase transition has been proposed as the basis for the specific tissue tropisms observed during parvovirus teratogenesis, and to explain the resistance of the adult animal as compared to the fetus or neonate (Margolis and Kilham, 1975). In support of this idea, several studies have shown that resistant adults can be rendered sensitive to disease by inducing some tissue to undergo an abnormal proliferative response. Thus partial hepatectomy prior to H-1 infection will render the normally resistant adult rat susceptible t o a form of viral hepatitis in which the sites of viral attack are localized in the regenerating margins of the liver (Ruffolo et ul., 1966). Susceptibility to a similar viral disease can also be brought about by inducing mitotic activity in the liver with carbon tetrachloride or by infection with the parasite Cysticercus fuscioluris (Kilham et ul., 1970; Kilham and Margolis, 1975). Osteolytic parvovirus strains will also infect healing bone fractures, causing defective callus formation in normally resistant adult hamsters (Baer et al., 1971). In addition to these studies where regenerative proliferation can be shown to provide a target tissue for viral attack in the adult, a number of reports suggest that these viruses can interfere with or suppress a second type of abnormal proliferation in their hosts, namely, neoplastic disease. The fact that most of the early parvovirus isolates were obtained

94

SUSAN F. COTMORE AND PETER TATTERSALL

from tumor tissues superfically suggested that these viruses might have some causal association with neoplastic disease in their hosts. Indeed, after the original isolations by Kilham of rat virus (RV) and by Toolan of H-1, attempts to isolate these viruses from nonmalignant tissues by the same procedures proved unsuccessful. Since that time, however, many studies on the long-term influence of infection on the host animal have shown the opposite conclusion to be true, that parvoviruses interfere with, and in some cases appreciably suppress, tumor formation in their hosts. In one long-term study (Toolan, 1967) involving large numbers of hamsters, those which had survived H-1 infection at birth without pathological effect had a fivefold lower spontaneous tumor rate than their uninfected siblings. Furthermore, the rate among animals exhibiting the mongoloid-like craniofacial lesions associated with neonatal H-1 infection was fivefold lower still. Infection with autonomous parvoviruses has also been shown directly to suppress tumor formation by a number of viruses and carcinogens. For instance, RV was shown to suppress leukemia induction by Moloney leukemia virus in rats (Bergs, 19691, and H-1 infection of hamsters has been shown to suppress tumor formation by both adenovirus (Toolan and Ledinko, 1968) and dimethylbenzanthracene (Toolan et al., 1982). The mechanism underlying parvoviral oncosuppression is not understood, and until recently there have been no in uitro systems available with which to dissect this phenomenon at the cellular level. Although proliferative activity appears to be a prerequisite for target organs, it is clear that not all tissues which turn over rapidly are necessarily subject to parvoviral attack. While most adult tissues are mitotically quiescent compared to the fetus and neonate, many, such as gut epithelium and the hematopoeitic system, contain large numbers of cycling cells. One might expect these cells, which are essential for the host organism’s well-being and survival, to be targets for parvovirus attack in the adult. The sparing of these tissues by the majority of parvovirus strains is underlined by the existence of a small subset of parvoviruses which frequently cause fatal disease in adult animals by bringing about the extensive destruction of gut epithelium and, in some cases, cells of the reticuloendothelial system. The most notable examples of this type of pathogenic behavior involve strains of the feline panleukopenia/mink enteritidcanine parvovirus serotype (Siegl, 1984b3, Aleutian disease virus of mink (Porter and Cho, 1980; Hadlow et al., 1983), and a recently isolated RV strain which causes fatal disease in young adult rats (Coleman et al., 1983). As described in a later section, in uitro studies on target cell specificity have provided significant support for the hypothesis that lytic virus growth is modulated by developmentally regulated components

AUTONOMOUSLY REPLICATING PARVOVIRUSES

95

operating in the host at the cellular level. Mohanty and Bachman (1974) reported that the actively dividing cells of the early mouse embryo are resistant to killing by minute virus of mice (MVM). Murine embryonal carcinoma cells, the stem cells of teratocarcinoma, are resistant to the prototype strain, MVM(p), as are many of their differentiated derivatives (Miller et al., 1977; Tattersall, 1978b). However, when these cells are induced to differentiate in vztro, they give rise to at least one differentiated cell type, resembling a fibroblast, which supports productive MVM(p) replication (Tattersall, 1978b). Taken with the reciprocal restriction on MVM(p) and MVM(i) (the immunosuppressive variant) replication in each other’s productive host cell type, described later, these studies demonstrate that cell cycling, although necessary, is not sufficient for the lytic, productive replication of individual parvovirus strains, and that the differentiated state of the host cell is of paramount importance. One of the most dramatic examples of a limited host tissue repertoire is exhibited by the recently characterized human parvovirus B19 (Summers et al., 1983; Cotmore and Tattersall, 19841, which is targeted for a specific subset of erythroid progenitors in bone marrow (Mortimer et al., 1983). This agent circulates in the human population causing erythema infectiosum (fifth disease), a common rubelliform rash of childhood (Anderson et al., 1984,1985). When individuals with hereditary hemolytic anemia are infected with B19, they develop an aplastic crisis associated with massive parvoviremia, probably due to infection of the large numbers of cycling target cells present in the hyperplastic marrow characteristic of such anemias (Pattison et al., 1981; Saarinen et al., 1986). The fact that differences in pathogenic potential exist not only between virus serotypes, but also between virus strains of the same serotype (reviewed in Tattersall and Cotmore, 1986) suggests that a particular tissue tropism is not an invariant property of each virus. Thus, the isolation of additional strains of minute virus of mice and rat virus, MVM(i) and RTV, respectively, as immunosuppressive agents from transplantable tumors (Bonnard et al., 1976;Campbell et al., 1977; Engers et al., 1981) lends support to the suggestion that a mutable genetic component of the virus could play a role in determining the type of differentiated cell the virus can lytically infect (McMaster et al., 1981; Tattersall and Bratton, 1983). The possibility that parvoviral suppression of oncogenic transformation is not merely another reflection of their requirement for dividing cells is afforded by a recent study of MVM interference with SV40 transformation reported by Mousset and Rommelaere (1982). They described the isolation of a BALB/c 3T3 mouse fibroblast variant

96

SUSAN F. COTMORE AND PETER TATTERSALL

which is resistant to cell killing by MVM(p) by virtue of an intracellular block to viral replication. Although this cell line is identical to its MVM-sensitive parent in its susceptibility to SV40 transformation, coinfection or superinfection of these mutant cells with MVM drastically reduces the ability of SV40 to successfully transform them. Furthermore they demonstrated that these mutant cells, once transformed by SV40 in the absence of MVM, were now capable of supporting a productive infection with MVM. Several aspects of the natural history of autonomous parvoviruses are beginning t o be understood in some detail, mostly through the analysis of tissue culture analogs of the pathogenic processes observed in the whole animal. The purpose of this article is t o review the current state of knowledge of autonomous parvovirus structure and replication, and to explore at the molecular and cellular levels, the strategies employed by these viruses to parasitize their various hosts.

11. VIRAL STRUCTURE AND ORGANIZATION A. Structure of the Virion 1 . Morphology Parvoviruses are isometric, nonenveloped particles, 20-25 nm in diameter, which contain a single-stranded DNA genome of around 5000 nucleotides (Siegl et al., 1985). Electron micrographs show icosahedral particles made up of multiple capsomers, which are easily distinguished from those of most other virus groups by their very small size (Fig. 1).Infectious particles resolve in isopycnic cesium chloride gradients into two forms: a major species with a buoyant density of 1.41 g/cm3 (termed light fulls) and a minor species with a density of 1.45-1.47 g/cm3 (heavy fulls) which is probably an infectious precursor form of the mature particle (Tattersall, 1978a). Both species sediment at 110 S, have molecular weights of 5.5-6.2 x lo6 (Siegl et al., 1985), and have a particle to infectivity ratio of 200-400 to 1 (Paradiso, 1981). Although infectious 1.41 g/cm3 virus is frequently the major form of particle isolated from infected animals, virus stocks grown in tissue culture can contain large numbers of empty protein capsids and defective particles which contain submolar amounts of DNA. The molecular weight of the empty capsids has been estimated at 4.2 x lo6, and they band as sharp peaks a t 70 S in velocity gradients and at 1.32 g/cm3 in cesium chloride (Ward and Tattersall, 1982). Although the relative proportions of full and empty

AUTONOMOUSLY REPLICATING PARVOVIRUSES

97

FIG.1. Structure of the parvovirus virion. On the right is an electron micrograph of purified MVM@)virions, stained with 3% phosphotungstic acid (instrument magnificaThe horizontal bar represents 50 pm. The left panel shows proteins of tion ~80,000). MVM(p) full virions (0 and empty capsids (e) analyzed by SDS-polyacrylamide gel electrophoresis. The viral proteins VP-1, VP-2, and VP-3 are 83,64, and 62 kDa, respectively. Markers (m) are 130, 100, 77, 68, 57, 53, and 49 kDa, from top to bottom. (Data taken from Tattersall et al., 1976.)

particles can vary depending upon both the cell type and the virus, empty virions usually outnumber infectious particles by between 2and 50-fold (Tattersall, 1978a). Defective particles have a normal capsid component of 4.2 x lo6 Da, but also contain a variable, subgenomic length of DNA such that their buoyant densities range from 1.33 to 1.39 g/cm3 (Ward and Tattersall, 1982).

2. Biochemical Structure The DNA in infectious particles makes up 19-32% of the total mass; the capsid proteins probably make up most, if not all, of the residuum. These particles do not appear to contain lipids, carbohydrates, cellular or virally coded enzymes, or low molecular weight histone-type proteins (Siegl et d.,1985). It is not known if polyamines are present to

98

SUSAN F. COTMORE AND PETER TATTERSALL

assist in stabilizing the DNA by charge neutralization, as has been shown to be the case for members of the Densovirus genus of the Parvoviridae (Kelly and Elliott, 19771, but it seems likely that in at least one subgroup of the autonomous viruses a histone-like function is provided by a specialized, highly basic region at the amino terminal of the largest capsid polypeptide (Tattersall et al., 1977). Some reports suggest that one end of viral DNA may be covalently associated with protein, but there is no clear evidence as to the nature or function of such an interaction (Muller and Siegl, 1983a; Chow et al., 1986). Classically parvoviruses are regarded as highly stable particles, being resistant to extraction with lipid solvents, pH ranges between 3 and 9, heating to 56°C for 60 minutes, and exposure to the relatively high salt concentrations used in cesium chloride isopycnic gradients (Siegl et al., 1985). It seems likely, however, that prolonged storage and repeated freeze-thawing may compromise virus integrity so that, while remaining infectious, some viruses are no longer stable to extremes of heat or high salt concentrations. Thus, for example, reports that repeatedly frozen and thawed samples of the human virus B19 are inactivated by heating to 56°C for 5 minutes (Young et al., 1984) may not reflect the real stability of the virus in uiuo. Viral characteristics discussed so far are probably common to at least most members of the Paruouirus genus. However, analysis of the structure of the viral capsid provides a clear indication that the autonomous viruses are divided into a number of rather disparate groups. The largest group identified to date shares the broad features of capsid structure exhibited by the type species of the genus, Kilham’s rat virus (RV). This group includes H-1, MVM, LuIII, PPV, and members of the FPV serotype (see Table I for acronyms). Other autonomous viruses such as BPV, LPV, ADV, and B19 each exhibit a number of characteristics which distinguish them from the RV-like viruses. As discussed in another section, there is evidence to suggest that BPV and LPV may be structurally similar to each other, but the other serotypes appear superficially quite unalike, and it may be that the autonomous Paruouirus genus comprises a number of relatively distantly related subgroups. In addition, there are a number of reported virus isolates, obtained from various animal species such as the goose, chicken, shrimp, horse, and human, which appear to belong to this genus but for which there is little structural information currently available; it is therefore impossible to say if these are unique viruses or if they are closely related to any of the previously recognized virus types. Members of the RV-like group have three major capsid polypeptides in the mature virion (Fig. 11, of which two, VP-l(83-86 kDa) and VP-2 (64-66 kDa), appear to be primary translation products (Cotmore et

AUTONOMOUSLY REPLICATING PARVOVIRUSES

99

123456789

-83 -64

-25 -23 FIG.2. Immunoprecipitation of MVM and LuIII in uitro translation products. Autoradiograph of a 10%SDS-polyacrylamide gel showing the total [36S]methionine-labeled in vitro translation products of cytoplasmic mRNA from uninfected 324K cells (lane l), LuIII-infected 324K cells (lane 21, and MVM-infected A9 cells (lane 3). Translation products of LuIII-infected cell RNA (lanes 4-6) or MVM-infected cell RNA (lanes 7-9) were immunoprecipitated with anti-LuIII capsid serum (lane 6), anti-MVM capsid serum (lane 7), serum from a hamster severely infected neonatally with the related virus H-1 (lanes 5 and 8), or nonimmune hamster serum (lanes 4 and 9).

al., 1983; Rhode and Paradiso, 19831, while the third, VP-3 (60-62 kDa), is derived by proteolytic cleavage of VP-2 (Figs. 1and 2; Clinton and Hayashi, 1976; Tattersall et al., 1977). Virus preparations grown in some cell types may also have a small amount of a fourth polypeptide of approximately 50 kDa (Tattersall et al., 1976) which is probably a proteolytic fragment of one of the higher molecular weight capsid polypeptides. Tryptic and chymotryptic peptide analysis has shown

100

SUSAN F. COTMORE AND PETER TATTERSALL

that the entire amino acid sequence of VP-2 is present in VP-1 (Tattersall et al., 1977; Brown and Salzman, 1984). In addition to sequences in common with VP-2, VP-1 contains an amino-terminal region of approximately 17,000 Da which contains a large number of basic amino acids. This basic region naturally has a profound influence on the isoelectric point of the molecule such that while VP-2 and VP-3 have p1 values between 6.6 and 7.2 (for MVM or and H-1,) the isoelectric forms of VP-1 (MVM) exhibit p1 values between 8.2 and 8.5 (Peterson et al., 1978). All three capsid peptides show low levels of phosphorylation and each can be resolved by two-dimensional gel electrophoresis into two t o four distinct species which differ in their isoelectric point by 0.05 of a pH unit (Peterson et al., 1978). Recently Molitor et al. (1985) showed that the PPV capsid peptides VP-1 and VP-2 immunoprecipitated from [32P]orthophosphate-labeledcell extracts contain substantial amounts of 32P in the form of phosphoserine while purified (cesium chloridebanded) preparations of virus contain little, if any, label. In Peterson’s study the 32P-labeled forms of the viral proteins did not comigrate with the major protein species present, and it is tempting to speculate that the phosphorylated forms might be viral proteins which have been posttranslationally modified in different ways or to different extents in order for them to function as transient intermediates in virion assembly or maturation (Peterson et al., 1978). Although all the viruses in the RV group have capsids with a fixed amount of the largest structural polypeptide VP-1 (between 12 and 18%of the total capsid protein), the relative proportions of VP-2 and VP-3 can vary dramatically for each virus from preparation to preparation (Tattersall, 1978a; Paradiso, 1981). Highly purified empty particles contain only the primary translation products VP-1 and VP-2, but all purified preparations of infectious virus also contain at least a few copies of VP-3, a proteolytic cleavage product derived from VP-2 (Fig. 1).If empty virus particles are exposed to the proteases trypsin or chymotrypsin, the structural polypeptides remain intact, but if infectious particles are treated with these proteases, VP-1 remains intact while VP-2 can be almost quantitatively cleaved to a VP-3-like peptide (Tattersall et al., 1977). Although these proteases fail to mimic the in uiuo cleavage exactly, they do cut in the same part of the molecule, trypsin cutting about 20 amino acids from the amino terminal of VP-2 (Paradiso et al., 1984). The resulting tryptic product is effectively indistinguishable from authentic VP-3 when analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Thus the presence of DNA in the virion is associated with a change in the conformation or posttranslational modification of VP-2 such that a

AUTONOMOUSLY REPLICATING PARVOVIRUSES

101

protease-sensitive site is exposed to the external environment. It has yet to be proved whether this change in virion structure, together with the inevitable cleavage of at least a few of the VP-2 molecules, is an essential part of the infectious process. As will be discussed later, sequences immediately downstream of the cleavage site, which form the amino terminal of the new VP-3 molecule, are so unusual and so highly conserved between different viral serotypes that such a role seems likely. Cleavage of VP-2 to VP-3 occurs progressively in cell culture: when total full virions (1.41-1.47 g/cm3) were purified. from MVM-infected cell nuclei at various times after the addition of [35Slmethionine, SDS-PAGE showed a change in the major capsid protein of the particle from VP-2 to VP-3 with increasing time (Tattersall et al., 1976). Paradiso (1981) also traced the progressive conversion of VP-2 to VP-3 in uiuo over a period of 24 hours after infection using iodinated input H-1 virus. Cells were separated into crude nuclear and cytoplasmic fractions at each time point, and analysis of these fractions showed that at any particular time after infection both nuclear and cytoplasmic virus had approximately the same relative amounts of these two capsid proteins. Thus major conversion of VP-2 to VP-3 is not required for, or even associated with, penetration of the virus to the cell nucleus, nor is there any evidence for the presence of a specific peptidase at a particular locus in the cell, for example at the cell membrane. Since particle preparations containing relatively little VP-3 appear to be as infectious in plaque assay as those in which VP-2 has been extensively cleaved, it seems likely that cleavage of even a minor proportion of these molecules may be sufficient to render the virus highly infectious, after which the fate of the others may become immaterial. At present the situation remains unresolved, and site-directed mutagenesis studies will probably be required to establish conclusively the biological significance of the VP-2 to VP-3 cleavage. Estimates of the molecular weight of the viral capsid and the molecular weight and stoichiometry of the individual capsid polypeptides predict a particle made up of 60-72 protein molecules of which 6-9 are VP-1. Since the entire amino acid sequence of VP-2 is contained within that of VP-1, the fact that in full virion preparations the larger molecule cannot be cleaved by proteases, whereas most if not all of the smaller species can, suggests that the common body region of these two molecules do not occupy exactly equivalent positions in the mature virion, but that each has its own unique conformation or chemical modification. Although morphologically the capsid appears to be made up of a number of capsomers (Fig. 1)details of its icosahedral structure remain uncertain. Chemical cross-linking studies with H-1 show that

102

SUSAN F. COTMORE AND PETER TATTERSALL

in the empty capsid at least one domain of the VP-1 molecules must be sufficiently closely packed to permit efficient cross-linking of a high proportion of these molecules to each other rather than to the more abundant VP-2 proteins (Paradiso, 1983). Unfortunately, we do not know if this cross-linking occurs exclusively at a particular position in the protein chain, and so do not know if these results mean that the VP-1 molecules are physically clustered into distinct capsomers in the virion, or simply that particular lysine-rich regions of the VP-1 molecules, such as their basically charged amino-terminal domains, are juxtaposed at the internal surface of the empty capsid where they might be preferentially available for cross-linking. Small-angle neutron scattering and model-building studies by Wobbe et al. (1984) tend to support the latter conclusion. These studies suggest that empty RV capsids contain two concentric shells of protein and that the inner shell, which makes up 11%of the scattering mass of the capsid (approximately 440 kDa of protein), has a higher content of basic amino acids than the outer shell. They also suggest a molecular weight of 4.0 x lo6 for RV, and a capsid made up of 60 subunits of which 10 are VP-1. Although superficially incompatible with a 32-capsomer structure previously proposed on the basis of electron microscopic observations (Vasquez and Brailovsky, 1965; Karasaki, 1966), a 60-subunit particle could possibly give rise to such an external morphology if each protein has two domains on the surface, one of which is close to the vertex (fivefold axis) and the other close to the center of the icosahedral face (threefold axis). In this case one would observe 32 capsomers, of which 20 would be face clusters of domains and 12 would be vertex clusters. Coat proteins with multiple domains provide a plausible way of overcoming the dilemma set by Casper and Klug (1962) who showed that icosahedral particles with 32 capsomers cannot be constructed with less than 180 physically equivalent structural units. In contrast to those of the RV-like viruses, infectious BPV and LPV virions contain four polypeptides. BPV virions purified from infected tissue culture cells have proteins with apparent molecular weights of 80,000, 72,000, 62,000, and 60,000 which make up 6.8, 4.6, 76.8, and 11.8%,respectively, of the total capsid protein (Lederman et al.,19831, while infectious LPV particles contain polypeptides of 96, 85, 75, and 70 kDa, comprising 5, 8, 78, and 9%of the total protein mass, respectively (Matsunaga and Matsuno, 1983). Interestingly, the capsids of LPV are reported to be 27-28 nm, unusually large for a parvovirus (Matsunaga et al., 1977), and this coincides with the relatively higher molecular weights of each of the individual LPV capsid proteins. Empty particles of LPV appear to contain relatively little of the 70-kDa

AUTONOMOUSLY REPLICATING PARVOVIRUSES

103

polypeptide (estimated at 2% of the total), while purified full particles contain a rather variable amount of this species (4-14% of the total) which can be increased dramatically by in uitro exposure to trypsin (but not to chymotrypsin), suggesting that the 70-kDa peptide maybe derived in uiuo by proteolytic degradation of the 75-kDa species. Partial proteolysis of the three higher molecular weight proteins of the PBV capsid with Staphylococcus aureus V8 protease revealed extensive amino acid sequence overlap (Lederman et al., 1983), indicating that, as discussed later for the RV group viruses, the various capsid proteins are derived from overlapping transcripts encoding the same region of the viral genome. Similarly, the three higher molecular weight capsid proteins of LPV give comparable partial V8 maps when analyzed in a two-dimensional system (Matsunaga and Matsuno, 1983). Although the structural similarities between the capsid and noncapsid proteins of BPV and LPV (discussed in Section II,C,4) suggest a close phylogenetic relationship, direct comparative studies have yet to be reported. However, these structural features and the antigenic and DNA hybridization analyses reported later do make it is clear that BPV is only very distantly related to the RV group viruses, or t o a possible third group of autonomous viruses for which the type species is the recently characterized human virus B19. When purified from human plasma, the B19 virion contains two major capsid proteins of 83 and 58 kDa, of which the latter is by far the predominant species, constituting at least 80% of the total protein mass (Cotmore et al., 1986). Analysis of the DNA sequence of the coat protein genes of B19 (discussed in Section II,C,3) confirms that the capsid structure of this virus must also be substantially different from that of the RV-like viruses. However, prokaryotic expression studies have shown that a region of continuous open reading frame in the B19 genome, encoding some 284 amino acids, contains antigenic determinants present on both the 83- and 58-kDa polypeptide. Thus, as with the other viruses in this genus, the B19 capsid proteins appear to share overlapping amino acid sequences (Cotmore et al., 1986). Aleutian disease virus (ADV) exhibits a fourth type of capsid structure. ADV-Gorhan and Utah-1 strains of ADV can be grown in uitro in Crandell feline kidney cells (CRFK) with variable efficiency. When propagated in uitro the capsid of these viruses have two polypeptides of 85 and 75 kDa, which share overlapping peptide composition and of which the 75-kDa species is slightly the more abundant (Bloom et al., 1982).However, when in uitro grown ADV-Gorham particles were first exposed to trypsin and the still-intact full particles then reisolated by centrifugation on cesium chloride gradients prior to SDS-PAGE analysis, the 85- and 75-kDa proteins had been degraded in situ to a

104

SUSAN F. COTMORE AND PETER TATTERSALL

number of much lower molecular weight protein species (Aasted et al., 1984).Similarly, when virions from Utah 1,Pullman, and DK strains of ADV were purified from infected mink tissues, the only structural polypeptides which could be identified had molecular weights between 27,000and 30,000. Although the in uitro trypsin-treated virions have a somewhat reduced infectivity, sera from infected mink contain high titers of antibodies directed specifically against the proteolyzed form of the virus, suggesting that in uiuo this type of degradation is common (Aasted et al., 1984).Such extreme sensitivity of the structural proteins to trypsin is clearly unique in the autonomous parvoviruses, but lack of sequence information, antigenic comparisons, or DNA hybridization studies makes it impossible to assess how closely ADV is related to other members of the group. 3. Antigenic Structure Classically parvoviruses were identified and differentiated from each other by the specificity of the antigenic determinants exhibited on their intact capsids. Since most of these viruses are known to agglutinate erythrocytes from one animal species or another, serotypes are frequently conveniently determined by antibody-mediated inhibition of hemagglutination (HAI). Although disparities can occur, overall results from serum neutralization studies usually mimic HA1 data, probably because the multiple loci which determine these parameters are either overlapping or at least closely juxtaposed on the virion surface. On the basis of such tests the parvoviruses have been separated into a number of distinct serotypes (see Table I) which share practially no HA1 or neutralizing antigens. However, if used with heterospecific antisera, these tests are rather insensitive to minor antigenic drift between virus isolates or host range variants in a particular antigenic group, since they average the influence of antibodies directed against many different determinants. In some serotypes, most notably the FPV group, panels of monoclonal antibodies overcome these limitations and permit the fine dissection of antigenic drift in a changing virus population. Using 13 monoclonal antibodies raised against CPV and eight monoclonal antibodies against FPV, Parrish and Carmichael (1983)were able to demonstrate clear differences between four viruses in this serotype (FPV, CPV, MEV, and RPV), and between many individual isolates of the same subtype. However, of the total 21 neutralizing antibodies used in this study 14 reacted with most isolates, reaffirming the close overall antigenic relationship of these host range variants to each other. When fourth and fifth generation cell culturepassaged stocks of CPV and FPV were added to cells in the presence of

AUTONOMOUSLY REPLICATING PARVOVIRUSES

105

various individual, homologous monoclonal antibodies, preexisting virus variants could usually be selected at frequencies between 10-4 and In all cases these variants now failed to react with the selecting antibody, but reacted to different degrees with other monoclonals. Studies of this type together with competitive binding assays between the various antibodies allowed these authors to cluster the monoclonals into groups which influenced the binding of other members of the same group to a greater or lesser extent, but which did not influence binding of antibodies from another group. In this way they showed that the capsid surface of the FPV serotype viruses contains several different, but overlapping, neutralizing antigenic sites, each comprising many different, overlapping neutralizing epitopes. Repeated passage of a CPV isolate (CPV-a) in dog kidney cells in uitro, followed by 10 passages in NL feline kidney cells, gave rise to a virus which could still grow in uitro in cat cells but had lost the ability to replicate productively in dog cells (Parrish and Carmichael, 1986). By constructing recombinant viral genomes between this mutant and a wild-type CPV isolate (CPV-d),Parrish and Carmichael mapped both the antigenic and the host range differences to a small region of the capsid gene. They then showed by DNA sequencing that there were only two single base changes in this region and that these altered the amino acids specified by two adjacent codons. Target cell specificity is dealt with in greater detail in a later section, but it seems appropriate to point out here that the surface structure of the viral particle, as monitored by the expression or absence of certain antigenic configurations, may have a dramatic influence on the ability of the virus to replicate in a particular host cell type, and that this capsid-mediated specificity may well involve intracellular interactions with host cell factors, as well as, or rather than, differences in binding to a specific cell surface receptor. All 21 of the murine monoclonal antibodies obtained in the Parrish and Carmichael study (1983) were capable of plaque neutralization. Since there was no obvious pressure applied to ensure the selection of such antibodies, this suggests that a very high proportion of the antigenic determinants accessible on the surface of the virus can influence infectivity, and thus that the surface structure of one virus serotype is likely to be quite different from that of another. However, nonneutralizing surface epitopes do exist in CPV, as demonstrated by Burtonboy et al. (1985). Of the 40 anti-CPV rat monoclonal antibodies obtained by these authors, four did not neutralize viral infectivity, and these same four failed to influence binding of any of the neutralizing antibodies. Exactly how the neutralizing epitopes are constructed remains to be

106

SUSAN F. COTMORE AND PETER TATTERSALL

determined. Molitor et al. (1983) showed that SDS-polyacrylamide gel-purified VP-1, VP-2, and VP-3 molecules from PPV, after partial renaturation in 50% glycerol, could evoke neutralizing antibodies when injected into rabbits. However, when extensive linear protein sequences from the capsid gene of CPV were cloned into prokaryotic expression vectors and the fusion proteins thus obtained used to immunize rabbits, no neutralizing antibodies were obtained, although these fusion proteins proved highly immunogenic for other types of capsid specific antibody (Smith and Halling, 1984). These prokaryotic fusion proteins lacked the glycine-rich amino-terminal region of VP-2 (discussed in Section II,C), and Paradiso has shown that a cyanogen bromide fragment of the H-1 VP-2 molecule which includes this region can evoke neutralizing antibodies, while other capsid-derived peptides do not (P. Paradiso, personal communication). It thus seems likely that in the RV-like viruses there is at least one neutralizing linear epitope which involves the amino-terminal region of VP-2. Significantly, those neutralizing monoclonals which were tested failed to detect viral capsid proteins in Western blots (Parrish and Carmichael, 1983). This suggests that although a linear fragment of VP-2 may be able to elicit a neutralizing antibody, the majority of epitopes involved in neutralization depend upon the correct tertiary, and perhaps quaternary, structure of the capsid proteins.

B . Structure of the Viral Genome 1 . DNA Structure and Sequence Each autonomous parvovirus virion contains a single copy of a linear, nonpermuted DNA molecule about 5 kilobases in length in which a long single-stranded coding region, comprising over 90% of the genome, is bracketed by shorter terminal palindromic regions capable of folding into hairpin duplexes (Bourguignon et al., 1976). The packaged strand can be predominantly of one polarity (invariably the complement of the coding sense) or a mixture of strands of both polarities packaged in separate virions (Siegl et al., 1985). By convention the genome is always drawn with the coding sense (mRNA 5’ t o 3’ direction) going from left to right. For those viruses packaging predominantly a single-sense DNA strand, this places the 3’ end of virion DNA on the left and its 5’ end on the right as depicted in Fig. 3. This convention will be used throughout this article, to assign ends to monomeric units of both single and double-stranded forms of the viral genome. Several autonomous parvovirus DNA molecules have been molecularly cloned and partially or completely sequenced. The viruses for

AUTONOMOUSLY REPLICATING PARVOVIRUSES

107

FIG. 3. Structure of the viral termini. The 5149 nucleotide viral DNA strand of MVM(p) is depicted in (A) showing the 4828 nucleotide single-stranded region in the center of the molecule and the duplex terminal structures. The nucleotide sequence of the 3’ palindrome is illustrated in (B)in the hairpin configuration. Of the 115 nucleotides in this region, 104 are base paired. The bubble in the duplex stem, created by the mismatch between nucleotides 25-26 and 88-91, is conserved in a number of serotypes. The nucleotide sequence of the 5’ palindrome (shown in C, D, E, and F)exists in two orientations, denoted “flip” and “flop,” which are the inverted complement of each other. The inboard sequence, depicted in (C), is a perfect palindrome in both forms, but the terminal regions of the duplex (D and E) contain a n unpaired loop (nucleotides 5024-5026 in flip, 5067-5069 in flop) and have a different terminal triplet (nucleotides 5047-5049 in flip, 5044-5046 in flop). In both forms shown in (C), (D), and (E), 200 of the 206 nucleotides are base paired, but a cruciform structure, shown in (F), is also possible in which 194 of the residues are base paired.

108

SUSAN F. COTMORE AND PETER TATTERSALL

which complete DNA sequences are available are MVM(p), 5149 nucleotides (Astell et al., 1983,1986);MVM(i), 5085 nucleotides (Sahli et al., 1985;Astell et al., 1986);and H-1, 5176 nucleotides (Rhode and Paradiso, 1983).Detailed sequence information about the viral termini of B19 is still lacking, but the sequence of the entire coding region is now known, and this genome is at least 5300 and probably closer to 5500 nucleotides long (Shade et al., 1986).Sequence information presented recently for BPV suggests that its genome is also about 5500 nucleotides long (Schull et al., 1985).Size estimates for other viruses based predominantly on the electrophoretic mobility of the denatured viral strands or restriction endonuclease fragments of the replicative forms suggest that few if any viruses have genomes smaller than MVM(i) or larger than B19 (e.g., Cotmore and Tattersall, 1984).This careful conservation of genome length suggests that there are fairly rigid constraints on the maximum length of DNA which can be encapsidated, although we know of no experimental data which directly address this problem. Although the genomes of all autonomous parvoviruses share certain common characteristics diagnostic of the virus family, a somewhat superficial analysis of capsid structure has already suggested that this genus may contain a number of disparate virus groups, and analysis of the viral genome lends additional support to this suggestion. In the absence of full DNA sequence information, DNA hybridization and heteroduplex mapping studies can provide some insight into the genetic relationship of one virus serotype to another, but little of this work has been fully quantitated and much of it remains unpublished. Of the studies available, one by Banerjee et al. (1983)used heteroduplex mapping of RF DNA from MVM, RV, H-1, and LuIII to show that complementary strands from these viruses were capable of reannealing along a continuous stretch covering approximately 70% of their genome length, predominantly involving the left-hand end of viral DNA. In contrast, the genome of BPV completely failed to hybridize to these viruses or to AAV under the same annealing conditions. Southern blot hybridization data from our laboratory essentially confirm these observations and also show that CPV and PPV share limited homology with the rodent group. This, once again, allows us to cluster the rodent viruses with LuIII, CPV, and PPV into a single group of rather obviously related viruses (designated the RV-like viruses), but leaves BPV by itself. The RV-like viruses also show trace homology when hybridized with nick-translated probes derived from the B19 genome, although in this case the cross-hybridization is extremely weak, rarely exceeding 0.1% of the homologous reaction (Cotmore and Tattersall, 1984). Interestingly, in the same series of experiments, these B19

AUTONOMOUSLY REPLICATING PARVOVIRUSES

109

probes failed to hybridize to BPV or AAV-2 genomes even at a stringency of T, -35°C. Clearly B19 is only very distantly related to any of the Parvoviridae examined so far, although recent evidence suggests that there may be an antigenically distinct human enteric virus to which it cross-hybridizes rather efficiently (Clewley, 1985). Both the 3’ and 5’ termini of all parvovirus genomes analyzed to date contain palindromic nucleotide sequences which can exist in the form of relatively stable hairpin duplexes in the single-stranded form of viral DNA. In the rodent viruses MVM, RV, H-1, and H3, the 3’ hairpin structure of virion DNA comprises 115 or 116 nucleotides (Astell et al., 19791, while the 5’ hairpins of MVM and H-1 contain 207 and 242 nucleotides, respectively (Astell et al., 1983; Rhode and Klassen, 1982; Rhode and Paradiso, 1983). Unlike the adeno-associated viruses, this group of autonomously replicating viruses do not possess inverted terminal repeats, but have a unique primary DNA sequence at each terminus. Initiation and termination of DNA replication occur in the terminal regions, as discussed later, and the efficient replication and encapsidation of Type I defective genomes of MVM, which have very large internal deletions, demonstrate that all critical cis-acting sites necessary for these processes are located entirely within 200 to 300 nucleotides of each genomic terminus (Faust and Ward, 1979). The 3‘ palindromes of MVM(p), H-1, RV, and H3 have been sequenced and compared (Astell et al., 1979), and although there are some minor differences, the sequence of the first 150 nucleotides of each virus is essentially the same. Since sequence homology at this level is not maintained throughout the viral DNA, this suggests a strong pressure to conserve this particular region. As shown in Fig. 3B, 104 of the first 115 nucleotides of MVM can be base paired to form a stable Y-shaped hairpin, and it may well be that it is the constraint implied in maintaining this overall conformation which resists genetic drift, rather than absolute linear sequence requirements. A bubble in the duplex stem created by a mismatch between nucleotides 25-26 and 88-91 is a common feature in all four viruses, suggesting that it may be an essential element of the terminal structure. Restriction mapping and DNA sequence studies showed that the right-hand termini of MVM and H-1 RF DNA can exist in two alternative sequence orientations, termed “flip” and “flop,” which occur with equal frequency in the monomeric, double-stranded, replicative form (RF) of the genome isolated from infected cells (Astell et al., 1983; Rhode and Klaassen, 1982). The 5’ end of single-stranded MVM viral DNA has also been shown to exist in both orientations (Astell et al., 1985). As seen in Figs. 3D and E, these two forms are only apparent because the terminal sequences have small asymmetric loops near

110

SUSAN F. COTMORE AND PETER TATTERSALL

the axis of symmetry of otherwise perfect palindromes. Sequence inversions of this type are also found in both terminal repetitions of packaged AAV DNA, but are not found at the left-hand ends of either RF or virion DNA from MVM, although these termini also contain asymmetries. The presence or absence of such inversions is of primary importance when considering possible models of DNA replication, as will be discussed below. The RF form of MVM(p) DNA has at least a n additional 18 nucleotides at its right-hand terminus which are not present on the 5’ end of the DNA strand encapsidated into virions (Astell et al., 1985). The additional nucleotides are the complement of a sequence in the viral genome (residues 4923 and 4940) that lies immediately to the 3’ side of the 5’ hairpin structure, and it seems likely that, rather than nicking at the normal cutting site for virion DNA (nucleotide 5149) during the formation of RF DNA the 5’ nick occurs on the complementary strand 18 bases from nucleotide 5149 a t nucleotide 5167. This observation is most easily explained by suggesting that for MVM the potential sitespecific nickases involved in the replication of the 5’ ends of RF and virion DNA are different, as will be discussed later. In contrast to the dissimilar sequences found a t each end of the RVlike viruses, the partial sequence information available from molecularly cloned, albeit deleted, forms of the B19 termini (Shade et d., 1986) suggests that these are extensively homologous and may even be slightly imperfect inverted terminal repeats. They also appear to be very much longer than those of the rodent viruses, with the left-hand palindrome comprising at least 178 nucleotides and that a t the righthand end at least 240 nucleotides. We have recently cloned a form of the right-hand terminus which is at least 70-80 bases longer than this, making the probable length of the hairpin sequence around 320 nucleotides. Unfortunately, using plasmid DNA to obtain sequence data for viral termini can be problematic since the central regions of many palindromes become deleted, and plasmids carrying such deletions appear to be preferentially replicated by the host bacteria (Merchlinsky et aZ., 1983; Boissy and Astell, 1985). The exact mechanism by which this occurs is not known but a t least two deleted forms of the 5‘ terminus of MVM(p) and MVMW are found which appear to result from recombination between tandem sets of 7- and ll-base pair direct repeats in a site-specific manner (M. Merchlinsky, personal communication). The most severely deleted form is known to have lost 99 base pairs between nucleotides 4997 and 5095, and both it and the less stable intermediate deletion fail to give virus when transfected into eukaryotic cells. This suggests that extensive 5’ palindrome sequences are required for productive replication in uiuo, but it is not yet clear if

AUTONOMOUSLY REPLICATING PARVOVIRUSES

111

it is the actual sequence or merely the size and structure of the terminus which is essential. For AAV there is evidence to suggest that both of these features influence the efficiency of viral replication (Samulski et al., 1983). Imperfect direct and inverted repeat sequences of various lengths and dispersed location are a common feature of all the parvoviru genomes analyzed to date. Hogan and Faust (1984) analyzed the sequence of three Type I defective MVM genomes which ranged in size from 2.7 to 3.3 kilobases. In each case the deletion occurred between pairs of perfectly homologous 4-to 10-base direct repeats such that one copy of the repeated sequence was lost and the other remained behind at the deletion junction. The repeated Sequences in each case were different and had an A + T content between 50 and 80%. One particularly striking example of a large repetitive element was identified by Rhode and Klaasen (1982) just inboard from the start of the righthand terminal palindrome in H-1, and similar sequences have also been found in CPV (Rhode, 1985b1, FPV (Carlson et al., 1985), and MVM(p) (Sahli et al., 1985; Astell et al., 1986). The 55- to 65-base pair A:T rich sequences exist as tandem direct repeats in these viruses, but, as shown in Fig. 4,sequences homologous to the repeat units in H-1 and MVM(p) exist as single copies in H3 and MVM(i), respectively. A defective mutant of H-1, called Dl-1, has three copies of this sequence, and a series of defective genomes generated by repeated passage of H-1 a t high multiplicity in NB cells showed evidence of multiple 60base insertions a t this map position (Rhode, 1978). It is not clear if duplications of these particular sequences convey any selective advantage to the virus or if they are simply a common but viable accident resulting from the particular location of the sequences in the viral genome. 2 . Organization of the Genome In all parvoviruses analyzed to date, all of the protein coding regions appear to be clustered on one of the DNA strands which is by definition the plus strand. In the case of parvoviruses which encapsidate strands of one sense, this coding strand is the complement of the virion DNA molecule. Where transcription data are available, it confirms the transcriptional use of only one strand. In MVM we have specifically looked for minor protein species encoded by the other (viral) strand using prokaryotic expression of isolated gene fragments, but have so far been unable to find any evidence for them (S. F. Cotmore and P. Tattersall, unpublished results). Figure 5, adapted from Shade et al. (1986), diagrams the blocks of open reading frame available in the plus strand of the sequenced virus genomes. In each case there are two

112

SUSAN F. COTMORE AND PETER TATTERSALL

MVM(p)

MVM( 1 )

-

H-1

H -

H-3 H-1

DII

FPV CP v

-~AATGTTAGATAGAATAAGAAGATCATGTATAATGAATAAAAGGGTGGAAGGGTGGTTGGTAGGT -TAATGTTAGATAGAATAAGAAGATTATGTATAATGGGTAAAAGGGTGGAAGGGTGGTTGGTTGGT

-AATAAATATAGTTAGTTGGTTAATGTTAGATAGAATATAAAAAGATTTTGTATTT

-TCAACTAGCACCTAGAAAATTATATTAATATACTTACTATGTTTTTATGGTTATTACATA -TCAACTAGCACCTAGAAAATTATATTAATATACTTACTATGTTTTTATG~TTATTACATA : : ; t t t t : : :

: 1 t t t t t : t : t 1 1 1 1 1

t l l l l l l t t l t t t 1 t : : t l t l l t t I t l

I!

-TCAACTAGCAGGTAGAAAAlTATA~TAACATACTTACTATGTTTTTATGTTTATTAGAT

-TCAACTAGCACCTAGAAAATTATATTAATATACTTACTATGTTTTTATGTTTATTAGAT

FIG.4. Repetitive elements located near the right-hand palindrome. A region (0) located just inboard of the right-hand palindrome in the RV group viruses may contain one or two copies of a 55- to 65-base perfect or slightly imperfect repeat sequence. Three copies of such a repeat are found in the defective H-1 genome H1 D11. The nucleotide sequences present in each of these repeats are indicated a t the bottom of the diagram, showing that although they are derived from the same region of the genome, they are not invariably homologous. Thus the repetitive element in MVM(p) initiates a t a TAAT sequence located in the middle of the H-1 repeat (at nucleotide 21 in the figure) and involves sequences up to 30 nucleotides downstream of those reiterated in H-1. Double dots (9 between nucleotide sequences denote identity. Although the sequences are diagrammed as physically separate blocks, in the genome they are closely juxtaposed, being contiguous or separated by just a few nucleotides.

large open reading frames (ORFs) which together span almost the entire genome, and a number of smaller ORFs, the exact size and location of which vary somewhat from virus to virus. As will be discussed further in the next section, in each case the long, left-hand ORF is known to encode a major, nonstructural protein, while the right-

113

AUTONOMOUSLY REPLICATING PARVOVIRUSES

B19

1

MVM

2

2 3 1

3 1

H-1

2 3

FPV

I

2

3

CPV

2

3

AAV

1

2 3

t

kb

1 I

I

I

I

1

2

3

I

I

I

4

5

I

FIG.5. Comparison of known DNA sequences of members of the Parvoviridae. The sequence of the coding strand of each virus, as far as currently available, is shown as a series of vertical bars representing predicted translational termination codons in each of the three reading frames. (Adapted from Shade et al., 1986.)

hand ORF provides most of the sequence expressed in the various capsid polypeptides. Comparison of the published sequences of MVM(p) and MVMW shows that there are 175 (Sahli et al., 1985) or 163 (Astell et al., 1986) nucleotide differences between these two strains, depending on the exact source of MVM(i). In the sequence derived by Astell there are 129 transitions and 34 transversions giving rise t o a total of 27 amino acid changes in the two major open reading frames of which 22 are conservative. Data from the Sahli group provide essentially the same picture, although the details vary slightly. Figure 6 illustrates the influence of these nucleotide changes on the amino acid sequences expressed in the two major ORFs. Most of the nucleotide differences in these regions are third-base changes and so do not influence protein sequence. However, the distribution of these changes is nonrandom, and very few are found between nucleotides 2000 and 3000. This indicates that there are additional constraints operating in this region which deter even third-base changes, and one obvious way to account for such conservation is to suggest that the small ORFs located in frame 2 also encode protein. This has been shown to be the case for the region between map units 38 and 44 in

114

SUSAN F. COTMORE AND PETER TATTERSALL

R3 reading frame

I

1

3

noncoding silent conservative radical map units kilobases

A I I

+

’

,

,,

10

20

30

b a ”

40

,-

50

I

I

1

2

3

80

70

60

I

I

90

100

I

I

4

5

FIG.6. Comparison of the genomes of MVM(i) and MVM(p). Genetic map of MVM showing the distribution of nucleotide differences between the allotropic variants MVM(p) and MVM(i). The viral genome is displayed with the 3’ end of virion singlestranded DNA, the negative strand with respect to transcription, on the left a t zero map units. At the top of the diagram the three major cytoplasmic transcripts R1, R2, and R3 are represented by double lines with single wavy lines indicating their polyadenylated tails, and single straight lines indicating the introns spliced out in the production of mature message. Beneath the transcripts a block diagram in which vertical bars represent translation termination codons shows the regions of open reading frame in all three frames of the transcribed (or complementary) DNA strand. A horizontal line labeled “noncoding” uses vertical bars to indicate nucleotide differences between MVM(p) and MVM(i) in the noncoding region of the genome, while the three lines below it similarly indicate changes in the coding sequences. Nucleotide differences which do not change the encoded amino acid (designated “silent”) are depicted separately from those which cause “conservative” or “radical” changes in the specified amino acid. Bars labeled “a” and “b” indicate the nature of the amino acid changes in the two positions where the viral genome is known to encode proteins in both the available open reading frames. “a” denotes a radical change in the sequence of NS-2 coincident with a conservative change in NS-1, and “b” denotes a conservative change in the sequence of NS-1 which does not affect the sequence of NS-2 (see Fig. 13 for further explanation). (Adapted from Astell et al., 1986.)

reading frame 2 of MVM and H-1 (Cotmore and Tattersall, 1986a), but proteins encoded by the other small ORFs have yet to be identified. CPV, FPV, and B19 lack a small ORF homologous to that unknown to encode protein in MVM and H-1 (frame 2, map units 38-44), but conservation of nucleotide sequence between CPV and FPV in various other small ORFs strongly suggest that some of these minor open frames may well be utilized to encode protein. Obviously, factors other than dual open reading frames can influence the distribution of viable nucleotide changes between highly ho-

AUTONOMOUSLY REPLICATING PARVOVIRUSES

115

mologous viruses. For example, absolute sequence conservation could be required to preserve the environment surrounding multiple potential splice sites, if splices of different efficiency were required to provide the necessary viral proteins in the correct stoichiometric amounts, as discussed later, or to preserve transcription or replication signals. Conversely, phenotypic changes between such viruses must be mediated by nucleotide changes, even though such changes may be very minor. In this context it is interesting to note that there is a “hot spot” for base changes between the two host range variants of MVM located around nucleotide 3580 (map unit 701, in the coding region of the viral structural proteins. Unlike the changes described in Section III,A,3 in the host range mutant of CPV (Parrish and Carmichael, 1986) this region contains eight changes within 25 nucleotides, but all eight occur in the third base position and do not influence the amino acid sequence of the capsid proteins. The significance, or otherwise, of this sequence will be discussed further in the section on virus-host interactions. In eukaryotes, RNA polymerase I1 transcription control regions are usually characterized by a TATA box which occurs approximately 30 nucleotides upstream of the actual site of initiation of the RNA chains. While the precise function of this sequence is not clear, it is generally thought to specify the start site for transcription. The plus-strand sequences of MVM and H-1 contain several of these signals, one of which is located upstream of each of the known transcription start sites at map units 4 and 39 (nucleotides 177 and 1977 in MVM, 180 and 1979 in H-l), while the published sequence for CPV also contains a TATAAA box a t nucleotide 318, in an equivalent position to the TATAAA boxes a t map unit 39 in MVM and H-1. In FPV this sequence has been mutated to TGTAAAT (a less usual, but not unprecedented promoter sequence), and there is a TAAAATA sequence not found in MVM, H-1, or CPV located 15 bases upstream of the TGTA site. A major FPV transcript has been shown to initiate just downstream of these signals (Carlson et al., 1985) and it seems probable that one or other of them constitutes part of the major promoter region. A second eukaryotic consensus sequence, GGPyCAATCT (CAAT) characteristically located around 30 nucleotides upstream of the TATA box, has also been implicated in the efficient initiation of RNA polymerase I1 transcripts in uiuo. The known H-1, MVM, CPV, and FPV promoters lack this sequence in the appropriate position, although the promoters at map unit 39 in MVM and H-1 have a CCAAT sequence 87 nucleotides upstream of the TATAAA box. However, in CPV and FPV the sequence in the equivalent position is mutated to TGAAT, suggesting that such signals are not essential for efficient transcription in

116

SUSAN F. COTMORE AND PETER TATTERSALL

these viruses, and emphasizing the rather unusual nature of the viral control regions. There is now good evidence to suggest that the virally coded NS-1 protein in some way up-regulates the rate of transcription from the promoter at map unit 39 in H-1 (Rhode, 1985~1,and may also be able to influence the rate of transcription from its own promoter at map unit 4 (Rhode, 1985a). Whether these interactions are direct or indirect remains to be seen, but preliminary deletion-mutant studies suggest that a region extending approximately 116 nucleotides upstream of the TATA box at 1979 may be involved (Rhode, 1935a). Weiher et al. (1983) suggested a consensus sequence for the SV40 TTT transcriptional enhancer of (GITGG A A A (GI based on point mutations in the SV40 72-base pair repeat and comparisons with other viral enhancers. The left-hand promoter of MVM has such a sequence immediately upstream of the TATA box at map unit 4 (GTGGTTT, nucleotides 164-170). This sequence also forms part of a 13 out of 16 nucleotide fit to the enhancer consensus described by Khoury and Gruss (1983), and overlaps an 8 of 11 fit homology to the human adenovirus 5 E l a enhancer sequence described by Hearing and Shenk (1983). Although this whole sequence may act as part of the viral promoter, when it is inserted into a plasmid carrying the chloramphenicol acetyltransferase (CAT) gene driven from the SV40 early promoter it is unable to influence expression, in mouse cells, of the indicator gene in the absence of the SV40 72-base repeat enhancer element (E. M. Gardiner and P. Tattersall, unpublished observations). H-1 has a somewhat different sequence in this particular location but reasonable approximations to the various putative “enhancer” consensus sequences can be found scattered throughout both genomes. However, none of the sequences in MVM(p) or MVM(i) appears able to exert a cis-acting influence on the transcription of CAT from the SV40 early promoter following transfection into mouse fibroblasts, mouse lymphocyte x fibroblast hybrid cells, or 324K cells, a human newborn kidney cell transformed by SV40 which is a productive host of both MVM strains (E. M. Gardiner and P. Tattersall, unpublished results). Similar lack of evidence for a “classical” viral enhancer sequence was It should also be noted that constructs reported for H-1 (Rhode, 1985~). containing the CAT gene driven by the MVM(i) P4 promoter are inactive when transfected into fibroblasts or lymphocyte x fibroblast hybrids derived from their natural host, the mouse. However, they are almost as active as enhancer-containing SV40 CAT constructs when transfected into human 324K cells, suggesting that there is a radical difference between MVM gene control in cells of these two species, perhaps involving the tumor virus enhancer-like sequences located

AUTONOMOUSLY REPLICATING PARVOVIRUSES

117

upstream of the P4 promoter (R. Moir and P. Tattersall, unpublished observations). The polyadenylation signals (AATAAA) of MVM, H-1, and probably CPV and FPV are restricted to the extreme right-hand end of the genome [positions 4602,4755,4820,and 4885 in MVM(p), and positions 4716,4833, and 3896 in H-11. However, MVM(i)retains only the signals at nucleotides 4602 and 4885, while CPV and FPV have a single consensus sequence in a position equivalent to the most downstream signal of MVM. Recent 3’ mapping studies of the major MVM(p) transcripts suggest that despite the potential for heterogeneity which results from the presence of multiple polyadenylation signals, all transcripts proceed to the same map position, downstream of nucleotide 4885 (Pintel and Pintel, 1985). In contrast to the relatively simple situation in the RV-like viruses, a search of the plus strand of B19 reveals multiple potential transcription start and polyadenylation signals, as shown in Fig. 7. Five TATA sequences are clustered at the extreme left-hand end of the viral genome, in the position of the single sequence seen in the rodent viruses, and there are additional TATA boxes a t nucleotides 1225,2247, 2308, and 2986. Interestingly three of these have appropriately spaced upstream CAAT sequences, but at present there is no direct information as to which actually function as part of transcriptional promoter sequences. Six potential polyadenylation signals are found at nucleotides 1303,1872,2935,4168,4307, and 4990, but again, how many of these are used remains to be established.

C. Coding Strategy of the Viral Genome 1 . Transcription

Early analysis of H-1 transcription suggested that the genome functioned as a single transcriptional unit giving rise to multiple spliced transcripts (Green et al., 1979). Subsequent studies have shown that this is not the case, and, although the most complete analysis of viral transcription has been carried out in MVM@) (Pintel et al., 19831, the general picture which is emerging may well be common to most, if not all, the RV-like viruses. MVM(p) encodes two overlapping transcription units with separate promoters near the left end (map unit 4) and middle (map unit 38) of the viral genome. As seen in Fig. 6, three major spliced and polyadenylated RNAs have been identified, 4.8,3.3, and 3.0 kilobases in length (designated R1 to R3) which constitute approximately 10-15, 15-20, and 65-70%, respectively, of the total

SUSAN F. COTMORE AND PETER TATTERSALL

118

I

2 3

reading frame

met frame 1

metframe2 V

met frame 3

I I

b l

P I

II

I I

II

11

IIIIII I I I I i l

1

,

1

I I

I

TATA AATAAA map units nucleotides

20

40 I

60 I

80

I I

1

I

I

1000

2000

3000

I

4000

100 I

5000

FIG.7. Coding strategy of the B19-Au genome. The bar diagram represents translation termination signals in all three reading frames of what is presumed to be the plus strand of the B19 genome. Lines labeled “met frames 1-3” use vertical bars to indicate the positions of methionine codons in the three reading frames, and boxed segments represent the major regions of open reading frame (ORF) in this strand. V denotes the potential initiation codons in the two major ORFs. Positions of TATA sequences and polyadenylation signals (AATAAA) are aligned for comparison with these blocks of coding sequence. (Adapted from Shade et al., 1986.)

MVM specific RNA as analyzed by quantitation of Northern blots (Pintel et al., 1983). All three species are transcribed from the virion minus strand of DNA and all have a short intron sequence between map units 46 and 48 removed. Transcripts R1 and R2 originate from the promoter a t map unit 4 and the most abundant RNA (R3) is transcribed from the promoter at map unit 38. A fourth transcript (R4) of 1.8 kilobases has also been described, but the proportion and size of this species vary from preparation to preparation, and it seems possible that such mRNAs may be the product of defective, subgenomic forms of the virus. Similar minor and rather variable RNA species (1.25 and 0.95 kb) have been described in RV (Mitra et al., 19831,while the major RNA species encoded by this virus, estimated to be 4.7,3.4, and 3.0 kb by Northern blot analysis, appear to correspond reasonably well in size, stoichiometry, and general genomic location to those described in MVM. It is also probable that the 4.8-, 3.0-, and 2.8-kb RNAs identified by Green et al. (1979) for H-1 correspond to the R1, R2, and R3 transcripts of these viruses, although the

AUTONOMOUSLY REPLICATING PARVOVIRUSES

119

initial reports of the genomic origin of these RNAs appears now to be incorrect. The major transcript present in FPV-infected Crandell feline kidney cells has also been shown to originate from a promoter at around map unit 40, and to encode almost the entire right-hand half of the genome (Carlson et al., 1985). It presumably corresponds to the 3.0kb R3 transcript of MVM and, like this RNA, contains two spliced exon sequences of 270 and 2500 base pairs separated in the genome by a small intron a t around map unit 48 (Carlson et al., 1985). There are no published transcription data for other members of the genus, but while the RV group are likely to employ somewhat similar strategies, multiple TATA boxes and polyadenylation signals distributed throughout the genomes of B19 (Shade et al., 1986) and BPV (Schull et al., 1985) suggest that these viruses may give rise to a rather different series of RNAs. 2. Translation I n uitro translation of mRNA from cells infected with MVM, H-1, or LuIII giver four major virally coded proteins (Cotmore and Tattersall, 1986a). In MVM two of these comigrate on SDS-polyacrylamide gels with the viral capsid proteins VP-1 and VP-2 (83 and 64 kDa, respectively), and are indistinguishable from these proteins purified from viral particles both antigenically and by peptide map analysis (Cotmore et al., 1983). Synthesized in uitro in approximately the same 15 ratio as the VP-1 and VP-2 species found in assembled empty capsids, these molecules appear to be the primary translation products of the viral structural proteins. The other two proteins, designated NS-1 and NS-2, are not related to the capsid proteins, but are recognized by sera from animals infected with a number of different autonomous parvovirus serotypes. The NS-1 protein made in uitro comigrates with VP-1 (83 kDa), while the NS-2 polypeptide has a n apparent molecular weight of 25 kDa. Figure 2 shows the in uitro translation products of MVM and LuIII cytoplasmic RNA before and after immune precipitation with sera directed against the viral capsids (lanes 6 and 7) and with a serum which recognizes all four of the virally coded proteins obtained from a hamster infected a t birth with the related parvovirus H-1 (lanes 5 and 8). In Fig. 6 the map coordinates of the three major transcripts of MVM are shown aligned with the viral genome, and with the major blocks of open reading frame in the plus strand. All RNA species proceed to a polyadenylation signal a t the right-hand end of the genome and thus i t is not surprising that RNAs selected by hybridization to DNA sequences from this region encode all four major viral translation products (Fig. 8). In contrast, RNA selected by hybridization to a cloned

120

SUSAN F. COTMORE AND PETER TATTERSALL

1 2 3 4 5 6 7 8 9

-83 -64

25 FIG.8. In uitro translation of hybrid-selected MVM-specific mRNA. Autoradiograph of a 10% SDS-polyacrylamide gel showing [35Slmethionine-labeled proteins immunoprecipitated from the in uitro translation products of mRNA from MVM(p)-infected A9 cells with anti-capsid serum (lane a), serum from a hamster infected neonatally with H-1 (lane 9), or nonimmune serum (lane 7). Translation products of mRNA selected by hybridization to DNA fragments from the 5' (right-hand) end of the viral strand (nucleotides 4342-5149, lanes 1 and 2) or the 3' (left) end of the viral strand (nucleotides 1411, lanes 5 and 6) yield all four virally coded proteins upon precipitation with anticapsid serum (lanes 1 and 5) and anti-H-1 infected hamster serum (lanes 2 and 6), although only a relatively small proportion of the total mRNAs encoding capsid proteins hybridize to the left end fragment. In contrast, mRNA purified by hybridization to DNA from the major intron of R2 (nucleotides 1084-1659, lanes 3 and 4) only programs the synthesis of the 83-kDa NS-1 polypeptide, as demonstrated by precipitating its translation products with anti-capsid serum (lane 3) or anti-H-1 infected hamster serum (lane 4).

AUTONOMOUSLY REPLICATING PARVOVIRUSES

121

DNA sequence from the major intron of the R2 transcript gave a single polypeptide of 83 kDa, which was shown to be the major nonstructural protein (NS-1). Since the only mRNA which encodes this sequence is the 4.8-kb R1 transcript, this identifies NS-1 as the single product of R1. Similarly, hybridization of infected cell RNA to cloned sequences from the left-hand end of the genome should select the R1 and R2 transcripts exclusively, and by comparison with the products of the R1 transcript alone (i.e., NS-1) should allow us to identify the products of R2. However, as seen in Fig. 8, the synthesis of all four viral gene products is directed by RNA selected in this manner, although only relatively small amounts of the structural polypeptides VP-1 and VP-2 (lane 5 ) are synthesized from transcripts annealing to this region of the genome. As will be discussed in a later section there is good reason to believe that the entire coding sequence of the structural polypeptides maps downstream of the promoter a t map unit 38, and can be encoded by R3 (3 kb) transcripts (Paradiso et al., 1984; LabieniecPintel and Pintel, 1986). We introduce Fig. 8 only to emphasize that it is not yet proved that the capsid genes are synthesized exclusively from R3 transcripts, and it could be that a minor proportion of both VP-1 and VP-2 molecules are encoded by R2 RNAs arising from the promoter at 4 map units. Since this promoter is thought, although not proved, to be operating at maximum efficiency prior to the activation of the R3 promoter, this might suggest an early function for capsids in viral transcription or DNA replication. Intriguing though we find this possibility, we recognize that in uitro data such as that presented in Fig. 8 could be artifactual, resulting from the in uitro breakdown of R2 (but not R1) transcripts or because sequences in the capsid genes efficiently crosshybridize to the cloned 411-base pair PstI fragment taken from the left end of the genome although we cannot find obvious candidate sequences for such cross-hybridization. Immune precipitation of polysomes prepared early after the onset of S phase with antisera directed against the capsid polypeptides should resolve this dilemma by identifying, or failing to identify, a 3.3-kb transcript encoding the structural gene products. Rhode and Paradiso (1983) identified a n 84-kDa nonstructural protein homologous to the NS-1 of MVM in the translation products of RNA from cells infected with H-1 and showed that in uitro translation of the mRNA encoding this protein could be arrested by adding cloned DNA from the left-half of the viral genome, while in uitro translation of the capsid proteins could be arrested by adding DNA from the right half of the genome. In these experiments DNA sequences hybridizing to noncoding regions of the message did not appear to impair translation; moreover it was difficult to tell if a small

122

SUSAN F. COTMORE AND PETER TATTERSALL

percentage of the capsid transcripts were inhibited by sequences encoded by R2 upstream of the promoter at map unit 39. Thus although these studies help to map the coding sequences for the capsid proteins to the right half of the-genome, they do not shed any additional light on the actual mRNA species involved. 3. Structural Polypeptides As mentioned previously, peptide map analysis reveals that the capsid proteins of all the autonomous parvoviruses analyzed to date are made up from overlapping amino acid sequences, such that the larger molecules contain the entire sequence of all smaller capsid species, but have unique amino-terminal extensions (Tattersall et al., 1977; Brown and Salzman, 1984). Paradiso et al. (1984) used the fact that the VP-2 to VP-3 cleavage can be mimicked in uitro with trypsin to enable them to identify the amino-terminal peptide of the H-1 VP-2 molecule generated by trypsin or cyanogen bromide cleavage. Although these peptides proved to have blocked amino termini, their amino acid compositions indicated that the VP-2 molecule initiated at the AUG at position 2797 in the H-1 genome. This location was then confirmed by sequencing the amino-terminal region of VP-3 created by this tryptic cleavage. The AUG at 2797 is the first such codon in the major open reading frame in the right half of the genome, but it is approximately 400 nucleotides downstream of the start of the open frame. Thus ribosomes loading onto R3 transcripts, which in MVM have been shown to start at position 2005 -+ 5 (Ben Asher and Aloni, 19841, might have to traverse some 790 nucleotides of mRNA and to read through three upstream AUGs (in alternative reading frames, at nucleotides 2289, 2335, and 2361) in order to initiate the translation of the most abundant viral gene product at 2797. However, these three AUGs are now known to be spliced out of the predominant form of the R3 transcript (which presumably encodes VP-21, leaving a sequence of 694 & 5 nucleotides upstream of the initiating AUG (Jongeneel et al., 1986). In MVM the situation is slightly more complex since there is an additional AUG at position 2504 in frame 2 which is located immediately upstream of 180 bases of open reading frame, but there is no evidence available at present to suggest that this initiation codon is ever used. In addition to being absent from the genomic sequences of H-1, CPV, and FPV, this methionine codon is located in the sequence CCUAUGA, and thus is in a very unfavorable environment for acting as a translational initiator codon, according to the optimal consensus sequence ACCAUGG recently established by Kozak (1986). The methionine codons a t nucleotide numbers 2286 and 2794 in the MVM(p) sequence, which are presumed to start VP1 and VP2, respectively,

123

AUTONOMOUSLY REPLICATING PARVOVIRUSES

v

..a

n s

n-i

n s D c I E T n

PPV CPV

BPI 819

v v

v E R A A ~ v e R s A D-S ~ S D ~ ~ V . . ~ P D C ~ ~ . . . nP A~ V InR T n s o c r v . . ~ ~ ~ c c a . . . ~ R ~E vR n A I

nvn(p)

D G I

s

(I

P D s c R I P D I

c

I A

v n

R

s

A I R

..A

.

---

~ ~ ~ ~ [ G ~ s e l S l c I v cs Il r ~n v S

R

R

S

vn

S

[

~

~

S

s

I ~~

-.. T~

I

~

S

~

~

~

S

~

G

G

G

G

G

G

C

.O... ~ S

~

V

~

~

S

~

~

~

S

~

G

G

G

G

G

G

G

]

S

~

V

~

S l

~

*

-

-

V V m S m S V[o]RC]RICSmVOT T

s

v n s

A E A s I

~

A

~

n s]v

I Ss

FIG.9. Protein sequence a t the amino termini of the capsid proteins VP-2 and VP-3. The VP-2 polypeptides of MVM, H-1, FPV, and CPV initiate a t a methionine codon (V) and encode 15-18 amino acids upstream of the first arginine residue. In each case a second arginine is separate from the first by two amino acids. Exposure of full virions to trypsin cleaves VP-2 a t both these sites (V) releasing a tripeptide and the longer aminoterminal oligopeptide, and leaves an extremely glycine-rich hydrophobic sequence as the new amino terminus of VP-3. The capsid gene of BPV contains a n homologous, glycinerich sequence, although as yet we do not know its position in the structural proteins. The B19 capsid does not contain an exactly comparable sequence, but when the putative amino terminus of the 58-kDa capsid protein (VP-2) is aligned with the second tryptic cleavage site of the RV group viruses it is clear that this region of the B19 VP-2 polypeptide is similarly uncharged and contains a smaller cluster of glycine residues.

each fit this consensus three bases out of four, being ACGAUGG and ACCAUGA, respectively. Furthermore, these two methionine codons have a purine located at position -3, whereas the problematic codon a t nucleotide 2504 has a pyrimidine at this position, which accentuates the unfavored nucleotides a t the other sites (Kozak, 1986). It is of interest in this respect that the environment of the methionine codon presumed to start the NS-1/NS-2 polypeptide fits the Kozak consensus sequence perfectly in all of the autonomous parvoviruses for which the sequence of that part of the genome is currently available. Trypsin digestion of full H-1 particles cuts the VP-2 molecule at two sites, releasing a partial cleavage product of 20 amino acids and two peptides of 17 and 3 amino acids from the amino terminal of VP-2 (Paradiso et al., 1984). This leaves a sequence which is extraordinarily rich in glycine a t the amino terminal of VP-3. In MVM, CPV, and FPV similar tryptic cleavages can be predicted and, as seen in Fig. 9, each leaves a contiguous stretch of uncharged amino acids, highly enriched for glycine residues, a t the new amino terminal. The accessibility of these two cleavage sites to trypsin suggests that this region of the molecule is located on or close to the outside surface of the virion. Although the enzyme(s) which carry out the reaction in uiuo do not appear to have exactly the same specificity as trypsin, VP-3 molecules generated in uitro by tryptic cleavage exactly co-electrophorese with authentic VP-3 molecules on SDS-polyacrylamide gels (Tattersall et al., 1977). Thus while the in uiuo cleavages may not occur precisely at

...

n

...

124

SUSAN F. COTMORE AND PETER TATTERSALL

the arginine residues illustrated in Fig. 9, they must occur within two or three amino acids of this residue. In H-1 and MVM, glycine occurs at 13 positions in the first 24 residues of trypsin-derived VP-3 molecules, and in CPV and FPV at 14 of the first 24 residues, Preliminary DNA sequence data for BPV (Schull et al., 1985) have also identified a similar sequence located immediately downstream of an arginine residue in what is presumed to be the capsid gene. As seen in Fig. 9 this region, which exhibits 68% exact homology with the equivalent sequence in H-1, contains 12 glycines in the first 24 residues, Since few sequences in BPV appear to share this level of homology with the RV group viruses, the conservation of this peptide strongly suggests that in BPV a postassembly cleavage of one of the virion proteins is likely to occur. The capsid genes of B19 and the adeno-associated viruses do not, however, contain an exactly equivalent glycine-rich sequence, and there is no evidence for postassembly cleavage of virion proteins. Labieniec-Pintel and Pintel (1986) cloned a double-stranded segment of MVM(p), spanning nucleotides 1086-5146, into a bovine papillomavirus shuttle vector and used this construct to transform C127 mouse cells. The only MVM-specific RNAs expressed in continuous cell lines carrying this plasmid originated from the MVM promoter at map unit 38, but these cells synthesized high levels of both VP-1 and VP-2, proving that the entire coding sequence of both capsid proteins can be translated from R3 transcripts. Since both of these are primary translation products, there must be a mechanism by which the translation of the cell can select two different AUGs. Recently, Jongeneel et al. (1986) showed that in cDNA clones derived from MVM(i) transcripts, nucleotides 2280 and 2377 are juxtaposed. [Note: MVM(p) nucleotide numbers used throughout, unless otherwise indicated.] As seen in Fig. 10, 5’ and 3’ consensus splice sequences occur at the positions necessary to mediate such a translocation, and R3 transcripts carrying this sequence would contain only the apparently unused AUG at position 2504 upstream of the VP-2 initiation codon at 2974. These R3 transcripts presumably encode the major capsid protein VP-2. However, in order to synthesize VP-l, an initiation codon substantially upstream of this site must be used and cDNA sequencing studies from two laboratories (Jongeneel et al., 1986; Morgan and Ward, 1986) indicate that this is achieved by using a different arrangement of splice sites. Thus transcripts encoding VP-1 apparently ignore the 5’ splice site used for VP-2 and so encounter an additional AUG triplet at nucleotide 2286 which falls in the intron of VP-2. Protein synthesis presumably initiates at this AUG and allows translation of a decapeptide (MAPPAKRAKR) encoded in frame 3 before the RNA splices t o a

125

AUTONOMOUSLY REPLICATING PARVOVIRUSES A

r

1 1

L.

1 6 3 U

1

1

I

A I I

I

I I

6

A

I

ll ll

1

V

6

I***

V*** F N

I 1

I I

R I P

*.*

E

I H 1 1 1 6 1 1 I T 1 F*** (IS-2)

&J

IlP-I1

23g

7 6 6 n 6 6 T 6 6 6 6 T A ~ A A T 6 ~ A A ~ A C ~ 6 ~CCT6AAATCAcTT66lVTA ACA A C C A A C C A C C C C A T A A ~ A C A A A ~ A A T 6 6 A C A A A A T 6 T AAhcTTTA6T6AACCAAAAT 2360

P

n666T6CCTC... IACCCACGGA6...

-

FIG. 10. Anatomy of the small splice of MVM. A small splice, occurring at the position indicated on the translation termination codon map of MVM, is present in all of the cytoplasmic transcripts of the virus. In approximately 85%of the transcripts the 5’ donor sequence a t position 2280/1 (immediately downstream of the translation termination codon for NS-1 in frame 3) is spliced to the 3‘ acceptor a t 237617 (both designated “a”),while in the remaining 15%of transcripts the 5’ and 3‘ splice sites marked “b” (at nucleotides 2316/7 and 2398/9 respectively) are used. Differential expression of the capsid proteins VP-1 and VP-2 is controlled by this alternative splicing strategy. VP-1, encoded by transcripts with a splice of the less common “ b type, initiates a t the ATG in frame 3 a t position 2286 and expresses 10 amino acids from this region before splicing into frame 1 a t the beginning of a major open reading frame which occupies most of the right half of the viral genome. VP-2, encoded by the more abundant transcripts with splices of type “a,” does not initiate until well downstream of the coding region detailed here, since the splice removes all of the possible ATG codons. R2 transcripts which contain splices of type “b” terminate translation upstream of the 5’ splice site, but transcripts with splices of type “a” program an NS-2 molecule in which the sequences expressed in frame 2 on either side of the splice are joined. Translation of NS-2 would then terminate with an amber codon at position 2396.

126

SUSAN F. COTMORE AND PETER TATTERSALL

new reading frame using alternative 5’ and 3’ splice sites at nucleotides 231617 and 239819, respectively. This allows the decapeptide in frame 3 to be spliced into frame 1near the beginning of the major open reading frame known to encode the bulk of the capsid protein. This splice removes 82 nucleotides from the genomic sequence and thus gives rise to an mRNA 14 nucleotides longer than the major R3 transcript encoding VP-2. Whether all cell types in all species generate the VP-1 and VP-2 splices with the same differential efficiency at all times remains to be investigated, but it is remarkable that these particular splice sites should, apparently invariably, be selected, given the plethora of theoretically acceptable 3’ and 5’ consensus splice sequences in this region of the genome (Astell et al., 1983). Assuming VP-1 initiates at nucleotide 2286 and terminates in frame 1 at nucleotide 4555, it will have a coding sequence of 2187 nucleotides and contain 729 amino acids, while VP-2 molecules initiating at position 2794 and coterminating with the VP-1 proteins at 4555 will express 1761 nucleotides, encoding a protein of 587 amino acids. Both of these sequences are thus reasonably compatible with the apparent molecular weights of VP-1 and VP-2 (83,000 and 64,000) determined by SDS-polyacrylamide gel electrophoresis. Moreover, according to this scheme, the VP-l-specific sequence, which in MVM is responsible for a difference of 19 kDa between VP-1 and VP-2 and for a change in pZ from 7.0-7.2 to 8.2-8.5 (Peterson et al., 1978), comprises 142 amino acids of which 15 are lysine (10.6 mol%) and 9 arginine (6.3 mol%). In B19 the organization and expression of the capsid genes may be somewhat different. Two major structural proteins of 83 kDa (VP-1) and 58 kDa (VP-2) have been described, and prokaryotic expression studies confirm that both of these polypeptides map to the large open reading frame in the right half of the viral genome (nucleotides 24414787, see Fig. 7; Cotmore et al., 1986). Antibodies directed against the protein sequence encoded between nucleotides 2897 and 3749 recognize both of these molecules, suggesting that they share overlapping sequence, but a t present no direct transcription data are available, since there is no adequate in vitro growth system for the virus. As shown in Fig. 7, there is an AUG codon located at the beginning of the right-hand ORF at nucleotide 2444. The next in-frame AUG is at position 3125, and this whole stretch of potential coding sequence terminates at nucleotide 4787. If the B19 VP-1 were to initiate a t the first position and its VP-2 at the second, this would give molecules of 781 and 554 amino acids, which is clearly compatible with the observed apparent molecular weights. However, there are three other AUGs upstream of nucleotide 2444 but downstream of the TATA box at 2308 which is presumed to be part of the promoter for this transcription

AUTONOMOUSLY REPLICATING PARVOVIRUSES

127

unit. Thus, by analogy with the RV group viruses, there may well prove to be a small splice in this region, removing some of the redundant AUGs, or perhaps even juxtaposing a short sequence from a n alternative open reading frame upstream of the major ORF as the amino terminus of VP-1. In a VP-2 transcript originating from the same promoter and initiating at nucleotide 3125, the problem of multiple AUG codons would be much more dramatic than in MVM, since the RNA would contain 16 out-of-frame potential initiator codons between the 5' end of the mRNA and the proposed initiation site. It is therefore of particular interest that B19 appears to have a n additional promoter in this region, with a CAAT box at 2950 and a TATA box at 2986. The codon at nucleotide 3125 is the first AUG downstream of this promoter, and it would thus seem from simple perusal of the sequence that in B19 the mRNAs for VP-1 and VP-2 molecules may possibly be transcribed from different promoters. Comparison of the DNA sequence encoding the capsid genes of B19, MVM, and AAV-2 reveals a conserved sequence in the VP-l-specific region of these molecules (Shade et al., 1986). Over a stretch of 30 amino acids, encoded by nucleotides 2407-2496 in frame 1 for MVM(p), 2815-2905 in frame 1 for B19, and 2341-2431 in frame 1 AAV, these viruses share homologies at the protein level of 47% (between B19 and AAV), 50% (between B19 and MVM), and 70% (between MVM and AAV). The homology between AAV and the other two viruses stops a t this point (perhaps because the bulk of the AAV capsid sequence is encoded in a different ORF) but the homology between MVM and B19 continues for a significant distance, with 50%homology over 44 amino acids to nucleotide 2947 in B19 and 38% homology over 81 amino acids to nucleotide 3058 in B19. Although the function of this conserved region is currently obscure, its presence lends support to the suggestion that the sequence upstream of nucleotide 3125 in the major right-hand ORF of B19 encodes the VP-l-specific region, despite the fact that this amino acid sequence in B19 lacks the high concentration of basic residues and marked hydrophilic character seen in the VP-l-specific regions of the other autonomous parvoviruses analyzed to date. 4 . Nonstructural Polypeptides

The first parvoviral nonstructural protein was identified in CRFK cells infected with the ADV-G isolate of Aleutian disease virus (Bloom et al., 1982). This protein, which had a n apparent molecular weight of 71,000 and could be labeled in uztro with [32Plorthophosphate, was compared to the two ADV capsid proteins (85 and 75 kDa) by twodimensional chymotryptic map analysis and shown to have a very

128

SUSAN F. COTMORE AND PETER TATTERSALL

different amino acid sequence. Serum from mink with progressive Aleutian disease, but not normal, uninfected mink had high levels of antibody against this protein which were maintained throughout the course of the disease. Subsequently, high molecular weight nonstructural proteins were also identified in the in uitro translation products of both H-1 and MVM-infected cell mRNA. These polypeptides were shown to be of viral origin and to be synthesized from transcripts which encode the major open reading frame in the left half of the viral genome (Rhode and Paradiso, 1983; Cotmore et al., 1983).In MVM, this protein, which is designated NS-1, was shown to originate from a 4.8-kb R1 transcript synthesized from the promoter at map unit 4 (Cotmore et al., 1983). The 5’ end of this RNA has been mapped to nucleotide 201 2 5 (Ben Asher and Aloni, 1984) and the first AUG encountered lies in the major open reading frame at position 261 in the MVM(p) sequence. Prokaryotic expression studies confirm this AUG as the probable initiation codon for NS-1, since antibodies directed against bacterially synthesized proteins encoded in this ORF upstream of the RNA start site (nucleotides 60-198) do not detect these sequences in the in uitro translation products of MVM-infected cell mRNA, while antibodies directed against the sequence encoded between nucleotides 225 and 534 efficiently precipitate both NS-1 and a second nonstructural protein (NS-2) which will be considered in detail later (see Fig. 11; Cotmore and Tattersall, 1986a). The whole of NS-1 is probably encoded from a continuous sequence in this major ORF, terminating at nucleotide 2277 in MVM(p). The predominant form of the 46-48 map unit splice is known to occur downstream of this site, deleting nucleotides 2281-2376 in MVM(p1 (Jongeneel et al., 19861, and thus it is highly unlikely that the NS-1 sequences are spliced into any of the minor ORFs downstream of the 2277 termination codon. Moreover, DNA fragments from sequences located upstream, but not from sequences located downstream, of the termination codon were able to arrest the in uitro translation of the equivalent polypeptide, called NCVP1, encoded by H-1 (Rhode and Paradiso, 1983). As seen in Fig. 11, antibodies raised against bacterial fusion proteins expressing the MVM(p) sequences in frame 3 between nucleotides 225-534 (fragment A), 591-957 (fragment B), and 1110-1628 (fragment C) all precipitate NS-1, confirming that the protein is the product of an R1 transcript encoding sequences throughout this entire region. An MVM NS-1 molecule specified in this way would contain 672 amino acids, which is somewhat less than would be expected from the apparent molecular weight of 83,000 but remains within possible experimental limits. The NS-1 polypeptides of the RV-group viruses show extensive anti-

AUTONOMOUSLY REPLICATING PARVOVIRUSES

129

FIG.11. Immunoprecipitation with monospecific antibodies. A block diagram of the translation termination codons in all three frames of the transcribed DNA strand is marked A, B, C, and D to indicate the sequences and reading frames of the fragments expressed as fusion proteins in bacteria. Antibodies raised against these purified bacterial proteins were used to immunoprecipitate the [35Slmethionine-labeled in vztro translation products of uninfected and MVM(p)-infected A9 cell mRNA, and the proteins analyzed on a 10% SDS-polyacryalamide gel as shown. Proteins immunoprecipitated from the total translation products (T)of mRNA from uninfected (u) and MVM(p)infected (i) A9 cells with antibodies against fragment A (nucleotides 225-534), B (591897), C (1110-1638), and D (2075-2291) are compared with those precipitated using antibodies directed against the capsid proteins (encoded by the region marked E). (Adapted from Cotmore and Tattersall, 1986a.)

genic conservation, in that sera from animals infected with any one of the serotypes will usually react with the NS-1 molecules, but rarely the capsid proteins, of any other member of the group. However, this antigenic conservation is not evenly spread throughout the molecule, but is particularly concentrated in a central “domain.” As seen in Fig. 12, comparative sequence analysis has identified a region in the mid-

AAV 019 YVY 11-1 FPV

crv

BPV AAV

G

K

R

~

~

T

L

F

~

G

P

A

~

T

G

K

~

N

L

A

K

A

/I

,111

I,

, I

I,

8 ,

I

Ill

I

I

I

I

111

111

1,1111

I ,

t t

I ,

I,,

I

I,

I

,

1 1 1

t i l l , ,

I,

I

, , , I ,

I,

I,

I

I,,

,I

I

I t

$ 2

I,

~

~

A

I,,

BI9

GGlAAGAAAAATACAClGTGGlTTTAlGGGCCGCCAAGlACAGGAAAAACAAACllGGCAAlGGCCAllGClAAAAGlGllCCAGlAlAlGGCAlGGlTAAC

YVY H-l PPV

GGCLAAAGAAATAClGllllATllCAlGGACCAGCCAGCACAGGCAAAlClAlTAllGCACAAGCCATAGCACAAGCAGllGGCAAlGllGGllGClAlAAT GGCAAAAGGAACAClGlGClCTllCACGGACCAGCCAGCACAGGCAAAlClAllAllGCACAAGCCAlAGCACAAGCAGllGGlAAlGllGGllGllACAAl GGlIAAAGAAAlACAGllClTlllCAlGGACCAGCAAGlACAGGAAAATClAllAllGCTC~AGCCAlAGCACAAGClGlGGGlAAlGllGGllGllAlAAl

npv

GGGAAAAGAAAllClAClCTlTlCTATGGGCCGGCGAGCACAGGGAAAACCAATCTCGClAAGGCCAlClGCCACGCAGlGGGGClATACGGGlGCGlGAAC

crv

I

GGCAAGAGGAACACCATClGGClGllTGGGCClGCAAClACCGGGAAGACCAACAlCGCGGAGGCCAlAGCCCACAClGTGCCCllClACGGGlGCGlAAAC I

I,,,,,

t i ,

1 1 1 ,

I,

I

,

I,,

I l l , ,

I t

1 1 1

i l l

/ I ,

AAV 019 YVM 11-1

FPV CPV 0PV

A

A

~

N

N

V

N

N

K

F

P

~

F

F

N P

O F

C

T

N

N

D

K

A

N P

L

I

N

W

K

I

E

U

I

f

A

L

U

G

N ~

F E

G

V

V

V

N

V

F

K

A

~

AAV

lGGACCAAlGAGAAClllCCCllCAACGACTGTGlCGACAAGAlGGTGAlCTGGlGGGAGGAGGGGAAGAlGACCGCC~AGGlCGlGGAGlCGGCCAAAGCC

U19

lGGAATAAlGAAAAClllCCAlllAAlGAlGTAGCAGGAAAAAGCllGGlGGlClGGGATGAAGGlAllAllAAGlClACAAllGlAGAAGClGCAAAAGCC

I,**

,,'I,

Illlllil I,

I , , ,

ItI,t,,,,,*,,,,,,,

I t

I

I

I

I

II

I

8 ,

I

X I

, I

/ I / , /

I

I , , ,

t

I

1 1

X I

I I I I I I I

I

8

I

1

:I

I :

* I 2

8 2

I

I t

I f I l i t

a:::::

I

nvu GCAGCCAAlGlAAACTllCCAlllAAlGAClGlACCAACAA~AACllGAlllGGGlAGAAGAAGClGGlAAClllGGACAGCAAGTAAACCAGlllAAAGCC GClGCCAA1GlGAAClTlCCAlllAAlGAClGlACCAACAAAAACllGAlllGGGlGGAAGAAGClGGlAAClllGGCCAGCAAGlAAACCAAllCAAAGCT H-l IPV CPV 0PV

G C A G C A A A ~ G ~ A A A ~ ~ ~ ~ C C A ~ ~ ~ A A ~ G A C ~ ~ ~ A C C A A ~ A A A A A ~ ~ ~ A A ~ ~ ~ G ~ A ~ T ~ A A G A A G C ~ G ~ ~ A A C ~ ~ ~

CACAACAACAAACAGlllCCllllAACGAlGCACCCAACAAAAlGATCCTGlGGlGGGAGGAA

AAV BIP YVY 11-1 IPV CPV UPV

l

C

S

G

~

l

l

R

l

D

D

V

V

K

K

G

G

K

G

S

K

~

C

S

K

V

l

K

l E

E

P

P

l

l

P

P

V

V

l

l N

T

Y T

l

l

N

E

Y

E

N

I

W l

l

l

I

I

AAV

AllClCGGAGGAAGCAAGGlGffiCGlGGACCAGAAAlGCAAGTCClCGGCCCAGAlAGACCCGAClCCCGlGAlC6lCACCTCCAACACCAACAlGlGCGCC

019

A ~ ~ ~ ~ A G G C G G G C A A C C C A C C A G G G ~ A ~ A T C A A A A A A ~ G C G ~ G ~ A A G ~ ~ ~ A G C ~ G ~ ~ C C ~ G ~ A ~ ~ A C C ~ G ~ ~ ~ ~ ~ A ~ A A C

YVY 11- I trv CPV IlPV

AlllGClClGGlCAAAClAllCGCAllGAlCAAAAAGGAAAAGGCAGCAAACAGAllGAACCAACACCAGlCAlCAlGACCACAAAlGAGAACAllACAGlG

, ! I

I

ii

I ,

I

I X I I I I /

I Z I I

I

I t

$ 8

I I I

I ,

I

/,,,

I I I I l i I I I I

,,,*

: I , , , *

,

,

8

, , # I t

I

! I Z i

Ili

i I 3 i i l l

I,:,

IIiIiii

i

A~~T~T~CTGGCCAAACCA~ACGCA~~GA~CAAAAAGGAAAAGGCAGCAAACAGA~~GAACCAACA~CAG~~A~~A~GACCACCAACGAGAAC A l l I G l 7 C l C G A C A A A C A A l T A G A A l l G A l C A A A A A G G T / l l

GATCAAAAACGlAAAGGAAGlAAGCAAAllGAACCAAClCCAGlAAllAlGACAACGAAlGAAAAlAlAALAAll

AAV 819

YVY 11-1 FPV CPV UPV

I V V

R R

I l

G C

C C

E E

E E

R R

P P

E E

H H

T l

V V

P P

R I

I

R R

O O

R R

Y Y

L W L N L N

I I l

H K M

L L

V V

R C C

T K K

L L L

P P P

G D F C G D F G G D F G

AAV

G~GA~~GACGGGAAC~CAACGACC~~CGAACACCAGCAGCCGT~~CAAGACC~GA~G~~CAAA~TTGAAC~CACCCGCCG~C~~~A~CA~GA

819

GllGlAAGCGGGAACAClACAACAAClGlACAlGClAAAGCCllAAAAGAGCGCAlGGlAAAGllAAAClllACl6lAA6ATGCAGCCCTGACAlGGGG

/ I

YVY 11-1

FPV CP v APV

I

I l l 8 S l l

$

8

,

I,

I

,,I

I

I

I,

,111

I ,

1 1 ,

I

I ,

I ,

I

,

I /

I

1

I,,* I

I l Z

AUTONOMOUSLY REPLICATING PARVOVIRUSES

131

dle of the molecule which appears to highly conserved among all members of the Parvoviridae examined to date (Shade et al., 1986). In the 405 base sequence from this region illustrated in Fig. 12, the levels of nucleotide homology are 53%between MVM and AAV-2,52% between MVM, and B19 and 52% between B19 and AAV-2, while a t the protein level these homologies are 51,41, and 51%, respectively. Among members of the RV group viruses these homologies are much more extensive, with MVM and H-1 being distinguished by a single amino acid, and FPV/CPV differing from MVM and H-1 at six residues only (95.5% homology). Presumably this sequence has been conserved because it forms a vital part of the active site of the molecule in one or more of the functions shared by this protein in all the parvoviruses. Of particular interest is the sequence corresponding to the consensus G(X),GKT/S(X),-,I/L/V, which is located near the beginning of the homology region, in MVM between nucleotides 1455 and 1496.This consensus has been recognized as a feature of purine triphosphate binding sites present in proteins of a number of prokaryotes, eukaryotes, and their viruses (Walker et al., 1982; Gay and Walker, 1983; Gorbalenya et al., 1985; Gill et al., 1986). This suggests that whatever common function(s1 is mediated by the conserved region, it may well involve the hydrolysis of adenosine or guanine triphosphate. Antibodies directed against bacterial proteins encoded by a n MVM DNA sequence that includes part of the conserved region (nucleotides 1110-1638, Fig. 11)cross-react well with the NS-1 molecules of H-1, LuIII, RV, PPV, and CPV synthesized in uitro or in uiuo, but fail to react with the translation products of BPV or AAV-2 mRNA, showing that the most conserved residues are not immunodominant, at least in the denatured form of the protein. Possibly one or more oligopeptides could be synthesized from this region which would be capable of elicit-

FIG.12. The highly conserved region of parvoviral NS-1 genes. Homology between NS-1 genes of a helper-dependent (AAV-2) and various helper-independent parvoviruses. A region of conserved amino acid sequence is shown along with the corresponding nucleotide sequence. Nucleotide homology between B19 and AAV-2 or MVM is indicated by short vertical bars. At the protein level homology among B19, AAV-2, and MVM is indicated by enclosing the homologous amino acids in boxes. The corresponding sequences from other autonomous parvoviruses are given for comparison. The nucleotide sequences in this figure begin a t nucleotide 1293 (AAV-2), 1390 (Bl9-Au), 1428 (MVMp), 1431 (H-11, 1 (CPV), and 345 (FPV) in the published sequences. The exact position of this sequence in the BPV has yet to be determined. At the top of the diagram the major transcripts of MVM and a block diagram of the translation termination codons in all three frames of the transcribed strand of MVM are aligned and marked with a double-headed arrow to indicate the position of this conserved sequence in the genome.

132

SUSAN F. COTMORE AND PETER TATTERSALL

ing antibodies that could identify the NS-1 proteins of the entire Parvoviridae family and which, more importantly, might block whatever function is being performed by this protein domain. DNA sequence comparisons of the NS-1 protein genes of FPV/CPV and MVM downstream of this homology region (toward the carboxy terminal of the polypeptide) reveal a substantial decline in the level of conservation (to around 57% if a few small gaps in the nucleotide lineup are allowed), but the homology between these viruses still remains considerably higher than that seen among MVM, B19, and AAV-2, which is only about 16%. Moreover, antibodies directed against the amino-terminal domain of MVM NS-1 (between nucleotides 225 and 534) react efficiently with the homologous molecules from H-1 and LuIII, but entirely fail to react with the NS-1 of CPV (Cotmore and Tattersall, 1986a). Unfortunately, the DNA sequence for this region of the CPV or FPV genome is not currently available, so it is not possible to say whether this lack of cross-reactivity represents intense antigenic drift or the complete absence of an analogous region in the CPV molecule. Antibodies directed against bacterial proteins expressing a sequence from the middle of the large ORF in the left end of the B19 genome (between nucleotides 1072 and 2044) recognized noncapsid proteins of 71, 63, and 52 kDa in the tissues of a fetus infected transplacentally with B19 (Cotmore et al., 19861, but it is not yet clear whether these are primary translation products or processed forms of the nonstructural proteins. These antibodies did not recognize the NS polypeptides of the RV group viruses or of BPV, even though the expressed peptide contained the highly conserved 135 amino acid sequence discussed previously (nucleotides 1390-1794 in the B19 sequence). This major ORF is preceded by a cluster of TATA sequences at the left end of the genome (at nucleotides 257, 319, 321, 323, and 4121, one or more of which presumable function as part of a transcriptional promoter, but no CAAT sequences are present upstream of any of these putative promoters. An AUG at position 436 could initiate translation of a nonstructural protein, giving rise to a primary translation product of 671 amino acids terminating at nucleotide 2445. Such a protein might well migrate with an apparent molecular weight of 71,000, even though its calculated molecular weight would be 86,000. The left-hand ORF in B19 contains an internal promoter-like sequence (with a TATA box at map unit 24, nucleotide 1225, and a CAAT box at 11951, which, if used to generate a transcript, could encode a protein initiating at an AUG codon at position 1288 and presumably coterminating with the larger polypeptide at 2445. Such a molecule would contain the carboxy-terminal387 amino acids of NS-1 (including the 135-amino acid homology region) and would have a calculated molecular weight of

AUTONOMOUSLY REPLICATING PARVOVIRUSES

133

49,000. At present, however, we do not know if either the 63- or the 52kDa protein seen in uiuo are produced in this way. Interestingly AAV-2 is known to have a functional promoter at around this position (map unit 18) in its equivalent to the NS-1 gene. Preliminary sequence data show that BPV has TATA boxes at map units 4 and 13, and a large open reading frame which extends through most of the left half of the genome and includes the NS-1 homology region detailed in Fig. 12 (Schull et al., 1985). Nevertheless, a high molecular weight nonstructural protein corresponding to this sequence has yet to be identified in uitro or in uiuo.Instead, a major virally coded 27-kDa nuclear phosphoprotein (designated NPl), which is not a constituent of the viral capsid, has been identified both in uiuo and in the in uitro translation products of BPV-specific mRNA (Lederman et al., 1983, 1985). This protein is not obviously related to the bulk of the capsid proteins, but antigenic cross-reactions and peptide mapping studies suggest that it may share amino acid sequence with the amino terminal of VP-1. NP1 has recently been shown to bind DNA sequences from the 3' end of the BPV genome (Lederman et al., 1985). Polypeptides of 25 and 22-kDa which appear to be virally coded but which are not found in intact virions, were also identified in LPVinfected cells (Matsunaga and Matsuno, 1983), but at present it is not known if either of these correspond to the NP1 protein of BPV. A second nonstructural protein (NS-2) has been identified in the in uitro translation products of H-1, MVM, and LuIII mRNA and in RVinfected cells in culture (Cotmore and Tattersall, 1986a, and unpublished observations). As seen in Figs. 2 and 11, when synthesized in uitro from viral mRNA this protein has an apparent molecular weight of around 25,000. Prokaryotic expression studies reveal that the NS-2 molecule specified by MVM, depicted schematically in Fig. 13, shares a common amino-terminal region with NS-1, encoded in frame 3, but utilizes a small block of alternative ORF in frame 2, located between nucleotides 2075 and 2291, to encode the carboxy-terminal half of the molecule (Cotmore and Tattersall, 1986a). This suggests that NS-2 is likely to be the product of a 3.3-kb R2 transcript which arises from the promoter at the left end of the genome and has two intervening sequences (map units 10-40 and 46-48) spliced out (see Fig. 6). Recently Jongeneel et al. (1986) isolated and sequenced a cDNA clone, derived from MVM(i) mRNA, in which nucleotide 514 was followed by nucleotide 1990 [MVM(p)nucleotide numbers], with the intervening sequence of 1475 bases deleted. Such a splice would join 84 amino-terminal residues encoded in frame 3 downstream of the AUG at 261, to 96 residues encoded in frame 2 between nucleotides 1991 and 2278, an arrangement which is clearly compatible with the organization of

134

SUSAN F. COTMORE AND PETER TATTERSALL map units 25

0

100

75

50 I

A A

R1

R

2

0

w w

A

R3

NS-2

1 0

1

1000

2Ooo

1

3000

4ooo

I

t

5000

nucleotides

FIG.13. Coding strategy of the MVM genome. The cytoplasmic transcripts of MVM, denoted R1, R2, and R3, are aligned beneath a line diagram of the viral DNA strand which illustrates the extent of the 3’ and 5’ terminal hairpin palindromes and the positions of the two promoters at map units 4 and 38. Major blocks of open reading frame in the transcribed DNA strand are depicted for each of the three possible reading frames ( F l , F2, and F3) and the sequences encoded in the viral proteins NS-1, NS-2, VP-1, and VP-2 are illustrated with numerals to designate which reading frame is expressed in each part of the molecule.

NS-2 as determined by the expression studies and shown in Fig. 13. This particular splice, which is summarized in Fig. 14, is somewhat unexpected since the splicing apparatus of the cell would have to ignore the potential 5’ splice sites at nucleotides 494, 518, and 539, which fit well to the consensus sequence determined from other eukaryotic proteins (Mount, 19821, while recognizing the sequence AAA/GCAAGT, which is extremely unusual although not unique (e.g., Fischer et al., 19841, in having a C instead of a T at position 2 in the intron. The cDNA clone isolated by Jongeneel et al. (1986) also contained the minor splice which juxtaposes nucleotides 2280 and 2377 [MVM(p) nucleotide numbers], described as alternative splice “a” in Fig. 10. This would leave the NS-2 protein coding sequence in frame 2 beyond the minor splice for another six amino acids before terminating at an

AUTONOMOUSLY REPLICATING PARVOVIRUSES A

I

L

I I

2"'L

lUS-II3P

.,... ...

0

V W

R

S

CCTGATTGGTCA

,

135

A

I

I I

L

A I I

L O G A K 0 A U S Y K F T I F G 1 L 1 I I4 0 T E K I A S O P E 1 I A E P K I P 1 P I U L L O I A If S P F 1 1 P K 5 1 P L S 0 1950 tOl0

K

E

AAACTGGGCGGA

CAAA

115-21 lWS-ll

TGCCAACTC

. G G ~ C i ~ ~ C C ~ G i ~ i i T G ~ C c c c c c i ~ G ~ = ~ ~ c G c i i c ~ G

FIG.14. Anatomy of the large splice of MVM. The R1, R2, and R3 transcripts of MVM are superimposed on a diagram showing the translation termination codons in all three frames of the transcribed DNA strand. The position of the 3' and 5' splice junctions determining the major intron of the R2 transcript are indicated, together with the genomic nucleotide sequence at these junctions. An unusual 5' splice site, a t nucleotides 5141515, in which the bases GC substitute for the more usual GT sequence at the intron side of the site, is used despite the presence nearby of a number of other potential 5 donor sequences (marked by vertical lines) which fit more closely to the proposed consensus sequence (Mount, 1982). The 3' splice site used by R2, a t nucleotides 1989/1990, transfers the coding sequence from frame 3 into frame 2. Other consensus 3' sites in this vicinity (marked by vertical lines) could transfer the coding sequence into alternative reading frames, but these sequences are also ignored. The R1 transcript is not spliced in this region of the genome.

amber codon at nucleotide 2396. In the absence of this minor splice or if the alternative splice "b" as described in Fig. 10 is used, the protein would still terminate seven residues downstream of nucleotide 2280 at position 2300. This is of interest because, as discussed previously (in Section II,C,3), it has been shown that while most R3 transcripts use this 5' splice site, the 15%or so which encode VP-1 do not. This implies

136

SUSAN F. COTMORE AND PETER TATTERSALL

that NS-2 might exist in at least two forms with slightly different carboxy-terminal hexapeptides. If the NS-2 transcripts use the splice sites described by Jongeneel et al. (19861,a protein starting a t position 261 would contain 188 amino acids and have a calculated molecular weight of 25,000, which is clearly in good agreement with the apparent molecular weight of NS-2 estimated from SDS-polyacrylamide gels. NS-2 molecules translated from mRNAs which contain the map unit 46-48 splice alternative “a” would terminate in frame 2 a t a single amber termination codon. However, after this single terminator frame 2 is open for another 96 codons to nucleotide 2667. As mentioned previously in the discussion of Fig. 6, evidence from comparative sequence studies of MVM(p) and MVMW suggest that this alternative ORF downstream of the minor splice is used by the virus to encode essential protein, and it could well be that the translation apparatus of the cell is able to read through a proportion of these amber codons in a manner similar to that found for the translation of the protease encoded by the gug-pol gene of murine leukemia virus (Yoshiyuki et al., 1985). Such natural suppression would give an NS-2 derivative molecule of 285 amino acids and a calculated molecular weight of 37,000. Alternatively, the NS-2 molecules may actually terminate at the amber triplet at position 2396, and the downstream sequence be used to encode, in reading frame 2, a 6-kDa peptide originating from the in-frame AUG at nucleotide 2504. However, while the H-1, FPV, and CPV sequences do not contain an AUG at an equivalent position, all of these viruses do contain a somewhat homologous small alternative open reading frame in this region, making the latter suggestion seem less probable. Wherever it is actually located, the peptide originating from this open frame must confer unique properties on the whole protein, since it is exceedingly rich in leucine (20.8 mol%) and threonine (18.1mol%),but contains only 11.1mol% basic and 5.6 mol% acidic amino acids. Bacterially synthesized proteins containing this sequence [nucleotides 2441-2654 in MVM(p)l aggregate extensively even after boiling in 2% SDS (S. F. Cotmore and P. Tattersall, unpublished observation). However, to date the existence of either the 6kDa protein or the 37-kDa derivative of NS-2 as natural products of MVM gene expression remains hypothetical. Antibodies raised against the amino-terminal (nucleotides 225-534) or carboxy-terminal (nucleotides 2075-2291) domains of the MVM NS-2 fail to recognize homologous molecules in the translation products of CPV-infected cell mRNA (Cotmore and Tattersall, 1986a), and there is, as yet, no evidence that comparable proteins are encoded by FPV or CPV. Analysis of the DNA sequence of CPV or FPV (Fig. 6) shows that, as in MVM, there is a small alternative ORF located just

AUTONOMOUSLY REPLICATING PARVOVIRUSES

137

upstream of the map unit 46-48 splice (between nucleotides 288 and 566 in CPV), but this sequence terminates upstream of the 5’ splice site and, when compared to the amino acid sequence expressed in the carboxy-terminal half of NS-2, shows only very weak homology (a maximum of 28% over a region of 72 amino acids). Such observations indicate that the transcription patterns and organization of the nonstructural genes of FPV and CPV may be substantially different to those determined for MVM and H-1.

111. VIRAL LIFE CYCLE A . Early Events

Productive infection is initiated by absorption of the virion to specific cell surface receptors. The exact nature of these receptors is unknown, but N-acetylneuraminic acid residues appear to play an essential role in the binding of MVM, since pretreating cells with neuraminidase prevents binding (D. Stanick and P. Tattersall, unpublished results). Pretreatment with trypsin also abolishes binding, suggesting that the receptors involve or are displayed on a protein backbone, but at present it is not established whether the receptor for a particular virus serotype is a single molecular species or a number of different molecules which all carry specific carbohydrate side chains. MVM is known to bind efficiently to a number of different cell types, from various species, and binding studies indicate that such cells frequently have between lo5 and 5 x lo5 specific binding sites per cell (Linser et al., 1977; Spalholz and Tattersall, 1983). This remarkable abundance of receptor sites would tend to suggest that multiple protein species may be utilized, but expression of the receptor is nonetheless under developmental control, since some differentiated cell types, such as those of the B lymphocyte lineage, appear to lack receptors for MVM (Spalholz and Tattersall, 1983) and are completely resistant to virus infection. As discussed in the previous section, a proportion of the VP-2 molecules in full virions of the RV group viruses are cleaved to yield VP-3 molecules, thus exposing hydrophobic amino-terminal protein sequences near the virion exterior. Empty particles cannot be cleaved at this site in VP-2, suggesting that they exhibit a different surface topography. However, in MVM, both full and empty particles compete equally for the receptor sites which lead t o infection (Spalholz and Tattersall, 1983), indicating that this topographic difference does not affect the initial interaction with the host cell.

138

SUSAN F. COTMORE AND PETER TATTERSALL

Binding of virus to the cell surface can occur at 4°C and is apparently reversible, since labeled virions can be competitively displaced from the cell surface by subsequent addition of cold virus (Linser et al., 1977). Bound virus can also be rapidly removed with EDTA or trypsin, a fact which permits the temporal dislocation of binding 2nd penetration steps in experimental systems. Internalization takes place through what appear to be coated pits (Linser et al., 19771, and electron micrographs show cytoplasmic virus clustered within multivesicular bodies and heterophagosomes, with the viral particles arrayed as a lining around the membrane or in paracrystalline aggregates (Richards et al., 1977). Since a single virion is capable of initiating infection (Tattersall, 1972; Tattersall and Bratton, 1983), it seems probable that many of the input virions and empty capsids observed a t the high multiplicities used in these studies might not have been destined for the cell nucleus. Much of the observed phagosome-enclosed virus may thus have been in the process of being inactivated and degraded, rather than being transported to the nucleus prior to DNA replication. Paramyxoviruses are similarly engulfed through coated pits and pass via coated vesicles to a n acidified endosome compartment, but virus destined to infect the cell is never exposed to lysosomal enzymes. Instead, the acid pH in the endosome induces conformational changes in the hemagglutinin glycoprotein of the viral envelope, exposing a hydrophobic amino-terminal protein sequence which penetrates the endosomal membrane and initiates fusion of the lipid bilayers of the virus and endosome, whereupon uncoated viral RNA is ejected into the cell cytoplasm (White et al., 1983). Although an exactly homologous mechanism could not be used by nonenveloped viruses, both paramyxoviruses and parvoviruses are faced with the problem common to all viruses of translocating across the limiting membrane of the host cell, and the exposure by both virus groups of hydrophobic protein sequences near the surface of their infectious particles suggests that somewhat analogous processes could be employed. However, for MVM, lowering the pH of the culture medium does not induce bound virus to interact irreversibly with the outer lipid bilayer of the cell, and exposure to lysosomotropic bases does not protect the cell from viral infection (D. Stanick and P. Tattersall, unpublished observations). Thus the pH shift which is presumably encountered by the virus in the vacuoles of the cytoplasm does not appear to contribute to the infectious process, and it is not known a t present how the virus eventually penetrates the plasma membrane or is translocated t o the cell nucleus. The protein sequence PPKKKRKV appears to mediate translocation of SV40 large T antigen to the nucleus (Kalderon et al., 19841, and a similar basic, proline-rich sequence MAPPAKRAKR occurs at the

AUTONOMOUSLY REPLICATING PARVOVIRUSES

139

amino terminus of the VP-1 capsid proteins of MVM, H-1, CPV, and FPV (see Fig. 101, but whether or not this sequence is influential either in the passage of input virus to the nucleus, or in the intranuclear accumulation of de novo synthesized capsids later in infection remains to be investigated. Finally, there are no published experimental data which directly addresses the problem of how or where the viral DNA is released from its capsid. The absence of large amounts of free, cytoplasmic, singlestranded DNA tends to suggest that either relatively few of the incoming virions are uncoated andlor that this event occurs inside the nucleus. Electron micrographs show that when purified full MVM particles are exposed to formamide the DNA strand is extruded from the capsid, while remaining attached to it at one end (Bourguignon et al., 1976). The nature of this capsid-DNA interaction is not clear, nor is the mechanism by which the DNA is expelled, but such observations imply that diminishing ionic interactions within the particle might mediate this event. Moreover it is even possible that the first step in viral DNA replication, synthesis of the complementary strand, could occur while the incoming strand is still associated with the capsid, provided this association involves the 5’ end of the viral DNA. The possibility that the parental genome is uncoated in the nucleus is strengthened by the observations, to be discussed below, that the incoming capsid is in some way involved in the initiation of viral gene expression. During the course of a normal lytic infection, the MVM genome does not integrate into the host DNA (Richards and Armentrout, 19791, although whether such events ever occur during prolonged nonlytic interactions, such as the restrictive infections to be described later (Tattersall and Bratton, 19831, still needs further analysis. However, in the absence of a helper adenovirus, AAV genomes integrate with high efficiency into host cell DNA (Cheung et al., 19801, and it will be interesting to determine whether autonomous viruses, such as B19, which have closely related terminal repeat sequences a t each end of their genomes, rather than the unique termini found in MVM and H-1, share this integrative ability with AAV.

B. Gene Expression The processes of viral entry and intracellular accumulation can proceed synchronously in all the cells of an infected population irrespective of their position in the cell cycle (Siegl and Gautschi, 1973a; Rhode, 1973). In contrast, viral DNA replication and gene expression are entirely dependent upon one or more cellular functions expressed

140

SUSAN F. COTMORE AND PETER TATTERSALL

transiently during the S phase of the cell cycle (Tennant et al., 1969; Tattersall, 1972; Siegl and Gautschi, 1973b; Rhode, 1973). Since parvoviruses are unable to induce resting cells to enter S phase but must wait for their host cell to embark on DNA replication of its own volition (Tattersall, 19721, infecting randomly dividing cell populations leads to an asynchronous round of replication in which it is impossible to trace the progression of specific viral activities with any precision. To resolve this problem, a number of different regimes have been employed which induce a more or less synchronous cell cycle in parvovirus-infected cells (Rhode, 1973; Siegl and Gautschi, 1973a; Richards et al., 1977; Parris and Bates, 1976; Ward and Dadachanji, 1978; Rommelaere et al., 1981; Hardt et al., 1983). Unfortunately, none of these techniques, used individually, gives a population of cells which are sufficiently tightly coordinated as they enter S phase to permit fine dissection of the viral life history. Better resolution can be achieved by using a double-block schedule, such as that illustrated in Fig. 15. In these experiments A9 cells were first allowed to accumulate in Go by starving them of isoleucine for 48 hours (Ley and Tobey, 1970). Release of the isoleucine block was accompanied by addition of the DNA polymerase (Y inhibitor aphidicolin (Pedrali-Noy et al., 1980), so that over the course of the next 20 hours, cells leaving Go somewhat asynchronously progressively accumulated at the G,/S phase boundary. As seen in Fig. 15, uninfected cells exposed to this regime began DNA synthesis rapidly upon removal of the inhibitor, and exhibited what appears to be an extremely well-coordinated, biphasic, 8-hour S phase, in contrast to the 20-hour S phase found by Hardt et al. (1983) using the isoleucine block alone. MVM virions, added to the cells along with the aphidicolin, were able to penetrate the cell, accumulate in the nucleus, and become uncoated, in the 20 hours before release of the replication block, and thus the viral genome was ready to replicate and express its genes as soon as the cells initiated DNA synthesis. It is clear from the data for NS-1 expression presented in Fig. 16 that de novo synthesis of the viral proteins was dependent upon removal of the aphidicolin block. In the Western blot analysis used here, capsid proteins were detected at all times postinfection because this technique cannot discriminate between newly synthesized and input virus (added at 30 m.0.i. per cell), but pulselabeling and immunoprecipitation studies confirm that neither NS-1 nor the capsid polypeptides are synthesized during the 20 hours preceding S phase. However, within 2 hours of removing the inhibitor, de novo synthesis of NS-1, VP-1, and VP-2 can be detected, and this reaches maximal levels for all three proteins over the course of the next 6 hours, well before the major burst of duplex viral DNA replica-

AUTONOMOUSLY REPLICATING PARVOVIRUSES

141

FIG.15. Kinetics of DNA replication in a highly synchronized infection. DNA replication in uninfected (0) and MVM(p)-infected (@) A9 cells following synchronization by successive isoleucine deprivation (-ile) and aphidicolin (+aph) inhibition of DNA polymerase a activity. Points on the horizontal axis indicate time (in hours) before or after removal of aphidicolin (at 0 hours). Infected cells received MVM(p) (30 m.o.i.1 at the time of addition of aphidicolin (-20 hours). The lower panels show total [3Hlthymidine-labeled DNA synthesized by uninfected (u) and infected (i) cells during 1-hour periods as indicated. Total cellular DNA was extracted, embedded in agarose, and electrophoresed on a 1%agarose gel. H, High molecular weight DNA;D, dimer, and M, monomer duplex MVM replicative forms; S, single-stranded progeny viral DNA.

142

A

SUSAN F. COTMORE AND PETER TATTERSALL

U 0 2 4 6

B

U0246

-VPI

-VP2,3

FIG. 16. Kinetics of viral protein synthesis in highly synchronized infection. Western blot analysis of viral proteins synthesized early in S phase in synchronized uninfected (u),and MVM(p)-infected A9 cells (from the experiment described in Fig. 15). Samples of infected cells were taken a t 2-hour intervals a t and following release from the aphidicolin block, and transfers were probed with a rabbit antibody directed against the protein sequence encoded between nucleotides 1110 and 1637 in the MVM NS-1 gene (A) or against gradient-purified MVM capsids (B).

tion (Fig. 15,and S. F. Cotmore and P. Tattersall, unpublished observations). Clearly an event which occurs a t the very beginning of S phase controls viral gene expression, and it is tempting to speculate that this event is the synthesis of the complementary DNA strand which thus provides a duplex template for viral transcription. As will be discussed in the next section, there is considerable controversy as to whether DNA polymerase a is responsible for synthesis of the complementary strand in uzuo, and it may be that aphidicolin simply exerts its inhibitory effect indirectly by preventing other cellular events which influence expression of the viral proteins. Synthesis of the complementary DNA strand occurs when both MVM(p) and MVMW infect A9 cells (Spalholz and Tattersall, 19831,

AUTONOMOUSLY REPLICATING PARVOVIRUSES

143

but transcription and protein synthesis are confined to MVM(p). Thus, having a duplex transcription template is not suficient to ensure gene expression, since developmentally regulated host cell factors appear to exert an overriding influence. Mechanisms involved in this type of regulation form a n area of considerable current interest and will be discussed in more detail later. As mentioned above, if MVM is used to infect highly synchronized cell populations, proteins encoded by both halves of the viral genome are synthesized almost simultaneously within the first 2 hours of S. Since a t least the majority of the capsid proteins are encoded by R3 transcripts arising from the promoter a t map unit 38, while NS-1 is synthesized from an R1 transcript arising at map unit 4, this almost simultaneous early expression of all the major viral proteins appears to suggest that both promoters are operational during this period. However, proof of this will require kinetic analysis of the viral messenger RNAs, since it is formally possible that a small proportion of the capsid proteins synthesized early in infection could be derived from mRNAs arising a t the promoter a t map unit 4 (Fig. 8, discussed in Section 11,C,2).Paradiso (1984) also found that in unsynchronized SV40-transformed human newborn kidney (NB) cells infected with H-1, de novo expression of all the major viral proteins was first detected simultaneously, in this case 9 hours after infection, and all reached peak levels of synthesis at around the same time, 18 hours postinfection. In contrast, Molitor et al. (1985) clearly demonstrated synthesis of the NS-1 protein of porcine parvovirus (PPV) 5-7 hours after infection of a n asynchronous population of swine testis (ST)cells, but was unable to detect capsid proteins until 9-11 hours postinfection. As in the previous studies, transcriptional data are not available, and detection of the protein species is indirect, relying on the ability of various antisera to precipitate the relevant proteins with equal efficiency. Nevertheless, the data appear dramatic, and it would seem that some viruses in some cell types can express and accumulate substantial levels of NS-1 prior to synthesis of detectable amounts of the capsid polypeptides. In all three viruses (MVM, H-1, and PPV) expression of the structural proteins appears to remain a t peak levels long after synthesis of NS-1 has started to decline. Such observations raise questions about the significance of the two separate transcription units known to be operational in the RV group viruses, but without precise kinetic transcription data, we can do little more than speculate on the potential for phasing gene expression from these two promoters during the viral life cycle. Parvovirus genomes appear to lack classical “early” and “late” transcription units, since it is doubtful if any RNA is made prior to synthesis of the parental

144

SUSAN F. COTMORE AND PETER TATTERSALL

complementary DNA strand. Moreover, in MVM and H-1, de novo synthesis of NS-1 and the capsid proteins is so tightly linked in time that, if these proteins are indeed the products of different transcription units early in infection, it is unlikely that the activity of one promoter could be entirely dependent upon interaction with a protein specified by the other. Nevertheless, in DNA transfection studies, synthesis of proteins encoded from the P38 promoter of H-1 was shown to be up-regulated 10- to 20-fold in NB cells by the simultaneous expression of NS-1 (Rhode, 1985~). It has also been suggested that NS-1 can down-regulate transcription from its own promoter (Rhode, 1985a),but if so kinetic studies of protein synthesis rates suggest that this is likely to be a very slow type of regulation which perhaps only operates when large concentrations of NS-1 are present in the cell. Another interesting possibility is that the nonstructural proteins might be able to influence transcription from cellular promoters. Early studies by Parris and Bates (1976) showed that the rates of RNA (and protein) synthesis in hydroxyurea-synchronized fetal bovine spleen cells were relatively depressed in the first 4 hours following infection with BPV, became elevated at 6 hours postinfection, presumably as a result of viral RNA synthesis, and then progressively decreased so that by 17 hours the rate of RNA synthesis was only 20%of control levels. Since one of the ways in which the virus could manipulate its environment for its own advantage would be to suppress expression of some, but not all, cellular genes, quantitative and qualitative analysis of host-specific transcription following infection might well prove informative, but no data are available at the present time. When naked DNA fragments containing either of the MVM promoters were transcribed in nuclear extracts in uitro, both promoters appeared to function with equal efficiency (Beard et al., 1985). However, when nucleoprotein complexes isolated from infected A9 cells, also digested with EcoRI to permit discrimination between the two products, were similarly transcribed, the promoter at map unit 4 was 10 times more active than that at map unit 38. Although in uitro experiments are difficult to interpret, these studies reinforce the idea that in uiuo the immediate protein environment of the transcription template is likely to influence profoundly the relative activity of the two promoters. Ben Asher and Aloni (1984) isolated transcription complexes from MVM-infected A9 cells, cut them with restriction enzymes, and incubated them in uitro in the presence of all four ribonucleoside triphosphates t o generate in uiuo-initiated, run-off transcripts. Under these conditions the products of both promoters once again appeared to be synthesized with equal efficiency. However, when uncleaved viral

AUTONOMOUSLY REPLICATING PARVOVIFWSES

145

transcription complexes or isolated nuclei were incubated in uitro they failed to give the expected full-length transcripts derived from the left-hand promoter, but instead gave a major band of 142 nucleotides mapping between positions 201 and 342 in the genome. A similar transcript has also been identified by this research group in uiuo (Resnekov and Aloni, 19851, and they suggested that in the infected cell both promoters may be equally active but that a regulatory mechanism which they call “attenuation” functions to depress synthesis of full-length transcripts arising from the left-hand promoter (by terminating them at residue 142) and thus permits the apparently more abundant expression of RNAs arising at map unit 38. They postulated that “attenuation” results from the tendency of transcripts arising from this region to fold into two mutually exclusive stem and loop structures, designated “attenuation” and “readthrough” in Fig. 17, one of which is followed immediately by a run of uridylic acid residues typical of a prokaryotic transcription termination signal. Although transcription and translation are not coupled in eukaryotes, as it is in the prokaryotic systems in which attenuation was first described (Yanofsky, 19811, Ben-Asher and Aloni (1984) suggested that a similar, albeit rather more sluggish, feedback mechanism could modulate MVM transcription if one of the viral gene products acts as an “attenuator” or “anti-attenuator” by somehow stabilizing one of the alternative RNA conformations. They suggested that the polymerase slows down at the stem and loop structure and pauses at the polyuridylic acid sequence. If the RNA folds into the “attenuator” configuration this could then render the template-transcript interaction exceptionally unstable, inducing termination and, perhaps release of the RNA. Interestingly, transcription complexes and nuclei used in this study were only isolated late in infection (after 24 hours), and although such a system could hardly result in rapid regulation (involving as it does the splicing and passage of one transcript to the cytoplasm, translation, and passage of the viral protein back to the transcription complex in the nucleus), it could be involved in down-regulating R 1 and R2 synthesis late in the lytic cycle. Ben Asher and Aloni (1984) suggested that the alternative stem and loop conformations discussed above might also be present at the 5’ end of full-length viral mRNA. This could influence translation of R 1 and R2 transcripts since in the attenuator conformation the AUG initiation codon used by NS-1 and NS-2 (nucleotide 261) is available for translation initiation, while in the readthrough conformation it is sequestered in the stem. By interacting with and stabilizing one or other of these configurations, viral proteins could theoretically influence translation of the nonstructural proteins, but this idea has yet to be

146

SUSAN F. COTMORE AND PETER TATTERSALL

A ATTENUATION

YY)

2?

GAUGAAGUUU U GUUUUUA U-AU G. c - ~ 4 0Termination

280-c uC

G.U

*U

G-C

sites

a; a,*::: 0 AG

G

-- 6 Kcol

C -G m-AA-’ C-GA c-G!330

270-AtG

2” G A

tC-G

c

G G

A

C A A

G-C U-A

177

G-C U-A 220-C - G U-A-240

,-PI

iATAA

U-A G -C

260-c

G -C U-A 30-U -A A-U A

CU A A

G

A

A

G

C 2P ACUUAUCUUUUCUUUCAu-AAGAGAGUAA 250

A - W A

G ‘ A A

AG--14Kcol

AG-- 19 Kcol

B READTHROUGH

A

A

G

G

U

U

A

u-290

G

U

U 2805-G

U

C A

G G-C U-A G-C U-A

T*TAA

I-mI

210

C-340

A

G

‘77

AUUUGUUUUUA 330

220-C-G U-A-240

$A : “A-U 260C-G C-G A-U A-U-310 A C ~ - ~ A

k - G A

G

A C C A A

u G

ACUUAUCUUUUCUUKA”-~AGAGAGUA+~-~A

250

320

FIG.17. Potential secondary structures at the 5’ end of RNAs synthesized from the P4 promoter of MVM(p).Transcripts initiating a t nucleotide 201 ? 5 (marked i-201) can fold into either the “attenuation” configuration (A) or the “readthrough” configuration (B).The attenuation structure contains a putative termination signal (marked 2 and 3) and transcription of the 142-nucleotide RNA is thought to cease a t the adjacent poly(U) sequence (marked “termination sites”).Full-length RNAs adopting this configuration in

AUTONOMOUSLY REPLICATING PARVOVIRUSES

147

explored experimentally. Certainly it seems possible that viral gene products could influence translation preferences and rates in the infected cell, but so far there have been no published studies into this aspect of host cell manipulation by the virus. As discussed in Section II,C,3 and Fig. 10, differential expression of VP-1 and VP-2 (and possible also of carboxy-terminally modified forms of NS-2) is regulated by splicing full-length nuclear precursor RNAs in two different ways. Although this appears to be a highly reliable system, we have no idea how it operates or how it invariably selects particular splice signals from the myriad such sequences which appear to be available. In uiuo the NS-1 protein of MVM accumulates in the nucleus of the infected cell where, early in infection, it is found in two major forms, one of which comigrates with its in uitro translation product (83 kDa) while the second, a phosphoprotein, migrates with a slightly higher apparent molecular weight, 84,000-85,000 (Cotmore and Tattersall, 1986b). Similar phosphorylated forms of NS-1 with abnormally high apparent molecular weights have also been identified in H-l- and PPV-infected cells (Paradiso, 1984; Molitor et al., 1985); in each case phosphoserine was shown to be the predominant phosphorylated amino acid, although phosphothreonine residues were also detected in the NS-1 of PPV. In PPV the unphosphorylated (or poorly phosphorylated) forms of NS-1 synthesized early in infection could be chased into the higher molecular weight form over the course of the next 1-2 hours (Molitor et al., 1985). Compared, for example, to the almost instantaneous cytoplasmic phosphorylation of the polyoma capsid protein VP-1 (Garcea et al., 19851, the phosphorylation of NS-1 seems a rather protracted process, suggesting that it may well occur after NS-1 has been translocated into the nucleus. The functional significance of this NS-1 modification is unknown, but it appears to be associated with events occurring early in infection, since phosphorylated forms of NS-1 appear soon after this molecule is first detected in the infected cell nucleus (Paradiso, 1984; s. F. Cotmore and P. Tattersall, unpublished observations), but are much less abundant late in infection. A truncated form of NS-1 (around 60 kDa) also accumulates in the nucleus of most infected cells late in infection (S. F. Cotmore and P. Tattersall, ~

~~

~

the cytoplasm leave the AUG codon (boldface) at nucleotide 261-263 exposed. In the readthrough configuration transcription does not terminate a t nucleotides 342-350, permitting synthesis of full-length RNAs. Cytoplasmic transcripts folding in this way sequester the AUG (nucleotides 261-263) used to initiate NS-1 and NS-2 translation and may thus interfere with the synthesis of these proteins. (Reproduced with permission from Ben-Asher and Aloni. 1984.)

148

SUSAN F. COTMORE AND PETER TATTERSALL

unpublished results). We do not know if this abbreviated molecule plays any essential role in the biology of the virus or if it is an inevitable by-product of one of the activities of full-length NS-1, but for MVM(p) the same processed form appears in every cell type examined so far. NS-1 probably carries out a number of quite different functions in uiuo. Frameshift mutations introduced into the NS-1 coding sequence of an infectious plasmid clone of MVM DNA prevented excision and replication of the viral sequences following transfection of this construct into eukaryotic cells, but similar mutations in the capsid genes, while inhibiting single-strand DNA and virion synthesis, permitted high levels of duplex DNA replication (Merchlinsky, 1984). This suggests that the NS-1 proteins play an essential role in the later stages of viral DNA replication, perhaps in part by cutting or nicking the DNA at specific sites, as discussed later. The NS-1 protein also appears to be able to up-regulate transcription from the viral promoter at map unit 38 (Rhode, 1 9 8 5 ~and, ) perhaps, to down-regulate the activity of both its own and a number of foreign eukaryotic promoters (Rhode, 1985a). Figure 15 shows that, 2 hours into S phase in highly synchronized, MVM-infected populations of A9 cells, the rate of cellular DNA replication begins to decline dramatically, although the peak of viral DNA replication is not destined to occur until much later. Viral proteins are fairly abundant in the cell by this time and it seems likely that one or more of them is responsible for this effect. The capsid polypeptides do not impair cell viability since Pintel and colleagues were able to isolate stable cell lines which both expressed these proteins and assembled them into virions (Pintel et al., 1984; LabieniecPintel and Pintel, 1986).However, repeated attempts to obtain cell lines expressing the nonstructural polypeptides have failed (Pintel et al., 1984; R. Moir, S. F. Cotmore, and P. Tattersall, unpublished results), and Rhode has shown that the cotransfection into eukaryotic cells of DNA encoding NS-1 together with constructs bearing various selectable genes effectively abolishes the outgrowth of colonies expressing the selectable marker (Rhode, 1985a). Thus NS-1 expression appears to impair the long-term viability of the host cell, but whether it does this by a direct effect on DNA replication or by disturbing cellular transcription remains t o be determined. The amino-terminal domains of NS-1 and NS-2 share 84 amino acids of protein sequence, but they each have unique carboxy-terminal peptides (Cotmore and Tattersall, 1986a). The cellular location and function(s) of NS-2 have yet to be determined, but it seems possible that these molecules may help to regulate one or more of the activities of NS-1 by competing for structures which interact with the common amino-terminal sequence, while

AUTONOMOUSLY REPLICATING PARVOVIRUSES

149

mediating individual, and perhaps antagonistic, regulatory functions via their dissimilar carboxy-terminal domains.

C. DNA Replication Viral DNA replication apparently passes through three distinct phases: synthesis of the parental complementary DNA strand, amplification of duplex DNA forms, and excision and concomitant packaging of progeny, single-strand viral DNA. As already discussed, synthesis of complementary strands is probably dependent on a cellular function expressed early in S phase. Exactly how long this function remains available has yet to be determined, but experiments with parasynchronized cells infected at various times in S (Siegl and Gautschi, 1973a) suggest that it persists for some hours. Although much of the input virus may never reach the nucleus, Wolter et al. (1980) showed that when high multiplicities of [3H]thymidine-labeled MVM(p) were used to infect A9 cells, by mid-S phase each cell contained an average of 50-100 copies of labeled double-strand DNA, most if not all of which had been derived by direct conversion of the input single strands. Such experiments require the use of relatively high levels of labeled virus, but as the virus can infect cells with single-hit kinetics (Tattersall and Bratton, 1983), it may be that in many successfully infected cells there are far fewer copies of duplex DNA available early in S phase to act, for example, as transcription templates. A9 cells, released from a single thymidine replication block and simultaneously infected with 15-20 PFU/cell of 32P-labeled MVM(p) virions, were monitored for conversion of input strand to duplex DNA by Ward and Dadachanji (1978). By 4 hours postinfection, 99% of the input virus remained in a single-strand form, but between 6 and 16 hours postinfection the percentage of input label present in double-strand DNA increased from 2 to 10% of the total added. Although cells parasynchronized in this way may gradually enter S over a rather prolonged period of time, de novo synthesis of complementary strands as late as 16 hours postrelease suggests that this must normally be a relatively protracted process and that duplex DNA is likely to continue accumulating in the nucleus over several hours at the beginning of S. Exposure of infected cells to 5-bromo-2’-deoxyuridine (BUdR) during this phase of the cell cycle blocked subsequent viral DNA replication, capsid synthesis, and production of progeny virus (Rhode, 1974; Wolter et al., 1980). However, the same inhibitor did not affect host DNA synthesis or the replication of duplex viral DNA if added later in S phase, and viral capsid synthesis and infectious virus production

150

SUSAN F. COTMORE AND PETER TATTERSALL

appeared almost normal. Since this analog is efficiently incorporated into complementary-strand DNA (Rhode, 19741, and BUdR-treated cells are as efficient as untreated cells in converting incoming singlestrand DNA to a duplex form (Wolter et al., 1980), the reported absence of viral antigen synthesis suggests that this inhibitor may well exert its effect by inhibiting transcription of the first duplex DNA forms. As seen in Fig, 15, major amplification of duplex DNA occurs relatively late in S, and generally continues for several hours after uninfected control cells have ceased DNA replication (Fig. 15; Parris and Bates, 1976; Hardt et al., 1983). Although it is formally possible that this form of synthesis is delayed simply because it is dependent upon prior accumulation in the nucleus of high levels of the viral nonstructural proteins (and any changes these products might effect on cellular processes), this explanation seems unlikely since restrictively infected cells still replicate MVM DNA to 520% of the normal level seen in productive infections (Spalholz et al., 1983) despite a massive reduction in viral transcription (discussed in Section 111,E).A more intriguing possibility is that this phase of DNA replication requires a second cellular function which is only expressed late in S phase, but at present there is no evidence of what this function might be, or what role it performs in the normal cell. The subsequent excision and displacement of single strands seem to be entirely dependent upon the availability of viral capsids (Muller and Siegl, 1983a; Merchlinsky, 1984) and are driven by ongoing viral DNA replication (Richards et al., 1977). However, provided these criteria are met, this mode of replication initiates soon after multimeric duplex synthesis has become well established (see Fig. 15). Infection of synchronized fetal bovine spleen cells with BPV resulted in changes in the levels and patterns of expression of the cellular DNA polymerases (Y and y, but not p, during the cell cycle (Pritchard et al., 1978). In particular, DNA polymerase (Y activity was threefold greater in infected versus mock-infected cells during the period of maximal viral DNA synthesis, and closely paralleled viral replication at other times, while DNA polymerase y activity remained slightly elevated throughout infection. In viuo all phases of viral DNA replication appear to be influenced by the DNA polymerase a inhibitor aphidicolin (Hardt et al., 1983; Robertson et al., 1984; Gunther et al., 1984; see Fig. 151, but such observations, while compatible with a role for this enzyme in rival replication, are subject to the criticism that they could be monitoring indirect effects of the polymerase, such as allowing progression through S phase, rather than its direct involvement in the synthesis of viral DNA. In vitro studies using specific

AUTONOMOUSLY REPLICATING PARVOVIRUSES

151

polymerase inhibitors and nuclear extracts (Pritchard et al., 1981; Gunther et al., 1984) or isolated nuclei (Kollek et al., 1982) from infected cells further implicated DNA polymerase a in the later stages of viral DNA replication, although a late role for DNA polymerase y was also suggested (Kollek et al., 1982). During cellular DNA replication, elongation of new DNA strands at the replication fork is believed to involve relatively continuous synthesis of the leading DNA strand and discontinuous, RNA-primed synthesis of the lagging strand. Both of these functions are thought to be carried out predominantly by DNA polymerase (Y acting in concert with various cofactors (Denhardt and Faust, 1985). Tseng et al. (1979) analyzed the structure of nascent, replicating H-1 DNA by labeling cells in uiuo with L3H1thymidine for various intervals at the time of maximum viral duplex DNA replication. Even with labeling periods of only 20-30 seconds, these authors were unable to detect the small DNA fragments, migrating at 4-5 S in denaturing gradients, which characterize discontinuous synthesis of mammalian cellular DNA and correspond to the Okazaki fragments of Escherichia coli and the coliphages (Okazaki et d., 1969). However, longer forms of newly replicated DNA were present which hybridized with approximately equal efficiency to both viral and complementary-strand DNA. This suggests that, at least late in infection, a relatively continuous mode of synthesis is used for replicating both strands of H-1. In uitro a number of polymerases of both prokaryotic and eukaryotic origin are able to use the base-paired 3’-hydroxyl group at the left end of viral DNA to initiate continuous synthesis of an essentially fulllength complementary strand (Bourguignon et al., 1976; Cotmore and Tattersall, 1984). Kollek and Goulian (1981) fractionated extracts of NB cells and human placenta looking for an activity which would mimic this conversion of single-strand H-1 DNA in uitro. The only such activity detected in this study was shown to be mediated by DNA polymerase y. Crude cell fractions enriched for DNA polymerase (Y or f3 did not catalyze the conversion, but the authors could not rule out the possibility that the in uitro conditions provided did not adequately mimic those normally encountered by these replicative enzymes in the infected cell. Faust et al. (1984) achieved a 4300-fold purification of a form of DNA polymerase a from Ehrlich ascites cells which was active in the ATP (GTP)-dependent conversion of MVM single-strand DNA to a duplex form. Purification of this polymerase complex was monitored using an assay based on the ability of the enzyme preparation to repair gaps which had been created by DNase I treatment of calf thymus DNA (Faust,, 1984), and it contained a tightly bound oligoribonucleo-

152

SUSAN F. COTMORE AND PETER TATTERSALL

tide polymerase, generally referred to as the “primase.” The complementary strand of MVM DNA synthesized in uitro using this complex was discontinuous (Faust et al., 1985). In all, 17 3’-hydroxyl termini were identified which were clustered at six sites, each of which was located 2-14 nucleotides upstream of CCAACC, CCACCC, or CCAATT sequences in the viral genome. Each individual fragment was shown to contain a n oligoribonucleotide chain averaging 5-7 nucleotides in length, and it was proposed that in this system hexanucleotides having the general formula C2A,,[C2-, or T21, termed JI sequences, are involved in terminating DNA synthesis and/or in the de novo initiation of RNA-primed chains by DNA polymerase a. Such sequences are peculiarly abundant in MVM. Computer analyses by J. W. Bodnar and D. C. Ward (personal communication) show that while the pentanucleotides ACCAA and AACCA each occur over 60 times in the MVM genome and concatameric forms of these oligomers are widespread, the same pentamers are not particularly abundant in a number of other viral genes or genomes or in various cellular genes. Moreover, in MVM not all such sequences are confined to regions near a primase initiation site (Faust et al., 1985). Since there is, as yet, no evidence for such markedly discontinuous synthesis of the complementary strand of MVM in uiuo,the exact significance of these sites must remain in question. Whichever enzyme is eventually shown to be responsible for this synthesis, such sequences remain of interest because many purified replicative enzymes exhibit similar patterns of arrest sites (Weaver and DePamphalis, 1982; Kaguni and Clayton, 1982).For example, of the five sites at which mouse mitochondria1 replication pauses in uiuo,three correspond to sites that arrest purified Drosophila DNA polymerase a in uitro (Kaguni and Clayton, 1982),but mitochondrial DNA synthesis is normally carried out in uiuo by DNA polymerase y (Clayton, 1982). Rommelaere and Ward (1982) showed that prior UV irradiation of the encapsidated single-strand MVM DNA used to infect normal A9 cells resulted in a dose-dependent, single-hit inhibition of replicative form (RF) formation due to the introduction of absolute blocks to the elongation of the newly synthesized complementary strand. However, exposing cells to UV light prior to infecting them with UV-irradiated MVM enhanced the fraction of input DNA which could be converted to full-length RF. Although the exact mechanism by which irradiated cells circumvent the block to polymerase activity in this system is still not clear, de novo protein synthesis was shown to be required to permit expression of the modified phenotype. Studies monitoring the reversion of a UV-treated temperature-sensitive mutant of H-1 (ts6) showed that prior exposure of the host cell to UV light, 2-nitrofuran deriva-

AUTONOMOUSLY REPLICATING PARVOVIRUSES

153

tives, or UV-treated, replication-defective forms of SV40 DNA (Cornelis et al., 1981,1982; Su et al., 1981) promoted recovery of H-1 replication, but that the progeny virus exhibited an enhanced mutation frequency. Subsequently, it was shown that cells in which this mutation-prone S.O.S. replication mechanism had been induced were able to overcome the absolute block to H-1 complementary-strand synthesis which results from the presence of apurinic or apyrimidinic sites in the DNA, such as might be induced in uiuo by heating or following the decay of lesions produced by ionizing radiation and various chemical mutagens (Cornelis et al., 1985). Apurinic or apyrimidinic sites can also occur spontaneously in the cell under physiological conditions and such mutagenic escape mechanisms may well contribute to the rather frequent emergence of new parvovirus strains. Ward and Dadachanji (1978) showed that in uiuo full-length complementary strands of MVM are synthesized from the 3’-hydroxyl group of the input viral strands to yield covalently closed hairpin duplex molecules. Early in infection (6 or 8 hours postinfection in parasynchronized cells) much of the input-labeled, duplex DNA in the cell existed in this closed form, but by 16 hours postinfection nicks had been introduced into all but 18%of the converted input strands. These same authors demonstrated that 32P-labeled viral replicative form DNA extracted from asynchronous cultures of A9 cells 38 hours after infection with MVM(p) migrated in neutral agarose gels as singlestranded forms and as monomeric, dimeric, and tetrameric duplex molecules. However, in denaturating gels these forms collapsed to single strands (64% of the labeled DNA) and monomer duplexes (32%), showing that the multimeric replicative form (RF) intermediates contained nicks which divided the DNA into unit length or twice unit length fragments. The restriction endonuclease EcoRI cuts the MVM genome twice, giving two easily distinguished terminal fragments, but EcoRI digestion of gel-purified monomer RF yielded two forms of each of the terminal fragments which had different electrophoretic mobilities. When the digestion mixture was heat-denatured, quench-cooled, and reelectrophoresed only the terminal fragments with the higher mobility from each end of the genome spontaneously reannealed, allowing them to migrate in their previous position, while the other form collapsed to the mobility expected for single-strand molecules of this size. These were termed the “extended” and “turnaround” forms of the termini. The presence of spontaneously renaturing fragments from both the right and left ends of the genome is compatible with the suggestion, to be developed later, that progeny DNA synthesis occurs exclusively via a self-priming mechanism. The larger, noncovalently closed forms of each terminus might then be explained by the intro-

154

SUSAN F. COTMORE AND PETER TATTERSALL

duction of a site-specific nick somewhere in the 3‘ palindrome, and strand-displacement synthesis from the newly exposed 3’-hydroxyl to complete the shorter strand, thus yielding two complementary copies of the terminal hairpin. At the left end of the genome (the 3’ end of the viral strand) extended and turnaround forms were found with approximately equal frequency, but at the right end there were around nine copies of the extended form for each turnaround fragment. EcoRI digestion of dimer and tetramer RF molecules yielded similar extended and turnaround forms from both ends of the genome, but additional duplex fragments containing head-to-head copies of the left-hand EcoRI fragment (designated “dimer bridge”) and tail-to-tail copies of the right-hand fragment (“tetramer bridge”) were also identified. Upon heating these palindromic duplex fragments melted to comigrate with turnaround fragments from the appropriate end of monomer RF. Thus dimers and tetramers are palindromic arrangements of unit-length duplex DNA molecules arranged in head-to-head (viral 3’ to viral 3’) and tail-to-tail (viral 5’ to viral 5 ’ ) orientations, respectively. Pulsechase experiments further revealed that these intermediates were metabolically active and eventually gave rise to single-strand viral DNA (Ward and Dadachanji, 1978). Although all infected cells also contain a population of partially replicated molecules with regions of both single-strand and duplex DNA (Tattersall et al., 1973), these molecules are usually heterogeneous and do not migrate upon gel electrophoresis as distinct bands which are easily distinguishable from the major replicative forms. However, in some infected cell types predominant partially replicated dimeric forms have been observed. One such molecule, observed by Faust and Gloor (1984) in MVM-infected Ehrlich ascites and 3T6 cells, appeared to migrate as an 8-kb duplex on neutral gels but gradually dissociated to give 5-kb duplex molecules and, more or less, unit-length single strands. Although we do not know if these structures are true replicative intermediates, any model for parvoviral replication needs to be able to account for their formation. A small proportion of relaxed circles and unit-length, double-stranded linear molecules with lassolike structures at their right-hand termini were also observed in one elsctron microscopic study of intracellular MVM DNA (Bratosin et al., 1979), but the origin and significance of these molecules remain unclear. In uiuo initiation and termination of H-1 DNA replication appear to occur close to the viral termini (Rhode, 1977; Singer and Rhode, 1977). Type I defective particles of MVM have been described (Faust and Ward, 1979) in which as much as 90-95% of the internal, singlestrand region of the wild-type genome has been deleted, leaving only

AUTONOMOUSLY REPLICATING PARVOVIRUSES

155

sequences derived from within about 5 map units (250 nucleotides) of the molecular termini. These defectives, which invariably retain both of the duplex terminal palindromes intact, were selectively amplified during serial undiluted passage and therefore contain all the critical cis-acting functions necessary for the replication of MVM. In addition, plasmid clones of MVM DNA lacking substantial regions from either of the two termini failed to give replicating linear molecules following transfection into eukaryotic cells (Merchlinsky et al., 1983) and thus, for MVM, both unique terminal palindromes must be retained relatively intact. The experimental data described here have been drawn almost exclusively from early studies on MVM, and it is perhaps surprising that although many virus serotypes have been shown to have, for example, monomer and dimer duplex replicative forms with a mixture of covalently closed and extended-form termini, detailed analyses of the structure and progression of replicative intermediates are not available for any other virus in the genus. Since this information is essential in the development of models for viral DNA replication, all such models rely heavily on the known structure of MVM and cannot be adequately cross-checked by reference to other viruses. This is unfortunate since sequence analysis suggests that all members of the genus share a common ancestral form (Shade et al., 1986), and it would be surprising if they employed very different replication strategies. Despite the considerable advances in molecular technology seen during the last few years, there have been relatively few observations made since the mid-1970s which radically influence our view of the viral replication process, and the model proposed at that time (Tattersall and Ward, 1976) still forms the basis for more recent proposals. However, two significant additions must be made in the light of DNA sequence studies into the structure of the termini of viral and replicative form DNA from MVM (Astell et al., 1985). As discussed earlier in Section II,B, the termini of all autonomous parvoviruses analyzed to date are imperfect palindromes, containing a few asymmetrical nucleotides which are mispaired in the hairpin form. The presence of these mismatched residues allowed the identification of two sequence orientations termed “flip” and “flop,” in the right-hand termini of both viral and RF molecules of MVM (see Fig. 3 and Section II,B), while only a single sequence was present at the left end of either DNA molecule (Astell et al., 1985). Molecularly cloned plasmid forms of the MVM genome containing either of the two possible 5’configurations gave wild-type virus with equal proportions of both forms upon transfection into A9 or 324K cells (Merchlinsky, 1984), showing that replication of a single 5’ DNA sequence inevitably regenerates both se-

156

SUSAN F. COTMORE AND PETER TATTERSALL

quences. This, therefore, implies that in vivo the mechanism used to replicate the 5’ (right-hand) end of viral DNA must be different from, and less conservative than, that used to copy the 3’ end. Since the right-hand end of RF DNA was also shown to be at least 18 nucleotides longer than that of viral DNA (Astell et al., 1983, the nicking site, and possibly also the enzymes involved in generating the 5’ end of viral and RF DNA could be different. Similar length differences are not seen at the 3’ (left-hand) ends of viral and RF DNA. The limited genetic capacity of the parvoviral genome and the fact that these viruses depend upon host functions expresses transiently during the S phase of the cell cycle imply that for many aspects of their DNA replication the viruses depend upon the synthetic machinery of the cell. Replicating the ends of linear DNA molecules so that a single, perfect copy of each telomere is obtained at every division presents particular problems, and since these rather complex structures are known t o be of overriding importance for parvoviral DNA replication, most models focus on explaining their regeneration. A “rolling-hairpin” system of self-priming at the genomic termini (Tattersall and Ward, 1976), coupled with a hairpin transfer mechanism which copies the palindromic sequence at the 5‘ end of one strand on to the 3’ end of its complementary strand, forms the basis for most models. A series of site-specific nicks and ligations must also be postulated and, since few details are available about either these or the enzymes and auxiliary proteins involved, an almost endless variety of marginally different mechanisms can be proposed. A particularly comprehensive and provocative “modified rollinghairpin” model for MVM replication has been put forward by Astell and colleagues (1982, 1985) based on the DNA sequence studies mentioned above, and we have incorporated the main features of that model here since it provides a plausible explanation for many of the observed structures and makes novel predictions about the nature of the enzymes used to generate the 3’ termini of the viral strands. Here we have further extended this replication scheme to include some alternative interpretations of currently available data. According to this model (Fig. 18) the 3’ terminal palindrome of the incoming V strand (step 1) is used by a host polymerase to prime synthesis of a compleFIG. 18. Modified “rolling-hairpin”model for parvoviral DNA replication. The autonomous parvovirus genome is represented by a series of sequence blocks denoted by letters. Each upper- and lowercase letter denotes sequences complementary to one another. The 3‘ end is shown on the left of the parental genome in the top left-hand corner, and is represented by the letter A. The letters B and G denote the asymmetrical central portions of the left- and right-hand terminal palindromes, respectively. Vpar and Vprog

157

AUTONOMOUSLY REPLICATING PARVOVIRUSES 4.

B

BC

Q

T

7

0

Vmr

A

C

G

e,

C

A

B(

V

Qe f E F

V

0

f

)G

f

)G

C

ELA

)G E

V

0

w

-

F

P e fdi G

C

A

C

A

BC

V

0

P

~ ~ 0 ............

E

V

B

0 9 A f

Qef

e6

G F E

C

C

A b o

C

V

A b o

f

E

F

g -e

C

Vpr

e

m

t

)P

f

XI

E F

B

C

V 4

C

E F g f e

QTf 4

'b B

G F E

V

A 0 b9 A

E F a t e

C

A b

6 .

\1;

t t

E F gg ff ee

E F

o@

A

C

e f G FE,

0

V

E F g f e ' O

B(

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

0 9 A I.''

e f G F E

4

V

@V

CC

A b

0 %

denote parental and progeny viral single-strand sequences, respectively, whereas the complementary strand is represented by the letter C.Sequences at the 5' end of mature virion DNA are represented by the letter fi while the letter e denotes the extra 18-base sequence found on the 5' end of this strand in RF DNA molecules. The thick lines represent newly synthesized DNA, with the arrowhead denoting the 5' to 3' direction of DNA synthesis. The symbols T and P represent terminal protein/site-specific nucleases, and x represents the topoisomerase-like reaction, discussed in the text.

e

158

SUSAN F. COTMORE AND PETER TATTERSALL

mentary strand, giving rise to a monomer duplex RF which is covalently closed at the left-hand end. Astell et al. (1985) suggested that the 5’ hairpin is then displaced and copied directly (as in steps 4 and 51, although it seems equally possible that this step could be preceded by ligation of the 5’ and 3’ ends of the molecule (step 2), giving rise to a transient “collapsed-closed-circular”intermediate. We postulate that this intermediate could be the substrate for the sequencespecific nicking reaction, shown in step 3, which reopens the molecule at a point 18 nucleotides inboard of the original viral strand 5’ end. For simplicity, we propose that this reaction is catalyzed by the RF terminal protein denoted by the circled T, to be discussed below, which remains covalently attached to the new 5’ end. The 3’ OH at this nick is then used as a primer for the stranddisplacement synthesis, shown in steps 4 and 5, producing an RF molecule with a covalently closed turnaround terminus at the left-hand end and an extended-form right-hand end, including the terminal protein attached outboard of the extra 18-bp segment. Strand-displacement synthesis, such as that postulated for steps 4 and 5 , may require the use of single-strand DNA binding proteins, but such proteins have yet to be identified in viuo. At step 6 , a rabbit-eared structure is formed and displacement synthesis occurs (step 7) to produce a dimer RF (step 8)in which the 3‘ ends of the viral strands overlap in the center of the 10-kb linear duplex molecule, while the 5’ ends of the viral strands are located at the molecular termini. As will be seen, involvement of a dimer intermediate at this point is essential for the regeneration of viral strands with a single sequence orientation at their 3’ termini. To achieve this, Astell and colleagues (1985) proposed that a nick is introduced, as shown in step 9, specifically in the parental V strand sequence opposite the 3’ end of the newly synthesized V strand, by a topoisomerase-like terminal protein, represented by the T in a square. Since it is crucial for the model that this nick is confined to the parental V strand, the authors proposed that strand preference may be determined by an asymmetry which results from mismatched nucleotides in the center of the main stem of palindrome as detailed in Fig. 19, and discussed below. The 3‘ OH of this nick is then used for the extension of the parental V strand in a 5’ to 3’ direction, displacing the original 3’ palindrome of the parental V strand, as shown in step 10, enabling it to fold into a hairpin configuration, and regenerating an exact copy of this sequence in its original position. The authors proposed that the same site-specific nickase remains covalently attached to the 5’ end of the nicked DNA during displacement and progressive reformation of the hairpin structure until it encounters a second copy of its DNA recognition sequence

AUTONOMOUSLY REPLICATING PARVOVIRUSES

159

located at the 3’ end of the viral progeny strand (marked with a circled

X in the dimer produced by step 10).It then introduces a second nick at

this new site and reseals a complete 3’ hairpin (step 11, right monomer), while itself being transferred to the 5’ end of the complementary strand (step 11, left monomer). Such a mechanism would be compatible with the observed 1:l incidence of extended and turnaround forms at the left end of monomer RF molecules, and the invariable occurrence of unit-length nicks in dimer RF, and would generate progeny in which the 3’ palindrome of the viral strand contained a single DNA sequence. The topoisomerase invoked for this reaction would produce both the duplicate, site-specific nicks and the necessary ligation and would be left attached to the complementary RF strand. At least a proportion of the 5’ ends of both viral and complementary strands of RF DNA are known to be covalently attached to protein, although relatively little is known about these terminal polypeptides. Moreover reactions of this type are known to be carried out by the gene A protein of the bacteriophage +X174 (Sanhueza and Eisenberg, 19841, although a comparable enzyme has yet to be described in eukaryotic cells. At this point molecules such as that shown as the right-hand product of step 11 can recycle into the pool of replicating RF by undergoing steps 6 onward, as indicated by the long arrow at the right side of Fig. 18. An important result of this synthetic loop is the observed generation of the two forms of the right-hand palindrome, designated FGf and Fgf for “flip” and “flop”, at the 5’ end of progeny viral strands (underlined in Fig. 181, and the conservation of the single sequence, designated ABa, at the 3‘ ends of all mature viral strands. Molecules such as that shown as the left-hand product of step 11 can either regenerate monomers and dimers by rabbit-ear formation at either end of the genome, or go on to produce packaged genomes as shown in step 12. In this process maturation of the correct viral 5’ ends might occur via site-specific nicking of the viral strand by a putative viral strand terminal protein (circled P ) at a position 18 nucleotides inboard of the RF terminal protein. While the 18-nucleotide fragment remaining attached to the RF terminal protein could then prime displacement synthesis of viral strands, the P protein might direct packaging of displaced progeny viral strands into preformed capsids, a process which is known to be concomitant with and dependent upon viral strand DNA synthesis (Richards et al., 1977; Muller and Siegl, 1983a). The P protein would subsequently be cleaved off during final maturation of progeny virions. An interesting alternative to this packaging mechanism is suggested by the recent finding that the 5’ ends of package virion DNA are “ragged,” even if recently derived from a molecularly cloned viral

160

SUSAN F. COTMORE AND PETER TATTERSALL

genome (Gardiner and Tattersall, 1987). This terminal heterogeneity is biased, in that while the majority of 5’ ends lie 18 or 17 nucleotides inboard of the RF terminus, there is also a series of minor species extending down to 14 nucleotides inboard of the RF terminus, but no cuts are made further inboard than 18 nucleotides. This suggests that the terminal protein might itself loop back on the packaged viral single strand and free itself by cleaving to produce the mature 5’ end in a length-dependent rather than sequence-specific fashion. The predominant cut at 18 nucleotides may represent both the optimal and the maximum distance required for such looping back. In this case we would suggest that the primer required for the initiation of stranddisplacement synthesis of progeny viral strands is provided by rabbitear formation as shown in step 6 of Fig. 18. If this mechanism is correct, any viral strands still attached to their terminal proteins would not show the 18-base 5’ deletion first described by Astell et al. (1982). However, several alternative mechanisms can also be postulated at this point, since there are no sufficiently detailed short-term, pulse-labeling studies of the very earliest replicative form intermediates and, as discussed below, we know so little about the proteins associated with the 5’ ends of single-strand and replicative form DNA. Astell and colleagues (1985) proposed that the same terminal protein catalyzes the processing events at both 3’ and 5’ ends of the genome. The major reason for this is the presence of two short, but fairly conserved regions of DNA sequence near the putative nicking sites at both termini (Astell et al., 1982). As seen in Fig. 19 one of these putative recognition sequences (CTTATCA at the 3’ end and CTATTCA at the 5’ end) lies at the actual cut site, while the other, ACCAAC, is located, 13 or 5 bases, respectively, upstream of this site and could represent the initial binding site of the protein. Since the mechanism proposed for replication of the 3’ ends of viral DNA requires that the enzyme binds to its recognition sequence in the parental V strand before it encounters that in the progeny V strand, the authors invoked differential control of binding by sequence differences upstream of the ACCAAC sequence which occur because of the imperfect nature of this palindrome. As seen in Fig. 3B and discussed in Section 11, the left-hand terminal sequence is unusually well conserved in a number of different parvovirus genomes and, when folded into a Y configuration, invariably gives a two- to three-base mismatched bubble at around nucleotides 25-26 and 89-91. Thus six bases upstream of the ACCAAC sequence in the parental V strand is the sequence GAA, while upstream of this box in the progeny V strand GAA is replaced by the dinucleotide TC. It is supposed that the first sequence could be absolutely required for initial binding of the en-

161

AUTONOMOUSLY REPLICATING PARVOVIRUSES par

F E

GC

a 8 A

C

e

f

A b a

Vprog

E

F

I

C

9

C

..OTOC10T011TOCl-CTTOTACCAACCAOTCAAOATTTTTACTATTCOCCll~TCCCTCllATTTOOTT. ... 5 ‘ I A

5 : . .ClCOTClCTTlCOT-OAACATOOTTOO~ClOTTCTlAlAAT~ATlAOCOO~ClOOOAOTTTllACCAl. .3’

3‘.

v V 1 5’. ..llllCTCllCCAlOICTlCTOTCTATTClOTOllCCAlCTOAlCClTTlOTlTTlCTlTOTTTTT 3‘. .. T T T T O l O T T O O T T C T O l T O----_ lClO~TAlOT * C ----A C T T O4-O T T O l C T T ~IO T l A T C l T l l T O l T l C ~ l l A l ~ t 5140

j

4910

4910

*pros

FIG. 19. Proposed nick sites in MVM RF. A diagram of the duplex dimer replicative intermediate of MVM(p), the product of step 8 in Fig. 18, shows the organization of the parental (par) and progeny (prog) viral-sense (V)strands and their complementary sequences (C). Viral strands are oriented with their 3’ ends overlapping in the middle of the duplex and their 5’ ends at the termini. Nucleotide sequences of the boxed regions, marked i, ii, and iii, are shown below. It is suggested that the nickase recognizes the sequences underlined with broken arrows. The nick is made near the 3’ end of the sequence 5’-CTTATCA-3’ (i and ii) or 5’-CTATTCA-3’ (iii), but to accomplish this the enzyme must first bind to the DNA via the sequence 5‘-ACCAAC-3’, recognized in conjunction with the triplet 5‘-GAA-3’(both in the nicked strand). Asymmetry in the 3’ palindrome replaces this GAA triplet with 5’-TC-3’at the 3’ end of the V progeny strand (ii), inhibiting de novo binding of the 3’ nicking enzyme. Thus the initial cut at the 3’ end of viral DNA in the dimer would always be made in the V parental strand, where the recognition sequences are perfect (marked “yes” in i), and the equivalent cut site in the V progeny strand (marked “no” in ii) would only be recognized once the nickase is brought into close proximity by virtue of its attachment to the opposite DNA strand. Panel iii shows the arrangement of similar sequences around the inboard end of the right-hand RF palindrome (V),as discussed in the text. (V) denotes the 5’end of mature virion single-strand DNA.

162

SUSAN F. COTMORE AND PETER TATTERSALL

zyme. At the 5’ end of the viral strand the sequence GAA lies immediately upstream of the ACCAAC sequence and so is once again available as part of the binding site. In viruses such as AAV-2 where the DNA sequences near the two potential nick sites in the center of the dimer molecule are identical, the nickase could theoretically bind to either strand of the dimer with equal frequency and would thus generate termini in both “flip” and “flop” orientations. There have been reports of proteins associated with the 5’ ends of RF DNA from a number of autonomous parvoviruses (Hayward et al., 1978; Revie et al., 1979; Astell et al., 1982; Chow et al., 1986; Wobbe and Mitra, 19851, and with the 5’ ends of LuIII and MVM single-strand viral DNA (Muller and Seigl, 1983a; Chow et al., 1986). A candidate protein for the RF-specific nickase of MVM was identified by Astell and her colleagues (1982)to be covalently attached to the 5’ ends of the extended forms of both the plus and minus DNA strands. The acid and base stability of the protein-DNA linkage suggested that it was not mediated via a serine, threonine, or amide residue, and the authors suggested that a phosphodiester bond between the 5’-terminal deoxynucleotide and a tyrosine residue in the protein would provide the observed stability. Iodinated DNA-protein complexes released a 60-to 65-kDa protein upon DNase digestion (Astell et al., 1982). Revie et al. (1979) had previously estimated that a protein associated with H-1 RF molecules had a molecular weight of 60,000-80,000 from a decrease in the buoyant density of RF DNA associated with the presence of bound protein. However, more recent experiments by P. R. Paradiso (personal communication) identified a 95-kDa protein tightly associated with H-1 RF DNA, and Wobbe and Mitra (1985) demonstrated both a 90and a 40-kDa species specifically complexed with each of the terminal restriction fragments of RV RF DNA, which were not associated with internal restriction fragments from the same molecules. The latter is clearly a very important control given the relative abundance of viral proteins with molecular weights around 60,000-80,000 present in the nuclei of infected cells and, presumably, associated with the replicating DNA. As yet there is insufficient evidence to assess whether these terminal proteins are encoded by the virus or the host cell.

D . Virion Assembly and Maturation The proteins reputed to be associated with viral single-strand DNA, and possibly involved in packaging, have proved even more elusive than those associated with intracellular RF DNA molecules.. Buoyant density studies (Revie et al., 1979) failed to reveal a protein attached to H-1 viral DNA, and those investigators who have evidence for such a

AUTONOMOUSLY REPLICATING PARVOVIRUSES

163

protein associated with MVM single strands (Chow et al., 1986; M. Gunther and P. Tattersall, unpublished observations) report that the fraction of DNA complexed with protein varies substantially from preparation to preparation. Muller and Siegl (1983a) had previously shown that, unlike most mature forms of virial DNA, newly synthesized viral strands of LuIII prepared in an in uitro replication system were unable to penetrate a neutral agarose gel, suggesting that they were complexed with protein. Thus, it may be that removal of any associated protein frequently occurs soon after encapsidation, but can vary in efficiency from one infection to another or from one cell type to another. At present, we do not know if the same polypeptides are associated with both viral and RF DNA, or what the exact functions of the viral strand terminal proteins are, since, although i t seems likely that they are involved in cutting the DNA at specific sites during replication, they may equally well influence packaging of the nascent single strands or maturation of the newly assembled virion. Using Brij-58 extracts of LuIII-infected cells, Muller and Siegl (1983a) were able to show that pulse-labeled viral DNA could be recovered from 110 S particles within minutes of its synthesis, but that the newly assembled full particles banded exclusively at 1.44 g/cm3 and were relatively unstable in cesium chloride. Despite extensive proteinase K digestion, a large proportion of viral strands extracted from such particles within 10 minutes of their assembly showed abnormally low electrophoretic mobilities, compatible with (although not proving) the persistence of a covalently associated peptide. At later times this aberrant mobility pattern was lost. This suggests that the terminal protein may remain attached to the 5’ end of the viral strand as it is encapsidated, but is rapidly removed thereafter. Maturation was also reflected by a change in virus stability in cesium chloride and a shift in density from 1.44 g/cm3 (heavy fulls) to 1.41 g/cm3 (light fulls) around 10-15 minutes after encapsidation, but it is not clear whether these changes were dependent upon, or even related to, removal of the terminal peptide from the viral strand. Similar changes in the buoyant density of full virions have been observed during the maturation of several other parvoviruses in uiuo (Richards et al., 1977; Paradiso, 1981), but these shifts are generally reported to occur over a few hours, rather than minutes. As yet no correlations have been made between these in uiuo density shifts and the presence or absence of a covalent DNA-protein linkage, and, as discussed previously, we know very little about the nature of the structural changes which accompany virion maturation. Muller and Siegl were also able to identify a transient population of precursor nucleoprotein complexes which could be chased into 110 S particles. These complexes sedimented between

164

SUSAN F. COTMORE AND PETER TATTERSALL

70 and 100 S, were unstable in cesium chloride, and dissociated in the presence of 1M NaCl (Muller and Siegl, 1983b). Electron microscopy of fixed complexes revealed structures consisting of DNA threads associated with capsid-like particles. Viruses such as MVM and H-1 package predominantly (<99%) a single-sense DNA strand which is the complement of that encoded into RNA, and are thus, technically, negative-strand DNA viruses. Although this characteristic was initially presumed to be a diagnostic feature of the entire genus, more recent studies show that some autonomous parvoviruses, notably LuIII and B19, encapside both strands with approximately equal efficiency (Muller and Siegl, 1983a; Summers et al., 1983; Bates et al., 1984; Cotmore and Tattersall, 1984). Since LuIII packages strands of both polarity, while H-1 grown in the same cell type only encapsidates the minus strand (Bates et al., 1984), the selection of strands is predominantly virus coded rather than host specified. Type I defective genomes of MVM, which may contain little more than the genomic termini, are efficiently encapsidated (Faust and Ward, 1979) and thus any cis-acting DNA sequences or threedimensional structures which control this process must be located in, or near, these regions. Moreover, Type I1 defectives, which only contain sequences from the 5‘ end of viral DNA, are also packaged (Faust and Ward, 19791, suggesting that they also contain any such control signals. Candidate sequences have yet to be identified, but the fact that MVM packages even 1%of the plus strand indicates that the controlling signals may be somewhat ambiguous. Since virion strand synthesis and packaging are concomitant processes (Richards et at., 1977; Muller and Siegl, 1983a), encapsidation presumably proceeds in a 5’ to 3’ direction, and it is therefore reasonable to speculate that specific packaging signals occur in the right-hand palindrome of MVM which are similar, but not identical, to structures in the left-hand palindrome. Thus encapsidation would usually start at the 5’ end of the minus or virion strand, but occasionally the analogous signal at the left end of the genome would be recognized, and the plus strand packaged instead, once again in a 5’ to 3’ direction. Such a mechanism would predict that viruses such as AAV-2 and B19, which have similar sequences at both termini, would package both strands with equal frequency, as in fact occurs. However, electron-microscopic analysis of self-annealed LuIII virion DNA suggests that the genomic termini of this virus are unique (Bates et al., 19841, and it will be interesting to see if detailed analysis of the terminal regions of this virus reveals any common structures or sequences which could serve as packaging signals. As discussed above, proteins covalently associated with the 5’ end of the viral DNA could also mediate strand selection and encapsida-

AUTONOMOUSLY REPLICATING PARVOVIRUSES

165

tion, but at present we can only speculate on these various possibilities. Relatively few of the preformed viral capsids are ever destined to package DNA and mature into infectious virus particles. Richards et al. (1977) observed that assembly of MVM particles in semisynchronized RT-7 cells began around 10 hours after the beginning of S phase, peaked 4-10 hours later, and was virtually over by 18 hours after the beginning of cellular DNA synthesis. Eventually a mixture of empty capsids and full virions is released from the cell, following nuclear degeneration and rupture of the plasma membrane by mechanisms as yet undefined. Infection of a productive host cell culture gives rise t o wholesale cell death. However, for many cell types, continued culture allows for the outgrowth of colonies of virus-resistant cells. Resistant cells may have several different phenotypes with respect to the step of virus replication which is affected, and these will be explored in the next section.

E. Nonproductive Virus-Host Cell Interaction Two characteristics of parvoviral infection, the requirement for host factors expressed only during S phase and the inability of the virus to stimulate resting cells to initiate DNA synthesis, were first established using cell culture systems as discussed previously, and are reflected in parvoviral disease of the whole animal as a preference for actively proliferating tissues as sites of attack. Not all virus-host cell interactions are productive, however, and some of these nonproductive interactions can still be lytic. A number of different types of nonproductive interactions have been investigated and in some cases one can learn much about particular virus-host relationships by studying systems in which the virus fails to grow. As discussed previously, productive infection is initiated by adsorption of the virion to specific cell surface receptors. The presence of functional receptors on the cell surface is apparently under developmental control, since some differentiated cell types lack them and are completely resistant to virus infection (Spalholz and Tattersall, 1983). When the requirement for transition through S phase is met in a receptor positive, productive host cell, infection results in wholesale cell death. However, for many cell types, continued culture allows for the outgrowth of colonies of virus-resistant cells (Linser and Armentrout, 1978; Stanick and Tattersall, 1987). Resistant cells may have several different phenotypes depending upon which step in virus replication is affected. Common phenotypes include stable loss of expression of the cell surface virus receptor, inability to support viral

166

SUSAN F. COTMORE AND PETER TATTERSALL

transcription, and inability to carry out an early step in either virus penetration or transport of input virus to the nucleus (Linser and Armentrout, 1978; D. Stanick and P. Tattersall, unpublished observations). Although dissection of the host genes affected in such resistant cell variants is in its infancy, much information has already been gained about the cellular contribution to virus growth by studying the infection of naturally nonproductive cell types in uitro. Such virushost interactions can usefully be divided into three categories: restrictive, abortive, and cryptic. Restrictive infection, which is perhaps the best characterized of these, is exemplified by studies on the prototype (p) and immunosuppressive (i) variants of MVM. These virus strains differ from one another by being restricted for growth in one another’s productive host cell type, fibroblasts, and T lymphocytes, respectively (Tattersall and Bratton, 1983). The two viruses are serologically indistinguishable and have been designated allotropic variants of the MVM serotype, to indicate their distinct target cell specificities. Somatic cell hybrids between these two cell types are productive hosts for both viruses (Tattersall and Bratton, 1983). These results suggest that each restrictive host cell lacks a factor necessary for the productive replication of the restricted virus strain, and that these factors are expressed as a function of the differentiated state of the cell. Biochemical studies have demonstrated that these host factors operate at an intracellular level, and that both virus strains bind to, and compete with one another for, functional receptors on both cell types (Spalholz and Tattersall, 1983; Spahlholz et al., 1983). The first steps of virus growth are relatively unimpaired in restrictive infection (Spalholz and Tattersall, 1983),and viral genomes are replicated to 5-20% of the equivalent levels found in productive interaction, but no viral capsid proteins, progeny virions, or singlestranded DNA are detectable (Tattersall and Bratton, 1983; Spalholz et al., 1983). Recent studies show that the primary defect in restrictive infection is a dramatic decrease in the initiation of viral transcription (Spalholz et al., 1987), suggesting that the viral locus controlling strain-dependent target cell specificity, called the allotropic determinant, is a regulatory element which interacts with a developmentally regulated host factor to turn on viral gene expression. It is quite surprising, therefore, that genetic mapping of the allotropic determinant, using recombinants constructed between the two genomes, has located the fibrotropic determinant in MVM(p) to a region located between nucleotides 3522 and 3757, that is, in the VP-2 coding sequence of the capsid gene (E. M. Gardiner and P. Tattersall, unpublished observations). This 235-nucleotide sequence, though only 4.5% of the genome,

AUTONOMOUSLY REPLICATING PARVOVIRUSES

167

contains nearly 10% of the single-base differences between MVM(p) and MVM(i), and interestingly includes, in addition to 12 silent changes, one conservative amino acid difference and two of the three radical amino acid changes between the capsid genes of these two viruses. In all restrictive virus-cell combinations examined in culture, there is a small subpopulation of cells, continually being replenished during the growth of the culture, which is able to support fully productive, lytic virus replication. The remainder of the culture restricts virus replication at the level of transcription, and infection has little or no effect on cell growth rate or viability (Tattersall and Bratton, 1983). Infection of such cultures with a restricted virus can therefore result in a persistent carrier culture, which continually produces low levels of virus without apparent cytopathic effect. During passage of such cultures the low level of virus replication can give rise to a virus mutant, called an extended host range (hr)mutant, which has gained the ability to transcribe and replicate its genome in the formerly restrictive host cell type (Ron et al., 1984). Such hr mutants may also preexist in a virus stock, and can be isolated directly as plaques on appropriate, normally restrictive cell monolayers. We have recently mapped the mutation in one such isolate, hr 101, a mutant of MVM(i) which is capable of growing productively in mouse fibroblasts, and found it to lie within the same small genomic region as the fibrotropic determinant of MVM(p), discussed above. Preliminary sequence data suggest that the mutation results in a radical amino acid change in the VP-1IVP-2 common region of the coat protein, although it is not the same change as either radical amino acid difference between MVM(p) and MVM(i) (R. Moir and P. Tattersall, unpublished observations). Interestingly, Parrish and Carmichael (1986) have reported the isolation and mapping of a CPV host range mutant, selected by continued growth in feline cells, which has lost its ability to grow in canine cells. This mutation lies in the region of the CPV coat protein gene equivalent to that spanned by the two radical changes between MVM(p) and MVM(i),and also results in a radical amino acid substitution. It is not known, however, at what point in the normal replicative cycle the CPV mutant is blocked in canine cells. Taken together, these results strongly suggest that the capsid protein plays a role in the target cell specificity of the virus. Since in MVM this is mediated after entry of the virus but before expression of the coat protein gene, this implies that it is the incoming capsid which interacts with a developmentally regulated host cell factor to initiate viral transcription. The appearance of hr mutants in a persistently infected restrictive cell culture results in a cytopathic crisis, with concomitant massive

168

SUSAN F. COTMORE AND PETER TATTERSALL

virus production. As this interaction is now productive, there is often a rapid selection of cellular variants which are resistant to the new virus mutant. It appears that many of these cellular variants can be restrictive for the new virus and the whole cycle can begin over again, and can even occur several times (Ron and Tal, 1985). In some types of nonproductive virus-cell interactions, especially those across species barriers (for instance, the infection of normal human fibroblast cell lines with rodent parvoviruses), significant expression of viral nonstructural and capsid antigens has been observed in the absence of detectable progeny virus production (de Foresta et al., 1985a,b; P. Tattersall and S. F. Cotmore, unpublished observations). These interactions are termed abortive infections. In abortive infection the virus fails to establish its own DNA replication program, and the infection may or may not result in cell death, depending upon the particular combination of virus strain and cell line. All cells in the culture appear to be able to sustain viral transcription and antigen expression, often to levels found in productively infected cells (de Foresta et al.,1985b; P. Tattersall and S. F. Cotmore, unpublished observations). This type of interaction is of particular interest since transformation of the cell by viral agents, such as SV40, convert the abortive interaction to a fully productive one. Dual SV40 and parvovirus infections are not productive, however, implying that transient SV40 early gene expression is not sufficient for reversal of the abortive phenotype, and that the establishment of the fully transformed state is required for these cells to support a productive parvovirus growth cycle (P. Tattersall and J. Bratton, unpublished observations). Some mouse cell strains, both selected resistant mutants and apparently normal lines, possess this transformation-sensitive phenotype with respect to MVM infection (Mousset and Rommelaere, 1982; J. Rommelaere, personal communication). The final distinct type of interaction between virus and host cell occurs as a direct result of the requirement for the host to traverse S phase and the inability of the virus to stimulate resting cells to do so. Thus, infection of naturally resting but otherwise productive host cells leads to a situation we denote cryptic infection. Examples of this are the infection of unstimulated porcine peripheral blood lymphocytes with PPV (Paul et al., 1979) or murine splenic T cells with the immunosuppressive strain of MVM (T. Molitor, D. Winograd, A. Smith, and P. Tattersall, unpublished observations). In these cases, activation of cryptically infected cells with concanavalin A or antigen results in active virus replication, leading to lysis of the activated cell. Infection may then spread among other susceptible, cycling cells in the population.

AUTONOMOUSLY REPLICATING PARVOVIRUSES

169

The ability of a parvovirus to infect and kill its target cell, and produce infectious progeny virions, therefore depends upon the species and differentiated phenotype of the host cell, as well as its proliferative status.

ACKNOWLEDGMENTS We are very grateful to the numerous colleagues who communicated ideas and information prior to publication and hope that these contributions are adequately referenced in the text. We also thank D. Greenberg for her help in processing the manuscript. The authors were supported by Public Health Service Grants CA 16038 and CA 29303 from the National Cancer Institute.

REFERENCES Aasted, B., Race, R. E., and Bloom, M. E. (1984).J . Virol.51, 7-13. Anderson, M. J.,Lewis, E., Kidd, I. M., Hall, S. M., and Cohen, B. J. (1984).J . Hyg. 93, 85-93. Anderson, M. J., Kidd, I. M., and Morgan-Capner, P. (1985).k n c e t 2,663. Astell, C. R., Smith, M., Chow, M. B., and Ward, D. C. (1979).Cell 17, 691-703. Astell, C. R., Thomson, M., Chow, M. B., and Ward, D. C. (1982).Cold Spring Harbor Symp. Quant. Biol.47, 751-762. Astell, C. R., Thomson, M., Merchlinsky, M., and Ward, D. C. (1983).Nucleic Acids Res. 11, 999-1018. Astell, C. R., Chow, M. B., and Ward, D. C. (1985).J . Virol. 54, 171-177. Astell, C. R., Gardiner, E. M., and Tattersall, P. J. (1986).J . Virol. 57,656-669. Baer, P. N.,Garrington, G. E., and Kilham, L. (1971).J . Gerontol. 26,373-377. Banerjee, P. T., Olson, W. H., Allison, D. P., Bates, R. C., Snyder, C. E., and Mitra, S. (1983).J . Mol. Biol. 166, 257-272. Bates, R. C., Snyder, C. E., Banerjee, P. T., and Mitra, S. (1984).J. Viol. 49,319-324. Beard, P., Bruggman, H., and Doerig, C. (1985).EMBO Parvovirus Workshop, Grangeneuve, Switzerland. Ben-Asher, E., and Aloni, Y.(1984).J. Virol. 52, 266-276. Bergs, V. V. (1969).Cancer Res. 29,1669-1672. Berns, K. I., ed. (1984).‘The Parvoviruses.” Plenum, New York. Bloom, M. E., Race, R. E., and Wolfinbarger, J. B. (1982).J . Virol. 43,608-616. Boissy, R., and Astell, C. R. (1985).Gene 35, 179-185. Bonnard, G. D.,Manders, E. K., Campbell, D. A., Herberman, R. B., and Collins, M. J. (1976).J . Exp. Med. 143, 187-205. Bourguignon, G. J., Tattersall, P. J., and Ward, D. C. (1976).J. Virol. 20, 290-306. Bratosin, S.,Laub, O., Tal, J., and Aloni, Y. (1979).Proc. Natl. Acad. Sci. U S A . 76, 4289-4293. Brown, D. D., and Salzman, L. A. (1984).J . Virol.49, 1018-1020. Burtonboy, G., Delferiere, N., Surleraux, M., and Bodeus, M. (1985).EMBO Parvovirus Workshop, Grangeneuve, Switzerland. Campbell, D. A,, Staal, S. P., Manders, E. K., Bonnard, G. D., Oldham, R. K., Salzman, L. A,, and Herberman, R. B. (1977).Cell. Zmmunol. 33,378-391. Carlson, J., Rushlow, K., Maxwell, I., Maxwell, F., Winston, S., and Hahn, W. (1985).J. Virol. 55, 574-582. Casper, D. L. D., and Klug, A. (1962).Cold Spring Harbor Symp. Quunt. Bwl. 27, 1. Cheung, A., Hoggan, M. D., Hauswirth, W. W., and Berns, K.I. (1980).J . Virol.33,739748.

170

SUSAN F. COTMORE A N D PETER TATTERSALL

Chow, M., Bodnar, J. W., Polvino-Bodnar, M., and Ward, D. C. (1986).J . Virol.57,10941104. Clayton, D. A. (1982).Cell 28,693-705. Clewley, J. P. (1985). EMBO Parvovirus Workshop, Grangeneuve, Switzerland. Clinton, G. M., and Hayashi, M. (1976). Virology 74,57-63. Coleman, G. L., Jacoby, R. J., Bhatt, P. N., Smith, A. L., and Jonas, A. M. (1983). Vet. Pathol. 20, 49-56. Cornelis, J. J., Su, Z. Z., Ward, D. C., and Rommelaere, J. (1981).Proc. Natl. Acad. Sci. U S A . 78,4480-4484. Cornelis, J., Su, Z. Z., Dinsart, C., and Rommelaere, J. (1982).Biochimie 64, 677-680. Cornelis, J. J., Avalosse, B., Vos, J. M., Su, Z. Z., and Rommelaere, J. (1985).EMBO Parvovirus Workshop, Grangeneuve, Switzerland. Cotmore, S. F., and Tattersall, P. (1984).Science 226, 1161-1165. Cotmore, S. F., and Tattersall, P. (1986a). J . Virol. 58, 724-732. Cotmore, S. F., and Tattersall, P. (1986b). Virus Res. 4, 243-250. Cotmore, S. F., Sturzenbecker, L. J., and Tattersall, P. (1983). Virology 129, 333-343. Cotmore, S. F., McKie, V. C., Anderson, L. J., Astell, C. R., and Tattersall, P. (1986).J . Virol. 60, 548-557. De Foresta, F., Chen, Y. Q., Hertoghs, J. Avalosse, B., Buponchel, N., Cornelis, J. J., and Rommelaere, J. (1985a). EMBO Parvovirus Workshop, Grangeneuve, Switzerland. De Foresta, F., Deleys-Hertoghs, J., Cornelis, J., Klein, B., and Rommelaere, J. (1985b). CJ1. Soc. Biol. (Paris) 179, 276-282. Denhardt, D. T., and Faust, E. A. (1985).BioEssays 2, 148-154. Engers, H. D., Louis, J. A., Zubler, R. H., and Hirt, B. (1981).J . Zmmunol. 127, 22802285. Faust, E. A. (1984). Bull. Can. Biochem. Soc. XXI, 10-17. Faust, E. A., and Gloor, G. (1984). J . Virol. 49, 621-625. Faust, E. A., and Ward, D. C. (1979).J . Virol. 32, 276-292. Faust, E. A., Gloor, G., Macintyre, M.-F., and Nagy, R. (1984). Biochim. Biophys. Acta 781, 216-224. Faust, E. A., Nagy, R., and Davey, S. K. (1985). Proc. Natl. Acad. Sci. U.S.A.82,40234027. Ferm, V. H., and Kilham, L. (1965). Grolvth 29, 7-16. Fischer, H. D., Didgson, J. B., Hughes, S., and Engel, J. D. (1984).Proc. Natl. Acad. Sci. U S A . 81,2733-2737. Galton, M., and Kilham, L. (1966).Proc. SOC.Exp. Biol. Med. 122, 18-22. Garcea, R. L., Ballmer-Hofer, K., and Benjamin, T. L. (1985).J . Virol. 54, 311-316. Gardiner, E. M., and Tattersall, P. (1987).In preparation. Gay, N. J., and Walker, J . E. (1983).Nature (London) 301, 262-264. Gill, D. R., Hatfull, G. F., and Salmond, G. P. C. (1986). Mol. Gen. Genet. 205,134-145. Gorbalenya, A. E., Blinov, V. M., and Koonin, E. V. (1985).Mol. Genet. 11.30-36. Green, M. R., Lebovitz, R. M., and Roeder, R. G. (1979). Cell 17,967-977. Gunther, M., Prigent-Tiravy, S., and Wicker, R. (1984). J . Gen. Virol. 65, 2021-2031. Hadlow, W. J., Race, R. E., and Kennedy, R. C. (1983). Infect. Immun. 41, 1016-1023. Hardt, N., Dinsart, C., Spadari, S., Pedrali-Noy, G., and Rommelaere, J . (1983).J . Gen. Virol. 64,1991-1998. Hayward, G. S., Bujard, H., and Gunther, M. (1978).In “Replication of Mammalian Parvoviruses” (D.C. Ward and P. Tattersall, eds.), pp. 327-340. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Hearing, P., and Schenk, T. (1983). Cell 33,695-703. Hogan, A., and Faust, E. A. (1984). Mol. Cell. Biol. 4, 2239-2242.

AUTONOMOUSLY REPLICATING PARVOVIRUSES

171

Jongeneel, C. V., Sahli, R., McMaster, G. K., and Hirt, B. (1986).J. Virol. 59,564-573. Kaguni, L. S.,and Clayton, D. A. (1982).Proc. Natl. Acad. Sci. U S A . 79, 983-987. Kalderon, D., Richardson, W. D., Markham, A. F., and Smith, A. E. (1984).Nature (London) 311,33-38. Karasaki, S. (1966).J . Ultrastruct. Res. 16, 109. Kelly, D. C.,and Elliot, R. M. (1977).J. Virol. 21, 408-410. Khoury, G., and Gruss, P. (1983).Cell 33, 313-314. Kilham, L. (1961).Virology 13, 141-143. Kilham, L., and Margolis, G. (1975).Prog. Med. Virol. 20, 113-143. Kilham, L., Margolis, G., and Colby, E. D. (1970).J. Infect. Dis. 121, 648-652. Kollek, R., and Goulian, M. (1981).Proc. Natl. Acad. Sci. U.S.A. 78, 6206-6210. Kollek, R., Tseng, B. Y., and Goulian, M. (1982).J. Virol. 41, 982-989. Kozak, M. (1986).Cell 44,283-292. Labieniec-Pintel, L., and Pintel, D. (1986).J. Virol. 57, 1163-1167. Lederman, M., Bates, R. C., and Stout, E. R. (1983).J. Virol. 48, 10-17. Lederman, M., Bates, R. C., and Stout, E. R. (1985).EMBO Parvovirus workshop, Grangeneuve, Switzerland. Ley, K. D., and Tobey, R. A. (1970).J. Cell Biol. 47,453. Linser, P., and Armentrout, R. W. (1978). In “Replication of Mammalian Parvoviruses” (D. C. Ward and P. Tattersall, eds.), pp. 151-160. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Linser, P. Bruning, H., and Armentrout, R. W. (1977).J. Virol. 24,211-221. McMaster, G. K., Beard, P., Engers, H. D., and Hirt, B. (1981).J. Viml. 38, 317-326. Margolis, G., and Kilham, L. (1975).Prog. Med. Virol. 20, 144-179. Matsunaga, Y., and Matsuno, S. (1983).J. Virol. 45,627-633. Matsunaga, Y.,Matsuno, S., and Mukoyama, J. (1977).Infect. Zmmun. 18, 495-500. Merchlinsky, M. J. (1984).Ph.D. thesis, Yale University. Merchlinsky, M. J., Tattersall, P. J., Leary, J. J., Cotmore, S. F., Gardiner, E. M., and Ward, D. C. (1983).J. Virol. 47, 227-232. Miller, R. A., Ward, D. C., and Ruddle, F. H. (1977).J. Cell. Physwl. 91, 393-402. Mitra, R., Wali, T., Valdez, V., Fabisch, P., and Salzman, L. A. (1983).Virology 125,349360. Mohanty, S. B., and Bachmann, P. A. (1974).Infect. Immun. 9,762-763. Molitor, T. W.,Joo, H. S., and Collett, M. S. (1983).J. Virol. 45, 842-854. Molitor, T.W., Joo, H. S., and Collett, M. S. (1985).J. Virol. 55, 554-559. Morgan, W. R., and Ward, D. C. (1986).J. Virol. 60, 1170-1174. Mortimer, P. P., Humphries, R. K., Moore, J. G., Purcell, R. H., and Young, N. S. (1983). Nature (London) 302, 426-629. Mount, S. M. (1982).Nucleic Acids Res. 10,459-472. Mousset, S.,and Rommelaere, J. (1982).Nature (London) 300, 537-539. Muller, D.-E., and Siegl, G. (1983a).J. Gen. Virol. 64, 1043-1054. Muller, D.-E., and Siegl, G. (1983b).J. Gen. Virol. 64, 1055-1067. Okazaki, R., Okazaki, K., Sakake, K., Sugimoto, K., Kainuma, R., Sugino, A., and Iwatsuki, N. (1969).Cold Spring Harbor Symp. Quant. Biol. 33, 129. Paradiso, P. R. (1981).J. Virol. 39,800-807. Paradiso, P. R. (1983).J. Virol. 46,94-102. Paradiso, P. R. (1984).J. Virol. 52, 82-87. Paradiso, P. R., Williams, K. R., and Constantino, R. L. (1984).J. Virol. 52, 77-81. Parris, D.S.,and Bates, R. C. (1976).Virology 73, 72-78. Parrish, C.R., and Carmichael, L. E. (1983).Virology 129,401-414. Parrish, C . R., and Carmichael, L. E. (1986).Virology 148, 121-132.

172

SUSAN F. COTMORE AND PETER TATTERSALL

Pattison, J. R., Jones, S. E., Hodgson, J., Davis, L. R., White, J. M., Stroud, C. E., and Murtaza, L. (1981).Lancet 1, 664-665. Paul, P. S.,Mengeling, W. L., and Brown, T. T., Jr. (1979).Infect. Immun. 25, 10031007. Pedrali-Noy, G., Spadari, S., Miller-Faures, A., Miller, A. 0. A., Kruppa, J., and Koch, G. (1980).Nucleic Acids Res. 8, 377-387. Peterson, J. L., Dale, R. M. K., Karess, R., Leonard, D., and Ward, D. C. (1978).I n “Replication of Mammalian Parvoviruses” (D. C. Ward and P. Tattersall, eds.), pp. 431-445.Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Pintel, D., and Pintel, L. L. (1985). EMBO Parvovirus Workshop, Grangeneuve, Switzerland. Pintel, D.,Dadachanji, D., Astell, C. R., and Ward, D. C. (1983).Nucleic Acids Res. 11, 1019-1038. Pintel, D., Merchlinsky, M. J., and Ward, D. C. (1984).J. Virol. 52,320-327. Porter, D. D., and Cho, H. J. (1980).Compr. Virol. 16,233-256. Pritchard, C., Bates, R. C., and Stout, E. R. (1978).J. Virol. 27, 258-261. Pritchard, C., Stout, E. R., and Bates, R. C. (1981).J. Virol. 37, 352-362. Resnekov, O., and Aloni, Y. (1985). EMBO Parvovirus Workshop, Grangeneuve, Switzerland. Revie, D., Tseng, B. Y., Graftsom, R. H., and Goulian, M. (1979).Proc. Natl. Acad. Sci. U S A . 76, 5539-5543. Rhode, S. L., I11 (1973).J. Virol. 11, 856-861. Rhode, S.L.,I11 (1974).J . Virol. 14, 791-801. Rhode, S. L., I11 (1977).J. Virol. 21, 694. Rhode, S. L.,111 (1978).J . Virol. 27, 347-356. Rhode, S. L., I11 (1985a).EMBO Parvovirus Workshop, Grangeneuve, Switzerland. Rhode, S. L., I11 (1985b).J. Virol. 54, 630-633. Rhode, S. L., 111 (1985~). J . Virol. 55, 886-889. Rhode, S. L., I11 and Klaassen, B. (1982).J. Virol. 41, 990-999. Rhode, S. L., 111, and Paradiso, P. R. (1983).J. Virol. 45, 173-184. Richards, R. G., and Armentrout, R. W. (1979).J. Virol. 30,397-399. Richards, R., Linser, P., and Annentrout, R. W. (1977).J. Virol. 22,778-793. Robertson, A. T., Stout, E. R., and Bates, R. C. (1984).J. Virol. 49, 652-657. Rommelaere, J.,and Ward, D. C. (1982).Nucleic Acids Res. 10,2577-2596. Rommelaere, J., Vos, J.-M., Cornelis, J. J., and Ward, D. C. (1981).Photochem. Photobiol. 33,845-854. Ron, D., and Tal, J. (1985).J. Virol. 55, 424-430. Ron, D., Tattersall, P., and Tal, J . (1984).J. Virol. 52, 63-69. RufTolo, P. R., Margolis, G., and Kilham, L. (1966).Am. J . Pathol. 49, 795-824. Saarinen, U.M., Chorba, T. L., Tattersall, P., Young, N. S., Anderson, L. J., Palmer, E., and Coccia, P. F. (1986).Blood 67, 1411-1417. Sahli, R., McMaster, G. K.,and Hirt, B. (1985).Nucleic Acids Res. 13, 3617-3633. Samulski, R. J., Srivastava, A., Berns, K. I., and Muzczka, N. (1983).Cell 33, 135-143. Sanheuza, S.,and Eisenberg, S. (1984).Proc. Natl. Acad. Sci. U S A . 81, 4285-4289. Schull, B. C., Lederman, M., Chen, K. C., Moses, E. A., Stout, E. R., and Bates, R. C. (1985).EMBO Parvovirus Workshop, Grangeneuve, Switzerland. Shade, R. O., Blundell, M. C., Cotmore, S. F., Tattersall, P., and Astell, C. R. (1986).J. Virol. 58, 921-936. Siegl, G.(1984a).In “The Parvoviruses” (K. I. Berns, ed.), Chap. 8, pp. 297-362.Plenum, New York. Siegl, G. (1984b). In “The Parvoviruses” (K. I. Berns, ed.), Chap. 9,pp. 363-388.Plenum, New York.

AUTONOMOUSLY REPLICATING PARVOVIRUSES

173

Siegl, G., and Gautschi, M. (1973a).Arch. Ges. Virusforsch. 40, 105-118. Siegl, G., and Gautschi, M. (1973b).Arch. Ges. Virusforsch. 40, 119-127. Siegl, G., Bates, R. C., Berns, K. I., Carter, B. J., Kelly, D. C., Kurstak, E., and Tattersall, P. (1985). Intervirology 23, 61-73. Singer, I. I., and Rhode, S. L., I11 (1977). J . Virol. 21, 724. Smith, S., and Halling, S. M. (1984). Gene 29, 263-269. Spalholz, B. A., Bratton, J., Ward, D. C., and Tattersall, P. (1983).~In “Tumor Viruses and Differentiation” (A. Levine and E. Scolnick, eds.), pp. 315-328. Liss, New York. Spalholz, B. A., and Tattersall, P. (1983). J. Virol. 46, 937-943. Spalholz, B. A., Ward, D. C., and Tattersall, P. (1987). In preparation. Stanick, D., and Tattersall, P. (1987).In preparation. Su, Z. Z., Cornelis, J. J., and Rommelaere, J. (1981). Carcinogenesk 2, 1039-1043. Summers, J., Jones, S. E., and Anderson, M. J. (1983). J . Gen. Virol. 64, 2527-2532. Tattersall, P. (1972). J. Virol. 10, 586-590. Tattersall, P. (1978a).In “Replication of Mammalian Parvoviruses” (D. C. Ward and P. Tattersall, eds.), pp. 53-72. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Tattersall, P. (1978b).In “Replication of Mammalian Parvoviruses” (D. C. Ward and P. Tattersall, eds.), pp. 131-149. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Tattersall, P., and Bratton, J. (1983). J . Virol. 46, 944-955. Tattersall, P., and Cotmore, S. F. (1986).In “Viral and Mycoplasma Infections of Laboratory Rodents: Effect on Biomedical Research” (P. N. Bhatt, R. 0. Jacoby, H. C. Morse, and A. New, eds.). pp. 305-348. Academic Press, Orlando, Florida. Tattersall, P., and Ward, D. C. (1976). Nature (London) 263, 106-109. Tattersall, P., Crawford, L. V., and Shatkin, A. J. (1973). J. Virol. 12, 1446. Tattersall, P., Cawte, P. J., Shatkin, A. J., and Ward, D. C. (1976). J . Virol. 20,273-289. Tattersall, P., Shatkin, A. J., and Ward, D. C. (1977). J . Mol. Biol. 111, 375-394. Tennant, R. W., Layman, K. R., and Hand, R. E., J r . (1969).J . Virol. 4, 872-878. Toolan, H. W. (1960). Science 131, 1446-1448. Toolan, H. W. (1967). Nature (London) 214, 1036. Toolan, H. W. (1978).I n “Replication of Mammalian Parvoviruses” (D. C. Ward and P. Tattersall, eds.), pp. 161-176. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Toolan, H. W., and Ledinko, N. (1968). Virology 35,475-478. Toolan, H. W., Rhode, S. L., 111, and Gierthy, J. F. (1982). Cancer Res. 42, 2552-2555. Tseng, B. Y., Grafstrom, R. H., Revie, D., Oertel, W., and Goulian, M. (1979). Cold Spring Harbor Symp. Quant. Biol. 43, 263-270. Vasquez, C., and Brailovsky, C. (1965). Exp. Mol. Pathol. 4, 130. Walker, J. E. Saraste, M., Runswick, M. J., and Gay, N. J. (1982). EMBO J . 1, 945951. Ward, D. C., and Dadachanji, D. (1978).In “Replication of Mammalian Parvoviruses” (D. C. Ward and P. Tattersall, eds.), pp. 297-313. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Ward, D. C., and Tattersall, P. (1982).In “The Mouse in Biomedical Research” (H. L. Foster, J . D. Small, and J. G. Fox, eds.), Vol. 2, pp. 313-332. Academic Press, New York. Weaver, D. T., and DePamphilis, M. L. (1982). J . Biol. Chem. 257,2075-2086. Weiher, H., Konig, M., and Gruss, P. (1983). Science 219, 626-631. White, J., Keilian, M., and Helenius, A. (1983). Q. Reu. Biophys. 16, 151-195. Wobbe, C. R., and Mitra, S. (1985). Proc. Natl. Acad. Sci. U.S.A. 82, 633543339. Wobbe, C. R., Mitra, S., and Ramakrishnan, V. (1984). Biochemistry 23, 6565-6569.

174

SUSAN F. COTMORE AND PETER TATTERSALL

Wolter, S., Richards, R., and Armentrout, R. W. (1980). Biochim. Biophys. Acta 608, 420-431. Yanofsky, C. (1981). Nature fLondon) 289, 751-758. Yoshiyuki, Y., Katoh, I., Copeland, T. D., and Oroszlan, S. (1985). Proc. Natl. Acud. Sci. U S A . 82, 1618-1622. Young, N., Mortimer, P. P., Moore, J. G., and Humphries, R. K. (1984). J . Clin. Znuest. 73,224-230.

ADVANCES IN VIRUS RESEARCH, VOL. 33

REGULATION OF TRANSLATION BY POLIOVIRUS Nahum Sonenberg Department of Biochemistry and McGill Cancer Center McGill Univenity Montreal, Quebec, Canada H3G 1Y6

I. INTRODUCTION Viral infection of mammalian cells often results in the shutoff of macromolecular synthesis such as transcription and translation of host mRNAs (Bablanian, 1975; for recent reviews see Comprehensive Virology Vol. 19, 1984). These effects on host metabolism are generally believed to be of importance for a productive viral infection. Inhibition of cellular mRNA translation after viral infection is a highly discriminatory event in many systems (e.g., poliovirus, adenovirus, vaccinia) whereby translation of host mRNA is selectively suppressed. Elucidation of the mechanisms underlying this translational selectivity is of importance for the understanding of virus-host interactions and for designing strategies to combat virus infection. Unfortunately, there is to date only limited information to provide insights into these mechanisms, with the possible exception of poliovirus. Apart from their obvious relevance to virology, studies of the shutoff of host protein synthesis after viral infection should aid in elucidating certain aspects of normal cell physiology. The aim of this review is to discuss the latest developments in the understanding of the process by which poliovirus selectively inhibits host protein synthesis. Several reviews have been published during the past 20 years concerning the effects of picornavirus infection on host protein synthesis (Baltimore, 1969; Lucas-Lenard, 1979; Ehrenfeld, 1982), including a recent excellent review by Ehrenfeld (1984), which summarizes relevant work up to 1983. This review will emphasize more recent developments, especially with regard to the function of cap binding proteins in translation initiation and their modification following poliovirus infection. There is currently a general agreement about the mechanism by which poliovirus infection effects cessation of host protein synthesis. Consequently, I will emphasize the work that led to the model for shutoff, which best fits the available data, i.e., that a cap binding protein function which is an essential activity for 175

Copyright 0 1987 by Academic Press, h e . All rights of reproduction in any form reserved.

176

NAHUM SONENBERG

translation of cellular mRNAs, but somehow not required for poliovirus translation, is inactivated following poliovirus infection. The discussion will deal with possible viral mediators of the shutoff phenomenon. Several alternative models proposed to explain the shutoff phenomenon will be evaluated with respect to their likelihood of being applicable to poliovirus infection.

11. CAP BINDING PROTEINS OF EUKARYOTIC mRNAs A description of the mechanism by which poliovirus inhibits host mRNA translation requires an understanding of the proteins that interact with the 5' cap structure of cellular mRNAs during translation initiation. Table I lists the eukaryotic initiation factors and their probable function in translation.

TABLE I EUKARYOTIC INITIATION FACTORS AND THEIRPROBABLE FUNCTION IN TRANSLATION INITIATION^

Initiation factor eIF-1 eIF-2 eIF-2B eIF-3 eIF-4A eIF-4B eIF-4C eIF-4D eIF-4E (24K-CBP, CBP I) eIF-4F (CBP 11, CBP complex) eIF-5 eIF-6

Subunit molecular mass (kDa)

15 35, 50, 55 34, 40, 55, 65, 82 -10 subunits 28,000-160,000 50 80 17 15 24

24, 50, 220

150 25.5

Activity ? Ternary complex formation eIF-2 recycling

Subunit anti-association; binding to 40 S subunits mRNA binding to 40 S subunit; ATPase activity mRNA binding to 40 S subunit 60 S subunit joining 60 S subunit joining 5' cap recognition (subunit of eIF-4F) mRNA binding to 40 S subunit ATPase activity Release of eIF-2 and eIF-3; ribosome-dependent GTPase Subunit anti-association

A recent review on eukaryotic translation was published in the Annual Review of Biochemistry (Moldave, 1985).

TRANSLATION BY POLIOVIRUS

177

A . Identification of Cap Binding Proteins A “cap” structure, m7GpppX (where X = any nucleotide), is present at the 5‘ of all eukaryotic cytoplasmic (nonorganelle) mRNAs and most eukaryotic viral mRNAs, picornaviral and some plant mRNAs being notable exceptions (Shatkin, 1976). Numerous studies have demonstrated that the cap structure facilitates stable complex formation between 40 S ribosomal subunits and mRNA during translation initiation (for reviews see Shatkin, 1976, 1985; Filipowicz, 1978; Banerjee, 1980).It was anticipated that cap function would prove to be mediated via a cap-specific mRNA-protein interaction. Several groups have attempted to identify the polypeptides that interact specifically with the cap structure during mRNA translation. Early studies suggested that eukayrotic initiation factors eIF-2 and -4B specifically and directly bind the mRNA cap structure (Kaempfer et al., 1978; Shafritz et al., 1976, respectively). These studies were based on the ability of these IF to form cap analog-sensitive complexes with mRNAs that were retained on nitrocellulose filters. These studies were, however, criticized because of nonspecific inhibitory effects that cap analogs exhibited on RNA-protein complex formation (Sonenberg and Shatkin, 1978). In addition, with this technique it is difficult to assess the contribution of minor contaminants in the IF preparations to mRNA binding activity. A more direct approach to identify proteins that bind a t or near the cap structure was developed by Sonenberg and Shatkin (1977), who used periodate-oxidized [3Hlmethyl-labeled reovirus mRNA to promote chemical cross-linking of IF to mRNA. Schiff bases formed between the reactive dialdehyde of the m7G group, and free amino groups of protein can be stabilized by reduction with NaBH,CN. Cross-linked mRNAprotein complexes were treated with ribonuclease to degrade all but the cap portion and adjacent nucleotides of the mRNA. The labeled polypeptides were then resolved by SDS-polyacrylamide gel electrophoresis and visualized by fluorography. Using this approach, Sonenberg et al. (1978)initially identified a 24-kDa polypeptide in the high-salt wash of rabbit reticulocyte ribosomes (this fraction will be referred to as crude IF preparation throughout this review), which could be specifically crosslinked to the oxidized cap structure of several viral mRNAs. This polypeptide was consequently termed the 24K-CBP (cap binding protein) and will be referred to as such throughout this review. In subsequent publications it has been referred to as CBP I (Tahara et al., 1981) and eIF-4E (Altmann et al., 1985). More recently, Pelletier and Sonenberg (1985a) have shown that UV irradiation of IF preparation in the

178

NAHUM SONENBERG

presence of mRNA induced specific cross-linking of the 24K-CBP to the mRNA cap structure. A 24-kDa polypeptide could also become chemically cross-linked to oxidized m7GMP (Sonenberg et al., 1978) and to a benzophenone derivative of m7GTP by UV irradiation (Patzelt et al., 1983). A polypeptide of similar mobility on SDS-polyacrylamide gels and with identical cross-linking characteristics has been detected in IF preparations from mouse Ehrlich ascites cells (Sonenberg et al., 19781, human HeLa cells (Hansen and Ehrenfeld, 1981; Lee and Sonenberg, 19821, and other cell lines (I. Edery, F. Rozen, and N. Sonenberg, unpublished results). In addition, yeast (Altmann et al., 1985) and wheat germ (Lax et al., 1985; Seal et al., 1986) contain a similar-sized CBP. When IF preparations from several different sources (Sonenberg et al., 1978; Hansen and Ehrenfeld, 1981; Lee and Sonenberg, 1982) are cross-linked to the mRNA oxidized m7G group in the absence of ATPMg2+, the only polypeptide that cross-links specifically to the cap structure is the 24K-CBP. However, when cross-linking is performed in the presence of ATP-Mg2+,polypeptides of 50 and 80 kDa become covalently linked to the cap structure, in a cap-dependent manner (Sonenberg, 1981; Sonenberg et al., 1981). It is of importance that cross-linking of these polypeptides to an inosine-substituted reovirus mRNA can occur in the absence of exogenously added ATP-Mg2 (Lee et al., 1983), the possible significance of which will be discussed below (Section 11,B). In addition, two other polypeptides were reported to become cross-linked in an ATP-Mg2 -dependent manner: a 28-kDa polypeptide in IF preparations from rabbit reticulocytes and HeLa cells, and a 32-kDa polypeptide which is unique to HeLa cell preparations (Sonenberg, 1981; Lee and Sonenberg, 1982). Whereas little is known regarding the latter polypeptides, the identity of the 50- and 80-kDa polypeptides has been established. Using purified IF, Grifo et al. (1982) demonstrated that eIF-4A (50 kDa) and eIF-4B (80 kDa) when present together, but not by themselves, can cross-link to mRNA in a cap-specific, ATP-Mg2 -dependent fashion. Consequently they suggested that the cross-linked 50- and 80-kDa polypeptides in IF preparations are eIF-4A and eIF-4B, respectively. This suggestion was verified for the 50-kDa polypeptide by Edery et al. (1983) who showed that monoclonal antibodies raised against eIF-4A immonoprecipitated the cross-linked 50-kDa polypeptide in crude IF preparations. In more recent experiments, Milburn et al. (1986) showed that the 80-kDa polypeptide in crude IF that can be cross-linked photochemically to mRNA in a cap- and ATP-Mg2 -dependent manner (Pelletier and Sonenberg, 1985a) is recognized by antibodies directed against eIF-4B. +

+

+

+

TRANSLATION BY POLIOVIRUS

179

B . Structural Complexity of Cap Binding Proteins The 24K-CBP from rabbit reticulocyte has been purified to nearhomogeneity by affinity chromatography using m7GDP coupled to Sepharose 4B (Sonenberg et al., 1979). The 24K-CBP obtained by this technique was shown to possess biological activity, stimulating the translation of capped mRNAs but not naturally uncapped RNAs in a translation extract from HeLa cells (Sonenberg et al., 1980). A 24-kDa polypeptide that is most probably identical to the 24K-CBP was purified by measuring an activity that relieves translational inhibition caused by cap analogs in a rabbit reticulocyte lysate (Hellman et al., 1982). This polypeptide was shown to bind specifically to capped oligonucleotides. More recently, Webb et al. (1984) purified the 24KCBP from rabbit reticulocytes by an alternative affinity chromatography procedure, and performed biochemical characterization of the protein. They found that the 24K-CBP preparation from human erythrocytes and rabbit erythrocytes or reticulocytes contained up to five isoelectric variants (Rychlik et al., 1986). Previously only two isoelectric variants had been reported for rabbit reticulocyte 24K-CBP (Sonenberg et al., 1979). In addition, the 24K-CBP from rabbit reticulocytes can be phosphorylated on a serine residue (Rychlik et al., 1986). There is no evidence as yet for a regulatory pathway involving phosphorylation/dephosphorylation of the 24K-CBP. Many lines of evidence suggest that the 24K-CBP is also able to associate with other polypeptides to form a high molecular weight protein complex and that this complex is the biologically relevant form. It was observed that 24K-CBP (detected by cross-linking to mRNA) sediments as an -200-kDa complex in sucrose gradients containing 0.3 M KC1 (Bergmann et al., 19791, and subsequently Tahara et al. (1981) have isolated a protein complex by m7GDP affinity chromatography comprising major polypeptides of -50,55, and 220 kDa in addition to the 24-CBP. This complex was termed CBP I1 to distinguish it from the uncomplexed 24K-CBP that was termed CBP I. CBP I1 was shown to be functionally distinct from the 24-CBP in that it has an activity which can restore translation of capped mRNAs in extracts from poliovirus-infected cells as detailed below (Tahara et al., 1981). Using a similar protocol, Edery et al. (1983) have isolated a CBP complex structurally similar to CBP I1 which comprises polypeptides of 50 and 220 kDa ( ~ 2 2 0in ) addition to the 24K-CBP. This complex (Edery et al., 1983) is referred to as the CBP complex. Grifo et al. (1983) have isolated a similar complex using a conventional fractionation protocol and found that aside from having an activity that can restore cap-

180

NAHUM SONENBERG

dependent mRNA translation to extracts from poliovirus-infected cells, it can stimulate translation in a reconstituted rabbit reticulocyte protein synthesis system. It is of interest that translation of both capped (globin) and the naturally uncapped satellite tobacco necrosis virus (STNV)mRNAs is stimulated in this system by this factor (Grifo et al., 1983). This result, however, does not necessarily indicate that the complex is required for translation of all naturally uncapped RNAs. In light of the findings of Grifo et al. (19831, the complex has been given born fide initiation factor status as eIF-4F. Because it is not unequivocally clear that all subunits of this factor are required for activity we preferred to refer t o this factor as the “CBP complex.” The stoichiometry of the subunits in the CBP complex has not yet been established, but some work has been done toward their characterization. The 24-kDa subunit was found to be similar to the 24KCBP with regard to mRNA cross-linking, migration on SDS-polyacrylamide gels, and reactivity with antibody against the 24K-CBP (I. Edery, unpublished results). The -50-kDa polypeptide is very similar to eIF-4A as determined by two-dimensional gel analysis (Grifo et al., 19831, peptide map analysis, and immunoreactivity (Edery et al., 1983). It is of interest that a difference was found in the relative amounts of two closely migrating peptides, as revealed by peptide map analysis, between free (uncomplexed) eIF-4A and the eIF-4A component of the CBP complex (Edery et al., 1983). Consequently, it is possible that a modification of eIF-4A correlates with its association with the 24K-CBP and p220 to form the CBP complex. Several studies indicate that the eIF-4A component of the CBP complex is involved in the ATP-mediated melting of mRNA secondary structures, as will be discussed below (Section 1,D).The p220 component of the CBP complex had not been previously characterized with respect to the cap recognition process. However, its integrity is apparently crucial for the function of the CBP complex, because in poliovirus-infected cells its proteolysis most likely results in inactivation of the CBP complex (Section 111,C).

C . Function of the CBP Complex Many observations are consistent with the hypothesis that a CBP(s), by utilizing energy derived from ATP hydrolysis, mediates the melting of mRNA 5’ noncoding region to facilitate ribosome binding (Sonenberg, 1981). Kozak (1980) and Morgan and Shatkin (1980) have shown that reovirus mRNA in which inosine is substituted for guanosine, with consequent reduction in stability of mRNA secondary structure, showed a reduced requirement for ATP and the cap structure for

TRANSLATION BY POLIOVIRUS

181

ribosome binding. Moreover, naturally occurring uncapped RNAs such as the plant cowpea mosaic virus RNA also exhibit a reduced dependence on ATP for ribosome binding (Jackson, 19821, consistent with the idea that the requirements for the cap structure and ATP hydrolysis are related aspects of translation initiation. In other experiments it was shown that a monoclonal antibody with anti-CBP activity can inhibit initiation complex formation with native reovirus mRNA but not with inosine-substituted reovirus mRNA (Sonenberg et al., 1981). Also consistent with the hypothesis that the cap functions by mediating the melting of mRNA 5’ secondary structure are findings obtained with the plant alfalfa mosaic virus 4 (AMV-4) RNA, which is capped and codes for the virus coat protein. Translation of this RNA in in uitro translation systems appears to be almost independent of the cap structure (van Vloten-doting et al., 1977; Sonenberg et al., 1981; Gehrke et al., 1983).Also, ribosome binding and translation of this mRNA are resistant to inhibition by high salt concentrations that inhibit ribosome binding and translation of other mRNAs [tobacco mosaic virus (TMV), vesicular stomatitis virus (VSV), and reovirus (Herson et al., 1979; Lee et al., 1983; Gehrke et al., 1983; Edery et al., 198413. These findings were interpreted to indicate that at higher salt concentrations, which stabilize mRNA secondary structure (Holder and Lingrel, 19751, there is an increased dependence on CBP for translation. However, AMV-4 RNA was shown by nuclease digestion experiments to be predominantly single strand at its 5’ terminus and hence presumably less dependent on the cap structure and the CBP complex for translation (Gehrke et al., 1983). Several recent studies further support the proposed mechanism of action of the CBP complex in which it melts 5‘ mRNA secondary structure in an ATP-Mg2+-dependent reaction. (1) Cross-linking of eIF-4A and eIF-4B, in crude IF preparations, to native mRNA is dependent on ATP-Mg2 (Sonenberg, 1981; Sonenberg et al., 1981). When inosinesubstituted mRNA (which is less structured as compared to native mRNA) is used, cross-linking of these polypeptides occurred in the absence of exogenously added ATP-MgZ+ (Lee et al., 1983). [In contrast to these results, Tahara et al. (1983) showed that cross-linking of purified eIF-4A and eIF-4B to inosine-substituted reovirus mRNA is dependent on the addition of ATP-Mg2 , A resonable explanation for this apparent discrepancy is that cross-linking of eIF-4A and eIF-4B does require low levels of ATP-Mg2+ which happen to be present in crude IF preparations, but not in purified eIF-4A and eIF-4B preparations.] (2) Messenger RNA with reduced secondary structure (inosinesubstituted reovirus mRNA and AMV-4 RNA) can function in extracts prepared from poliovirus-infected cells in which the CBP complex is +

+

182

NAHUM SONENBERG

inactivated (see below), strongly indicating that the CBP complex is involved in melting mRNA secondary structure (Sonenberg et al., 1982). (3)The CBP complex (or eIF-4F) was shown to relieve translational competition between mRNAs, suggesting that it is a limiting factor in translational initiation (Ray et al., 1983; Sarkar et al., 1984). Consequently, it was predicted according to the model that mRNAs with excessive secondary structure at their 5’ terminus would require more of the CBP complex for translation and would be translated relatively inefficiently. This prediction was verified by Pelletier and Sonenberg (1985b), who inserted nucleotide sequences having dyad symmetry into the 5’ noncoding region of the herpesvirus thymidine kinase gene and showed that its mRNA translational efficiency was reduced in direct proportion to the increased degree of secondary structure in the mRNA. Direct studies to demonstrate that the CBP complex can relieve the reduced translational efficiency are in progress in our laboratory. (4) Ray et al. (1985) recently demonstrated that incubation of reovirus mRNA with the CBP complex (eIF-4F) resulted in increased sensitivity to single-strand-specific nuclease digestion. This activity was ATP dependent and was inhibited by cap analogs. In addition these authors demonstrated that eIF-4F could melt a short (20 base pairs) RNA-DNA heteroduplex. These activities of eIF-4F were lost when the eIF-4A component of eIF-4F was ablated from the complex (see below).

D. Function of the eIF-4A Component of the CBP Complex The component of the CBP complex that is most likely directly responsible for its apparent RNA melting activity is eIF-4A, active as part of the complex. Several lines of evidence are consistent with this possibility: (1)eIF-4A was shown to possess ATPase activity that is dependent on the presence of polynucleotides (Grifo et al., 1984). Similarly, eIF-4F (CBP complex) was shown to possess ATPase activity which was polynucleotide dependent, but unlike that of eIF-4A it could be inhibited by the addition of cap analogs. Furthermore, eIF-4B stimulated the in vitro ATPase activity of either eIF-4A or eIF-4F (Grifo et al., 19841, which is in accordance with the findings that eIF-4B is required for the proposed CBP complex melting activity (Edery et al., 1983; Ray et al., 1985). (2) The eIF-4A component of the CBP complex was shown directly to have an ATP binding site by photoafinity labeling of the complex with [ w ~ ~ P I A TorP dATP (Sarkar et al., 1985). Peptide map analysis of the labeled eIF-4A subunit revealed only one one major 32Plabeled peptide, suggesting only one ATP binding site (G. Sarkar, unpublished results). It was found that UV irradiation-induced cross-

TRANSLATION BY POLIOVIRUS

183

linking of ATP to the eIF-4A component of the CBP complex was -60fold more efficient than cross-linking of ATP to uncomplexed eIF-4A. Furthermore, Sarkar et al. (1985)showed that uncomplexed eIF-4A and eIF-4B could stimulate the cross-linking of ATP to eIF-4A. Based on these results, Sarkar et al. (1985) suggested that the eIF-4A component of the CBP complex is responsible for the generation of energy that is consumed in the mRNA melting process. Seal et al. (1983) reported that the ATP analog 5’-fluorosulfonylbenzoyladenosine, which can covalently cross-link to protein ATP-binding sites, strongly inhibited the activity of uncomplexed wheat germ eIF-4A. (3) Ray et al. (1985) demonstrated that when rabbit reticulocyte eIF-4F (CBP complex) was passed through a phosphocellulose column, the eIF-4A component can be separated from the rest of the complex. The authors showed that

elF-46 ATP C A W + Pi

FIG.1. Model for the function of eIF-4A, eIF-4B, and the CBP complex (eIF-4F) in translation initiation. eIF-4A* represents the eIF-4A component of the CBP complex.

184

NAHUM SONENBERG

eIF-4F depleted of eIF-4A is no longer active in destabilization of mRNA higher structure. In conclusion, there are several lines of evidence indicating that the CBP complex (eIF-4F) functions to melt 5’ secondary structure of eukaryotic mRNAs in an ATP-mediated process. The apparent unwinding activity is stimulated by uncomplexed eIF-4A and eIF-4B and is most probably contained in the eIF-4A component of the CBP complex. Melting of mRNA 5’ secondary structure presumably facilitates subsequent ribosome binding (Sonenberg et al., 1983) or migration of ribosomes (Kozak, 1983). Although the focus of the preceding discussion is the proposed function of the CBP complex in melting mRNA 5’ secondary structure, several other possible roles of the complex are noteworthy of consideration. (1)The CBP complex possibly interacts with eIF-3 (Trachsel et al., 1980; Hansen and Ehrenfeld, 1981), thus providing a possible means for the ribosome-eIF-3 complex to attach to the mRNA. (2) The CBP complex might physically interact with eIF-4B, thus enabling the latter factor to attach to the mRNA (Edery et al., 1983).It is conceivable that more activities could be assigned to the CBP complex once its molecular mechanism of action is clarified. Figure 1shows a schematic model of the mode of function of CBP complex, eIF-4A, and eIF-4B in initiation of translation.

111. INVOLVEMENT OF THE CBP COMPLEX IN THE SHUTOFF OF HOST AFTER POLIOVIRUS INFECTION mRNA TRANSLATION

A . Early Studies It is safe to conclude that host mRNAs are not functionally impaired following poliovirus infection. Poliovirus infection neither induces the degradation of host mRNAs (Leibowitz and Penman, 1971; Kaufmann et al., 1976) nor causes detectable changes in the patterns of host mRNA capping, methylation, and polyadenylation (Fernandez-Munoz and Darnell, 1976). Furthermore, Ehrepfeld and Lund (1977) demonstrated that host mRNA extracted from infected cells remains functional in a wheat germ translation system. It has been shown that the lesion in the process of host protein synthesis is at the level of translation initiation (Kaufmann et al., 1976). Subsequent studies by Helentjaris and Ehrenfeld (1978) demonstrated that crude IF preparations from poliovirus-infected cells can stimulate translation of poliovirus RNA but not host mRNA in HeLa cell extracts. In contrast to the ubiquitous presence of the cap structure in eu-

TRANSLATION BY POLIOVIRUS

185

karyotic cellular mRNAs, poliovirus RNA is not capped (Hewlett et al., 1976; Nomoto et al., 1976) and its translation most probably proceeds by a cap-independent mechanism. It was proposed that this difference provides the basis for the translational discrimination between capped cellular mRNAs and uncapped poliovirus RNA after infection with poliovirus (Fernandez-Munoz and Darnell, 1976; Rose et al., 1978; Helentjaris and Ehrenfeld, 1978). An attractive hypothesis was that specific inactivation of a protein that recognizes the cap structure and is required for translation of host mRNAs, but not poliovirus RNA, could be responsible for the shutoff phenomenon. In subsequent studies attempts were made to identify the component in crude IF from infected cells that is impaired as the result of infection.

B. The CBP Complex Is Inactivated in Poliovirus-Infected Cells The first report demonstrating that a specific initiation factor is inactivated after poliovirus infection was by Rose et al. (19781,who used VSV mRNA as a model for host mRNA in a cell-free translation system. It had been previously determined that superinfection of VSV-infected cells by poliovirus results in inhibition of VSV mRNA translation in a similar if not identical manner to that of host mRNA (Doyle and Holland, 1972; Ehrenfeld and Lund, 1977). Rose et al. (1978) examined the ability of different IF preparations to restore the translation of VSV mRNA in extracts from poliovirus-infected cells and found that eIF-4B was the only IF which had this activity. In contrast to the conclusions drawn from this experiment, Helentjaris et al. (1979) showed that eIF-4B purified from poliovirus-infected cells was active in a reconstituted reticulocyte translation system. In contrast, eIF-3 purified from these cells was inactive in this system. As stated above (Section II), both of these IF preparations (eIF-3 and eIF-4B) were found to contain the 24K-CBP, as assayed by the cross-linking technique (Sonenberg et al., 1978).Later it was also determined that eIF-3 and eIF-4B cofractionate with the CBP complex (eIF-4F) through several purification steps (Grifo et al., 1983). Furthermore, Trachsel et al. (1980) showed that the 24K-CBP isolated by several purification steps copurified with an activity that can restore translation of capped mRNAs in extracts from poliovirus-infected cells (restoring activity). However, the activity obtained by these workers was unstable. Subsequently it was reported that a multisubunit protein complex of -8-10 S that was purified by m7GDP affinity chromatography could restore translation of capped mRNAs in extracts from poliovirus-infected cells, and that this activity was stable (Tahara et al., 1981).This complex, which was termed CBP I1 to distinguish it from the 24K-CBP (termed CBP I), consisted of three

186

NAHUM SONENBERG

major polypeptides in addition to the 24K-CBP, and further work led to its characterization as eIF-4F and the CBP complex as described above (Section 11,B). The ability of the CBP complex to restore translation of capped mRNAs in extracts from poliovirus-infected cells strongly suggested that this protein is somehow inactivated as a consequence of poliovirus infection. Several lines of evidence are consistent with this contention: (1)eIF-4A and eIF-4B are neither structurally modified (Duncan et al., 1983) nor functionally impaired (Helentjaris et al., 1979) after poliovirus infection. However, these proteins could not cross-link to the mRNA 5’ cap structure when the chemical (Lee and Sonenberg, 1982; Lee et al., 1985a) or photochemical (eIF-4B in this case) cross-linking procedure was used with IF preparations from poliovirus-infected cells (Pelletier and Sonenberg, 1985a). These findings are in accord with the interpretation that cross-linking of eIF-4A and eIF-4B to the cap structure is dependent on the interaction of the CBP complex with the cap structure (Edery et al., 1983). Furthermore, addition of the CBP complex (which is free of eIF-4B) to crude IF preparations from poliovirus-infected cells restored cross-linking of eIF-4B in the infected extract (Lee et al., 1985a). (2) Etchison et al. (1984) have demonstrated, using a fractionated translation system, that an activity attributable to the CBP complex is inactivated in poliovirus-infected cells. (3)Messenger RNAs that are less dependent on the cap structure or CBP complex for translation can still function in extracts prepared from poliovirus-infected cells (Sonenberg et al., 1982; Edery et al., 1984), indicating that the function impaired in poliovirus-infected cells is related to the CBP complex. These results present, however, only indirect evidence for the inactivation of the CBP complex and do not point to the mechanism of inactivation. The first report to describe a specific modification of the CBP complex was by Etchison et al. (1982). These authors prepared an antiserum against a 220-kDa polypeptide (termed p220) that copurified with preparations of eIF-3 under conditions in which the CBP complex fractionates with eIF-3. This antiserum reacted with a 220-kDa polypeptide present in extracts prepared from mock-infected cells, but not in poliovirus-infected cells. Instead, in extracts from poliovirus-infected cells, the antibody recognized two to three polypeptides of 110130 kDa. Figure 2 shows a representative immunoblotting result obtained with an anti-p220 antibody. The authors proposed that these polypeptides are cleavage products of p220. Because anti-p220 antibody also recognizes the 220-kDa polypeptide of the CBP complex, it was proposed that proteolysis of the CBP complex p220 by a poliovirusdependent protein results in the shutoff of host protein synthesis (Etchison et al., 1982). To demonstrate directly that the CBP complex

TRANSLATION BY POLIOVIRUS

187

FIG.2. Immunoblot analysis of p220 in extracts from mock- and poliovirus-infected cells. Cell extracts were resolved on a 6%polyacrylamide gel and transferred to nitrocellulose paper. The paper was reacted with anti-p220 antibody followed by 126I-labeled protein A and autoradiography as described by Lee et al. (1985a).

in poliovirus-infected cells is indeed modified, Lee et al. (1985a) passed fractions from mock- and poliovirus-infected cells on a m7GDPagarose column. They isolated a modified CBP complex from poliovirus-infected cells consisting of the 24K-CBP and putative cleavage products which comigrated with the 110- to 130-kDa cleavage products detected by immunoblotting of infected extracts. However, the presence of eIF-4A in this CBP complex varied among preparations, consistent with recent results by Ray et al. (1985) and D. Etchison (personal communication) that the eIF-4A component of the CBP complex can be readily removed from the complex. It is quite possible that the eIF-4A component of the CBP complex dissociates more readily from the modified complex than the native complex. This raises the interesting possibility that this event is important for the inactivation of the CBP complex.

C. Mechanism of CBP Complex Inactivation A key question concerns the mechanism by which proteolysis of the CBP complex results in loss of function. Several reports attempted to address this question. Hansen and Ehrenfeld (1981) reported that the

188

NAHUM SONENBERG

extent of cross-linking of the 24K-CBP to mRNA is similar in IF preparations from poliovirus-infected and mock-infected cells. Subsequently, they showed that most of the 24K-CBP in uninfected cells, capable of being cross-linked to the cap structure, sediments as a large complex probably in association with eIF-3 (Hansen et al., 1982). However, following poliovirus infection, cross-linked 24K-CBP was found only at the top of the gradient, apparently as a noncomplexed entity. In contrast to these results, Lee and Sonenberg (1982) have demonstrated that cross-linking of the 24K-CBP is drastically reduced in IF preparations from poliovirus-infected cells and this reduction is also observed after fractionation of ribosomal high-salt wash on gradients (K. Lee unpublished observations). Identical results were obtained when a photochemical cross-linking technique was used (Pelletier and Sonenberg, 1985a). The source of discrepancy between the results of Hansen et al. (1982) and Lee and Sonenberg (1982) is not immediately clear. However, it was also shown that the extent of cross-linking of the 24K-CBP to capped mRNA is at least 10-fold greater when the 24K-CBP exists as part of the CBP complex than when it is in uncomplexed form (Lee et al., 1985a). Therefore, it seems likely that the 24KCBP in the putative modified CBP complex from poliovirus-infected cells behaves like the uncomplexed 24-kDa polypeptide in the crosslinking to mRNA. Two groups have also examined whether modifications of the 24KCBP also contribute to its reduced cross-linking ability. This was of interest in light of a report by James and Tershak (1981) that several polypeptides including a 24-kDa polypeptide become phosphorylated in poliovirus-infected CV-1 cells. Lee et al. (1985a) reported that the 24K-CBP purified from the S-100 fraction of mock- or poliovirus-infected cells consisted of the same single major species on two-dimensional gels in both cases. In addition, they suggested that the different isoelectric variants of the ribosome-associated 24K-CBP have similar activities, but this has still not been rigorously examined. In a more recent study, Buckley et al. (1986) analyzed total 24K-CBP by twodimensional SDS-polyacrylamide gel electrophoresis and found no differences in the isoelectric variants after poliovirus-infection as compared to mock-infected 24K-CBP. It seems safe to conclude that poliovirus infection does not cause any modification of the 24K-CBP, and that the inactivation of the CBP complex is therefore a consequence of cleavage of the 220-kDa polypeptide.

D . Poliovirus Mediator of p220 Cleavage A challenging question concerns the nature of the protease that is implicated in the cleavage of p220. There are a priori at least three

TRANSLATION BY POLIOVIRUS

189

possibilities: (1)one of the poliovirus proteases that is involved in the processing of the poliovirus polyprotein precursor or a n undiscovered poliovirus protease is also responsible for the cleavage of p220; (2) a cellular protease is induced on poliovirus infection; or (3)a poliovirus activity renders p220 a substrate for a cellular protease. These possibilities were tested by two groups (Lloyd et al., 1985,1986: Lee et al., 1985b).The only characterized poliovirus protease is 3C (P3-7c according to the old nomenclature; the nomenclature used in this review is according to Rueckert and Wimmer, 19841,which is responsible for the cleavage between Gln-Gly amino acids pairs during processing of the viral polyprotein (Hanecak et al., 1982). Its direct involvement in cleavage of p220 was ruled out by Lloyd et al. (1985) who showed that protease 3C does not copurify with the p220 proteolyzing activity, and antibodies against 3C which inhibit the cleavage of poliovirus polyprotein in uitro do not inhibit the in uitro cleavage of p220. Lee et al. (1985b) performed similar experiments and arrived at the same conclusions. In addition Lee et al. (1985b) also examined the effect of antibodies against 2C (P2-X) [claimed to be a protease (Korant et al., 1979), although it has not been confirmed in another study (Hanecak et al., 1982)l on in uitro cleavage of p220 and found no effect. More recently it became apparent that a different poliovirus protein, 2A (P2-8), is involved in the cleavage of two Tyr-Gly amino acid pairs that are not cleaved by protein 3C (Toyoda et al., 1986). This protease has a consensus sequence Gly-Asp-Cys-Gly-Gly that is found in protein 3C (Blinov et al., 1985) and in several cellular serine proteases (Neurath et al., 1967: Blinov et al., 1984; in poliovirus proteins 2A and 3C, cysteine substitutes for serine in the consensus sequence fqund in serine proteases). Experiments showed that antibodies directed against protein 2A partially inhibit (-60%) the cleavage of the TyrGly amino acid pair that lies between poliovirus proteins 3C’ and 3D’. Evidence that protein 2A is implicated in the cleavage of p220, and consequently in the shutoff of host protein synthesis after poliovirus infection, was reported by Bernstein et al. (1985). They obtained a mutant that contained an extra amino acid (leucine) in the aminoterminal portion of the P2 region. The effects of this mutation on virus replication when tested in both HeLa and CV-1 cells were striking. It was found that in both cell lines the mutant virus generated small plaques at all temperatures apparently as a consequence of the low levels of viral protein synthesized. Furthermore, in both cell lines the mutant virus appears to have lost the ability to mediate the selective inhibition of host protein synthesis. In an attempt to correlate the characteristic discrimination between host and poliovirus RNA translation in poliovirus-infected cells with the cleavage of p220 of the CBP complex, Bernstein et al. (1985) mea-

190

NAHUM SONENBERG

sured the level of p220 in mutant vs wild-type infected cells by an immunoblot assay. They found that in both HeLa and CV-1 cells, p220 was not cleaved following infection with mutant virus, though it was readily cleaved following wild-type virus infection of these cells. These observations provide the first indication for a causal relationship between cleavage of p220 and the selective suppression of host protein synthesis after poliovirus infection. In addition, these experiments suggest that protein 2A is the mediator of p220 cleavage. It is not entirely clear, however, whether protein 2A cleaves p220 directly or whether it activates a cellular protease that plays a role in the normal turnover or regulation of p220. In this regard it has been observed that very low levels of cleavage products of p220 sometimes appear in uninfected cells (I. Edery, unpublished observations; E. Ehrenfeld, personal communication). Moreover, recent experiments (Lloyd et al., 1986) indicate that 2A might not be involved directly in the cleavage of p220, because (1) antibodies directed against 2A that inhibit cleavage of 3CD into 3C’ and 3D’ had no effect on the in uitro proteolysis of p220 and (2) p220 proteolyzing activity did not copurify with 2A, but instead copurified with a cellular protein of -50 kDa (E. Ehrenfeld, personal communication). Thus, it appears that the 2A protein does not cleave the CBP complex directly. There was one important difference in the pattern of host protein synthesis in mutant poliovirus-infected HeLa versus CV-1 cells. Whereas host protein synthesis in CV-1 cells was not reduced early in infection, in HeLa cells a shutoff of host protein synthesis did occur about 3 hours after infection. However, in this case there was concomitant cessation of viral protein synthesis. This apparent nondiscriminatory translational inhibition is distinguishable from wildtype mediated inhibition by several criteria (Bernstein et al., 1985). Recent experiments suggest that the early shutoff of protein synthesis in HeLa cells infected with the mutant virus is due to phosphorylation of the a subunit of eIF-2 (H. Bernstein, personal communication). It is of interest that similar findings were also reported for adenovirus-2 deletion mutants in the region coding for VA RNA I, which upon infection permit the phosphorylation of the eIF-2 a subunit, resulting in inhibition of both host and adenovirus protein synthesis (Reichel et al., 1985: Siekierka et al., 1985: Schneider et al., 1985). It was suggested that the function of VA RNA I is to counteract the cellular response to adenovirus infection which involves eIF-2 phosphorylation. Related phenomena were reported for influenza virus (Katze et al., 1984) and vaccinia virus (Whitaker-Dowling and Youngner, 1984) and might apply to many other viruses. Therefore, it is possible that poliovirus, like other viruses, has evolved mechanisms to suppress eIF-2 phosphorylation that is induced during infection. This activity

TRANSLATION BY POLIOVIRUS

191

might be associated with poliovirus protein 2A and rendered inactive in the mutant virus. Studies to characterize the p220 proetolytic activity in an in uitro assay demonstrated that the activity is temperature dependent (proteolysis did not occur at 0°C) and heat sensitive. In addition, several serine and thiol protease inhibitors abolished the proteolytic activity (Sonenberg et al., 1983).

E . Criticism of the Model The model for the implication of CBP complex cleavage in the shutoff of host protein synthesis following poliovirus infection has been questioned (Alonso and Carrasco, 1982a). The criticisms were based on the findings that poliovirus superinfection of the parainfluenza SV5infected cells did not cause the suppression of SV5-encoded protein, although SV5 mRNAs are capped (Choppin and Holmes, 1967). Similarly, Semliki Forest virus-encoded proteins were synthesized in cells coinfected with poliovirus (Alonso and Carrasco, 1982a). These findings could be explained in light of the proposed model for the RNA melting activity of the CBP complex, if the mRNAs of SV5 and Semliki Forest viruses are devoid of extensive secondary structure in their 5’ region. In that case these mRNAs will be less dependent on the CBP complex for their translation and would be able to translate in poliovirus-infected cells. This situation is analogous to that described for AMV-4 RNA which could be translated in cell extracts from poliovirus-infected cells, in spite of the fact that it is naturally capped (Sonenberg et al., 1982: see Section 11,C).Thus, it should be of interest to determine if the mRNA 5’ sequences of SV5 and Semliki forest viruses are less structured as compared to cellular mRNAs. TO EXPLAIN POLIOVIRUS INHIBITION OF IV. ALTERNATIVEMODELS HOSTPROTEIN SYNTHESIS

Several different models have been proposed to explain the mechanism of the shutoff of host protein synthesis that occurs after poliovirus infection. Notwithstanding the drawbacks of these models, I will summarize them and point out their weaknesses.

A . Double-Strand RNA Inhibition of Protein Synthesis Ehrenfeld and Hunt (1971) showed that extracts prepared from poliovirus-infected cells when added to a reticulocyte lysate inhibit pro-

192

NAHUM SONENBERG

tein synthesis. Extracts from mock-infected cells did not have this activity. The activity responsible for this inhibition was shown to be double-stranded poliovirus RNA. Later studies showed, however, that addition of double-stranded poliovirus RNA to HeLa cell-free extracts inhibited the translation of both cellular and viral mRNAs (Celma and Ehrenfeld, 1974), apparently by inducing a double-strand RNA-activated kinase that phosphorylates the a subunit of eIF-2. The important conclusion from these experiments was that poliovirus doublestrand RNA is not responsible for the selective inhibition of host mRNA synthesis. These findings also bear on the observations of H. Bernstein (Section II1,D) that poliovirus infection might induce a double-strand RNA-activated kinase, and that during a normal infection a poliovirus-coded activity prevents this process.

B . Alteration of Intracellular Ion Concentration after Poliovirus Infection According to a model that was promulgated by Carrasco and Smith (1976, 19771, infection by viruses causes changes in the cell membrane, rendering it leaky to ions. The consequent perturbation of the normal monovalent ion gradient results in the increase of intracellular Na concentration. It was proposed that the binding of a viral coat protein to the membrane is responsible for the changes in the ion gradient (Carrasco, 1977). The increase of Na+ in the cell would then lead to inhibition of cellular protein synthesis, as has been shown in both in vivo (Saborio et al., 1974) and in vitro (Carrasco and Smith, 1976) translation systems. Translation of poliovirus RNA was shown to be relatively resistant to inhibition by high salt concentrations (Nuss et al., 1975). This increased Na+ concentration would result in the preferential translation of viral RNA. The major drawback of this hypothesis is that the shutoff of cellular protein synthesis in poliovirus-infected cells occurs early in infection (1-2 hours) when no perturbation of Na ,K -ATPase activity or intracellular Na concentration was detectable (Nair et al., 1979; Lacal and Carrasco, 1982, 1983; Shaeffer et al., 1982). Moreover, Bossart and Bienz (1981) showed that increased ion concentration inhibited poliovirus polypeptide chain elongation to a higher extent that initiation, in vitro. They concluded that increased ion concentrations in poliovirus-infected cells inhibited both poliovirus and host protein synthesis. It must be stressed that this hypothesis might still explain the shutoff of host protein synthesis that occurs after infection with other members of the picornavirus group, especially encephalomyocarditis (EMC) virus, since in the EMC case there is a positive correlation between the shut+

+

+

+

TRANSLATION BY POLIOVIRUS

193

off of host protein synthesis and the increase of Na+ concentration in the cell (Lacal and Carrasco, 1982, and see below). However, when the shutoff of host protein synthesis induced by EMC virus is rapid, as described by Jen and Thach (1982), it cannot be explained by membrane leakiness models, because membrane leakiness can be demonstrated only later in infection.

C. Competition between Viral and Host mRNAs It is believed that the shutoff of host protein synthesis by EMC and mengovirus is achieved at least partly by virtue of the ability of their RNA to successfully compete with cellular mRNAs for binding to components of the translational machinery and probably also due to the large amounts of viral RNA that are synthesized during infection. These arguments are supported by reasonable data. However, it is clear that this model could not apply to poliovirus as was suggested by Nuss et al. (1975) for the following reasons. (1) The shutoff occurs early in infection when no appreciable amounts of poliovirus RNA are synthesized; (2) shutoff occurs in the presence of guanidine which drastically inhibits poliovirus replication (Holland, 1964; Penman and Summers, 1965); and (3) poliovirus RNA competes poorly with other mRNAs in in uitro translation systems (Brown and Ehrenfeld, 1980; Rose et al., 1978). For example, Rose et al. (1978) have shown that VSV mRNA outcompetes poliovirus mRNA in an in uitro translation system prepared from HeLa cells, but initiation of VSV translation is shut off efficiently in uiuo by poliovirus superinfection. Nevertheless, it is possible that poliovirus RNA might be translated efficiently in uiuo (see Section V,B).

D. Inhibition of Host mRNA Synthesis by Poliovirus Structural Components Cooper et al. (1973) have suggested the existence in poliovirus-infected cells of an hypothetical regulator particle that they termed “equestron.” This particle was postulated to bind to the 40 S ribosomal subunits and to the replication complexes of viral RNA. This would result in the inhibition of host protein synthesis and efficient translation of viral RNA. In a search for viral proteins that would have these properties, Wright and Cooper (1974) found that the viral structural polypeptides VPO, VP1, and VP3 were bound to ribosomes. Similar results were reported for Ehrlich ascites cells infected with EMC virus (Medvedkina et al., 1974) and mengovirus-infected L cells (Manak et al., 1975). However, there is probably no significance of these findings

194

NAHUM SONENBERG

for the shutoff of host protein synthesis, since shutoff can be achieved without detectable synthesis of viral capsid polypeptides (Holland, 1964), indicating again that it is achieved by a catalytic and not stoichiometric mechanism. Several other studies attempted to show an involvement of the structural proteins of poliovirus in the translational shutoff of capped mRNAs. Steiner-Pryor and Cooper (1973) obtained, by a chemical mutagenesis protocol, temperature-sensitive mutants that were mapped to the capsid protein region. These mutants were shown to be defective in the shutoff reaction. Unfortunately, it was not excluded that secondary mutations had occurred in other parts of the genome. V. TRANSLATIONAL INHIBITION BY OTHERPICORNAVIRUSES The translational inhibition caused by other picornaviruses has been extensively studied by several groups. It appears, however, that only rhinoviruses have adopted a mechanism similar to that of poliovirus for the shutoff of host protein synthesis. Etchison and Fout (1985) found that the p220 polypeptide of the CBP complex is cleaved in rhinovirus 14-infected HeLa cells, and that this cleavage roughly correlates with the decrease in the rate of host protein synthesis after infection. In addition, they showed that crude preparations containing eIF-3-CBP complexes could restore translation of capped globin mRNA in a fractionated translation system. However, purified CBP complex had no stimulatory activity in this system, whereas it restored activity in a similar system from poliovirus-infected cells (Etchison et al., 1984). It was proposed that in rhinovirus-infected extracts there is a higher level of p220 proteolyzing activity than in similar preparations from poliovirus-infected cells, but the level was not measured. Experiments conducted with the cardioviruses EMC virus (Mosenkis et al., 1985) and mengovirus (R. Duncan and J. W. Hershey, personal communication) demonstrated that p220 is not cleaved during infection with these viruses. In this regard it is noteworthy that the nucleotide and amino acid sequence of protein 2A of poliovirus is not homologous to its EMC virus counterpart. This is in contrast to homology with other proteins, including the capsid proteins and protease 3C. Thus, it is possible that the EMC virus and poliovirus 2A protein are functionally different. These results are consistent with numerous reports which are discussed below suggesting that the mode of shutoff of host protein synthesis by poliovirus is different from that of cardioviruses, where several models, which might not be mutually exclusive, were invoked to explain the shutoff phenomenon. However,

TRANSLATION BY POLIOVIRUS

195

the models that are outlined below are not proved and more work needs to be carried out to test their validity.

A . Competition Model Infection of several cell lines including HeLa, MOPC ascites, Krebs ascites, and SC-1 fibroblasts with EMC virus results in inhibition of host protein synthesis, but as opposed to the precipitous inhibition that is induced by poliovirus, this inhibition is gradual and occurs concomitantly with an increase in viral protein synthesis. Consequently, the shutoff occurs late in infection (6-7 hours; Lawrence and Thach 1974; Jen et al., 1978). A similar characteristic pattern was observed after infection of L cells with mengovirus (Otto and Lucas-Lenard, 1981). In addition, protein synthesizing extracts prepared from these cells translated capped cellular mRNAs and EMC virus or mengovirus RNAs with similar efficiencies (Lawrence and Thach, 1974; Svitkin et al., 1974; Abreu and Lucas-Lenard, 1976; Jen et al., 1978; Hackett et al., 1978). Furthermore, crude IF preparations from EMC virus-infected cells stimulated globin mRNA translation in a fractionated translation system prepared from mouse cells, in contrast to IF preparations from poliovirus-infected cells (Jen et al., 1980). Taken together these results are consistent with the hypothesis that EMC virus and mengovirus RNAs possess a stronger affinity than host mRNAs for a translational component that is a limiting factor in translation. Two initiation factors were reported as candidates for such a limiting factor: eIF-4B and eIF-2. [The function of eIF-4B in initiation of translation was described above. Eukaryotic initiation factor 2 is required for the formation of a ternary complex consisting of eIF-2, met-tRNA, and GTP (e.g., Staehelin et al., 1975). In addition, it is claimed that eIF-2 binds specifically to a region of the mRNA encompassing the initiator AUG and thus plays a role in directing ribosomes to this region (Kaempfer et al., 1981).1Initiation factor 4B was shown to relieve competition between EMC virus and globin RNAs in an in uitro translation system from mouse ascites cells (Golini et al., 1976) and, in a second study, this factor was shown to bind with higher affinity to EMC virus RNA than to cellular mRNAs, using a nitrocellulose filter binding assay (Baglioni et al., 1978). However, these studies are problematic for several reasons: (1)eIF-4B preparations used in those studies were only partially pure and were shown later to contain variable amounts of the 24K-CBP or the CBP complex (Sonenberg et al., 1978; Bergmann et al., 1979; Trachsel et al., 1980: Grifo et al., 1983). Indeed, as indicated above several studies have shown that the CBP complex (eIF-4F) can relieve translational com-

196

NAHUM SONENBERG

petition between mRNAs (Ray et al., 1983; Sarkar et al., 1984). (2) The experiments involving nitrocellulose filter binding of IF-mRNA complexes can be challenged because nonspecific complexes between RNA and proteins can be trapped by this technique and nonspecific complex formation can be inhibited by the addition of cap analogs (Sonenberg and Shatkin, 1978). (3) It has been demonstrated by Staehelin et al. (1975) that translation of EMC virus RNA in a reconstituted in vitro translation system requires three times more eIF-4B than that required for the translation of globin mRNA, thus vitiating the concept that translation of EMC virus RNA requires less eIF-4B due to the higher affinity of viral RNA for this factor (Baglioni et al., 1978). In other studies, Rosen et al. (1982) demonstrated by using the nitrocellulose filter technique that mengovirus RNA has a 30-fold higher affinity for eIF-2 than globin mRNA. In addition, the interaction between mengovirus RNA and eIF-2 was more resistant to inhibition by high salt than the interaction between globin mRNA and eIF-2. It was, therefore, concluded that the latter observations support a model for the shutoff of host protein synthesis that invokes mengovirus outcompeting cellular mRNAs for eIF-2. Furthermore, in the infected cell there is an increase in ion concentration, augmenting the competitive advantage of mengovirus RNA.

B . Other Models for Shutoff by Cardioviruses The competition model for the translational shutoff phenomenon cannot explain some observations reported in several systems. (1) EMC virus infection of mouse L cells causes a precipitous shutoff of host protein synthesis which is kinetically similar to the inhibition induced by poliovirus. The shutoff occurs before appreciable amounts of viral RNA are synthesized, therefore ruling out a competition model (Jen and Thach, 1982). Jen and Thach (1982) attempted to determine the mode of translational discrimination in this system and found that, in contrast to extracts from poliovirus-infected cells, extracts from EMC virus-infected L cells translated capped mRNAs with an efficiency similar to extracts from mock-infected cells. However, they concluded that the distribution of an activity that stimulated translation of capped mRNAs is changed in extracts from EMC virus-infected L cells, specifically, that the S-200 fraction in these cells is significantly enriched with respect to an activity that stimulated globin but not EMC mRNA translation in a reconstituted translation system prepared from mouse cells that was deficient in cap translation activity.

TRANSLATION BY POLIOVIRUS

197

The authors further argue that under the in uitro translation conditions a redistribution of a CBP-like activity takes place, correcting the lesion in the translation process in uiuo. It was proposed that alterations in the infected cells disrupt initiation complexes of cellular but not viral mRNAs, followed by breakdown of polysomes. (2) Rosen et al. (1981) showed that translation of mengovirus RNA in a template-dependent reticulocyte lysate is resistant to inhibition by concentrations of dsRNA that completely block translation of globin or ascites mRNA. This was explained on the basis that mengovirus RNA binds eIF-2 with much higher affinity than dsRNA and consequently protects eIF-2 from inactivation by dsRNA. In contrast, binding of eIF-2 t o globin mRNA is weaker than to dsRNA and therefore in this case eIF-2 is inactivated and rendered unavailable for globin mRNA translation. (3)Yet another model for the inhibition of protein synthesis by mengovirus was recently proposed by Pensiero and Lucas-Lenard (1985), who presented evidence that extracts prepared from mengovirus-infected L cells are less active in translation than extracts prepared from mock-infected cells. None of the known IF had the ability to restore translation when added to extracts from infected cells, as would be expected according to Rosen et al. (1981). In a n attempt to define the lesion in protein synthesis, extracts were prepared at different times after infection. Late in infection (4-6 hours), polysomes were shown to contain an inhibitor of translation initiation; treatment of these polysomes with 0.5 M KC1 restored their translational activity (Pensiero and Lucas-Lenard, 1985). Cell extracts prepared early after infection (1hour) contained a kinase activity which phosphorylates the o! subunit of exogenously added eIF-2 (J. M. LucasLenard, personal communication). However, the significance of this phosphorylation is not clear since host protein synthesis at this time of infection is affected only marginally. These results are not consistent with the concept of a discriminatory mechanism for the inhibition of host protein synthesis, and one has to invoke again the assumption that mengovirus RNA has a competitive edge over cellular mRNAs. The summary of the latest results leaves the reader bamed by the variety of mechanisms invoked to explain the shutoff of host protein synthesis that is elicited by EMC- and mengoviruses in different cell types. It is therefore important to elucidate the molecular mechanisms involved in this phenomenon, possibly by the use of defined mutants derived from the currently available infectious cDNA clones of EMC virus and mengovirus in a fashion analogous to the study described by Bernstein et al. (1985).

198

NAHUM SONENBERG

One important question raised by these studies is why EMC virus and mengovirus, both belonging to the picornavirus family and having naturally uncapped RNAs, apparently employ different strategies from poliovirus to induce the shutoff of host protein synthesis. This is all the more puzzling when one considers the fact that EMC virus and mengovirus RNAs are translated by a cap-independent manner (Weber et al., 1976; Canaani et al., 1977), as is poliovirus RNA. These viruses can replicate in poliovirus-infected cells (McCormick and Penman, 1968; Detjen et al., 1981; Alonso and Carrasco, 1982b) and EMC virus RNA can be translated in extracts from poliovirus-infected cells (Bonatti et aE., 1980). One possible way to explain these differences is that poliovirus RNA is a feeble mRNA in that it translates inefficiently as was demonstrated in uitro in a reticulocyte lysate translation system (Shih et al., 1978) and therefore cannot compete with host mRNAs, whereas EMC virus and mengovirus RNAs translate very efficiently, and therefore compete favorably with host mRNA. Accordingly, the efficient translation of the latter RNAs is not contingent on inhibition of host mRNA translation. However, a major caveat in this explanation is that it is not clear that poliovirus mRNA is indeed a weak mRNA in uiuo. The existing evidence points to the contrary, because poliovirus translation occurs when EMC virus-infected HeLa cells are superinfected with poliovirus although EMC RNA virus is regarded as a “strong” RNA, which favorably competes with host mRNA (Detjen et al., 1981; Alonso and Carrasco, 1982b). One possible explanation for the apparently low translational eficiency of poliovirus RNA in uitro was offered by Daniels-McQueen et al. (19831, who reported that poliovirus RNA translation in a reconstituted translation system from Ehrlich ascites cells required high levels of eIF-4A, whereas EMC virus RNA translation was much less dependent on this factor. These findings are at variance with those of Blair et al. (1977) who found the opposite effect, namely that EMC virus and mengovirus RNA translation in an ascites reconstituted translation system required the addition of eIF-4A, whereas translation of poliovirus RNA was unaffected by its addition. It is unlikely that uncomplexed eIF-4A plays a major regulatory role in translation of poliovirus RNA, since it normally is the most abundant IF. The ratio of eIF-4A to ribosomes is 3 to 1, while other IFs are present at much lower factor-to-ribosome ratios of -0.5-1 to 1 (Duncan and Hershey, 1983). Therefore, it is expected that eIF-4A would saturate poliovirus RNA, making it as efficient as EMC virus RNA. Clearly, more work is required to elucidate the diverse requirements for translation of the different picornavirus mRNAs.

TRANSLATION BY POLIOVIRUS

199

REMARKS AND PERSPECTIVES VI. CONCLUDING The specific inhibition of host protein synthesis after poliovirus infection serves as a classical example of the ability of viruses to direct the cellular macromolecular synthesis machinery to their advantage. In spite of the fact that this phenomenon was first documented about 25 years ago (Franklin and Baltimore, 1962; Zimmerman et al., 1963), we are just now beginning to understand the molecular mechanisms involved in this remarkable biological process. The reasons for this long and dificult research history is that eukaryotic protein synthesis is one of the most complex processes that takes place in the cell, and thus can be regulated at many stages. Only complete understanding of this regulation will lead to better understanding of the mechanisms by which viruses interdict host protein synthesis. One striking example of our ignorance in the understanding of the control of protein synthesis is the role of the cellular cytoskeleton in translation. Lenk et al. (1977)have suggested that protein synthesis in eukaryotes takes place on a cytoskeleton framework (Wolosewick and Porter, 1976). It was proposed that following poliovirus infection, host mRNAs are released from the cytoskeleton and replaced by poliovirus RNA (Lenk and Penman, 1979). It is striking, however, that release of host mRNAs from the cytoskeleton following virus infection is unique to poliovirus and does not occur with other viruses such as VSV or reovirus, although they also cause the inhibition of host protein synthesis (Bonneau et al., 1985). This raises the interesting possibility that cleavage of the CBP complex p220 component is perhaps implicated in mRNA dissociation from the cytoskeleton. This example again demonstrates that better understanding of cell structure and function is required for the elucidation of poliovirus regulation of host protein synthesis. Future research will undoubtedly concentrate on the purification of the poliovirus or cellular activity that is directly responsible for the translational shutoff. Availability of the protein would enable the characterization of the p220 cleavage reaction. It would be of great interest to find specific inhibitors of this reaction or to isolate mutant forms of p220 that cannot undergo cleavage after poliovirus infection. Another direction of research is to elucidate the mechanism by which poliovirus RNA is translated by a cap-independent mechanism. This is an intriguing question, because this mechanism is unique to a small number of viral RNAs. Preliminary results from work done in our laboratory (J. Pelletier) in collaboration with V. Racaniello and G. Kaplan (Columbia University) suggest the existence of a sequence in the 5’ noncoding region of poliovirus RNA that confers the cap inde-

NAHUM SONENBERG

pendence of poliovirus translation. It is also possible that initiation of translation on poliovirus RNA occurs by direct binding of ribosomes t o the initiator AUG (internal initiation). This hypothesis has been suggested previously (Perez-Bercoff,1982) and is consistent with the observation that ribosomes do not accumulate on the long leader sequence of poliovirus RNA when elongation of protein synthesis is blocked. It is also of interest that the 5’ noncoding sequence of poliovirus RNA contains an adenine-uridine sequence (about 80 nucleotides) just preceding the initiator codon (Dorner et al., 1982). In a very recent report Herman (1986) demonstrated internal initiation of translation on a VSV mRNA, thus supporting the contention that internal initiation of translation can occur on eukaryotic mRNAs. Finally, research is being conducted in several laboratories to understand why the modified CBP complex cannot function in initiation of translation of capped cellular mRNAs.

ACKNOWLEDGMENTS This review was prepared during my sabbatical stay in Dr. David Baltimore’s laboratory at the Whitehead Institute. I am grateful to Dr. Baltimore for his support and encouragement. I am indebted to Y. Sonenberg for typing this manuscript. I thank A. J. Shatkin, E. Ehrenfeld, K. Lee, I. Edery, J. Pelletier, H. Bernstein, P. Sarnow, K. Kirkegaard, and G. Daley for their useful comments. Research done in the author’s laboratory was supported by the Medical Research Council and the National Cancer Institute of Canada. The author was supported by a Terry Fox Cancer Research Award from the National Cancer Institute of Canada.

REFERENCES Abreu, S. L., and Lucas-Lenard, J. (1976).J. Virol. 18, 182-194. Alonso, M. A., and Carrasco, L. (1982a).EMBO J . 1, 913-917. Alonso, M.A., and Carrasco, L. (1982b).J . Gen. Virol. 61,15-24. Altmann, M., Edery, I., Sonenberg, N., and Trachsel, H. (1985).Biochemistry 24,60856089. Bablanian, R. (1975).Prog. Med. Virol. l9,40-83. Baglioni, C., Simili, M., and Shafritz, D. A. (1978).Nature (London) 275, 240-243. Baltimore, D. (1969).In “The Biochemistry of Viruses” (H. B. Levy, ed.), pp. 101-176. Dekker, New York. Banejee, A. K. (1980).Microbiol. Rev.44, 175-205. Bergmann, J.E., Trachsel, H., Sonenberg, N., Shatkin, A. J.,and Lodish, H. F. (1979).J. Biol. Chern. 254, 1440-1443. Bernstein, H.D., Sonenberg, N., and Baltimore, D. (1985).Mol. Cell. Biol. 5,2913-2923. Blair, G. E., Dahl, H. H. M., Truelsen, E., and Lelong, J. C. (1977).Nature fLondon) 265, 651-653. Blinov, V. M., Gorbalenya, A. E., and Donechenko, A. P. (1984).Dokl. Akad. Nauk USSR 279,502-505. Blinov, V. M., Donechenko, A. P., and Gorbalenya, A. E. (1985).Dokl. Akad. Nauk USSR 281,984-987. Bonatti, S., Sonenberg, N., Shatkin, A. J., and Cancedda, R. (1980).J. Biol. Chem. 255, 11473-11477.

TRANSLATION BY POLIOVIRUS

201

Bonneau, A. M., Darveau, A., and Sonenberg, N. (1985).J. Cell. Biol. 100,1209-1218. Bossart, W., and Bienz, K. (1981).Virology 111, 555-567. Brown, B. A., and Ehrenfeld, E. (1980).Virology 103, 327-339. Buckley, B., and Ehrenfeld, E. (1986).Virology 152,497-501. Canaani, D., Groner, Y.,and Revel, M. (1977).FEBS Lett. 64,326-331. Carrasco, L. (1977).FEBS Lett. 76, 11-15. Carrasco, L., and Smith, A. E. (1976).Nature (London) 284, 807-809. Celma, M. L., and Ehrenfeld, E. (1974).Proc. Natl. Acad. Sci. U S A . 71, 2440-2444. Choppin, P.W., and Holmes, K. V. (1967).Virology 33,442-451. Cooper, P.D., Steiner-Pryor, A., and Wright, P. J. (1973).Zntervirology 1, 1-10. Daniels-McQueen, S.,Detjen, B. M., Grifo, J. A., Merrick, W. C., and Thach, R. E. (1983). J . Biol. Chem. 258,7195-7199. Detjen, B. M., Jen, G., and Thach, R. E. (1981).J. Virol. 38, 777-787. Dorner, A. J., Dorner, L. F., Larsen, G. R., Wimmer, E., and Anderson, C. W. (1982).J. Virol. 42, 1017-1028. Doyle, S., and Holland, J. (1972).J. Virol. 9,22-28. Duncan, R., and Hershey, J. W. B. (1983).J. Biol. Chem. 258, 7228-7235. Duncan, R., Etchison, D., and Hershey, J. W. B. (1983).J . Biol. Chem. 258,7236-7239. Edery, I., Humbelin, M., Darveau, A., Lee, K. A. W., Milburn, S., Hershey, J. W. B., Trachsel, H., and Sonenberg, N. (1983).J . Biol. Chem. 258, 11398-11403. Edery, I., Lee, K. A. W., and Sonenberg, N. (1984).Biochemistry 23,2456-2462. Ehrenfeld, E. (1982).Cell 28, 435-436. Ehrenfeld, E. (1984).Compr. Virol. 19,177-221. Ehrenfeld, E., and Hunt, T. (1971).Proc. Natl. Acad. Sci. U S A . 68, 1075-1078. Ehrenfeld, E., and Lund, H. (1977).Virology 80,297-308. Etchison, D., and Fout, S. (1985).J. Virol. 54, 634-638. Etchison, D., Milburn, S. C., Edery, I., Sonenberg, N., and Hershey, J. W. B. (1982).J. Biol. Chem. 257, 14806-14810. Etchison, D., Hansen, J., Ehrenfeld, E., Edery, I., Sonenberg, N., Milburn, S., and Hershey, J. W.B., (1984).J. Virol. 51, 832-837. Fernandez-Munoz, R., and Darnell, J. E. (1976).J . Virol. 18, 719-726. Filipowicz, W.(1978).FEBS lett. 96, 1-11. Franklin, R. M.,and Baltimore, D. (1962).Cold Spring Harbor Symp. Quant. Biol. 27, 175-198. Gehrke, L., Auron, P. E., Quigley, G. J., Rich, A., and Sonenberg, N. (1983).Biochemistry 22,5157-5164. Golini, F., Thach, S. S., Birge, C. H., Safer, B., Merrick, W. C., and Thach, R. E. (1976). Proc. Natl. Acad. Sci. U.S.A. 73, 3040-3044. Grifo, J. A., Tahara, S. M., Leis, J. P., Morgan, M. A., Shatkin, A. J.,and Merrick, W. C. (1982).J . Biol. Chem. 257,5246-5252. Grifo, J. A., Tahara, S. M., Morgan, M.A., Shatkin, A. J., and Merrick, W. C. (1983).J. Biol. Chem. 258,5804-5810. Grifo, J. A,, Abramson, R. D., Satler, C. A., and Merrick, W. C. (1984).J. Biol. Chem. 259, 8648-8654. Hackett, P. B., Egberts, E., and Traub, P. (1978).Eur. J . Biochem. 83, 341-352. Hanecak, R., Semler, B. L., Anderson, C. W., and Wimmer, E. (1982).Proc. Natl. Acad. Sci. U.S.A. 79,3973-3977. Hansen, J. L., and Ehrenfeld, E. (1981).J. Virol. 38,438-445. Hensen, J. L., Etchison, D., Hershey, J. W. B., and Ehrenfeld, E. (1982).J. Virol. 42, 200-207. Helentjaris, T., and Ehrenfeld, E. (1977).J. Virol. 21, 259-267. Helentjaris, T., and Ehrenfeld, E. (1978).J. Virol. 26, 510-521.

202

NAHUM SONENBERG

Helentjaris, T., Ehrenfeld, E., Brown-Luedi, M. L., and Hershey, J. W. B. (1979).J . Biol. Chem. 264, 10973-10979. Hellman, G. M., Chu, L.-Y., and Rhoads, R. E. (1982).J. Biol. Chem. 257,4056-4062. Herman, R. (1986).J . Virol. 58, 797-804. Herson, D., Schmidt, A., Seal, S. N., Marcus, A., and van Vloten-Doting, L. (1979).J . Biol. Chem. 254, 8245-8249. Hewlett, M. J., Rose, J. K., and Baltimore, D. (1976).P m . Nutl. Acud. Sci. U S A . 73, 327-330. Holder, J. W., and Lingrel, J. B. (1975).Biochemistry 14, 4209-4215. Holland, J. J. (1964).J. Mol. Biol. 8, 574-581. Jackson, R. J. (1982).In “Protein Biosynthesis in Eukaryotes” (R. Perez-Bercoff, ed.), pp. 362-418. Plenum, New York. James, L. A,, and Tershak, D. R. (1981).Can.J . Microbiol. 27, 28-35. Jen, G., and Thach, R. E. (1982).J. Virol. 43, 250-261. Jen, G., Birge, C. H., and Thach, R. E. (1978).J. Virol. 27,640-647. Jen, G., Detjen, B. M., and Thach, R. E. (1980).J. Virol. 35, 150-156. Kaempfer, R., Rosen, H., and Israeli, R. (1978).Proc. Natl. Acad. Sci. U.S.A. 75, 650654. Kaempfer, R., Van Emmelo, J., and Fiers, W. (1981).Proc. Natl. Acud. Sci. U.S.A. 78, 1542-1546. Katze, M.G.,Chen, Y.-T., and Krug, R. M. (1984).Cell 37, 483-490. Kaufmann, Y., Goldstein, E., and Penman, S. (1976).Proc. Nutl. Acud. Sci. U.S.A. 73, 1834-1838. Korant, B., Chow, N., Lively, M., and Powers, J. (1979).Proc. Nutl. Acad. Sci. U S A . 76, 2992-2995. Kozak, M. (1980).Cell 19, 79-90. Kozak, M. (1983).Microbiol. Rev.47, 1-45. Lacal, J. C.,and Carrasco, L. (1982).Eur. J . Biochem. 127,359-366. Lacal, J. C., and Carrasco, L. (1983).J . Gen. Virol. 64, 787-793. Lawrence, C.,and Thach, R. E. (1974).J . Virol. 14,598-610. Lax, S., Fritz, W., Browning, K., and Ravel, J. (1985).Proc. Nutl. Acad. Sci. U.S.A.82, 330-333. Lee, K. A. W., and Sonenberg, N. (1982).Proc. Nutl. Acad. Sci. U.S.A. 79,3447-3451. Lee, K. A. W., Guertin, D., and Sonenberg, N. (1983).J . Biol. Chem. 258, 707-710. Lee, K. A. W., Edery, I., and Sonenberg, N. (1985a).J. Virol. 54, 515-524. Lee, K. A. W., Edery, I., Hanecak, R., Wimmer, E., and Sonenberg, N. (1985b).J . Virol. 55,489-493. Leibowitz, R., and Penman, S. (1971).J . Virol. 8,661-668. Lenk, R., and Penman, S. (1979).Cell 16,289-301. Lenk, R., Ransom, L., Kaufmann, Y., and Penman, S. (1977).Cell 10,67-78. Lloyd, R. E., Etchison, D., and Ehrenfeld, E. (1985).Proc. Nutl. Acud. Sci. U S A . 82, 2723-2727. Lloyd, R. E., Toyoda, H., Etchison, D., Wimmer, E., and Ehrenfeld, E. (1986).Virology 160, 299-303. Lucas-Lenard, J. M. (1979).In “The Molecular Biology of Picornaviruses” (R. PerezBercoff, ed.), pp. 73-99. Plenum, New York. McCormick, W., and Penman, S. (1968).J. Virol. 2,859-864. Manak, M., Abreu, S., and Lucas-Lenard, J. (1975).ZNSERM 47, 387-396. Medvedkina, D. A., Scarlat, I. V., Kalinina, N. O., and Agol, V. I. (1974).Zntervirology 39, 4-8. Milburn, S.C . , Pelletier, J., Sonenberg, N., and Hershey, J . W. B. (1986).In preparation.

TRANSLATION BY POLIOVIRUS

203

Moldave, K. (1985).Annu. Rev. Biochem. 54, 1109-1149. Morgan, M. A., and Shatkin, A. J. (1980).Biochemistry 19,5960-5966. Mosenkis, J . S., Daniels-McQueen, S., Janovec, S., Duncan, R., Hershey, J. W. B., Grifo, J. A., Merrick, W. C., and Thach, R. E. (1985).J. Virol. 54,643-645. Nair, C . N., Stower, J. W., and Singfield, B. (1979).J . Virol. 31, 184-189. Neurath, H., Walsh, K. A., and Winter, W. P. (1967).Science 158, 1638-1644. Nomoto, A., Lee, Y. F., and Wimmer, E. (1976).Proc. Natl. Acad. Sci. U S A . 73,375380. Nuss, D. L., Oppermann, H., and Koch, G. (1975).Proc. Natl. Acad. Sci. U.S.A.72,12581262. Otto, M. J., and Lucas-Lenard, J. (1980).J. Gen. Virol. 50, 293-307. Patzelt, E., Blaas, D., and Kuechler, E. (1983).Nucleic Acids Res. 11, 5821-5835. Pelletier, J., and Sonenberg, N. (1985a).Mol. Cell. Biol. 5,3222-3230. Pelletier, J., and Sonenberg, N. (1985b).Cell 40,515-526. Penman, S.,and Summers, D. (1965).Virology 27,614-620. Pensiero, M. N., and Lucas-Lenard, J. M. (1985).J. Virol. 56, 161-171. Perez-Bercoff, R. (1982).In “Protein Biosynthesis in Eukaryotes” (R. Perez-Bercoff, ed.), pp. 245-252. Plenum, New York. Ray, B. K., Brendler, T. G., Adya, S.,Daniels-McQueen, S., Kelvin Miller, J., Hershey, J. W. B., Grifo, J . A., Merrick, W. C., and Thach, R. (1983).Proc. Natl. Acad. Sci.U S A . 80,663-667. Ray, B. K., Lawson, T. G., Kramer, J. C., Cladaras, M. H., Grifo, J. A., Abramson, R. D., Marrick, W. C., and Thach, R. E. (1985).J . Biol. Chem. 260,7651-7658. Reichel, P.A.,Merrick, W. C., Siekierka, J., and Mathews, M. B. (1985).Nature (London) 313,196-200. Rose, J . K., Trachsel, H., Leong, K., and Baltimore, D. (1978).Proc. Natl. Acad. Sci. U.S.A. 75, 2732-2736. Rosen, H., Knoller, S., and Kaempfer, R. (1981).Biochemistry 20, 3011-3020. Rosen, H., Di Segni, G. and Kaempfer, R. (1982).J. Biol. Chem.257,946-952. Rueckert, R. R., and Wimmer, E. (1984).J. Virnl. 50, 957-959. Rychlik, W., Gardner, P. R., Vanaman, T. C., and Rhoads, R. E. (1986).J. Biol. Chem. 261, 71-75. Saborio, J. L., Pong, S. S., and Koch, G. (1974).J. Mol. Biol.85, 195-211. Sarkar, G.,Edery, I., Gallo, R., and Sonenberg, N. (1984).Biochim. Biophys. Actn 783, 122-129. Sarkar, G., Edery, I., and Sonenberg, N. (1985).J . Biol.Chem. 260,13831-13837. Schneider, R. J., Safer, B., Munemitsu, S. M., Samuel, C. E., and Shenk, T. (1985).Proc. Natl. Acad. Sci. U S A . 82,4321-4325. Seal, S.N., Schmidt, A., and Marcus, A. (1983).Proc. Natl. Acad. Sci. U.S.A. 80,65626565. Seal, S. N., Schmidt, A., Marcus, A., Edery, I. and Sonenberg, N. (1986).Arch. Biochem. Biophys. 246, 710-715. Shaeffer, A,, Kuhne, J., Zibirre, R., and Koch, G. (1982).J. Virol. 44,444-449. Shafritz, D. A., Weinstein, J. A., Safer, B., Merrick, W. C., Weber, L. A., Hickey, E. D., and Baglioni, C. (1976).Nature (London) 261, 291-294. Shatkin, A. J. (1976).Cell 9,645-653. Shatkin, A. J . (1985).Cell 40,223-224. Shih, D. A., Shih, C. T., Kew, O., Pallansch, M., Rueckert, R., and Kaesberg, P. (1978). Proc. Natl. Acad. Sci. U.S.A. 75, 5807-5811. Siekierka, J., Mariano, T. M., Reichel, P. A., and Mathews, M. B. (1985).Proc. Natl. Acad. Sci. U.S.A. 82, 1959-1963.

NAHUM SONENBERG Sonenberg, N. (1981).Nucleic Acids Res. 9, 1643-1656. Sonenberg, N., and Shatkin, A. J. (1977).Proc. Natl. Acad. Sci. U.S.A. 74, 4288-4292. Sonenberg, N., and Shatkin, A. J. (1978).J. Biol. Chem. 253, 6630-6632. Sonenberg, N., Morgan, M., Merrick, W. C., and Shatkin, A. J. (1978).Proc. Natl. Acad. Sci. U.S.A.75,4843-4847. Sonenberg, N., Rupprecht, K., Hecht, S., and Shatkin, A. J. (1979).Proc. Natl. Acad. Sci. U S A . 76,4345-4349. Sonenberg, N., Trachsel, H., Hecht, S., and Shatkin, A. J. (1980).Nature (London) 285, 331-333. Sonenberg, N., Guertin, D., Cleveland, D., and Trachsel, H. (1981).Cell 27, 563-572. Sonenberg, N., Guertin, D., and Lee, K. A. W. (1982).Mol. Cell. Biol. 2, 1633-1638. Sonenberg, N., Edery, I., Darveau, A,, Humbelin, M., Trachsel, H., Hershey, J. W. B., and Lee, K. A. W. (1983).In “Protein Synthesis” (A. K. Abraham, T. S. Eikhom, and I. F. Pryme, eds.), pp. 23-43. Humana Press, Clifton, N.F. Staehelin, T., Trachsel, H., Erni, B., Boschetti, A., and Schreier, M. H. (1975).Proc. FEBS Meet. 10th 39, 309-323. Steiner-Pryor, A., and Cooper, P. D. (1973).J. Gen. Virol. 21, 215-225. Svitkin, Y.V., Ugarova, T. Yu., Ginevskaya, V. A,, Kalinina, N. O., Scarlat, I. V., and Agol, V. I. (1974).Intervirology 4, 214-220. Tahara, S. M., Morgan, M. A., and Shatkin, A. J. (1981).J . Biol. Chem. 256,7691-7694. Tahara, S . M., Morgan, M. A,, and Shatkin, A. J. (1983).J . Biol. Chem. 258, 1135011353. Toyoda, H.,Nicklin, M. J. H., Murray, M. G., Anderson, C. W., Dunn, J. J., Studier, F. W., and Wimmer, E. (1986).Cell 45,761-770. Trachsel, H., Sonenberg, N., Shatkin, A. J., Rose, J. K., Bergmann, J. E., Gordon, J., and Baltimore, D. (1980).Proc. Natl. Acad. Sci. U.S.A. 77, 770-774. Van Vloten-Doting, L., Bol, J., Neeleman, L., Rutgers, T., Van Dalen, D., Castel, A., Bosch, L., Marbaix, G., Huez, G., Hubert, E., and Cleuter, Y. (1977).In “NATO Advanced Studies on Nucleic Acids and Protein Synthesis in Plants” (L.Bogorad and M. Weil, eds.), pp. 387-411. Plenum, New York. Webb, N. R., Chari, R. V. J., De Pillis, G., Kozarich, J. W., and Rhoads, R. E. (1984). Biochemistry 23, 177-181. Weber, L.A., Feman, E. R., Hickey, E. D., Williams, M. C., and Baglioni, C. (1976).J. Biol. Chem. 251, 5657-5662. Whitaker-Dowling, P., and Youngner, J. S. (1984).Virology 137, 171-181. Wolosewick, J. J., and Porter, K. R. (1976).Am. J. Anat. 147,303-324. Wright, P. J., and Cooper, P. D. (1974).Virology 59, 1-20. Zimmerman, E. F., Heeter, M., and Darnell, J. E. (1963).Virology 19, 400-408.

ADVANCES IN VIRUS RESEARCH,VOL. 33

DISEASE INDUCTION BY PLANT VIRUSES L. C. van Loon Department of Plant Physiology Agricultural University 6703 BD Wageningen, The Netherlands

I. INTRODUCTION Viruses are responsible for severe diseases in plants resulting in major losses in commercially important crops. Little is known about how these diseases are induced and why and how symptoms arise. The intimate relationship between host cellular metabolism and virus replication precludes both the study of disease-inducing determinants in uitro and effective disease control once a plant has been infected. Thus, an understanding of the mechanisms of disease induction can be derived only from comparisons of the cytological, physiological, and biochemical differences between healthy and infected plants and in this way measures for controlling viral pathogenesis may eventually be devised. For disease symptoms to become apparent, a virus has to multiply and spread. Single cells may become infected but with restriction of viral replication to directly inoculated cells within an organ or tissue the infection will pass unnoticed. Only when a virus is able to move and spread to neighboring cells can it initiate reactions on a suffliciently large scale for macroscopic symptoms to become apparent. Isolated protoplasts, when infected, often show cytological alterations related to viral replication. However, neither necrotic reactions nor developmental abnormalities are expressed in the absence of a cell network. Hence, the pathogenic action of a virus can be recognized only by its expression through the plant, and disease has to be studied as a plant phenomenon. Since virus multiplication and spread are necessary for symptoms of disease to develop, it can be anticipated that the rate and extent to which these processes occur are primary determinants of symptom severity and that they may also influence symptom type. However tolerance may occur, in which viruses multiply and spread through the plant while the plant remains almost normal in appearance. Additional factors must exist, therefore, that underlie the pathogenic ac205

Copyright 0 1987 by Academic Press, Inc. All righta of reproduction in any form reserved.

206

L. C . VAN LOON

tion of a virus. In any specific host such viral determinants may interact, on the one hand, with the factors governing accumulation and spread and, on the other hand, with host factors involved in plant reactions. Numerous changes in the metabolism of plants upon infection with viruses have been documented. However, such changes usually become evident only after symptoms have become discernable, i.e., far too late to be involved in the induction of disease. Furthermore, during inoculation only very few cells appear to function as entry sites for the virus and, thus, any changes occurring early after viral penetration may escape detection due to the large amount of surrounding cells that are still healthy and exhibit normal characteristics. Nevertheless, the nature of the symptoms induced by viruses, as well as the study of mutants that induce variant symptoms, can provide clues to what kind of underlying processes may be involved in regulating viral pathogenesis. Besides virus multiplication, none of these processes has yet been studied in any detail. Based on the scattered evidence available I shall, therefore, attempt to speculate on the mechanisms of disease induction and the regulation of pathogenesis and symptom expression in virus-infected plants, and be deliberately provocative in presenting a hypothesis which, in my opinion, is worth testing.

11. THEINTERACTIONS OF VIRUSESWITH THEIRHOSTS

A. Induction of Symptoms The symptoms induced by viruses vary from hardly visible local discolorations to severe perturbations of growth and development or even death of the whole plant. Any of these symptoms may be provoked by other conditions, but their distribution is usually highly specific and often confined to certain organs or tissues (Bos, 1978). Chlorosis and mosaics only occur in tissues with functional plastids; malformations such as enations may be confined to veins; some viruses only affect generative organs. Such specific relationships may be caused by the distribution of the virus within the host. Viruses appear to be transported both from cell to cell and within the vascular bundles, and their site of introduction into the plant and the direction and rate of their spread may be responsible for the differential manifestation of disease symptoms. Alternatively, symptoms may arise due to a special sensitivity of certain organs, tissues, or organelles. Electron microscopic examination of stained thin sections has revealed various cytological effects, usually associated with the presence

DISEASE INDUCTION BY PLANT VIRUSES

207

of virus particles. These may range from membrane proliferation and vacuolation of the cytoplasm to the disintegration of all cell organelles. Often organelles such as nuclei, chloroplasts, or mitochondria are specifically affected. Many plant viruses induce distinctive nuclear or cytoplasmic inclusions containing virus particles, host constituents, and protein products of viral origin. Potyviruses induce pinwheels, scrolls, and laminated aggregates containing virus-coded, nonstructural polypeptides (Dougherty and Hiebert, 1980a,b). Such cytopathic structures and similar viroplasms apparently function as sites of virus synthesis and assembly (Martelli and RUSSO,1977). Some viruses are exclusively found in the nucleus, others are in chloroplasts, but the majority are located in the cytoplasm, where they may be associated with certain membranous structures. It is questionable whether replication always occurs at those subcellular locations where complete viral particles can be observed. Using antisera against subunits of the viral replicase complex, Dorssers et al. (1984) established that cowpea mosaic virus (CPMV) replication is associated with the membranes present in the cytopathic structures induced in infected cells (De Zoeten et al., 1974). Similarly, Bov6 and Bov6 (1985) have demonstrated that turnip yellow mosaic virus (TYMV)replicates at the chloroplast envelope of Chinese cabbage leaf cells. Attempts to localize the site of replication of other viruses by in situ hybridization with cloned cDNA probes have not yet provided definitive answers. So far, it can only be said that the occurrence of virus particles in vacuoles, in which synthesis of nucleic acids and proteins cannot take place, clearly indicates that the subcellular localization of a virus can be subject to change. Hence, its presence in certain cells and association with specific organelles in electron micrographs of ultrathin sections from infected tissues provide at most only a starting point for determining where the damaging action of the virus may be initiated. Different viruses, or even isolates that on the basis of their serological relationship are considered strains of a single virus, may induce very different types of symptoms on a single plant species, indicating that viruses contain genetic information specifying symptom type. In fact, the occurrence of differential symptoms on a test host is the basis for the recognition that different pathogenic agents are involved, and before the advent of molecular virology, virus classification has been largely based on host plant and symptom type (e.g., tobacco mosaic virus). However, in different plant species, the same virus may induce very different types of symptoms. Often, different symptoms are induced even in different varieties or cultivars of a single plant species. Two conclusions can be drawn from these generalizations: (1)the virus appears to contain pathogenic determinants for more than a single

208

L. C. VAN LOON

symptom type, and (2) a susceptible plant has various abilities to react to virus infection. Within single plant species such diversity can be exploited for selection and breeding for resistance, the main strategy still followed to control virus diseases. The phenotypic expression of symptoms depends on the genetic constitution of both the host plant and the virus and is characteristic of the particular combination of both. Varying this combination can yield insight into the complexity of the disease-inducing functions of the virus and of the pathways involved in the reactions of the infected plant. In principle, inoculating a single virus on a range of test hosts and observing the symptoms induced could show how many types of symptoms can be specified by the virus. Conversely, inoculating a single test host with various viruses would give an indication of the spectrum of possible reaction types of the plant. However, unlike in fungal and bacterial diseases where the metabolism of the pathogen gives rise to macromolecules and metabolites that may be transported to host cells and affect host metabolism, for instance by functioning as toxins, viruses do not possess a metabolism of their own. Their interdependency on host cell metabolism may give rise to symptoms that are unique to the particular combination of virus and host. Because both the viral and the host plant contribution to symptom development have a genetic basis, genetic analysis is the method of choice to analyze the determinants of symptom induction and expression. The use of molecular biological techniques such as nucleic acid sequencing and DNA cloning will greatly facilitate the elucidation of the genes involved in disease determination. Already, the nucleotide sequences of several viroids as well as of some viruses have been determined, yielding information about their structure and capacity to code for protein. The major challenge now is to learn the biological function(s) of virally coded proteins, the nature and complexity of the molecular structures of the host involved in host-virus recognition and initiation of disease, and the way in which defense reactions are activated or suppressed.

B . Viral Properties Specifying Symptom Type 1 . Effects of Mutagenesis The study of the types of symptoms specified by a virus has been facilitated by the possibility of generating phenotypical mutants. Since the majority of plant viruses possess single-stranded (ss)RNA as the genetic material, any nonlethal mutational event will lead to a nucleotide change that may affect the amino acid sequence of a virus-

DISEASE INDUCTION BY PLANT VIRUSES

209

coded protein or a regulatory function of the RNA. Mutants can be most easily distinguished on the basis of altered symptomatology, and many mutant viral strains have been described which are characterized by an altered phenotype on a specific host. As early as 1958, Mundry and Gierer observed that mutation with nitrous acid of tobacco mosaic virus (TMV) strain uulgare led to a loss of the ability to induce systemic mosaic symptoms on Java tobacco. Instead, local necrotic lesions, characteristic of a hypersensitive response, were formed. This transition was apparently accomplished by only one out of 180200 possible deaminations and, because uridylic acid cannot be deaminated, the size of the corresponding cistron was estimated to comprise 250-300 nucleotides. In a screening experiment, the total number of viable mutants exceeded the percentage transition from systemic mosaic symptoms to local lesions by about a factor of eight (Mundry, 1965). These mutants could be classified in 12 different groups on the basis of symptom type on Samsun and Java tobacco (Mundry and Gierer, 1958). Although selection of contaminating variant strains could not be excluded, a mathematical analysis indicated that the experimental data could not be explained by selection of surviving variants, but were both quantitatively and qualitatively in accordance with mutation (Mundry, 1965). As the total mutation rate was roughly eight times that of generating local lesion mutants on Java tobacco, the number of nucleotides involved was estimated to be 2000-2400. A single hit out of about 3000 nucleotides was found to be lethal (Schuster and Schramm, 19581, implying that about 4000 nucleotides take part in vital functions. Together these numbers account for all of the TMV genome, and indicate that roughly one-third of the genetic information can somehow specify symptom type. Mutants of the common strain of TMV producing local lesions on Nicotiana syluestris were likewise induced by Kado and Knight (1966). After stripping the protein subunits to various degrees from one end of the virus with sodium dodecyl sulfate (SDS), they exposed the partially stripped virus to nitrous acid. A pronounced rise in the ratio of local lesion mutants occurred when about 75% of the RNA had been stripped. On these grounds, they defined the mutation site to lie at approximately 75%from the 3’ end of the viral RNA. However, Wilson et al. (1976) and Ohno and Okada (1977) established that preferential stripping by SDS occurred from the 5’ rather than the 3’ end and thus, the mutation must have occurred at about three-quarters of the length of the TMV RNA from the 5’ end. The cistron alledgedly responsible for specifying local lesions on N . syluestris was named the “local lesion gene” (Kado and Knight, 1966). Despite many attempts, in uitro the reverse mutation, of a strain

210

L. C. VAN LOON

causing necrotic lesions on Java tobacco or N.syluestris to one inducing systemic mosaic symptoms, has never occurred. Spontaneously arising natural mutants overcoming localization have occasionally been reported (e.g., Holmes, 1952; Yarwood, 1979; Lakshman et al.,1985). In most cases, selection of preexisting variant strains has not been ruled out. However, systemic variants of CPMV overcoming hypersensitivity in the cowpea cv. Early Red appear to be generated by spontaneous mutation (De Jager and Wesseling, 1981).Nevertheless, Jones and Dawson (1978) have argued that reversion of mutants to wild-type phenotype is an event sufficiently rare that essentially pure stocks of mutants can easily be maintained. Mutation is more likely to impair the expression of genetic information than to restore it. The rather common transition from a systemic mosaic-inducing to a local lesion strain of TMV and the inability to generate the reverse mutation in vitro can best be interpreted as the loss of a function enabling the virus to circumvent a hypersensitive reaction and to spread systemically throughout the entire plant. Hence, the term “local lesion gene” seems misleading, as the gene might be necessary for systemic spread rather than tissue localization. Temperature sensitivity of systemic spreading in the nitrous acid mutant Ni 2519 (Jockusch, 1966a, 1968; Taliansky et al.,1982a) likewise suggests that an active viral function is required to evade localization in the host. Natural variants of TMV that have lost the ability to spread systemically generally are of low apparent infectivity. Holmes (1952)has suggested that infectivity and ability to spread systemically are closely linked or actually controlled by a single mechanism. The presence in the viral genome of a function determining systemic spreading appears to be widespread if not general, as evidenced by the occasional appearance of chlorotic or necrotic spots on leaves exhibiting mosaic symptoms. RNA viruses show relatively high mutation frequencies due to the absence of proofreading during replication (Holland et al., 19821, resulting in a heterogeneous viral RNA population but with one predominant genotype. Isolates of TMV strain U1 generated via a single local lesion from minimum-dose inoculation were found to contain two classes of variants: (1)distantly related strains of TMV which are presumed to have been passaged as contaminants along with the U1 strain, and (2) variants that because of their close sequence homology to the U1 strain were considered to have arisen by spontaneous mutation. These variants were isolated from chlorotic or necrotic spots on the leaves of U1-infected plants (Garcia-Arena1 et al., 1984) and had apparently lost the ability to spread systemically. Such observations can only be reconciled with the easy maintenance of es-

DISEASE INDUCTION BY PLANT VIRUSES

211

sentially pure stocks (Jones and Dawson, 1978), if the variants are of sufficiently lower fitness to be naturally selected against. However, in a restrictive host, variants escaping restriction will be favored. Indeed, occasionally variant strains have been selected that overcome the resistance in particular host plants, such as the strain NCP of southern bean mosaic virus on certain cowpea lines (McGovern and Kuhn, 1984). This strain must likewise contain determinants specifying systemic movement within these particular hosts. 2 . Pseudorecombinants In multicomponent viruses, individual components from different strains or characterized mutants can sometimes be exchanged without loss of infectivity. Such pseudorecombinants provide information on which of its components genetic information for symptom type is located. In an isolate of alfalfa mosaic virus (AMV) spontaneous mutants continually arise that have lost the ability to systemically invade bean plants. By in uitro recombination of the four components of wildtype virus, this property could be assigned to the M component which contains RNA 2, the next to the largest RNA type (Dingjan-Versteegh et al., 1972). From the naturally occurring strain 425, which likewise induces only necrotic local lesions on bean plants, a mutant was produced by UV irradiation of the M component that induced systemic chlorosis in both bean and cowpea. This mutant multiplied to substantially higher levels in cowpea protoplasts than the parent strain and the relative amount of the M component was strikingly increased. This effect could be suppressed by the addition of strain 425 M component to the mutant inoculum. The progeny induced both systemic chlorosis and necrotic local lesions in bean, indicating that both types of RNA had been replicated. Since supplementation with the M component of the parent strain eliminated the altered balance between the viral nucleoproteins but not the altered phenotype, it appears that RNA 2 harbors at least two functional domains, one of which seems to be related to the induction of a host resistance response (Roosien et al., 1982, 1983). If more than a single character is being screened, generally determinants can no longer be assigned to a single component. Thus, in AMV, the kinds of symptoms on tobacco are determined by RNA 3 (DingjanVersteegh et al., 1972; Hartmann et al., 1976). In cowpea chlorotic mottle virus (CCMV)and the related brome mosaic virus (BMV) RNAs 2 and 3 determined local and primary lesion morphology on Chenopodium hybridum, whereas CCMV RNA 2 was responsible for systemic symptom changes in cowpea (Bancroft and Lane, 1973). Howev-

212

L. C.VAN LOON

er, RNA 1 was found to influence symptoms in susceptible cowpeas and to control systemic invasion of cowpea cultivars that are resistant to some strains of the virus (Wyatt and Kuhn, 1980). In cucumber mosaic virus (CMV), the ability to systemically infect bean, pea, or cowpea is likewise determined by RNA 2 (Marchoux et al., 1975; Edwards et al., 19831, but the property of moving systemically in Nicotianu tabacum has been assigned to RNA 1 (Lakshman and Gonsalves, 1985). Local lesion size on Vignu spp. has been attributed variously to RNA 3 and RNA 2, and the induction of necrotic local lesions on cotyledons of Cucurbita pep0 to RNA 1 (Lakshman and Gonsalves, 1985),whereas local lesion size on this host as well as symptom type on N. tabacum cv. Xanthi-nc, Beta vulgaris, and Zea mays are linked to RNA 3 (Marchoux et al., 1975). By exchanging the genomic RNAs of three strains of CMV and inoculating the pseudorecombinants to 10 selected host plant species, Rao and Francki (1982) confirmed that all three viral RNAs are involved in symptom induction. Some host reactions were controlled by only one specific RNA component, whereas others could be controlled by either one or another. Some reactions were also identified as being controlled by the interaction of two or perhaps even three components. Determinants for lesion size and color on Vigna cylindrica and dwarfing of tobacco were shared by top, M, and B components of tobacco streak virus (TSV) (Fulton, 1975). Whereas RNA 1 of sweet clover necrotic mosaic virus (SCNMV)was essential for systemic infection of sweet clover, its RNA 2 complemented RNA 1 of red clover necrotic mosaic virus (RCNMV) in causing local infection. RNA 1 of clover primary leaf necrosis virus when combined with RNA 2 of SCNMV or RCNMV acquired the ability to infect white clover and also caused symptoms characteristically different from those induced by the parental viruses on kidney beans (Okuno et al., 1983). RCNMV RNA 2 also determined lesion morphology in cowpea and ability to invade the plants systemically (Osman et al., 1986). RNA 2 of tobacco rattle virus (TRV) strain OR contains a determinant to induce yellow symptoms (Lister and Bracker, 1969). This ability was lost in an isolate in which about 150 nucleotides of this RNA were deleted but which was otherwise about 90% homologous (Robinson, 1983). For raspberry ringspot virus, the smaller RNA likewise determines systemic yellowing symptoms in Petunia hybrida, while the larger RNA determines infectivity on Lloyd George raspberry, the severity of systemic symptoms in Chenopodium quinoa, and the ability to systemically infect Phaseolus vulgaris (Harrison et al., 1974). For a different nepovirus, cherry leaf roll virus, the B component similarly determined the ability to induce systemic symptoms in Gomphrena

DISEASE INDUCTION BY PLANT VIRUSES

213

globosa (Jones and Duncan, 1980;Haber and Hamilton, 1980). However, whereas Jones and Duncan (1980) found the B component to also determine the lesion type and severity of systemic symptoms in Chenopodium and Nicotiam spp., Haber and Hamilton (1980) observed the M component to be the primary determinant of symptom type in these latter hosts. Both RNA-containing components of CPMV carry determinants for symptom type (De Jager, 1978). B component RNA is able to replicate independently from the presence of M component (Goldbach et al., 1980) but is incapable of spreading to surrounding cells (Rezelman et al., 1982). Indeed, as shown in Fig. 1, a mutation in the M component could lead to the loss of the ability to systemically invade bean plants. However, a mutation in B RNA caused the virus to remain localized in cowpea cv. Early Red. Recently, evidence has been presented that B RNA determines infectivity on cowpea, as a mutant was isolated which was able to grow in bean but not in cowpea (Evans, 1985). Based on in uitro recombination, supplementation, and reassortment tests, De Jager (1976, 1978; De Jager and Breekland, 1979) characterized several mutants showing atypical local lesions on Pinto bean and Early Red cowpea as well as defective symptom development on the systemic hosts Noordhollandse bruine bean and Blackeye cowpea. Both local lesion morphology and systemic spreading in the four test hosts were affected by mutations in either the M or the B component, indicating that both viral components specify functions controlling symptom development. Individual mutants showed changes in symptoms and multiplication in more than one host, suggesting that the same genetic determinants govern several phenotypic properties. However, in all cases in which the ability to spread systemically in a certain plant host was lost, symptoms on other hosts became less severe. Lesions on hosts reacting hypersensitively to the virus remained smaller or became chlorotic rather than necrotic, chlorotic lesions spread less, and mosaic symptoms occurring in other hosts were milder. In the few cases where the extent of multiplication of these mutants has been determined, it was found to be less than that of wildtype virus, suggesting that reduced virus multiplication may underlie the apparent pleiotropic effect of quite a few mutations in CPMV. 3. Viral Multiplication In general, however, there is no correlation between the rate of virus multiplication and symptom severity. For instance, both TMV and barley yellow dwarf virus induce severe diseases but yields of extractable virus are typically 5-10 mg and 50-150 ng per gram of tissue, respectively (Zaitlin, 1979). In virus-host combinations leading

FIG.1. (A) On bean, wild-type cowpea mosaic virus (CPMV) (left) causes chlorosis and veinal necrosis on the inoculated leaves, and veinal necrosis, mosaic, and malformations on the leaves that develop after inoculation. An artificially derived mutant in the M component (right) induces the same symptoms in the inoculated leaves but systemic spread does not occur. (B)In Early Red cowpeas, local necrotic lesions develop on inoculated leaves after infection both with CPMV strain Sb (left) and with the naturally occurring variant isolate Sb-S8 (right). Only with Sb-S8 do systemic symptoms develop and this property resides in the B component. (Courtesy of Dr. C. P. de Jager.)

DISEASE INDUCTION BY PLANT VIRUSES

215

to chlorosis and mosaic symptoms, loss of chlorophyll does not necessarily accompany infection and virus multiplication. Both protoplasts inoculated in uitro and attached leaves inoculated whan nearly fully mature often remain symptomless, even when they accumulate large amounts of virus. Host genotype can modify expression of symptoms independently from virus replication (Kuhn et al., 1981). The synthesis of viral RNA and protein requires nucleotides and amino acids as well as energy derived from photosynthesis and respiration (Kano, 1985). In general and at least initially, viral multiplication does not occur at the expense of host RNA and protein synthesis. Eventually, nucleotides and amino acids may become available through increased RNA and protein turnover. In the absence of any specific preexisting nutritional stress, however, symptoms are not a direct consequence of exhaustion of the host due to its being forced to synthesize additional nucleic acid and protein (Matthews, 1981). In tobacco developing mosaic symptoms as a result of TMV infection, protein synthesis was reduced by up to 75% during rapid virus multiplication but later recovered. Since the virus did not cause any alteration in the concentration of host polyadenylated mRNA, it appears that viral protein synthesis does compete with host protein synthesis (Fraser and Gerwitz, 1980). However, closely related strains of the same virus may multiply in a particular host to give a similar final concentration of virus and yet may have different effects on cell constituents and induce symptoms of greatly differing severity (Matthews, 1980). Thus, although infection of tobacco with the P6 strain of CMV causes a severe chlorosis in contrast to the mild symptoms induced by the W strain, the time course and extent of accumulation of infectivity were essentially similar for both strains (Roberts and Wood, 1981a). The “masked strain” of TMV does not induce visible symptoms on tobacco, although it multiplies to significant levels. In contrast, the fluuum strain and several nitrous acid mutants (Jockusch, 1966a) induce severe yellow mosaic symptoms but multiply far less than the common strain. In studies by Whenham et al. (1985)the extent of virus accumulation of several strains of TMV was determined together with the severity of mosaic symptoms and shoot growth reduction in tobacco. Symptom severity was unrelated to virus accumulation but a correlation coefficient of 0.73 was found between reduced growth and virus content, suggesting that increased virus multiplication is associated with a reduction in relative growth rate. Multiple regression analysis further indicated that in tobacco virus multiplication could account for about half of the variation in growth. However, in tomato variation in growth was not related to variation in virus multiplication

216

L. C. VAN LOON

(Fraser et al., 1986). Since TMV multiplied to higher concentrations overall in tobacco than in tomato, the implied importance of extent of virus multiplication in the growth control of infected tobacco may reflect the very high level of virus synthesis. These observations indicate that higher virus multiplication can increase symptom severity, but this conclusion may hold only for related strains or mutants sharing comparable determinants for symptom type on one specific host, This point is dramatically illustrated by the modulation of symptoms through the presence of viral satellites. Viral satellites are small nucleic acids that can replicate only in conjunction with the nucleic acids of certain “helper” viruses in certain hosts. CMV-associated RNA 5 (CARNA 5) is encapsidated with and dependent on, but not part of, the viral genome, and its presence in CMV strains changes the characteristic symptoms induced in tomato from chlorosis and “fern leaf’ syndrome to lethal systemic necrosis (Kaper and Waterworth, 1977). In contrast, in several other host species disease symptoms are attenuated (Waterworth et al., 1979; Habili and Kaper, 1981). Under the latter conditions, CMV synthesis can be almost totally suppressed. However, partial suppression of viral synthesis in tomato leads to symptom changes rather than their attenuation. Several variants of CARNA 5 have been described that differ in their effects on symptoms of CMV in different hosts (e.g., Collmer et al., 1983). The satellite RNA 2 of lucerne transient streak virus (LTSV) alone does not infect Chenopodium amaranticolor, whereas the viral RNA 1 does. The latter alone induces chlorotic lesions but when RNA 2 is added, the lesions induced are necrotic (Jones et al., 1983). A satellite RNA (RNA C) associated with turnip crinkle virus increased the severity of symptoms in plants infected with the genomic RNA A (Altenbach and Howell, 1981). Likewise, a viroid-like RNA behaving as a satellite of velvet tobacco mottle virus is responsible for the severe symptoms in N. clevelandii induced by field isolates (Francki et al., 1986). The nettlehead disease of hop is associated with the presence of a satellite in arabis mosaic virus isolates; when transmitted to C. quinoa, such isolates induce unusually severe symptoms. The severity of these symptoms was related to the amount of satellite RNA in virus prepared from the plants (Davies and Clark, 1983). A low molecular weight RNA associated with tomato bushy stunt virus infection altered symptom expression in several experimental hosts. Notably, attenuation in N . clevelandii was associated with the suppression of low molecular weight RNA replication (Hillman et al., 1985). Thus, differences in the extent of virus multiplication can explain some but certainly not all variations in symptom severity.

DISEASE INDUCTION BY PLANT VIRUSES

217

During the replication of ssRNA viruses, double-stranded structures function as intermediates. The postulated “replicative form” and the “replicative intermediate” do not have counterparts in the metabolism of normal, healthy cells. It has been supposed, therefore, that these structures could play a role both in determining whether a plant is susceptible and in inducing in susceptible plants reactions that lead to specific disease symptoms. However, it is difficult to see how the many single nucleotide changes that underlie specific changes in symptom type upon mutation can have a basis at the rather nonspecific structural level of double-stranded RNA. Alternatively, one would have to suppose that sequence-specific stretches function as determinants, leading in each case to a different specific type of interaction. Enzymes responsible for the formation of double-stranded structures, RNA-dependent RNA polymerases, are already present in noninfected plants (Fraenkel-Conrat, 1979). Proteins interacting with doublestranded RNA might thus be present in healthy plants and could, in principle, be sequestered upon infection by the viral components, leading to alterations in metabolism. However, the occurrence of doublestranded RNA and the role of RNA-dependent RNA polymerases in healthy plants are unknown, so that evidence for such mechanisms is entirely lacking. Moreover, host plant specificity is commonly lost at the cellular level. Apart from a few exceptions where the resistance of host plant lines toward a particular virus is maintained in protoplasts, individual plant cells are capable of replicating essentially any virus. Thus, in single cells susceptibility rather than resistance is the common condition. Apparently, virus replication occurs in a way analogous to the fundamental processes governing nucleic acid metabolism in every plant cell. Disease symptoms, on the contrary, require the interaction of a number of interconnected cells and are fully dependent on a direct exchange of information between those cells.

C . Virus-Coded Proteins and Their Possible Functions 1 . The Number of Virus-Coded Proteins Is Limited Each viral RNA specifies one or more translational products that may be candidates for symptom-inducing factors. However, the number of proteins that are coded for by a typical plant virus appears much smaller than the variety of symptom types that may be induced on different hosts or that are displayed by series of mutants on a single host. All “stable” viruses contain genes for one or more coat proteins and for at least one large polypeptide that is considered to play a role

218

L. C. VAN LOON

in replication. Besides, viruses transmitted by vectors code for one or more proteins apparently necessary for arthropod or nematode transmission. Since these viral products are needed for the basic functions of virus multiplication and transmission, their involvement in inducing symptoms may at first sight seem less likely. However, it is not inconceivable that symptoms are a secondary consequence of the interaction of one or more of these proteins with specific cellular components. a. Coat Protein. Von Sengbusch (1965) postulated that the amino acid sequence of the coat protein of TMV influences symptom type in tobacco. Severe yellow mosaic symptoms develop when the plant is infected with mutants that, with regard to the parent strain, uulgure, exhibit a lower electrophoretic mobility under alkaline conditions. All these mutants were found to be temperature sensitive in the sense that at 32°C only a fraction of the stable infectious material produced at 23°C was recovered (Jockusch, 1966a,b, 1968). Conversely, not all temperature-sensitive mutants displayed a change in electrophoretic mobility and associated severe yellow mosaic symptoms. All temperature-sensitive mutants were characterized by a defect in the assembly of RNA and protein at the restrictive temperature. The effect is caused by a n amino acid substitution in the coat protein, which at 32°C results in its accumulation in the cell in an insoluble, denatured form. Von Sengbusch (1965) supposed that those mutants that induce severe yellow mosaic, by their altered surface charges, might influence the pH of the cell, causing increased breakdown of chlorophyll. In contrast, Jockusch and Jockusch (1968) assumed that the formation of insoluble aggregates of coat protein or virus particles, through adsorption to membrane proteins and destruction of cell organelles, would lead to yellowing and death of the cell. Their hypothesis was supported by the fact that increased temperature led to more pronounced yellow symptoms. The occurrence of such mechanisms has been disputed by Fraser (1969) on the grounds of his observation that degradation of chloroplast ribosomal RNA precedes that of chlorophyll in tobacco leaves infected with the flauurn strain of TMV, and the fact that some mutants produce insoluble, denatured coat protein at elevated temperature but do not induce yellow symptoms. Although the coat proteins of different strains of a virus usually differ in amino acid sequence and composition, it is improbable that such differences are primarily responsible for the variety of symptoms induced. Thus, the differing responses of various tomato cultivars to a wide range of TMV isolates were not correlated with the amino acid composition of their coat proteins (Dawson et al., 1979). Despite induc-

DISEASE INDUCTION BY PLANT VIRUSES

219

ing very different symptoms in tobacco, the common strain and the masked strain of TMV possess identical coat proteins. Likewise, many mutants have been characterized that have identical coat proteins but induce very different symptoms. However, the effectiveness of the N’ gene for resistance to TMV in tobacco varies with virus strains differing in coat protein properties (Fraser, 19831, suggesting that altered coat protein can modulate the reaction of the host. From a TMV mutant with insoluble coat protein inducing bright yellow primary chlorotic symptoms on sensitive tobacco, a spontaneous submutant was isolated that was symptomless and appeared truly free of coat protein (Sarkar and Smitamana, 1981). Yet, it still produced local lesions on Xanthi-nc tobacco, although the lesions were significantly smaller than those of the wild strain. Upon generation of pseudorecombinants of multipartite viruses, symptom type generally does not segregate together with the coat protein gene (Habili and Francki, 1974). Thus, the short particle of TRV specifies the coat protein, whereas the long particle can replicate by itself and this replication is accompanied by spread of the virus RNA through the infected plant and the production of symptoms (Lister, 1968). b. Transport Protein. The inability of the nitrous acid mutant Ni 2519 of TMV to spread and produce systemic symptoms at 32°C was suggested by Jockusch (1968) to reside in a protein necessary for the spreading of the virus infection in host tissue. This mutant does not have an amino acid substitution in the coat protein and is distinguished at 23°C by inducing small local lesions on Xanthi-nc tobacco. Dorokhov et al. (1984) found that a protein different from TMV coat protein formed a nucleoprotein complex with the RNA of the temperature-sensitive coat-protein mutants Ni 118 and fZauum of TMV at the restrictive temperature. They speculated that this protein was virus coded and involved in the process of transport of TMV RNA from cell t o cell. It is now known that TMV codes for four polypeptides of 183, 126, 30, and 17.5 kDa, respectively, the 183-kDa polypeptide being the readthrough product of the 126-kDa gene (Goelet et al., 1982). The 17.5-kDa polypeptide is the coat protein. In the analysis of the proteins synthesized in uiuo and in uitro from two closely related tomato w a i n s of TMV, one of which (Lsl; Nishiguchi et al., 19781, like Ni 2519, is defective in cell-to-cell movement at the restrictive temperature, a slight difference was revealed only in the 30-kDa polypeptide of the two strains (Leonard and Zaitlin, 19821, representing the replacement of a proline by a serine codon (Ohno et al., 1983). Ni 2519 was also found to be characterized by a point mutation, leading to the replacement of a glycine for an arginine in the 30-kDa polypeptide

220

L. C. VAN LOON

(Zimmern and Hunter, 1983). However, whereas Lsl replication is not temperature sensitive, Ni 2519 produces coat protein but not intact virus particles at the nonpermissive temperature (Bosch and Jockusch, 1972; Taliansky et al., 1982a). Ni 2519 forms local lesions on N. syluestris, in contrast to the parental strain, which initiates a systemic infection (Taliansky et al., 1982a). This characteristic is expressed at the permissive temperature, whereas the inability to spread systemically is expressed exclusively at the restrictive temperature. The position of the “local lesion gene” defined by Kado and Knight (19661, where mutations result in local lesion formation on N . syluestris, is not sufficiently well defined to be inconsistent with part of the 30-kDa coding sequence, although a location in the part coding for the 183-kDa protein cannot be ruled out. Recently, a fifth translational product of TMV, of 54 kDa, encompassing the carboxy terminus of the 183-kDa protein, has been found to be specified by a subgenomic RNA isolated from polyribosomes of infected tissues (Sulzinski et al., 1985). Although this part of the 183kDa protein shows amino acid homology to the RNA-dependent RNA polymerase of poliovirus (Kamer and Argos, 1984), it might not only function in virus replication but also play a role in symptom formation. Interestingly, the mutations in AMV giving rise to local lesions on bean have been linked to RNA 2, whose 90-kDa protein product is also implicated in viral replication (Nassuth et al., 1981; Berna et al., 1984). Since the character of necrotic lesion formation in viruses with a multipartite genome or in pseudorecombinants can oRen be attributed to components other than those implicated in systemic spreading, these two functions are more likely to depend on different protein products, although the processes of cell-to-cell spreading and local lesion induction may be linked biochemically. For TMV the evidence suggests that the 30-kDa protein is involved in cell-to-cell spread of infectivity within the infected plant and that, perhaps, when its function is impaired, the resulting restriction of virus movement may be phenotypically expressed by the induction of local lesions. The 30-kDa protein is basic (Meshi et al., 1982) and may bind to RNA, thereby facilitating the transport of viral RNA within the plant (Dorokhov et al., 1983, 1984). Thus far, TMV is the only virus for which a transport function has been linked to a specific translational product. However, a similar functional genomic organization may exist in other viruses specifying proteins of roughly similar molecular weights. BMV RNAs 1 and 2 are sufficient to sustain replication in protoplasts but the presence of RNA 3 is necessary for infectivity and spread in whole plants. B component RNA of CPMV is able to replicate by itself but incapable of spreading to surrounding cells, whereas M

DISEASE INDUCTION BY PLANT VIRUSES

22 1

RNA encodes one or more proteins essential for the transport of the virus (Rezelman et al., 1982). Although M RNA codes for the capsid proteins, coat protein seems to only stabilize virus during long-distance transport in the plant and a different function has to be implicated to explain cell-to-cellmovement (Atabekov and Dorokhov, 1984). Further indirect evidence that a transport function is carried by different viruses stems from complementation experiments, in which a helper virus provides a product enabling a defective virus to spread systemically. Thus, the temperature-sensitive TMV mutant Lsl was able to spread systemically at the restrictive temperature in plants preinfected with the temperature-resistant strain U1 or the coat-protein mutants Ni 118 or flauurn (Dorokhov et al., 19841, but was also aided by the unrelated potato virus X (PVX) (Taliansky et al., 1982b). Such virus-specific transport functions may play a role in controlling host range (Taliansky et al., 1982~):Tm-2 tomato lines, which are resistant to TMV at the whole plant but not at the protoplast level, became susceptible to TMV after preinfection with PVX. In contrast, Tm-1 tomato lines, whose protoplasts do not support TMV multiplication, remained resistant under these conditions. Likewise, cotton plants resistant t o TSV were systemically invaded by the virus upon infection with the unrelated cotton anthocyanosis virus (Costa, 1969). BMV was aided in systemic invasion of tomato and bean plants by the presence of TMV and dolichos enation mosaic virus, respectively. Similarly, TMV replicates and moves systemically in barley infected with barley stripe mosaic virus (BSMV) (Hamilton and Dodds, 1970; Dodds and Hamilton, 19721, although barley is considered to be highly resistant to TMV. TMV is restricted from spreading in cowpea or cotton but infected cells accumulate normal levels of virus, indicating that these plant species support TMV replication in those cells which receive virus during mechanical inoculation. However, the infective principle is unable to move from these original centers (Sulzinski and Zaitlin, 1983). This so-called subliminal infection constitutes an effective principle of resistance under field conditions, unless helper viruses provide a complementary transport function. Viruses which normally do not infect a certain plant species might thus become able to do so when such plants are already infected. Under those circumstances, it is likely that also the disease syndrome will be altered. c. Polymerase. The rather exceptional situation that plant species or cultivars are resistant to replication of some viruses at the protoplast level suggests that in those instances the viral polymerase cannot form a functional replicase complex. The resulting immunity is the only effect that may be linked directly to polymerase. With multipartite viruses, variations in symptoms have been shown to be associated

222

L. C. VAN LOON

with components specifying putative polymerase proteins, but at least in the case of AMV, more than one functional domain has been postulated to explain apparently unrelated effects on virus multiplication and symptom type (Roosien et al., 1983). However, the loss of virulence of many mutants, as found in the case of TMV (Jockusch, 1966a) and CPMV (De Jager, 1978), is often associated with reduced virus multiplication. More generally, it can be expected that alterations leading to a less efficient polymerase reduce the rate of virus multiplication and, consequently, may cause milder symptoms or a reduction in the extent of virus spread. Such effects would be quantitative rather than qualitative, however, and should not be expected to give rise to changes in symptom type. 2 . Absence of Specific Symptom-Inducing Proteins Taking TMV as an example of the group of small ssRNA viruses, the occurrence of no more than four different translational products precludes the possibility that the various symptom types induced on different plants by strains and mutants of the same virus are specified by different specific symptom-inducing proteins. As the 183 (54)-kDa protein is the presumed polymerase protein, the 30-kDa protein is involved in cell-to-cell transport, and the 17.5-kDa protein is the capsid, only the 126-kDa protein has not been assigned a definite function. In uitro, additional overlapping protein products can be formed that are coterminal with the main polypeptides and may arise through internal initiation or premature termination. Since eukaryotic mRNAs are generally monocistronic, and specific subgenomic RNAs have been identified in uiuo that code for the 54-kDa, 30-kDa, and coat proteins but no additional protein products have been discerned in infected leaves or protoplasts, it is unlikely that the products found only in uitro can play any role in uiuo in maintaining a stable specific phenotype. Unlike TMV, other viruses may use a more complicated strategy to produce essentially similar types of polypeptides. Members of the comovirus group specify proteases needed for cleavage of the primary polyprotein products into their constituent polypeptides. A small protein, Vpg, covalently linked to the 5’ end of various viral RNAs may serve a role in replication. In tobacco ringspot virus (TRSV) only, the presence of Vpg is required for viral infectivity (Mayo et al., 1982). Potyviruses and the DNA-containing cauliflower mosaic virus (CaMV) code for proteins present in intracellular inclusion bodies (Dougherty and Hiebert, 1980b; Xiong et al., 1982). The latter virus has the capacity to code for eight different proteins (Xiong et al., 1984). Its genomic

DISEASE INDUCTION BY PLANT VIRUSES

223

organization to some extent resembles that of animal retroviruses (Howell, 1985). The insertion of short DNA sequences that induced frameshift mutations either had no effect, abolished infectivity, or retarded the development of symptoms (Dixon et al., 1983). In-phase insertions likewise had no effect or destroyed infectivity. However, when in the amino-distal portion of the region coding for the inclusion body protein, they reduced symptom severity (Daubert et al., 1983). This is the first evidence that a protein product necessary for virus maturation has a function in inducing disease. Satellite viruses such as CARNA 5 and the related peanut stunt virus-associated RNA 5 (PARNA 5 ) each have two open reading frames that may specify small polypeptides. So far, these products have not been detected in infected plants. However, infection of tomato with either CMV in the presence of CARNA 5 or one of several viroids leads to the accumulation of a host-specific protein whose distribution within the plant correlates with symptom severity. The disease-associated protein has not been detected in tomato plants infected with TMV or CMV in the absence of CARNA 5 but appears restricted to viroid and satellite RNA infection (Galindo et al., 1984). Satellite RNAs such as PARNA 5, LTSV RNA 2, and the satellite of TRSV show structural similarity to viroids. PARNA 5 has several regions of 90% sequence homology with various plant viroids, including sequences of the conserved central region of most viroids (Collmer et al., 1985).Viroids do not have a protein coat and do not specify any proteins; yet, they may cause severe diseases. Their disease-inducing ability has been linked to a small structural domain in their sequence of no more than 240-380 nucleotides (Stinger, 1982;Schnolzer et al., 1985). Of 17 naturally occurring variants of citrus exocortis viroid (CEV), isolates inducing mild and severe symptoms on tomato formed two separate classes of sequence variants that differed by a minimum of 26 nucleotides in two regions of the native structure. Either one or both of these regions appear responsible for the variation in pathogenicity (Visvader and Symons, 1985).Different strains of potato spindle tuber viroid (PSTV) that induce symptoms on potato varying from very mild to very severe likewise show only a small number of nucleotide substitutions, presumably no more than 2-10 (Dickson et al., 1979).In those strains that have been fully sequenced, four base changes each differentiate the mild and the severe strain from the intermediate type strain (Stinger, 1982).Diseases may thus be induced by naked RNAs that do not code for proteins. This raises the possibility that viruses might likewise induce disease at the RNA level, for instance by folding into a specific secondary structure. Some viruses

224

L.C.VAN LOON

possess a tRNA-like structure at their 3’ ends. It is unlikely that this structure plays a role in translation (Hall, 1979) and a function has still to be assigned.

111. THE GENETICS OF HOST-VIRUSINTERACTIONS A. The Concept of the Gene-for-Gene Relationship On the one hand, determinants for symptom type have been assigned to most or all components of multipartite viruses. On the other hand, viruses code for only a few proteins. Proteins coded by different RNA components have been linked to essentially similar types of symptoms in different host plants (e.g., systemic mosaic induced by CPMV in bean or cowpea is determined by the M and B component of the virus, respectively; Fig. 1).Based on these observations one can suppose that symptoms arise as a result of the interactions of more than a single determinant, the relative importance of each varying in different hosts. Alternatively, in different hosts symptoms may be a consequence of different initial interactions between viral products and plant constituents. As long as the mechanism of symptom induction is unknown, the first possibility is difficult to explore experimentally. In contrast, the second possibility could, in principle, be assessed by searching for complexes of viral proteins with cellular structures. On the basis of genetic analyses similar kinds of interactions have been postulated to occur in the cultivar-specific recognition between host plants and pathogenic fungi and bacteria. Since in virus-host plant interactions the resulting symptom type likewise depends on the genetic constitution of both partners, genetic analysis can provide clues to the nature and the complexity of the host-virus relationship. Such an analysis requires the availability of genetically well-characterized differential host cultivars which react differently to at least one strain of the virus, and virus strains which induce different symptoms on at least one host cultivar. The information gained is increased exponentially when more virus strains and/or host cultivars can be involved in the analysis. In the gene-for-gene relationship originally developed to explain the interactions between plant cultivars and physiological races of a pathogenic fungus (Flor, 1971;Ellingboe, 19841,incompatibility between plant and pathogen is the result of specific recognition. Such recognition is supposed .to give rise to the activation of defense mechanisms, thereby limiting the progress of the pathogen and conferring resistance on the host. Compatibility is seen as due to the failure to

DISEASE INDUCTION BY PLANT VIRUSES

225

activate the defense mechanisms. This hypothesis has been remarkably successful in explaining most, though not all, cultivar-specific resistance in fungal infections (Sidhu, 1975).In the simplest terms the relationship consists of a dominant gene for avirulence, A l , in the pathogen, which is complemented by a specific dominant gene for resistance, R l , in the plant. The product of the gene for avirulence of the pathogen (“elicitor”) would interact with the product of the gene for resistance of the plant (“receptor”),leading to a functional complex. In this combination the plant and the pathogen are incompatible. Absence of the avirulence gene product, a l , and/or of the resistance gene product, r l , leads to compatibility and disease (Table Ia). Disease also results when a mutation in the pathogen produces a new avirulence gene, A2 (now a “virulence” gene), that is not matched by the available resistance genes carried by the host cultivar. However, this new avirulence gene may be complemented by another resistance gene, R2, present in a different cultivar (Table Ib). Pathogens may possess multiple avirulence genes matching an equal number of resistance genes in the host, and these can be either allelic or nonallelic. This scheme can explain the frequent occurrence of constitutive mutations from avirulence to virulence in fungi and the often-observed phenomenon that fungal temperature-sensitive mutants become virulent at the restrictive temperature. TABLE I SIMPLE SCHEME OF A GENE-FOR-GENE RELATIONSHIP ENCOMPASSING ONE AND Two LOW Pathogen avirulence gene

Host cultivar resistance gene (a) One locus R1 -

A1

+

a1

(b) Two loci RIr2

C A2 AI

(

mutation to new gene)

-

+

rl

+ + rl R2

+

-

a Symbols: t , compatible interaction; -, incompatible interaction.

226

L.C.VAN LOON

Resistance to viruses is often monogenically controlled, and thus a n equally simple initial interaction between a virus and its host could be envisaged. Incompatibility might be equated with a hypersensitive reaction, compatibility with systemic disease. However, for a virus to be effectively localized, necrosis is not required. Confinement of an infecting virus to chlorotic local lesions or symptomless starch lesions proves that virus localization can be separated from necrosis. This is also evidenced by the occurrence of diseases in which systemic spreading of the virus is accompanied by necrosis. Thus, incompatibility is expressed whenever the virus is localized and compatibility when it spreads systemically, independent of the phenotypic form in which symptoms are expressed.

B . Applications to Selected Host-Virus Combinations 1 . Tobacco-TMV Genetically well-studied host-virus combinations comprise tobaccoTMV, tomato-TMV, and bean-bean common mosaic virus (BCMV).In these combinations several host cultivars with different genes for resistance and many virus isolates or mutants inducing different symptoms on these cultivars have been analyzed. In tobacco all natural varieties are sensitive to TMV (Holmes, 1960) and develop systemic mosaic symptoms after infection with the common strain (uulgare, U1, OM). However, tobacco species, varieties, or cultivars that possess the dominant gene N‘ localize all strains of TMV except the common one (Valleau, 1952), whereas the partially dominant gene ns confers resistance toward ribgrass mosaic virus (RMV, Holmes’ ribgrass strain of TMV, US) (Weber, 1951). Some mutants that have lost the ability to spread systemically are localized in all genetic backgrounds, even in those that apparently do not possess genes for resistance to TMV. The common strain is localized in Nicotianu glutinosa. The resulting hypersensitive reaction is controlled by a single dominant gene N . By chromosome substitution the Hg chromosome carrying the N gene has been incorporated into the N . tabacum cultivars Samsun, White Burley, and Xanthi, replacing the H chromosome and giving rise to the TMV-resistant cultivars Samsun NN (Holmes, 1938), White Burley NN (Valleau, 1952), and Xanthi-nc (Takahashi, 1956), respectively. Since the genes N ’ , ns, and N are taken to be allelic (Weber, 1951; Valleau, 19521, the four phenotypes can be differentiated as shown in Table 11. In this table the virus strains listed are progressively less able to spread in hosts carrying less effective alleles for resistance. An

DISEASE INDUCTION BY PLANT VIRUSES

227

TABLE I1

INTERACTIONSBETWEEN TOBACCO PLANTS CARRYING DIFFERENT ALLELES FOR RESISTANCEAND SELECTED STRAINS AND MUTANTS OF TMVa Host allele TMV strain

N

N’

Common, U1, uulgare MDGTMVb, U2, flauurn RMV, U8 Ni 2338, 1952D

-

+ -

-

ns

t

+

-

n’ (n)

+ + *= -

a Symbols: + , compatible interaction, systemic disease; -, incompatible interaction, localized reaction.

b c

Mild dark-green tobacco mosaic virus (Wetter, 1984). Semisystemic ringspotting.

essentially similar relationship exists between Capsicum and TMV (Tobias et al., 1982). This scheme is at variance with the model of the gene-for-gene relationship outlined above, in that incompatibility occurs in the absence of an allele for resistance in the host [i.e., Ni 2338 or 1952D on n’ (n) tobaccos]. Moreover, mutations in the virus lead from compatibility to incompatibility in any given genetic background. The mutation in TMV uulgare leading to the transition from systemic mosaic symptoms to local lesions on Java tobacco (N’) and the generation of the mutants Ni 2338 and 1952D causing local lesions irrespective of the host allele present imply that the virus has lost the ability to overcome the effects of the resistance genes. The fact that until now no strain of TMV has overcome the allele N indicates that the virus cannot mutate easily from incompatibility to compatibility, a conclusion supported by the rare occurrence of systemic variants of strains localized in other genetic backgrounds. Thus, virulence of the virus is mutated to avirulence, showing that virulence is a positive function (“dominant”). When virulence is lost or, due to the presence of a resistance allele, cannot be expressed, the virus invariably remains localized within necrotic lesions. Hence, the resistance reaction of the host is triggered irrespective of the presence of a functional product of an (a)virulence gene. It further appears that the positive function conferring virulence on the virus suppresses the resistance reaction of the host. If this is so, the existence of different strains or groups of strains of a virus that are able to systemically infect only host cultivars with a specific genetic constitution may be explained by assuming that only in those com-

228

L. C. VAN LOON

binations does a specific and functional interaction between a product of the virus and a factor in the host ensue. TMV possesses such a function, which enables it to spread systemically, and which has been attributed to the 30-kDa protein. This function has been implicated in host plant specificity and might perhaps also enable the virus to circumvent localization. The resistance alleles in tobacco do not appear to completely nullify the effect of the spreading factor of the virus, however, as lesions usually progress to a variable extent from the infection center. Rather, they seem to be just effective enough to prevent the virus from escaping. The extent of spreading is as much a function of the virus strain as it is of the genetic background of the host. Thus, RMV in n' (n) tobaccos causes yellow ringspot-type lesions to which the virus remains confined. Upon inoculation of two lower leaves, the virus is able to invade the two to four leaves directly above, but young developing leaves remain symptomless and no virus can be recovered from them. In N hosts, however, RMV lesions remain extremely small. Such intrinsic differences in the rate of spread of different virus strains constitute quantitative variations modulating the qualitative outcome of the host-virus interaction. Above 28°C the hypersensitive reaction due to the N gene is no longer expressed. However, if infected tissues are held at 20°C for 24 hours and then the temperature is raised to 30°C, necrotic lesions subsequently develop (De Laat and Van Loon, 1983). This observation indicates that not necrotization itself, but its initiation is inhibited at elevated temperature. At this temperature RMV infects n' (n)tobaccos systemically; young, developing leaves show yellow mottling symptoms and contain large amounts of virus. The mutant 1952D induces large, round, yellow flecks without necrosis (Van Loon, 1972); Ni 2338 likewise causes spreading symptoms (Jockusch, 1966a). These observations show, on the one hand, that temperature affects the expression of host alleles similarly and, on the other hand, that none of the virus strains or mutants is defective in spreading. The apparent overall gain in virulence is hard to explain by the acquisition of additional resistance-suppressing activity of all virus strains at 30°C. Rather, it appears that a host function is impaired that is responsible for triggering the resistance reaction in all genetic backgrounds. For the virus to be able to suppress the resistance reaction of the host, the resistance reaction has to be triggered first. This triggering is likely to be caused by events occurring early in infection (De Laat and Van Loon, 1983). Specificity does not appear to be required at this stage, except for a basic complementarity enabling further reactions to be initiated. It must be concluded that the interaction between tobacco

DISEASE INDUCTION BY PLANT VIRUSES

229

and TMV involves two stages: (1)a basic recognition event common to all tobaccos, by which the presence of the virus triggers mechanisms leading to virus localization, and (2) a superimposed recognition scheme dependent on specific “resistance” alleles of the host, which can be suppressed by complementary viral products. In the latter case the viral 30-kDa protein might be involved, although i t is difficult t o see how. What viral factor could be involved in basic recognition is not clear, although one might speculate that this step involves the formation of a functional virus-replicase complex. Since this step should be the same in all virus strains, it is likely that different viral functions are needed in the two stages of the recognition, and thus at least two viral products, probably proteins, may be implicated in the initiation of systemic disease. Independently from the N locus, the A locus determines sensitivity of tobacco to TMV. The dominant allele A is found in many wild species and in all varieties of N. tubacum except Ambalema, which is almost tolerant (Nolla, 1935). Tolerance is due to the presence of two recessive genes designated rml and r , (Nolla, 1938). Since N . tubucum is an amphidiploid, Valleau (1952) considered these factors identical and described them as a. It is not clear whether the absence of the dominant A allele obliterates only visible symptoms. Apparently, virus multiplication is also reduced. This could explain why the hypersensitive reaction of tobacco plants possessing the factor N but lacking the A allele appears more effective; local lesions appear later and remain smaller. Such plants are immune to some strains of the virus (Holmes, 1960). Determination of the rate of virus multiplication in Ambalema protoplasts could clarify what factors are involved. Apart from the functions described above, the virus must possess determinants for the phenotypic expression of mosaic symptoms, various types of which have been distinguished. These may be specified by yet another viral function. In view of the small coding capacity of virus genomes, it is likely that individual gene products have more than one function. Thus, in spite of the error-prone replication of RNA genomes, mutant gene products fit to well perform these different functions will be scarce and new virus strains may not readily arise. 2 . Tomato-TMV A more elaborate but largely similar scheme applies to the interaction between tomato and TMV. The dominant alleles Tm-2 and Tm-22 confer resistance toward all isolates except those designated as tomato strains 2 and 22, respectively (Table 111).However, another nonallelic host gene, Tm-1, suppresses visible mosaic symptoms and strongly reduces virus multiplication from the time of inoculation. This sug-

L.C. VAN LOON

230

TABLE I11 GENE-FOR-GENE RELATIONSHIP BETWEEN TOMATO AND

TMVaJ

Host genotype

TMV strain

a b

t

Tm-1

Tm-2

Tm-22

Tm-112

Tm-1/22

Derived from Fraser (1986). Symbols: t , systemic disease; -, localized reaction.

gests that the product of this resistance gene acts at an early step of virus replication (Fraser and Loughlin, 1980). This notion is supported by the rather exceptional character of Tm-1 resistance in that it is also expressed when isolated protoplasts are inoculated (Motoyoshi and Oshima, 1977). Tm-1 might thus regulate basic recognition whereas the action of Tm-2 may be sought at the superimposed recognition level. Tomato strain 1 overcomes the action of the Tm-1 gene, indicating that recognition can take place. Tm-1 appears to specify a temperature-sensitive function. If plants are grown at a constant 33"C, TMV strain 0 multiplies as well in Tm-1 hosts as in susceptible (+) hosts. However, visible symptoms are still suppressed (Fraser and Loughlin, 1982). This suggests that the Tm-1 gene affects at least two different viral functions.

3. Bean-BCMV BCMV can be differentiated into 10 pathotype groups on the basis of the reactions produced on nine host groups of bean (Drijfhout et al., 1978) (Table IV). Host groups 1-5 are presumed to contain recessive alleles i of the so-called inhibitor gene and show mosaic but never systemic necrosis, whereas cultivars in groups 6-9 carry the dominant allele I and show systemic necrosis but never mosaic. Apart from this distinguishing host characteristic the viral pathotypes are differentiated on the basis of their ability to systemically invade the nine host groups. It would be extremely improbable that the BCMV genome carries enough information to circumvent nine different genetic blocks conferring resistance to each of the nine differential host types. However, the interactions between host differentials and virus strains can be explained by an apparent gene-for-gene relationship encom-

23I

DISEASE INDUCTION BY PLANT VIRUSES

passing six recessive host genes for resistance and four positive pathogenicity functions of the virus determining systemic disease. The allelic host genes bc-l and bc-I2 confer resistance to virus strains possessing nonallelic pathogenicity genes P1 and P12; the allelic genes bc-2 and b ~ - confer 2 ~ resistance to strains possessing the possibly allelic genes P2 and P22. A fifth resistance gene, b c 3 , present in some selected breeding lines, is not complemented by a pathogenicity gene. So far, lines carrying this gene are resistant to all BCMV strains. This interaction scheme is the most elaborate described so far for any virus-host plant relationship. BCMV and bean appear to interact in a genuine gene-for-gene relationship (Table IV). However, in a mirror image of the classical scheme for fungus-host plant interactions, virulence is a positive function and resistance is recessive. Moreover, the presence of any resistance gene in the host confers resistance to all strains of the virus unless these strains possess the corresponding pathogenicity gene(s). This means that also in the combination bean-BCMV resistance is triggered but suppressed by complementary viral functions. In the absence of a resistance gene, however, compatibility appears to be basic, since no pathogenicity gene seems necessary to explain the ocTABLE IV GENE-FOR-GENE RELATIONSHIP BETWEEN

BEANAND BCMVarb

Host group

Strain group (pathotype)

I

-

I1 P1 I11 P2 IVa P1.12 Ivb P I .12 Va P1.2 vb P 1 2 VIaP1.122 VIbP1.122 VII P1.12.22

1 -

2

bc-1

3 bc-12

4 bc-2

5 bc-12.22

6'

bc-1

bc-u, i

+ + + + + + + + + +

+ + + + + + + +

-

-

-

+

-

+ + +

8 -

9 bc-12

bC-U+, Z

-

+ +

7c bc-1

-

-

-

-

-

-

-

+ + + + -

-

+

+ -

+ + +

-

-

-

+ + + +

+ + + +

-

-

-

-

-

+ -

Derived from Drijf'hout (1978); Drijf'hout et al. (1978). Symbols: +, systemic disease; -, localized reaction. c Pathotypes Ivb and Vb cause necrotic tip kill of most or all plants of host group 6 (sensitive) but systemic necrosis on only some plants of host group 7 (variably sensitive). a

b

232

L. C.VAN LOON

currence of systemic infection in host group 1 infected by pathotype I. Indeed, as in protoplasts, the virus may not trigger resistance under those conditions. Alternatively, it might do so but then it must possess a pathogenicity gene (PO)additional to the ones listed and needed to suppress the triggering of resistance reactions after basic recognition takes place. In any case, dual recognition is required to explain the relationship between BCMV and bean: a basic recognition triggering resistance, on which is superimposed a gene-for-gene type recognition to suppress resistance and induce systemic disease. According to this scheme, the existence of at least three and perhaps four different viral functions, one of which presumably occurs in two allelic forms, must be assumed. Furthermore, additional interactions exist, as genes bc-u and I (host groups 6-9) modify the reaction type. Hence, BCMV must also carry several genetic determinants involved in pathogenesis.

C . How Do Pathogenic Determinants and Resistance Genes Act? Pairs of genes exist in the virus and in the host that determine recognition and enable the virus to overcome the resistance and induce systemic disease. Whenever a resistance gene in the host is not matched by a corresponding pathogenicity gene of the virus, the phenotypic result is usually similar, i.e., virus localization, and the many metabolic alterations occurring in the host will also be similar and may even be identical. This can be understood because basic recognition, triggering resistance reactions, will be similar in all combinations. Matching host and viral genes will prevent the resistance from being triggered and lead to a mechanism essentially switching off the resistance response. Whatever the viral genes involved, the host genes may be assigned a regulatory character and a priori it makes no difference whether they are inherited in a dominant or in a recessive fashion. The examples given above illustrate that recessive resistance genes are not exceptional. In fact, in many plant species both dominant and recessive genes have been identified that, most often singly and sometimes in combination, restrict virus movement and confer resistance against quite a number of economically important viruses (Fraser, 1986). In spite of extensive investigations, so far products of resistance genes have not been identified. If resistance genes have regulatory functions, their products might be present in only very small amounts, or they might not even be transcribed at all. In spite of their small genome size, viruses must harbor several pathogenic determinants. This conclusion corroborates the assignment of determinants to more than a single and sometimes all components of multipartite viruses. However, there does not appear to be room for

DISEASE INDUCTION BY PLANT VIRUSES

233

symptom-inducing proteins, nor has any specific symptom-inducing character been defined as due to the proteins that are known to be coded for. In susceptible host plants, viroids may cause a variety of symptoms largely encompassing those induced by viruses, although they do not code for proteins. The bulk of viroid infectivity is found within the nucleus and closely associated with the chromatin (Diener, 1971). This nuclear location and the inability to act as mRNA suggest that viroid pathogenesis is caused by a direct interference of viroid RNA with the regulation of gene expression. Since viroids contain sequences identical to those of the small nuclear U1 RNA functioning in mRNA splicing, it has been proposed that viroids induce disease by interfering with cellular mRNA processing. Notably the extent of complementarity of different viroid isolates to splice junctions of mRNA precursors correlates with the severity of disease caused in chrysanthemum plants (Dickson, 1981). However, none of the U1 RNA sequences is present in the plus or minus strand of avocado sunblotch viroid, so that at least this viroid must be assumed to exert its pathogenic function in a different way. The location of mature viroids in the nucleolus (Schumacher et ul., 1983) and the presence of U1 RNA in ribonucleoprotein particles further disfavor this model of pathogenesis (Riesner and Gross, 1985). Furthermore, one might expect that interference with such a basic mechanism as mRNA splicing would lead to gross aberrations and possibly seedling death, but in many wild host plant species, viroid infection and replication do not produce any recognizable disease symptoms. Rather, symptoms of viroid infection seem to result from hormonal malfunction, resulting in abnormal cell development (Sanger, 1982). Stunting and leaf rugosity of cucumber seedlings infected by hop stunt viroid (HSV)were associated with significantly lower levels of endogenous indoleacetic acid (IAA). Female flower formation, which is causally associated with high endogenous auxin levels, was delayed (Yaguchi and Takahashi, 1985). CEV induces stunting, epinasty, and leaf rugosity in several hosts. Such plants contain decreased amounts of gibberellins (Rodriguezet ul., 1978), exhibit impaired root formation apparently as a result of insensitivity to auxin (Duran-Vila and Semancik, 19821, and show increased ethylene production (Conejero and Granell, 1986). Interestingly, this syndrome can be reproduced in Gynuru auruntzacu in the absence of viroid by foliar sprayings with silver nitrate. Since silver ions strongly increased ethylene production while at the same time supposedly interfering with ethylene action, perturbations in growth regulator metabolism appear to be central in viroid pathogenesis. Growth regulators influence transcription and translation, thereby controlling plant growth and development. If

234

L.C . VAN LOON

genes conferring resistance to virus infection are regulatory control elements in host DNA, their function might be perturbed through alterations of plant hormone metabolism or, more directly, interference with their functioning might be reflected by changes in growth regulator content, distribution, and metabolism.

IV. ALTERATIONS IN HOSTPLANT METABOLISM A. Common Features in Symptom Type Variations A complementary approach to the question how viruses cause disease can be taken by studying the cytological, physiological, and biochemical alterations in host plant metabolism accompanying symptom development. Since viruses possess only a small number of pathogenic determinants, the many metabolic alterations occurring once symptoms have become apparent must arise in a coordinated manner, perhaps as the result of a single initial event. In near-isogenic tobacco cultivars, reacting to TMV with systemic mosaic symptoms and a hypersensitive reaction, respectively, vastly different metabolic alterations occur. Yet, these differences have to be attributed to a single distinctive host gene (e.g., Section III,B,l). Starting from determinations of the many differences between infected and healthy plants, patterns may be sought that appear connected with the type and severity of the developing symptoms. Working backward in time, factors governing these patterns may become apparent and point to even earlier initiating principles. However, even upon inoculation with a very concentrated virus suspension, at most 0.1% of the leaf cells become infected (Matthews, 1981). This makes it more and more difficult to discern physiological and biochemical alterations at earlier times after infection. Nevertheless, some patterns have emerged, particularly because many of the metabolic alterations connected with symptom type and severity can be essentially similar to those occurring after infection with fungi or bacteria (Van Loon and Callow, 1983). Such observations indicate that common factors are involved in pathogenesis and symptom expression, irrespective of the type of infecting pathogen. 1 . Chlorosis and Necrosis Wounding of plant parts or exposure to toxic chemicals may kill plant cells rapidly, leaving whitish-gray desiccated flecks. Viral necrosis usually takes the form of reddish-brown to black lesions or streaking, indicating that killing of the tissue proceeded sufficiently slowly for

DISEASE INDUCTION BY PLANT VIRUSES

235

oxidative enzymes to be activated or induced. The combined action of polyphenoloxidases and peroxidases leads to the polymerization of phenols, yielding pigments responsible for the color of the developing necrotic spots. Such necrotic spots and many of the accompanying physiological and biochemical alterations can be reproduced in the absence of virus by treating plants locally with high concentrations of ethephon, a growth regulator from which ethylene is released in plant tissues (Van Loon, 1977).Since ethylene is produced in copious amounts near the time of the appearance of virus-induced necrosis, this plant hormone appears t o act as a natural regulator (Van Loon, 1983).However, the natural precursor of ethylene, l-aminocyclopropane-l-carboxylic acid, is less active than ethephon, and inhibitors of ethylene synthesis, such as aminoethoxyvinylglycine,are only partially effective in inhibiting the metabolic alterations associated with the development of necrosis. Since upon ethylene release from ethephon, hydrochloric and phosphoric acid are produced in stochiometric amounts, resulting damage to plant cell membranes appears to be a second factor involved in the induction of necrosis. A necrosis-inducing factor secreted by TMVinfected tobacco cells (Hooley and McCarthy, 1980)might fulfill such a role. Necrosis can be induced artificially in systemically infected plants by treatments that damage cell membranes (Ohashi and Shimonura, 1982).The resulting decompartmentalization stimulates aromatic biosynthesis as well as the induction of further defense reactions (Van Loon, 1983). Localization of bean yellow mosaic virus in chlorotic lesions on Tetragonia expunsa is also accompanied by a strong increase in ethylene production (Gaborjanyi et al., 1971). When such plants were exposed to 1% C0,-a competitive inhibitor of ethylene action-they did not develop chlorotic lesions and infection remained symptomless. However, further gassing with ethylene or spraying with ethephon led to the appearance of necrotic specks within the chlorotic lesions (Bailiss et ul., 1977). These observations suggest that the occurrence of chlorotic or necrotic lesions represents a quantitative rather than a qualitative difference. High concentrations of ethylene increase membrane permeability (Suttle and Kende, 1980)and may thus substitute for the necrosis-inducing factor which acts by damaging cell membranes. The occurrence of necrotic ringspotting may be interpreted as due to a time dilation between initiation of ethylene production and occurrence of membrane damage. On the other hand, systemic necrosis may simply be the reflection of the path along which the virus progressed in the tissue, initiating the necrotic reaction in its wake. These considerations illustrate that various forms of chlorotic and

236

L.C.VAN LOON

necrotic symptoms may be viewed as resulting from the spatial and temporal variations in just two factors: increased ethylene production and induction of a necrosis-inducing factor. So far, it is not clear how these factors originate and interact. However, during a hypersensitive reaction both increased ethylene production and the occurrence of the necrosis-inducing factor are very early events. They may precede visible necrosis by as much as 24 hours and develop concomitantly with the initiation of the resistance response (De Laat and Van Loon, 1983; Van Loon, 1983).

2. Mosaics Chlorosis, yellowing, and mosaic result from a local or fairly general loss of chlorophyll. Alterations in yellowing leaf parts often appear cytologically, physiologically, and biochemically similar to those occurring in noninfected plants after leaves have fully expanded and enter the phase of senescence, or when leaves are detached and therefore become subject to accelerated aging. In typical mosaics light green or yellow infected parts alternate with dark-green, virtually virusfree, leaf areas. The light parts exhibit many characteristics of senescing leaves; the dark areas, in contrast, are physiologically young. Mosaic symptoms usually start after systemic infection of developing leaves, but chlorosis and yellowing may occur concomitantly in fully expanded leaves, for instance after infection of tobacco with CMV (Ehara and Misawa, 1975)or lettuce with beet western yellows virus (BWYV) (Tomlinson and Webb, 1978). In Chinese cabbage infected with TYMV chloroplasts become rounded and clumped together in the cell, and may eventually develop large vesicles or fragment. At such stages, photosynthetic activity may be considerably lowered (Matthews 1980,1981). On an area basis, in TMV-infected tobacco photosynthesis is not reduced. Chlorophyll content in the light green areas is decreased, but this loss is compensated for by a higher chlorophyll content in the dark green areas. Moreover, the light reactions of photosynthesis themselves are often stimulated in light green areas due to a lower amount of antenna chlorophyll per reaction center. Virus-infected variegated leaves from Tolmiea menziesii plants, while having severely lowered chlorophyll levels, had photosynthetic rates only slightly less than those of control leaves (Platt et al., 1979). However, in many other mosaic or yellows diseases, photosynthesis can be severely reduced, even when calculated on a chlorophyll basis. Yellowing such as induced by beet yellows virus in sugar beet can decrease photosynthesis up to about 50% (Spikes and Stout, 1955;Hall and Loomis, 1972). In peanut infected with peanut green mosaic virus, leaves transitorily

DISEASE INDUCTION BY PLANT VIRUSES

237

develop a light green mosaic. Although chlorophyll loss is relatively small, photosynthesis is severely reduced due to an inhibition of electron transport a t the reducing side of photosystem I1 (Naidu et al., 1984a). Although it has been suggested that this inhibition results from a decrease in the content of plastochinon (Naidu et al., 1984b), i t is equally possible that the synthesis of the rapidly turning over protein that binds the electron acceptor QB is reduced (cf. White and Brakke, 1983).This protein readily disappears under stress conditions. Moreover, it is synthesized on chloroplast ribosomes, which, like in senescence, are rapidly lost in mosaic-diseased leaves (Fraser, 1969). In mosaic-diseased tobacco leaves the light parts appear to contain less cytokinins than the dark areas (Sziraki and Balazs, 1975). Low cytokinin content as well as increases in abscisic acid (ABA) and ethylene are common features of natural leaf senescence. Ethylene may be increased in leaves displaying chlorosis or yellowing but not in tissue exhibiting ordinary mosaic. In contrast, in the latter ABA is increased (Whenham and Fraser, 1981). Whereas only about 1%of the ABA was found outside the chloroplasts in control plants, up to 78% appeared to be located in the cytoplasm in severely mosaic-diseased leaves (Whenham et al., 1985). These observations are suggestive of a direct effect of TMV on chloroplast properties. However, similar effects can occur in leaves as a result of osmotic stress. Thus, the specificity of this response remains questionable. Nevertheless, the increase in cytoplasmic ABA is sufficient to be held responsible for the accompanying reduction in shoot growth rate. Reduction in leaf area was correlated with a reduction in the number of cells per leaf. ABA did not affect cell elongation (Whenham and Fraser, 1981). This hormone similarly mimicked the inhibition of leaf initiation in TYMVinfected Chinese cabbage (Fraser and Matthews, 1983). The regulation of natural leaf senescence is a nuclear-controlled cytoplasmic process (Thomas and Stoddart, 1980), and thus there is no need to invoke a chloroplastic localization or a specific effect on chloroplasts for the virus to induce mosaic symptoms. However, Reinero and Beachy (1986) recently reported the presence of TMV coat protein in both stroma and thylakoid membranes of systemically infected tobacco leaves. Chloroplasts isolated from leaves infected with the masked strain contained approximately 10-50 times less coat protein than did chloroplasts isolated from leaves bearing common mosaic symptoms, although the levels in total cell extracts were similar. The coat proteins of the two strains are identical and the different concentrations reached inside chloroplasts must, therefore, reflect differences in the virus-host relationship due to other viral functions. ABA may accelerate leaf senescence, particularly when cytokinin

238

L.C. VAN LOON

levels are low, and so the increase in cytoplasmic ABA may well be (one of) the factorb) responsible for the apparent senescence in the light-colored leaf parts. Somehow, the dark-colored leaf parts escape from the virus and its effects. Only very few of such “green islands” have a clonal origin (Carlson and Murakishi, 1978). Often, virus may at some stage penetrate into the dark green areas and initiate additional yellowing. Apparently, a dynamic equilibrium exists in the tissue that depends on the exchange of hormonal signals. As for ABA, its increase seems to be due to de novo synthesis, but the triggering mechanism has not been elucidated. 3. Morphogenetic Disturbances The appearance of mosaic symptoms in TMV-infected tobacco is developmentally controlled; i.e., only newly emerging plant parts are sensitive to the pathogenic action of the virus. Leaves longer than 1.5 cm at the time of infection do not develop mosaic symptoms (NilssonTillgren et al., 1969). Malformations of mosaic-diseased leaves arise because growth in dark green parts is not, or only slightly, inhibited as compared to that in light green parts. Stem length is reduced and developing flowers may show color breaking. Growth reduction, color deviations, wilting, hypo- and hyperplasia, malformations of specific organs or whole plants, as well as such abnormalities as tumor formation, premature leaf drop, pollen sterility or abortion, seed abortion, abnormal secretion, and graR incompatibility are all symptoms indicative of aberrant hormone metabolism. For instance, stunting of cucumber seedlings infected with CMV has been associated with a reduction in the content of endogenous gibberellins (GAS) (Bailiss, 1974; Aharoni et al., 1977) and altered metabolism of applied GA, (Ben-Tal and Marco, 19801, as well as increases in the endogenous levels of ABA (Aharoni et al., 1977) and ethylene (Marco et al., 1976). This example illustrates well the complex effects of infection on the hormone balance of the tissue. Symptoms of virus infection thus appear to reflect disturbances of hormonally controlled growth and development. As symptom expression is genetically determined, it must be subject to regulatory control mechanisms functioning similarly as in healthy plants. However, owing to the presence of the pathogen, genetic information is expressed in an untimely or uncoordinated way, resulting in abnormal patterns of growth and development. The nature and extent of these aberrations are modulated by host genes, virus strain, and environment. As indicated earlier, the developmental stage may greatly affect plant reactions. Environmental variables such as light, temperature, moisture, and nutrition have

DISEASE INDUCTION BY PLANT VIRUSES

239

pronounced effects on the growth and development of plants and likewise influence virus symptoms. Plant hormones function as coordinating messengers in the adjustment of growth and development to changes in the internal and external environment. If viruses somehow affect plant hormone metabolism, interactions with other modulating factors are likely to influence hormone levels in various ways, leading to variation in the symptoms expressed.

B . Attempts to Influence Symptom Type by Hormone Applications Apart from the examples cited above, comparatively little conclusive data are available on hormone changes in virus-infected plants and their relationship to pathogenesis and symptom expression (Fraser and Whenham, 1982).Effects of exogenous hormones on virus multiplication and symptoms of disease are highly varied, depending on plant developmental stage, hormone concentration, and time of application with respect to infection. Attempts to change symptom type by growth regulator applications have not been successful so far. However, some regulators quantitatively influence symptom severity and occasionally may even mask symptoms altogether. In some cases hormone applications can mimic virus symptoms. Particularly ABA and ethylene may reduce growth, promote senescence, and cause wilting, epinasty, and abscission. Since auxins, GAS, and cytokinins commonly counteract these effects of ABA and ethylene, spraying with these hormones, especially at high concentrations, often reduces symptom severity. Thus, GA reversed the stunting effect caused by rice tungro virus (RTV) on rice (Thomas and John, 1981);however, usually GAS only partly counteract virus-induced stunting. In cotton, the cytokinin kinetin, the auxins IAA and 2,4-dichlorophenoxyaceticacid, and ethephon all increased the capacity of TMV to replicate (Cheo, 1971).Since high levels of auxin stimulate ethylene synthesis in the plant, the stimulatory action of the auxins was supposed to be due to ethylene. In contrast, ethephon did not stimulate TMV accumulation in systemically infected Physalis floridanu. Carbendazim (methylbenzimidazole-2-ylcarbamate; MBC), which has a cytokinin-like action, alleviated symptoms of RTV on rice (Thomas and John, 1980) and prevented degeneration of the chloroplasts in BWYV-infected lettuce (Tomlinson and Webb, 1978).However, MBC failed to prevent symptom formation in BWYV-infected Claytonia perfoliata, in lettuce infected with lettuce mosaic virus, or in tobacco infected with CMV. In contrast, spraying TMV-infected tobacco with MBC was reported to completely mask the mosaic symptoms (Tomlinson et al., 19761, as well as significantly reduce virus ac-

240

L. C. VAN LOON

cumulation (Fraser and Whenham, 1978a). MBC was ineffective in suppression of TMV multiplication when supplied to mature leaves but caused a persistent inhibition of TMV multiplication lasting well through leaf maturity and senescence, if it entered the leaf while this was still very young. This indicates that the effect of MBC is dependent on leaf developmental stage (Fraser and Whenham, 1978131, similarly to the mosaic-inducing action of the virus. To investigate the effect of hormones on the development of the mosaic pattern in TMV-infected Samsun tobacco plants, the growing shoot tip and developing leaves were treated every 2-3 days during the period of systemic infection with solutions of each of the five classes of hormones. Although auxins, GAS, and cytokinins participate in regulating normal leaf growth and development, none of the treatments had any influence on the nature of the mosaic pattern developing on consecutive leaves. Masking of mosaic symptoms was observed with MBC and even more with the auxin naphthylacetic acid (NAA). Masking was due to a darkening of the light green areas, indicative of formation of near-normal amounts of chlorophyll, but the mosaic pattern itself remained unaltered. However, at the high concentrations of NAA necessary to achieve this effect, growth inhibition and epinasty occurred due to auxin-induced ethylene production. These results corroborate those of Nichols (1952) that spraying tobacco plants with auxins retarded development of symptoms and decreased severity of mosaic symptoms, but that no concentration of growth regulator was found to prevent all mosaic symptoms without causing injury. In contrast, the natural cytokinin zeatin and GA, antagonized the effect of NAA and increased symptom severity. GA, increased stem length in both healthy control and TMV-infected plants to about the same extent, indicating that exogenous GA, was not able to alleviate the virus-induced stunting. Under these conditions, both ABA and ethylene had only marginal effects (L. C. van Loon and E. Linders, unpublished observations). Thus, symptom severity was influenced in opposing ways by different regulators but symptom type and more specifically the delineation between light and dark areas were not changed. Apparently, the mosaic pattern is fixed at a very early stage of leaf development. In such newly developing leaves, when hormone metabolism is redirected in different ways in neighboring groups of cells, it is clearly impossible to restore normal conditions by the necessarily crude way of external application. What is needed are methods to determine endogenous concentrations of hormones at the cellular or even subcellular level, which will be a formidable task, given their often low concentrations.

DISEASE INDUCTION BY PLANT VIRUSES

241

C . Interference of Viral Products 1 . Early Events in Virus-Host Recognition To determine what happens between the time when the virus has entered the cell and when changes in cellular metabolism become first detectable, early events in virus-host recognition have to be sought. Although not a specific pathogenicity determinant, virus multiplication is a prerequisite for symptom development. When plants are inoculated with very low amounts of virus, symptoms may appear considerably later than when a massive dose of inoculum is applied, and systemic symptoms may be absent (Pilowsky et al., 1981)Conversely, the frequency and severity of systemic infection may increase with inoculum concentration (Edwards and Agrios, 1981). Probably, replication to a certain level is necessary to ensure the availability of a sufficient quantity of pathogenesis-determiningmolecules. For successful systemic infection, these molecules are necessary to avoid or suppress the eventual induction of resistance reactions, and to perturb the synthesis, metabolism, or action of the hormonal compounds that regulate normal growth and development. There is very little information on how virus infection causes alterations in host growth substance levels. Furthermore, little is known of the mechanism of action of hormones in healthy plants (Trewavas, 1981). However, in selected plant systems, growth regulators have been shown to regulate specific gene expression at the transcriptional level, suggesting that at least part of their action may be sought in the nucleus. Whatever its mechanism (Van Loon, 19831, resistance to viruses appears to be a nonspecific incompatibility phenomenon. Since most plants are resistant to most viruses, all plants possess resistance mechanisms against viruses. How far these are expressed upon infection depends on the particular combination of host and virus. Although plants with genetically determined differential sensitivity to one or more viruses react with resistance reactions only after infection with specific viruses or strains, virus localization in chlorotic or necrotic lesions engenders sufficiently similar physiological and biochemical changes to consider the reactions associated with resistance as largely nonspecific. In contrast, in susceptible interactions, the interaction appears specific. Whereas changes in protein patterns in plants reacting with hypersensitive necrosis are virtually identical, protein alterations were found to be at least quantitatively dissimilar in plants exhibiting symptoms varying from a slight mottling to bright yellow mosaic due to various viruses (Van Loon, 1972;White and Brakke, 1983).

242

L. C. VAN LOON

Since systemic infection appears to require selective suppression of the activation of resistance reactions that are genetically controlled, the virus might also directly influence gene expression, perhaps through a viral product interfering directly with host regulatory control mechanisms. It has been postulated repeatedly that viral (or viroid) RNA may interact with host DNA (e.g., Mundry, 1965; Zaitlin, 1979). One type of interaction of plant viruses with the host genome is the so-called aberrant ratio (AR) effect. This effect is produced in maize when pollen from a parent infected with BSMV, wheat streak mosaic virus, or corn lily fleck virus and carrying dominant traits mainly for kernel properties is used to fertilize a line resistant to the virus and carrying the corresponding recessive alleles. A low frequency of the progeny gave significant distortion of the expected ratios for one or more of the genetic markers, suggesting that virus infection resulted in inactivation of plant genes (Sprague et al., 1963; Sprague and McKinney, 1966, 1971). The AR effect was inherited in a stable manner in infected lines, with a low frequency of reversion to normal ratios. Once established, the effect was transposible to other loci. Although the genetic constitution of the original parents was questioned (Brakke et al., 1981) and it was suggested that at least some of the deviation could result from a lowered transmission of gametes carrying dominant traits through the male but not the female side (Nelson, 19811, an analysis of the mutation of the Adhl gene following infection with BSMV established the presence of an insertion at the Adh locus. This insert did not appear to contain sequences homologous to the BSMV genome. Rather, BSMV infection seemed to mobilize endogenous but “dormant” transposable elements (Mottinger et aZ.,1984). Hirai and Wildman (1967) demonstrated that treatment of small tobacco plants with actinomycin D, a n inhibitor of DNA-dependent RNA synthesis, led to the development in young leaves of mosaic symptoms apparently similar to those induced by the common strain of TMV. These results also suggest that symptoms and disease can arise from the interference of the infecting virus with host gene expression. It may be hypothesized, therefore, that symptom induction requires the recognition of host functions involved in the regulation of plant growth and development. The resulting interference could resemble hormonal disturbances in as far as the virus affects hormone-controlled processes. Under those conditions, exogenously applied growth regulators might partly compensate for these disturbances, depending on the nature, extent, and localization of the interference. Such a hypothesis may also explain why different regulators, or the same regulator applied at different times or concentrations, can have very

DISEASE INDUCTION BY PLANT VIRUSES

243

diverse effects and may significantly alter pathogenesis and symptom expression. In an extension of this hypothesis, the molecular basis of sensitivity to viral infection may be sought at the level of transcription of host DNA. Selective gene expression appears to be under the control of the proteins complexed with the DNA in the chromatin. Particularly the nonhistone chromatin proteins are seen as responsible for regulating gene activity (Cartwright et al., 1982). As a test of our hypothesis, we therefore investigated whether changes in chromatin-associated proteins occur in mosaic-diseased tobacco leaves. 2. Association of Viral Proteins with Host Chromatin The micromosaic characteristic of TMV-infected tobacco leaves precludes the physical separation of light- and dark-colored leaf parts. For that reason, an isolation method had to be sought that would ensure the recovery of a representative sample of nuclei from the whole leaf. Using young, developing leaves which when systemically infected develop the symptoms typical of the disease, this was achieved by tissue homogenization in a large volume of medium in an Omnimixer, followed by repeated grinding in a Potter homogenizer. With this method, up to 45% of the nuclei were released intact (Van Telgen and Van Loon, 1983). Since the yield of nuclei from TMV-infected leaves was typically only slightly lower than that from healthy controls, a selective extraction appeared rather unlikely. Nuclear proteins must thus have been derived from both the light green and the dark green tissues. Gel electrophoresis of purified chromatin protein preparations revealed that systemically infected leaves contained two additional polypeptides with apparent mass of 116 and 20 kDa, respectively (Fig. 2). The 20-kDa protein was identified serologically as the viral coat protein (Van Telgen et al., 1985a). It was only loosely associated with the chromatin and preferentially released upon incubation in 6 M urea. In contrast, the 116-kDa protein remained tightly bound under these conditions and was only dissociated together with all other chromatinassociated proteins in the presence of salt (Van Telgen et al., 1985b). By protease V8 mapping it was established that this protein was identical to the virally coded 126-kDa protein (Van Telgen et al., 1985~). No other differences between the chromatin protein profiles of healthy and TMV-infected leaves were discernable, indicating that these virus-induced changes were entirely virus specific. CMV also induced a virus-specific change in that its coat protein became associated with host chromatin in tobacco (Van Telgen et al., 1985a). During systemic infection with TMV, the virus was first detectable

244

L. C. VAN LOON

1

2

3

11694 -

67 -

43 -

30-

c

20 14 -

FIG.2. Electrophoretic patterns in 0.1% SDS-containing 7.5-17.5% linear gradient polyacrylamide gels of chromatin-associated proteins of Samsun tobacco grown under natural light a t 20-22°C (lane 2) and mosaic-diseased leaves of TMV-infected plants grown at 31-32°C under fluorescent light only (lane 3).The positions of the virally coded 126-kDa protein and the coat protein are indicated by large arrows. The additional band a t 31 kDa (small arrow) is due to the difference in light conditions. Lane 1 contains the marker proteins p-galactosidase (116 kDa), phosphorylase b (94 kDa), bovine serum albumin (43 kDa), carbonic anhydrase (30 kDa), trypsin inhibitor (20 kDa), a-lactalbumin (14 kDa). (Courtesy of Dr. H. J. van Telgen.)

in the young leaves 4 days after fully grown lower leaves had been inoculated. Vein-clearing became evident after 7 days. The 126-kDa protein became discernable between 120 and 144 hours after inoculation, coinciding with the appearance of the first visible symptoms. Although it was also present both in the soluble protein and in the sedimentable membrane fractions, its concentration in nuclei was calculated to be about eightfold higher than in the cytoplasm, suggesting that it is preferentially associated with the chromatin (Van Telgen et al., 1985b). These properties of the 126-kDa protein are highly suggestive of a possible regulatory role in pathogenesis and symptom expression in systemically infected leaves. This conclusion is rein-

DISEASE INDUCTION BY PLANT VIRUSES

245

forced by the recent finding of Nishiguchi et al. (1985) that the attenuated strain L,,A of TMV differs from the virulent ancestral strain L by 10 base substitutions. Seven of these occurred in the third bases of in-phase codons and did not influence amino acid sequences. Only three, at positions 1117,2349, and 2754 in the common reading frame of the 126- and 183-kDa proteins, resulted in amino acid changes. An intermediate strain, L,,, differed from strain L only in nucleotide 1117. This change corresponds to the replacement of a cysteine by a tyrosine residue. Since the virulence of L,, is much closer to that of L,,A than to that of L, it seems that the base substitution at position 1117 is primarily responsible for the attenuation of virulence. In protoplasts, the 126-kDa protein is synthesized very early in infection (Siege1 et al., 19781, indicating that it may indeed function in symptom induction. Also coat protein, though only loosely bound, may not be without effect in view of the apparent relationship between coat protein properties and resistance gene expression (e.g., Section II,C,l,a). CMV coat protein is more basic than TMV coat protein and might be more tightly associated with host chromatin. Interestingly, Roberts and Wood (1981~) described that in tobacco the severe yellow mosaic-inducing P6 strain of CMV synthesized considerably more coat protein than the mild W strain. In tobacco leaves systemically infected with TMV the 126-kDa protein constitutes up to about 0.6% of the amount of protein associated with the chromatin. Assuming the protein to be present in all leaf cells-which is presumably an overestimation, as not all cells will be infected-this would correspond to about 2.106 molecules per cell nucleus. This is far more than seems necessary to interfere with the expression of a few specific genes such as those determining resistance, and suggests that the virus might affect gene expression in a more general way. Recently, Evans et al. (1985) used the 8-azidopurine analogs 8-N3ATP and 8-N3GTP to photolabel nucleotide-binding sites induced by TMV in tobacco. They found the 126-kDa protein to become photolabeled and contain (a) nucleotide binding site(s) with low micromolar affinity constants for all ribo- and deoxyribonucleotides. Although they suggested that the protein forms part of an RNA-dependent RNA polymerase, these findings are equally consistent with its chromatin-binding property. Since the 126-kDa protein has a rather high isoelectric point and is positively charged at physiological pH, it might affect the chromatin structurally. On the other hand, whereas the protein was clearly present among the chromatin-associated proteins of TMV-infected Samsun tobacco plants at 30°C, it was not discernable in the chromatin of mosaic-diseased leaves of Samsun NN tobacco at this temperature. As small quantities of the protein fall

246

L. C . VAN LOON

below the level of detection, i t cannot yet be decided how significant this difference between the two tobacco cultivars is. Mosaic symptoms are induced only in young, developing leaves. If the association between viral proteins and host chromatin is instrumental in symptom induction, the proteins can only interfere with the expression of genes that function specifically early in leaf development. It must be assumed, then, that if the proteins associate with the chromatin in leaves where these genes have already been transcribed, no symptoms can be induced. Fully grown leaves, when infected, also contain 126-kDa protein in their chromatin but the amounts are considerably less than in systemically infected leaves (Van Telgen et al., 1985b). The TMV-coded 126-kDa protein has approximately 20-30% amino acid homology with similarly sized translational products of AMV and BMV (Haseloff et al., 1984; Cornelissen and Bol, 1984). This homology suggests that other plant viruses code for proteins that may similarly interact with their hosts. One may speculate, therefore, that association of one or more virally coded proteins with host chromatin may be a more general strategy of plant viruses to exploit host cells and, consequently, cause disease. Although the subcellular localization of TMV multiplication is still being disputed, it seems improbable that the association of some of the virally coded proteins with the host chromatin constitutes an exclusive requirement for viral replication. However, viruses that do replicate and mature in the nucleus, as shown for instance for pea enation mosaic virus (Powell et al., 19771, several rhabdoviridae (Francki, 1973), and the DNA-containing CaMV and geminiviruses (Gracia and Shepherd, 1985; Howell, 1985), would have at least some of their proteins already at the site where interference with host transcription and, perhaps, mRNA processing can take place. Virus particles of several small isometric viruses accumulate in the nucleus in addition t o the cytoplasm. Other viruses, such as :he potyviruses, induce intranuclear inclusions containing protein of viral origin (for review, see Matthews, 1981). Nonhistone chromatin proteins contain sequences which make them accumulate preferentially in nuclei. It will be interesting to see if comparable sequences can explain the nuclear occurrence of specific virally coded proteins. Viroids are also present in the nucleus, probably complexed in the nucleolus via protein-nucleic acid interactions (Schumacher et a1., 1983). For PSTV it has been suggested that host DNA-dependent RNA polymerase I1 is involved in the viroid replication process (Rackwitz et al., 1981). Thus, it appears that viroids are synthesized outside the nucleolus and later become associated with this subnuclear component. It is conceivable, therefore, that viroids cause symptoms by di-

DISEASE INDUCTION BY PLANT VIRUSES

247

rectly interacting with host DNA, thereby interfering with normal gene expression. Gel electrophoretic techniques, although able to resolve chromatin proteins from tobacco into up t o 350 components (Van Telgen and Van Loon, 1985), do not allow detection of tiny amounts of individual proteins. Hence, it is extremely difficult to determine when virally coded proteins first enter the nucleus and associate with the chromatin. Protoplasts may be infected semisynchronously . However, protoplasts do not develop s7mptoms. Since the stage of leaf development affects the rate of accumulation of the 126-kDa TMV-coded protein associated with the chromatin, it may well be that the interaction in protoplasts will turn out to be different, and thus, the information obtained from protoplast studies will be of limited value. To be able to support extensive virus multiplication, protoplasts have to be isolated from fully grown leaves. Moreover, isolation of protoplasts from young, developing tobacco leaves has so far proved beyond our technical capabilities (Van Telgen and Van Loon, 1983). Differential temperature treatment may enhance the synchrony of virus replication in intact plants. Dawson et al. (1975) inoculated the fully grown leaves of a tobacco plant with TMV and kept these leaves at a temperature of 27"C, at which rapid virus multiplication takes place. The young, noninoculated leaves were kept at a low temperature (3-12'0, at which no virus synthesis occurs. Under these conditions virus inoculum was transported from the inoculated to the noninoculated leaves. When the plants were subsequently placed at 25°C in their entirety, almost synchronous virus multiplication occurred in the young leaves. In a similar approach, Roberts and Wood (1981b) successfully increased the synchrony of CMV replication and symptom expression in tobacco. Such procedures obviously will be of great value in the study of disease induction.

V. CONCLUDING REMARKS To relate viral properties to disease symptoms, viral mutants offer distinct advantages over naturally occurring strains which may differ considerably in nucleotide sequence. Statistically, alteration of a single base may completely change symptom type or severity, indicating that any interaction of a viral product with a host constituent may be highly specific. It is quite reasonable to suppose that a single amino acid substitution may profoundly influence the interaction of a viral protein with, for instance, host DNA. Similarly, a single base change in host DNA may be decisive in altering the binding characteristics of

248

L.C.VAN LOON

a viral protein. Apart from the few examples where resistance is expressed at the protoplast level, virus replication in plant cells does not seem to require specific interactions beyond those supplying the subcellular environment necessary for nucleic acid and protein synthesis. However, at the whole-plant level, virus multiplication and spread more often than not are impeded as a result of the activation of resistance mechanisms. How and at what level resistance is being triggered is unknown at present, but the expression of whole plant resistance requires an activation of the cells neighboring the point of entry of the virus. This activation involves changes in the patterns of transcription and translation, must be nuclear controlled, and is probably mediated by diffusible messengers with hormone-like properties. Systemic infection may result when the virus is able to specifically suppress this response, perhaps by interfering directly with the genetic control mechanisms regulating resistance. At least two properties of the virus are essential for successful systemic infection: (1)the virus must multiply and spread sufficiently rapidly to neighboring cells to be able to suppress the activation of resistance reactions by the chemical messengers diffusing from infected cells, or to remain ahead when it cannot, and (2)the virus must be able to specifically interact with the genetic control mechanisms. Resistance, whatever its phenotypic character (chlorotic or necrotic lesions, slow virus movement), is the inevitable consequence of insufficient adaptation of a virus to its host in these two respects. Since all virally coded proteins, perhaps with the exception of coat protein, have been implicated in virus multiplication and spread, variations in these proteins may, at least indirectly, affect symptom expression. These considerations can explain why all parts of the genetic information of viruses appear t o contain determinants for pathogenicity. In cross-protection infection with one strain of a virus renders a plant apparently immune to subsequent infection with a different strain of the virus. Multiplication of the second virus is suppressed and symptoms are slighter, milder, or inapparent. Although coat protein has been implicated in cross-protection (Sherwood and Fulton, 1982;Abel et al., 1986),the phenomenon can be induced by a coat-protein-free mutant of TMV as well as by mutants with nonfunctional coat proteins (Sarkar and Smitamana, 1981); it also occurs between viroids, which do not possess a protein coat (Niblett et al., 1978).Although usually the primary effect in cross-protection appears to be on virus multiplication, replication of the superinfecting strain may occur without symptoms becoming apparent, indicating that cross-protection does not represent complete immunity but rather insensitivity (Fulton, 1982). Such a condition can be explained if, for a virus to spread and symptoms to be produced, a viral protein has to interact with host DNA. A superinfect-

DISEASE INDUCTION BY PLANT VIRUSES

249

ing strain of the virus may find its binding sites on the DNA occupied by the corresponding product of a previously inoculated strain. Crossprotection is closely confined to the tissues in which the inducing virus is present. Apparently, the superinfecting virus is unable to interact with the regulatory controls involved in symptom induction. In contrast, acquired resistance, in which following a primary infection resistance to a challenging pathogen is enhanced, is entirely nonspecific. This condition results from the activation of defense reactions in response to the primary infection and usually extends to plant parts distant from the site where the pathogen is present, although these do not show symptoms. No changes in chromatin proteins were discernable in young, systemically resistant tobacco leaves (Van Telgen et al., 1985a). The induced resistance is solely a plant reaction and can be regarded as an increased capacity to make use of existing resistance mechanisms (Van Loon, 1983). The development of methods to clone cDNA copies of RNA viruses has greatly facilitated the sequencing of viral genomes. This offers the possibility of determining exactly what base substitutions must be responsible for altered symptoms induced by viral mutants. Further manipulation of viral genomes can be achieved by site-directed mutagenesis. Such methods have already revealed transitions abolishing the infectivity of HSV (Ishikawa et al., 1985) and PSTV (Tabler and Sanger, 1985; Owens et al., 1986). Plasmid DNA and/or their in uitro transcripts containing full copies of PSTV (Cress et al., 19831, BMV (Ahlquist et al., 1984) and TMV (Knorr et al., 1986; Y. Okada, personal communication) have proved infectious and to give rise to RNA transcripts identical to the viroid and virus, respectively. A systematic survey which nucleotides are required for infectivity, replication, and pathogenesis can reveal the molecular nature of the pathogenic determinants of the virus and their relationship to symptom type and severity. This will also allow further testing of the hypothesis that symptom induction depends on a specific interaction between a viral protein and host DNA. In this way, study of the molecular basis of pathogenesis and symptom expression may contribute to a better understanding of normal cellular processes and their control.

ACKNOWLEDGMENTS In developing these ideas, over the years I have been aided by discussions with Drs. J. Bruinsma, R. S. S. Fraser, R. W. Goldbach, C. P. de Jager, A. van Kammen, K. W. Mundry, H. J. van Telgen, and P. J. G. M. de Wit. I also thank J. Bruinsma, C. P. de Jager, and A. van Kammen for their criticisms during the preparation of this paper.

REFERENCES Abel, P. P., Nelson, R. S., Barun, D., Hoffmann, N., Rogers, S. G., Fraley, R. T., and Beachy, R. N. (1986). Science 232, 738-743.

250

L. C. VAN LOON

Aharoni, N., Marco, S., and Levy, D. (1977).Physiol. Plant Pathol. 11, 189-194. Ahlquist, P., French, R., Janda, M., and Loesch-Fries, L. S. (1984).Proc. Natl. Acad. Sci. U S A . 81,7066-7070. Altenbach, S. B., and Howell, S. H. (1981).Virology 112, 25-33. Atabekov, J. G., and Dorokhov, Yu.L. (1984).Adv. Virus Res. 29, 313-364. Bailiss, K. W. (1974).Physiol. Plant Pathol. 4, 73-79. Bailiss, K. W., BalBzs, E., and KirBly, Z. (1977).Acta Phytopathol. Acad. Sci. Hung. 12, 133-140. Bancroft, J. B., and Lane, L. C. (1973).J. Gen. Virol. 19, 381-389. Ben-Tal, Y.,and Marco, S.(1980).Physiol. Plant Pathol. 16, 327-336. Berna, A., Briand, J. P., Stussi-Garaud, C., Godefroy-Colburn, T., and Hirth, L. (1984). Ann. Virol. (Znst. Pasteur) 135 E,285-296. Bos, L. (1978).“Symptoms of Virus Diseases in Plants.” Pudoc, Wageningen. Bosch, F. X.,and Jockusch, H. (1972).Mol. Gen. Genet. ll6,95-98. B o d , J. M.,and Bov6, C. (1985).Physiol. Veg. 23, 741-748. Brakke, M. K., Samson, R. G., and Compton, W.A. (1981).Genetics 99,481-485. Carlson, P. S.,and Murakishi, H. H. (1978).Plant Sci. Lett. 13,377-381. Cartwright, I. L., Abmayr, S. M., Fleischmann, G., Lowenhaupt, K.,Elgin, S. C. R., Keene, M. A., and Howard, G. C. (1982).Crit. Rev. Biochem. 13, 1-86. Cheo, P. C. (1971).Phytopathology 61,869-872. Collmer, C. W., Tousignant, M. E., and Kaper, J. M. (1983).Virology 127, 230-234. Collmer, C. W., Hadidi, A., and Kaper, J. M. (1985).Proc. Natl. Acad. Sci. U S A . 82, 3110-3 114. Conejero, V., and Granell, A. (1986).Physiol. Mol. Plant Pathol. (in press). Cornelissen, B. J. C., and Bol, J. F. (1984).Plant Mol. Bwl. 3, 379-384. Costa, A. S.(1969).Phytopathol. 2.85, 219-230. Cress, D. E.,Kiefer, M. C., and Owens, R. A. (1983).Nucleic Acids Res. 11,6821-6835. Daubert, S., Shepherd, R. J., and Gardner, R. C. (1983).Gene 25, 201-208. Davies, D. L., and Clark, M. F. (1983).Ann. Appl. Biol. 103,439-448. Dawson, J. R. O., Rees, M. W., and Short, M. N. (1979).Ann. Appl. Biol. 91, 353-358. Dawson, W. O.,Schlegel, D. E., and Lung, M. C. Y.(1975).Virology 65,565-573. De Jager, C. P. (1976).Virology 70, 151-163. De Jager, C. P. (1978).“Genetische Analyse van Cowpea-Mozaiekvirusmutanten.”Thesis, Agricultural University, Wageningen. De Jager, C. P., and Breekland, L. (1979).Virology 99,312-318. De Jager, C. P., and Weaseling, J. B. M. (1981).Physiol. Plant Pathol. 19, 347-358. De Laat, A. M. M., and Van Loon, L. C. (1983).Physiol. Plant Pathol. 22,261-273. De Zoeten, G. A., Assink, A. M., and Van Kammen, A. (1974).Virology 59, 341-355. Dickson, E. (1981).Virology 116,216-221. Dickson, E.,Robertson, H. D., Niblett, C. L., Horst, R. K., and Zaitlin, M. (1979).Nature (London) 277,60-62. Diener, T. 0. (1971).Virology 43,7549. Dingjan-Versteegh, A., Van Vloten-Doting, L., and Jaspars, E. M. J. (1972).Virology 49, 716-722. Dixon, L. K., Koenig, I., and Hohn, T. (1983).Gene 25, 189-199. Dodds, J. A., and Hamilton, R. I. (1972).Virology 60,404-411. Dorokhov, Y. L., Alexandrova, N. M., Miroshnichenko, N. A., and Atabekov, J. G. (1983). Virology 127, 237-252. Dorokhov, Yu. L., Alexandrova, N. M., Miroshnichenko, N. A., and Atabekov, J. G. (1984).Virology 137, 127-134. Dorssers, L., Van der Krol, S., Van der Meer, J., Van Kammen, A., and Zabel, P. (1984). P m . Natl. Acad. Sci. U S A .81, 1951-1955.

DISEASE INDUCTION BY PLANT VIRUSES

25 1

Dougherty, W. G., and Hiebert, E. (1980a).Virology 104,174-182. Dougherty, W. G., and Hiebert, E. (1980b).Virology 104,183-194. Drijfhout, E. (1978).“Genetic Interaction between Phnseolus vulgaris and Bean Common Mosaic Virus with Implications for Strain Identification and Breeding for Resistance.” Pudoc, Wageningen. Drijfhout, E., Silbernagel, M. J., and Burke, D. W. (1978).Neth. J.Plant Pathol. 84,1326. Duran-Vila, N.,and Semancik, J. S. (1982).Phybpathology 72, 777-781. Edwards, M. C., and Agrios, G. N. (1981).Phytopathology 71, 7-11. Edwards, M. C.,Gonsalves, D., and Prowidenti, R. (1983).Phytopathology 73,269-273. Ehara, Y . , and Misawa, T. (1975).Phytopathol. Z.84,233-252. Ellingboe, A. H. (1984).Adv. Plant Pathol. 2, 131-151. Evans, D. (1985).J . Gen. Virol. 66, 339-343. Evans, R. K., Haley, B. E., and Roth, D. A. (1985).J. Biol. Chem. 280,7800-7804. Flor, H. H. (1971).Annu. Rev. Phytopathol. 9, 275-296. Fraenkel-Conrat, H. (1979).Trends Biochem. Sci. 4, 184-186. Francki, R. I. B. (1973).Adv. Virus Res. 18,257-345. Francki, R. I. B., Grivell, C. J., and Gibb, K. S. (1986).Virology 148,381-384. Fraser, L. G., and Matthews, R. E. F. (1983).Plant Sci. Lett. 29, 67-72. Fraser, R. S. S. (1969).Mol. Gen. Genet. 106, 73-79. Fraser, R. S. S. (1983).Physiol. Plant Pathol. 22, 109-119. Fraser, R. S. S. (1986).Crit. Rev. Plant Sci. 3,257-294. Fraser, R. S. S., and Gerwitz, A. (1980).J. Gen. Virol. 46, 139-148. Fraser, R. S. S., and Loughlin, S. A. R. (1980).J. Gen. Virol. 48, 87-96. Fraser, R. S. S., and Loughlin, S. A. R. (1982).Physiol. Plant Pathol. 20, 109-117. Fraser, R. S. S., and Whenham, R. J. (1978a).J. Cen. Virol. 39, 191-194. Fraser, R. S. S., and Whenham, R. J. (1978b).Physiol. Plant Pathol. 13,51-64. Fraser, R. S.S., and Whenham, R. J. (1982).Plant Growth Regul. 1, 37-59. Fraser, R. S. S., Gerwitz, A., and Morris, G. E. L. (1986).Physiol. Mol. Plant Puthol. 29, 239-249. Fulton, R. W. (1975).Virology 67, 188-196. Fulton, R. W. (1982).In “Active Defense Mechanisms in Plants” (R. K. S.Wood, ed.), pp. 231-245. Plenum, New York. Gaborjanyi, R., Balhzs, E., and Kirhly, Z. (1971).Acta Phytopathol. Acud. Sci. Hung. 4, 51-55. Galindo, A. J., Smith, D. R., and Diener, T. 0. (1984).Physiol. Plant Pathol. 24, 257275. Garcia-Arenal, F., Palukaitis, P., and Zaitlin, M. (1984).Virology 132, 131-137. Goelet, P., Lomonossoff, G. P., Butler, P. J. G., Akam, M. E., Gait, M. J., and Karn, J. (1982).Proc. Natl. Acad. Sci. U S A . 79,5818-5822. Goldbach, R., Ftezelman, G., and Van Kammen, A. (1980).Nature (London) 286, 297300. Gracia, O.,and Shepherd, R. J. (1985).Virology 146, 141-145. Haber, S., and Hamilton, R. I. (1980).J. Gen. Virol. 50, 377-389. Habili, N., and Francki, R. I. B. (1974).Virology 61,443-449. Habili, N.,and Kaper, J. M. (1981).Virology 112,250-261. Hall, T.C.(1979).Znt. Rev. Cytol. 60,1-26. Hall, A. E., and Loomis, R. S. (1972).Plant Physiol. 50, 576-580. Hamilton, R. I., and Dodds, J. A. (1970).Virology 42,266-268. Harrison, B.D.,Murant, A. F., Mayo, M. A., and Roberts, I. M. (1974).J.Gen. Virol. 22, 233-247. Hartmann, D., Mohier, E., Leroy, C., and Hirth, L. (1976).Virology 74,470-480.

252

L. C. VAN LOON

Haseloff, J., Goelet, P., Zimmern, D., Ahlquist, P., Dasgupta, R., and Kaesberg, P. (1984).Proc. Natl. Acad. Sci. U S A . 81,4358-4362. Hillman, B. I., Morris, T. J., and Schlegel, D. E. (1985).Phytopathology 76, 361-365. Hirai, A., and Wildman, S. G. (1967).Virology 31, 721-722. Holland, J., Spindler, K., Horodyski, F., Grabau, E., Nichol, S., and VandePol, S. (1982). Science 215, 1577-1585. Holmes, F. 0. (1938).Phytopathology 28, 553-561. Holmes, F. 0.(1952).Indian Phytoputhol. 5,8-16. Holmes, F. 0.(1960).Virology 12, 59-67. Hooley, R., and McCarthy, D. (1980).Physiol. Plant Pathol. 16,25-38. Howell, S. H. (1985).Crit. Rev. Plant Sci. 2,287-316. Ishikawa, M., Meshi, T., Okada, Y., Sano, T., and Shikata, E. (1985).J . Biochern. 98, 1615-1620. Jockusch, H. (1966a).2.Vererbungsl. 98,320-343. Jockusch, H. (1966b).2. Vererbungsl. 98,344-362. Jockusch, H.(1968).Virology 35,94-101. Jockusch, H.,and Jockusch, B. (1968).Mol. Gen. Genet. 102,204-209. Jones, A. T.,and Duncan, G. H. (1980).J . Gen. Virol. 50, 269-277. Jones, A. T.,Mayo, M. A., and Duncan, G. H. (1983).J . Gen. Virol. 64, 1167-1173. Jones, G. E.,and Dawson, W. 0. (1978).Intervirology 9,149-155. Kado, C. I., and Knight, C. A. (1966).Proc. Natl. Acad. Sci. U S A . 55, 1276-1283. Kamer, G., and Argos, P. (1984).Nucleic Acids Res. 12, 7269-7282. Kano, H. (1985).Plant Cell Physiol. 26, 1241-1249. Kaper, J. M., and Waterworth, H. E. (1977).Science 196,429-431. Knorr, D. A., Beck, D. L., and Dawson, W. 0. (1986).J . Cell. Biochem. Suppl. IOD,294. Kuhn, C. W., Wyatt, S. D., and Brantley, B. B. (1981).Phytopathology 71, 1310-1315. Lakshman, D. K., and Gonsalves, D. (1985).Phytopathology 75, 758-762. Lakshman, D. K.,Gonsalves, D., and Fulton, R. W. (1985).Phytopathology 75,751-757. Leonard, D. A., and Zaitlin, M. (1982).Virology 117,416-424. Lister, R. M. (1968).J . Gen. Virol. 2,43-58. Lister, R. M., and Bracker, C. E. (1969).Virology 37, 262-275. McGovern, M. H., and Kuhn, C. W. (1984).Phytopathology 74,95-99. Marchoux, G., Marrou, J., Devergne, J. C., Quiot, J. B., Douine, L., and Lot, H. (1975). Meded. Fac. Landbouwwetenschuppen Gent 40,59-72. Marco, S., Levy, D., and Aharoni, N. (1976).Physiol. Plant Puthol. 8, 1-7. Martelli, G.P.,and Russo, M. (1977).Adu. Virus Res. 21, 175-266. Matthews, R. E.F. (1980).In “Comprehensive Virology, Vol. 16;Virus-Host Interactions, Viral Invasion, Persistence and Disease” (H. Fraenkel-Conrat and R. R. Wagner, eds.), pp. 297-359. Plenum, New York. Matthews, R. E. F. (1981).“Plant Virology,” 2nd Ed. Academic Press, New York. Mayo, M. A., Barker, H., and Harrison, B. D. (1982).J . Gen. Virol. 59, 149-162. Meshi, T.,Ohno, T., and Okada, Y. (1982).J . Biochern. 91, 1441-1444. Motoyoshi, F., and Oshima, N. (1977).J . Gen. Virol. 34, 499-506. Mottinger, J. P., Johns, M. A., and Freeling, M.(1984).Mol. Gen. Genet. 195,367-369. Mundry, K. W. (1965).In “Reproduction: Molecular, Subcellular, and Cellular,” pp. 6594.Academic Press, New York. Mundry, K. W., and Gierer, A. (1958).2.Vererbungsl. 89,614-630. Naidu, R. A., Krishnan, M., Ramanujam, P., Gnanam, A., and Nayudu, M. V. (1984a). Physiol. Plant Pathol. 25, 181-190. Naidu, R. A., Krishnan, M., Nayudu, M. V., and Gnanam, A. (1984b).Physwl. Plant Pathol. 25, 191-198. Nassuth, A., Alblas, F., and Bol, J. F. (1981).J. Gen. Virol. 53, 207-214.

DISEASE INDUCTION BY PLANT VIRUSES

253

Nelson, 0. E. (1981).Maydica 26, 119-131. Niblett, C. L.,Dickson, E., Fernow, K. H., Horst, R. K., and Zaitlin, M. (1978).Virology

91, 198-203.

Nichols, C. W. (1952).Phytoputhology 42, 579-580. Nilsson-Tillgren, T., Kolehmainen-SevBus, L., and Von Wettstein, D. (1969).Mol. Gen. Genet. 104, 124-141. Nishiguchi, M., Motoyoshi, F., and Oshima, N. (1978).J. Gen. Virol. 39, 53-61. Nishiguchi, M.,Kikuchi, S., Kiho, Y.,Ohno, T., Meshi, T., and Okada, Y. (1985).Nucleic Acids Res. 13, 5585-5590. Nolla, J. A. B. (1935).J. Agric. Uniu. Puerto Rico 19, 29-49. Nolla, J. A. B. (1938).J. Hered. 29,42-48. Ohashi, Y.,and Shimomura, T. (1982).Physiol. Plant Puthol. 20, 125-128. Ohno, T., and Okada, Y. (1977).Virology 76, 429-432. Ohno, T., Takamatsu, N., Meshi, T., Okada, Y., Nishiguchi, M., and Kiho, Y. (1983). Virology 131, 255-258. Okuno, T., Hiruki, C., Rao, D. V., and Figueiredo, G. C. (1983).J. Gen. Virol. 64,1907-

1914.

Osman, T. A. M., Dodd, S. M., and Buck, K. W. (1986).J . Gen. Virol. 67, 203-207. Owens, R. A.,Hammond, R. W., Gardner, R. C., Kiefer, M. C., Thompson, S. M., and Cress, D. E. (1986).Plant Mol. Eiol. 6,179-192. Pilowsky, M., Frankel, R., and Cohen, S. (1981).Phytopathology 71,319-323. Platt, S. G., Henriques, F., and Rand, L. (1979).Physwl. Plant Pathol. 16, 351-365. Powell, C. A., De Zoeten, G. A., and Gaard, G. (1977).Virology 78, 135-143. Rackwitz, H.-R., Rohde, W., and Sanger, H. L. (1981).Nature (London) 291, 297-301. Rao, A. L. N., and Francki, R. I. B. (1982).J. Gen. Virol. 61, 197-205. Reinero, A., and Beachy, R. N. (1986).Plant Mol. BioZ. 6,291-301. Rezelman, G.,Franssen, H. J.,Goldbach, R. W., Ie, T. S., and Van Kammen, A. (1982).J . Gen. Virol. 60,335-342. Riesner, D., and Gross, H. J. (1985).Annu. Rev. Biochem. 54,531-564. Roberts, P. L.,and Wood, K. R. (1981a).Physiol. Plant Pathol. 19, 99-111. Roberts, P. L.,and Wood, K. R. (1981b).Phytopathol. 2.102, 114-121. Roberts, P. L.,and Wood, K. R. (1981~). Phytopathol. 2.102, 257-265. Robinson, D. J. (1983).J. Gen. Virol. 64, 657-665. Rodriguez, J. L.,Martinez, J. L. G., and Flores, R. (1978).Physiol. Plant Pathol. 13,355-

363.

Roosien, J., and Van Vloten-Doting, L. (1982).J. Gen. Virol. 63, 189-198. Roosien, J., Sarachu, A. N., Alblas, F., and Van Vloten-Doting, L. (1983).Plant MoZ. Biol. 2, 85-88. Sanger, H. L. (1982).In “Encyclopedia of Plant Physiology, New Series, Vol. 14B;Nucleic Acids and Proteins in Plants 11” (B. Parthier and D. Boulter, eds.), pp. 368-454. Springer-Verlag, Berlin. Sarkar, S., and Smitamana, P. (1981).Naturwissenschften 68,145-146. Schnolzer, M.,Haas, B., Ramm, K., Hofmann, H., and Sanger, H. L. (1985).EMBO J . 4,

2181-2190.

Schumacher, J., Sanger, H. L., and Riesner, D. (1983).EMEO J. 2, 1549-1555. Schuster, G.,and Schramm, G. (1958).2.Nuturforsch. 13b,697-704. Sherwood, J. L.,and Fulton, R. W. (1982).Virology 119, 150-158. Sidhu, G. S. (1975).Sci. Prog. 62, 467-485. Siegel, A., Hari, V., and Kolacz, K. (1978).Virology 85,494-503. Spikes, J. D., and Stout, M. (1955).Science 122,375-376. Sprague, G.F., and McKinney, H. H. (1966).Genetics 54, 1287-1296. Sprague, G. F., and McKinney, H. H. (1971).Genetics 67, 533-542.

L. C. VAN LOON Sprague, G. F., McKinney, H. H., and Greeley, L. (1963).Science 141, 1052-1053. Sulzinski, M.A., and Zaitlin, M. (1982).Virology 121, 12-19. Sulzinski, M. A., Gabard, K. A., Palukaitis, P., and Zaitlin, M. (1985).Virology 145,132140. Suttle, J. C., and Kende, H. (1980).Plant Physiol. 85, 1067-1072. Sziraki, I., and BalBzs, E. (1975).In “Current Topics in Plant Pathology (2.Kirhly, ed.), pp. 345-352. Akademiai Kiad6, Budapest. Tabler, M., and Sanger, H. L. (1985).EMBO J . 4,2191-2199. Takahashi, W. N.(1956).Phytopathology 46,654-656. Taliansky, M. E., Atabekova, T. I., Kaplan, I. B., Morozow, S. Yu., Malyshenko, S. I., and Atabekov, J. G. (1982a).Virology 118,301-308. Taliansky, M. E., Malyshenko, S. I., Pshennikova, E. S., Kaplan, I. B., Ulanova, E. F., and Atabekov, J. G. (1982b).Virology 122,318-326. Taliansky, M. E., Malyshenko, S. I., Pshennikova, E. S., and Atabekov, J. G. (1982~). Virology 122,327-331. Thomas, H.,and Stoddart, J. L. (1980).Annu. Rev. Plant Physiol. 31, 83-111. Thomas, J., and John, V. T. (1980).Plant Dis. 64,402-403. Thomas, J., and John, V. T. (1981).Phytopathol. 2.101, 168-174. T6bias, I., Rast, A. T. B., and Maat, D. Z. (1982).Neth. J . Plant Pathol. 88,257-269. Tomlinson, J. A,, and Webb, M. J. W. (1978).Physwl. Plant Pathol. 12, 13-18. Tomlinson, J. A., Faithfull, E. M., and Ward, C. M. (1976).Ann. Appl. Bwl. 64, 31-41. Trewavas, A. (1981).Plant Cell Environ. 4, 203-228. Valleau, W. D. (1952).Phytopathology 42,40-42. Van Loon, L. C. (1972).“Pathogenese en Symptoomexpressie in Viruszieke Tabak; Een Onderzoek naar Veranderingen in Oplosbare Eiwitten.” Thesis, Agricultural University, Wageningen. Van Loon, L. C. (1977).Virology 80,417-420. Van Loon, L. C. (1983).In ‘The Dynamics of Host Defence” (J. A. Bailey and B. J. Deverall, eds.), pp. 123-190. Academic Press, Sydney. Van Loon, L. C., and Callow, J. A. (1983).In “Biochemical Plant Pathology” (J. A. Callow, ed.), pp. 385-414. Wiley, Chichester. Van Telgen, H. J., and Van Loon, L. C. (1983).2.Pflanzenphysiol. 112,171-180. Van Telgen, H. J., and Van Loon, L. C. (1985).J . Plant Physiol. 118, 285-296. Van Telgen, H. J., Van der Zaal, E. J., and Van Loon, L. C. (1985a).Physiol. Plant Pathol. 26, 83-98. Van Telgen, H. J., Van der Zaal, E. J., and Van Loon, L. C. (1985b).Physiol. Plant Pathol. 26,99-109. Van Telgen, H. J., Goldbach, R. W., and Van Loon, L. C. (1985~). Virology 143,612-616. Visvader, J. E., and Symons, R. H. (1985).Nucleic Acids Res. 13, 2907-2920. Von Sengbusch, P. (1965).Z.Vererbungsl. 96,364-386. Waterworth, H.E., Kaper, J. M., and Tousignant, M. E. (1979).Science 204,845-847. Weber, P. V. V. (1951).Phytopathology 41,593-609. Wetter, C. (1984).Phytopathology 74, 1308-1312. Whenham, R.J., and Fraser, R. S. S. (1981).Physiol. Plant Pathol. 18, 267-278. Whenham, R. J., Fraser, R. S. S., and Snow, A. (1985).Physiol. Plant Pathol. 26, 379387. White, J. L., and Brakke, M.K. (1983).Physiol. Plant Pathol. 22, 87-100. Wilson, T. M. A., Perham, R. N., Finch, J. T., and Butler, P. J. G. (1976).FEBS Lett. 64, 285-289. Wyatt, S. D., and Kuhn, C. W. (1980).J. Gen. Virol. 49, 289-296. Xiong, C., Muller, S., Lebeurier, G., and Hirth, L. (1982).EMBO J. 1, 971-976.

DISEASE INDUCTION BY PLANT VIRUSES

255

Xiong, C., Lebeurier, G., and Hirth, L. (1984). Proc. Natl. Acad. Sci.U S A . 81,66086612. Yaguchi, S., and Takahashi, T. (1985). 2.Pflanzenkmnkh. Pflanzenschutz 92,263-269. Yarwood, C . E. (1979). Adu. Virus Res. 25, 169-190. Zaitlin, M.(1979). In “Plant Disease, An Advanced Treatise” (J. G. Horsfall and E. B. Cowlings, eds.), Vol. IV,pp. 257-271. Academic Press, New York. Zimmern, D., and Hunter, T. (1983). EMBO J . 2, 1893-1900.

This Page Intentionally Left Blank

ADVANCES IN VIRUS KESEARCH, VOL. 33

THE DIANTHOVIRUSES: A DISTINCT GROUP OF ISOMETRIC PLANT VIRUSES WITH BIPARTITE GENOME C. Hiruki Depattment of Plant Science University of Alberta Edmonton, Alberta, Canada T6G 2P5

I. INTRODUCTION In 1981 the International Committee on Taxonomy of Viruses (ICTV) approved the establishment of a new plant virus group named dianthoviruses (Matthews, 1982). All member viruses have a bipartite genome and share the following general properties. Virions contain single-stranded genomic RNAs of approximate molecular weights (M,) 1.5 x lo6 (RNA-1)and 0.5 x lo6 (RNA-2)and a coat protein of M, 3840 x lo3. The particles measure 31-35 nm in diameter and sediment at 130-135 S as a single component. Members of this group have a moderate host range, but cause characteristic necrosis, both local and systemic, often associated with severe stunting, a potential cause of reduction in crop yield. While knowledge of the biochemistry of the dianthoviruses is scarce, the amount of information is increasing. This review assesses current knowledge of the dianthoviruses, emphasizing their unique bipartite genome strategy.

11. DIANTHOVIRUSES The group consists of three viruses at the present time: the type member, carnation ringspot virus (CRSV), red clover necrotic mosaic virus (RCNMV), and sweet clover necrotic mosaic virus (SCNMV). However, it is quite reasonable to expect that the number of members of the dianthovirus group will increase in the near future as our understanding and interest in this group increase.

A . Carnation Ringspot Virus Kassanis (1955) first described CRSV isolated from Dianthus caryophyllus in the British Isles. A virus reported to cause anjermozaiek 251

Copyright 0 1987 by Academic Press, Inc. All rights of reproduction in any form reserved.

258

C. HIRUKI

in the Netherlands (Noordam et al., 1951) is considered to be CRSV (Kassanis, 1955). Since then the virus has been found in the United States (Brierley and Smith, 1957),Denmark (Kristensen, 19571, Canada (Kemp, 19641, France (Devergene and Cardin, 1967),East and West Germany (Casper, 1976; Richter et al., 19781, Australia (Sutton, 1974), New Zealand (Bennet and Milne, 1976), USSR (Kilevits, 1976; Rybalko and Kharuta, 1978), and Finland (Bremer and Lahdenpera, 1980). Its widespread distribution is suspected in other areas of the world where carnations are grown. The occurrence of CRSV strains and comparative studies of their physicochemical properties have been reported (Tremaine and Ronald, 1976). CRSV is commonly found in a mixed infection with carnation mottle virus from which it can be readily distinguished by indicator plant tests using Gomphrena globosa and Dianthus barbatus (Kemp, 1964; Hollings and Stone, 1965).

B . Red Clover Necrotic Mosaic Virus Musil(1969a) first reported the occurrence of RCNMV infecting red clover (Trifolium pratense) in Czechoslovakia. The virus was distributed in hilly regions in north Slovakia, central Bohemia, and Moravia (Musil, 1981). Different isolates or strains of RCNMV have been reported from Canada (Ragetli and Elder, 1977; Rao and Hiruki, 1985), Czechoslovakia (Musil, 196913; Musil and Gallo, 1982a1, Great Britain (Hollings and Stone, 1974; Frame et al., 1976; Bowen and Plumb, 1979; Car, 1979), Australia (Lyness et al., 1981; Gould et al., 1981), New Zealand (Forster, 1981), Sweden (Gerhardson and Lindsten, 19731, Ireland (Anonymous, 1980), and Poland (Kowalska, 1974; Blaszczak and Micinski, 1980; Blaszczak, 1981). C. Sweet Clover Necrotic Mosaic Virus SCNMV was isolated in central Alberta, Canada, in 1979, from sweet clover with necrotic mosaic symptoms (Hiruki et al., 1981, 1984~). The distribution of the virus is limited t o Alberta (Hiruki, 19861, and a serologically distinguishable strain was isolated from alfalfa growing in the same area (Inouye and Hiruki, 1985).

111. DISEASESCAUSEDIN PLANTS Symptoms of CRSV infection in carnation and D . barbatus (Sweet William) include leaf mottle, ringspots, stunting, and distortion, sometimes with leaf-tip necrosis. In such plants, the production of distorted

THE DIANTHOVIRUSES

259

flowers is not uncommon (Hakkaart, 1964). In experimental infection of D . barbatus, the primary symptoms are localized chlorotic spots and necrotic flecks, often with necrotic rings and ringspots, about 4-7 days after inoculation. Systemic symptoms that develop later are chlorotic mottle and streaking and irregular necrotic areas (Hollings and Stone, 1970). The frequent occurrence of CRSV in orchard trees and weeds, very significant epidemiologically, has been documented from West and East Germany. Kegler et al. (1977) isolated a virus later identified as CRSV (Richter et al., 1978) in pear trees with stony pit symptoms. Fritzsche and Schmelzer (1967) found the same virus in naturally growing Stellaria media and reported its transmission by nematodes for the first time. Rude1 et al. (1977) also found S. media plants naturally infected by CSRV in vineyards. Following the report of Casper (1976) on the occurrence of CSRV in a plum tree, detailed investigations were made on the virus isolated from different cultivars of apple, pear, and sour cherry (Kleinhempel et al., 1980). Disease symptoms of apple plants infected by CRSV are not very specific as the plants are often subject to multiple infections with other viruses. When the tests were positive for CRSV, the apple cultivars ‘Clivia,’ ‘Golden Delicious,’ and ‘Spartan’ as well as ‘Kola Crab’ and ‘Spy 227,’ also had apple Spy decline syndrome. In one case CRSV was isolated from ‘Spy 227’ with symptoms of pear red mottle (Kegler et al., 1977). One of the possible reasons that CRSV has not been reported from elsewhere is that the virus occurs in fruit trees in very low concentrations or does not spread to other parts from infected tissue to become fully systemic. Furthermore, coupled with the presence of inhibitory system(s) or virus inactivating system(s), the detection of CRSV is often very difficult. In apple plants, detection of CRSV was successful only after concentrating the virus by ultracentrifugation and by testing it on suitable assay plants (Kegler et al., 1977). CRSV was isolated from pear trees showing symptoms of pear stony pit (Kegler et al., 1977). CRSV was also isolated from the sour cherry cultivar ‘Schattenmorelle’ which had cherry decline syndrome. However, whether CRSV is involved in the development of cherry decline is uncertain. To satisfy Koch’s postulates, experiments on the return inoculation of CRSV from infected herbaceous plants to healthy young fruit tree seedlings were conducted using partially purified and concentrated preparations of known virus isolates. One year after inoculation CRSV was detected by Chenopodium quinoa tests from the inoculated plants of ‘Spy 227,’ ‘Kola Crab,’ Pyronia ueitchii, and Malus coronaria, of which only M.coronaria developed detectable symptoms of a mild epinasty (Kegler et al., 1977). CRSV has been isolated from

260

C. HIRUKI

plants with symptoms of apple Spy decline, pear stony pit and/or pear red mottle, but its causative relationships with these diseases have not been established. RCNMV is readily sap-transmissible to a relatively wide range of herbaceous plant species including nonleguminous plants (Musil, 1969a; Gerhardson and Lindsten, 1973; Ragetli and Elder, 1977). In naturally infected and in inoculated red clover, severe mottle, distortion, and necrotic blotches appear in the leaf tissue. In the field, infection of red clover is frequently associated with various degrees of stunting, together with chlorotic and necrotic spots and leaf deformation which are visible only in cool seasons. These symptoms are generally masked in summer. The natural occurrence of RCNMV is known in red clover (Musil, 1969a; Gerhardson and Lindsten, 1973; Hollings and Stone, 1974; Ragetli and Elder, 1977; Blaszczak and Micinski, 1980), sweet clover (Musil, 1969a1, white clover (Lyness et al., 1981; Forster, 19811, and alfalfa (Lyness et al., 1981). Natural hosts of SCNMV are sweet clover and alfalfa (Hiruki et al., 1984c; Inouye and Hiruki, 1985). Other leguminous species growing in proximity to SCNMV-infected sweet clover were free from virus infection upon testing by ELISA and infectivity tests (Hiruki, 1986). The virus occurs naturally on both Melilotus officinalis and Melilotus alba, causing ringspot and systemic veinal necrosis associated with mosaic. Consequently, infected plants generally display severe stunting. The virus causes mild stunting without severe systemic necrosis on alfalfa (Medicago satiua).

IV. PHYSICAL, CHEMICAL, AND BIOCHEMICAL PROPERTIES Physical, chemical, and biochemical properties of the members of the dianthovirus group are very similar and, therefore, only a general description will be given here. Any specific features of individual viruses will be treated separately when it is suitable to do so.

A . Virus Purification All three viruses of the dianthovirus group are very stable in uitro and reach high concentration in infected plants. Consequently there are no serious problems associated with purification when ordinary precautions are exercised. Some of the representative procedures are described in the following sections.

THE DIANTHOVIRUSES

261

1 . Carnation Ringspot Virus There are several procedures developed in different laboratories. Hollings and Stone (1965) recommend the use of systemically infected Nicotianu clevelandii leaves as a virus source for a high yield. Infected tissues were minced in phosphate buffer (pH 7.6) containing 0.1% thioglycolic acid (w/v U1.25). Extracted sap was mixed with n-butanol to 8.5% of the total volume and the mixture stored overnight at 2°C. The virus was then separated by differential centrifugation (108,000g for 90 minutes followed by 10,000 g for 20 minutes). Kalmakoff and Tremaine (1967) and Tremaine and Ronald (1976) isolated CRSV by the following method. The virus was increased in cowpea (Vignu unguiculatu cv. Blackeye). About 10 days after inoculation, the leaves were harvested and frozen. After clarification of the sap at pH 5, host components were removed by treatment with ethanol, 20% by volume of the sap. The virus was precipitated with polyethylene glycol 6000 followed by differential centrifugation. Henriques and Morris (1979) applied Bockstahler and Kaesberg’s procedure (1962) for purifying CRSV from infected N . clevelandii. Kuhne et al. (1983) purified CRSV from sap of Phaseolus vulgaris cv. Pinto which was harvested 4-6 days after inoculation. Infected primary leaves, 100 g, were homogenized in 0.01 M potassium-sodium phosphate buffer in the presence of 0.001 M MgSO,, pH 5.0. The sap was clarified by mixing with 100 ml chloroform. After brief centrifugation the aqueous phase, pH 5.0, was incubated at 4°C for 1hour. The virus was pelleted at 80,000 g for 120 minutes, and at 110,000 g for 90 minutes after removing insoluble material by low-speed centrifugation. The resulting virus suspension was dialyzed and was further purified by sucrose density gradient centrifugation ( 1 0 4 0 % w/v in the phosphate buffer) for 180 minutes at 65,000 g at 5°C. The virus fraction was subjected to gel filtration on Epidex B2. For isolation of CRSV from naturally infected fruit trees as well as experimentally infected P . vulgaris, Kegler et al. (1977) applied the procedure of Steere (1956). Extraction buffer was 0.01 M Tris-HC1 containing 0.05 M MgSO,, 0.1% Na, SO,, 0.1% ascorbic acid, and 1%nicotine, pH 8.5.

2. Red Clover Necrotic Mosaic Virus The virus is readily purified from systemically infected N . clevelandii or from inoculated leaves of P . vulgaris (Hollings and Stone, 1974; Hollings et al., 1977; Gould et al., 1981).Systemically infected leaves of N . clevelandii plants were harvested 18-21 days after inoculation (710 days with P . vulgaris),homogenized in 0.1 M phosphate buffer, pH

262

C. HIRUKI

7.4, containing 0.1%thioglycolic acid (2 ml buffer/g tissue). The resultant sap preparation was emulsified with an equal volume of 1:l mixture of chloroform and butanol. After one or more cycles of differential centrifugation (10,000 g for 10 minutes; 78,000 g for 60 minutes), the final pellets were resuspended in a small amount of 0.02 M phosphate buffer, pH 7.4, or distilled water. Virus yields up to 50 mg virus/100 g leaf tissue were obtained (Gould et al., 1981). Under high glasshouse temperatures during summer, Nicotiana spp. often gave low yields of the virus; plants kept in growth chambers at 20°C and 16 hours/day illumination at a light intensity of about 5000 lux consistently gave high virus yields (Hollings and Stone, 1977). Further purification was achieved by sucrose density gradient centrifugation. Satisfactory purification was also achieved by applying 1.0-1.5 ml of partially purified preparation to columns (85 X 1.5 cm) of controlled pore glass beads (70 mm pore size) and eluting the virus in the void volume by adding 0.04 M phosphate buffer (Barton, 1977). Marcinka et al. (1969) applied the following procedure for RCNMV purification. Phaseolus vulgaris cv. Saxa plants were harvested 7 days after inoculation at about 22"C, and were ground in 0.007 M phosphate-buffered saline, pH 7.2 (PBS) (1 ml PBS/g leaf tissue). Extracted sap was shaken for 15 minutes with chloroform at a 1:l volume ratio. After centrifugation at 4000 g for 30 minutes, the virus was pelleted from the aqueous phase at 90,000 g for 120 minutes. The pellet was resuspended in PBS, left overnight at 4"C,and clarified at 4000 g for 30 minutes and the supernatant portion was kept for 3 days at 4°C. After removing the precipitate by another low-speed centrifugation, the virus was pelleted at 90,000 g for 120 minutes. For further purification the partially purified virus preparation was subjected to sucrose density gradient centrifugation at 37,000 g for 5 hours at 4°C. Morris-Krsinich et al. (1983) used the following simplified procedure. Inoculated leaves of N . clevelandii were harvested 5-7 days after inoculation and were ground in 3 volumes of 0.5%K,HPO, and 0.5%bentonite solution (pH 8.6) containing 0.2%thioglycolic acid. The extracted sap was centrifuged at 10,000 g for 10 minutes and the resulting supernatant centrifuged at 78,000 g for 90 minutes. Pelleted virus, suspended in 1% phosphate buffer, pH 7.0, was layered onto sucrose gradients that had been prepared by freezing 25%sucrose in phosphate buffer and thawing at room temperature for 90 minutes before use. The virus, fractionated from a single band after centrifuging the gradients at 130,000 g for 120 minutes, was concentrated by centrifugation at 368,000 g for 60 minutes and resuspended in water.

THE DIANTHOVIRUSES

263

3. Sweet Clover Necrotic Mosaic Virus

Hiruki et al. (1984~)used the following procedure for the purification of SCNMV and found that it was also useful for purifying other dianthoviruses. Infected P . vulgaris cv. Red Kidney or N . clevelandii was homogenized in a Waring blender in cold 0.1 M phosphate buffer (pH 7.0), containing 0.5% ascorbic acid, at a 1:2 ratio (w/v). The homogenate was clarified by chloroform-butanol (1:l) and the virus was precipitated by polyethylene glycol 6000 and sodium chloride at 8% and 0.4% (w/v), respectively, followed by two cycles of differential centrifugation (108,000 g for 120 minutes and 9000 g for 10 minutes). For further purification, linear 10-40% sucrose gradients were prepared in 0.025 M phosphate buffer, pH 7.0. A single band, formed during centrifugation at 64,000 g for 90 minutes, contained the virus. An average yield of 30 mg virus/lOO g of infected tissue was obtained.

B . Electron Microscopy of Purified Virus Preparations Particle morphology of the various dianthoviruses in purified preparations is very similar when viewed in the electron microscope after negative staining treatment. In both RCNMV and SCNMV, particles stained in uranyl acetate have rounded outlines and granular surfaces with a diameter measuring 35 nm (Fig. 1A). Slightly lower values ranging from 29 to 31 nm have also been reported (Hollings and Stone, 1965; Bowen and Plumb, 1979; Kuhne and Eisbein, 1983). It is known that virus particles from the same preparations measure about 31 nm when stained with 2% phosphotungstate (Fig. 1B; Kalmakoff and Tremaine, 1967; Gould et al., 1981; Hiruki et al., 19844. The particles of the dianthoviruses are morphologically indistinguishable from those of the tombusviruses and a number of spherical viruses which have not yet been taxonomically classified, such as galinsoga mosaic virus (Hatta et al., 1983).However, their identification can be serologically established by immunosorbent electron microscopy (Chen et al., 1984; Gould et al., 1981). Three strains of CRSV referred to as A, N, and R display strain-specific aggregation properties which can be studied by electron microscopy (Tremaine et al., 1976). CRSV-A can form clusters of 12 virus particles arranged in an icosahedral symmetry which may be linked to form larger clusters containing 23, 34,45, or 56 virus particles. CRSV-N and CRSV-R form another kind of aggregate, a two-dimensional net, which is temperature dependent, the latter virus requiring higher concentrations or higher temperatures. The diameter of the 12-particle aggregate measures two and one-half to

264

C.HIRUKI

FIG. 1. Electron micrographs of purified SCNMV particles. (A) Stained with 2% uranyl acetate. (B) Stained with 2% potassium phosphotungstate. Bars represent 100 nm. (From Hiruki et al., 1984c.)

THE DIANTHOVIRUSES

265

three times the size of the monomer, depending on the axis of the aggregate measured and on the amount of stain found in the electron micrograph. A similar phenomenon is not known for other dianthoviruses.

C . Properties of Virus Particles Dianthovirus particles sediment at about 135 S as a single component (Hollings and Stone, 1970,1977).The sedimentation coefficient is influenced, with a range of 125-140 S, by the virus concentration, the age of the virus preparations, the method of virus purification (Hollings and Stone, 19651, and the pH of virus preparations (Kuhne and Eisbein, 1983; Hiruki et al., 1984~).Certain strains of CRSV are known to swell at pH 7.0 and have a reduced sedimentation coefficient (Tremaine and Ronald, 1976). Both CRSV and RCNMV have a buoyant density in CsCl of about 1.36 g/cm3 (Hollings and Stone, 1977; Gould et al., 1981; Kuhne and Eisbein, 1983). CRSV particles have an M, of about 7 x lo6 as determined by physical methods (Kalmakoff and Tremaine, 1967). Assuming that a dianthovirus particle consists of 180 protein subunits of 38,000 Da and a single molecule RNA-1 of 1.5 x lo6 Da or three molecules of RNA-2 of 0.5 x lo6 Da, a theoretical molecular mass for an intact virion is 8.3 x lo6 Da. A T=3 symmetry (Casper and Klug, 1962) is suggested by the presence of 180 protein subunits. A subunit protein contains approximately 347 amino acid residues (Kalmakoff and Tremaine, 1967). The amino acid composition differs for the A, N, and R strains of CRSV (Tremaine et al., 1984). Particles of CRSV-N swell above pH 7.0 in the absence of divalent cations. Particles are readily disrupted at pH 7.0 at room temperature in the presence of low concentrations of the anionic detergent SDS and dissociate into RNA and protein. Particles of CRSV-N are stabilized by pH-dependent protein-protein interactions and by RNA-protein interactions as well as Mg2+ (Tremaine and Ronald, 1976). These are characteristics of the bromoviruses which Kaper (1972,1973) grouped in the cucumber mosaic virus profile. CRSV-A and CRSV-R are more stable in SDS than CRSV-N (Tremaine and Ronald, 1976). The diffusion coefficient value (D20,wx 10-7 cm2/sec) for CRSV has been reported to be 1.48 (Kalmakoff and Tremaine, 1967). The partial specific volume of CRSV was calculated to be V = 0.693 ml/g from percentage nucleic acid, amino acid, and nucleotide composition (Kalmakoff and Tremaine, 1967). The composition of RNA in CRSV calculated from recovered nucleotides and protein was 20.48% (Kalmakoff and Tremaine, 1967). An extinction coefficient of 6.46 cm2/mg at 260 nm was calculated from the absorbance of the purified CRSV before hydrolysis

266

C. HIRUKI

and the yield of amino acids and nucleotides after hydrolysis (Kalmakoff and Tremaine, 1967). Physicochemical properties of the dianthoviruses are summarized in Table I.

D . Properties of Coat Protein and Nucleic Acid For electrophoretic analyses, nucleoprotein of the dianthoviruses can be dissociated by heating to 100°C for a few minutes in 30-50 mM Tris-HC1 buffer (pH 6.8) containing 1.8-2.0% SDS and 1.0-1.8% 2mercaptoethanol (Morris-Krsinich et al., 1983; Okuno et al., 1983). The molecular weight of the coat protein of CRSV, consisting of 347 amino acid residues, was chemically estimated to be 38,000 (Kalmakoff and Tremaine, 1967). The values estimated by polyacrylamide gel electrophoresis for SCNMV and RCNMV were in the range of 37,500 to 41,000 (Fig. 2A, Hiruki et al., 1984c; Rao and Hiruki, 1985). A translation product of RCNMV RNA-1 in rabbit reticulocyte lysate was reported as having a molecular weight of 39,000, and was shown to be coat protein (Morris-Krsinich et al., 1983). There is no evidence for more than one coat protein species in dianthoviruses. Infective RNA from CRSV was isolated by sucrose density gradient centrifugation from dianthovirus particles after dissociation by adding 1M Tris containing 0.2 M disodium ethylenediaminetetraacetate (EDTA), and 10% SDS to a virus preparation at a 1 : l O ratio (v/v) (Dodds et al., 1977). Intact RNA was also isolated from all three member viruses of the dianthovirus group, including their strains by the following method (Okuno et al., 1983). Purified virus preparations were mixed with an equal volume of extraction medium consisting of 0.05 M Tris-HC1, pH 7.8, 0.005 M EDTA, 0.05 M NaC1, 1%SDS, and 0.1% bentonite. Water-saturated phenol was added to the mixture and shaken vigorously for 15 minutes at room temperature. The subsequent aqueous phase was treated once with phenol and three times with ether. Ethanol-precipitated RNA was suspended in sterile water containing 18%glycerol. Two RNA species were found in such preparations with molecular weights ranging from 1.35 x lo6 to 1.55 X lo6 for RNA-1 and 0.5 x lo6 to 0.6 x lo6 for RNA-2 by polyacrylamide gel electrophoresis using nondenaturing conditions. RNA-1 and RNA-2 of several strains of RCNMV had molecular weights of about 1.5 x lo6 and 0.5 x lo6, respectively, under denaturing conditions (Fig. 2B, Gould et al., 1981; Rao and Hiruki, 1985; Osman et al., 1986). The base compositions of unfractionated RNA of CRSV were G = 25.85 2 0.72%, A = 27.26 2 0.55%,C = 22.68 .t 0.23%, U = 24.18 0.68%. The comparative base analysis of RNA-1 and RNA-2 has not

*

267

THE DIANTHOVIRUSES TABLE I

PROPERTIES OF DIANTHOVIRUSEW~ Property

CRSV

Diameter (nm)

31; PTA (10,12,16)

M, virion

RNA-2 M, protein

7.07 x 106 (10) G:A:C:U x 26:27:23:24 (10) 20 (10) 1.5 X lo6 (3) 0.5 x lo6 (3) 38 x 103 (10)

Protein

ao (5)

RNA base ratio RNA (%)

M, RNA-1

Buoyant density

(cm2/mg) EzsolEzso

128 (11) 129 (12) 132 (10) 135 (8) 1.358 g/ml in CsCl (12)

6.46 (10) 1.61 (10)

Isoelectric point Diffusion coefficient (Dzo,w x 10-7 cmZ/sec)

RCNMV

SCNMV

27; PTA (9) 30; FTA (2,5,9) 31; PTA (1,6) 34; UA (6)

33; PTA (7) 35; UA (7)

1.5 x 106 (6) 0.5 x 106 (6) 40 x 103; H (9)

1.4 x 106 (7) 0.6 X 106 (7) 38 x 103 (7)

125 (1) 135 (6) 139 (2)

126 (7)

41 x 103; sw (9) 41 X 103; C (15)

1.37 g/ml in CsCl (6) 1.66 g/ml in CsCl (9) 1.72 g/ml in CsCl (2) 5 (6) 1.66 (9) 1.72 (2) 5.05, A (4) 4.83, B (4) 4.61, C (4)

1.80 (7)

1.48 (10)

Abbreviations: PTA, potassium phosphotungstate; UA, uranyl acetate; H, English isolate; SW, Swedish isolate; C, Canadian isolate; A, B , and C refer to serotypes. Numbers in parentheses are reference numbers: (1)Blaszczak and Micinski (1980); (2) Bowen and Plumb (1979); (3) Dodds et al. (1977); (4) Gallo and Musil (1984); (5) Gerhardson and Lindsten (1973); (6) Gould et al. (1981); (7) Hiruki et al. (1984~); (8) Hollings and Stone (1970);(9) Hollings and Stone (1977);(10) Kalmakoff and Tremaine (1967); (11)Kleinhempel et al. (1980);(12) Kuhne and Eisbein (1983); (13) Marcinka et al. (1969); (14) Ragetli and Elder (1977); (15) Rao and Hiruki (1985); (16) Tremaine (1961).

268

C.HIRUKI

FIG.2. Electrophoretic analysis of coat protein and RNA of RCNMV isolates. (A) Coat proteins from RCNMV-AUS (lane 1)and RCNMV-C (lane 2) in 9% discontinuous SDS-polyacrylamide gel system. Bands in lane 3 are molecular weight markers containing (from top) bovine plasma albumin (66,000), ovalbumin (45,000),trypsinogen (24,000), and tobacco mosaic virus coat protein (17,500).(B) RNA components on a 3% polyacrylamide gel containing 7 M urea. Lanes 1-3: RNAs from (1)RCNMV-AUS, (2) RCNMV-C, and (3) brome mosaic virus. (From Rao and Hiruki, 1985.)

been reported. The nucleic acid homology study of RNA-1 and RNA-2 of RCNMV using the RNA-cDNA hybridization procedures (Taylor et al., 1976) showed that RNA-1 and RNA-2 have no sequence homology between them (Gould et al., 1981). Using unseparated RNA of a n English isolate of RCNMV (RCNMV-H) and a Czechoslovakian isolate TpM-34 of RCNMV (RCNMV-341, Osman et al. (1986) distinguished the RNA species of each isolate by Northern hybridization. Under high stringency conditions at 55"C, the cDNAs hybridized only with their homologous RNA.

THE DIANTHOVIRUSES

269

V. SEROLOGICAL STUDIES In studying the serological properties, workers reported considerable differences between certain RCNMV isolates and the respective antisera (Musil, 1969b; Gould et al., 1981). While Musil and Gallo (1982a) distinguished three serotypes among them, recent results on a large collection of RCNMV isolates from different parts of the world have suggested that they can be grouped into four serotypes (Rao et al., 1985). Similar serological diversity may exist among naturally occurring CRSV and SCNMV. However, the number of isolates studied is rather limited and our understanding of their serological relationship must await future elucidation. A . Polyclonal Antibodies Dianthoviruses are generally good immunogens. Rabbits immunized by one intravenous followed by two intramuscular injections with CRSV immunogen yielded antisera with specific titers in precipitin tube tests of 1/4,096 to 1/32,768 (Hollings and Stone, 1970). The virus antigen reacts well with antiserum both in precipitin tube and in gel diffusion tests and can be detected in crude sap of carnation by gel diffusion tests (Kemp, 1964). In both the tube test (Hiruki et al., 1984~) and gel diffusion tests (Tremaine et al., 1976, 1984; Gould et al., 1981; Hiruki et al., 1984c; Rao et al., 1985) polyclonal antibodies were used for studying serological relationships of dianthoviruses and strain differentiation. Specific precipitates in tubes are granular and a single band of precipitate forms in gel diffusion tests between RCNMV and its antiserum (Fig. 3). An extremely remote, but relatively consistent, cross-reaction against CRSV antiserum with viruses of the tymovirus group was reported (Berks and Querfurth, 1972) but not confirmed by other workers. The use of enzyme-linked immunosorbent assay (ELISA) for the detection of plant viruses is well documented (Clark and Adams, 1977; Clark, 1981). Working on SCNMV, Hiruki et al. (1984~)found that 1 pg of coating y-globulinlml and 1/3000 dilution of the conjugate were optimal to detect virus up to 10 ng/ml. However, in double-antibody sandwich ELISA (direct ELISA), the immunoglobulin (Ig)of each test serum must be purified and coupled to enzyme. Furthermore, the high specificity of the test can preclude detection of related strains of the same virus (BarJoseph and Salomon, 1980; Koenig, 1978). The use of indirect ELISA makes it possible to evaluate strain relationships of the

270

C. HIRUKI

FIG. 3. Serological reaction of dianthoviruses in immunodiffusion plate. Central well contained antiserum to RCNMV-C;peripheral wells were charged with antigens (1) RCNMV-E (England), (2) RCNMV-34 (Czechoslovakia),(3) RCNMV-AUS (Australia), (4) RCNMV-SW (Sweden), (5) RCNMV-C (Canada), and (6) RCNMV-48 (Czechoslovakia). (From Rao and Hiruki, 1985.)

virus in question (Koenig, 1978). In indirect ELISA a n enzyme-labeled anti-Ig is used as a second antibody to detect the antigen-antibody complex on the solid phase (Crowther and Abu-El Zein, 1979). This eliminates the need for specific enzyme conjugates for each antigen to be tested and lessens problems associated with extreme specificity of direct ELISA. The relative sensitivities of antigen detection were compared for direct and indirect ELISA using purified CRSV (Lommel and Morris, 1982). Indirect ELISA was capable of detecting the virus at a concentration of 1.6 pg/ml in contrast to the detection limit by direct ELISA of 8 pg/ml, indicating that indirect ELISA had a greater sensitivity than direct ELISA for the detection of CRSV (Fig. 4). Direct ELISA was used in detecting SCNMV in field specimens and the results were in good agreement with those from sap inoculation of indicator plants (Hiruki, 1986). However, in the study of serological relationships between dianthoviruses, direct ELISA failed to detect any reaction between a conjugate containing rabbit anti-RCNMV-C antibodies and other dianthoviruses (Hiruki et al., 198413). Highly purified virus preparations were used to provide a series of twofold dilutions for each virus. The results indicated the high specificity of direct ELISA with polyclonal RCNMV-C antibody because when heterologous antibodies were used both for coating the plates and in the enzyme conjugates, only the homologous antigen could be detected. In similar tests antisera prepared to RCNMV-AUS, RCNMV-SW, SCNMV, and CRSV also showed high specificity (Hiruki et al., 1984b). In contrast, indirect ELISA (Hiruki et al., 1984b) was less discriminating than the direct

271

Virus ng/ml

FIG. 4. A comparison of the relative sensitivity of direct (sandwich) and indirect ELISA. Detection of purified CRSV by direct ELISA (0) and by indirect ELISA (0). Each point is the mean of five replications with a maximum standard deviation of 0.01. (From Lommel and Morris, 1982.)

ELISA and was used with polyclonal RCNMV-C (Rao and Hiruki, 1985) and CRSV antibodies with homologous and heterologous antigens (Fig. 5 ) . Purified preparations of RCNMV-AUS reacted with both CRSV and RCNMV-C antisera. These tests also showed the distant serological relationship between RCNMV-C and SCNMV. These results support the idea that the dianthovirus group consists of viruses that differ considerably in their serological relationship. CRSV is only distantly related to the other two members, the relationship being detected only by indirect ELISA (Figs. 5 , 8; Hiruki et al., 1984b). In quantitative comparisons of the direct and indirect ELISAs on pure CRSV and crude plant extract preparations (Lommel and Morris, 19821, the slopes of the plots of A,,, nm versus sap dilution and pure virus concentration were similar in direct ELISA (Fig. 6A). This result indicates that the presence of host proteins in the direct test does not affect quantitative detection of CRSV. In contrast, results with indirect ELISA showed a depression of absorbance at the lowest sap dilution (highest concentration of host proteins) and an apparent enhancement of absorbance at the highest dilution compared to samples containing only pure CRSV (Fig. 6B).Thus, indirect ELISA would not be suitable for accurate quantitation of CRSV in crude plant extracts.

C. HIRUKI SCNMV

4

CRSV

FIG.5. Histograms showing the reaction of RCNMV-C and CRSV antibodies with some dianthoviruses in indirect ELISA on plates not coated with antibodies. (O), RCNMV-C antibody; (HI, CRSV antibody. The virus concentrations (ng/ml) were (a) 2000, (b) 1000, (c) 500, (d) 250, (e) 125, (062.5, and (g) 31. (From Hiruki et al.,1984b, typographic error in unit corrected.)

B . Monoclonal Antibodies The usefulness of antibodies was greatly facilitated by the discovery by Kohler and Milstein (1975) of a technique for producing monoclonal antibodies. The principle of monoclonal antibody production is well known (Pollock et al., 1984; Sander and Dietzgen, 1984; Carter and ter Meulen, 1984). A short-lived antibody-producing cell isolated from an immunized animal is fused with a myeloma cell of sustained longevity to create a permanent cell system (hybridoma) for highly specific antibody production. Hybridomas can be frozen and recovered or injected into mice to produce ascites fluid containing large amounts of the monoclonal antibody of highly increased reactivity. SCNMV is one of the first plant viruses against which monoclonal antibodies have been generated. Hiruki (1982) reported the production of monoclonal antibodies by fusing mouse myeloma cells P3-X63-AgU1with spleen cells derived from BALB/c mice immunized with purified SCNMV. Twentyone monoclones which secreted monoclonal antibodies of high titers against SCNMV were selected for characterization (Hiruki et al., 1984a; Table 11.) The class and subclass of immunoglobulins of the monoclonal antibodies secreted by these cultures were determined to be IgG,, for 15 clones and IgG, for 6 clones. The ability of monoclonal antibodies to differentiate between di-

273

THE DIANTHOVIRUSES

sqoo

lop0

-

Virus ng/ml 290 4.0 A Sandwich

9

pwevirus virus 6 tissue 0 infected tissue A

m

A

\

d

E

C

v)

0

t I

al

B Indirect

0

C

a

: u)

n

U

2

1

B O-0

5

6

0 -

25

125

6 2S

3125

Tissue Dilution

FIG.6. CRSV by (A) direct (sandwich) ELISA and (B)indirect ELISA a t different stages of purity. The following samples were serially fivefold diluted in the plates: (A) pure CRSV at 5 Fg/ml, (H)CRSV at 5 pg/ml with carnation protein added a t 200 pg/ml, (0)CRSV-infected carnation tissue, and (0) healthy carnation tissue. Each point is the mean of five replications with a maximum standard deviation of 0.01. (From Lommel and Morris, 1982.)

274

C. HIRUKI TABLE I1 IMMUNOGLOBULIN ISOTYPINC A N D HOMOLOGOUS ASSAYSWITH MONOCLONAL ANTIBODIESPRODUCED AGAINST SWEET CLOVER NECROTIC MOSAIC VIRUS^ Mouse ascites fluid

Hybridoma 1A2 2A2 2c2 4A2 4Dl 4G1 4H5 5G4 5G5 6B2 6E6 6F1 6H3 7B3 7c3 7E2 9A2 9A3 lOBl llGl 12A1

Ig isotype

Medium supernatant, RPHI 2048 256 4 32 2 64 2 32 16 32 16 4096 512 16 16 512 128 32 64 128 4

RPHI

PHA

Mouse 1

Mouse 2

Mouse 3

409,600 409,600 3,200 12,800 200 1,600 400 3,200 3,200 6,400 400 12,800 1,600 3,200 6,400 409,600 12,800 3,200 25,600 3,200 800

409,600 12,800 1,600 1,600 400 3,200 200 51,200 6,400 6,400 800 51,200 6,400 3,200 6,400 409,600 1,600 3,200 3,200 6,400 3,200

40,000h 40,960,000 64,000 320,000 640,000 64,000 4,000 16,000 160,000 32,000 4,000 1,280,000 32,000 256,000 32,000 16,000 1,280,000 16,000 32,000 128,000 16,000

a As determined by reverse passive hemagglutination inhibition (RPHI) or passive hemagglutination (PHA) tests. (From Hiruki et al.,1984a). b Figures represent the reciprocal of the highest dilution in RPHI or PHA test.

anthoviruses was measured by indirect ELISA (Fig. 7). The results clearly showed the diversity in the reactivity of different monoclonal antibodies with different dianthoviruses. Although CRSV had not been proven to be serologically related to other dianthoviruses either by immunodiffusion tests (Ragetli and Elder, 1977; Hiruki et al., 1984c) or by immunosorbent electron microscopy (Chen et al., 1984) using polyclonal antibodies, it reacted positively against SCNMV monoclonal antibodies 1A2, 2C2, and 6B2 (Hiruki and Figueiredo, 1985). A range of reactivities among members of the dianthovirus group was found against SCNMV monoclonal antibody 1A2 (Fig. 8).In another test, RCNMV-AUS and RCNMV-E could not be differentiated by using 6B2. At present, except for CRSV, the chemical structure of

275

THE DIANTHOVIRUSES

1.4 1.2

1.0 Y (


0.8 0.6

0.4 0.2

IC-9 10-7 10-5 10-3 Monoclonal antibody dilution

FIG.7. A comparison of monoclonal antibodies from clones 1A2,2C2, and 6B2 reacting specifically with SCNMV.(From Hiruki and Figueiredo, 1985.)

0.7 0.6

0.5 y1

0

2'

0.4 0.3 0.2 0.1

0.5

1

2

4

(ugfml)

Ant fgen

FIG.8. Comparative binding of dianthoviruses with a monoclonal antibody from clone 1A2 tested by indirect ELISA. (From Hiruki and Figueiredo, 1985.)

276

C. HIRUKI

the coat protein of dianthoviruses has not been elucidated. Studies of amino acid sequences in the coat protein of other dianthoviruses and the selection of single point mutants will enhance the usefulness of monoclonal antibodies in our understanding of the antigenic structure of dianthoviruses. VI. HOSTRANGEAND SYMPTOMATOLOGY The relative value of host range and symptomatology in the study of virus properties has diminished over the years. Nevertheless they provide useful information in assessing interrelationships among member viruses when considered together with other characteristics of the viruses (Hamilton et al., 1981). It should be emphasized that our understanding of virus and host interactions, which involve intricate genetic systems of both virus and host plant, is meager. Specific knowledge of host range and symptomatology in relation to the function of certain selected genomes will be useful in analyzing the results of such virus-host genome interactions. The range of both cultivated plants and weeds that become naturally infected with dianthoviruses appears to be wider than had generally been thought. For example, CRSV, initially isolated from D . barbatus (Kassanis, 1955), was later reported to occur naturally on fruit trees such as plum (Casper, 19761, apple, pear, and sour cherry (Kegler et al., 1977; Kleinhempel et al., 1980). Weed species growing between the infected fruit trees were also found to be naturally infected by CRSV. These include Urtica wens (Kleinhempel et al., 19801, Poa annua (Kleinhempel et al., 1980), and S . media (Richter et al., 1978; Fritzsche et al., 1979; Kleinhempel et al., 1980).

A. Natural Isolates CRSV has been isolated mainly from different cultivars of Dianthus spp. in various parts of the world. However, there are no comprehensive studies on the natural isolates of different origins as to their host range and symptomatology. The only distinctive isolates that have been reported are three biochemically selected isolates (A, N, R) of CRSV (Tremaine et al., 1976). Several natural isolates of RCNMV have been reported from different parts of the world such as Czechoslovakia (Musil, 1969a,b), Poland (Kowalska, 1974; Blaszczak and Micinski, 1980), Sweden (Gerhardson and Linsten, 1974), Great Britain (Hollings and Stone, 1974; Bowen and Plumb, 1979), Canada (Ragetli and Elder, 1977; Rao and Hiruki,

277

THE DIANTHOVIRUSES

19851, New Zealand (Forster, 19811, and Australia (Lyness et al., 1981).However, only a limited study was made of selected isolates for host and virus interactions under comparable growth conditions. In recent studies on two natural isolates of SCNMV conducted under comparable greenhouse conditions, it was found that a new alfalfa isolate was more pathogenic on alfalfa than the sweet clover isolate of SCNMV (Inouye and Hiruki, 1985).

B . Experimental Studies CRSV infected 62 of 96 plant species tested by manual inoculation (Hollings and Stone, 1965).Of these infected plants, systemic invasion occurred in only 16 species. At least 47 of 106 plants, in 15 species of 30 families tested, were susceptible to different strains of RCNMV (Hollings and Stone, 1977). SCNMV infected 16 of 25 species of both legumes and nonlegume plant species (Hiruki et al., 1984~). Table I11 shows representative symptom descriptions on selected TABLE I11 COMPARATIVE SYMPTOMATOLOGY OF SELECTED DIANTHOVIRUSES~.~ Plant species= Phaseolus uulgaris ‘Red Kidney’ Vigna unguiculata ‘Blackeye’ Chenopodium amaranticolor Cucumis sativus Tetragonia expansa Gomphrena globosa Nicotiana cleulandii

CRSVd

RCNMV-C‘

RCNMV-AUSf

SCNMVg

s,c,rs;sy,vn

s,c,LL;sy,y,ns

l,c,LL;sy,y,ns

r,b,rs;sy,vn

r,b,LL;sy,vn

c,LL;sy,m

r,b,LL;sy,vn

r,b,LL;sy,vn

l,c,LL,c,h

s,c,LL

l,LL,c,h

LL

l,c.LL,c,h, LL;sy,m L,ns;sy,m,ld ns;sy,m

s,LL,c,h LL NI NI

l,LL,c,h LL;sy,m r,b,LL ns;sy,m

LL LL;sy,m r,b,LL ns;sy,m

From Rao and Hiruki (1985). Coded symptom descriptions: b = brown, c = chlorotic, h = halo, 1 = large, Id = leaf distortion, L = local, LL = local lesions, m = mosaic, ns = necrotic spots, r = reddish, rs = ringspot, s = small, sy = systemic, vn, veinal necrosis, y = yellow, and NI = plants produced no symptoms and virus could not be recovered by back-inoculation to susceptible hosts. c A minimum of six plants were inoculated for each test species, and symptoms were recorded 12-15 days after inoculation. Before inoculation, all plants were kept in the dark for at least 12 hours to increase sensitivity. d Carnation ringspot virus. e Red clover necrotic mosaic virus (Canadian isolate). f Red clover necrotic mosaic virus (Australian isolate). 1 Sweet clover necrotic mosaic virus. a

b

27 8

C. HIRUKI

species useful as diagnostic, propagative, and quantitative assay plants for the dianthoviruses. VII. REPLICATION In inoculated French bean leaves or N . cleuelandzi leaf tissue kept at 22"C, RCNMV multiplies rapidly; particle numbers and infectivity reach peak values in about 2 days and change little during the next 4 days. Temperature greatly affects the rate of accumulation of RCNMV particles in leaves. The replicative forms of CRSV were studied by gel electrophoresis (Dodds et al., 1977; Henriques and Morris, 1979; Morris, 1983).dsRNAs isolated from CRSV-infectedcowpea leaves were of two major species of 3.0 x lo6 and 1.0 x 106 Da and a minor species of 1.2 x lo6 Da (Fig. 9). Unmelted dsRNA replicative form preparations were not infectious on Chenopodium amranticolor but were infectious after melting (Dodds et al., 1977).The isolation of dsRNA replicative forms which melted into ssRNA molecules of the same size as RNA-1 and RNA-2 is further proof of a bipartite genome in CRSV. Two ssRNAs recognized in melted dsRNA of 1.0 x lo6 Da were similar in size t o RNA-2 and to the minor ssRNA component that was found consistently in fresh CRSV preparations, although the role of this minor RNA component in CRSV infectivity was not determined. Translation of the bipartite genome of RCNMV (New Zealand isolate D) in rabbit reticulocyte lysate resulted in the synthesis of three major translation products of 39,000, 36,000, and 34,000 Da (MorrisKrsinich et al., 1983). The 39,000-Da translation product was shown to be RCNMV coat protein by peptide mapping after partial proteolysis and by specific immunoprecipitation with anti RCNMV y-globulins. The 36,000-Da product, of undetermined function, was also translated from RNA-1 while the 34,000-Da product appeared to be derived from RNA-2. Using genomic RNA preparations of RCNMV-AUS, CRSV, and their reassortant R1C2 (RNA-1 from RCNMV and RNA-2 from CRSV) for translation, similar results were obtained (Lommel, 1983). With regard to the smallest labeled protein of 20,000 DA, MorrisKrsinich et al. (1983) suggested that it was caused by translation of degraded genome fragments. In contrast, Lommel(1983) believed that the protein was an intrinsic translation product because it was discrete and reproducible. Regarding coat protein coded on RNA-1, the level of synthesis of the protein from sucrose gradient fractions containing less than full-length RNA-1 was more pronounced than from RNA-1. This means that a

THE DIANTHOVIRUSES

279

FIG. 9. Absorbance profiles of CRSV RNAs in 2.4% polyacrylamide gels. The samples were electrophoresed at 8 mA/gel for 2.25 hours in 0.04 M Tris, 0.02 M sodium acetate, 0.001 M disodium EDTA, 0.1% SDS, pH 7.8. The direction of electrophoresis was from left to right. The samples applied were 100 pl of (A) TMV as a reference at 0.5 mg/ml, dissociated in 0.02 M Tris, 0.001 M disodium EDTA, 1% SDS, pH 9.0; (B)CRSV at 0.2 mg/ml, dissociated in 0.1 M Tris, 0.02 M disodium EDTA, 1%SDS, pH 8.0; (C) CRSV dsRNA purified by two cycles of CF-11 cellulose chromatography; (D) CRSV dsRNA heated for 60 seconds in 0.003 M EDTA, pH 7.4, then cooled in an ice bath. (From Dodds et al., 1977.)

discrete subgenomic RNA species coding for the coat protein may be involved. Morris-Krsinich et al. (1983) suggested that the presumptive subgenomic species coding for the coat protein is encapsidated in RCNMV particles or is generated from full-length RNA-1 during the translation process. The hybridization experiments by Lommel(1983) between virus-induced dsRNA and genomic RNA demonstrated that, of the three dsRNAs produced during replication, dsRNA-1 and dsRNA-2 (3.0 x lo6 and 1.0 x lo6 Da) were replicative forms for the production of new genomic RNA-1 and RNA-2. The dsRNA-3 (1.0 x lo6 Da) was related to the production of subgenomic RNA species from RNA-1. This replication strategy involving subgenomic RNA for the coat protein coding appears to be similar to that known for a number of other plant viruses (Atabekov and Morozov, 1979; Davies, 1979). When the replicative form dsRNAs of RCNMV, CRSV, and their reassortant R1C2 were electrophoresed under comparable conditions, the subgenomic dsRNAs which code for coat protein had the same electrophoretic mobility. In contrast, the replicative form dsRNAs for

280

C. HIRUKI

RNA-2 that code for the protein of 34,000 Da had significantly different mobilities (Lommel, 1983). An interesting finding which emerged in this study was that CRSV dsRNA-3, which is the replicative form of CRSV RNA-2 (one of the donor RNAs for the R1C2 reassortant), is much smaller than the replicative form of R1C2 dsRNA-2. This fact suggests that a process of evolution of CRSV RNA-2 is in operation which is required for the establishment of the stable reassortant RlC2. On the basis of the hybridization of dsRNA and in uitro translational work (S. A. Lommel, 1983, and personal communication; MorrisKrsinich et al., 1983), a schematic model of the genome organization and translation strategy of the dianthoviruses may be presented (Fig. 10). Since recent cDNA cloning work has demonstrated the presence of an additional cistron on RNA-1 coding for a protein of 60,000 Da (S. A. Lommel, personal communication), it appears that dianthovirus RNA-1 contains at least the information for three proteins of 36,000, 39,000, and 60,000 Da, of which only the protein of 39,000 Da is immunoprecipitated by coat protein antiserum. It is assumed that the cistrons for the proteins of 60,000 and 36,000 Da are located on the 5’ end of RNA-1 and are translated from the putative translation initiation point on the 5’ end of RNA-l. The genome of RNA-1 may be divided at some stages of the virus replication into subgenomic RNA with a previously closed cistron at its 5’ end. The genome coding for the coat protein of 39,000 Da is located within RNA-1 toward the 3’ end separate from the genome for the proteins of 36,000 and 60,000 Da, probably lacking its own translation initiation point. RNA-2 is capable of serving as its own messenger RNA at least for the synthesis of the 34,000-Da protein. Its functional role, however, is not well understood. It has been shown that RNA-2, together with RNA-1, is involved in the development of certain symptoms (Okuno et al., 1983). Another strong possibility is that the 34,000-Da protein coded by RNA-2 has a function similar to the proteins of 30,00035,000 Da that are produced by many plant viruses and are thought to be involved in virion transport (Atabekov and Morozov, 1979; Davies, 1979; Rutgers et al., 1980; Leonard and Zaitlin, 1982). Recent evidence from genome analysis of RCNMV reassortants, reported by Osman et al. (1986) is supportive of this view, though further work is needed to be conclusive in this respect. If RNA-2 encodes only the virion transport protein, virus replication in single cells might require only RNA-1, because RNA-1 is large enough to encode the coat protein, a replicase and other protein(4 (Fig. 10). Analyses of the protoplasts inoculated with individual RNA species of dianthovirus should resolve

THE DIANTHOVIRUSES

281 X B

P pBRROO8 1.6kb pBRR201 1.Okb 5‘

3’, 4.3 kb

RNA-1

. . . . . . _ 1 . .

60 K

7

Replicare7

1-1

pBRROO5 1.4kb

’5 ’

3’ RNA-2 1.3ib

tr.....rrr 36K

3BK

Capsid

P CDNA

clone

Genomic RNA

Protein

34K

Movement

h n c t ion

FIG. 10. Strategy of genome expression of RCNMV-AUS. Location of restriction sites on the cDNA and organization of the RCNMV-AUS genome are shown. The exact regions responsible for coding for the 60K and 36K proteins are not conclusively determined. (After s. Lommel, personal communication.)

this hypothesis and such a study is in progress (Pappu and Hiruki, 1986).

VIII. GENOME REASSORTMENT STUDIES

A . Dianthoviruses Contain Bipartite Genome Dianthoviruses are the most recently recognized group of “bipartite genome viruses,” which contain a genome of two ssRNA species, RNA-1 (M,1.5 x 106) and RNA-2 (M,0.5 x 109, both of which are required for infectivity. A single capsid polypeptide molecule. M, 38,000-41,000, is encoded by RNA-1 (Dodds et al., 1977; Gould et al., 1981; Okuno et al., 1983). The bipartite genome viruses such as dianthoviruses are the simplest in genome segmentation and thus particularly amenable to genetic studies. Reconstruction of a new set of hybrid viruses (reassortant, after Lane, 1979) with genome segments from different origins and evaluation of their biochemical and biological functions would allow us to correlate each genome segment with a particular property. Several attempts in this direction have been quite successful and generated interesting information.

B . Infectivity To demonstrate the divided genome nature of a given virus, nucleic acid components of the virus must be purified to the extent where each

282

C. HIRUKI

of the genome components is noninfectious by itself. Demonstration of high infectivity upon mixing such noninfectious components is proof of the existence of a divided genome in the virus and of the high purity of the genome components involved. However, it must be remembered that it is often difficult to obtain the RNA preparations free from contamination of other size classes. This is especially true of larger RNA components. Gould et al. (1981) monitored by cDNA hybridization the purity of fractionated RNA-1 and RNA-2 of RCNMV-AUS separated by electrophoresis in polyacrylamide gels under denaturing conditions. Noninfectious RNA-2 was found to be contaminated with 5.1% by weight of nucleotide sequence derived from RNA-1. RNA-1 was infectious when contaminated by only 2.3% by weight of RNA-2. However, its contamination was significantly reduced (0.063%)by reelectrophoresis in polyacrylamide gel. First evidence for the bipartite genome of dianthoviruses was presented by Dodds et al. (1977) who reported that the addition of noninfectious RNA-2, relatively free from RNA-1, to RNA-1 preparations markedly increases infectivity. Gould et al. (1981) also showed that mixing RNA-1 and RNA-2 of RCNMV-AUS results in substantial increases in infectivity. Hiruki et al. (1981)found that SCNMV similarly contains a bipartite genome, indicating that the virus has a functionally divided genome and requires both RNA-1 and RNA-2 to initiate infection. In dealing with a divided genome virus, it is important to demonstrate that each nucleic acid component is not only essential in infection but also contributes genetically to the progeny. Individual RNA components can be isolated from dianthoviruses of different origins and a new set of genome reassortants can be created for analyses of certain genetic markers. Okuno et al. (1983) tested compatibility of genome segments from SCNMV, RCNMV-SW, and RCNMV-C. Any combination of RNA-1 and RNA-2 from these viruses was highly infectious, indicating that RNA-1 and RNA-2 are compatible not only between strains of the same virus but also between certain member viruses of the same group. However, the compatibility of RNA-1 and RNA-2 of different origirls is not always bilateral. For instance, in testing reassortants between CRSV and RCNMV-AUS, Lommel(1983) found that while the heterologous combination of RCNMV-AUS RNA-1 and CRSV RNA-2 was infectious, the reciprocal combination failed to initiate infection. Similar results were obtained in combinations of different strains of RCNMV. Rao and Hiruki (1987) demonstrated that a heterologous combination of RCNMV-34 RNA-1 and RCNMV-48 RNA-2 is highly infectious whereas the reciprocal combination is not.

THE DIANTHOVIRUSES

283

C. Aggregation of Coat Protein Two strains of CRSV, A and N, were distinguished by their sedimentation properties in sucrose density gradients (Tremaine et al., 1976). When CRSV-A particles were treated at pH 5, all the particles aggregated irreversibly into clusters of 12 virus particles and linked aggregates. Absorbance scans of these preparations centrifuged into sucrose density gradient columns showed a distinctive sedimentation pattern and single virus particles were not detected. CRSV-N preparations contained only single virus particles after being subjected to the same treatment. In mixtures, both strains of the virus were readily detected after the treatment. When single-lesion isolates of each homologous and heterologous combination of RNA-1 and RNA-2 were propagated on cowpea, purified on sucrose density gradients, and analyzed by the same method, it was found that the aggregation phenotype was determined by the RNA-1 of CRSV-A (Dodds et al., 1977).The two strains of CRSV can also be distinguished by their reaction with 0.5% SDS at pH 5.0 (Tremaine and Ronald, 1976). Particles of CRSV-N were dissociated but CRSV-A particles were not. Tests of reassortants from homologous and heterologous combinations of RNA-1 and RNA-2 from CRSV-A and CRSV-N indicated that the pH 5 SDS-sensitive phenotype was determined by RNA-1 of CRSV-N.

D . Serological Specificity A series of tests was conducted on serological specificity of the progeny viruses originating from heterologous combinations of RNA-1 and RNA-2 among SCNMV, RCNMV-SW, and RCNMV-C. The serological specificity of reassortant viruses was identical with that of the donor virus of RNA-1 as tested by ELISA (Okuno et al., 1983). Similar tests by immunosorbent electron microscopy on the serological specificity of the reassortant viruses confirmed that RNA-1 codes for coat protein of dianthoviruses (Fig. 11; Chen et al., 1984). A stabilized reassortant virus from RCNMV RNA-1 and CRSV RNA-2 had a coat protein serologically identical with RCNMV. It was confirmed that the coat protein gene resides on RNA-1 for CRSV and RCNMV (Lommel, 1983). RCNMV-H and RCNMV-34 could be readily distinguished by gel immunodiffusion. Their heterologous reassortant, RNA-1 from RCNMV-34 and RNA-2 from RCNMV-H, gave a strong reaction to RCNMV-34 antiserum, but no reaction with RCNMV-H antiserum. Another reassortant, RNA-1 from RCNMV-H, RNA-2 from RCNMV-34

FIG.11. Electron micrographs of dianthoviruses and their reassortant preparations tested by immunosorbent electron microscopy. The Formvar-carbon coated grids in (A), (B),(C), and (D)were treated with SCNMV antiserum prior to mounting virus specimens. The grids in (E)were precoated with RCNMV-SW antiserum. Bar represents 100

THE DIANTHOVIRUSES

285

reacted with RCNMV-H, but not with RCNMV-34 antiserum. Both reassortants were serologically identical with their parental viruses from which RNA-1 derived (Osman et al., 1986). E . Electrophoretic Mobility

Reassortants derived from heterologous combinations of RNA-1 and RNA-2 among SCNMV, RCNMV-SW, and RCNMV-C were stable and their progeny viruses maintained their RNA components without change in the electrophoretic mobility (Okuno et al., 1983). The electrophoretic mobilities of the coat proteins of the progeny viruses derived from single-lesion isolation were similar to those of the viruses providing RNA-1 (Okuno et al., 1983). When the replicative dsRNAs of RCNMV, CRSV, and their reassortant (RlC2) were electrophoresed in parallel, the subgenomic RNAs (coat protein messengers) all had the same mobility, whereas the mobility of RNA-2 of the stabilized reassortant R1C2 was slower than that of RNA-2 of parental CRSV, suggesting a molecular evolution during stabilization of the progeny virus (Lommel, 1983).

F. Symptomtology The symptoms of virus infection are not a simple gene product, but rather the result of the action of one or more gene products (Lane, 1979). Nevertheless, genes which affect symptoms are of great interest to those who wish to identify the gene product. In an attempt to detect genetic determinants for certain symptoms on the two genomic RNAs of SCNMV, RCNMV-SW, and RCNMV-C, several plant species known to respond differently to their parental viruses were selected. Tests of reassortants on C. amranticolor, cowpea, and Red Kidney bean plants indicated that the lesion type and systemic symptoms were usually similar to those induced by the donor viruses of RNA-1. However, in a few cases such as the reassortant ClS2, large concentric necrotic lesions developed on Red Kidney bean leaves while their RNA-1 donor RCNMV-C caused only small whitish necrotic spots

nm. (A) Purified SCNMV preparation decorated with SCNMV antiserum (dilution 1/256). (B)Purified SCNMV preparation undecorated with antiserum. ( C )Sap preparation containing SCNMV and RCNMV-SW extracted from cowpea leaf tissue with mixed infection, decorated with SCNMV antiserum (dilution 11256). (D)As in (C) but decorated with RCNMV-SW antiserum (dilution 1/256). (E)Sap preparation containing progeny virus particles of a reassortant (RCNMV-SW RNA-1 + SCNMV RNA-2), decorated with RCNMV-SW antiserum (dilution 1/256). (From Chen et al., 1984.)

C. HIRUKI

286

TABLE IV

FUNCTIONS OF GENOME SEGMENTS OF DIANTHOVIRUSES~ RNA-1

RNA-2

Codes for protein of M, 39K (2,3) Codes for protein of M, 36K (2,3) Controls aggregation specificity (1) Controls serological specificity (2,4,5) Controls host range specificity (2) Controls systemic invasion of sweet clover at 26"Cb (4)

Codes for protein of M, 34K (2,3) Controls systemic invasion of cowpea (5)

0 Numbers in parentheses are reference numbers: (1)Dodds et al. (1977); (2) Lommel (1983); (3)Morris-Krsinich et al. (1983); (4) Okuno et al. (1983); (5)Osman et al. (1986). b Several other types of symptoms are determined by interaction between RNA-1 and RNA-2 (2,4).

(Okuno et al., 1983). On cowpea leaves at 26"C, RCNMV-C caused small whitish necrotic spots while the reassortant ClS2, like SCNMV and the reassortant SlC2, caused brownish ringspots. Only SCNMV caused acute death of the inoculated leaves. On sweet clover, SCNMV (SlS2) and a reassortant (SlR2) caused severe systemic mosaic symptoms at both 25 and 17"C, whereas RCNMV (RlR2) caused local and, rarely, systemic symptoms of necrotic spots at 17°C. These results indicated that S1 is essential for systemic invasion of sweet clover at 26°C. A reassortant (RlS2) caused symptoms similar to those caused by RCNMV (RlR2) at 17°C but localized infection with necrotic spots at 26°C. It suggests that S2 complements R1 in causing local infection at 26°C. Evidence that the symptom expression at 26°C results from interaction between RNA-1 and RNA-2 is provided also by several reassortants tested on several host species (Okuno et al., 1983). With regard to the function of RNA-2, Osman et al. (1986)reported that RNA-2 determines the morphology of lesions induced by RCNMV isolates and their reassortants in cowpea and their ability to invade the plant systemically. A summary of principal functions of RNA-1 and RNA-2 is shown in Table IV.

IX. CYTOPATHOLOGY Light microscopy. In light microscope study, amorphous inclusions containing darkly stained granular material were found in the cytoplasm of infected cowpea epidermal cells when stained with Phloxine

THE DIANTHOVIRUSES

287

B 5 days after inoculation with SCNMV (Hiruki et al., 1984~). Large amorphous aggregates consisting of spherical virus particles were observed in thin sections of SCNMV-infected cells. These were similar to the amorphous bodies observed in the light microscope (Hiruki et al., 1984~). Electron microscopy. The cytoplasm of D. barbatus cells infected with CRSV contained numerous virus particles 47 hours or more after inoculation (Fig. 12A). They were readily observed in large accumulations by 160 hours after inoculation (Weintraub et al., 1975). Intracellular crystals, which may have been formed by desiccation of the tissue resulting from necrotization, were seen in older infections of cowpea (Fig. 12B). A number of abnormal structures were observed in the nuclei of CRSV-infected cowpea and D. barbatus mesophyll cells (Fig. 13).Large aggregates in the nuclei, as well as in the cytoplasm, were found readily in the systemically infected leaf and root tissue about the time that systemic symptoms began to appear (Weintraub et al., 1975). Similar observations were made on thin-sections from a N . clevelandii plant infected with RCNMV (Francki et al., 1985)and from a Red Kidney bean leaf infected with SCNMV (C. Hiruki and M. Chen, unpublished data). In close contact with or in proximity to virus aggregates were tubules, the lengths of which it was not possible to determine, scattered throughout the nuclei of both CRSV-infected cowpea and D. barbatus. Upon close examination of the aggregates, single rows of virus particles were found in association with only a trace of what appeared to have been tubules. These tubules were present in both the nuclei and the cytoplasm of infected cowpea whereas in D. barbatus they were found only in the nuclei (Weintraub et al., 1975). No such tubules were found in tissues infected by RCNMV (Francki et al., 1985; C. Hiruki and M. Chen, unpublished data). A possibility that the presence of these cytoplasmic tubules was an artifact caused by disruption of the nuclear membrane above or below the level of the section could not be ruled out. These nuclear tubules resulting from CRSV are morphologically distinct and considerably wider than microtubules observed in dividing nuclei (Burgess, 1970; Esau and Gill, 1969; Cronshaw and Esau, 1968; Pickett-Heaps, 1967). Whether these tubules have a function as a site of viral replication or assembly remains to be determined. In leaf mesophyll cells of N . clevelandii infected with RCNMV, Francki et al. (1985) found numerous virus particles scattered in the cytoplasm and vacuole. While nuclei of infected cells also occasionally contain a few virus particles, they have never been observed to form crystals in infected cells. However, the formation of irregular virus aggregates was seen in heavily infected cells.

288

C . HIRUKI

FIG.12. Transmission electron microscopy of thin sections of CRSV-infectedplants. (A) A D . barbatus leaf cell containing large aggregates of CRSV particles, 47 hours after primary infection. (B) A cowpea leaf cell containing cytoplasmic virus crystals, sectioned in equatorial plane of the CRSV particles, 65 hours after primary infection. n, Nucleus; v, virus. Bars represents 1 pm. (From Weintraub et al., 1975.)

THE DIANTHOVIRUSES

289

FIG.13. A thin section of CRSV-infected D. barbatus cell showing nucleus in primary infection. Tubules are scattered throughout nucleoplasm. Note aggregates of CRSV particles and normal protein crystals. cr, Protein crystal; n, nucleus; t, tubules; v, virus. A bar represents 1 pm. (From Weintraub et al., 1975.)

X. TRANSMISSION BY VECTORS Four kinds of vector transmission have been investigated: nematodes, fungi, insects, and dodder, although many problems still remain to be worked out. Nematodes. In soil transmission of CRSV, adult nematodes of Longidorus macrosoma and Xiphinema diversicaudatum were involved as vectors (Fritzsche and Schmelzer, 1967;Kegler et al., 1977).An additional vector species (Longidorus elongutus) was found by Fritzsche et al. (1979)in the root range of CRSV-infected apple trees and S. media, a weed in the orchard. In a greenhouse experiment using adult nema-

290

C. HIRUKI

todes from infected apple plants, virus transmission to Chenopodium murale as test plants was at a rate of about 10%. Longidorus spp. were found in soil layers to 40 cm depth, where the dense root system was located, with the highest population density at a rate of 25 nematodes/100 cm3 soil at a depth of 10 cm. In addition t o apple trees, weeds species in the orchard, such as Chenopodium spp., Poa annua, and S . media, all of which are known hosts of Longidorus spp. (Pietler, 19781, were also naturally infected by CRSV. Among 40 plants of Urtica urens in an orchard with CRSV-infected apple trees, 21 were CRSV-infected; similarly infected were 4 out of 40 P . annua plants (Kleinhempel et al., 1980). Similar nematode transmission studies have not been reported on CRSV occurring in other parts of the world, nor on RCNMV and SCNMV. Fungi. For RCNMV which is apparently soil-borne, the possibility of transmission by Olpidium brassicae, an obligatory root parasite with a wide host range, was suggested (Lange, 1977; Bowen and Plumb, 1979) and was tested in different laboratories. Gerhardson and Insunza (1979) reported that a Swedish isolate of RCNMV was transmitted t o N . clevelandii by planting the seedlings in soil which had been contaminated by growing virus-infected plants, or by adding virus suspension to the soil prior to planting. However, the presence of 0. brassicae (lettuce strain) in the soil merely increased the rate of transmission and was not essential as a vector for soil transmission. In similar tests using an Australian isolate of RCNMV and a lettuce strain of 0. brassicae, Lyness et al. (1981) were unable to demonstrate a vector relationship between them. Leggat (1981) also confirmed these reports by an additional test in which RCNMV was treated with polyamino acids, poly-L-ornithine or poly-D-lysine, prior to adding to suspensions of zoospores of 0. brassicae (lettuce strain). None of these treatments caused RCNMV to become transmissible by Olpidium zoospores. Hiruki (1986) reported that the presence of a tobacco strain of 0. brassicae was not essential for transmission of SCNMV in sand culture. Insects. None of several isolates of CRSV from Canada (Kemp, 1964),Britain (Hollings and Stone, 19651, and Germany (Kegler et al., 19771, nor a Swedish isolate of RCNMV (Gerhardson and Lindsten, 1973)were transmitted by Myzus persicae. In England, Acrythosiphon pisum, the only aphid species found on red clover, and glasshouse cultures of A . pisum and M . persicae did not transmit RCNMV (Bowen and Plumb, 1979). In tests with weevils, Sitona lineatus, Apion aestivum, Apion apricans, and Apion assimile did not transmit RCNMV to test plants after being confined on infected plants for 1-16

THE DIANTHOVIRUSES

291

days. Apion assimile, Apion flavipes, Apion uethiops, and S . lineatus from RCNMV-infected clover in the field also failed to transmit RCNMV (Bowen and Plumb, 1979).In Canada Sitona cylindricollus did not transmit SCNMV between sweet clover plants (Hiruki, 1986). Dodder. Kemp (1964)attempted transmission by Cuscutu campestris from infected D. barbatus to healthy D. barbatus and seedlings of Gypsophila paniculata, an experimental host of CRSV. No transmission occurred.

XI. ECOLOGICAL STUDIES Ecological studies of dianthoviruses encompass diverse areas including the production of ornamental crops such as carnations, orchards of apples, pears, sour cherries, and grapevines, and forage legumes such as alfalfa, red clover, white clover, and sweet clover plants. Cultivation practices and optimal growth conditions vary according to the crop in question. With regard to the mode of transmission of dianthoviruses, three kinds of virus transmission in soil have been investigated: (1) transmission by nematode vectors (Fritzsche, 1968;Fritzsche and Schmelzer, 1967; Fritzsche et at., 1979),(2) transmission by fungal vectors (Gerhardson and Insunza, 1979;Leggat, 1981;Lyness et al., 1981;Hiruki, 1986),and (3) transmission by drainage water without vectors (Kegler and Kegler, 1981; Hiruki, 1986). Since there is no evidence for transmission of dianthoviruses by fungi, it will not be discussed here. Although CRSV is much less widespread in carnation than formerly, it is capable of causing severe loss of flower quality and quantity when infection takes place (Kassanis, 1955;Hollings and Stone, 1965;Kemp, 1964;Hakkaart, 1964).Roguing, selection, and meristem culture (Rybalko and Kharuta, 1978)are useful measures in reducing the spread of disease, for carnation plants are not completely invaded by CRSV, and the disease symptoms are recognized in the initial cutting. CRSV in carnation spreads readily by careless handling and leaf and root contact when management of carnation stocks is neglected. However, there is no report that nematodes are involved as vectors of CRSV in infecting carnation under natural conditions. In contrast, CRSV infection in orchards in East Germany appears to be effected by nematodes and extensive studies were made on various aspects of CRSV transmission in the field. Longidorus macrosomu (Fritzsche, 1967),L. elongatus (Fritzsche et al., 19791,and X . diversicaudatum (Fritzsche and Schmelzer, 1967) are root-feeding ectoparasites and parasitize not only fruit trees in the orchards but also

292

C.HIRUKI

weed species which serve as volunteer hosts of both the virus and nematodes. Natural occurrence of CRSV in S. media from vineyards in the Rhine/Pfalz region was reported and X . diuersicaudatum was thought to be a possible vector (Rude1 et al., 1977). These nematodes are long-lived, have wide host ranges, and are capable of surviving in adverse conditions, even in the absence of host plants, for considerable periods. For example, Harrison and Hooper (1963) found that L. elongatus survived in moist soil at room temperature without plants for more than 2 years, the longest period tested. In the orchards, the highest population of Longidorus spp. was found at a depth of 10 cm from the soil surface, with proved depth of distribution up to 40 cm where the dense apple root system is available for feeding (Fritzsche et al., 1979). The nematodes migrate to the subsoil when topsoil conditions are hot or dry in summer or frozen in winter and return when environmental conditions are favorable to their feeding. The migration of nematodes is, nevertheless, very slow. Fritzsche (1968) estimated that Longidorus sp. migrated only a few millimeters per day. Harrison and Winslow (1961) calculated that a population of X . diuersicaudatum invaded uncultivated woodland at the rate of about 33 cm per year. In East Germany, monthly applications of slurry (rabbit excrement) t o the orchard soil during May to July significantly reduced a population of L . elongatus. In contrast, the population of S. media plants was considerably higher on the site where slurry was applied than on the one without slurry. Slurry application had no effect on the occurrence of CRSV in the soil (Fritzsche et al., 1982). Although several nematodes, including Longidorus caespiticola, Longidorus goodeyi, Helicotylenchus vulgaris, and Paratylenchus sp., were isolated from RCNMV-infested soil, apparently no virus transmission experiments were conducted (Bowen and Plumb, 1979). Similarly, the role of nematodes as a vector has not been investigated with SCNMV. With regard to vectorless transmission of dianthoviruses, Hollings and Stone (1977) reported that RCNMV was exuded into soil from roots of virus-infected N . cleuelandii and T . pratense and was capable of infecting healthy N . cleuelandii bait seedlings planted in the soil. Gerhaldson and Insunza (1979) found that drainage water collected from pots containing RCNMV-infected N . cleuelandii contained infectious virus particles. SCNMV also was detected from soil around an experimentally infected plant and from drainage water collected from the pots containing SCNMV-infected plants (Hiruki, 1986). Lyness et al. (1981) reported that all N . cleuelandii plants growing in autoclaved soil became infected when the roots were artificially damaged prior to inoculation by pouring infective sap onto the soil. Virus release from

THE DIANTHOVIRUSES

293

FIG. 14. Map of Alberta showing frequent occurrence of SCNMV (*I in the areas 3 and 4. 1, Light brown soil; 2, dark brown soil; 3, black soil; 4, gray wooded soil; dark jagged lines outline areas at present considered as nonagricultural land.

the roots of virus-infected plants and subsequent reinfection of neighboring plants were proved for CRSV (Kegler and Kegler, 1981). CRSV particles in the soil remained infective for at least 7 months without plants, or after autoclaving for 2 hours at 121°C but not at 137°C (Kegler and Kegler, 1981). In Alberta there are four major soil climatic areas which must be considered in deciding kinds, crop combinations, and seeding ratio of forage crops to be grown in a given area (Fig. 14). During a survey period from 1979 to 1983, SCNMV was found widespread in the central and northern areas of Alberta where major soil types are black soil and gray wooded soil, respectively. It must be noted that sweet clover is particularly recommended as an important component of a forage seed mixture for poorly structured gray wooded soil. SCNMV was found to be prevalent in this area. Out of 1014 samples including alfalfa, alsike clover, red clover, sweet clover, crown vetch, and white clover, SCNMV occured solely on sweet clover with the exception of one case of a SCNMV isolate on alfalfa (Hiruki, 1986). Recently the alfalfa isolate has been found to be a new pathogenic variant of

294

C. HIRUKI

SCNMV that is serologically distinct from the type isolate of SCNMV (Inouye and Hiruki, 1985). The fact that SCNMV was not isolated from other forage legume crops growing in the areas where SCNMV was prevalent appears to suggest that SCNMV is very well adapted to sweet clover under the local field conditions. This is supported by the fact that alfalfa, alsike clover, and white clover were not infected when sap-inoculated with SCNMV and maintained at 25 ? 2"C, while both M.officinalis and M.alba were infected with ease (Hiruki et al., 19844. White clover was also resistant to inoculation with RNA preparations isolated from SCNMV at greenhouse temperatures of 17°C as well as 26°C (Okuno et al., 1983). Evidence that the alfalfa isolate is serologically distinct from the type isolate of SCNMV appears to be an indication that serological diversity and host adaptation of SCNMV can occur in the field. Musil and Gallo (1982a,b) have reported the occurrence of three serotypes of RCNMV without definite geographical boundary in their distribution. At present, however, it is not certain whether SCNMV occurs in many serotypes in Alberta. Precise serological studies of field isolates, using monoclonal antibodies, are required. Many nematode-transmitted viruses are also transmitted through the seed in a range of weed and crop plant hosts (Lister and Murant, 1967). However, there is no reported evidence of seed transmission for dianthoviruses.

XII. CONCLUDING REMARKS Dianthoviruses have recently emerged as an interesting group of plant viruses, though there are many other equally important plant virus groups known to have multipartite genomes or to produce functional subgenomic RNAs. The structure of dianthoviruses is simple, and their RNAs, a bipartite genome, are compatible not only between strains of a virus but also between member viruses in the group. However, certain genome recombinations between dianthoviruses resulting in infectious reassortants are possible only unilaterally (Lommel, 1983; Rao and Hiruki, 1987). This is similar to the reported cases of reassortants between two distantly serologically related strains of tomato black ring virus (Randles et al., 19771, and between certain strains of tobacco rattle virus (Lister, 1969; Ghabrial and Lister, 1973).Although dianthoviruses share a unique feature with pea enation mosaic virus in that the coat protein genome resides on RNA-1, the viruses of the dianthovirus group sediment as a single component. Coupled with other attractive properties such as high virus yield and

THE DIANTHOVIRUSES

295

stability, the group offers an ideal experimental system to research basic problems related to virus structure, replication,and genome strategy. There are conceivable advantages and disadvantages of divided genome systems in dianthoviruses. Deletion of RNA nucleotides representing smaller genome parts (Shirako and Brakke, 1984a,b) is a possible disadvantage of certain divided genome systems. However, such a case has not yet been reported for dianthoviruses. The advantages of a bipartite genome system for long-term survival are at least threefold. First, it offers an increased opportunity for providing biologically advantageous reassortants to cope with intracellular and environmental pressures for survival. For example, the coat protein cistron which directly influences virus stability resides on RNA-1, and other physical, chemical, and serological properties of dianthoviruses are also under the genetic control of RNA-1. Therefore, dianthoviruses may effectively adjust themselves to changing chemical, physical, and biological conditions by means of genetic reassortment. A bipartite genome of a dianthovirus is assumed to be packaged into separate particles which sediment as a single virus component in ultracentrifugation without revealing any sign of molecular heterogeneity. If this is so, it follows that biological selection pressure to two classes of genomic RNA can act independently, again allowing rapid adjustment for survival. Second, a bipartite system assures that the chance of genetic assortment may be amplified or modified by its vector such as nematodes. Vector nematodes have been reported for CRSV in orchards (Fritzsche and Schmelzer, 1967; Kegler et al., 1977). However, the occurrence of such new field reassortants of CRSV is yet to be proved. The identification of three field serotypes of RCNMV in Czechoslovakia (Musil, 1969b; Musil and Gallo, 1982a,b) and two serotypes of SCNMV in Canada (Inouye and Hiruki, 1985) are very interesting examples of genetic diversity within a virus population. Experimental evidence from reassortment studies (Hiruki et al., 1981; Lommel and Morris, 1982; Okuno et al., 1983; Morris, 1983; Lommel, 1983; Chen et al., 1984; Osman et al., 1986) gives indirect support for a possibility that such field serotypes can result from genetic reassortment. Third, the bipartite system of dianthoviruses makes it possible for the separation of genome function in time sequence or intracellular location with regard to translation, replication, or packaging of a bipartite genome as nucleoprotein particles. Recent work on the translation strategy of RCNMV-AUS has indicated that RNA-1 encodes at least three proteins of 36,000, 39,000, and 60,000 Da, and RNA-2 encodes a protein of 34,000 Da (S.A. Lommel, personal communication). Several RNA-containing plant viruses encode proteins of 30,000-

296

C. HIRUKI

35,000 Da which are thought to play a role in the cell-to-cell translocation of virus particles (Davies and Hull, 1982; Zimmern and Hunter, 1983; Ohno et al., 1983). The protein of 34,000 Da encoded by RNA-2 of RCNMV-D (Morris-Krsinich et al., 1983) and CRSV (Lommel, 1983) may also fall into this category. Further progress in the studies of genome functions and replication of dianthovirus may be expected from protoplast research using individual RNA species. Theoretically, it is conceivable that complete readout of their sequence allows coding of proteins totaling approximately 165,000 Da by RNA-1 and 55,000 Da by RNA-2. This suggests, therefore, that a few more proteins of unknown function(s1 are yet to be discovered during the course of future investigations of virus replication. The subgenomic RNAs of dianthoviruses may be encapsidated or synthesized during replication of genomic RNAs. However, there is no information available yet whether they are translated in the same manner in uiuo as they are in uitro. With regard to choice of methodology for future studies, the outstanding ability of monoclonal antibodies to recognize minute antigenic differences in virus particles should be useful in several ways in studies of dianthoviruses. First, it will provide the means to delineate field serotypes and genetic reassortants in the laboratory in a reproducible manner. Second, monoclonal antibodies will be a powerful tool for understanding molecular structure of coat protein and particle-to-particle relationships of dianthoviruses. Recent studies have accumulated a wealth of information on aggregation behavior of several strains of CRSV (Kuhne et al., 1983; Tremaine and Ronald, 1976; Tremaine et al., 1976,1983,1984). These results will provide excellent background information for such studies. The structural analysis of epitopes and of conformational changes in coat protein by means of monoclonal antibodies is expected to open an entirely new field of research. Third, since CRSV is reportedly nematode transmitted, the mechanism of vector specificity, if any, can be precisely understood by application of monoclonal antibody technology. Fourth, cellular proteins involved in virus replication can be identified using monoclonal antibodies. Since the target protein of a monoclonal antibody can be detected by radioimmune precipitation or Western blot procedures (Carter and ter Meulen, 19841, a similar analysis can be made to identify cellular RNA polymerase. This approach will provide a means to investigate the transcription mechanism, virus replication, and involvement of host cell proteins in the process. In summary, the dianthoviruses are a unique group having simple structure and interesting biological properties. Future studies must address the issues of the relevance of the dsRNAs to the viral replica-

THE DIANTHOVIRUSES

297

tion process, the mode of genesis of subgenomic RNAs, determination of genome functions, and the nature of the viral translation products generated in uitro. The dianthoviruses can serve as an excellent system for such studies and will continue to yield many interesting results, which will help expand the scope of plant virology.

ACKNOWLEDGMENTS The author wishes to thank his colleagues and associates who provided him with information, unpublished data, and manuscripts prior to their publication. Special thanks are due to T. Tribe for preparation of the illustrations and Gina Figueiredo for editorial assistance. This work is supported in part by funds from the Natural Sciences and Engineering Research Council of Canada (Grants No. A3843,STR G1450,IC0145) and from the Alberta Agricultural Research Council (Grant No. 78-0038).

REFERENCES Anonymous (1980).Research Report 1978.Dublin. An Foras Taluntais. Rev. Plant Puthol. 59,481 (4884). Atabekov, J. G., and Morozov, S. Yu. (1979).Adu. Virus Res. 25, 1-91. BarJoseph, M., and Salomon, R. (1980).J. Gen. Virol. 47, 509-512. Barton, R. J. (1977).J. Gen. Virol. 35, 77-87. Bennet, P. R., and Milne, K. S. (1976).Actu Hort. 59,61. Bercks, R., and Querfurth, G. (1972).Phytoputhol. 2.75,215-222. Blaszczak, W.(1981).Zesz. Prob. Post. Nuuk Rol. 244, 199-208. Blaszczak, W.,and Micinski, B. (1980).Tug. Ber., Akud. Lundwirtsch. Wiss. DDJZ. (Berlin) 184, 367-375. Bockstahler, L. E., and Kaesberg, P. (1962).Biophys. J. 2, 1-9. Bowen, R., and Plumb, R. T. (1979).Ann. Appl. Biol. 91,227-236. Bremer, K.,and Lahdenpera, M. L. (1980).Ann. Agric. Fenn. 20,214-228. Brierley, P., and Smith, F. F. (1957).Phytoputhology 47, 714-721. Burgess, J. (1970).Protoplasm 71, 77-89. Car, A. J. H. (1979).Welsh Plant Breed. Sta. Rep. for 1978. 200-201. Carter, M. J.,and ter Meulen, V. (1984).Adv. Virus Res. 29, 95-130. Casper, D.L.D., and Klug, A. (1962).Cold Spring Harbor Symp. Quant. Bwl. 27,l-24. Casper, R. (1976).Mitt. Biol. Bundesunst. Land. Forstwirtsch. H 170,33-41. Chen, M. H., Hiruki, C., and Okuno, T. (1984).Can. J. Plant Puthol. 0,191-195. Clark, M. F. (1981).Annu. Rev. Phytopathol. 19,83-106. Clark, M. F., and Adams, A. N. (1977).J. Gen. Virol. 34,475-483. Cronshaw, J.,and Esau, K. (1968).Protoplasma 05,1-24. Crowther, J. R., and Abu-El Zein, E. M. E. (1979).J. Gen. Virol. 42,597-602. Davies, J. W.(1979).“Nucleic Acids in Plants” (T. C. Hall and J. W. Davies, eds.), Vol. 2, pp. 111-149.CRC Press, Boca Raton, Florida. Davies, J. W., and Hull, R. (1982).J. Gen. Virol. 61, 1-14. Devergene, J. C., and Cardin, L. (1967).Ann. Epiph. 18,65-83. Dodds, J. A,, Tremaine, J. H., and Ronald, W. P. (1977).Virology 83, 322-328. Esau, K.,and Gill, R. H. (1969).Can. J.Bot. 47,581-591. Forster, R. L.S. (1981).Res. Report DSZR N . Z.,p. 16. Frame, J., Harkness, R. D., and Hunt, I. V. (1976).J. Br. Grussl. SOC.31, 111-115. Francki, R. I. B., Milne, R. G., and Hatta, T. (1985).“Atlas of Plant Viruses,” Vol. 2,pp. 47-52. CRC Press, Boca Raton, Florida.

298

C. HIRUKI

Fritzsche, R. (1968). Biol. Zentrallpl. 87, 139-146. Fritzsche, R., and Schmelzer, K. (1967). Naturwissenschafkn 54,498-499. Fritzsche, R., Kegler, H., Thiele, S., and Gruber, G. (1979).Arch. Phytopathol. Pflanzenschutz 15, 177-180. Fritzsche, R., Thiele, S., and Kegler, H. (1982). Arch. Phytopathol. Pflanzenschutz 18, 319-325. Gallo, J., and M u d , M. (1984). Acta Virol. 28, 82-83. Gerhardson, B., and Lindsten, K . (1973). Phytopathol. 2.76, 67-79. Gerhardson, B., and Insunza, V. (1979). Phytopathol. 2.94, 67-71. Ghabrial, S. A., and Lister, R. M. (1973). Virology 62, 1-12. Gould, A. R., Francki, R. I. B., Hatta, T., and Hollings, M. (1981). Virology 108,499-506. Habili, N., and Francki, R. I. B. (1974). Virology 61,443-449. Hakkaart, F. A. (1964).Neth. J . Plant Pathol. 70, 53-60. Hamilton, R. I., Edwardson, J. R., Francki, R. I. B., Hsu, H. T., Hull, R., Koenig, R., and Milne, R. G. (1981).J . Gen. Virol. 54, 222-241. Harrison, B. D., and Hooper, D. J. (1963). Nematologica 9, 158-160. Harrison, B. D., and Winslow, R. D. (1961).Ann. Appl. Biol. 49, 621-633. Hatta, T., and Francki, R. I. B. (1984). J . Virol. Methods 9, 237-247. Hatta, T., Francki, R. I. B., and Grivell, C. J. (1983). J . Gen. Virol. 64, 687-692. Henriques, M. I. C., and Morris, T. J. (1979). Virology 99.66-74. Hiruki, C . (1982).Znt. Conf. Comp. Virol., 4th, BanK Canada Abstr. p. 57. Hiruki, C. (1986). Plant Dis. 70, 1129-1131. Hiruki, C., and Figueiredo, G. (1985). Acta Hort. 164, 217-224. Hiruki, C., Okuno, T., Rao, D. V., and Chen, M. H. (1981). Znt. Congr. Virol., 5th, Strusbourg Abstr. p. 235. Hiruki, C., Figueiredo, G., Inoue, M., and Furuya, Y . (1984a). J . Virol. Methods 8,301308. Hiruki, C., Rao, A. L. N., Furuya, Y., and Figueiredo, G. (1984b).J . Gen. Virol.66,227375. Hiruki, C., Rao, D. V., Chen, M. H., Okuno, T., and Figueiredo, G. (1984~).Phytoputhology 74, 482-486. Hollings, M., and Stone, 0. M. (1965). Ann. Appl. Biol. 56, 73-86. Hollings, M., and Stone, 0. M. (1970). CMVIAAB Descript. Plant Viruses (21). Hollings, M., and Stone, 0. M. (1974). Rep. Glasshouse Crops Res. Inst. for 1973, 122123. Hollings, M., and Stone, 0. M. (1977). CMVIAAB Descript. Plant Viruses (181). Inouye, N., and Hiruki, C. (1985). Ann. Phytopathol. SOC.Jpn. 51, 82. Kalmakoff, J., and Tremaine, J. H. (1967). Virology 33, 10-16. Kaper, J. M. (1972). RNA viruses: Replication and structure. FEBS Symp. 27, 19-41. Kaper, J. M. (1973). Virology 55,299-304. Kassanis, B. (1955). Ann. Appl. Biol. 43. 103-115. Kegler, G., and Kegler, H. (1981). Arch. Phytopathol. Pflanzenschutz 17, 307-323. Kegler, H., Verderevskqja, T. D., Proll, E., Fritzsche, R., Schmidt, H. B., Kalasjan, J. A., Kosakovskaja, 0. I., Kleinhempel, H., and Henmann, E. (1977). Arch. Phytopathol. Pflanzenschutz 13,297-310. Kemp, W. G. (1964). Can.J . Bot. 42, 45-55. Kilevits, M. A. (1976). Trudy Latv. S.-Kh. Akad. 100, 47-55. Kleinhempel, H., Gruber, G., and Kegler, H. (1980).Acta Phytopathol. Acud. Sci. Hung. 15, 107-111. Koenig, R. (1978).J . Gen. Virol. 40,309-318. Kohler, G., and Milstein, C. (1975). Nature (London)256,495-497. Kowalska, Cz. (1974). Rozczn. Nauk. Rol. 4(E), 89-122.

THE DIANTHOVIRUSES

299

Kristensen, H. R. (1957).Tidssk. Plunteaul. 61, 718-736. Kuhne, T., and Eisbein, K. (1983).Actu Phytopathol. Acud. Sci. Hung. 18, 101-110. Kuhne, T., Proll, E., and Reichenbacher, D. (1983).Actu Phytopathol. Acad. Sci. Hung. 18,85-99. Lane, L. C. (1979).In “Nucleic Acids in Plants” (T. C. Hall and J. W. Davies, eds.), Vol. 2. pp. 65-110. CRC Press, Boca Raton, Florida. Lange, L. (1977).“Plantesygdomme i Danmark 1976.”Arsoversigt. Leggat, F.W. (1981).Ph.D. thesis, Univ. of Queensland, Australia. Leonard, D. A., and Zaitlin, M. (1982).Virology 117,416-424. Lister, R. M. (1969).Fed. Proc. Fed. Am. SOC.Exp. Biol. 28, 1875-1889. Lister, R. M., and Murant, A. F. (1967).Ann. Appl. Bwl.59,49-62. Lommel, S. A. (1983).Ph.D. thesis, Univ. of California, Berkeley. Lyness, E. W., Teakle, D. S., and Smith, P. R. (1981).Aust. Plant Puthol. 10, 6-7. Marcinka, K., M u d , M., and Mrena, E. (1969).A& Virol. 13, 247-250. Matthews, R. E. F. (1982).Znteruirology 17, 1-199. Morris, T. J. (1983).“Current Communications in Molecular Biology Plant Infectious Agents, Viruses, Viroids, Virosoids, and Satellites,” pp. 80-83. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Morris-Krsinich, B. A. M., Forster, R. L. S., and Mossop, D. W. (1983).Virology 124,349356. Musil, M. (1969a).Biologia 24, 33-45. Musil, M. (1969b).Actu Virol. 13, 226-234. Musil, M. (1981).Zesz. Prob. Post. Nuuk Rol. 244, 67-80. M u d , M., and Gallo, J. (1982a).Actu Virol. 26, 497-501. Musil, M., and Gallo, J. (1982b).Acta Virol. 26, 502-505. Noordam, D., Thung, T.H., and van der Want, J. P. H. (1951).T&dschr. P1. Ziekt. 57,l14. Ohno, T., Takamatsu, N., Meshi, T., Okada, Y., Nishiguchi, M., and Kiho, Y. (1983). Virology 131,255-258. Okuno, T., Hiruki, C., Rao, D. V., and Figueiredo, G. (1983).J . Gen. Virol. 64, 19071914. Osman, T.A. M., Dodd, S. M., and Buck, K. W. (1986).J. Gen. Virol. 67,203-207. Pappu, H. R., and Hiruki, C. (1986).Phytopathology 76, 1121. Pickett-Heaps, J. D. (1967).Aust. J . Biol. Sci. 20, 1539-551. Pietler, K. (1978).Thesis, Wilhelm-Pieck-Univeritat, Rostock. German Democratic Republic. Pollock, R. R., Teillaud, J. L., and ScharfF, M. D. (1984).Annu. Rev. Microbwl. 38,389417. Ragetli, H. W. J., and Elder, J. (1977).Can. J . Bot. 55, 2122-2136. Randles, J. W., Harrison, B. D., Murant, A. F., and Mayo, A. (1977).J . Gen. Virol. 36, 187-193. Rao, A. L. N., and Hiruki, C. (1985).Plunt Dis. 69,959-961. Rao, A. L.N., and Hiruki, C. (1987).J . Gen. Virol. 68, 191-194. Rao, A. L. N., Ohki, S. T., Lakshman, D. K., and Hiruki, D. (1985).Phytopathology 75, 1357. Richter, J., Kleinhempel, H., Gruber, G., and Kegler, H. (1978).Arch. Phytopathol. Pflunzenschutt 14,411-412. Rudel, M., Querfurth, G., and Paul, H. L. (1977).Nuchrichtenbl. Dtsch. Pflanzenschutz 29, 59-60. Rutgers, T., Salerno-Rife, T., and Kaesberg, P. (1980).Virology 104, 506-509. Rybalko, A. E., and Kharuta, L. G. (1978).Fkwl. Biokhirn. Kult Rust. 10, 535-540. Sander, E.,and Dietzgen, R. G. (1984).Adu. Virus Res. 131-163.

300

C. HIRUKI

Shirako, Y., and Brakke, M. K. (1984a). J . Gen. Virol. 65, 119-127. Shirako, Y.,and Brakke, M. K. (1984b). J . Gen. Virol. 65, 855-858. Steere, R. L. (1956). Phytopathology 46,60-69. Sutton, J. (1974). Aust. Plant Puthol. Soc. Newslett. 3, 21. Taylor, J. M., Illmensee, R., and Summers, J. (1976).Biochirn. Biophys. Acta 442,324330. Tremaine, J. H. (1961).Can. J . Bot. 39, 1705-1709. Tremaine, J. H.,and Ronald, W. P. (1976). J . Gen. Virol. 30, 299-308. Tremaine, J. H.,Ronald, W. P., and Valcic, A. (1976). Phytoputhology 66,34-39. Tremaine, J. H.,Ronald, W. P., and McGauley, E. M. (1983). Phytoputhology 73, 12411246. Tremaine, J. H., Ronald, W. P., and McGauley, E. M. (1984). Phytopathology 74, 161165. Weintraub, M., Ragetli, H. W., and Leung, E. (1975). J . Phytornorphol. 25, 288-304. Zimmern, D.,and Hunter, T. (1983). EMBO J . 2, 1893-1900.

ADVANCES IN VIRUS RESEARCH, VOL. 33

BANANA BUNCHY TOP: A N ECONOMICALLY IMPORTANT TROPICAL PLANT VIRUS DISEASE James L. Dale Agricultural Research Laboratories Department of Primary Industries Indooroopilly, Queensland 4068, Australia

I. INTRODUCTION The banana is one of the world’s most important crops. It is grown in all types of tropical agricultural systems from small, mixed, subsistence gardens to very large company-owned monocultures, and the export trade in bananas is considerable. The control of diseases in bananas accounts for much of the effort in producing the crop. Banana bunchy top is one of the most important of these diseases and by far the most important virus disease. Yet banana bunchy top virus (BBTV) which is limiting or threatening the production in one-quarter of the world’s banana growing areas has, in recent years, been virtually neglected by plant virologists. In contrast to most economically important plant viruses, relatively little is known about BBTV. This was not always the case; in the 19208, knowledge about BBTV was very much “state of the art” and the control of the disease in Australia provides one of the most dramatic and successful examples of plant virus control using phytosanitary methods. In this article, I will explore the possible reasons why this virus has been so neglected, as well as what is known about it, how it has been controlled in Australia, and finally what possible directions future research could take to reduce the often devastating effects of the disease. Limited BBTV Research: Possible Explanations

There are probably a number of reasons why research on banana bunchy top virus has been so limited. First, the virus does not occur in all banana-producing countries. The “infected” areas produce about 26%of the world’s bananas but only 12%of the world’s banana export trade, of which the Philippines contributes 10%. Most of the bananas produced in these areas are for local consumption. Thus, the disease 301

Copyright 0 1987 by Academic h, Inc. All rights of reproduction in any form reserved.

302

JAMES L. DALE

and, therefore, production problems in these countries do not substantially affect the price of bananas paid by importing countries. The developed areas of North America, Europe, and Japan import 88% of the world banana export trade, of which all but 4% comes from the Caribbean and Central and South America, area where BBTV does not occur. Second, the large majority of countries where the disease occurs are subtropical or tropical and have had little or no access to well-equipped and staffed plant virus laboratories. The possible exception to this is Australia. During the 1920s and 1930s a considerable effort was put into BBTV research and as a result the disease has been effectively controlled (Section IV). However, because of this success, there has been little incentive to fund a continuing large-scale research program. Finally, and probably the least significant reason, is that both the host (banana) and the virus are difficult to work with under glasshouse conditions. Commercial banana cultivars do not produce seed and, therefore, experimental plants have to be vegetatively propagated, making mass propagation more tedious and the resultant plants less suitable than seedlings. Also, the banana contains considerable amounts of latex and tannins which, in other plants, are known to interfere with virus extraction and purification (Brunt and Kenten, 1963).There are no known alternative hosts of BBTV that would overcome or alleviate these problems. Finally, the available evidence suggests that BBTV is a luteovirus and as such should replicate only in the phloem; since luteoviruses normally multiply in plants only to a very low concentration, purification of even the most well studied of these viruses such as potato leafroll virus or barley yellow dwarf virus has proved dificult (Rochow and Duffus, 1981). There is evidence (Dale et al., 1986) that BBTV may occur in much lower concentration than barley yellow dwarf virus. Considering the probable reasons outlined above, it is easier to understand why so little research has been put into this virus even though in the last 25 years our understanding of plant viruses and plant virology has exploded.

11. THEDISTRIBUTION OF BANANAS AND THE GEOGRAPHICAL HISTORYOF BBTV Taxonomically, bananas belong in the family Musaceae, a monocotyledonous family of only seven genera. There are two main types, the dessert banana which is eaten ripe and raw and the plantain which

BANANA BUNCHY TOP

303

is eaten green but cooked. Both types are most conveniently referred to as Musa x paradisiuca L. as the true origin of the various clones is very difficult to ascertain. The other species of the Musa genus cultivated commercially is abaca or Manila hemp ( M u m textilis) which is grown as a fiber crop. Bananas, plantains, and abaca will be considered at least to some extent in this review. Bananas are propagated vegetatively usually from suckers or “bits” (Simmonds, 1966),each pseudostem producing one bunch before being replaced by the strongest sucker growing from the base. Bananas originated in Southeast Asia and their cultivation probably dates back to the earliest settled agriculture in the region. Therefore, banana cultivation may be tens of thousands of years old. The earliest written records of banana cultivation are from India in the Epics of the Pali Buddhist canon of 500-600 BC (Reynolds, 1951).There are also written records of bananas in Indonesia about 350 BC and China about AD 200. The movement of bananas to the Mediterranean and Africa is not well documented. It is assumed that the plant was not brought to the Mediterranean until the Mohammedan conquest of AD 650 (Simmonds, 1966).Time of arrival in Africa is considerably disputed; suffice to say that bananas were widely cultivated in that continent by the time of the first European explorers in the fifieenth century. Spread from Southeast Asia to Polynesia was probably around AD 1000.The Portuguese took bananas from West Africa to the Canary Islands in the early fifteenth century and from there to the Caribbean and Central and South America before the end of that century. By 1983,the Caribbean and Central and South America produced 46% of the world’s production but contributed a massive 86% of the world banana exports, a trade that began in the early nineteenth century and expanded rapidly during that century. The documented history of banana bunchy top disease is very recent in comparison to the history of banana cultivation. The first written reports of the disease were from Fiji in 1889 and 1890 although it was probably present there as early as 1879 (Magee, 1953). The disease spread rapidly across Fiji and production declined from 778,000 bunches in 1892 to 114,000in 1895;however, floods were responsible for at least some of this decline. The industry recovered to some extent mainly due to the introduction of more tolerant varieties and the expansion of production areas. A survey by Magee in 1937 (Magee, 1953) revealed that the incidence of BBTV in plantations and gardens was between 5 and 30%. The virus has become entrenched in Fiji and still remains a major problem. BBTV was reported from Taiwan as early as 1900 (Sun, 1961)and Egypt in 1901 (Magee, 1953;Fahmy, 1924) but the origins of these

304

JAMES L. DALE TABLE I GEOGRAPHICAL DISTRIBUTION OF BANANA BUNCHYTOP VIRUS Location Confirmed reports American Samoa Australia Bonin Is. China Congo Egypt Ellice Is. Fiji Gabon Guam India Kirabati Philippines Sri Lanka Taiwan Tonga Vietnam Wallis Is. Western Samoa Unconfirmed reports Bangladesh Burma Hong Kong Kampuchea Laos New Caledonia Okinawa Pakistan Sabah (North Borneo) United Arab Republic Zaire

Reference Magee (1967) Magee (1927) Gadd (1926) Anonymous (1979) Wardlaw (1961) Magee (1953) Campbell (1926) Magee (1927) Manser (1982) Beaver (1982) Magee (1953) Shanmuganathan (1980) Ocfemia (1926)(abaca) Castillo and Martinez (1961)(banana) Magee (1953) Sun (1961) Magee (1967) Vakili (1969) Simmonds (1933) Magee (1967) Foure and Manser (1982) Buddenhagen (1968) Buddenhagen (1968) Stover (1972) Stover (1972) Buddenhagen (1968) Buddenhagen (1968) Buddenhagen (1968) Magee (1967) Buddenhagen (1968) Manser (1982)

outbreaks are unknown. The disease first appeared in Sri Lanka (Petch, 1913) and Australia (Magee, 1927) in 1913;both infections probably originated from the importation of diseased suckers from Fiji. Further reports came from the Bonin, Ellice, and Wallis Islands during the 1920s.Bunchy top was probably brought to India from Sri Lanka about 1940 (Magee, 1953) and is now present in the states of Andhra Pradesh, Gujarat, Kerala, Maharashtra, Tamil Nadu, and Uttar Pradesh (Chenulu, 1984;Singh, 1979).The history of bunchy top

FIG.1. Banana-producingareas of the world and distribution of banana bunchy top virus.

306

JAMES L. DALE

in the Philippines dates back to 1910 when abaca bunchy top was first recorded (Ocfemia, 1926) but was not reported on bananas until 1960 (Castillo and Martinez, 1961). The exact geographical distribution of BBTV is extremely difficult to determine. It is fairly certain that the virus occurs in at least 19 countries, with possible reports from 11 other countries (Table I). Thus, the disease is confined to the continents of Asia, Africa, and Australia as well as a number of islands of the Pacific (Fig. 1).Many large producers of bananas and plantains from these areas have not recorded BBTV including Indonesia, Malaysia, Thailand, and Papua New Guinea, but considering their proximity to infected areas, these areas would be most under threat. The African nations, which produce 11%of the world’s bananas and a massive 62% of the world’s plantains, should also be considered to be under threat because of the presence of BBTV in the Congo and Gabon. The virus obviously did not travel with bananas to the New World as BBTV has not been recorded in the Caribbean or Central and South America. 111. THE CHARACTERISTICS OF THE VIRUS

A. Symptoms The diagnosis of BBTV depends virtually exclusively on the recognition of the symptoms of infection and, rarely, with further confirmation by aphid transmission. It is fortuitous, therefore, that the symptoms of an advanced BBTV infection in bananas is reasonably distinctive. The identification of the initial symptoms, however, is rather more difficult. The first symptom-bearing leaf develops dark green streaks of variable length in the leaf veins, midribs, and petioles (Fig. 2a). This streaking and dotting is often described as a “Morse code” pattern. As well as these symptoms, subsequent leaves become progressively dwarfed and develop marginal chlorosis or yellowing (Fig. 2b). As the disease develops the leaves become more upright and crowded or bunched at the apex of the plant, hence the name of the disease (Fig. 3a and b). Depending on when the plant becomes infected, it may produce no fruit or the bunch may not emerge from the pseudostem. Magee (1939) found that the phloem of infected plants is disorganized with excessive and irregular divisions in the phloem and associated parenchyma. Many of these small cells become chlorophyllous, giving rise to the dark green dots and streaks in the leaves, midribs, and petioles. Symptoms develop more quickly at high temperatures

FIG. 2. Whole-plant symptoms of banana bunchy top virus: (a) dwarfed plant with upright leaves showing marginal yellowing; (b) advanced symptoms of infection with narrow, upright leaves.

FIG.3. Banana leaf symptoms of infection: ( a )leaf with marginal yellowing and dark green streaking of leaf veins and midrib; (b) “Morse code” pattern of streaks and dots on the leaf.

BANANA BUNCHY TOP

309

than at low temperatures both in the field (Sun, 1961) and in controlled environment cabinets (Dale et al., 1986).

B . Host Range and Varietal Susceptibility The reported host range of BBTV appears t o be limited to six species of which five are members of the Musaceae. There is considerable confusion surrounding the taxonomy of the Musa genus because of the high incidence of polyploidy and therefore speciation is difficult, if not impossible, to define. However, Magee (1927)was able to transmit BBTV to five “species” of the Musaceae, namely, Musa x paradisiuca L. (both dessert and plantain types), M . textilis Nee, M . banksii F. Muell., the Fe’i banana, and Ensete uentricosurn (Welw.) E. E. Cheesm, but was unable to infect Strelitzia sp. (Musaceae), Rauenala sp. (Musaceae), Canna spp. (Cannaceae), Solanurn tuberosurn (Solanaceae), or Zea mays (Gramineae). The recent report that Colocasia esculenta (Ram and Summanwar, 1984)can be symptomlessly infected with BBTV may have important epidemiological significance as this species is widespread in many areas where the virus occurs and has been reported as a natural host of Pentalonia nigroneruosa (Mehta et al., 19641,the aphid vector of BBTV (Magee, 1927).Therefore, it is essential to confirm that Colocasia esculenta is a host and whether it is a natural reservoir of the virus. Other non-Musa species such as Cannu sp. (Vakili, 1969) and Heliconia sp. (Magee, 1967)have been implied as virus reservoirs based on bunchy-top-like symptoms in the field, but these have not been confirmed by transmission experiments. All species, cultivars, or types in the genus Musa that have been inoculated have been found to be susceptible to BBTV. Both Magee (1948)and Jose (1981)found that there was considerable variation between cultivars of banana in the number of test plants infected under experimental conditions. In Australia, Magee (1927,1948)reported that the dessert cultivar Gros Michel was more resistant to infection than the widely grown Dwarf Cavendish, and of the 19 cultivars tested by Jose (1981),Kanchikela and Venattukunnan were the most resistant.

C. Virus Strains The existence of strains of BBTV is not well defined. Abaca bunchy top virus (ABTV) is very probably a strain of BBTV based on the similarity of biological properties of both viruses. ABTV causes a yellows-type disease in abaca (M.textilis) and is transmitted in a per-

310

JAMES L. DALE

sistent manner by P. nigroneruosa but is not transmissible by sap (Ocfemia, 1926,1930).BBTV has been shown to infect abaca (Magee, 1927)causing similar symptoms to ABTV; however, ABTV does not infect banana (Ocfemia and Buhay, 1934).The exact relationship of these two viruses has not yet been resolved. There has been considerable interest shown in the reaction of some plants of the Fijian banana cultivar Veimama. Magee (1948)described a number of Veimama plants in Australia which showed very mild symptoms of BBTV infection. Suckers taken from these plants could be superinfected with an Australian isolate of BBTV but only when inoculated with three times as many infective aphids as usually required, and then only 40% of plants developed normal BBTV symptoms. Healthy Veimama plants could be as easily infected with an Australian isolate as Dwarf Cavendish, developed the same symptoms, and did not recover. Magee was unable to transmit BBTV from the mildly infected Veimama plants. In a yield trial, mildly infected Veimama plants produced about 25% less than healthy plants (Magee, 1967).The phenomenon of recovery of the Veimama cultivar has also been observed in Fiji (Magee, 1948).The most probable explanation of this phenomenon is the presence of a mild or attenuated strain of BBTV (Magee, 1967;Simmonds, 1966).Further work will be required to verify this. Symptom variation has been reported in natural infections. Vakili (1969)differentiated a “banana strain” and an “abaca strain” on a Giant Cavendish cultivar in South Vietnam.

D . Virus Etiology Banana bunchy top virus has been classified as a probable luteovirus by Matthews (1982)and Rochow and Duffus (1981)although there is no information reported on the morphology or biochemical composition of the virions of this virus. The probable classification is based on the biological characters, namely, that the virus (1)causes a yellows-type disease, (2) damages and disorganizes the phloem, (3) cannot be transmitted by sap inoculation, and (4)is transmitted by aphid (P.nigroneruosa) in a persistent manner. Magee (1927)demonstrated that BBTV was not sap transmissible but was aphid transmitted. That the virus was transmitted in a persistent manner was not reported by Magee until 1940. More recently, we have provided additional evidence that BBTV is a luteovirus (Dale et al., 1986).We have consistently isolated doublestranded RNA (dsRNA) from virus-infected plants but not from corre-

BANANA BUNCHY TOP

311

sponding healthy plants. The dsRNA with the highest molecular weight, which we call BT-1, has a molecular weight of about 4.4 x 106. Luteoviruses have single-stranded RNA as their genomic nucleic acid and this RNA has a molecular weight of about 2.0 x lo6 (Matthews, 1982). Therefore, luteoviruses should have dsRNA as their replicative form and the largest of the dsRNAs should have a molecular weight of about 4.0 x lo6. Such dsRNAs have also been extracted from plants infected with barley yellow dwarf virus (Gildow et al., 1983) and beet western yellows virus (Falk and Duffus, 19841, both well-characterized luteoviruses. This, together with the similarity between the total dsRNA pattern of BBTV (Fig. 4) and barley yellow dwarf virus (Gildow et al., 1983), provides additional strong evidence that BBTV is a luteovirus. Further, a higher concentration of dsRNA was extracted from plants grown at 30°C than at 25"C, and this concentration peaked 23 days after inoculation and then decreased dramatically (Dale et al., 1986). We were able to extract only about one-sixth to one-tenth as much dsRNA from BBTV infected bananas as we could from barley yellow dwarf-infected oats. If the concentration of virions in an infected plant is correlated to the concentration of dsRNA, which is probable, then it is more understandable why insufficient BBTV virions have been purified or visualized for further characterization or even to be certain of their identity. This is despite attempts by a number of researchers and the application of recently developed techniques for purifying luteoviruses and other phloem-limited viruses (Takanami and Kubo, 1979). However, in our laboratory, we have observed luteovirus-like particles in purified preparation from BBTV-infected plants (J.L. Dale and G. M. Behncken, unpublished). The isolation of BBTV-specificdsRNA should make possible the molecular cloning of at least part of the BBTV genome. This, used as a probe, could greatly assist in determining the optimum conditions for virion concentration in infected plants and for purifying the virions from those plants.

E . Diagnosis of Infected Plants Successful control of most diseases depends to a large extent on the accurate diagnosis of that disease. Historically, the diagnosis of BBTV has been based on recognition of symptoms. Unfortunately, because of the inherent difficulties in using symptoms as the only diagnostic technique, this has led to considerable confusion about the occurrence and distribution of the virus and has been a limiting factor in the eradication of BBTV from Australia (Allen, 197813). Confusion has

312

JAMES L. DALE

FIG.4. Gel electrophoresis of dsRNA extracts of (A) BBTV-infected bananas, (B) fungal dsRNA markers, and (C) Fiji disease virus (FDVbinfected sugarcane. (From Dale et al., 1986, by permission of the Society for General Microbiology.)

arisen by misdiagnosis of other virus diseases of banana such as cucumber mosaic virus (Mehta et d., 1964). Magee (1953) reported the presence of a disease in North Borneo (Sabah) very similar to that caused by BBTV but the identity of this disease was questioned by Reinking (1950) and Simmonds (1966) and has not been resolved. The limitations of using visual diagnosis of BBTV in the control program in Australia have been highlighted by Allen (197813). The incubation period of the disease, that period from inoculation to appearance of symptoms in the newly emerged leaf, has been reported variously as 19 days in summer to 125 days in winter (Allen, 1978a) and 23-72 days in summer to 50-131 days in winter (Sun, 1961).It is

BANANA BUNCHY TOP

313

this latent period, when infected plants cannot be detected visually, that allows the disease to persist between inspections (Section IV,A). Summanwar and Marathe (1982) developed a diagnostic technique based on the reaction of infected tissue with 2,3,5-triphenyltetrazolium chloride. The technique is not BBTV specific as it is reported to also detect infectious chlorosis (cucumber mosaic virus) in bananas. Improved diagnosis will almost certainly depend on the development of an antiserum or nucleic acid probe.

F. Epidemiology The basic epidemiology of BBTV appears to be relatively simple. The virus is transmitted in space by its aphid vector, P . nigroneruosa, and in infected planting material and, in time, in its perennial host, the banana. Dispersal internationally and over long distances intranationally is in infected material, and the virus has probably been introduced to most countries this way. Dispersal by aphids is essentially confined to short distance spread. Allen (1978b) showed that in two case studies, the mean distance of new infections from their source of inoculum in an established plantation was 17.2 m, about two-thirds of new infections were within 20 m, and all but about 1%were within 86 m. In a study of spread of BBTV between plantations, Allen and Barnier (1977) found that a new plantation (planted with healthy planting material) had a greater than 88% chance of becoming infected within 12 months if it was adjacent to an infected plantation, that this dropped to a 27% chance if the new plantation was 50-1000 m distant, and none of 19 new plantations that were more than 1000 m from an infected plantation became infected within 12 months. Naturally infected alternate hosts outside the Musaceae have not been conclusively identified although observations on possible such hosts have been made (Section 111,B). If these species are hosts of BBTV then it is important to determine the impact of this on the epidemiology of the disease. In the infected banana-growing areas of Australia, wild or uncultivated Musaceous hosts of the virus are very few in number and are probably of little consequence in the epidemiology of the disease. However, in the more tropical growing areas of Southeast Asia and the Pacific, such Musaceous alternate hosts are much more prevalent and may be a significant reservoir of the virus. The banana aphid, P. nigronervosa, the only recorded vector of BBTV, occurs in practically all banana-producing countries and, while it has been recorded on other hosts, survives and multiplies indefinitely on its primary host, the banana. Therefore, practically all banana-producing countries free of the disease already have all the components of a BBTV epidemic except the virus.

314

JAMES L. DALE

IV. CONTROL: THEAUSTRALIAN EXPERIENCE The history of banana bunchy top disease in Australia can be dated back to 1913 when the disease was first reported in the Tweed River district of New South Wales (Fig. 5 ) . The most probable origin of the infection was Fiji, from where suckers had been imported to the Tweed district just previous to that year. The disease had been recorded in Fiji from 1889. This supposedly single introduction began a saga unequaled in the history of plant pathology in Australia. The banana industry in Australia was expanding rapidly from 1913 to about 1926.This was due, first, to the very high prices being paid for bananas, as previously most were imported and, second, many exservicemen from the First World War were resettled in the banana growing areas of northern New South Wales. Of course, with the rapid expansion of the area under bananas came a n insatiable demand for planting material and with it the involuntary and, for the most part, unrecognized spread of BBTV. The disease quickly moved southward and northward (Fig. 51, first destroying the industry around Terranora before becoming well entrenched in the Mullumbimby district. By 1922,BBTV had reached Bangalow and by 1923,it had reached the Richmond River. The movement of the disease north was equally dramatic. The first record in Queensland was at Currumbin in 1916.By 1922,112plantations covering 1250 acres were affected in the Currumbin and Tallebudgera area. Despite a Queensland Government proclamation in 1921 prohibiting movement of planting material north from Cunumbin and Tallebudgera, BBTV had moved north across the Brisbane River by 1923 and by 1925 had reached the Caboolture River. Planting material from an infected plantation had been sent north of the Caboolture River to Kilcoy (one plantation), Beerwah (one), Yandina (one), and to Innisfail in north Queensland (two plantations). The outbreaks at Kilcoy and Innisfail were quickly eradicated. The economic effects of the disease were as dramatic as its spread. In New South Wales, practically no plantation was free of BBTV in the Tweed and Brunswick areas by the end of 1925.Over 800 plantations (representing 5000 acres) had been deserted and between 1922 and 1926,90% of the banana area had gone out of production. This was despite a rapid expansion in plantings to compensate for those areas abandoned due to BBTV infection. The New South Wales production figures dramatically show the demise of the industry (Fig. 6) with peak production of 460,000cases in 1922 crashing to 140,000cases in 1925.The industry in southeast Queensland fared no better. The Currumbin and Tallebudgera districts were probably the most severely

BANANA BUNCHY TOP

Q U E E N S L A N D

(i

ViCTORlA

315

I

<-,j

FIG.5. Areas in New South Wales and Queensland where BBTV has been recorded.

316

JAMES L.DALE

4

X fn Y

fn 4 0

-

2

z

P L L

a

0

0

a &

1

1919

1920

1921

1922

1923

1924

1925

YEAR

FIG.6. Annual production of bananas in New South Wales from 1919 to 1925.

affected. At Currumbin, the number of producing plantations fell from 100 in 1922 to only 4 in 1925 and over the same period production had plummetted from 4,400 tons to 110 tons. Thus, by 1925,BBTV had virtually destroyed the industry in New South Wales and southern Queensland. The first published report, by Darnell-Smith (1919),of a plant pathological investigation of banana bunchy top was unable to associate any fungus, bacterium, or nematode with the disease. Possible causes were attributed to “a physiological condition, adverse conditions of soil o r climate or running out of the stock.” He did, however, recommend that planting material be taken only from healthy plants which showed no signs of the disease. Little progress was made in determining the cause for a further 5 years, except for a few casual observations. Darnell-Smith and Tryon (1923)wrote, “Some notoriety has attached to the pronouncement that a particular insect-the banana aphis (Pentulonia neruosu)-serves as the communicator of the disease between one plant and another, or even is its primal cause.” A

BANANA BUNCHY TOP

317

number of farmers put forward hypotheses as to the cause of the disease and the best known of these was Mr. J. Marks. His hypothesis was that banana bunchy top was caused by aphids; if the aphids were controlled with a kerosene emulsion then no disease would develop. An experiment was set up to test this in November, 1922,with $20,000 going to Mr. Marks if no BBTV developed in the treated plot. Unfortunately after 14 months 20% of plants in the treated plot were infected but this was considerably better than the control plot of 65% infection. Darnell-Smith (1924)tried limited aphid transmission without success but did not discard this as an avenue for further research. As the disease situation deteriorated, in 1923 cooperation between the New South Wales and Queensland Governments and the Commonwealth Government was decided upon. In May, 1924,the formation of the Bunchy Top Investigation Committee was announced. In October, 1924, two experimental plots were planted with planting material from north Queensland (assumed to be BBTV free); one plot was on land previously planted to bananas and the other on virgin land. In January, 1925,the disease appeared and spread rapidly through the virgin plot whereas the incidence in the replant plot was low (Magee, 1927). The major difference between the plots was the high aphid population in the virgin plot. This prompted Magee to transfer some of these aphids to healthy caged plants in the glasshouse. Within 23 days, symptoms of BBTV developed on the healthy plants (Magee, 1927).By 1927,Magee (1927)had demonstrated that (1)banana bunchy top was a virus, probably of the potato leafroll virus type, (2)the virus was transmitted by the aphid P . nigroneruosa, (3)the virus was not sap transmitted, (4)the virus damaged the phloem region of the vascular system which was probably the site of “infection,” and ( 5 ) the virus was present in the suckers of infected plants and this provided the primary means of spread. Virtually all this information was collected in 12 months, a remarkable feat considering the little that was known of plant viruses at that time and especially considering that aphid transmission of sap-transmissible plant viruses was first reported for cucumber mosaic virus in 1916 (Doolittle, 1916) and of non-sap-transmissible viruses (potato leafroll virus) in 1920 (Oortwijn Botjes, 1920).Thus, in 1927,the characterization of BBTV was as advanced as any of the non-sap-transmissible plant viruses, a situation that has certainly not continued in the ensuing 60 years. Based on his research results, Magee (1927)put forward a number of recommendations for the control of BBTV in Australia, and these were divided into two main directions. The first was the protection of uninfected areas by exclusion and the second, the

318

JAMES L. DALE

rehabilitation of infected areas by eradication. His recommendations were as follows: a. Exclusion 1. Prohibition of movement of vegetative parts of Musa spp. north or to any uninfected or lightly infected area. 2. Prohibition of movement of vegetative parts of Musa spp. within an infected area. 3. Trade in planting material by license only. 4. All plantations to be registered. 5. Destruction of all Musa spp. not in registered plantations. 6. Immediate eradication of infected plants in lightly infected areas. 7. Systematic inspection by growers of plantations in lightly infected areas. 8. Prohibition of movement of fruit from infected areas. 9. Declare BBTV a notifiable disease in uninfected areas. 10. Survey all banana plantations in Queensland. 11. Start an extension program to disseminate information about BBTV and its control. b. Eradication 1. Clean up all deserted plantations. 2. Clean up all infected plantations. 3. Destroy nonplantation bananas. 4. Declare the owner liable for eradication of infected plants. 5. Discourage further planting within the heavily infected areas until a clean certificate could be given for the whole area. 6. After 12 months, completely destroy all bananas in an area or plantation following an unfavorable report. 7. Future replanting with planting material from certified BBTV-free areas. Magee (1927) thought that the control of the aphid vector was unrealistic and that there was little hope of finding a resistant or immune cultivar. The control recommendations of the Bunchy Top Investigation Committee as put forward by Magee (1927) were quickly implemented. Legislation for plant diseases in Australia is a responsibility of the individual states and so the legislation and mechanisms for controlling BBTV differ between New South Wales and Queensland. The final outcome, however, has been essentially the same.

BANANA BUNCHY TOP

319

A . The Campaign in New South Wales The first proclamations and regulations under the New South Wales Plant Diseases Act were officially announced in November, 1927 (Magee, 1936).In summary, (1) quarantine areas were declared, (2) the movement of planting material was restricted, (3) permits were required before planting bananas, (4) procedures were laid down for the eradication of infected bananas, and (5) neglected or abandoned plantations were to be destroyed. The onus was on the farmer to regularly inspect his plantation and eradicate any infected plants. In February, 1928, four inspectors were employed by the New South Wales Department of Agriculture to advise and assist growers with their inspections and t o ensure that farmer obligations under the regulations were carried out. For this purpose, the infected areas were divided into zones which to a certain extent depended on the level of infection. The scheme showed early signs of success as the area planted to bananas steadily increased from a low 3340 acres in 1928 to 6256 acres in 1931 (Fig. 7) and there was little BBTV infection in new areas. However, in 1932, banana prices rapidly increased with the resultant enormous increase in banana plantings. By 1934, 22,289 acres were under bananas but at the same time, prices fell and banana growing had become uneconomic. By 1935, there were numerous serious outbreaks of BBTV and the entire control program was in jeopardy. There were many reasons for the failure of the campaign (Eastwood, 1946): (1) the inspectors could not efficiently enforce the regulations over the vastly increasing acreage, (2) the number of neglected and abandoned plantations increased as the industry became uneconomic, (3) there were delays in dealing with negligent farmers, and there were disputes over the ownership of neglected plantations, and finally (4) it was found that many farmers were unable to recognize the symptoms of BBTV infection except in its most advanced form. In 1935, the control campaign was reorganized with a cooperative effort between the New South Wales Banana Growers Federation (a farmer’s organization) and the Department of Agriculture. The essential difference was that the Banana Growers Federation financed the employment and transport of personnel to regularly inspect all plantations and to treat all infected plants. The finance was raised by a levy on production. The Department of Agriculture was responsible for the hiring and training of these personnel as well as organizing and directing the campaign. Since 1937, the campaign has been a remarkable success to the point that BBTV does not limit the production of bananas in New South Wales. The number of infected plants detected

320

0 1926

JAMES L. DALE

1

I

1

1

1

I

1930

1936

1940

1946

1060

1965

1

1060

I

1966

VEAR

FIG.7. Area planted to bananas in New South Wales, 1928 to 1966.

fell dramatically in the first 5 years of the new campaign from 1937 to 1941 (Table 11), from then on the decline was steady but consistent until about 1960. Since then, even though BBTV has not been eradicated, the number of infected plants has oscillated around a low level. Recently, Allen (1978a,b) studied the factors which influence the success of the roguing program in New South Wales. Originally, success was attributed to the early detection of infected plants. Allen (1978a), however, found that success actually involved four random variables (incubation period, relative infection rate, detection efficiency, and eradication efficiency) and one fixed variable (interval between inspections). The incubation period varied from 19 days in summer to 125 days in winter, and relative infection rate variation could be attributed to changes in host susceptibility or, more probably, to fluctuations in aphid activity. Detection efficiency was greatest in summer because symptoms develop between inspections more quickly and symptoms are more easily recognized during the summer growth phase. Eradication efficiency appeared to depend on the labor demands on the farmer. Allen (1978b) also addressed the problem of continued infection in

BANANA BUNCHY TOP

321

TABLE I1 ANNUALNUMBEROF BBTV-INFECTED PLANTSDETECTED IN NEWSOUTH WALES Year

Number

1937 1938 1939 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984

42,305 26,755 15,114 8,276" 5,550" 3,692a 3,417" 1,338" 969" 1,138" 1,851" 1,7430

a

Annual average of the 5-year period.

plantations due to the inability to visually detect recently infected plants not yet displaying symptoms. He found that the distribution of BBTV in a plantation was localized and was related to the mean flight distance of the aphid vector. Allen observed that the mean distance of spread of BBTV from the source of infection was 17.2 m. He calculated that about 25% of new infections would be within 5 m of the source of infection, 45% within 10 m, and 70% within 20 m. This information suggested that a great many more infected plants would be eradicated if all plants, although apparently healthy, were destroyed within 1020 m of a source of BBTV. The effect of this on ultimate total eradication is not presently known.

B . The Queensland Campaign The Queensland Campaign was commenced with the passing of the Banana Industry Protection Act in 1929 under which the Banana Industry Protection Board was set up. The Board consisted of both grower and government representatives. The essential elements of the campaign were the same as the original New South Wales approach with the onus on the farmer to detect and destroy infected plants along with permits required to plant and for planting material, restricted movements of planting material, and eradication of neglected or abandoned plantations (Colborne, 1953). The BBTV-free areas of north Queensland were protected by establishing a buffer zone just north of

322

JAMES L. DALE

the infected areas but free of infection itself. The movement of any banana material to any part of the state north of this buffer zone was completely prohibited. This isolation of the north has been an effective control except for one outbreak of BBTV at Innisfail in north Queensland in 1954.The area was quarantined and an eradication campaign was instigated. Three years later, the area was declared BBTV free and no further outbreaks have occurred. Again, the campaign in southern Queensland has been very successful in limiting the disease to a level where it is not causing any significant economic loss. Eradication, however, has not been accomplished. The history of BBTV in Queensland and New South Wales is an excellent example of the application of legislation to effect field control of a virus disease. Success was probably due to a number of factors including (1) the absence of virus reservoirs other than bananas and the small number of wild bananas to act as a reservoir, (2) the knowledge that the primary source of infection and spread was planting material and secondary spread was by aphid, (3) cultivation of bananas in relatively small and localized plantations rather than a widely distributed subsistence crop, and (4) cooperation of most farmers.

DIRECTIONS V. THE FUTURE:POSSIBLE Most attempts to control or eradicate banana bunchy top virus have been made in commercial banana-producing countries. In Australia, a control program has successfully contained the disease at a very low level in southeast Queensland and the north coast of New South Wales for about 50 years. However, the disease still requires continual monitoring. Eradication in Australia has not been achieved and there has been one isolated outbreak around Innisfail in north Queensland. Control measures were implemented in the late 1920s and have changed little since then. The basis for control came from the discovery that (1)banana bunchy top was caused by a virus, (2) this virus was transmitted by the aphid, P. nigronervosa, (3) the virus could be disseminated in infected vegetative planting material, and (4)there were no known non-musaceous alternate hosts. For the campaign to be successful, little more information about the virus was required and, indeed, in the ensuing 60 years little more has been discovered. Most research has been confined to verification of the identity of the disease in particular areas, varietal reactions to infection, some epidemiology, and more recently isolation of the dsRNAs associated with infection. In many countries, control by eradication may not be feasible be-

BANANA BUNCHY TOP

323

cause of the cultural practices of those countries. In Australia bananas are grown primarily in plantations which are strictly controlled. Very small numbers of bananas are grown in home gardens and these often for decoration as much as consumption. In BBW-infected areas, the number of bananas growing wild is limited. In contrast, the situation in most southeast Asian and Pacific Island nations is of widespread cultivation of bananas in home gardens and these are grown as a n important component of diet. As well, there are often large numbers of bananas and other Musa spp. growing wild. In these situations, control by eradication becomes increasingly difficult without major social upheaval. In considering what directions future research and control methods may take, it is necessary to consider not only what can be achieved using the technology available now but also what may become available in the near future. The present or “first-generation” measures for controlling plant viruses depend heavily upon a knowledge of the biological properties of the virus such as natural host range and mode of transmission and in many cases it is not necessary to have isolated or characterized the virus. First-generation control measures include eradication; supply of virus-free planting material; breeding or selection for immunity, resistance, or tolerance; vector control; disease avoidance; and mild strain protection. Eradication combined with the supply of virus-free planting material has been used with success in Australia but may not be applicable in most other “infected” countries. Of the other methods, banana breeding is extremely difficult because the commonly grown cultivars produce little or no seed. In addition, most banana breeding is centered in Central and South America where BBTV does not occur and therefore there is no incentive to include BBTV resistance in a breeding program, and, also none of the banana cultivars so far tested have been found to be BBTV immune and therefore no immune parent is available. Some varietal differences in susceptibility to BBTV infection have been reported (Magee, 1948; Jose, 1981). Based on present evidence, resistance to BBTV in bananas may have to be sought using the techniques of somaclonal variation, mutation inducement, and protoplast fusion to develop genetic variability. These techniques are already being used in Queensland and elsewhere to attempt to develop resistance to other banana diseases. Vector control is expensive, requires a reasonable degree of management expertise, and probably would not be successful for long-term control. Avoidance is more applicable to annual crops where planting times can be manipulated to avoid population peaks of vectors. The little-used method of mild strain protection may provide a n avenue for

324

JAMES L. DALE

research in the absence of success of other methods. Such approaches have been suggested previously (Buddenhagen, 1968)and strain variation has been recognized (Section 111,C). A sensitive assay system based on an antiserum or cloned dsDNA probe for BBTV would greatly assist research on this virus and ultimately control of the disease. To date, the inability to produce a satisfactory purified virus preparation has precluded the development of a BBTV-specific antiserum. Similarly, a reliable probe is not yet available. The development of improved, rapid BBTV detection methods would benefit field identification but their most immediate benefit would be in research, particularly in breeding or germplasm improvement for BBTV resistance, and in quarantine to ensure freedom from BBTV in tissue-cultured bananas now being increasingly used as primary planting material in several countries. The “second-generation” control methods will, in all probability, utilize recombinant DNA technology such as gene transfer. This may well depend on the identification and isolation of BBTV resistance genes in wild Musa species or from outside the genus Musa or the family Musaceae. Success may also depend on the manipulation of virus genes and, therefore, will require an intimate knowledge of the viral genome. For many plant viruses, such information has been accumulating for the last 20 years, with exponential growth in the last 10 years. Even though there is no recognized transformation system for the Musaceae (or for most monocots), such systems are likely to be developed in the forseeable future. However, a great deal more research is necessary on banana bunchy top virus before advantage could be taken of such a development.

ACKNOWLEDGMENTS I wish to acknowledge the very capable assistance of Judith Parry in preparing this review and also D. M. Persley and G. M. Behncken for critically examining the manuscript. My research in this field has been generously supported by the COD Banana Sectional Group Committee, the Banana Industry Protection Board, and the Rural Credits Development Fund of Australia.

REFERENCES Allen, R. N. (1978a).Aust. J . Agric. Res. 29,535-544. Allen, R. N.(1978b). Aust. J . Agric. Res. 29, 1223-1233. Allen, R. N.,and Barnier, N. C. (1977).N.S.W. Plant Dis. Surv. 46,27-28. Anonymous, (1979). “Pests and Diseases of Agricultural Crops in China,” Vol. 2. Agricultural Publisher, Peking. Beaver, R. G. (1982). Plant Dis. 66, 906-907. Brunt, A. A., and Kenten, R. H. (1963). Virology 19,388-392. Buddenhagen, I. W.(1968). F A 0 Plant Protect. Bull. 16,17-31.

BANANA BUNCHY TOP

325

Campbell, J . G. (1926).Annual Report of the Fiji Department of Agriculture (1925). Castillo, B. S.,and Martinez, A. L. (1961).FA0 Plant Protect. Bull. 9,74-75. Chenulu, V. V. (1984).Indian Phytopathol. 37, 1-20. Colborne, F. (1953).Queensl. Agric. J . 77,85-90. Dale, J. L.,Phillips, D. A., and Parry, J. N. (1986).J. Gen. Virol. 67, 371-375. Darnell-Smith, G.P. (1919).Agric. Gaz.New South Wales 30, 809-814. Damell-Smith, G. P. (1924).Queensl. Agric. J . 21, 169-179. Darnell-Smith, G. P., and Tryon, H. (1923).Queensl. Agric. J . 19, 32-33. Doolittle, S.P. (1916).Phytopathology 6, 145-147. Eastwood, H.W. (1946).Agric. Gaz. New South Wales 67, 643-646. Fahmy, T. (1924).Min. Agric. Eg. Bull. (30). Falk, B. W., and Duffus, J. E. (1984).Phytopathology 74, 1224-1229. Foure, E., and Manser, P. D. (1982).Fruits 37,409-414. Gadd, C. H.(1926).Trop. Agric. 66,3. Gildow, F.E.,Ballinger, N. E., and Rochow, W. F. (1983).Phytopathology 73,1570-1572. Jose, P. C. (1981).Agric. Res. J . Kerula 19, 108-110. Magee, C. J. P. (1927).Counc. Sci. Ind. Res. Bull. (30). Magee, C. J . P. (1936).J . Aust. Inst. Agric. Sci. 2, 13-16. Magee, C. J. P. (1939).New South Wales Dept. Agric. Sci. Bull. (67). Magee, C. J. P. (1948).J. Aust. Inst. Agric. Sci. 14, 18-24. Magee, C. J. P. (1953).J. Proc. R . Soc. New South Wales 87, 3-18. Magee, C. J. P. (1967).South Pac. Comm. Tech. Pap. (150). Manser, P. D. (1982).FA0 Plant Protect. Bull. 30, 78-79. Matthews, R. E.F. (1982).Intervirology 17, 1-200. Mehta, P. R.,Joshi, N. C., Rao, M. H., and Renjhen, P. L. (1964).Sci. Cult. 30,259-263. Ocfemia, G. 0. (1926).Phytopathology 16, 894. Ocfemia, G. 0. (1930).Am. J . Bot. 17,1-18. Ocfemia, G. O.,and Buhay, G. G. (1934).Philipp. Agric. 22,267-280. Oortwijn Botjes, J. G. (1920).Dissertation, University of Wageningen. Petch, T. (1913).Trop. Agric. 41,427. Ram, R. D., and Summanwar, A. S. (1984).Cum. Sci. 63, 145-146. Reinking, 0.A. (1950).Plant Dis. Rep. 34,66-69. Reynolds, P. K. (1951).J. Am. Orient. SOC.Suppl. 12,28. Rochow, W. F., and Duffus, J. E. (1981).In “Handbook of Plant Virus Infections. Comparative Diagnosis” (E. Kurstak, ed.), pp. 147-170. Elsevier, Amsterdam. Shanmuganathan, N. (1980).FA0 Plant Protect. Bull. 28, 29-38. Simmonds, N. W. (1933).“Report on Visit to Samoa.” Dept. of Agriculture, Fiji. Simmonds, N. W. (1966).“Bananas,” 2nd Ed. Longmane, London. Singh, R. H. (1979).Indian J . Mycol. Plant Pathol. 9,253-254. Stover, R. H. (1972). “Banana, Plantain and Abaca Diseases.” Commonwealth Mycological Institute, Kew, Surrey, England. Summanwar, A. S., and Marathe, T. S. (1982).Curr. Sci. 61,47-49. Sun, S.-K. (1961).Spec. Publ. College Agric. Taiwan Univ. 10, 82-109. Takanami, Y.,and Kubo, S. (1979).J. Gen. Virol. 44,153-159. Vakili, N. G.(1969).Plant Dis. Rep. 63, 634-638. Wardlaw, C.W. (1961).“Banana Diseases.” Longmans, London.

This Page Intentionally Left Blank

ADVANCES IN VIRUS RESEARCH,VOL. 33

APPROACHES TO THE STUDY OF VECTOR SPECIFICITY FOR ARBOVIRUSES-MODEL SYSTEMS USING CULTURED MOSQUITO CELLS Victor Stollar Department of Molecular Genetics and Microbiology University of Medicine and Dentistry of New Jersey Robert Wood Johnson Medical School Piscatoway, New Jersey 08854

I. INTRODUCTORY REMARKS A . Transmission of Viral Diseases by Arthropod Vectors The demonstration by Walter Reed and his co-workers that yellow fever is transmitted to man by mosquitoes represented a landmark in the history of medicine (Reed et al., 1901;Reed, 1902). Not only did this work make possible practical measures to eliminate this dreaded disease in urban centers, but it also provided the first example of a human disease shown to be caused by a virus, or more precisely in terms of that time a filterable agent. Thus, the first virus shown to cause disease in man was an arthropod-borne virus. By 1978, 389 viruses could be referred to as definitely, probably, or possibly arthropod-borne (Berge, 1975;Karabatsos, 1978).Of this large total, however, only a small number are known to cause disease in man or domestic animals. Most of the arthropod-borne viruses (arboviruses) fall into five taxonomic groups (Schlesinger, 1980):(1) family Togaviridae, genus A1 phavirus; (2) family Flaviviridae; the flaviviruses until recently were classified as family Togaviridae, genus Fluuivirus; (3)family Bunyaviridae; (4) family Rhabdoviridae; and ( 5 ) family Reoviridae. All of these viruses are RNA viruses, and all replicate in the cytoplasm. With respect to human disease, the best known of the arboviruses fall into the first three families listed above and include, for example, eastern and western equine encephalitis viruses (EEE, WEE) in the family Togaviridae, yellow fever, dengue, and Japanese encephalitis viruses in the family Flaviridae, and California encephalitis virus in the family Bunyaviridae. Although some of these viruses are neurotropic and can cause encephalitis, various other arboviruses are trop327

Copyright 0 1987 by Academic Press, Inc. All rights of r e p d u d i o n in any form reserved.

328

VICTOR STOLLAR

ic for different organ systems and cause other types of pathology including hemorrhagic fever, hepatitis, arthritis, and skin rashes (Shope, 1980; see also Fields, 1985). From these brief remarks, i t is clear that arthropod-borne viruses are a diverse group, in both their physical and chemical structure and strategies of replication, as well as the disease syndromes with which they are associated. For this reason, the term arbouirus is no longer used, as it once was, in a taxonomic sense. Rather, the term arbouirus is used only to denote a biological property of certain viruses. To define the term more specifically, arboviruses are those viruses which, in order to be maintained in nature (1) must be alternately transmitted between a vertebrate host and a n arthropod host and (2) must be able to replicate well in both the vertebrate host and the arthropod host. With a few exceptions, most arboviruses cannot be transmitted directly from one arthropod host to another (the major exception is transovarial transmission), or from one vertebrate host to another (Chamberlain, 1980). The specific vertebrate and arthropod species which serve as hosts vary depending on the virus. Among the vertebrate hosts are reptiles, birds, and mammals, including man. The most common arthropod hosts are mosquitoes, ticks, and sandflies.

B . Specificity of Vectors and Viruses Despite the variety of vertebrate hosts and arthropod vectors which can be infected with arboviruses, each arbovirus shows a remarkable specificity with respect to its natural hosts, infecting only one or a small number of vertebrate and arthropod species (Chamberlain, 1980). Each arbovirus, therefore, has found an ecological niche involving a vertebrate host and an arthropod vector, in both of which it can replicate well and between which it can cycle. The availability of suitable vertebrate hosts and efficient arthropod vectors may explain, at least partially, the geographical localization seen with the various arboviruses, for example, the distribution of eastern equine encephalitis virus in the coastal states of the eastern United States but not in the midwest or western parts of the country.

C . Goals of Research into Vector Specificity and Approaches to These Goals One of the important goals of research in vector biology is to determine the physiological and biochemical basis of vector specificity, in other words, to identify those properties which determine the efficien-

VECTOR SPECIFICITY

329

cy with which a vector can transmit an infectious agent. To be more specific, why, in areas where many mosquito species abound, is yellow fever transmitted efficiently mainly by Aedes aegypti, and what are the properties of Aedes aegypti which are associated with this specificity? Questions such as these, concerning vector specificity, should properly begin with investigations at the level of the intact organism, i.e., the female adult mosquito. From work done in many laboratories (see Chamberlain, 1980; Hardy et al., 1983, for reviews) much has been learned concerning the replication of arboviruses in their mosquito vectors. For a vector to function efficiently in the transmission of virus, it must first bite an infected vertebrate host and ingest sufficient virus so that replication will occur in the vector. The first cells in the vector to be infected are the epithelial cells of the mesenteron (midgut). Following efficient replication in these cells, the virus must be able to spread through the basal lamina of the mesenteron into the hemocoel and eventually to the salivary glands, reaching a sufficiently high titer in the salivary gland secretions so that when the vector bites the next vertebrate host, enough virus will be injected to initiate replication in that host. Thus, to understand how and why vectors are able to transmit viruses, there must be knowledge of (1) the physiology and biochemistry of mosquitoes, including their biting habits, (2) the molecular biology of viral replication, and (3) the details of viral replication in the mosquito vector. For example, it is clearly necessary to know in what sites the virus replicates, to which tissues it spreads, and the levels of virus which are produced. Continuing at the level of the whole organism, we wish to identify those host properties which influence the amount of virus made and the ability of the virus to disseminate in the vector, ultimately to the salivary glands. These problems, however, do not lie within the scope of this article. Instead the reader is referred again to the review by Hardy et al. (1983) who have summarized very clearly what is known concerning the intrinsic factors which determine the vector competence of mosquitoes for arboviruses. Much of the work which these authors describe was carried out in their own laboratory. To cite an interesting example of vector specificity, this group has identified variants of Culex tarsalis mosquitoes which vary widely in their ability to function efficiently as vectors for western equine encephalitis virus (WEEV); and they have been able to localize the block to viral replication and spread in the vector-incompetent variants, to a “mesenteronal barrier.” Furthermore. they have demonstrated by genetic studies involving crosses between vector-competent and vector-incompetent mosquitoes that the ability to transmit WEEV is dominant.

330

VICTOR STOLLAR

Another example of vector specificity is provided by the species, Culex quinquefasciatus, which is very resistant to oral infection with WEEV, an alphavirus, but can serve as an excellent vector for St. Louis encephalitis virus, a flavivirus. Information obtained from studies such as these has been and will continue to be extremely important with respect to the understanding of vector specificity. Ultimately, however, to explain efficient viral replication in an intact organism and the ability of a vector to transmit virus, it will also be necessary to understand the ability of specific tissues or organs to support efficient viral replication at the cellular level. It is with studies at the cellular level that this article is mainly concerned.

D. Questions Concerning the Interaction between Virus and Host Cell If we wish to study virus-vector relationships at the level of the individual cell (with cells derived from vector species), what are the questions we would like to ask and what is the information we would like to obtain? I would suggest the following: (1)How are the efficiency of viral replication, the viral yield, and the outcome of infection (cell death vs cell survival) influenced by pinoperties of the host cell and by properties of the virus? (2) What are the structures, functions, and TABLE I

FUNCTIONS AND METABOLITES REQUIRED FOR THE REPLICATION OF ARTHROPOD-BORNE BY THE HOSTCELLQ RNA VIRUSESWHICHAREPROVIDED General 1. Energy supply-ATP 2. Protein synthesizing system 3. Endoplasmic reticulum and Golgi system for transport of membrane proteins More specifically 1. Substrates for virus-coded enzymes For polymerization of RNA: ATP, GTP, UTP, CTP For capping and methylation of 5’terminus of viral messenger RNA: GTP, S-adenosylmethionine 2. Enzymes needed to carry out modification of viral membrane proteins (these enzymes are associated with the endoplasmic reticulum and Golgi system) Glycosylating enzymes Acylating enzymes 3. ?? Requirement for a host protein factor to constitute a functional viral RNA polymerase a All arthropod-borne viruses have an RNA genome and replicate in the cytoplasm. The items listed above would, of course, not apply only to arthropod-borne viruses; most but not all would in fact apply to other cytoplasmic RNA viruses.

VECTOR SPECIFICITY

331

metabolites which the cell must provide in order for the virus to multiply? See Table I for a partial listing of these cellular contributions. In the material which is summarized in the following sections, I shall attempt to provide, from studies done in our laboratory, some information bearing on these fundamental questions. I have not attempted, however, in this review to summarize all the work done in our laboratory or by other groups concerning the replication of arboviruses in arthropod cell cultures. This area has been reviewed most recently in a collection of papers edited by Yunker (1986; see also Stollar, 1980, and Brown and Condreay, 1986).

11. SINDBIS VIRUS REPLICATION IN VERTEBRATE AND IN MOSQUITO CELLS-A MODELSYSTEM Sindbis virus (SV) (family Togaviridae, genus Alphavirus) has a wide host range and is able to multiply in cell lines or primary cell cultures derived from many different vertebrate species. The properties of the virus are briefly summarized in Table 11. Most commonly, SV is studied in BHK 21 Syrian hamster cells, or in chick embryo fibroblasts (CEF). In both cell types, as in most vertebrate cell systems, viral replication is accompanied by extensive cytopathic effect. SV also replicates well in cells of arthropod origin, including a variety of mosquito cell lines as well as in Drosophila cells (Stollar, 1980). Most work with SV in mosquito cells has been done with cells derived from the Aedes albopictus (Ae. albopictus) line originally established by Singh (1967). Viral yields are high in both mosquito and vertebrate cells and levels of 109 PFU/ml are readily obtained. In contrast to the results with vertebrate cells, no cytopathic effect was observed in the first experiments with SV or Semliki forest virus (also family Togaviridae, genus Alphavirus) in Ae. albopictus cells, although high yields of virus were obtained (Stevens, 1970; Davey and Dalgarno, 1974). As will be discussed below, CPE was subsequently demonstrated in certain clonally derived populations of Ae. albopictus cells. The comparative study of SV replication in vertebrate and mosquito cell cultures provides us with a simplified model for viral replication at the level of the whole organism, i.e., in the vertebrate host and the vector, and enables us to examine the features of a virus which make it possible to replicate in cells derived from hosts so widely separated phylogenetically as mosquito and man. Much of the work which we have done involves the study of mutant or variant cells and viruses. Host-range or host-restricted viral mutants are proving to be of special interest. For example, if a wild-type,

332

VICTOR STOLLAR TABLE I1

SINDBIS VIRUS^ Taxonomy Family Togaviridae, genus Alphavirus Other members of this genus: EEE, WEE, VEE, SFVb All of these viruses are mosquito transmitted All replicate in the cytoplasm Genome Single-stranded RNA of positive polarity 11,700 nucleotides RNA is polyadenylated a t 3' end, and capped and methylated a t 5' terminus Structure of the virion Enveloped, spherical particles with spikes on surface Approximately 65 nm in diameter Organization of glycoprotein subunits, and of nucleocapsid indicate an icosahedral structure Virion contains three proteins C Nucleocapsid protein, associated with viral RNA E l These are glycosylated envelope proteins that

+I

E2

make up the spikes.

,

a For details concerning the properties and replication of Sindbis and related viruses, the reader is referred to Schlesinger (1980) and Strauss et al. (1984). b Eastern equine encephalitis virus, western equine encephalitis virus, Venezuelan encephalitis virus, Semliki forest virus.

or standard, virus can multiply in two different cell types, A and B (or mosquito and human), and we are able to derive a mutant virus which is able to multiply in one of these cell types but not the other, then it must follow that these two cell types differ in some fundamental way crucial to viral replication. If we can identify the specific difference between the two cell types, we can learn something new about the conditions that are specifically required in the host cell for virus to replicate. Perhaps from such information, methods will ultimately be found which will make it possible to interrupt the cycle of viral transmission by their specific vectors.

111. ROLEOF THE HOSTCELL Two examples will be cited which illustrate an important influence of the host mosquito cell in determining the outcome of viral infection.

VECTOR SPECIFICITY

333

A . CPE(+) and CPE(-) Clones of Ae. albopictus Cells As noted above, in the first experiments involving the infection of uncloned Ae. albopictus cell populations, SV failed to produce any visible cell damage or even to interfere with cell growth, in spite of the fact that high yields of virus were produced. Later, however, after cells had been adapted to a defined medium and clonally derived populations were obtained, it became evident that the original cell population with which we were working was in fact heterogeneous and contained cells which differed markedly in their response to viral infection (Sarver and Stollar, 1977). Certain clones showed extensive CPE, which was generally marked at 34°C but minimal at 28°C (Fig. 1). Other clones showed no CPE while still others showed intermediate effects. No significant difference was found, however, between the 24-

FIG. 1. Light microscopy of Ae. albopictus C7 cells, a CPE(+) clone, 26 hours after infection with SV. CPE was very marked at 34°C; at 28°C only minimal changes were seen relative to the control or uninfected culture. A typical CPE(-) cell culture 24 or 26 hours after infection with SV, would be indistinguishable from a mock-infected culture (see also Fig. 5).

334

VICTOR STOLLAR

hour yields of infectious virus from the CPE(+) and CPE(-) clones. These cells thus provide an unusual system in which different clones make similar amounts of virus but vary in their ability to survive virus infection. Such a system is clearly very attractive for exploring the mechanisms whereby cells are killed following infection with SV. Several other features of the SV-infected Ae. albopictus cells are worthy of note. Although there was no difference in the 24-hour yields of infectious virus between the CPE(+) and CPE(-) cells, it was consistently observed that the amount of viral RNA synthesized (as mea-

CI AIS INFECTED

l145

t

/

37eC

13 12 II f

e

10

x 9 I

48 7

6 5 4

3 2 I

HOURS POSTINFECTION

FIG.2. Virus-directed RNA synthesis in SV-infected Ae. albopictus C7 [CPE(+)] and AIS C3 [CPE(-)] cells at 28, 34, and 37°C. Two hours after infection cultures were treated with actinomycin D (5pg/ml); at 3 hours [3Hluridine was added (final concentration 10 pCi/ml). At the indicated times duplicate plates were harvested, and incorporation of [3H]-uridine into TCA-precipitable material was determined. ( 0 )Infected C7 cells, temperature as indicated in figure; (A)infected AIS C3 cells at 28,34, and 37°C; all mock-infected C7 and AIS C3 cells at 28, 34, values were within the range shown; (0) and 37°C. (From Sarver and Stollar, 1977.)

335

VECTOR SPECIFICITY

I

AIS N

54-

32-

I2

4

6

8 10 12 14

FIG. 3. Protein synthesis in infected and mock-infected C7 [CPE(+)I and AIS N [CPE(-)] Ae. albopictus cells. Cells were infected with SV or mock-infected and then incubated at 34°C; at the indicated times the cultures were pulse-labeled for 30 minutes with [3Hlleucine, following which incorporation into TCA-precipitable material was measured. (From Sarver, 1978.)

sured by incorporation of [3Hluridine into acid-insoluble material) in the CPE(+ cells was severalfold higher than in the CPE(-1 cells (Fig. 2). Furthermore, in the former the rate of viral RNA synthesis was higher at 34°C than at 28"C, whereas no such temperature effect was seen in the CPE(-) cells. Finally, these initial experiments demonstrated that host RNA and protein synthesis were markedly decreased in infected CPE(+) cells but little, if at all, in the CPE(-) cells (Fig. 3) (Sarver, 1978). Tooker and Kennedy (19811, using Semliki Forest virus and cell clones derived from the same cell population as our clones, obtained results somewhat different from ours in that, with a few exceptions, virus yield was considerably higher in CPE(+) cells than in CPE(-) cells. We were therefore prompted to examine the time course of virus production in CPE(+) and CPE(-) cells more closely (Fig. 4A). As before, we found in our more recent experiments (Tatem and Stollar, 1986) no difference between the 24-hour yields of virus from CPE(+) and CPE(-) cells. We did observe, however, that viral replication in the CPE(+) C7 OuaR cells proceeded more rapidly during the early period after infection (i.e., up to 10 hours) and then tended to level off,

VICTOR STOLLAR

"f

HOURS POST INFECTION

FIG.4. Growth curves of SV in parental and hybrid Ae. albopictus cells. Cells were infected with SV a t an input multiplicity of 100 PFU/cell and incubated a t 34°C. Samples of medium were harvested at the indicated times and titrated by plaque assay on chick embryo fibroblast cultures. (A) Parental cells: C7 OuaR cells are CPE(+); Ama 1811 cells are CPE(-). (B)Five different hybrid cell lines generated by fusion of C7 OuaR and Ama 1811 cells. (From Tatem and Stollar, 1986.)

whereas in the CPE(-) Ama 1811 cells replication was slower in the earlier period, but showed less of a tendency to plateau within the first 24 hours after infection. This difference could be expressed semiquantitatively by the increase in viral titer between 4 and 8 hours after infection. In CPE(-) cells the increase was about aO-fold, whereas in CPE(+) cells it was about 150-fold. The more rapid rate of viral replication during the early time period in CPE(+) cells is consistent with our earlier finding of a higher rate of RNA synthesis in these cells during this same time period. Although we still do not know the biochemical basis for the virus-induced cell injury in CPE(+) cells, our results do show that the CPE(+) phenotype in Ae. albopictus cells is associated with both a n increased rate of viral RNA synthesis and an increased rate of viral replication during the first 10 hours after infection. Recent experiments involving fusion of CPE( + ) cells with CPE( -1

VECTOR SPECIFICITY

337

cells and the selection of hybrids (Tatem and Stollar, 1986) have indicated that the CPE(+) phenotype is dominant in the hybrid cells (Fig. 5). In these experiments, ouabain-resistant CPE(+) cells were fused with a-amanitin-resistant CPE(-) cells and the hybrids selected in medium containing both ouabain and a-amanitin. Of eight hybrids isolated, all were CPE(+). Like the parental CPE(+) cells, each of four hybrids tested supported a rapid rate of replication during the 4- to 8hour period after infection (Fig. 4B). These experiments with CPE(+) and CPE(-) cells lead us to conclude that some property of the host cell plays an important role in regulating the rate of viral RNA synthesis and secondarily that of viral replication. Although it seems quite likely that the increased level of viral RNA synthesis in CPE(+) cells is causally related, in some way, to the induction of CPE, there is little information about possible mechanisms which might be involved. It is of interest, nevertheless, to address the question of how the host cell could affect the rate of viral RNA synthesis. Two possibilities can be considered. (1)Since the host cell provides the substrates needed for the synthesis and posttranscriptional modification of viral RNA, the level of these substrates in the infected cell could affect the rate of viral RNA synthesis. (2) Alternatively the host cell may influence the level of viral RNA synthesis by directly regulating the expression of the viral RNA polymerase. With respect to the first possibility, there are five low molecular weight substrates known for the enzymes encoded by SV: ATP, GTP, UTP, CTP, and S-adenosylmethionine (ado met). ATP, GTP, UTP, and CTP are required by the viral RNA polymerase; GTP is also required by the capping enzyme, and ado met by the RNA cap methylase. It is assumed that each of these enzyme activities is carried out by a viruscoded protein (see Cross, 1983, re specificity of the methylase). According to one possible hypothesis, if the level of one or more of these key metabolites is limiting in CPE(-) cells, this would slow the rates of viral RNA synthesis and viral replication, and thus prevent the appearance of CPE. Although ado met is not directly involved in RNA synthesis, without sufficient ado met, progeny viral RNA molecules in the infected cell would not be methylated, and therefore would not be translated; the nonstructural viral proteins which are involved in the synthesis and modification of viral RNA would not be made, and RNA synthesis would therefore soon cease. With respect to the SV RNA polymerase, two possibilities can be envisaged by which the host cell could affect its function. First, as with polio virus (Dasgupta et al., 1980), there may be a host protein subunit required for viral RNA polymerase activity; if so, then the level of

FIG.5. Light microscopy of cell monolayers 24 hours after infection with SV at 34°C. C7 OuaR cells are CPE( + ); Ama 1811 cells are CPE(-). H-307 and H-309 are two different hybrid cells obtained by fusion of C7 OuaR and Ama 1811 cells; both H-307 and H-309 as well as the other hybrid cells tested are CPE(+). (From Tatem and Stollar, 1986.)

VECTOR SPECIFICITY

339

polymerase activity would be influenced by the availability of that host protein. Alternatively, if host proteolytic enzymes are needed to cleave any of the polypeptide precursors to the four nonstructural proteins (however, see Strauss et al., 19841, one or more of which would constitute the viral RNA polymerase, the activity of these proteolytic proteins could vary in the different cell types, and thus influence the amount or activity of viral enzyme generated, which would be available to synthesize viral RNA. Although any of the above possibilities would provide a reasonable explanation for how the host cell might regulate viral RNA synthesis, there is no evidence favoring one or another hypothesis. As an aside, it should be noted that with two exceptions (Mims et al., 1966; Lam and Marshall, 1968; Houk et al., 19851, the replication of arthropod-borne viruses in arthropod vectors has not been associated with any cytopathic effect or cell destruction. Indeed for the virus, cell destruction sufficient to impair viability of the vector or of certain tissues would be self-defeating. Interestingly, in one example where cell damage has been documented in the vector (Ae. aegypti females infected with Semliki forest virus showed cytopathological changes in the salivary glands) the ability to transmit virus by bite was reduced (Lam and Marshall, 1968). Although the cytopathic effect seen in cultured mosquito cells would therefore appear to have no general counterpart in the intact vector, our studies have clearly shown that the host cell can strongly influence the level of viral RNA synthesis. If such effects and variations in the levels of viral RNA synthesis also occurred in individual organisms of a given vector species they could affect the amount of virus produced and hence the efficiency with which an individual mosquito or strain of mosquitoes would function as a vector.

B . Response to Methionine Deprivation A second manner in which an important host effect on the outcome of infection can be demonstrated is by depriving cells of methionine (Stollar, 1978). When BHK or chick cells were infected with SV and incubated in methionine-free medium, normal yields of virus were obtained. In contrast, when SV-infected Ae. albopictus cells were maintained without methionine. no viral RNA synthesis could be detected and there was no production of infectious virus (Fig. 6). Any explanation for the effect of methionine deprivation must begin with a consideration of the role of methionine in cellular metabolism. First, methionine is required as an amino acid building block for proteins. Second, methionine is required for the synthesis of S-adenosyl-

340

VICTOR STOLLAR

I

3

6

d9

[ Methionine]

12

15

0

pg/ml

FIG.6 . Yields of SV from chick embryo fibroblasts (CEF), BHK cells, and C7 Ae. albopictus cells as a function of the methionine concentration in the medium. Cells were infected with SV (10PFU/cell) and incubated at 28°C in medium containing the indicated concentrations of methionine. After 24 hours samples of medium were harvested and assayed by plaque titration on CEF. (From Stollar, 1978.)

methionine (ado met) which in turn serves two functions. It is the methyl donor for nearly all methylation reactions in the cell (Cantoni, 19751,including methylation of viral and cellular mRNAs; in addition, it is an essential precursor involved in the synthesis of the polyamines spermine and spermidine (Tabor and Tabor, 1984).Even under conditions of methionine deprivation, the reduction in overall protein synthesis did not appear to be sufficient to account for the total inhibition of viral replication in infected mosquito cells; also cycloleucine, which inhibits the synthesis of ado met, potentiated the effect of methionine deprivation not only in mosquito cells but also in chick embryo fibroblasts (Stollar, 1978).These findings lead us to believe that the lowering of ado met is the critical factor with respect to the inhibition of viral replication in methionine-starved Ae. albopictus cells. Although polyamines are necessary for the replication of Semliki Forest virus (Tuomi et al., 1980),we favor the idea that the critical consequence of a lowered level of ado met is the inability to methylate viral RNAs, and that this is the reason why viral replication is inhibited in methionine-deprived Ae. albopictus cells. Evidence consistent with this view has come from studies of Ae. albopictus cells infected with vesicular stomatitis virus (VSV).Under conditions of methionine

VECTOR SPECIFICITY

341

FIG. 7. Effect of methionine starvation on host cell and virus-specific polysomes. Cytoplasmic extracts of VSV-infected and mock-infected Ae. albopictus cells maintained in media without serum, and with or without methionine, were prepared and fractionated by sucrose gradient centrifugation. Gradient fractions were collected and the absorbance was recorded by continuous monitoring at 254 nm (the direction of sedimentation was from right to left). The VSV-infected cells were labeled from 4 to 6 hours after infection in the presence of actinomycin; the mock-infected cultures were not treated with actinomycin. The Roman numerals indicate polysomal and nonpolysomal pools from which viral RNA was extracted. (From Gillies and Stollar, 1981.)

deprivation VSV, like SV, was unable to replicate in these cells (Gillies and Stollar, 1981). With VSV, however, some viral RNA synthesis was detected under these conditions, especially when cells were treated with actinomycin. This RNA synthesis,may represent mainly primary transcription, i.e., synthesis of messenger RNA transcribed off the parental genome. Significantly, these RNAs were largely non-polysome-associated (Figs. 7 and 8) and in the presence of S-adenosylhomocysteine were nontranslatable in a reticulocyte lysate (Fig. 8). This inhibition of translation by S-adenosylhomocysteine indicated that these viral RNAs could be translated only if they were methylated in the reticulocyte translation system. Both the nonassociation with polysomes and the requirement for methylation in the reticulocyte lysate could be explained by a failure to methylate VSV mRNAs in cells deprived of methionine. If it is indeed the level of ado met that explains the differences between the effects of methionine deprivation in mosquito and vertebrate cells, there must be significant differences either in the pool size

342

VICTOR STOLLAR

FIG. 8. In uitro translation of polysomal and nonpolysomal RNA from VSV-infected Ae. albopictus cells maintained in medium with or without methionine. RNA was extracted from the pools shown in Fig. 7 and translated in an mRNA-dependent reticulocyte lysate containing [3sS]methionine. The labeled translation products were then analyzed by electrophoresis through an SDS-containing polyacrylamide gel. Because viral RNA synthesis is decreased in methionine-starved Ae. albopictus cells, in order to detect the viral protein bands a larger volume of the FWA preparations was used for the pool I11 and pool IV reactions than for the pool I and I1 reactions (fivefold more). For the pool I11 and pool IV reactions + and - indicate the presence or absence of S-adenosylhomocysteine in the in uitro protein synthesizing system. When methionine was present in the medium, the bulk of the viral mRNA was in the polysomal fraction, pool 11,but in the absence of methionine most of the viral RNA was in the nonpolysomal fraction, pool 111. (From Gillies and Stollar, 1981.)

VECTOR SPECIFICITY

343

of ado met or in the rates of synthesis, turnover, and utilization in the two cell types. Little is known concerning the metabolism of methionine and ado met in mosquitoes or mosquito cells. Clearly, however, the availability of methionine in certain cells of the mosquito could play an important role in determining the efficiency with which virus can replicate. IV. ROLE OF THE VIRUS In this section I shall describe work with viral mutants which relates to how the properties of a virus can determine its ability t o replicate in cells derived from either the arthropod vector or a vertebrate host. A . Host Range Viral Mutants Which Are Restricted in Ae. albopictus Cells Following mutagenesis of SV with 5-azacytidine, Kowal and Stollar (1981) isolated two viral mutants which at 34°C showed a marked restriction in Ae. albopictus cells; both mutants were able, however, to replicate normally in chick cells. The host restriction was demonstrated both by efficiency of plaquing assays on the two cell types and by yield experiments. When titered on chick cells at 34.5"C, stocks of these two mutants, SV mutants 35 and 58, gave titers comparable to that of standard virus (SVstd)(Table 111). In contrast, when they were assayed at the same temperature on mosquito cells, the titers were, in one case, close to 106-fold and in the other case at least 107-foldlower than that of sv,,d (Table 111). One-step growth curves also revealed a clearcut difference between sv,,d and SV mutants 35 and 58 (Fig. 9). Chick embryo fibroblasts (CEF) and Ae. albopictus cells (clone C7) produced between lo3 and lo4 PFU per cell by 16 hours after infection with SV,,; similar amounts were produced in CEF infected by SV mutants 35 or 58, at 34°C. However, when Ae. albopictus C7 cells were infected with the mutant viruses, the yield at 16 hours postinfection was no more than 1PFU/cell. The subsequent demonstration that the infectivity of RNA from SV mutants 35 and 58 was markedly reduced in Ae. albopictus cells, compared to the infectivity registered on CEF monolayers, indicates that the block to replication of these mutants must be at some step later than adsorption. As already noted, the initial experiments with SV mutants 35 and 58 were done at 34.5"C, a temperature at which SV,, replicates well in both vertebrate and mosquito cells. Although the mutant viruses replicated and plaqued well in vertebrate cells at this temperature, at 40°C

TABLE III EFFICIENCY OF PLATING OF SINDBIS Vmus MUTANTS ON CHICKEMBRYO AND Ae&s albopictus CELLS~ Infectivity (PFUlml) when assayed on

Virus stock

SV& SV mutant 35 SV mutant 58 SV 35Rlc

CEF (34.5"C) 4.2 x 6.0 x 2.5 x 4.3 x

109 109 109 10'0

Ae. albopictus (34.5"C)

CEF (40°C)

1.4 x 108 1.7 x 104 <2.5 X 102 4.5 x 108

3.4 x 109 7.0 x 105 2.0 x 1046 2.5 x 10'0

Eficiency of plating

Ae. albopictus (34.5"C)EEF (34.5"C) 3.3 2.8
From Kowal and Stollar (1981). These were minute plaques. c SV 35R1 is a revertant of mutant 35 selected on the basis of its ts+ character. a

b

x x x x

10-2 10-6 10-7 10-2

CEF (40"C)ICEF (34.5"C) 0.81 1.2 x 10-4 8.0 x 10-6 0.58

345

VECTOR SPECIFICITY c

4

8

12

16

20

24

4

8

12

HOURS AFTER

16

20

24

hO*

SVOl 58

4

e

I2

I6

o;

24 '10'2

INFECTION

FIG. 9. Replication of SV host range mutants in cultures of chick embryo and mosquito cells at 34.5"C. Cultures of CEF (01,C7 Ae. albopictus cells (A),and AIS N Ae. albopictus cells) . ( were infected with SV,M, SV mutant 35, or SV mutant 58 and incubated a t 34.5"C. Samples of medium were taken at the indicated times and titrated by plaque assay on CEF at 34.5"C. The results are expressed as PFU released per cell. (From Kowal and Stollar, 1981.)

both their growth (Table IV) and plaquing efficiency (Table III) in vertebrate cells were severely impaired. Thus in addition to being host restricted, they were also temperature sensitive (ts) in the permissive cell (in fact, the temperature-sensitive property was also evident in Ae. albopictus cells; compare the yields at 28 and 345°C in Table IV). A similar association between host restriction and ts phenotypes has been described with mutants of VSV (Simpson and Obijeski, 1974). The fact that when ts revertants of SV mutant 35 were selected they were no longer host restricted (Kowal and Stollar, 1981)strongly suggests that the same lesion is responsible for both phenotypes. This last observation facilitated further study of these mutants, since characterization of a ts phenotype which is conditionally lethal is easier than that of a mutant which is simply unable to grow in a given cell type under any conditions. Measurement of viral RNA synthesis in infected cells demonstrated +

346

VICTOR STOLLAR TABLE IV

REPLICATION OF MUTANTS OF SINDBIS VIRUSIN CHICKEMBRYOFIBROBLASTS AND Ae. albopictus CELLSA T DIFFERENT TEMPERATURESQ Titer of infectious virus (PFU/ml) a t 16 hours Virus

Cells

1 hour

28°C

34.5"C

SVSt.3 SV mutant 35 SV mutant 58

Ae. albopictus Ae. albopictus Ae. albopictus

2.0 x 105 1.2 x 105 8.6 x lo4

6.8 x lo8 7.5 x 107 3.1 x 107

2.7 x 108 3.3 x 105 2.6 x 105

Titer at 6 hours

SVstd

SV mutant 35 SV mutant 58

CEF CEF CEF

1.0 x 106 9.5 x 104 9.8 x 104

34.5"C

40°C

1.6 x loy 1.1 x 109 9.0 x 108

2.1 x 109 9.8 x 105 6.1 x lo6

The titer of infectious virus was determined by plaque assay on CEF monolayers a t 28°C. From Kowal and Stollar (1981). Q

that SV mutants 35 and 58 were RNA-; i.e., they were unable to make viral RNA under restricted conditions, in CEF at 40°C or in Ae. albopictus cells at 345°C. Taking advantage of the ts phenotype, complementation tests were carried out, which showed that both mutants could be assigned to group F of the SV ts mutants (Table V). This is one of four groups containing RNA - mutants. Interestingly, another group F mutant, ts6, also showed some host restriction in Ae.albopictus cells (Renz and Brown, 19761,but the effect was much less marked than was the case with SV mutants 35 and 58. At the present time it is still not possible to name the precise function of the gene product which is defective in the group F mutants, nor is it possible to associate with any certainty this gene product with any one of the four nonstructural proteins, NSP1, 2,3,or 4 (Strauss et al., 1984).One can therefore only speculate as to the nature of the defect in SV mutants 35 and 58. We have suggested, however, that if a functional SV RNA polymerase requires a host protein subunit in addition to a virus-coded protein, an alteration in the virus-coded protein could adversely affect the binding of the host cell protein. The differential effect in the different cell types could then be attributed to slight differences between the host protein in the mosquito and verte-

347

VECTOR SPECIFICITY TABLE V VIRUSMUTANTS35 AND COMPLEMENTATION ANALYSISOF SINDBIS 58 AND SINDBISVIRUSTEMPERATURE-SENSITIVE MUTANTS~ Complementation POUP A, RNAB, RNAc , RNA+ D, RNA+ E, RNAf F, RNAG, RNA-

Complementation indexb

Virus mutant

Mutant 35

Mutant 58

ts4 tall ts2 tslO ts20 ts6 ts7 Mutant 58

51 43 130 ntc 41 1.4, 1.2, 0.W 170 0.85, 0.6

38 82 23 nt 23 0.5, 3.5, 2.1 1150 -

a The complementation assays were performed in CEF cells at 40"C, and virus was titered on CEF monolayers a t 28°C. From Kowal and Stollar (1981). b The complementation index is defined as the yield of infectious virus from cultures infected with the indicated pairs of mutants, divided by the sum of the yields of infectious virus from cultures infected with each mutant alone (Burge and Pfefferkorn, 1966). c nt, Not tested. d Results of multiple determinations.

brate cells such that the binding of the host cell protein was impaired in the former but not in the latter. Similarly, one might postulate that the same alteration in the viral protein which leads to a defective interaction with the host protein in Ae. albopictus cells at 34°C also prevents efficient binding of the host protein in CEF at 40°C. An alternate explanation for the host restriction in SV mutants 35 and 58 might center about an altered virus-coded RNA cap methylase. First, a defective methylase has been associated with a host range phenotype in the case of certain VSV mutants (Horikami and Moyer, 1982). Second, we have already described above how mosquito and vertebrate cells differ in their ability to support SV replication under conditions of methionine deprivation, and we have suggested that this difference reflects variations in the metabolism of ado met in the two cell types. Although the precise nature of the defect in SV mutants 35 and 58 is still unclear, these mutants should eventually prove extremely useful in identifying specific host cell functions and proteins which are necessary for viral replication.

348

VICTOR STOLLAR

B . A Viral Mutant Which Is Restricted in Vertebrate Cells The Sindbis virus mutant SVap15/21was obtained following plaque purification of a virus population which had been serially passaged 15 times in Ae. albopictus cells at 345°C (Durbin and Stollar, 1984). There was, however, no significant difference between the properties of SVap15/21and the uncloned population, SVap15,from which it was derived. Figure 10a demonstrates the growth restriction of SVap15/21in BHK cells. At 34.5"C the yield of the mutant was less than 1%of that of sv,td and at 40°C there was no detectable yield of progeny virus at all. In contrast, in Ae. albopictus cells at 34.5"C, SVap15/21grew just as well as the standard virus, SV,, (Fig. lob). The two viruses could also be distinguished on the basis of plaque size on CEF at 34.5"C. sv,td plaques measured 2-3 mm in diameter, whereas SVaplS/21plaques were less than 0.5 mm. At 40°C SVap15/21 did not generate any plaques on CEF (PFU/ml at 40°C + PFU/ml at 3 4 . 5 ~=
10'

lo3

.

10'

r

102

.

lo3

-

10'

-

ld . A W

-I wI

v3 U

b

l

2 101 -

-

U.

n

n lo-'

-

10-2

I

1

-

10-1

-

0"

1

. 2

1

6

B

10

12

4

6

12

16

2022

HOURS POST INFECTION

FIG.10. Replication of SVap16/21 and SV,M in (a) BHK and (b) Ae. albopictus mosquito cells. Cells were infected with virus at a multiplicity of 10 PFU/cell. BHK cells were incubated at 34.5 or 40 "C and Ae. albopictus cells at 34.5"C. Samples of medium were taken at the times indicated and assayed by plaque titration on CEF. (0) SVap16/21 SVap16/21at 40°C; (0)SV,M at 34.5"C; (M)SV,u at 40°C. (From Durbin a t 345°C; (0) and Stollar, 1984.)

349

VECTOR SPECIFICITY TABLE VI COMPLEMENTATION OF SVaplSial WITH RNA+ SV ts MUTANTSQ Complementation group

Virus mutantb

C

ts2 tslO ts20

D

E

Complementation index in mixed infection with SVaplS/21c

3.7;6.0 66;49 0.9; 0.6

a Complementation experiments were performed in CEF cultures at 39.5”C,and virus yields were assayed on CEF at 28°C. Modified from Durbin and Stollar (1984). 6 Mutants of Burge and Pfefferkorn (1966). c The complementation indices are from two separate experiments.

chick cells at 34.5 and 40°C showed no impairment relative to SV,,infected cells. Thus the block in replication was late. Complementation tests with SVap15/21and representative RNA ts mutants (Table VI) enabled us to assign it to group E, of which the prototype mutant is ts20. ts20, at the nonpermissive temperature, is unable to cleave the precursor protein, PE2, and generate the envelope protein E2 (Jones et al., 1974). Examination by gel electrophoresis of the viral glycoproteins, pulselabeled in BMK cells with [35S]methionine,revealed that PE2 encoded by SVap15/21had a molecular weight about 3000 greater than the PE2 encoded by SV,, (Fig. 11)(Durbin and Stollar, 1984). In contrast with what has been observed with ts20 under nonpermissive conditions, during a chase at 34.5”C, SVap15/21 PE2 was processed to E2 at the same rate as was sv,td PE2. However, as was the case with PE2, SVap15/21E2 was larger than the Sv&d E2 (Fig. 11).Again, the difference in molecular weight was about 3000. Treatment of infected BHK cells with tunicamycin showed that the PE2 “aglycoproteins,” made in cells infected with sv,td or with SVap15/21,were indistinguishable in size (Fig. 12, lanes C and D). This finding indicated that the increased size of the SV ap15/21 (P)E2 was due to an increase in glycosylation. When the two viruses were grown in mosquito cells, E2 of SVap15/21was again larger than that of SVSM.Thus, the increase in size of the mutant (P)E2 was observed in both vertebrate and mosquito cells even though the mutant virus was restricted only in the former and was able to grow normally in the latter. Fluorescent antibody experiments, using antibody to whole virions and nonpermeabilized cells, indicated that the viral antigens were reaching the cell surface; E l and PE2 exist as a complex in infected cells +

350

VICTOR STOLLAR

FIG.11. Pulse-chase labeling of SVap15/21 viral proteins. BHK cells were infected (100 (lanes H-N) and incubated at 34.5"C. At 6 PFUkell) with SV,d (lanes A-G)or SVap15/11 hours after infection cells were labeled for 15 minutes with [36Slmethionine, and then either prepared directly for electrophoresis (lanes A and H)or first chased with medium containing an excess of unlabeled methionine for 15 minutes (lanes B and I), 30 minutes (lanes C and J),45 minutes (lanes D and K), 60 minutes (lanes E and L),75 minutes (lanes F and M),or 90 minutes (lanes G and N). Proteins were alkylated and electrophoresed without reduction of disulfide bonds in order to optimally resolve E l and E2. The polypeptide marked B is the binary envelope protein; i.e., uncleaved PE2-El and C is the capsid protein. (From Durbin and Stollar, 1984.)

(Rice and Strauss, 1982). Thus, the block to replication of SVap15/z1in vertebrate cells does, in this respect, resemble what has been described for ts20 at the nonpermissive temperature (Jones et al., 1974). The findings above indicate that in vertebrate cells the growth of SVap15/21is blocked at a very late stage, perhaps at assembly. This suggestion was confirmed by electron microscopy of infected chick cells (Fig. 13). Whereas in SV,,-infected cells virions were seen outside the cell and in various stages of budding, in SV,p15,zl-infected cells there was little or no evidence of budding. Instead, large numbers of nucleocapsids were seen concentrated just beneath the plasma membrane. It appeared therefore that the increased glycosylation of PE2 in

VECTOR SPECIFICITY

351

FIG. 12. Effect of tunicamycin on synthesis of viral proteins. BHK cells were infected with SV,M (lanes A and C) or SVeplS,~l (lanes B and D)and incubated at 345°C. Four hours after infection tunicamycin was added to the cultures shown in lanes C and D. Cell cultures were labeled with [3W]methionine between 6 and 6.5 hours postinfection; lysates were then prepared and analyzed by SDS-polyacrylamide gel electrophoresis. (From Durbin and Stollar, 1984.)

SV,,,,,,,-infected cells in some way interfered with the final stage of viral morphogenesis, envelopment. How might increased glycosylation of E2 interfere with its function? Either its function might be compromised simply by the increase in the total mass or charge of the carbohydrate or alternatively it is the glycosylation at or very close to a critical site on PE2 which is deleterious. The question then arises why the glycosylation of E2 should be harmful in vertebrate cells but not in mosquito cells. A reasonable explanation for this differential effect is suggested by the work of Hsieh et al. (1983)who showed that although mosquito cells, like vertebrate cells, synthesize asparagine-linked oligosaccharides, they do

VECTOR SPECIFICITY

353

not make the complex type of oligosaccharides. Instead, at sites on SV glycoproteins where complex type oligosaccharides are added in vertebrate cells, only a small neutral oligosaccharide is added when the virus infects mosquito cells. Thus, depending on the host cell in which the virus replicates, there are profound differences in the chemical structure of the oligosaccharide component of the SV glycoproteins. Simply put, glycosylation in mosquito cells is not the same as glycosylation in vertebrate cells. Although we still do not understand precisely how hyperglycosylation interferes with the function of E2, the work of Hsieh et al. (1983)is helpful in explaining why increased glycosylation can be deleterious in vertebrate cells but not in mosquito cells. The results presented above indicate that four properties can be associated with SVap15/21.It is host restricted, it is temperature sensitive, it makes a P(E2) protein which is hyperglycosylated, and it produces small plaques. What evidence is there that the hyperglycosylation is causally related to any or all of the other three properties of SVap15/21?The study of viral revertants proved useful in getting at this question. Although our stocks of SVap15/21(these stocks are grown in mosquito cells) contained no detectable ts or large-plaque revertants (PFU/ml at 40°C + PFU/ml at 34.5% = <1.3 x by passaging SVap15/21 twice on chick cells at 34.5”C,stocks can be obtained in which largeplaque and ts revertants make up roughly 20-50% of the virus population (Table VII). (After further passage these revertants become the predominant population.) By taking such a stock and picking plaques on mosquito cells (although on chick cells SV,, and SVap15/21can readily be distinguished by plaque size, on mosquito cells their plaques are indistinguishable), it is possible to obtain a random group of cloned virus populations, about half of which may represent revertants. SVap15/21was therefore passaged twice on chick cells after which 26 plaques were picked from infected mosquito cell monolayers. Each of the cloned virus populations was then characterized with respect to the four properties associated with the SVap15,21mutant (Table VIII). An excellent correlation was found between reversion to the PE2 size characteristic of SVStdand loss of the host restriction in chick cells. Of 16 isolates tested that produced SV,,-size PE2,all produced yields in chick cells greater than 10% of the yield of SV,,, i.e., they were no +

+

FIG. 13. Electron micrographs of SV-infected chick cells. Cells were infected with 100 PFU of SVd (A) or SVap16121 (B)per cell and incubated at 346°C. At 7.5 hours postinfection, cells were fixed and prepared for election microscopy. (From Durbin and Stollar. 1984.) Bar = 100 nm.

354

VICTOR STOLLAR TABLE VII

REVERSION OF PLAQUE SIZEAND ts PHENOTYPES WITH PASSAGE OF SVap15i21u Yield (PFU/ml) at assay temperature of Chick cell passage level OC

1 2 3

34.5"C

40°C

Small plaques

Large plaques

1.6 x 109 5.0 x 107 8.5 x 106 (Not detectable)

(Not detectable) 1.2 x 106

8.4 x 106 3.0 x 108

Small plaquesb

<2 x 6.8 x 3.1 x 1.1 x

104 105 106

108

SVap15/21 passaged at 34.5"Con chick cells. From Durbin and Stollar

(1984). b c

Only small plaques were produced at 40°C by t s f revertants. Stock SVap15/21grown in mosquito cells.

longer host restricted; and of eight isolates tested which still produced the large PE2, seven produced yields less than 1%of that of SV,,, and in the other case the yield was about 2-3%. Thus these isolates retained the host restriction phenotype. Of 15 large-plaque and 10 ts+ revertants, all had also reverted to the SVStd-sizePE2. There were two revertants, numbers 14 and 18,which reverted with respect to the PE2 size and host restriction but remained ts and produced small plaques. Three other isolates, numbers 10, 13, and 15, reverted with respect to three of the properties, but remained ts. Thus, among these isolates, of the various phenotypic properties examined, that which correlated best with loss of the host restriction was the reversion to the SV,,-size PE2. As a follow-up to the biological characterization of SVap15,21and its revertants, nucleotide sequence determinations were performed (Durbin and Stollar, 1986). First, it was demonstrated that SV,,, which was derived from SV,, and which is the strain of SV we use in our laboratory, has three potential glycosylation sites on P(E2) instead of two as is the case with SV,, (Fig. 14). This confirms the work of Cancedda et al. (1981) who found earlier by examination of viral glycoproteins and glycopeptides that E2 of SV&d (referred to by them as SV,) bears, instead of two N-linked glycans as is the case with SV,,, a third N-linked glycan. Nucleotide sequencing showed that this new glycosylation site was generated by the change of G-9324 to A, converting the codon in E2 for aspartate-232 to a codon for asparagine (Durbin and Stollar, 1986). Since the usual signal for asparagine-linked glycosylation is Asn-X-Ser/Thr (Hubbard and Ivatt,

VECTOR SPECIFICITY

355

TABLE VIII

PHENOTYPIC ANALYSISOF ISOLATESOBTAINED AFTER PASSAGE OF SVaD15/21ON CHICKCELLS@ Isolate number

Yield (PFU/ml)C

Plaque size

Temperature sensitivity

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 SV,~

2.6 x 107 3.4 x 107 (NT) 1.0 x 107 4.4 x 104 2.0 x 107 6.3 x 104 4.0 x 107 1.4 x 107 6.0 x 106 6.2 x 104 7.2 x 104 1.8 x 107 6.3 x 108 8.0 x 106 1.2 x 107 2.8 x 107 2.2 x 109 5.1 x 104 1.2 x 107 4.6 x 104 4.7 x 104 1.4 x 107 4.2 x 104 1.2 x 106 3.6 x 107 4.6 x 107

L

(NT) ts (NT) ts+

std

W"

mut

L L L

S

+

PE2 size Std Std Std

L

ts

mut

L L L

ts ts+ ts ts ts

mut mut

S

S S L

S L L L

S S

L S S L

S S L L

+

+

ts ta ta ts ts+ ts ts ts ts+ ts ts ts+ ts

+

ts

ts ts

+

+

Std

Std Std

Std

Std Std Std Std Std

std mut Std

mut mut Std

mut mut std Std

a From Durbin and Stollar (1984).There is an error in Table 4 of Durbin and Stollar (1984),from which this table is taken. Isolate number 24 should have a mutant-size, not a standard-size PE2. This has been corrected in the table above. b SVap15/21was passaged twice on chick cells as described in the text. The resultant stock was plaqued at 34.5"C on Ae. albopictus cells and 26 plaques were picked at random. Each plaque isolate was then grown into a small stock, also at 34.5% in Ae. albopictus cells. The titers of these stocks ranged from 3 x 108 to 1 x 109 PFU/ml, as assayed on chick cells at 34.5"C.Each clonally derived stock was characterized with respect to growth, plaque size (L, large; S, small), and temperature sensitivity on chick cells (ts, temperature sensitive; ts+ , not temperature sensitive), and with respect to PE2 size in BHK cells (std, standard size; mut, mutant size). c Yield (PFU/ml) from CEF 6 hours after infection with 10 PFU/cell and incubation at 34.5"C.

356

VICTOR STOLLAR 196 232 2 5 318

c

n

i

I

I I

1

n

ap15/21

std

HR

FIG. 14. Map of glycosylation sites in glycoprotein E2. The hydropathicity of glycoprotein E2 (HR SP strain; sequence from Strauss et al., 1984) was analyzed by an IBM PC running a BASIC computer program based on that described by Kyte and Doolittle (1982). Hydrophobic and hydrophilic regions are represented by upward and downward displacements, respectively. The horizontal lines represent the 423 amino acid residue sequence of E2 (N terminus leftmost) and correspond to nucleotides 86319899 in the viral RNA nucleotide sequence. The two large peaks in the C-terminal region are thought to constitute the hydrophobic root anchoring the glycoprotein in the envelope (Rice and Strauss, 1982). Each vertical line designates a glycosylation site, and the number above each vertical line indicates the amino acid residue in E2 to which the carbohydrate is attached; the symbols below the numbers indicate the type of glycan added at each site in vertebrate cells: c for complex, n for neutral, and ? for uncertain (Mayne et al., 1985; Cancedda et al., 1981). As shown, SVHR,SV,a, and SVap15/21have two, three, and four glycosylation sites, respectively. The fourth glycosylation characteristic of SVap15/21occurs in a relatively hydrophobic region. (From Durbin and Stollar, 1986.)

1981),and since amino acid residue 234 is threonine, the change at residue 232 from aspartic to asparagine generated a glycosylation site. Turning to SVap15/21,conversion a t residue 9460 of the U (seen in SV,, and SV,,d) to C changed residue 277 of E2 from isoleucine to threonine (Table IX) (Durbin and Stollar, 1986). Since there is an asparagine at residue 275,the threonine at residue 277 generated a fourth potential glycosylation site on E2. This means that there are now SV variants available with two, three, or four potential glycosylation sites on E2 (Fig. 14).Furthermore, the work by Cancedda et al. (1981)and Durbin and Stollar (1984)indicated that all of these sites are in fact glycosylated. With respect to the third glycosylation site, that which is characteristic of Sv&d,there is no evidence that it affects replication or host range in any way. In contrast, the fourth glycosylation site, that which is characteristic of SVap15/21,clearly does affect

TABLE IX SUMMARY OF NUCLEOTIDE SEQUENCES AND PREDICTED AMINOACIDSEQUENCES AT THE ADDITIONAL GLYCOSYLATION SITEIN E2 OF SVap15/21 AND OF ITSDERIVED HR+ REVERTANTS"

Virus SVHR SVStd SVap15/21

Nucleotide sequenceb (9453-9461)

Amino acid sequenceb (275-277)

hr

gl

ts

sp

AAU GUA AUA AAU GUA AUA AAU GUA ACA

Asn Val Ile Asn Val Ile Asn Val Thr

X

X

X

X

Class I Revertant 2 6 16 17 20 23 26 4 8 9 Class I1 Revertant 10 13 15

U.. U.. U..

Class 111 Revertant 14 18

. . . . . . .U.

. . . . . . .U . . . . . . . .U . . . . . . . .U.

. . . . . . .u . . . . . . . .U. . . . . . . .U . ..A . . . . . . .. A . . . . . .

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

Ile Ile Ile Ile Ile Ile Ile

......

.u. . . . . . .

Lys Lys Ile

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

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

Tyr Tyr

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

......

. . . . . . .U .

. . . . . . .U .

Tyr

Phenotypesc

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

Ile Ile

X X X

x x

x x

a The nucleotide sequences of the indicated isolates were determined, as described by Both and Air (1979),using a synthetic DNA primer complementary to nucleotides 94879504.Throughout the approximately 200 nucleotides (9287-9486)determined by this procedure, the sequences of SVsa and all the isolates derived from it (i.e., SVap15/21and the 15 revertants) were identical except for the differences indicated in this table. The only difference in this region between the sequences of SV,M and its derivatives and the sequences of SVHR is the mutation at nucleotide 9324 (see text), accounting for the additional glycosylation site in E2 of SV,M (Cancedda et al., 1981). Except for these differences, the sequence agreed with that published for SVHR by Rice and Strauss (1981).From Durbin and Stollar (1986). b Where not explicitly indicated, the nucleotides and amino acids in the sequences of the revertants were similar to those in the corresponding positions of SVap15/21. c Where marked with an X the isolates are similar in phenotype to SVap15/21(hr, gl4, ts, sp). If unmarked, the isolate is similar to SVHRand SV.a (hr+, g12 or gl3, ts+, lp). The following phenotype descriptors are used. The host-restricted growth pattern typical of SVap15/21(Durbin and Stollar, 1984) is denoted hr, while revertants to the host range typical of SVsu are denoted hr . SV strains that encode E2 bearing two, three, or four glycans are denoted g12, gl3, and gl4, respectively. Temperature-sensitive virus strains are denoted ts; revertants to viability at 40°C are denoted ts+. The small-plaque phenotype typical of SVap15/21(minute plaques on chick cells at 34.5"C)is denoted sp, while revertants to the larger plaque size characteristic of SV,u are denoted lp. +

358

VICTOR STOLLAR

viral replication and morphogenesis but strictly in a host-dependent manner. Relevant sequences of 15 of the h r + revertants (revertants which were no longer host restricted) referred to above were also examined (Table IX) (Durbin and Stollar, 1986). In every case, the fourth glycosylation site, characteristic of SVap15/21,was lost either by change of Asn-275 to lysine, isoleucine, or tyrosine, or by change of Thr-277 back to the original isoleucine. The fact that all hr to hr revertants lost the glycosylation site at Asn-275 and not at any of the other three sites strongly supports the idea that it is the glycosylation at this specific site on PE2 which is deleterious to SVap15/21 rather than the total amount of carbohydrate. Interestingly, in three cases (revertants 10, 13, and 15 where the glycosylation site was lost by changing asparagine-275 to tyrosine) the host restriction was lost but the revertants remained temperature sensitive. Two other revertants, 14 and 18, which were h r + and had lost the fourth glycosylation site (they reverted to the SV,, sequence at amino acid residues 275-277 of E2), retained the ts and small-plaque phenotypes. In these revertants, however, in addition to losing the fourth glycosylation site in E2,C-9099 was changed to A, resulting in conversion of Arg-157 to serine. This change would therefore appear to be responsible for the ts and smallplaque properties of these isolates. In summary, the nucleotide sequence studies of SVap15/21and its revertants point strongly to a causal relationship between the hyperglycosylation phenotype (Gly4) and host restriction in vertebrate cells. In nature, a virus with such a phenotype would be expected to replicate well in the mosquito vector, but upon infection of a vertebrate host the transmission cycle would come to a dead end. +

C. A Viral Mutant Which Is Able to Replicate in Methionine-Starved Ae. Albopictus Cells In an earlier section, experiments were described which showed that SV replication in Ae. albopictus cells, but not in vertebrate cells, was dependent on the presence of methionine in the medium. This dependence was attributed primarily to a requirement for concentrations of ado met in the cell which would suffice for the methylation of viral RNA. If, as seems likely, the enzyme responsible for the methylation of SV RNA is virus coded (Cross, 19831, we postulated that it might be possible to select for viral mutants which produce an altered RNA methylase with an increased affinity for ado met, and would thus be able to replicate in methionine-starved Ae. albopictus cells. Accordingly,

VECTOR SPECIFICITY

359

lo8

0 2 4 6 8 10 12 14 16 18 Passage Number

FIG. 15. Titers of SV during passage on Ae. albopictus cells in medium containing a reduced concentration of methionine. Cells were infected with SV (1 PFUIcell) and then incubated at 34.5”C in serum-free medium containing 5 pA4 methionine (normal methionine concentration = 100 CLM). After 24 hours, virus was harvested and passaged without dilution on a second culture of Ae. albopictus cells. This procedure was continued for a total of 17 passages. Cells were incubated in medium containing 5 pkf methionine for the first 10 passages, and then in the absence of methionine for the remaining 7 passages. Virus from each passage was titered by plaque assay on CEF. The “passage 0” value refers to the titer of the starting virus stock,SV,u. (From Durbin and Stollar, 1985.)

SV,, was serially passaged without dilution in mosquito cells in medium containing only 5 pM methionine (normal medium contains 100 p M methionine). Such a reduction in methionine normally reduces the 24-hour yield of standard virus about 1000-fold. By seven passages it was evident that the level of virus was increasing (Fig. 15) compared to the low levels in the initial passages, and by the tenth passage the level of virus produced was similar to that obtained from cells maintained in 100 pM methionine. The virus was then passaged another seven times, but in methionine-free medium, generating the stock SV,,,,. Figure 16A illustrates the plaquing efficiency of this virus at different passage levels, and again demonstrates the loss of the methionine requirement with increasing passage level. Clones were obtained by plaque purification of s v L M 1 7 and characterized. The 24hour yield from Ae. albopictus cells of one of these clones, SV,,17,1?, was only 15-fold less in methionine-free medium than in normal medium (a reduction to 7.2 x lo8 PFU/ml from 1.1 x 1O1O PFU/ml). With SVStdunder similar conditions, there is typically greater than a 1000fold reduction in the final level of virus. Also, whereas the plaquing efficiency of SVStdon Ae. albopictus cell cultures showed a greater than a 1000-fold reduction with 20 p M in the overlay medium instead

360

VICTOR STOLLAR A

B

100

-

10'

E

L

/

I

103

I

10

100

I

10

100

[Methionine] in Agorose Overlay (pM)

FIG. 16. Plaquing efficiency of SVLMand SVsd as a function of methionine concentration in the agarose overlay medium. SV was assayed by plaque formation on mosquito cells, but with varying concentrations of methionine in the agarose overlay. (A) The numbers in parentheses indicate the passage level of the virus shown in Fig. 15. These virus stocks were uncloned. The plaquing efficiency of SVLMZwas very sensitive to the methionine concentration, whereas that of SVm17 was not affected even when the methionine was reduced from 100 to 1 p,M. (B)Plaquing efficiency of SV,u and of two isolates cloned after 17 passages in Ae. albopictus cells (10 passages in 5 phl methionine and 7 passages without methionine). With respect to the cloned isolates, the first number refers to the number of passages in methionine-deprived Ae. albopictus cells, ~,~~, and the second number is an arbitrarily assigned clone designation. S V L M ~which was passaged three times in BHK cells after the 17 passages in Ae. albopictus cells and then cloned, replicated much better in vertebrate cells than did SVLM17/17. Both viruses, however, showed the low methionine-resistance (LM)phenotype. (From Durbin and Stollar, 1985.)

of 100 pM, SVLM17117 showed the same plaquing efficiency with 1 pM as with 100 pM methionine (Fig. 16B). Additional support for the idea that the requirement of SV for methionine relates to the level of ado met came from the demonstration that in chick cells S V L M 1 7 / 1 7 was cross-resistant to cycloleucine, a compound which inhibits the synthesis of ado met. To extend these studies, information was sought concerning the relative concentration of ado met required by the SVRtdviral RNA cap methylase and by the various mosquito cell methylases. It was shown that the growth of Ae. albopictus cells was not affected until the concentration of methionine was reduced to less than 6 pJ4, whereas maintenance of cells in less than 10 pJ4 methionine reduced the yield of virus at least 100-fold compared to the yield from cells in normal medium (100 p M methionine). This finding is consistent with the

1

2

3

FIG.17. Induction of heat-shock proteins. Mosquito cells were incubated in serumfree E medium containing 100 p M methionine (lanes 1 and 2) or no methionine (lane 3) at 34.5"C for 3 hours. The temperature was then raised to 40°C. Five minutes before the temperature shift the cells represented in lane 1were treated with actinomycin, and 20 minutes after the temperature shift the cells were labeled for 10 minutes with PHlleucine. Cell lysates were prepared and fractionated by electrophoresis on polyacrylamide gels, and the gel was fluorographed. "he proteins indicated by the labels correspond in molecular weight to the Drosophilu heat-shock proteins. (From Durbin and Stollar, 1985.)

362

VICTOR STOLLAR

proposition that cellular methylases can function at levels not sufficient for the viral RNA cap methylase. Further support for this idea was obtained by study of the heat shock proteins in mosquito cells. Induction of heat shock proteins is known to require transcription of specific genes (Ashburner and Bonner, 1979), i.e., the synthesis and presumably capping and methylation of the mRNAs for the heat shock proteins. Whereas methionine-deprived SV-infected Ae. albopictus cells failed to synthesize any detectable viral RNA at 3 hours after infection (viral RNA is readily detectable at this time when cells are maintained in normal medium), Ae. albopictus cells deprived of methionine for 3 hours were still able to synthesize heat shock proteins when the temperature of incubation was raised from 34.5 t o 40°C (Fig. 17). Thus it appears that in Ae. albopictus cells host cell mRNAs can be methylated under conditions in which viral RNAs are not. As to why the effect of methionine deprivation is so dramatic in mosquito cells and is not observed in vertebrate cells, there is not yet sufficient information to answer this question. We have observed, however, that following methionine deprivation the level of ado met drops more rapidly and to a greater extent in Ae. albopictus cells than in BHK cells (Bossie and Stollar, unpublished). To understand the effect of methionine deprivation more completely, it will be necessary to learn about the kinetics of ado met synthesis and utilization in the different cell types as well as about the pool size of ado met. Perhaps the enzyme which is involved in the synthesis of ado met, methionine adenosyltransferase (MAT),plays a crucial role. One can postulate, for example, that in the mosquito cells this enzyme has a lower affinity (higher K,) for methionine than does the corresponding enzyme in vertebrate cells, so that as the methionine level in the medium and the cell falls, the synthesis of ado met is reduced to a much greater degree than in vertebrate cells. To summarize, the level of methionine and presumably of ado met plays an important role in determining whether SV is able to replicate in Ae. albopictus cells. Mutants of SV have been obtained which are able to replicate well in Ae. albopictus cells deprived of methionine, i.e., under conditions where standard virus is severely restricted. We believe that these viral mutants code for an altered viral RNA cap methylase with an increased affinity (lower K,) for ado met.

V. CONCLUDING REMARKS The comparative study of SV and appropriate viral mutants in mosquito and vertebrate cell cultures has provided new insights into how

VECTOR SPECIFICITY

363

specific properties of both the host cell and the virus can determine the outcome of infection. Further, the observations we have made at the cellular level on the model systems described in this article suggest areas for investigation at the level of the intact mosquito vector. Perhaps the properties of the host cell and the virus which have been shown to influence virus production in cultured cells may also modulate the rate and level of virus production in the whole mosquito, and hence determine the efficiency with which different species of mosquitoes or genetic variants within a species are able to serve as vectors for a specific virus or for arthropod-borne viruses in general. Finally, just as insights from studies with viral infections of cultured cells may point to paths of enquiry with whole mosquitoes, observations made on viral replication and restriction in the arthropod vector may highlight questions and problems which can be profitably studied at the cellular level.

REFERENCES Ashburner, M., and Bonner, J. J. (1979). The induction of gene activity in Drosophilu bv heat shock. Cell 17, 241-254. Berge, T. O., ed. (1975). “International Catalogue of Arboviruses,” 2nd Ed. US.Dept. of Health, Education, and Welfare, Public Health Service, DHEW Publication No. (CDC) 75-8301. Both, G. W., and Air, G. M. (1979). Nucleotide sequence coding for the N-terminal region of the matrix protein of influenza virus. Eur. J. Biochem. 96, 363-372. Brown, D. T., and Condreay, L. D. (1986). Replication of alphaviruses in mosquito cells. In “The Togaviridae and Flaviviridae” (S. Schlesinger and M. J. Schlesinger, eds.), Ch. 7. Plenum, New York. Burge, B. W., and Pfefferkorn, E. R. (1966). Complementation between temperaturesensitive mutants of Sindbis virus. Virology 30, 214-223. Cancedda, R., Bonatti, S., and Leone, A. (1981). One extra oligosaccharide chain of the high-mannose class in the E2 protein of a Sindbis virus isolate. J. Virol. 38, 8-14. Cantoni, G. L. (1975). Biological methylation: Selected aspects. Annu. Rev.Biochem. 44, 435-451. Chamberlain, R. W. (1980). Epidemiology of arthropod-borne togaviruses: The role of arthropods as hosts and vectors and of vertebrate hosts in natural transmission cycles. In “The Togaviruses. Biology, Structure, Replication” (R. W. Schlesinger, ed.), Chap. 6. Academic Press, New York. in Semliki Cross, R. K. (1983). Identification of a unique guanine-7-methyltransferase Forest virus (SFV) infected cell extracts. Virology 130,452-463. Dasgupta, A., Zabel, P., and Baltimore, D. (1980). Dependence of the activity of the poliovirus replicase on a host cell protein. Cell 19, 423-429. Davey, M. W., and Dalgarno, L. (1974). Semliki Forest virus replication in cultured Aedes albopictus cells: Studies on the establishment of persistence. J.Gen. Virol. 24, 453-463. Durbin, R. K., and Stollar, V. (1984). A mutant of Sindbis virus with a host-dependent defect in maturation associated with hyperglycosylation of E2. Virology 136, 331344. Durbin, R. K., and Stollar, V. (1985). Sindbis virus mutants able to replicate in methionine-deprived Aedes albopictus cells. Virology 144, 529-533.

364

VICTOR STOLLAR

Durbin, R. K., and Stollar, V. (1986). Sequence analysis of the E2 gene of a hyperglycosylated, host restricted mutant of Sindbis virus and estimation of mutation rate from frequency of revertants. Virology 164, 135-143. Fields, B. N., ed. (1985). “Virology.” Raven, New York. Gillies, S., and Stollar, V. (1981). Biochemical characterization of vesicular stomatitis virus-infected Aedes albopictus cells deprived of methionine. Virology 112.3 18-327. Hardy, J. L., Houk, E. J., Kramer, L. D., and Reeves, W. C. (1983). Intrinsic factors affecting vector competence of mosquitoes for arboviruses. Annu. Rev. Entomol. 28, 229-262. Horikami, S. M., and Moyer, S. A. (1982). Host range mutants of vesicular stomatitis virus defective in in uitro methylation. Proc. Natl. Acad. Sci. U S A . 79, 7694-7698. Houk, E. J., Kramer, L. D., Hardy, J. L., and Chiles, R. E. (1985). Western equine encephalitis virus: In vivo infection and morphogenesis in mosquito mesenteronal epithelial cells. Virus Res. 2, 123-138. Hsieh, P., Rosner, M. R., and Robbins, P. W. (1983). Host-dependent variation of asparagine-linked oligosaccharides at individual glycosylation sites of Sindbis virus glycoproteins. J . Biol. Chem. 258, 2548-2554. Hubbard, S. C., and Ivatt, R. J. (1981). Synthesis and processing of asparagine-linked oligosaccharides. Annu. Rev. Biochem. 50, 555-583. Jones, K. J., Waite, M. R. F., and Bose, H. R. (1974). Cleavage of a viral envelope precursor during the morphogenesis of Sindbis virus. J . Virol. 13, 809-817. Karabatsos, N. (1978). Supplement to international catalogue of arboviruses including certain other viruses of vertebrates. Am. J . Trop. Med. Hyg. 27,372-440. Kowal, K. J., and Stollar, V. (1981). Temperature-sensitive host-dependent mutants of Sindbis virus. Virology 114, 140-148. Kyte, J., and Doolittle, R. F. (1982). A simple method for displaying the hydropathic character of a protein. J . Mol. Biol. 157, 105-132. Lam, K. S. K., and Marshall, I. D. (1968). Dual infections of Aedes aegypti with arboviruses. 11. Salivary gland damage by Semliki Forest virus in relation to dual infections. Am. J . Trop. Med. Hyg. 17, 637-644. Mayne, J . T., Bell, J. R., Strauss, E. G., and Strauss, J. H. (1985).Pattern of glycosylation of Sindbis virus envelope proteins synthesized in hamster and chicken cells. Virology 142, 121-133. Mims, C. A., Day, M. F., and Marshall, I. D. (1966). Cytopathic effect of Semliki Forest virus in the mosquito Aedes aegypti. Am. J . Trop. Med. Hyg. 15, 775-784. Reed, W. (1902). Recent researches concerning etiology, propagation and prevention of yellow fever, by the United States Army Commission. J . Hyg. 2, 101-119. Reed, W., Carroll, J., and Agramonte, A. (1901). Etiology of yellow fever: Additional note. J . Am. Med. Assoc. 36, 431-440. Renz, D., and Brown, D. T. (1976). Characteristics of Sindbis virus temperature-sensitive mutants in cultured BHK-21 and Aedes albopictus (mosquito) cells. J . Virol. 19, 775-781. Rice, C. M., and Strauss, J. H. (1982). Association of Sindbis virion glycoproteins and their precursors. J . Mol. Biol. 154, 325-348. Sarver, N. (1978). Sindbis virus-induced killing of Aedes albopictus cells: Properties of virus-induced cell injury and inhibition by Virazole. Ph.D. thesis submitted to the Graduate School of Rutgers, The State University of New Jersey. Sarver, N., and Stollar, V. (1977). Sindbis virus-induced cytopathic effect in clones of Aedes albopictus (Singh) cells. Virology 80,390-400. Schlesinger, R. W. (1980). Introduction. In “The Togaviruses. Biology, Structure, Replication” (R. W. Schlesinger, ed.), Chap. 1. Academic Press, New York. Shope, R. E. (1980). Medical significance of togaviruses: An overview of diseases caused

VECTOR SPECIFICITY

365

by togaviruses in man and in domestic and wild vertebrate animals. In “The Togaviruses. Biology, Structure, Replication” (R. w. Schlesinger, ed.),Chap. 3.Academic Press, New York. Simpson, R. W., and Obijeski, J. F. (1974).Conditional lethal mutants of vesicular stomatitis virus. I. Phenotypic characterization of single and double mutants exhibiting host restriction and temperature sensitivity. Virology 57, 357-368. Singh, K. R. P. (1967).Cell cultures derived from larvae ofAedes albopictus (Skuse) and Aedes uegypti (L).Curr. Sci. 36,506-508. Stevens, T. M. (1970).Arbovirus replication in mosquito cell lines (Singh) grown in monolayer or suspension culture. Proc. SOC.Exp. Biol. Med. 134,356-361. Stollar, V. (1978).Inhibition of Sindbis virus replication in Aedes albopictus cells deprived of methionine. Virology 91, 504-507. Stollar, V. (1980).Togaviruses in cultured arthropod cells. In “The Togaviruses. Biology, Structure, Replication” (R. W. Schlesinger, ed.), pp. 583-621. Academic Press, New York. Strauss, E. G., Rice, C. M., and Strauss, J. H. (1984).Complete nucleotide sequence of the genomic RNA of Sindbis virus. Virology 133,92-110. Tabor, C. W.,and Tabor, H. (1984).Polyamines. Annu. Rev. Biochern. 53, 749-790. Tatem, J., and Stollar, V. (1986).Dominance of the CPE(+) phenotype in hybrid Aedes albopictus cells infected with Sindbis virus. Virus Res. 5, 121-130. Tooker, P., and Kennedy, S. I. T. (1981).Semliki Forest virus multiplication in clones of Aedes albopictus cells. J . Virol. 37, 589-600. Tuomi, K., Mantyjarvi, R., and Raina, A. (1980).Inhibition of Semliki Forest and herpes simplex virus production in a-difluoromethylornithine-treated cells: Reversal by polyamines. FEBS Lett. 121,292-294. Yunker, C. E., ed. (1986).“Arboviruses in Arthropod Cells in Vitro.” CRC Press, Boca Raton, Fla., in press.

This Page Intentionally Left Blank

INDEX

A

Abaca bunchy top virus, 309-310 Aberrant ratio, in plants, viral infection and, 242 Abortive infections, with parvoviruses, 168 Abscissic acid, viral infections and, 237238,239,240 Adenosine triphosphate, cap binding proteins and, 178,180-181, 182-184 S-Adenosylmethionine, viral replication and, 339-343, 360,362 Aedes albopictus, CPE( + ) and CPE( - ) clones, infection with Sindbis virus and, 333-339 Alfalfa, sweet clover necrotic mosaic virus and, 293-294 Alfalfa mosaic virus, 211,220,222,246 pseudorecombinants of, 211 Alphavirus glycoproteins of, 8-10 cytoplasmic domain, 10-11 ectodomain, 12-13 transmembrane domain, 11-12 host range determinants of, 11 Amino acids, of TMV coat protein, symptoms and, 218-219 Antigenicity, of parvovirus nonstructural polypeptides, 128,129-132 Antigenic structure, of parvoviruses, 104-106 AphidW, spread of banana bunchy top virus and, 317 Aphidicolin, parvovirus gene expression and, 140-142 Arabis mosaic virus, 216 Arbovirus, definition of, 328 Assembly, of parvovirus virion, 162,164 Attenuation, transcription by parvovirus and, 145 Australia, control of banana bunchy top virus infection in, 314-318

367

campaign in New South Wales, 319322 Queensland campaign, 321-322 Auxins, viral infections and, 239-240 8-Azidopurines, binding of viral protein to chromatin and, 245

B Banana(&, distribution of, 302-303 Banana bunchy top virus characteristics of diagnosis of infected plants, 311-312 epidemiology, 313 host range and varietal susceptibility, 309 symptoms, 306-307 virus etiology, 310-311 virus strains, 309-310 control: the Australian experience, 314-3 18 campaign in New South Wales, 319321 Queensland campaign, 321-322 geographical history of, 303-306 limited research on, possible explanations for, 301-302 Barley mosaic virus, 220,221,246 Barley stripe mosaic virus, 221,242 Barley yellow dwarf virus, 213,215,302, 311 Bean-BCMV interrelations, genetics and, 230-232 Bean yellow mosaic virus, 235 Beet western yellows virus, 236,239,311 Beet yellows virus, 236 Brome mosaic virus, 211 5-Bromo-2’-deoxyuridine, DNA replication by parvovirus and, 149-150 Budding, of Sindbis virus mutant, 350351

368

INDEX

C Cap binding proteins identification of, 177-178 structural complexity of, 179-180 Cap binding protein complex eIF-4A component, function of, 182184 function of, 180-182 involvement in shutoff of host mRNA translation after poliovirus infection criticism of model, 191 early studies, 184-185 inactivation of complex by infection, 185-187 mechanism of inactivation, 187-188 poliovirus mediation of p220 cleavage, 188-191 Cap binding protein@)11, components of, 179, 180 Capsids, of parvovirus, polypeptides of, 98-104 Cardioviruses, translational inhibition by, 196-198 Carnation ringspot virus distribution of, 257-258 occurrence in other plants, 259-260 purification of, 261 Cauliflower mosaic virus, 222-223, 246 Cellb), polarized, viral glycoproteins and, 32-33 Cell cycle, parvovirus infection and, 140 Cherry leaf roll virus, 212-213 Chlorosis, viral infection and, 235-236 Chromatin, host, association of viral protein with, 243-247 Citrus exocortis viroid, 223, 233 Cleavage, of parvoviral DNA, 160 Clover primary leaf necrosis virus, 212 Coat protein of dianthoviruses aggregation of, 283 properties of, 266, 267, 278-279 of plant viruses, 218-219, 245 Coding strategy, of parvovirus genome nonstructural polypeptides, 127-137 structural polypeptides, 122-127 transcription, 117-1 19 translation, 119-122 Competition, translational inhibition by picornaviruses and, 195-196, 198

Conformation, of parvovirus RNA, 145147 Corn lily fleck virus, 242 Coronaviruses, glycoproteins of, 30-3 1 Cowpea chlorotic mottle virus, 211-212 Cowpea mosaic virus, 207, 210, 213, 220-221, 222, 224 Cross-linking, of parvovirus capsid polypeptides, 102 Cryptic infection, parvovirus and, 168 Cucumber mosaic virus, 212, 215,236, 238,243,245,312 pseudorecombinants, symptoms and, 212 satellites and, 216, 223 Culex quinquefasciatus, as vectors of encephalitis, 330 Culex tarsalis, variants, efficiency as vectors for western equine encephalitis virus, 329 Cycloleucine, replication of Sindbis virus and, 340, 360 Cytokinins, viral infection and, 237, 239, 240 Cytopathology, of dianthoviruses, 286289 electron microscopy, 287-289 light microscopy, 286-287 Cytoplasmic domain of alphavirus glycoprotein, 12-13 of influena virus hemagglutinin, 7-8 of rabies virus, 19-22 of Rous sarcoma virus, 24-25 of vesicular stomatitis virus, 19-22

D

Deletions, in parvovirus DNAs, 109, 110 Deoxyribonucleic acid of parvoviruses, 97-98, 103, 105, 139, 142 packaging of, 164 replication of, 149-162 structure and sequence, 106-111 plant, viral infection and, 242-243 Deoxyribonucleic acid polymerases, parvovirus infection and, 150-152 Diagnosis, of banana bunchy top virus infection, 311

INDEX Dianthoviruses cytopathology of, 286-289 electron microscopy, 282-289 light microscopy, 286-287 diseases caused in plants, 258-260 ecological studies, 291-294 genome reassortment studies aggregation of coat protein, 283 bipartite genome, 281 electrophoretic mobility, 285 infectivity, 281-283 serological specificity, 283, 285 symptomatology, 285-286 host range and symptomatology experimental studies, 277-278 natural isolates, 276-277 members of carnation ringspot virus, 257-258 red clover necrotic mosaic virus, 258 sweet clover necrotic mosaic virus, 258 properties of, 257, 265-266 electron microscopy, 263-265 protein and nucleic acid, 266-268 purification, 260-263 replication of, 278-281 serological studies of monoclonal antibodies, 272-276 polyclonal antibodies, 269-272 transmission by vectors, 289-291 Dodder, virus transmission and, 291 Domains, of glycoprotein of spleen focusforming virus, 25-28

E Early events of flavivirus replication, 51-52 of parvovirus life cycle, 137-139 Ecological studies, of dianthoviruses, 291-294 Ectodomain of alphavirus glycoprotein fusion activity, 10-11

host range determinants, 11 of influenza virus hemagglutinin fusion activity, 4-6 receptor binding site, 6 of rabies virus oligosaccharide sites, 14-16 receptor binding activity, 17

receptor binding and fusogenic activity, 16-17 of Rous sarcoma virus glycoprotein, 23-24 of vesicular stomatitis virus oligosaccharide sites, 14-16 receptor binding and fusogenic sites, 16-17 Electron microscopy, of purified dianthoviruses, 263-265 Electrophoretic mobility, reassortants of dianthoviruses and, 285 Enhancer sequence, parvovirus DNA and, 116-117 Enzyme-linked immunosorbent assay, dianthoviruses and, 269-272 Epidemiology, of banana bunchy top virus, 313 Ethylene, symptom production and, 235236,239 Etiology, of banana bunchy top virus, 310-31 1

F Flaviviruses historical background, 45-46 mutants of, 80-81 replication of early events, 51-52 glycosylation of virus-specified proteins, 58-61 identification of virus-specified proteins, 53-58 intracellular location of proteins and possible function, 61-65 persistent infection, ts mutants and interfering particles, 79-82 preliminary overview, 50-51 strategy of viral RNA synthesis, 7477 translation strategy, 65-74 viral messenger RNA, 52-53 virus maturation and release, 77-79 ribonucleic acid polymerase of, 65, 76 structure of virion morphology and composition, 46-47 proteins, 48-50 RNA, 47-48 Fungi pathogenic, virulence of, 224-225

370

INDEX

transmission of dianthoviruses and, 290 Fusion activity of alphavirus glycoproteins, 10-11 of influenza virus hemagglutinin, 4-6 Fusogenic activity of rabies virus, 16-17 of vesicular stomatitis virus, 16-17

G

Gene expression, in parvoviruses, 139149 Gene-for-gene concept, host-virus interactions and, 224-226 Gene sequence, of flavivirus, 66-67 Genome bipartite, of dianthoviruses, 281 of parvoviruses coding strategy of, 117-137 organization of, 111-117 Gibberellins, viral infection and, 238, 239, 240 Glycoprotein(s) of alphavirus, 8-10 cytoplasmic domain, 10-11 ectodomain, 12-13 transmembrane domain, 11-12 of other viruses coronaviruses, 30-31 herpesviruses, 31-32 influenza virus neuraminidase, 2829 nonenveloped viruses, 32 paramyxoviruses, 29-30 of Sindbis virus mutant, 349, 350-351, 353-354,356-358 viral functions of, 1 polarized cells and, 32-33 Glycosylation, of flavivirus-specified proteins, 58-61 Growth, of plants, TMV multiplication and, 215-216

H

Heat shock proteins, in mosquito cells, 362

Helper virus, parvoviruses and, 91 Hemagglutination, parvovirus serotype and, 104 Hemagglutinin, of influenza virus, 2-4 cytoplasmic domain, 7-8 ecto domain, 4-6 receptor binding site, 6 transmembrane domain, 6-7 Herpesviruses, glycoproteins of, 31-32 Homologies, in parvovirus polypeptides, 127 Hormones, symptoms of viral infection and, 239-240 Host age, parvovirus infection and, 93, 9495 functions of, parvovirus DNA replication and, 156 mRNA translation after poliovirus infection, CBP complex and criticism of model, 191 early studies, 184-185 inactivation of complex by infection, 185-187 mechanism of inactivation, 187-188 poliovirus mediation of p220 cleavage, 188-191 protein synthesis, alternative models for poliovirus inhibition of alteration of intracellular ion concentration, 192-193 competition between host and viral mRNAs, 193 double-stranded RNA and, 191192 inhibition of host mRNA synthesis by viral structural components, 193-194 Host cell, role in virus infection CPE(+) and CPE(-) clones of Aedes albopictus, 333-339 response to methionine deprivation, 339-343 Host-CBII interactions genetics of, plant viruses and applications to selected combinations, 226-232 concept of gene-for-gene relationship, 224-226 how do pathogenic determinants and resistance genes act?, 232-234 of parvovirus, nonproductive, 165-169

371

INDEX with virus, questions concerning, 330331 Host plant alterations in metabolism attempts to influence symptom type with hormones, 239-240 common features in symptom type variation, 234-239 interference of viral products, 241247 interaction of plant viruses with induction of symptoms, 206-208 viral properties specifying symptom type, 208-217 virus-coded proteins and possible functions, 217-224 Host range of banana bunchy top virus, 309 of dianthoviruses experimental studies, 277-278 natural isolates, 276-277 of flaviviruses, 50-51 Host range determinants, of alphavirus glycoproteins, 11 Hybridization studies, of parvovirus DNA, 108-109

I

Infection, persistent, with flavivirus, 7981 Infectivity, of bipartite genome of dianthoviruses, 281-283 Influenza virus hemagglutinin, 2-4 cytoplasmic domain, 7-8 ectodomain, 4-6 receptor binding site, 6 transmembrane domain, 6-7 neuraminidase of, 28-29 Initiation factors cross-linking to mRNA, 186 eukaryotic, 176 Insects, transmission of dianthovirus by, 290-291 Integration, of parvovirus DNA into host DNA, 139 Interference, flavivirus and, 81-82 Internalization, of parvovirus, 138 Ion concentration, intracellular, poliovirus infection and, 192-193

L Lettuce mosaic virus, 239 Lucerne transient streak virus, 216 Lysosome parvovirus virions and, 138

M Maturation of flavivirus, 77-78 of parvovirus virions, 163, 165 Membranes, flavivirus proteins and, 6162 Mengovirus, IF phosphorylation and, 197 Messenger ribonucleic acid(s) eukaryotic, cap binding proteins of, 176 function of CBP complex, 180-182 function of eIF-4A component of CBP complex, 182-184 identification of CBPs, 177-178 structural complexity of CBPs, 179180 of flavivirus, 52-53 host, involvement of CBP complex in shutoff of translation by poliovirus infection criticism of model, 191 early studies, 184-185 inactivation of CBP complex by infection, 185-187 mechanism of inactivation, 187-188 poliovirus mediation of p220 cleavage, 188-191 poliovirus, translation of, 184-185 secondary structure, translation and, 180, 182, 184, 191 viral, competition with host mRNA, 193 Methionine deprivation, replication of Sindbis virus in mosquito and, 339-343,358-362 Monoclonal antibodies, to parvoviruses, 105-106 Morphogenesis, viral infections and, 238-239 Morphology, of parvoviruses, 96-97 Mosaics, viral infection and, 236-238 Mosquito, virus multiplication in, 329

372

INDEX

Mosquito cells, replication of Sindbis virus in, 331-332 Mutagenesis, effects on plant viral properties specifying symptom type, 208-211 Mutants of parvovirus frameshift, 148 host-range, 167-168 of poliovirus, protease and, 189-190 temperature sensitive, of flavivirus, 80-81

N Necrosis, viral infection and, 234-235 Nematodes, transmission of dianthoviruses and, 289-290, 291292 Neoplasia(s), parvoviruses and, 94 Neuraminidase, of influenza virus, 2829 Nicking, of parvoviral DNA, 160-162 Nucleic acid synthesis, TMV multiplication and, 215 Nucleotide changes, in parvovirus DNA, 113-115 Nucleus flavivirus proteins and, 64 parvovirus virions and, 138-139 0

Oligosaccharide sites of rabies virus, 14-16 of vesicular stomatitis virus, 14-16 Open reading frames in parvovirus DNA, 112-114 parvovirus nonstructural polypeptides and, 128, 133 P p220, cleavage, poliovirus mediated,

188-191 Palindromic sequences, in parvovirus DNA, 109-110, 155-156 Paramyxoviruses, glycoproteins of, 2930

Parvovirus(es) autonomous, list of, 92 life cycle DNA replication, 149-162 early events, 137-139 gene expression, 139-149 nonproductive virus-host cell interaction, 165-169 virion assembly and maturation, 162-165 pathogenicity of, 91-93 serotype, pathogenic potential and, 95 structure of viral genome coding strategy, 117-137 DNA structure and sequence, 106111 organization of genome, 111-117 susceptibility to, factors affecting, 93 virion structure antigenic, 104-106 biochemical, 97-104 morphology, 96-97 Pathogenic determinants, action of, 232233 Pea enation mosaic virus, 246 Peanut green mosaic virus, 236-237 Phosphorylation of initiation factors, poliovirus infection and, 190, 192 parvovirus proteins and, 147 Photosynthesis, mosaics and, 236-237 Picornaviruses, inhibition of translation by, 194-195 competition model, 195-196 other models for cardioviruses, 196198 Plant viruses interactions with host induction of symptoms, 206-208 viral properties specifying symptom type, 208-217 virus-coded proteins and possible functions, 217-224 Poliovirus alternative models for inhibition of host protein synthesis alteration of intracellular ion concentration, 192- 193 competition between viral and host mRNAs, 193 double-stranded RNA and, 191-192

373 inhibition of host mRNA synthesis by viral structural components, 193-194 mRNA, translation of, 184-185 Poliovirus infection, involvement of CBP complex in shutoff of host mRNA translation and criticism of model, 191 early studies, 184-185 inactivation of complex by infection, 185-187 mechanism of inactivation, 187-188 poliovirus mediation of p220 cleavage, 188-191 Polyadenylation, in parvovirus DNA, 117 Polymerase, plant virus symptoms and, 221-222 Polypeptides nonstructural, of parvoviruses, 127137, 143 of parvovirus capsids, 98-104, 143 structural, of parvoviruses, 122-127 viral, inhibition of host mRNA synthesis and, 193-194 Potato leaf roll virus, 302 Potato spindle tuber viroid, 223 Potato virus X, 221 Proteaseb), of poliovirus, 189 Protein(s) flavivirus-specified, 48-49 glycosylation of, 58-61 identification of, 53-58 intracellular location and possible functions, 61-65 parvovirus virion assembly and, 163164 virus-coded, 217-218 absence of symptom-producing proteins, 222-224 number of proteins is limited coat, 218-219 polymerase, 221-222 transport, 219-221 Protein C, of flaviviruses, 50 Protein E, of flaviviruses, 49-50 Protein M, of flaviviruses, 50 Protein synthesis inhibitors, flavivirus and, 70 plant, TMV multiplication and, 215 Proteolysis of CBP complex, 186-188

of flavivirus polyproteins, 67-68 of parvovirus capsid polypeptides, 100101, 103, 104, 122, 123-124 Protoplasts, viruses and, 217, 247 Pseudorecombinants, of plant viruses, symptom type and, 211-213,219 Purification, of dianthoviruses, 260 carnation ringspot virus, 261 red clover necrotic mosaic virus, 261262 sweet clover necrotic mosaic virus, 263

R Rabies virus glycoproteins of, 13-14 cytoplasmic domain, 19-22 ectodomain, 14-17 transmembrane domain, 17-19 Raspberry ringspot virus, 212 Receptor(s), for parvoviruses, 137, 165 Receptor binding, of vesicular stomatitis virus, 16-17 Receptor binding activity, of rabies virus, 17 Receptor binding site, of influenza virus hemagglutinin, 6 Red clover necrotic mosaic virus, 212 distribution of, 258 purification of, 261-262 Release, of flavivirus, 78 Repeat sequences, in parvovirus DNA, 111 Replication, of dianthoviruses, 278-281 Resistance genes, action of, 232, 234 Restrictive infection, parvovirus and, 166-167 Retrovirus, glycoproteins of, 22-23 cytoplasmic domain, 24-25 ectodomains, 23-24 Revertants, of host range mutants of Sindbis virus, 353-354 Ribgrass mosaic virus, 226 Ribonucleic acid, see also Messenger ribonucleic acid of dianthoviruses, properties of, 266268 double stranded of banana bunchy top virus, 310311

374

INDEX

dianthoviruses and, 278-280 inhibition of host protein synthesis by poliovirus infection, 191-192 plant viruses and, 217 of flaviviruses, 47-48 strategy of synthesis, 74-77 of parvoviruses, 118-119, 144 translation of, 119-122 conformation of, 145-147 symptom production and, 223-224 Ribonucleic acid polymerase, of flavivirus, 65, 76 Ribonucleic acid synthesis in Sindbis virus-infected mosquito cells, 334-335, 337 of temperature-sensitive viral mutants, 345-346 Rice tungro virus, 239 Rolling hairpin model, of parvoviral DNA replication, 156-159 Rous sarcoma virus, glycoprotein cytoplasmic domain, 24-25 ectodomain of, 23-24

Spleen focus-forming virus, glycoprotein, domains of, 25-28 Splicing of parvovirus polypeptides, 124-127, 133-136 Sweet clover necrotic mosaic virus, 212 isolation of, 258 purification of, 263 Symptomatology, of dianthoviruses, 277 reassortants and, 285-286 Symptoms of banana bunchy top virus infection, 306-309 induction by plant viruses, 206-208 plant virus multiplication and, 213217 type, common features in variations chlorosis and necrosis, 234-236 morphogenetic disturbances, 238239 mosaics, 236-238 type, plant viral properties specifying effects of mutagenesis, 208-21 1 pseudorecombinants, 21 1-213 viral multiplication, 213-217

S

St. Louis encephalitis virus, vector of, 329 Semliki Forest virus, 331, see also Alphavirus replication in mosquito cells, 335-336 Serological specificity, bipartite genome of dianthoviruses and, 283, 285 Serological studies, of dianthoviruses monoclonal antibodies, 272-276 polyclonal antibodies, 269-272 Sindbis virus, see also Alphavirus host range mutants restricted in Aedes albopictus cells, 343-347 mutant restricted in vertebrate cells, 348-358 replication in mosquito CPE(+) and CPE(-) clones of Aedes albopictus, 333-339 methionine deprivation, 339-343 replication in vertebrate and mosquito cells, 331-332 Sodium ions, viral infections and, 192193 Soil, transmission of dianthoviruses and, 292-293

T

Temperature sensitivity, of Sindbis virus host range mutants, 343-345 Tobacco mosaic virus, 209, 210, 213, 215, 235,238, 239-240, 243-244,246 mosaics and, 236, 237 mutants, symptoms and, 209-210, 218 transport protein of, 220, 221 virus multiplication and, 215 Tobacco rattle virus, 212 Tobacco ringspot virus, 222, 223 Tobacco streak virus, 212 Tobacco-tobacco mosaic virus, interactions, genetics and, 226-229 Tomato bushy stunt virus, 216 Tomato-tobacco mosaic virus, interactions, genetics and, 229-230 Topoisomerase, model of parvoviral DNA replication and, 159 Transcription, of parvovirus genome, 117-119 Transcription complexes, of parvovirus, products of, 144-145

375

INDEX Transcription control regions, in parvovirus DNA, 115-116, 117 Transformation oncogenic, parvovirus and, 95-96 parvovirus infection and, 168 Translation inhibition by other picornaviruses, 194-195 competition model, 195-196 other models for cardioviruses, 196198 of parvovirus genome, 119-122 Translation strategy, of flavivirus, 6574 Transmembrane domain of alphavirus glycoproteins, 11-12 of influenza virus hemagglutinin, 6-7 of rabies virus, 17-19 of vesicular stomatitis virus, 17-19 Transport protein, plant virus symptoms and, 219-221 Turnip crinkle virus, 216 Turnip yellow mosaic virus, 207, 236, 237 U

Ultraviolet radiation flavivirus translation and, 70-72 parvovirus DNA replication and, 152153

V Varietal susceptibility, to banana bunchy top virus, 309 Vectors arthropod, transmission of viral diseases by, 327-328 specificity of, 328 goals of research into, 328-330 transmission of dianthoviruses and, 289-291 Velvet tobacco mottle virus, 216 Vertebrate cells, replication of Sindbis virus in, 331-332

Vesicles, flavivirus RNA synthesis and, 76 Vesicular stomatitis virus glycoproteins of, 13-14 cytoplasmic domain, 19-22 ectodomain, 14-17 transmembrane domain, 17-19 replication, methionine deprivation and, 340-341 Viral diseases, transmission by arthropod vectors, 327-328 Viral products, interference of association of viral protein with host chromatin, 243-247 early events in virus-host recognition, 241-243 Viral satellites, symptoms of infection and, 216, 223 Viral strains, of banana bunchy top virus, 309-310 Viroids nucleus and, 246 pathogenesis and, 233 Virus(es) arthropod borne, specificity of, 328 glycoproteins, functions of, 1 interaction with host, questions concerning, 330-331 nonenveloped, glycoproteins of, 32 role in infection host range mutants restricted in Aedes albopictus,343-347 mutant able to replicate in methionine-starved Aedes albopictus cells, 358-362 mutants restricted to vertebrate cells, 348-358 Virus-host recognition, early events in, 241-243

W Western equine encephalitis virus, vector of, 329 Wheat streak mosaic virus, 242

This Page Intentionally Left Blank