Sindbis virus pathogenesis: phenotypic reversion ... - Semantic Scholar
Journal of General Virology (1993), 74, 1691-1695. Printed in Great Britain
1691
Sindbis virus pathogenesis: phenotypic reversion of an attenuated strain to virulence by second-site intragenic suppressor mutations Randal J. Schoepp and Robert E. Johnston* Department o f Microbiology and Immunology, University o f North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599, U.S.A.
Monoclonal antibodies (MAbs) specific for the E2c neutralizing antigenic site on the Sindbis virus E2 glycoprotein define a pathogenesis domain that affects neonatal mouse virulence. Sequence analysis of E2c MAb escape mutants showed that the domain included E2 amino acids 62, 96 and 159. The pathogenesis domain is also influenced by changes at E2 position 114. Mutation of E2 residues Asn 62 to Asp or Lys 159 to Glu results in suppression of the attenuated phenotype conferred by a mutation from Ser to Arg at E2 position 114. Possible mechanisms of phenotypic
suppression within the E2c pathogenesis domain were investigated by using site-directed mutagenesis to determine the effects of specific combinations of positively charged, negatively charged and uncharged amino acid substitutions at E2 positions 62, 114 and 159. Phenotypic reversion to virulence by second-site suppressor mutations at E2 amino acids 62 or 159 was not dependent on ionic interaction with the residue at E2 114. Rather, suppression appeared to be the result of independent virulence effects mediated by specific residues.
Sindbis virus is the prototype of the alphavirus genus in the family Togaviridae. This group of arthropod-borne viruses includes the agents of eastern, western and Venezuelan equine encephalitis, which are of veterinary and public health importance (Schlesinger & Schlesinger, 1986). Sindbis virus has been instrumental in the elucidation of the molecular basis of alphavirus virulence in animal models, and numerous studies have implicated specific domains on the glycoprotein spikes of alphaviruses as important determinants of pathogenesis (Sefton et al., 1978; Davis et al., 1986; Olmsted et aL, 1986; Vrati et al., 1986; Lustig et al., 1988; Polo et al., 1988; Russell et al., 1989; Polo & Johnston, 1990, 1991; Glasgow et al., 1991 ; Tucker & Griffin, 1991 ; Schoepp & Johnston, 1992). The glycoprotein spikes are heterodimers composed of glycoproteins E1 and E2, which are anchored near their carboxy termini in the cell-derived lipoprotein envelope that surrounds the nucleocapsid (Strauss et al., 1969; Schlesinger et al., 1972; von Bonsdorff & Harrison, 1975; Rice & Strauss, 1982). The icosahedral nucleocapsid is composed of 240 copies of the C protein and surrounds the single-stranded, message-sense RNA genome (Strauss et al., 1969; Paredes et al., 1992). The glycoproteins are the means by which the virus interacts with the surrounding environment (Sefton et al., 1978) and are responsible for virus attachment and penetration (Bose & Sagik, 1970; Olmsted et al., 1984, 1986; Flynn et al., 1988), haemagglutination (Dalrymple et al., 1976; Schmaljohn et al.,
1983) and induction of an immune response (Schmaljohn et al., 1983). A number of single-site mutations that affect Sindbis pathogenesis in mice have been identified, predominantly in the 5" halves of the E1 and E2 glycoprotein genes (Davis et al., 1986; Lustig et al., 1988; Russell et al., 1989; Pence et al., 1990; Polo & Johnston, 1990; Tucker & Griffin, 1991). For example, the influence on virulence of the residue at position 114 in E2 was demonstrated in studies of the Sindbis virus mutant SB-RL, derived from Sindbis strain AR339 (SB). SB-RL is attenuated in neonatal mice and has an increased rate of penetration into BHK-21 cells (Baric et al., 1981; Olmsted et al., 1984, 1986). These phenotypic differences are produced by substitution of Arg at E2 position 114 in SB-RL for Ser found in SB (Davis et al., 1986). The phenotypic effects on E2 of Arg 114 were confirmed by site-directed mutagenesis (SDM) of a full-length cDNA clone of Sindbis virus from which infectious virus can be derived (Rice et al., 1987; Polo et al., 1988). In addition to Arg, substitution of Lys at E2 position 114 was strongly attenuating, and the substitution of Phe, Tyr or Trp attenuated virulence to a lesser degree (Polo & Johnston, 1991). Another of the phenotypic effects on E2 of an Arg 114 substitution is an increase in sensitivity to neutralization by monoclonal antibodies (MAbs) recognizing the E2c antigenic site (Davis et al., 1986; Polo et al., 1988). This observation suggested that the E2 114 locus might be a
0001-1489 © 1993SGM
1692
Short communication
constituent of, or be in the vicinity of, the E2c site. The E2c site was mapped by selecting neutralization escape mutants of SB and SB-RL with two E2c-specific MAbs, R6 and R13 (Pence et al., 1990). From SB and SB-RL 11 E2c MAb escape mutants were selected, all of which showed greatly reduced binding to either R6 or R13. Sequence analysis of the E2 and E1 genes of these mutants demonstrated that each contained a single mutation in E2 at either position 62, 96 or 159. SDM of E2 gene codons 62, 96 and 159 confirmed the participation of these loci in the E2c site (Schoepp & Johnston, 1992). Moreover, a general correlation was observed between disruption of the E2c site conformation by a particular amino acid substitution, as measured by MAb binding, and the reduction of virulence in neonatal mice. Therefore, Sindbis virus mutations in E2 gene codons 62, 96, 114 and 159 affect both virulence in neonatal mice and the binding or biological activity of E2c-specific MAbs, suggesting that these residues are constituents of a virion structure important for Sindbis pathogenesis (Pence et al., 1990; Schoepp & Johnston, 1992). Certain mutations in the E2c site affected pathogenesis by phenotypically suppressing the attenuating Arg mutation at position 114. Neutralization escape mutants in which Ash 62 was changed to Asp, or Lys 159 was mutated to Glu, suppressed the attenuated phenotype of E2 Arg 114 in the SB-RL genetic background, resulting in a reversion to virulence. E2 position 114 is the central amino acid in a stretch of uncharged and hydrophobic residues (Davis et al., 1986). The attenuated phenotype of Arg 114 mutants may result from the putative disruptive effect of a large, positively charged residue located within this hydrophobic region. If residue 114 interacts directly with residues 62 and 159 then the destabilizing effect of an Arg residue at position 114 could be neutralized completely or in part by a negatively charged amino acid at E2 62 or 159 to restore the structure and, consequently, a more virulent phenotype. Consistent with this hypothesis is the fact that the suppressing amino acids at E2 62 and 159 in the escape mutants were acidic (Pence et al., 1990). Alternatively, the amino acids at each of the three loci could affect the virulence phenotype independently, with the phenotype of the double mutants resulting from the algebraic sum of independent virulence effects. In this study, possible mechanisms of phenotypic suppression within the E2c pathogenesis domain were investigated by determining the effects of specific combinations of positively charged, negatively charged and uncharged amino acid substitutions at E2 positions 62, 114 and 159. SDM (Kunkel et al., 1987) of a full-length Sindbis virus clone (pTR5500) was utilized to produce a panel of mutant viruses with different amino acid substitutions at positions 62, 114 and 159 in E2 (Schoepp & Johnston,
1992). Four uracil-containing M13-based mutagenesis cassettes contained the glycoprotein genes E2 and El, and differed only in amino acid 114. The reference virus strains derived from these mutagenesis vectors were isogenic except for the substitution of Ser, Arg, Phe or Glu at E2 114. Mutagenesis with synthetic oligonucleotide primers, as described previously (Polo & Johnston, 1990, 1991; Schoepp & Johnston, 1992), was used to introduce an additional mutation at E2 position 62 or 159. The specific oligonucleotide primers were 'doped' within the sequences encoding E2 amino acid 62 [nucleotides (nt) 8814 to 8816] or 159 (nt 9105 to 9107) during synthesis: 5' CAGCAAGCGCA(C/G)(A/G)CAAGTACCGCTA 3' and 5' CGACCGTCTG(C/G)AAGAAACAAC 3'. Phosphorylated primers and the uracilcontaining templates were annealed, and the second strand was completed. The dsDNA was transformed into strain JM101 of Escherichia coli. Isolated plaques were screened for the presence of mutations by sequence analysis. A restriction fragment encompassing the E2 and El genes was reinserted into the full-length cDNA clone using unique restriction sites, StuI (nt 8571) and BssHII (nt 9804). Sequence replacements were monitored using a previously engineered restriction site marker that did not change the amino acid sequence (loss of the PstI site at nt 9119). Infectious RNA, transcribed from the SP6 promoter of linearized clones, was sequenced directly to confirm the mutation, and was then transfected into BHK cells to produce virus (Polo & Johnston, 1992). Each virus was subcutaneously inoculated into two to seven litters of 1-day-old CD-1 mice (averaging 10 animals per litter) at a dose of 100 p.f.u./animal. After a 14 day observation period, percentage mortality and mean survival time (MST) for those animals which died were calculated for each virus. Statistical analysis was by Student's t-test, and differences were considered significant if P < 0.001. Initial studies were designed to determine whether a charged amino acid substitution at E2 114 was required for an attenuated phenotype. Isogenic viruses that differed only by the amino acid encoded at E2 114 were inoculated into neonatal mice. The E2 114 construct, containing the wild-type Ser at this position, yielded a virus that caused 100% mortality and a short MST (4.7 + 0.7 days) characteristic of virulent SB virus strains (Table 1). This virus was designated the virulent reference virus strain, with Asn/Ser/Lys at E2 residues 62, 114 and 159, respectively. Similarly, the E2 Arg 114 construct, analogous to the attenuated SB-RL virus strain, was attenuated upon subcutaneous inoculation. This virus was designated the attenuated reference virus strain (with Asn/Arg/Lys as residues 62/114/159). Substitution of the uncharged, phenolic ring-containing Phe at
Short communication
Table 1. Viru&nceeffectsofsubstitut~ns at E21oci62, 114 or 159 E2 codon 62
114
159
Mortality (%)
MST (days)
Asn* Asn$ Asn Asn Asn Asn Asn Asn Asn Asn Asp Asp
Ser* Arg$ Phe Glu Ser Ser Arg Arg Phe Phe Arg Phe
Lys* Lys$ Lys Lys Glu Gin Glu Gin Glu Gln Lys Lys
100 (0)f 22 (3) 100 (0) 100 (0) 100 (0) 100 (0) 96 (5) 58 (37) 100 (0) 100 (0) 100 (0) 100 (0)
4.7 (0"7)f 11.8 (1.5) 7.4 (1'4) 5-1 (1.1) 3-5 (0'8) 5-1 (1.2) 6"9 (1,4) 7.5 (2.3) 5.4 (1.2) 5.4 (1-2) 6.2 (1.2) 5.3 (1.7)
* Wild-typeamino acid of virulent referencestrain. t Calculationswere based on inoculationof two to sevenlitters of neonatal mice (averaging 10 animals per litter). S.E.M. is given in parentheses. $ Aminoacid of attenuatedreferencestrain.
E2 114 yielded a virus that caused 100 % mortality yet resulted in significantly protracted survival compared to the virulent virus containing the wild-type amino acid (P < 0-001). In contrast, substitution of the negatively charged Glu yielded a virus that was no less virulent than wild-type. These results are consistent with a previous study in which a library of different amino acid substitutions were made at E2 position 114 in a slightly different genetic background (Polo & Johnston, 1991). The majority of the substitutions resulted in virulent viruses. Attenuated viruses could be divided into two groups, first, those which were strongly attenuating and had large, positively charged amino acids (Lys and Arg) and second, those which were more weakly attenuating and had mutations with large amino acids with ring structures (Tyr, Trp and Phe). However, neither Glu, as shown here, nor Asp were attenuating, suggesting that a negative charge did not confer an attenuating phenotype in this sequence context. Therefore, a charged amino acid at E2 position 114 was not required for attenuation, although large, positively charged amino acids produced the most pronounced attenuation phenotype. The genotype of the MAb escape mutant containing Glu 159 and Arg 114 (Pence et al., 1990) was reproduced in the full-length cDNA clone, and the second-site suppressing activity of the Glu 159 mutation was confirmed (Table 1). Substitution of Glu at 159 caused a partial reversion of the attenuated E2 Arg 114 phenotype toward virulence, although the double mutant was not as virulent as the wild-type Asn/Ser/Lys construct (Table 1). This virus caused 96% mortality in neonatal mice with a significantly shortened MST (P < 0"001) when compared to the attenuated reference virus containing
1693
amino acids Asn/Arg/Lys at E2 positions 62, 114 and 159, respectively. A Gin substitution also partially suppressed the E2 Arg 114 phenotype. However, as Gln 159 represented the most conservative uncharged substitution for the E2 Glu 159 suppressor mutation, this result suggested that negative charge at E2 position 159 was not required for suppression of the attenuated phenotype. Because suppression appeared to be independent of an ionic interaction between E2 residue 159 and attenuating residue 114, we determined the effect of substitution of Glu or Gln at E2 position 159 on the attenuation phenotype specified by an uncharged amino acid, E2 Phe 114. Substitution of either Glu or Gln for Lys at position 159, with an attenuating Phe at 114, resulted in a virulent virus with a significantly reduced MST. Both Glu and Gln suppressed Phe 114, just as they suppressed Arg 114. Therefore, with respect to E2 159, second-site suppression was independent of any charge interactions with the attenuating residue at E2 114. Analogous constructs were used to determine the requirements for suppressing amino acids at E2 62. The genotype of the MAb escape mutant containing E2 Asp 62 and E2 Arg 114 was reproduced in the full-length clone. Virus derived from constructs containing E2 Asp62 and E2 Arg 114 induced 100% mortality in neonatal mice. The MST was intermediate between the virulent Asn/Ser/Lys and attenuated Asn/Arg/Lys reference strains, and was significantly different from each. As with second-site mutations at E2 position 159, if suppression involved a mechanism independent of a charge interaction between the amino acids at E2 62 and E2 114, then one might expect E2 Asp 62 to suppress an attenuating Phe mutation at position II4. The substitution at E2 Asp 62 in conjunction with the E2 Phe 114 resulted in a suppression phenotype characterized by a reduced MST that was not statistically different (P < 0"1) from the virulent Asn/Ser/Lys virus (Table 1). Therefore, as demonstrated for E2 159, second-site suppression of E2 62 appeared to be independent of any charge interactions with attenuating residues at E2 114. The substitution of Glu for Lys at E2 t 59 suppressed the attenuated phenotype specified by either Arg or Phe at E2 114. The same mutation in an otherwise wild-type background also increased virulence as judged by a significant ( P < 0"001) decrease in MST (Table 1; Schoepp & Johnston, 1992). These data suggest that suppression results from an increase in virulence effected independently within the E2c site, rather than a specific interaction with E2 residue 114. The results presented here have several implications with regard to vaccine design. Live, attenuated virus vaccines have been highly successful in controlling viral
1694
Short communication
disease in human and animal populations. The major problem with such vaccines is reversion to a more virulent phenotype. With the construction of full-length cDNA clones of alphavirus RNA genomes, it is possible to minimize the problem of same-site reversion by incorporating multiple, independently attenuating mutations into genetically engineered, prototype alphavirus vaccines (Polo & Johnston, 1990; Davis et al., 1991). Candidate loci for inclusion in such vaccines can be chosen and attenuating codons at each locus can be designed such that two reverting mutations will be required at a single locus to replace the attenuating amino acid with one that specifies a virulent phenotype (Schoepp & Johnston, 1992). The possibilities for secondsite reversion to a more virulent phenotype are more difficult to predict and consequently more difficult to circumvent. Likewise, small differences in genetic background, even a single amino acid change, may significantly decrease the efficacy of an attenuating mutation. This would occur if the background difference was at a locus capable of second-site suppression of the attenuated phenotype. Second-site phenotypic suppression is classically intergenic; there is a physical interaction between two protein products to form an active complex. The primary mutation in one of the proteins disrupts the complex or prevents its function and the suppressing mutation in the other protein allows re-establishment of the complex and restoration of biological activity. However, with a multifactorial phenotype, such as virulence in an animal model, a primary attenuating mutation and a second-site suppressing mutation could occur in different proteins that are never physically associated. In this case, each protein could affect different aspects of virus replication in vivo. The primary mutation could decrease the efficiency of one viral function, whereas the suppressing mutation could specify a compensating increase in the efficiency of another function, resulting in a double mutant as virulent as the wild-type virus. A hypothetical example would be a primary glycoprotein gene mutation that decreased the binding of virus to neurons, paired with a compensating mutation in one of the polymerase subunits that increased virus yield. An analogous situation could occur in separate domains of a multifunctional protein. Similarly, intragenic second-site suppressing amino acid substitutions could interact directly with the amino acid substitution of the primary lesion to restore a functional structure. Alternatively, the primary and suppressing mutations could affect the same domain of the protein but could do so independently and with opposite effects. This appears to be the case with phenotypic suppression in the E2c neutralizing antigenic site. This domain may constitute the site on virions that
binds to receptors on neurons, based on studies with an anti-idiotype antibody raised against a MAb that competes with the E2c-defining MAbs (Ubol & Griffin, 1991). An attenuating mutation at E2 114 increases the sensitivity of the virus to neutralization by the E2cdefining MAbs, and E2c MAb escape mutants map to E2 residues 62, 96 and 159, suggesting that these four residues are in close proximity on the virion surface (Pence et al., 1990). Some amino acid substitutions at E2 positions 62 and 159 are attenuating (Schoepp & Johnston, 1992). However, E2 Asp 62 or Glu 159, the substitutions that most strongly revert the attenuating Arg 114 mutation, significantly increase virulence when situated in wild-type background sequences. One explanation consistent with the results is that these substitutions place the E2c domain in an enhanced conformation which more efficiently mediates interactions (such as binding) with key target cells in vivo. An attenuating mutation that could normally disrupt the wild-type domain might be less effective in disrupting the enhanced conformation. In summary, we have utilized the Sindbis virus system to define a finite number of mutations affecting attenuation and reversion and to assess the effects of specific combinations of amino acid substitutions on second-site phenotypic suppression. These studies have demonstrated that certain amino acid residues at E2 positions 62 and 159 can act as second-site suppressors relative to a primary attenuating mutation at E2 114. Second-site suppression in this case appears to be independent of charge interactions between E2 114 and residues at E2 positions 62 or 159. A complete understanding of the mechanism(s) of these second-site phenotypic reversions will require detailed structural analysis of the Sindbis glycoprotein spike, which as yet is not available. This work was supported by Public Health Service grants AI22186 and NS26681 from the NIH and by the U.S. Army Research and Development Command, Contract DAMD 17-91-C-1092. R.J. Schoepp was supported by an NIH Postdoctoral Fellowship, AI08317.
References BA~Ic, R. S., TRENT,D. W. & JOHNSTON,R. E. (1981). A Sindbis virus variant with a cell-determined latent period. Virology 110, 23~242. BOSE, H.R. & SAGIK, B.P. (1970). The virus envelope in cell attachment. Journal of General Virology 9, 159-161. DALRYMPLE,J. M., SCHLESINGER,S. & RUSSELL,P. K. (1976). Antigenic characterization of two Sindbis envelope glycoproteins separated by isoelectric focusing. Virology 69, 93-103. DAVIS, N. L., FULLER,F. J., DOUGHERTY,W. G., OLMSTI~D,R. A. & JOm~STON, R.E. (1986). A single nucleotide change in the E2 glycoprotein gene of Sindbis virus affects penetration rate in cell culture and virulence in neonatal mice. Proceedings of the National Academy of Sciences, U.S.A. 83, 6771 6775. DAVIS,N. L., POWELL, N., GREENWALD,G. F., WILLIS,L. V., JOHNSON, B.J.B., SMITH, J.F. & JOHNSTON, R.E. (1991). Attenuating mutations in the E2 glycoprotein gene of Venezuelan equine
Short communication
encephalitis virus: construction of single and multiple mutants in a full-length cDNA clone. Virology 183, 20-31. FLYNN, D. C., OLMSTEO, R.A., MACKENZIE, J. M., JR & JOHNSTON, R.E. (1988). Antibody-mediated activation of Sindbis virus. Virology 166, 82-90. GLASGOW, G. M., SHEAHAN, B.J., ATKINS, G.J., WAHLBERG, J. M., SALMtNEN, A. & LILJESTROM, P. (1991). Two mutations in the envelope glycoprotein E2 of Semliki Forest virus affecting the maturation and entry patterns of the virus alter pathogenicity for mice. Virology 185, 741-748. KUNKEL, T. A., ROBERTS, J. D. & ZAKOUR, R.A. (1987). Rapid and efficient site-specific mutagenesis without phenotypic selection. Methods in Enzymology 154, 367-382. LUSTIG, S., JACKSON, A. C., HAHN, C. S., GRIFFIN, D.E., STRAUSS, E.G. & STRAUSS, J.H. (1988). Molecular basis of Sindbis virus neurovirulence in mice. Journal of Virology 62, 2329-2336. OLMSTED, R.A., BARIC, R.S., SAWYER,B . k . & JOHNSTON,R.E. (1984). Sindbis virus mutants selected for rapid growth in cell culture display attenuated virulence for animals. Science 225, 424-427. OLMSTED, R. A., MEYER, W. J. & JOHNSTON, R. E. (1986). Characterization of Sindbis virus epitopes important for penetration in cell culture and pathogenesis in animals. Virology 148, 245-254. PAREDES,A. M., SIMON,M. N. & BROWN, D. T. (1992). The mass of the Sindbis virus nucleocapsid suggests it has T = 4 icosahedral symmetry. Virology 187, 329-332. PENCE, D. F., DAVIS, N. L. & JOHNSTON, R. E. (1990). Antigenic and genetic characterization of Sindbis virus monoclonal antibody escape mutants which define a pathogenesis domain on glycoprotein E2. Virology 175, 4149. POLO, J.M. & JOHNSTON, R.E. (1990). Attenuating mutations in glycoproteins E 1 and E2 of Sindbis virus produce a highly attenuated strain when combined in vitro. Journal of Virology 64, 4438-4444. POLO, J.M. & JOHNSTON, R.E. (1991). Mutational analysis of a virulence locus in the E2 glycoprotein gene of Sindbis virus. Journal of Virology 65, 6358-6361. POLO, J. M., DAVIS, N. L., RICE, C. M., HUANG, H. V. & JOHNSTON, R.E. (1988). Molecular analysis of Sindbis virus pathogenesis in neonatal mice by using virus recombinants constructed in vitro. Journal of Virology 62, 2124-2133. RICE, C.M. & STRAUSS, J. H. (1982). Association of Sindbis virion glycoproteins and their precursors. Journal of Molecular Biology 154, 325-348.
1695
RICE, C.M., LEVIS, R., STRAUSS, J.H. • HUANG, H.V. (1987). Production of infectious RNA transcripts from Sindbis virus cDNA clones: mapping of lethal mutations, rescue of a temperaturesensitive marker, and in vitro mutagenesis to generate defined mutants. Journal of Virology 61, 3809-3819. RUSSELL, D. L., DALRYMPLE, J. M. & JOHNSTON, R. E. (1989). Sindbis virus mutations which coordinately affect glycoprotein processing, penetration, and virulence in mice. dournal of Virology 63, 1619-1629. SeHLESINGER, S. & SCHLESINGER, M.J. (1986). The Togaviridae and Flaviviridae. New York: Plenum Press. SCHLESlNGER, M. J., SCHLESINGER, S. & BURGE, B. W. (1972). Identification of a second glycoprotein in Sindbis virus. Virology 47, 539-541. SCHMALJOHN,A. L., KOKUBUN,K. M. & COLE, G. A. (1983). Protective monoclonal antibodies define maturational and pH-dependent changes in Sindbis virus E1 glycoproteins. Virology 130, 144-154. SCHOEPP, R. J. & JOHNSTON, R. E. (1992). Directed mutagenesis of a Sindbis virus pathogenesis site. Virology 193, 149-159. SEFTON, B.M., WICKUS, G . G . & BURGE, B.W. (1978). Enzymatic iodination of Sindbis virus proteins. Journal of Virology 11,730-735. STANLEY, J., COOPER, S.J. & GRIFFIN, D.E. (1985). Alphavirus neurovirulence: monoclonal antibodies discriminating wild-type from neuroadapted Sindbis virus. Journal of Virology 56, 110-119. STRAUSS, J. H., BURGE, B. W., PFEFFERKORN, E. R. & DARNELL, J. E. (1969). Identification of the membrane protein and "core" protein of Sindbis virus. Proceedings of the National Academy of Sciences, U.S.A. 59, 533 537. TUCKER, P. C. & GRIFFIN, D. E. (1991). Mechanism of altered Sindbis virus neurovirulence associated with a single-amino-acid change in the E2 glycoprotein. Journal of Virology 65, 1551 1557. UBOL, S. & GRIFFITH, D.E. (1991). Identification of a putative alphavirus receptor on mouse neural cells. Journal of Virology 65, 6913-6921. VON BONSDORFF, C.H. & HARRISON, S.C. (1975). Sindbis virus glycoproteins form a regular icosahedral surface lattice. Journal of Virology 16, 141 145. VRATI, S., FARAGHER,S. G., WEIR, R. C. & DALGARNO, L. (1986). Ross River virus mutant with a deletion in the E2 gene: properties of the virion, virus-specific macromolecule synthesis, and attenuation of virulence for mice. Virology 151, 222-232.
(Received 23 November 1992; Accepted 25 March 1993)
1691
Sindbis virus pathogenesis: phenotypic reversion of an attenuated strain to virulence by second-site intragenic suppressor mutations Randal J. Schoepp and Robert E. Johnston* Department o f Microbiology and Immunology, University o f North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599, U.S.A.
Monoclonal antibodies (MAbs) specific for the E2c neutralizing antigenic site on the Sindbis virus E2 glycoprotein define a pathogenesis domain that affects neonatal mouse virulence. Sequence analysis of E2c MAb escape mutants showed that the domain included E2 amino acids 62, 96 and 159. The pathogenesis domain is also influenced by changes at E2 position 114. Mutation of E2 residues Asn 62 to Asp or Lys 159 to Glu results in suppression of the attenuated phenotype conferred by a mutation from Ser to Arg at E2 position 114. Possible mechanisms of phenotypic
suppression within the E2c pathogenesis domain were investigated by using site-directed mutagenesis to determine the effects of specific combinations of positively charged, negatively charged and uncharged amino acid substitutions at E2 positions 62, 114 and 159. Phenotypic reversion to virulence by second-site suppressor mutations at E2 amino acids 62 or 159 was not dependent on ionic interaction with the residue at E2 114. Rather, suppression appeared to be the result of independent virulence effects mediated by specific residues.
Sindbis virus is the prototype of the alphavirus genus in the family Togaviridae. This group of arthropod-borne viruses includes the agents of eastern, western and Venezuelan equine encephalitis, which are of veterinary and public health importance (Schlesinger & Schlesinger, 1986). Sindbis virus has been instrumental in the elucidation of the molecular basis of alphavirus virulence in animal models, and numerous studies have implicated specific domains on the glycoprotein spikes of alphaviruses as important determinants of pathogenesis (Sefton et al., 1978; Davis et al., 1986; Olmsted et aL, 1986; Vrati et al., 1986; Lustig et al., 1988; Polo et al., 1988; Russell et al., 1989; Polo & Johnston, 1990, 1991; Glasgow et al., 1991 ; Tucker & Griffin, 1991 ; Schoepp & Johnston, 1992). The glycoprotein spikes are heterodimers composed of glycoproteins E1 and E2, which are anchored near their carboxy termini in the cell-derived lipoprotein envelope that surrounds the nucleocapsid (Strauss et al., 1969; Schlesinger et al., 1972; von Bonsdorff & Harrison, 1975; Rice & Strauss, 1982). The icosahedral nucleocapsid is composed of 240 copies of the C protein and surrounds the single-stranded, message-sense RNA genome (Strauss et al., 1969; Paredes et al., 1992). The glycoproteins are the means by which the virus interacts with the surrounding environment (Sefton et al., 1978) and are responsible for virus attachment and penetration (Bose & Sagik, 1970; Olmsted et al., 1984, 1986; Flynn et al., 1988), haemagglutination (Dalrymple et al., 1976; Schmaljohn et al.,
1983) and induction of an immune response (Schmaljohn et al., 1983). A number of single-site mutations that affect Sindbis pathogenesis in mice have been identified, predominantly in the 5" halves of the E1 and E2 glycoprotein genes (Davis et al., 1986; Lustig et al., 1988; Russell et al., 1989; Pence et al., 1990; Polo & Johnston, 1990; Tucker & Griffin, 1991). For example, the influence on virulence of the residue at position 114 in E2 was demonstrated in studies of the Sindbis virus mutant SB-RL, derived from Sindbis strain AR339 (SB). SB-RL is attenuated in neonatal mice and has an increased rate of penetration into BHK-21 cells (Baric et al., 1981; Olmsted et al., 1984, 1986). These phenotypic differences are produced by substitution of Arg at E2 position 114 in SB-RL for Ser found in SB (Davis et al., 1986). The phenotypic effects on E2 of Arg 114 were confirmed by site-directed mutagenesis (SDM) of a full-length cDNA clone of Sindbis virus from which infectious virus can be derived (Rice et al., 1987; Polo et al., 1988). In addition to Arg, substitution of Lys at E2 position 114 was strongly attenuating, and the substitution of Phe, Tyr or Trp attenuated virulence to a lesser degree (Polo & Johnston, 1991). Another of the phenotypic effects on E2 of an Arg 114 substitution is an increase in sensitivity to neutralization by monoclonal antibodies (MAbs) recognizing the E2c antigenic site (Davis et al., 1986; Polo et al., 1988). This observation suggested that the E2 114 locus might be a
0001-1489 © 1993SGM
1692
Short communication
constituent of, or be in the vicinity of, the E2c site. The E2c site was mapped by selecting neutralization escape mutants of SB and SB-RL with two E2c-specific MAbs, R6 and R13 (Pence et al., 1990). From SB and SB-RL 11 E2c MAb escape mutants were selected, all of which showed greatly reduced binding to either R6 or R13. Sequence analysis of the E2 and E1 genes of these mutants demonstrated that each contained a single mutation in E2 at either position 62, 96 or 159. SDM of E2 gene codons 62, 96 and 159 confirmed the participation of these loci in the E2c site (Schoepp & Johnston, 1992). Moreover, a general correlation was observed between disruption of the E2c site conformation by a particular amino acid substitution, as measured by MAb binding, and the reduction of virulence in neonatal mice. Therefore, Sindbis virus mutations in E2 gene codons 62, 96, 114 and 159 affect both virulence in neonatal mice and the binding or biological activity of E2c-specific MAbs, suggesting that these residues are constituents of a virion structure important for Sindbis pathogenesis (Pence et al., 1990; Schoepp & Johnston, 1992). Certain mutations in the E2c site affected pathogenesis by phenotypically suppressing the attenuating Arg mutation at position 114. Neutralization escape mutants in which Ash 62 was changed to Asp, or Lys 159 was mutated to Glu, suppressed the attenuated phenotype of E2 Arg 114 in the SB-RL genetic background, resulting in a reversion to virulence. E2 position 114 is the central amino acid in a stretch of uncharged and hydrophobic residues (Davis et al., 1986). The attenuated phenotype of Arg 114 mutants may result from the putative disruptive effect of a large, positively charged residue located within this hydrophobic region. If residue 114 interacts directly with residues 62 and 159 then the destabilizing effect of an Arg residue at position 114 could be neutralized completely or in part by a negatively charged amino acid at E2 62 or 159 to restore the structure and, consequently, a more virulent phenotype. Consistent with this hypothesis is the fact that the suppressing amino acids at E2 62 and 159 in the escape mutants were acidic (Pence et al., 1990). Alternatively, the amino acids at each of the three loci could affect the virulence phenotype independently, with the phenotype of the double mutants resulting from the algebraic sum of independent virulence effects. In this study, possible mechanisms of phenotypic suppression within the E2c pathogenesis domain were investigated by determining the effects of specific combinations of positively charged, negatively charged and uncharged amino acid substitutions at E2 positions 62, 114 and 159. SDM (Kunkel et al., 1987) of a full-length Sindbis virus clone (pTR5500) was utilized to produce a panel of mutant viruses with different amino acid substitutions at positions 62, 114 and 159 in E2 (Schoepp & Johnston,
1992). Four uracil-containing M13-based mutagenesis cassettes contained the glycoprotein genes E2 and El, and differed only in amino acid 114. The reference virus strains derived from these mutagenesis vectors were isogenic except for the substitution of Ser, Arg, Phe or Glu at E2 114. Mutagenesis with synthetic oligonucleotide primers, as described previously (Polo & Johnston, 1990, 1991; Schoepp & Johnston, 1992), was used to introduce an additional mutation at E2 position 62 or 159. The specific oligonucleotide primers were 'doped' within the sequences encoding E2 amino acid 62 [nucleotides (nt) 8814 to 8816] or 159 (nt 9105 to 9107) during synthesis: 5' CAGCAAGCGCA(C/G)(A/G)CAAGTACCGCTA 3' and 5' CGACCGTCTG(C/G)AAGAAACAAC 3'. Phosphorylated primers and the uracilcontaining templates were annealed, and the second strand was completed. The dsDNA was transformed into strain JM101 of Escherichia coli. Isolated plaques were screened for the presence of mutations by sequence analysis. A restriction fragment encompassing the E2 and El genes was reinserted into the full-length cDNA clone using unique restriction sites, StuI (nt 8571) and BssHII (nt 9804). Sequence replacements were monitored using a previously engineered restriction site marker that did not change the amino acid sequence (loss of the PstI site at nt 9119). Infectious RNA, transcribed from the SP6 promoter of linearized clones, was sequenced directly to confirm the mutation, and was then transfected into BHK cells to produce virus (Polo & Johnston, 1992). Each virus was subcutaneously inoculated into two to seven litters of 1-day-old CD-1 mice (averaging 10 animals per litter) at a dose of 100 p.f.u./animal. After a 14 day observation period, percentage mortality and mean survival time (MST) for those animals which died were calculated for each virus. Statistical analysis was by Student's t-test, and differences were considered significant if P < 0.001. Initial studies were designed to determine whether a charged amino acid substitution at E2 114 was required for an attenuated phenotype. Isogenic viruses that differed only by the amino acid encoded at E2 114 were inoculated into neonatal mice. The E2 114 construct, containing the wild-type Ser at this position, yielded a virus that caused 100% mortality and a short MST (4.7 + 0.7 days) characteristic of virulent SB virus strains (Table 1). This virus was designated the virulent reference virus strain, with Asn/Ser/Lys at E2 residues 62, 114 and 159, respectively. Similarly, the E2 Arg 114 construct, analogous to the attenuated SB-RL virus strain, was attenuated upon subcutaneous inoculation. This virus was designated the attenuated reference virus strain (with Asn/Arg/Lys as residues 62/114/159). Substitution of the uncharged, phenolic ring-containing Phe at
Short communication
Table 1. Viru&nceeffectsofsubstitut~ns at E21oci62, 114 or 159 E2 codon 62
114
159
Mortality (%)
MST (days)
Asn* Asn$ Asn Asn Asn Asn Asn Asn Asn Asn Asp Asp
Ser* Arg$ Phe Glu Ser Ser Arg Arg Phe Phe Arg Phe
Lys* Lys$ Lys Lys Glu Gin Glu Gin Glu Gln Lys Lys
100 (0)f 22 (3) 100 (0) 100 (0) 100 (0) 100 (0) 96 (5) 58 (37) 100 (0) 100 (0) 100 (0) 100 (0)
4.7 (0"7)f 11.8 (1.5) 7.4 (1'4) 5-1 (1.1) 3-5 (0'8) 5-1 (1.2) 6"9 (1,4) 7.5 (2.3) 5.4 (1.2) 5.4 (1-2) 6.2 (1.2) 5.3 (1.7)
* Wild-typeamino acid of virulent referencestrain. t Calculationswere based on inoculationof two to sevenlitters of neonatal mice (averaging 10 animals per litter). S.E.M. is given in parentheses. $ Aminoacid of attenuatedreferencestrain.
E2 114 yielded a virus that caused 100 % mortality yet resulted in significantly protracted survival compared to the virulent virus containing the wild-type amino acid (P < 0-001). In contrast, substitution of the negatively charged Glu yielded a virus that was no less virulent than wild-type. These results are consistent with a previous study in which a library of different amino acid substitutions were made at E2 position 114 in a slightly different genetic background (Polo & Johnston, 1991). The majority of the substitutions resulted in virulent viruses. Attenuated viruses could be divided into two groups, first, those which were strongly attenuating and had large, positively charged amino acids (Lys and Arg) and second, those which were more weakly attenuating and had mutations with large amino acids with ring structures (Tyr, Trp and Phe). However, neither Glu, as shown here, nor Asp were attenuating, suggesting that a negative charge did not confer an attenuating phenotype in this sequence context. Therefore, a charged amino acid at E2 position 114 was not required for attenuation, although large, positively charged amino acids produced the most pronounced attenuation phenotype. The genotype of the MAb escape mutant containing Glu 159 and Arg 114 (Pence et al., 1990) was reproduced in the full-length cDNA clone, and the second-site suppressing activity of the Glu 159 mutation was confirmed (Table 1). Substitution of Glu at 159 caused a partial reversion of the attenuated E2 Arg 114 phenotype toward virulence, although the double mutant was not as virulent as the wild-type Asn/Ser/Lys construct (Table 1). This virus caused 96% mortality in neonatal mice with a significantly shortened MST (P < 0"001) when compared to the attenuated reference virus containing
1693
amino acids Asn/Arg/Lys at E2 positions 62, 114 and 159, respectively. A Gin substitution also partially suppressed the E2 Arg 114 phenotype. However, as Gln 159 represented the most conservative uncharged substitution for the E2 Glu 159 suppressor mutation, this result suggested that negative charge at E2 position 159 was not required for suppression of the attenuated phenotype. Because suppression appeared to be independent of an ionic interaction between E2 residue 159 and attenuating residue 114, we determined the effect of substitution of Glu or Gln at E2 position 159 on the attenuation phenotype specified by an uncharged amino acid, E2 Phe 114. Substitution of either Glu or Gln for Lys at position 159, with an attenuating Phe at 114, resulted in a virulent virus with a significantly reduced MST. Both Glu and Gln suppressed Phe 114, just as they suppressed Arg 114. Therefore, with respect to E2 159, second-site suppression was independent of any charge interactions with the attenuating residue at E2 114. Analogous constructs were used to determine the requirements for suppressing amino acids at E2 62. The genotype of the MAb escape mutant containing E2 Asp 62 and E2 Arg 114 was reproduced in the full-length clone. Virus derived from constructs containing E2 Asp62 and E2 Arg 114 induced 100% mortality in neonatal mice. The MST was intermediate between the virulent Asn/Ser/Lys and attenuated Asn/Arg/Lys reference strains, and was significantly different from each. As with second-site mutations at E2 position 159, if suppression involved a mechanism independent of a charge interaction between the amino acids at E2 62 and E2 114, then one might expect E2 Asp 62 to suppress an attenuating Phe mutation at position II4. The substitution at E2 Asp 62 in conjunction with the E2 Phe 114 resulted in a suppression phenotype characterized by a reduced MST that was not statistically different (P < 0"1) from the virulent Asn/Ser/Lys virus (Table 1). Therefore, as demonstrated for E2 159, second-site suppression of E2 62 appeared to be independent of any charge interactions with attenuating residues at E2 114. The substitution of Glu for Lys at E2 t 59 suppressed the attenuated phenotype specified by either Arg or Phe at E2 114. The same mutation in an otherwise wild-type background also increased virulence as judged by a significant ( P < 0"001) decrease in MST (Table 1; Schoepp & Johnston, 1992). These data suggest that suppression results from an increase in virulence effected independently within the E2c site, rather than a specific interaction with E2 residue 114. The results presented here have several implications with regard to vaccine design. Live, attenuated virus vaccines have been highly successful in controlling viral
1694
Short communication
disease in human and animal populations. The major problem with such vaccines is reversion to a more virulent phenotype. With the construction of full-length cDNA clones of alphavirus RNA genomes, it is possible to minimize the problem of same-site reversion by incorporating multiple, independently attenuating mutations into genetically engineered, prototype alphavirus vaccines (Polo & Johnston, 1990; Davis et al., 1991). Candidate loci for inclusion in such vaccines can be chosen and attenuating codons at each locus can be designed such that two reverting mutations will be required at a single locus to replace the attenuating amino acid with one that specifies a virulent phenotype (Schoepp & Johnston, 1992). The possibilities for secondsite reversion to a more virulent phenotype are more difficult to predict and consequently more difficult to circumvent. Likewise, small differences in genetic background, even a single amino acid change, may significantly decrease the efficacy of an attenuating mutation. This would occur if the background difference was at a locus capable of second-site suppression of the attenuated phenotype. Second-site phenotypic suppression is classically intergenic; there is a physical interaction between two protein products to form an active complex. The primary mutation in one of the proteins disrupts the complex or prevents its function and the suppressing mutation in the other protein allows re-establishment of the complex and restoration of biological activity. However, with a multifactorial phenotype, such as virulence in an animal model, a primary attenuating mutation and a second-site suppressing mutation could occur in different proteins that are never physically associated. In this case, each protein could affect different aspects of virus replication in vivo. The primary mutation could decrease the efficiency of one viral function, whereas the suppressing mutation could specify a compensating increase in the efficiency of another function, resulting in a double mutant as virulent as the wild-type virus. A hypothetical example would be a primary glycoprotein gene mutation that decreased the binding of virus to neurons, paired with a compensating mutation in one of the polymerase subunits that increased virus yield. An analogous situation could occur in separate domains of a multifunctional protein. Similarly, intragenic second-site suppressing amino acid substitutions could interact directly with the amino acid substitution of the primary lesion to restore a functional structure. Alternatively, the primary and suppressing mutations could affect the same domain of the protein but could do so independently and with opposite effects. This appears to be the case with phenotypic suppression in the E2c neutralizing antigenic site. This domain may constitute the site on virions that
binds to receptors on neurons, based on studies with an anti-idiotype antibody raised against a MAb that competes with the E2c-defining MAbs (Ubol & Griffin, 1991). An attenuating mutation at E2 114 increases the sensitivity of the virus to neutralization by the E2cdefining MAbs, and E2c MAb escape mutants map to E2 residues 62, 96 and 159, suggesting that these four residues are in close proximity on the virion surface (Pence et al., 1990). Some amino acid substitutions at E2 positions 62 and 159 are attenuating (Schoepp & Johnston, 1992). However, E2 Asp 62 or Glu 159, the substitutions that most strongly revert the attenuating Arg 114 mutation, significantly increase virulence when situated in wild-type background sequences. One explanation consistent with the results is that these substitutions place the E2c domain in an enhanced conformation which more efficiently mediates interactions (such as binding) with key target cells in vivo. An attenuating mutation that could normally disrupt the wild-type domain might be less effective in disrupting the enhanced conformation. In summary, we have utilized the Sindbis virus system to define a finite number of mutations affecting attenuation and reversion and to assess the effects of specific combinations of amino acid substitutions on second-site phenotypic suppression. These studies have demonstrated that certain amino acid residues at E2 positions 62 and 159 can act as second-site suppressors relative to a primary attenuating mutation at E2 114. Second-site suppression in this case appears to be independent of charge interactions between E2 114 and residues at E2 positions 62 or 159. A complete understanding of the mechanism(s) of these second-site phenotypic reversions will require detailed structural analysis of the Sindbis glycoprotein spike, which as yet is not available. This work was supported by Public Health Service grants AI22186 and NS26681 from the NIH and by the U.S. Army Research and Development Command, Contract DAMD 17-91-C-1092. R.J. Schoepp was supported by an NIH Postdoctoral Fellowship, AI08317.
References BA~Ic, R. S., TRENT,D. W. & JOHNSTON,R. E. (1981). A Sindbis virus variant with a cell-determined latent period. Virology 110, 23~242. BOSE, H.R. & SAGIK, B.P. (1970). The virus envelope in cell attachment. Journal of General Virology 9, 159-161. DALRYMPLE,J. M., SCHLESINGER,S. & RUSSELL,P. K. (1976). Antigenic characterization of two Sindbis envelope glycoproteins separated by isoelectric focusing. Virology 69, 93-103. DAVIS, N. L., FULLER,F. J., DOUGHERTY,W. G., OLMSTI~D,R. A. & JOm~STON, R.E. (1986). A single nucleotide change in the E2 glycoprotein gene of Sindbis virus affects penetration rate in cell culture and virulence in neonatal mice. Proceedings of the National Academy of Sciences, U.S.A. 83, 6771 6775. DAVIS,N. L., POWELL, N., GREENWALD,G. F., WILLIS,L. V., JOHNSON, B.J.B., SMITH, J.F. & JOHNSTON, R.E. (1991). Attenuating mutations in the E2 glycoprotein gene of Venezuelan equine
Short communication
encephalitis virus: construction of single and multiple mutants in a full-length cDNA clone. Virology 183, 20-31. FLYNN, D. C., OLMSTEO, R.A., MACKENZIE, J. M., JR & JOHNSTON, R.E. (1988). Antibody-mediated activation of Sindbis virus. Virology 166, 82-90. GLASGOW, G. M., SHEAHAN, B.J., ATKINS, G.J., WAHLBERG, J. M., SALMtNEN, A. & LILJESTROM, P. (1991). Two mutations in the envelope glycoprotein E2 of Semliki Forest virus affecting the maturation and entry patterns of the virus alter pathogenicity for mice. Virology 185, 741-748. KUNKEL, T. A., ROBERTS, J. D. & ZAKOUR, R.A. (1987). Rapid and efficient site-specific mutagenesis without phenotypic selection. Methods in Enzymology 154, 367-382. LUSTIG, S., JACKSON, A. C., HAHN, C. S., GRIFFIN, D.E., STRAUSS, E.G. & STRAUSS, J.H. (1988). Molecular basis of Sindbis virus neurovirulence in mice. Journal of Virology 62, 2329-2336. OLMSTED, R.A., BARIC, R.S., SAWYER,B . k . & JOHNSTON,R.E. (1984). Sindbis virus mutants selected for rapid growth in cell culture display attenuated virulence for animals. Science 225, 424-427. OLMSTED, R. A., MEYER, W. J. & JOHNSTON, R. E. (1986). Characterization of Sindbis virus epitopes important for penetration in cell culture and pathogenesis in animals. Virology 148, 245-254. PAREDES,A. M., SIMON,M. N. & BROWN, D. T. (1992). The mass of the Sindbis virus nucleocapsid suggests it has T = 4 icosahedral symmetry. Virology 187, 329-332. PENCE, D. F., DAVIS, N. L. & JOHNSTON, R. E. (1990). Antigenic and genetic characterization of Sindbis virus monoclonal antibody escape mutants which define a pathogenesis domain on glycoprotein E2. Virology 175, 4149. POLO, J.M. & JOHNSTON, R.E. (1990). Attenuating mutations in glycoproteins E 1 and E2 of Sindbis virus produce a highly attenuated strain when combined in vitro. Journal of Virology 64, 4438-4444. POLO, J.M. & JOHNSTON, R.E. (1991). Mutational analysis of a virulence locus in the E2 glycoprotein gene of Sindbis virus. Journal of Virology 65, 6358-6361. POLO, J. M., DAVIS, N. L., RICE, C. M., HUANG, H. V. & JOHNSTON, R.E. (1988). Molecular analysis of Sindbis virus pathogenesis in neonatal mice by using virus recombinants constructed in vitro. Journal of Virology 62, 2124-2133. RICE, C.M. & STRAUSS, J. H. (1982). Association of Sindbis virion glycoproteins and their precursors. Journal of Molecular Biology 154, 325-348.
1695
RICE, C.M., LEVIS, R., STRAUSS, J.H. • HUANG, H.V. (1987). Production of infectious RNA transcripts from Sindbis virus cDNA clones: mapping of lethal mutations, rescue of a temperaturesensitive marker, and in vitro mutagenesis to generate defined mutants. Journal of Virology 61, 3809-3819. RUSSELL, D. L., DALRYMPLE, J. M. & JOHNSTON, R. E. (1989). Sindbis virus mutations which coordinately affect glycoprotein processing, penetration, and virulence in mice. dournal of Virology 63, 1619-1629. SeHLESINGER, S. & SCHLESINGER, M.J. (1986). The Togaviridae and Flaviviridae. New York: Plenum Press. SCHLESlNGER, M. J., SCHLESINGER, S. & BURGE, B. W. (1972). Identification of a second glycoprotein in Sindbis virus. Virology 47, 539-541. SCHMALJOHN,A. L., KOKUBUN,K. M. & COLE, G. A. (1983). Protective monoclonal antibodies define maturational and pH-dependent changes in Sindbis virus E1 glycoproteins. Virology 130, 144-154. SCHOEPP, R. J. & JOHNSTON, R. E. (1992). Directed mutagenesis of a Sindbis virus pathogenesis site. Virology 193, 149-159. SEFTON, B.M., WICKUS, G . G . & BURGE, B.W. (1978). Enzymatic iodination of Sindbis virus proteins. Journal of Virology 11,730-735. STANLEY, J., COOPER, S.J. & GRIFFIN, D.E. (1985). Alphavirus neurovirulence: monoclonal antibodies discriminating wild-type from neuroadapted Sindbis virus. Journal of Virology 56, 110-119. STRAUSS, J. H., BURGE, B. W., PFEFFERKORN, E. R. & DARNELL, J. E. (1969). Identification of the membrane protein and "core" protein of Sindbis virus. Proceedings of the National Academy of Sciences, U.S.A. 59, 533 537. TUCKER, P. C. & GRIFFIN, D. E. (1991). Mechanism of altered Sindbis virus neurovirulence associated with a single-amino-acid change in the E2 glycoprotein. Journal of Virology 65, 1551 1557. UBOL, S. & GRIFFITH, D.E. (1991). Identification of a putative alphavirus receptor on mouse neural cells. Journal of Virology 65, 6913-6921. VON BONSDORFF, C.H. & HARRISON, S.C. (1975). Sindbis virus glycoproteins form a regular icosahedral surface lattice. Journal of Virology 16, 141 145. VRATI, S., FARAGHER,S. G., WEIR, R. C. & DALGARNO, L. (1986). Ross River virus mutant with a deletion in the E2 gene: properties of the virion, virus-specific macromolecule synthesis, and attenuation of virulence for mice. Virology 151, 222-232.
(Received 23 November 1992; Accepted 25 March 1993)
