Genetic and Phenotypic Changes Accompanying the Emergence of Epizootic Subtype IC Venezuelan Equine Encephalitis Viruses from an Enzootic Subtype ID Progenitor

Abstract

Recent studies have indicated that epizootic Venezuelan equine encephalitis (VEE) viruses can evolve from enzootic, subtype ID strains that circulate continuously in lowland tropical forests (A. M. Powers, M. S. Oberste, A. C. Brault, R. Rico-Hesse, S. M. Schmura, J. F. Smith, W. Kang, W. P. Sweeney, and S. C. Weaver, J. Virol. 71:6697–6705, 1997). To identify mutations associated with the phenotypic changes leading to epizootics, we sequenced the entire genomes of two subtype IC epizootic VEE virus strains isolated during a 1992–1993 Venezuelan outbreak and four sympatric, subtype ID enzootic strains closely related to the predicted epizootic progenitor. Analysis by maximum-parsimony phylogenetic methods revealed 25 nucleotide differences which were predicted to have accompanied the 1992 epizootic emergence; 7 of these encoded amino acid changes in the nsP1, nsP3, capsid, and E2 envelope glycoprotein, and 2 were mutations in the 3′ untranslated genome region. Comparisons with the genomic sequences of IAB and other IC epizootic VEE virus strains revealed that only one of the seven amino acid changes associated with the 1992 emergence, a threonine-to-methionine change at position 360 of the nsP3 protein, accompanied another VEE virus emergence event. Two changes in the E2 envelope glycoprotein region believed to include the major antigenic determinants, both involving replacement of uncharged residues with arginine, are also candidates for epizootic determinants.

Venezuelan equine encephalitis (VEE) virus is a member of the Alphavirus genus of the family Togaviridae, a group of enveloped, single-stranded, positive-sense RNA viruses (9, 29). The VEE virus genome is about 11.4 kb; the 5′ two-thirds encodes four nonstructural proteins (nsP1 through 4) required for replication of the RNA genome, and the 3′ one-third encodes the structural proteins (capsid and envelope glycoproteins E2 and E1). The genomic RNA is capped at the 5′ terminus with 7-methylguanosine and is polyadenylated at the 3′ terminus. The nonstructural proteins are translated from genomic RNA as one or two polyproteins (nsP1-3 or nsP1-4) and proteolytically cleaved to produce nsP1, nsP2, nsP3, and nsP4, as well as partially cleaved polyproteins. The structural proteins are processed from a polyprotein translated from a 26S subgenomic mRNA that is identical in sequence to the 3′-terminal one-third of the genomic RNA. The alphavirus virion contains an icosahedral nucleocapsid that consists of 240 copies of the capsid protein, surrounded by a lipid envelope which is derived from the plasma membrane of infected cells, in which glycoprotein E1 and E2 heterodimers are embedded (28, 29).

The VEE antigenic complex is one of three major alphavirus serogroups found in the New World and is comprised of six antigenic subtypes (33). The viruses of subtypes II through VI, as well as subtype I, varieties D to F, are not associated with major epidemics or equine epizootics and are therefore referred to as enzootic strains. In contrast, viruses belonging to subtypes IAB and IC have been isolated only during epizootic and epidemic VEE outbreaks involving up to hundreds of thousands of equines and people, with high rates of morbidity and mortality (33, 34, 38). VEE epidemics and epizootics have occurred sporadically in the Americas since the virus was first isolated in 1938 (1, 14). No outbreaks were reported between 1973 and 1992, prompting speculation that epidemic/epizootic serotypes IAB and IC had become extinct (33). However, after 19 years of absence, equine epizootics caused by subtype IC viruses recurred in Venezuela in 1992 (22) and in Venezuela and Colombia again in 1995 (38).

Phylogenetic studies of all VEE virus serotypes by limited genome sequencing indicate that epizootic subtype IAB and IC viruses have evolved independently at least three times from enzootic, equine-avirulent subtype ID-like strains that circulate continuously in lowland tropical forest habitats of northern South America (21, 37). Comparison of the genomic sequence of an IC strain isolated in a 1963 Venezuelan epidemic (P676) to that of an enzootic ID virus from Panama (3880) showed nucleotide differences yielding only 66 amino acid differences (13), also suggesting that ID strains may be progenitors of epizootic strains.

Partial genomic sequencing of two epizootic IC isolates (243937 and SH3) isolated during a 1992–1993 Venezuelan outbreak revealed that they are extremely closely related to two ID isolates (66637 and 66457) from Sinamaica, in northwestern Venezuela near the Guajira Peninsula. These viruses differ by only 11 of 817 nucleotides (1.3%) in partial PE2 sequences (21). This similarity and the deduced phylogenetic relationships provide strong evidence that the 1992–1993 outbreak originated from local enzootic ID viruses that mutated to become subtype IC. Because serotype ID viruses are believed to be avirulent for equines and do not generate sufficient viremia for epizootic transmission (32, 33), this phenotypic transition presumably included the acquisition of equine virulence. To identify the mutations responsible for these phenotypic changes, we sequenced the complete genomes of two subtype IC epizootic strains from the 1992–1993 Venezuelan outbreak and four closely related, sympatric, subtype ID enzootic isolates. Sequence comparisons and phylogenetic analyses indicated that one or more of seven amino acid changes in nsP1, nsP3, the capsid, and E2 may be responsible for epizootic emergence. Plaque size phenotypes of the IC isolates were distinct from those of the ID isolates, consistent with previous studies showing a difference between epizootic and enzootic phenotypes.

MATERIALS AND METHODS

Virus preparation, PCR amplification, and sequencing.

Six VEE virus strains were sequenced in this study. Two subtype IC epizootic strains (243937 and SH3) were isolated during an equine epizootic and epidemic that occurred in western Venezuela during 1992 and 1993 (22). Four ID enzootic strains (66637, 66457, 83U434, and ZPC738) were isolated from sentinel hamsters exposed to the virus in tropical lowland forests of western Venezuela or eastern Colombia at various times between 1981 and 1997 (5, 31) (Table 1). Strain ZPC738, isolated in 1997, was characterized antigenically as subtype ID with monoclonal antibodies in an immunofluorescence assay as described previously (23). Virus stocks were prepared on BHK cell monolayers at 37°C with a multiplicity of infection of 0.1 to 1.0 PFU per cell. After cytopathic effects were evident, RNA was extracted by using Trizol LS (Bethesda Research Laboratories, Bethesda, Md.) according to the manufacturer’s protocol, and reverse transcription-PCR was performed as described previously (2). Six pairs of oligonucleotide primers (Table 2) were designed to produce overlapping amplicons covering the entire VEE virus genome. First-strand cDNAs were synthesized with the antisense primers Mlu-25V, which primes synthesis at the 3′ poly(A) tail of the genomic RNA, and 5251(−). PCR products were cloned into the pCR2.1 vector (Invitrogen, San Diego, Calif.), and bacterial colonies were screened by restriction enzyme digestion. The cDNA of the 5′ terminus of the genome was obtained with antisense primer V-1245(−) and was tailed with dCTP acid at its 3′ terminus by using the 5′ rapid amplification of cDNA ends system (Bethesda Research Laboratories). PCR products and clones were sequenced with an Applied Biosystems (Foster City, Calif.) Prism automated DNA sequencing kit and sequencer according to the manufacturer’s protocol. Additional primers were synthesized for sequencing, and their sequences are available upon request. For each virus strain, two or more clones were sequenced to confirm the nucleotide differences found among virus strains. Discrepancies were resolved by sequencing PCR amplicons directly to identify the consensus nucleotides.

TABLE 1.

VEE viruses sequenced

Strain	Subtype	Date of isolation	Place of isolation	Host	Passage historya
66637	ID	19 November 1981	Zulia State, Venezuela	Sentinel hamster	SM1, V1
66457	ID	11 November 1981	Zulia State, Venezuela	Sentinel hamster	SM1, V1
83U434	ID	June 1983	Norte de Santander, Colombia	Sentinel hamster	CEC1, V1
ZPC738	ID	24 September 1997	Zulia State, Venezuela	Sentinel hamster	None
243937	IC	18 December 1992	Trujillo State, Venezuela	Horse	SM2
SH3	IC	9 January 1993	Trujillo State, Venezuela	Human	V1

^aNumber of passages in chicken embryo cells (CEC), a suckling mouse (SM), and Vero cells (V). 

TABLE 2.

Oligonucleotide primers used for genome amplification and expected sizes of PCR products

Nucleotide positions	Primer sequence (5′–3′)	Expected PCR product size (bp)
1–21	ATGGGCGGCGCATGAGAGAAG	1,713
1714–1696	GAAAGCACAGCGTAAGAGC
1526–1546	CGATGAGGCTAAGGAAGTGCG	2,513
4039–4019	CCCTCGCACCACATGATATG
3578–3598	CCCAGGCAAAATGGTTGACTG	1,673
5251–5228	AGGAATGGACCAGGATGAGCTAG
5065–5086	AGAACCAATCCACAGAGGGGAC	2,586
7651–7632	CATCGCCAGAAAAGGGTCG
7514–7533	ACYCTCTACGGCTRACCTRA	2,333
9847–9825	TGTTCCATAGGTGATCCAAGGAC
9747–9764	CGTGCCTAACTCCTTACC	1,708
Mlu-25V primer	TTACGAATTCACGCGTTTTTTTTT TTTTTTTTTTTTTTTTV

Phylogenetic analyses.

Nucleotide and deduced amino acid sequences were aligned by using the PILEUP program of the Genetics Computer Group (4). Phylogenetic analyses of the aligned sequences were performed with the PAUP maximum-parsimony program (30) to predict the ancestral sequences of viral progenitors. Confidence values were obtained by bootstrapping (6).

Plaque assays.

Plaque assays were performed on Vero cell monolayers as described previously (16). Each virus sample was diluted and added to six-well cell culture plates and incubated for 45 min at 34°C. Monolayers were then overlaid with 4 ml of Eagle’s minimal essential medium supplemented with 1% agar and 2% fetal bovine serum. After 46 to 48 h of incubation at 37°C, 2 ml of a second overlay containing 0.008% neutral red was added. Eighteen hours after the second overlay, the plates were examined, and the wells containing between 10 and 50 distinct plaques were selected for measuring plaque diameters. A total of 30 plaques was measured for each strain. For irregularly shaped plaques, two measurements were made at right angles to each other and averaged.

Nucleotide sequence accession numbers.

Nucleotide sequences were deposited in the GenBank library under accession no. AF004458, AF004459, AF004472, and AF100566 and updates U55360 and U55362.

RESULTS

Nucleotide and amino acid sequences of ID and IC VEE viruses.

The RNA genome of each of the VEE subtype ID and IC viruses we sequenced was 11,420 nucleotides in length, except for strain ZPC738, which had an adenosine insertion after nucleotide position 42 in the 5′ untranslated region, and strain 83U434, which had a 21-nucleotide insertion (seven codons) after position 5047 in its nsP3 gene (numberings are for strain 66637, GenBank accession no. AF004458). This insertion is also found in the Trinidad donkey (subtype IAB) (12) and P676 (subtype IC) genomes (13), as well as in the genome of strain 68U201 (subtype IE) (19), suggesting that an ancestor of the 66637, 66457, ZPC738, SH3, and 243937 strains underwent a 21-nucleotide deletion. A summary of the variable nucleotide positions and deduced amino acids is shown in Table 3. Comparisons revealed a total of 219 variable nucleotides, with 41 amino acid differences (Fig. 1). The nucleotide differences were scattered throughout the genomes of both ID and IC viruses. The greatest nucleotide sequence variability was found in the 6K (3.0%), E2 (2.7%), nsP3 (2.4%), E3 (2.3%), and nsP2 (2.2%) gene regions. The E3 and nsP3 genes also showed the greatest amounts of amino acid variation, at 3.4 and 2.0%, respectively. The ratio of transitions to transversions was 4.13:1, nearly identical to previous estimates of alphavirus substitutions (2, 36).

TABLE 3.

Nucleotides and amino acids differing among VEE viruses sequenced

Gene or genome region	Nucleotides differing (%)	Amino acids differing (%)
5′ untranslated	0/44 (0)
nsP1	19/1,605 (1.2)	6/535 (1.1)
nsP2	53/2,382 (2.2)	8/794 (1.0)
nsP3	39/1,647 (2.4)	11/549 (2.0)
nsP4	30/1,821 (1.6)	4/607 (0.7)
26S junction (untranslated)	0/38 (0)
Capsid	13/825 (1.6)	3/275 (1.1)
E3	4/177 (2.3)	2/59 (3.4)
E2	34/1,269 (2.7)	4/423 (1.0)
6K	5/168 (3.0)	1/56 (1.8)
E1	20/1,326 (1.5)	2/442 (0.4)
3′ untranslated	2/118 (1.7)
Total	219/11,420 (1.9)	41/3,740 (1.1)

FIG. 1.

Summary of variable amino acids among 1992–1993 epizootic Venezuelan IC and sympatric enzootic ID viruses. Dots indicate same amino acid as is present in strain 66637.

The nucleotide sequences of two of the enzootic subtype ID strains, 66637 and 66457, were nearly identical, with only 10 nucleotide differences and no amino acid differences. Strain ZPC738 had 108 nucleotide differences from these two strains, with an additional nucleotide insertion in the 5′ untranslated region. Twenty amino acid differences were found between ZPC738 and the two 1981 ID strains. Comparison of another ID strain, 83U434, with 66637 and 66457 revealed 116 nucleotide differences, including an in-frame 21-nucleotide insertion in nsP3, and a total of 27 amino acid differences (Fig. 1).

Analysis of the genomic sequences of the two subtype IC strains, 243937 and SH3, indicated that they differed by only 15 nucleotides and three amino acids: nsP3 position 463 (L [243937] versus H [SH3]), nsP4 position 2 (I [243937] versus T [SH3]), and E1 position 384 (K [243837] versus R [SH3]). Comparison of these IC sequences with the ID sequences revealed 75 to 140 nucleotide differences (0.7 to 1.2%) in pairwise comparisons and 15 to 26 amino acid differences (0.4 to 0.7%), with differences in all of the viral genes.

Relationships of ID and IC viruses.

To estimate the nucleotide and amino acid changes that accompanied the emergence of the 1992 epizootic IC phenotype, phylogenetic analyses of genomic nucleotide and complete polyprotein amino acid sequences were conducted. Five other genomic sequences published previously were included: the Trinidad donkey epizootic IAB strain isolated in 1943 and its TC-83 vaccine derivative (12), the epizootic IC strain P676 isolated in 1963 in Venezuela (13), the enzootic ID strain 3880 isolated in 1961 in Panama (13), and the enzootic IE strain 68U201 isolated in 1968 in Guatemala (19). Maximum-parsimony analyses revealed that the Venezuelan and Colombian ID viruses group closely with epizootic/epidemic IC and IAB viruses (Fig. 2), in agreement with the results of previous studies employing shorter sequences (21, 37). When the epidemiological phenotype of each strain (enzootic ID or IE versus epizootic IAB or IC) was used as a character and its change minimized in the internal tree branches, the independent evolution of three epizootic lineages from enzootic progenitors was predicted (Fig. 2). Examination for the amino acid changes predicted to have accompanied these three emergence events revealed 7 for the 1992 IC emergence, 23 for the subtype IAB emergence, and 24 for the 1963 subtype IC emergence (Fig. 2). The seven changes predicted to have accompanied the 1992 emergence included two amino acid changes in nsP1, two in nsP3, one in the capsid, and two in E2 (Table 4). None of these changes has been linked to laboratory attenuation of the TC-83 vaccine strain (12) or to other laboratory-generated, attenuated mutants (3). The changes in nsP1 included (i) a change from asparagine to aspartic acid at position 167, a region of the protein near the N terminus where mutations have been shown to affect the methyltransferase and guanyltransferase activities required for alphavirus RNA genome capping (17, 18, 26), and (ii) a change from leucine to phenylalanine at position 528, closer to a region known to affect minus-strand RNA synthesis (7, 25). The histidine-to-tyrosine change at position 212 of nsP3 lies within the N-terminal conserved domain believed to be required for viral RNA synthesis (7, 29), while the threonine-to-methionine change at position 360 is in the poorly conserved C-terminal region thought to be heavily phosphorylated in some alphaviruses (15, 20). The function of this region is unknown. The glycine-to-arginine change at capsid position 76 lies within the N-terminal domain that is highly positively charged, poorly conserved, and believed to interact electrostatically with the RNA genome (29). Both E2 envelope glycoprotein amino acid changes (Table 4) lie within or near an important span of amino acids at positions 182 through 207 shown previously to react with monoclonal antibodies that neutralize viral infectivity, block hemagglutination, and passively protect mice (8). Because the VEE virus sequences we determined were extremely closely related, the probability of actual, historical mutations being omitted in the terminal tree branches due to a failure to detect superimposed substitutions of the same nucleotide position is very low; both the Jukes-Cantor (10) and Kimura two-parameter (11) formulas for genetic distance correction estimated that less than 1% of substitutions were obscured in the direct sequence comparisons used for the maximum-parsimony analysis.

FIG. 2.

Maximum-parsimony phylogenetic tree, derived from complete nonstructural and structural polyprotein amino acid sequences, showing relationships among VEE virus strains. Branches are labeled to reflect predicted ancestral epidemiological phenotypes (enzootic or epizootic) obtained by minimizing changes in the tree. Numbers represent amino acid changes in terminal branches representing epizootic virus emergence. Diagonal lines show two pairs of amino acid changes shared by two epizootic emergence events. Bootstrap values were 100% for all nodes in the tree of identical topology obtained by nucleotide sequence analysis.

TABLE 4.

Nucleotide and amino acid changes associated with emergence of epizootic subtype IC VEE in 1992

Genome nucleotide no.	Gene and amino acid no.	Hypothetical ID progenitor	1992–1993 IC viruses	Amino acid change
161		T	C
464		G	A
543	nsP1, 167	A	G	N→D
1628	nsP1, 528	G	T	L→F
4665	nsP3, 212	C	T	H→Y
4667		C	T
5110	nsP3, 360	C	T	T→M
5768		C	T
6227		G	A
7217		C	A
7763	Capsid, 76	G	A	G→R
7810		T	C
8482		C	T
8680		T	A
9100		T	C
9116	E2, 193	G	A	G→R
9177	E2, 213	C	G	T→R
9259		T	C
9310		C	A
9872		T	C
9922		C	T
10216		C	T
10327		T	C
10438		G	A
11281		G	A

To identify evidence of genetic changes common to two or more of the three VEE virus emergence events, which might represent epizootic determinants under selection by equine or mosquito hosts resulting in convergent evolution, we compared the nucleotide and amino acid changes in the three branches representing the phenotypic transition from enzootic to epizootic (Fig. 2). The threonine-to-methionine change at nsP3 position 360 was predicted to have accompanied the independent evolution of both IC lineages, and the glutamic acid-to-valine change at nsP2 position 340 accompanied the emergences of both the 1963 IC genotype and the IAB viruses.

Plaque size phenotypes.

Previous studies by Martin et al. (16) revealed a strong correlation between plaque size on Vero cells and the epidemiological phenotype and equine virulence of VEE virus strains. All 87 epizootic viruses examined had mean plaque diameters of ca. 1 to 2 mm, while 61 enzootic strains had plaque diameters ranging from 2 to 4 mm. To determine if this phenotypic correlation is maintained for the more closely related enzootic and epizootic isolates that we have identified, we determined plaque sizes for three of the ID strains (66647, 66637, and ZPC738) and the two IC strains (SH3 and 243937) that we sequenced. The enzootic ID strains all grew to mean plaque diameters of about 4 mm, whereas the two epizootic IC strains produced plaques of about 2 mm (Table 5) (P < 0.01; Student’s t test). To determine if these enzootic and epizootic strains were phenotypically similar to others examined previously (16), we also determined sizes for the epizootic strains Trinidad donkey and P676, as well as the enzootic strains 3880 and Fe-37c (Everglades virus, subtype II). These strains showed approximately the same plaque sizes as reported previously (16), and no significant differences were observed when the less recently isolated epizootic strains Trinidad donkey and P676 were compared with SH3 and 243937 or when 3880 and Fe-37c were compared with strains 66457, 66637, and ZPC738 (Table 5) (P > 0.3; Student’s t test). These results support the previous conclusion of Martin et al. (16) that plaque size on Vero cells is a useful indicator of the epidemiological phenotype of VEE viruses. However, the epidemiological potential, rather than just the history of these viruses, needs to be confirmed before this marker can be fully evaluated.

TABLE 5.

Plaque sizes of epizootic and enzootic VEE virus isolates

Strain	Subtype	Epidemiological type	Mean plaque diam (mm) ± SD
Trinidad donkey	IAB	Epizootic	1.86 ± 0.93
243937	IC	Epizootic	1.59 ± 0.70
SH3	IC	Epizootic	1.79 ± 0.71
P676	IC	Epizootic	1.81 ± 0.84
66637	ID	Enzootic	4.06 ± 0.93
66457	ID	Enzootic	4.11 ± 0.95
ZPC738	ID	Enzootic	4.23 ± 0.74
3880	ID	Enzootic	4.67 ± 0.24
Fe-37c	II	Enzootic	3.50 ± 0.89

DISCUSSION

Although VEE virus has caused numerous equine epizootics and epidemics in the Americas in this century, the source of the epidemics and epizootics, variety IAB and IC viruses, and their mechanism of interepizootic persistence were unknown. Phylogenetic analyses indicated that epidemic/epizootic viruses probably arose several times from variety ID-like ancestors in northern South America (21, 37). The results presented here provide further evidence in support of this conclusion. Genomic sequences of enzootic subtype ID strains from western Venezuela and eastern Colombia showed a very close genetic relationship to epizootic subtype IC strains isolated from a 1992–1993 Venezuelan outbreak. The enzootic ID strains most closely related to epizootic IC viruses, based on partial PE2 sequences (21), were 66637 and 66457 from Zulia State near the Guajira Peninsula; these enzootic strains differed from the 1992–1993 IC isolates by a nucleotide divergence of only 1.4% in the PE2 region. In this paper, we report that the entire genomes of these viruses differ from that of the IC strains by only 0.8 to 0.9% at the nucleotide level. Strain ZPC738, isolated in 1997 from the Catatumbo region of southern Zulia State, is even more closely related to epizootic viruses and differs from these IC viruses by only 0.7%. The close proximity of the Catatumbo region to the epicenter of the 1992–1993 outbreak near Trujillo provides a stronger epidemiological link between enzootic virus circulation and epizootic emergence and suggests the possibility of future VEE virus emergences in western Venezuela.

The extremely close genetic relationship among these enzootic and epizootic VEE virus strains supports a hypothetical model for the genetic mechanisms of epizootic VEE virus emergence. The enzootic ID viruses from Venezuela are presumed to be equine avirulent, based on previous experimental studies with other subtype ID viruses (32). In addition, the natural history of VEE in western Venezuela suggests that these enzootic VEE viruses are not virulent for horses; unvaccinated equines reside directly adjacent to the forests where enzootic ID viruses circulate in the Catatumbo region, yet there is no history of equine encephalitis in this region. Natural vaccination by avirulent enzootic viruses is a hypothetical explanation for this observation. However, the avirulent nature of these Venezuelan enzootic ID viruses and lack of preexisting epizootic potential must ultimately be confirmed with experimental equine infections.

Our hypothetical model for VEE virus emergence involves the mutation of enzootic subtype ID viruses, resulting in enhanced equine viremia and disease, and selection of the virulent, epizootic phenotypes in equines and possibly epizootic mosquito vectors (34). High-titered equine viremia, a critical factor allowing efficient transmission by various epizootic mosquito vectors with only moderate oral susceptibility, is hypothetically caused by one or a few of the mutations we have predicted to have accompanied the 1992 emergence (Table 4). A likely candidate is the change at position 360 in nsP3 from threonine to methionine, a mutation predicted by phylogenetic analysis to also have accompanied the independent emergence of a different subtype IC lineage in the early 1960s (Fig. 2). Other likely candidates are the amino acid changes encoded by the nsP1, nsP3, capsid, and E2 genes (Table 4). The correlation between VEE virus serotype and the epizootic phenotype suggests that the E2 gene, the site of the major VEE virus antigenic determinants, including hemagglutination inhibition (8, 24), is also involved in epizootic emergence. However, the lack of any E2 mutations common to the three VEE virus emergence events delineated in our phylogenetic analysis (Fig. 2) indicates that different E2 mutations can probably generate the same IC serotype. The occurrence of two E2 mutations involving charge alterations on the surface of the E2 protein implies that these could affect the binding of the virus to cells and thus influence pathogenesis. These hypotheses need to be tested with congeneic mutants generated from infectious clones that we are now developing.

Elucidation of the minimum number of mutations required to generate the epizootic phenotype will be important in predicting the frequency at which future outbreaks will occur and the constraints on epizootic activity. If only a small number of mutations is required, the high mutation rate of RNA viruses like VEE virus would result in the frequent generation of viruses with epizootic potential. Assuming a mutation frequency of about 10⁻⁴ (estimated previously for another alphavirus, eastern equine encephalitis [35]), a VEE double mutant should occur in most naturally infected rodents, which develop viremia of up to 10⁸ PFU/ml after experimental infection with enzootic subtype ID viruses (39). Triple mutants would be expected to occur approximately once in every 1,000 to 10,000 infected hosts. Rates of epizootic mutant generation in enzootic Culex (Melanoconion) mosquito vectors should be lower because viral populations in mosquitoes rarely exceed about 10⁷ PFU (27). If only a relatively small number of mutations (i.e., two to three) is required for generation of the equine-virulent phenotype, ecological requirements for epizootic transmission may limit the frequency of outbreaks. Ecological requirements may include the availability of susceptible equines for local virus amplification or the transport of an infected equine harboring a mutant epizootic virus to another region with susceptible populations. The limited seasonal availability of large populations of mammalophilic mosquitoes may also limit the amplification of epizootic mutants.

A more complete understanding of VEE virus emergence and elucidation of the mutations needed to generate equine-virulent, epizootic virus strains from enzootic progenitors will require the use of infectious cDNA clones generated from the strains we have implicated in the 1992 emergence event, followed by mutagenesis and experimental equine infections for phenotypic characterization. These studies are now underway in our laboratory.