Effects of an Opal Termination Codon Preceding the nsP4 Gene Sequence in the O'Nyong-Nyong Virus Genome on Anopheles gambiae Infectivity

Abstract

The genomic RNA of an alphavirus encodes four different nonstructural proteins, nsP1, nsP2, nsP3, and nsP4. The polyprotein P123 is produced when translation terminates at an opal termination codon between nsP3 and nsP4. The polyprotein P1234 is produced when translational readthrough occurs or when the opal termination codon has been replaced by a sense codon in the alphavirus genome. Evolutionary pressures appear to have maintained genomic sequences encoding both a stop codon (opal) and an open reading frame (arginine) as a general feature of the O'nyong-nyong virus (ONNV) genome, indicating that both are required at some point. Alternate replication of ONNVs in both vertebrate and invertebrate hosts may determine predominance of a particular codon at this locus in the viral quasispecies. However, no systematic study has previously tested this hypothesis in whole animals. We report here the results of the first study to investigate in a natural mosquito host the functional significance of the opal stop codon in an alphavirus genome. We used a full-length cDNA clone of ONNV to construct a series of mutants in which the arginine between nsP3 and nsP4 was replaced with an opal, ochre, or amber stop codon. The presence of an opal stop codon upstream of nsP4 nearly doubled (75.5%) the infectivity of ONNV over that of virus possessing a codon for the amino acid arginine at the corresponding position (39.8%). Although the frequency with which the opal virus disseminated from the mosquito midgut did not differ significantly from that of the arginine virus on days 8 and 10, dissemination did began earlier in mosquitoes infected with the opal virus. Although a clear fitness advantage is provided to ONNV by the presence of an opal codon between nsP3 and nsP4 in Anopheles gambiae, sequence analysis of ONNV RNA extracted from mosquito bodies and heads indicated codon usage at this position corresponded with that of the virus administered in the blood meal. These results suggest that while selection of ONNV variants is occurring, de novo mutation at the position between nsP3 and nsP4 does not readily occur in the mosquito. Taken together, these results suggest that the primary fitness advantage provided to ONNV by the presence of an opal codon between nsP3 and nsP4 is related to mosquito infectivity.

O'nyong-nyong virus (ONNV) is an arthropod-borne virus (arbovirus) in the genus Alphavirus (family Togaviridae). Maintenance of arboviruses in nature requires a biological transmission cycle that involves alternating virus replication in a susceptible vertebrate host and a blood feeding arthropod. Although the vertebrate infection is acute and often associated with disease, continual transmission of virus is dependent on the establishment of a persistently infected state in the arthropod host. The general features of mosquito infection with arboviruses have been described (9). Susceptible mosquito vectors become infected with alphaviruses after ingestion of an infectious blood meal. The envelope glycoproteins present on the surface of the virion facilitate entry into the midgut epithelial cells, most likely through a specific receptor-ligand interaction (2, 9, 13, 25). Replication of virus in midgut epithelial cells may be followed by escape into the hemolymph. As virus enters the hemolymph, other tissues become accessible. Replication of virus in secondary tissues serves to further amplify virus in the hemolymph. If virus successfully infects and replicates in the salivary glands, virus can be transmitted through the saliva to a vertebrate host. Although alphaviruses are generally poorly vectored by anopheline mosquitoes, ONNV appears to represent an exception as Anopheles gambiae and A. funestus mosquitoes have been implicated as primary vectors during periods of epidemic transmission (4, 22, 36).

The ONNV genome is a positive-sense, nonsegmented, single-stranded RNA ∼11,800 nucleotides in length. The RNA has a 5′-terminal cap and a 3′-terminal poly(A) tract (reviewed in reference 31). The 5′ two-thirds of the genome encode the nonstructural or replicase proteins, while the structural genes are encoded in the 3′ one-third. Immediately after uncoating of the virion, four nonstructural proteins (nsP1 to -4) are translated from the genomic RNA (reviewed in reference 31). The polyprotein P123 results when translation is stopped at an opal termination codon (UGA) between nsP3 and nsP4. The polyprotein P1234 is produced when readthrough of the stop codon occurs (30). Readthrough occurs with 5 to 20% efficiency as determined by using in vitro translation reactions (5, 27). The structural genes are translated from a subgenomic mRNA (26S RNA) that is transcribed from an internal subgenomic promoter present in a full-length negative sense RNA copy. The negative-sense copy also functions as the template for production of new genomes (49S RNA).

Although all functions of the individual nonstructural proteins probably have yet to be determined, the topic has been well studied (14, 31). The nsP1 protein is required for initiating synthesis of viral RNA of negative polarity and also plays a role in capping genomic and subgenomic viral RNAs (1, 8, 35). The nsP2 protein possesses RNA helicase activity and also serves as the viral protease for the nonstructural polyprotein (6, 11, 23, 28, 33). The function of nsP3 is not completely understood, but it is essential to RNA replicase function and heavily phosphorylated (8, 19, 20, 34). The presence of a characteristic GDD motif has led to the assignment of nsP4 as the viral RNA polymerase, and genetic studies are consistent with this assignment (7, 15).

Sequence information initially indicated that the termination codon before nsP4 had been replaced with an arginine codon (CGA) in at least two alphaviruses, ONNV and Semliki Forest virus (SFV) (29, 32). For these viruses, it was presumed that only P1234 was generated during replication. However, passaging of ONNV prior to sequencing may have determined the codon at this particular locus (29). Consensus sequences obtained from later, lower passage, isolates indicate that ONNVs possess both sense and stop codons at this position (18). Further, when ONNV isolates found to have an opal termination codon are passaged in vertebrate cells (Vero), they rapidly acquire in its place an arginine codon (18). Taken together, these results suggest that ONNV quasispecies populations include both stop and arginine codons in equilibrium at this locus and that both sequences are somehow required for the survival of these viruses in nature. Replication of ONNVs in either vertebrate or arthropod hosts may determine the predominance of the codon at this particular locus in the quasispecies population. However, systematic studies testing this hypothesis have not yet been done in whole animals. We report here the results of the first study to investigate the significance of the opal stop codon to the alphavirus genome in a natural mosquito host. We used a full-length cDNA clone of ONNV to construct a series of mutants in which the arginine between nsP3 and nsP4 was replaced with an opal, ochre, or amber stop codon. These mutants were used to assess the effects of these changes on invertebrate infectivity in a natural mosquito vector of ONNV, A. gambiae.

MATERIALS AND METHODS

Site-directed mutagenesis and construction of infectious clone.

ONN mutant viruses were generated through site-directed mutagenesis of the full-length cDNA clone, pONN.AP3. Construction of pONN.AP3 has previously been described (3). Briefly, this clone was constructed by using cDNA generated from the SG650 ONNV genome (GenBank accession number AF079456). The SG650 strain was originally isolated from human serum during a 1996 outbreak in Uganda (18). The virus was passaged twice in African green monkey (Vero) cells prior to being deposited in the CDC arbovirus reference center in Fort Collins, CO. Virus rescued from pONN.AP3 contains a codon for the amino acid arginine (R; CGA) between nsP3 and nsP4. This codon corresponds to amino acid position 1897 in the translated SG650 genome sequence. This codon was changed to either an opal (Op; UGA), ochre (Oc; UAA), or amber (Am; UAG) stop using Stratagene's QuikChange site-directed mutagenesis kit (La Jolla, CA) according to the manufacturer's protocol. Briefly, paired oligonucleotides were used to introduce mutations in the double-stranded vector. Oligonucleotide sequences can be provided upon request.

Rescue of virus from infectious clones.

Infectious RNA transcripts were generated from plasmid templates linearized with NotI in a standard in vitro transcription reaction (12, 26). Baby hamster kidney (BHK-21) and A. albopictus mosquito (C6/36) cells were electrotransfected with RNA transcripts using a BTX ECM 630 ElectroCell Manipulator (BTX Instrument Division, Harvard Apparatus, Inc., Holliston, MA). Cells grown to 70 to 80% confluence were treated with trypsin, washed three times with Dulbecco phosphate-buffered saline, and used at a final concentration of 10⁷ cells/ml. Cells (4 × 10⁶) were mixed with RNA (10 μg) and pulsed twice. The settings were 460 V, 725 Ω, and 75 μF and 250 V, 25 Ω, and 325 μF for BHK-21 and C6/36 cells, respectively. Approximately 72 to 96 h after electroporation, 90% of the BHK-21 cells exhibited cytopathic effects (CPE). Tissue culture supernatant was harvested from BHK-21 and C6/36 cells at approximately 72 and 120 h after electroporation, respectively. The supernatant was clarified by centrifugation, the titer was determined, and the supernatant was stored at −70°C. The relative growth of the harvested viruses was compared by plaque assay analysis. Student t test was used to evaluate statistical differences between the quantities of virus rescued.

Infection of mosquitoes.

Three to five-day-old A. gambiae mosquitoes (G3 strain) were fed on a blood meal containing virus. Mosquitoes were reared in colonies under standard conditions (3). Immediately preceding preparation of the infectious blood meal, sheep blood (Colorado Serum Co., Boulder, CO) was washed with phosphate-buffered saline, and the erythrocytes were packed by centrifugation. Blood meals were prepared by mixing equal volumes of defibrinated sheep erythrocytes, a 10% solution of sucrose prepared in fetal bovine serum (FBS), and virus suspension. Blood meals were heated (37°C) and presented to mosquitoes using a Hemotek feeding apparatus (Discovery Workshops, Accrington, Lancashire, United Kingdom). Mosquitoes were allowed to feed for 1 h. Fully engorged females were separated and held in humidified chambers at 28°C until processing. Virus suspensions were diluted to equal titers in Dulbecco minimal essential medium containing 10% FBS plus nonessential amino acids, l-glutamine, and antibiotics prior to blood meal preparation. A portion of each infectious blood meal was frozen (−70°C), and back titers were determined later by plaque assay on African green monkey kidney (Vero) cells at 37°C. The back titers are reported in Tables 1 and 2.

TABLE 1.

Infectivity of recombinant ONNVs for A. gambiae^a

Virus	Rep.	Titer	Percentage of mosquitoes infected (n)
			Day 2	Day 4	Day 6	Day 8	Day 10
ONN.AP3 (R)	1	5.5	32.5 (40)	40.7 (59)	29 (38)	30.8 (39)	38.3 (60)
	2	5.5	20 (50)	31.4 (51)	39.2 (51)	35.3 (51)	41.2 (51)
ONN.AP3 (R→Op)	1	5.0	66.7 (3)	59.1 (44)	78 (18)	93.3 (15)	78 (50)
	2	5.0	54.2 (48)	82.4 (51)	78.4 (51)	84.3 (51)	73 (33)
pONN.AP3 (R→Am)	1	5.0	ND	ND	ND	ND	4 (50)

^aTiter = log₁₀ PFU/ml. n, Total number of mosquitoes ingesting blood meal assayed. ND, not determined due to an insufficient number of mosquitoes available. Rep., replicate.

TABLE 2.

Dissemination of recombinant ONNVs from the alimentary canal of A. gambiae

Virus	Rep.	Titer	Percentage of mosquitoes with disseminated infections (n)
			Day 2	Day 4	Day 6	Day 8	Day 10
ONN.AP3 (R)	1	5.5	0 (13)	0 (24)	0 (11)	8.3 (12)	8.7 (23)
	2	5.5	0 (10)	0 (16)	0 (20)	5.6 (18)	14.3 (21)
ONN.AP3 (R→Op)	1	5.0	0 (2)	11.5 (26)	7.1 (14)	14.3 (14)	15.4 (39)
	2	5.0	3.9 (26)	2.4 (42)	15 (40)	25.6 (43)	33.3 (24)
pONN.AP3 (R→Am)	1	5.0	ND	ND	ND	ND	50 (2)

^aTiter, log₁₀ PFU/ml. n, Total number of infected mosquitoes assayed. ND, not determined due to an insufficient number of mosquitoes available. Rep., replicate.

Processing and analysis of mosquitoes and tissues.

After predetermined holding periods, the mosquitoes were processed and analyzed for the presence of virus. Mosquitoes were cold anesthetized and decapitated. The heads and their corresponding bodies were then individually triturated in 300 μl of Dulbecco modified Eagle medium containing 10% FBS plus nonessential amino acids, l-glutamine, antibiotics, and antifungal agents. Supernatant was clarified by centrifugation and transferred to a clean tube. Clarified supernatant (100 μl) and BHK-21 cell suspension were added to individual wells of a 96-well plate and placed at 37°C. Cells were scored for the presence of CPE on days 2 to 5. The presence of CPE in a well containing supernatant from a body indicated infection. The presence of CPE in a well containing supernatant from the corresponding head indicated the infection had disseminated from the alimentary canal of the mosquito. The Fisher exact test was used for statistical analysis of differences between infection and dissemination rates in the cohorts under study.

Sequence analysis of viral RNA recovered from mosquitoes.

Viral RNA was isolated from triturated bodies and heads found to be positive in the CPE assays and then sequenced at the nsP3/nsP4 junction region. Viral RNA was isolated from supernatant by using the QIAamp viral RNA minikit as recommended by the manufacturer (QIAGEN, Valencia, CA). RNA was isolated from the supernatant of individually triturated bodies; however, the smaller amount of virus obtained from triturated heads required that positive supernatants be pooled during the extraction of RNA. Primers flanking codon 1897 were used in a reverse transcription-PCR to amplify a cDNA fragment from the viral RNA. The codon at position 1897 was then determined by sequencing the cDNA amplicon using primers internal to the amplification primers. Oligonucleotide sequences can be provided upon request.

RESULTS

Relative growth of ONNVs in vertebrate and invertebrate cell lines.

Growth patterns of all of the mutagenized ONNVs were compared to those of virus derived from the parental clone (arginine) in BHK-21 or C6/36 mosquito cells at 37 or 28°C, respectively (Fig. 1). Equivalent amounts of RNA were electrotransfected into cells. Culture medium was harvested from transfected cells at 72 (BHK) or 120 (C6/36) hours, and the titer of the virus present was determined by plaque assay on Vero cells at 37°C. In BHK-21 cells, the parental (arginine) virus and the amber termination mutant replicated to approximately 1.2 × 10⁷ PFU/ml. ONNV RNA possessing an opal termination codon produced almost 10-fold less virus, 1.7 × 10⁶ PFU/ml, in this cell type. The reduction in virus production compared to the arginine parental virus was significantly different (P < 0.001). Plaque assay repeatedly failed to detect any virus in the supernatant harvested from BHK-21 cells transfected with RNA of the ochre termination codon mutant. In C6/36 cells, the parental virus again replicated to the highest titer, 1.6 × 10⁷ PFU/ml. The ONNV amber termination mutant replicated to 2.7 × 10⁵ PFU/ml in this cell type. The production of ONNV by the opal termination codon variant was almost 10-fold lower, 4.2 × 10⁴ PFU/ml, than the amber termination mutant and nearly 500 times less than that of parental ONNV in C6/36 cells. Interestingly, the mosquito cells did yield a low level of virus (2.6 × 10³ PFU/ml) when transfected with the ochre termination codon mutant RNA. All three stop codon mutants produced significantly less virus in C6/36 cells compared to the arginine parent, with P values of <0.001 for all three comparisons. The opal and amber titers were also significantly different in the invertebrate cell type (P = 0.002), as were the opal and ochre titers (P = 0.002).

FIG. 1.

Relative growth of ONNVs in BHK-21 and C6/36 cells. Equivalent amounts of infectious RNA were electrotransfected into cells and harvested at equivalent times posttransfection. Titers obtained from the mutagenized ONNVs were compared to that of the parental virus possessing an arginine between nsP3 and nsP4.

Differences were also observed in the relative plaque sizes produced by various ONNVs (data not shown). The parental virus produced the largest plaque size when assayed on Vero cells. The ONNV opal variant and ochre mutant produced plaques of similar size, these also being the smallest. The ONNV amber mutant produced an intermediate plaque size. Plaque sizes were consistent between cell types; rescue of viruses from vertebrate or invertebrate cells made little difference.

Infectivity of ONNVs.

Specific infectivities of the ONNVs were examined by infecting A. gambiae mosquitoes with artificial blood meals containing either the respective individual viruses or a combination of viruses. Mosquitoes ingested artificial blood meals containing equivalent titers (∼5 log₁₀ PFU/ml) of ONN.AP3 (R), ONN.AP3 (R→Op), or ONN.AP3 (R→Am) virus. Low titers of virus obtained from the transfection of ONN.AP3 (R→Oc) RNA precluded the use of this mutant in these experiments. The results of these experiments indicated that the presence of an opal stop codon, upstream of nsP4, approximately doubled the infectivity of ONNV over that of virus possessing a codon for the amino acid arginine at the corresponding position. Using mean values, the time course experiment indicated that the opal virus infected 75.5% (n = 83) of the mosquitoes 10 days after ingestion of virus, whereas the arginine virus infected 39.8% (n = 111) of the mosquitoes at the same time point; statistical analysis indicated that these values were significantly different (P < 0.001). Transfected ONN.AP3 (R→Am) RNA yielded relatively high titers of virus in cell culture; however, this virus proved to be very inefficient at infecting A. gambiae mosquitoes when fed at a titer of 5 log₁₀ PFU/ml. The amber virus infected only 8.0% (n = 50) of mosquitoes at the 10-day time point. The results are summarized in Table 1. The advantage provided to ONNV by the opal stop was clearly not present when an amber stop was substituted. In CEF cells, SINV mutants containing an amber codon have previously been shown to underproduce the translational readthrough product nsP34 compared to SINV possessing an opal codon (21). The amber and ochre ONNV mutants were included to test the hypothesis that translational readthrough might occur less frequently when these viruses are used. It would appear that the infectivity of ONNV, for A. gambiae mosquitoes, is optimal under the levels of translational readthrough occurring when the leaky opal stop is present and cannot be improved by using alternative stop codons providing lower frequencies of readthrough.

The dissemination of viruses from the alimentary canal of the mosquito was also examined. Differences in the ability of these viruses to escape from the alimentary canal of this vector could translate to differences in the ability of these viruses to be transmitted to the vertebrate host. Mean values indicate the ONN.AP3 (R→Op) virus disseminated in 24.4% (n = 63) of infected mosquitoes at the 10-day time point, whereas the ONN.AP3 (R) virus disseminated in 11.5% (n = 44) of mosquitoes at the same time point (Table 2). However, these values were not significantly different (P = 0.2). Although dissemination of the opal virus was initially detected on day 2 and dissemination of the arginine virus was not detected until day 8, only at the day 6 time point did dissemination of the two viruses differ significantly (P = 0.045) (Table 2). Nevertheless, these results indicate the opal virus disseminates from the alimentary canal of the mosquito earlier than does the arginine virus. Two of the four mosquitoes that became infected with ONN.AP3 (R→Am) virus also had a disseminated infection (Table 2). Very little can be inferred from the results obtained using the amber mutant because of the small sample size; however, dissemination may have been due to a midgut leakage phenomenon (10).

To further assess the relative fitness levels of the ONN viruses, ONN.AP3 (R→Op) virus was mixed with ONN.AP3 (R) virus at a ratio of 1:9 and administered to A. gambiae mosquitoes in a blood meal containing approximately 5 log₁₀ PFU of total virus/ml, as determined from the back titer. This ratio was used because the opal was so much more effective than the arginine virus variant in individual infections. The infectivity of this mixture was assayed over time and compared to the results obtained using just the ONN.AP3 (R→Op) virus or the ONN.AP3 (R) virus individually. With 34% (n = 47) of mosquitoes becoming infected 4 days after the blood meal, the infectivity of this mixture resembled that of virus containing only the arginine codon at this time point. P values of the mixture, at day 4, versus the individual, mean, arginine or opal infections were 0.86 and <0.001, respectively. These results clearly demonstrate that the mixture follows an arginine-like infection rate early. However, by the 8-day time point, infectivity had increased to 68.3% (n = 41) and remained at this level at the 10-day time point (68.6%, n = 51). These results were not significantly different from those obtained with only ONN.AP3 (R→Op) virus at day 10 (P = 0.423) but were significantly different than results obtained with only the arginine virus at day 10 (P < 0.001). These results suggest that a significant fitness advantage is provided to ONNV by the alternative opal codon during replication in the alimentary canal of A. gambiae mosquitoes. Dissemination of virus from the alimentary canal occurred at a level comparable to that of ONN.AP3 (R) virus throughout the time course experiment: 0% at day 4 (n = 16), 7.1% at day 8 (n = 28), and 8.6% at day 10 (n = 35). Because there is significant mortality associated with holding infected or uninfected A. gambiae longer than a 10-day time course, it is possible that the level of dissemination observed may have increased beyond the period of time assessed here.

Sequence analysis of ONNVs.

Viral RNA was extracted from triturated mosquito bodies or heads and sequenced to determine whether codon usage upstream of nsP4 corresponded to that of the virus administered in the infectious blood meal. In the case of mixed infection, RNA was sequenced to identify codon usage in the predominant genomic virus sequence. A total of 46 samples were sequenced from mosquitoes infected with homogenous suspensions of either the arginine (n = 20) or opal (n = 26) virus. While it is likely that at least some of the viruses recovered from mosquito bodies and sequenced included viruses that had disseminated from the alimentary canal, one sequence for each respective virus was also derived from pooled infected mosquito heads (n = 8). Virus was sequenced directly from RNA extracted from mosquito homogenates and was not amplified in cell culture prior to the extraction. RT-PCR was used to amplify cDNA from viral RNA template recovered from the bodies or heads (disseminated virus) of mosquitoes previously found to be positive by observation of a CPE in BHK-21 cells.

For infections using homogeneous virus suspensions, changes in codon usage were not detected in any of the virus sequences examined at any time postinfection, nor was any mixed codon usage detected. These results indicate that, at least over a time course of 10 days, the pressure(s) responsible for the observed fitness advantage in A. gambiae do not result in de novo mutation at this particular locus of the ONNV genome. A total of 79 samples were sequenced from mosquitoes infected with mixed virus suspensions (arginine-opal; 9:1). Opal was the predominant codon detected in virus sequences recovered at 8 and 10 days postinfection. At 4 days postinfection, eight samples contained only arginine codons, three contained only opal codons, and five had a mixed population at the nsP3/nsP4 junction site. By 8 days postinfection, 17 opal codons were identified, whereas only seven arginine and four mixed populations were detected. On day 10, 16 samples contained only opal codons, 10 samples were arginine only, and 9 samples demonstrated evidence of mixed populations. These results, when combined with the data on infectivity, indicate that if both viruses are initially present in the blood meal, the selective pressures present are great enough for the opal virus to realize a fitness advantage.

DISCUSSION

The results reported here indicate that a locus near the nsP3/nsP4 junction has a major influence on ONNV infection of a natural mosquito host. A. gambiae mosquitoes were more frequently infected with the opal virus variant after ingestion in an infectious blood meal. ONNV containing the opal codon infected ca. 76% of the A. gambiae alimentary canals (Table 1). Approximately 40% of the A. gambiae mosquitoes ingesting the arginine virus developed alimentary canal infections (Table 1). The increase in the infectivity of the opal virus resulted from a single nucleotide substitution (CGA→UGA) in the nonstructural genes of otherwise isogenic genomes. Thus, contributions from the virus envelope glycoproteins on the observed differences can be effectively ruled out. This strongly suggests that effects on infectivity are related to the replication of virus. Although the scope of the present study limits conclusions regarding specific mechanisms, these results are consistent with several hypotheses proposing a regulatory role for the opal codon in alphavirus replication (17, 21, 31). Future studies will focus on the identification of a specific mechanism for the observed differences in infectivity.

Differences in infectivity between viruses containing a stop (opal) or the amino acid arginine suggest this particular locus influences the establishment of persistent infection in the mosquito host by regulating ONNV replication. In both C6/36 and BHK-21 cells the opal stop codon mutant produced significantly less virus than did the arginine parent. These results suggest that ONNVs possessing opal codons may more frequently establish persistent infections in A. gambiae by limiting replication in this host. Host responses may be involved. Although speculative, observations reported here appear to show some similarity to previously reported observations of RNA interference (RNAi) acting as an antagonist to ONNV replication in A. gambiae (16). RNAi is an evolutionarily conserved pathway by which double-stranded RNA (dsRNA) triggers the silencing of homologous mRNA sequences. Keene et al. demonstrated that A. gambiae were more readily infected with ONNV when the mosquito's Ago 2 protein, a core component of the RNAi machinery, was depleted. Future studies will investigate a potential relationship between the increased infectivity of the ONNV opal stop codon mutant and the mosquito's RNAi pathway.

We have previously reported a low frequency of dissemination from A. gambiae gut tissues using the SG650 strain of ONNV (3). The presence of the opal codon in ONNV does not appear to significantly alter those results. However, high levels of dissemination are only one component of vectorial capacity. Our results indicate that the opal virus disseminates from mosquito gut tissues more rapidly than does the arginine virus. Since naturally occurring ratios of virus variants present in quasispecies populations infecting mosquito vectors are unknown, this locus may affect the vectorial capacity by altering the speed of virus dissemination. The fitness advantage associated with the opal codon also may extend to tissues other than the gut, for example, the salivary glands or tissues involved in amplifying virus after dissemination from the alimentary canal. Differences in the infectivity of viruses containing an opal or arginine for these tissues could result in differences in the ability of these viruses to be transmitted to the vertebrate host. Nonetheless, Miller et al. have demonstrated that even incompetent vectors can initiate and maintain arboviral transmission if present in large enough numbers (24). The dramatic increases in infectivity associated with the opal codon at this locus could contribute to this type of epidemic transmission.

Sequence analysis of ONNV RNA extracted from mosquito bodies and heads indicated that codon usage at this position corresponded to that of the virus administered in the blood meal. These results suggest that while selection on ONNV variants is occurring, it is not significant enough to bring about a change in codon usage at the position between nsP3 and nsP4. Thus, the primary mutation pressures contributing to the evolution of this mechanism may be present in the vertebrate host. Nevertheless, evolutionary forces appear to have maintained in nature alphavirus genomes possessing both sense and opal codons at this particular locus. This suggests that both codons are associated with a fitness advantage but at different times in the transmission cycle. Our results demonstrate a clear fitness advantage in the mosquito host when an opal stop codon is present at this locus in the ONNV genome. Our results also suggest that this fitness advantage is primarily related to an effect on mosquito infectivity. Further understanding the molecular basis of the fitness advantage may provide valuable insights into understanding how alphaviruses establish persistent infections in the mosquito host.