Role of Anopheles Mosquitoes in Cache Valley Virus Lineage Displacement, New York, USA

Cache Valley virus (CVV) is a mosquitoborne virus that infects livestock and humans. We report results of surveillance for CVV in New York, USA, during 2000–2016; full-genome analysis of selected CVV isolates from sheep, horse, humans, and mosquitoes from New York and Canada; and phenotypic characterization of selected strains. We calculated infection rates by using the maximum-likelihood estimation method by year, region, month, and mosquito species. The highest maximum-likelihood estimations were for Anopheles spp. mosquitoes. Our phylogenetic analysis identified 2 lineages and found evidence of segment reassortment. Furthermore, our data suggest displacement of CVV lineage 1 by lineage 2 in New York and Canada. Finally, we showed increased vector competence of An. quadrimaculatus mosquitoes for lineage 2 strains of CVV compared with lineage 1 strains.

Cache Valley virus (CVV) is a mosquitoborne virus that infects livestock and humans. We report results of surveillance for CVV in New York, USA, during 2000-2016; full-genome analysis of selected CVV isolates from sheep, horse, humans, and mosquitoes from New York and Canada; and phenotypic characterization of selected strains. We calculated infection rates by using the maximum-likelihood estimation method by year, region, month, and mosquito species. The highest maximumlikelihood estimations were for Anopheles spp. mosquitoes. Our phylogenetic analysis identifi ed 2 lineages and found evidence of segment reassortment. Furthermore, our data suggest displacement of CVV lineage 1 by lineage 2 in New York and Canada. Finally, we showed increased vector competence of An. quadrimaculatus mosquitoes for lineage 2 strains of CVV compared with lineage 1 strains.

Virus Isolation
We processed mosquito pools as described (10,11). In brief, we homogenized pools in 1 mL of mosquito diluent containing 20% fetal bovine serum, 50 µg of streptomycin/mL, 50 U of penicillin, and 2.5 µg of amphotericin B/mL in phosphate-buffered saline in a Retsch Mixer Mill (https://www.retsch.com) set to 24 cycles/s for 2 min. We used viral stocks of 2 CVV strains isolated from cerebrospinal fluid of humans (strain Hu-2022) and from brain tissue (strain PA) (12,13) for RNA extraction. We used RNA extracted from brain tissues of a horse that died from neurologic disease and tested positive for CVV in this analysis. We homogenized placenta tissues from sheep (from a ewe that had given birth to a deformed lamb in a southern Ontario flock during 2011) and used them to infect Vero E6 cells for virus isolation. Cytopathic effect was observed after 6 days, and supernatant was harvested and used for RNA extraction and to generate virus stocks.

Primer Design and Reverse Transcription PCR
We used a standard PCR to identify CVV isolates as described (14). Beginning in 2012, we developed a real-time reverse transcription PCR (RT-PCR) by using new primers and probes (CVVF1, CVVR1, and CVV1 probe) to expedite the surveillance process (Table 1). A quantitative RT-PCR was developed according to manufacturer's protocol (Quanta Biosciences, https:// www.quantabio.com) with slight modifications.

Role of Anopheles Mosquitoes in Cache Valley Virus
The final volume of the reaction was 15 µL and consisted of 10 μL of master mixture and 5 μL of template. Each reaction contained 0.7 μmol/L of each forward and reverse primers and 0.3 μmol/L of probe. We performed real-time quantitation by using ABI Prism 7500 (Life Technologies, https://www.thermofisher.com). Cycling conditions were as follows: 3 min at 50°C, followed by 10 min at 95°C, then 40 cycles of alternating 95°C for 10 s and 60°C for 30 s. After introduction of CVV lineage 2, we developed new primers and probes (CVVF2, CVVR2, and CVV2 probe) for better detection (Table 1).

Maximum-Likelihood Estimation
We used maximum-likelihood estimation calculations to determine prevalence of CVV in mosquitoes. These calculations were based on a program developed by Brad Biggerstaff (https://www.cdc.gov/ westnile/resourcepages/mosqsurvsoft.html).

Sequencing
We chose representative CVV samples by county, year, and species and sent them to the National Microbiology Laboratory (Winnipeg, Manitoba, Canada) for full-genome sequencing. One PCR fragment was developed for the S segment, 3 for the M segment, and 5 for L segment (Table 2), and Sanger sequencing was performed by using Big-Dye version 3.1 on an ABI 3730X Analyzer L (both Thermofisher). Trace files were compiled by using SeqMan II (DNAStar, https://www.dnastar.com) to get consensus sequence for each segment. Alignments were generated by using ClustalW (https:// www.clustal.org) and MEGA4 software (15). Phylogenetic trees were generated by using the maximum-likelihood method in Geneious version 11.1.5 (https://www.geneious.com) and PhyML (http:// www.atgc-montpellier.fr) with the Jukes-Cantor substitution model. Robustness of the nodes was evaluated by performing 500 bootstrap replicates. Trees were rooted with the Fort Sherman virus S, M, and L segments (GenBank accession nos. KX100130, KX100131, and KX100132). Mean nucleotide distances between and within CVV lineages were calculated by using MEGAX software (https://berkstech.psu.edu).

Mosquito Vector Competence
A colony of unknown generations of An. quadrimaculatus mosquitoes (Orlando strain) was obtained from BEI   (16). Freshly propagated virus supernatant from infected Vero (African green monkey kidney) cultures were harvested at 48 h after infection (multiplicity of infection ≈1.0) and diluted 1:1 with defibrinated sheep blood and 2.5% sucrose mixture without freezing. In addition to undiluted supernatant, 10-fold dilutions from 1:10 to 1:10,000 were made in C6/36 maintenance medium (Eagle minimum essential medium containing 2% fetal bovine serum heatinactivated with 0.5 g/L of sodium bicarbonate plus 0.1 mmol/L nonessential amino acids plus 100 U/mL penicillin/streptomycin) before being mixed 1:1 with defibrinated sheep blood and a final concentration of 2.5% sucrose. Female mosquitoes (3-5 days old) were deprived of sugar for 1-2 hours and allowed to feed on CVV-defibrinated sheep blood-sucrose mixture for 30 min in a Hemotek membrane feeding system (Discovery Workshops, https://accrington.cylex-uk.co.uk) with a porcine sausage casing membrane at 37°C. After feeding, females were anesthetized with CO 2 and fully engorged mosquitoes were transferred to 0.6-liter cardboard containers and maintained with 10% sucrose at 27°C, 70% relative humidity¸ and a 12:12-h light: dark photoperiod. Infection, dissemination, and transmission assays were performed on days 6 and 15 after the infectious blood meal as described (17). On day 2 after feeding, because of the early time point, only infection and dissemination assays were performed. Dissemination rate is the proportion of mosquitoes with infected legs among infected mosquitoes; transmission rate is the proportion of mosquitoes with positive saliva among mosquitoes with disseminated infection. We compared infection, dissemination, and transmission rates among strains by using χ 2 analysis, followed by Bonferroni corrections for multiple comparisons in GraphPad Prism version 7.05 (GraphPad Software, https://www. graphpad.com). We used a TaqMan real-time reverse transcription to detect CVV by using primers and probe targeting both lineage 1 and 2 (Table 1).

CVV Surveillance
We sampled 1,842,352 female mosquitoes in 57,321 mosquito pools from 2000-2016, yielding a total of 255 CVV-positive pools. We compared MLE of prevalence by year ( Figure 2 In addition, we calculated prevalence for 10 mosquito species that had the highest number of CVV isolations. The 5 mosquito species with the highest MLE were An. punctipennis   where the rate was on average lower than those for the rest of the regions (0.04) (Figure 2, panel C).

Phylogenetic Analysis of CVV
We sequenced 48 CVV isolates representing various New York counties, hosts, and isolation dates and 3 isolates from Canada ( Table 3). Most of the CVV isolates were from mosquitoes, except 4 that were isolated from 2 humans, 1 sheep, and 1 horse. Phylogenetic analysis of CVV confirmed 2 distinct lineages (lineages 1 and 2) (Figure 3). Lineage 1 contained all CVV strains obtained during 2001-2010, and lineage 2 contained isolates obtained during 2011-2016. Segment reassortment between M and S was observed There was no evidence of spatial clustering of clades within the S, M, and L segments, except the reassortants, which all came from western New York regions (Cattaraugus, Chautauqua, and Allegany Counties). Mean genetic distance calculated as the number of nucleotide substitutions per site between lineage was 0.040 for the S segment, 0.074 for the M segment, and 0.051 for the L segment (Table 4). On average, there were more nucleotide substitutions for the M segment (0.074) than for the S (0.040) and L (0.051) segments.

Mosquito Vector Competence
We conducted vector competence assays with An. quadrimaculatus mosquitoes for 2 lineage 1 (NY10, NY25), 3 lineage 2 (15350152, 15330577, and Hu2011), and 4 reassortant (15041084, PA, 15041170, and 15060131) strains to determine whether there were differences between the lineages or between strains in the same lineage and to address effects of reassortment. We also hoped to determine whether vector competence was a potential mechanism of displacement of lineage 1 (Tables 5, 6). Our results indicate that lineage 1 strains are generally less infectious in An. quadrimaculatus mosquitoes because they had a 50% infectious dose ≈0.5-1.0 log 10 higher than that for lineage 2 (Table 5).
We also found decreased dissemination and transmission for lineage 1 strains of CVV compared with lineage 2 strains (p<0.05 by χ 2 test) (Tables 5, 6). We observed that CVV disseminated efficiently in An. quadrimaculatus mosquitoes by 2 days postfeeding. All mosquitoes infected with lineage 2 strains had disseminated virus, and dissemination of lineage 1 strains was more variable (Tables 5, 6). In addition, An. quadrimaculatus mosquitoes are a competent vector for the lineage 2 human strain but not for the human reassortant (PA) strain (lineage 1 L RNA segment and lineage 2 S and M RNA segments), which had a low dissemination rate. Except for the PA strain, An. quadrimaculatus mosquitoes were able to transmit CVV at day 6 postfeeding on an artificial blood meal with a high viral titer (6.0-7.0 log 10 PFU/mL). When mosquitoes were infected with a lower viral titer (4.0 log 10 PFU/mL), the infection rate decreased from 95%-100% to 12% for lineage 1, from 100% to 28%-64% for lineage 2, and from 85%-100% to 24%-52% for reassortants (Tables 5, 6).

Discussion
Consistent with the findings of Armstrong et al., who analyzed CVV strains from Connecticut (4), we identified substantial variability in CVV activity in New York during 2000-2016. In addition, in both states, CVV could be isolated from different mosquito genera, including Aedes, Anopheles, and Coquillettidia (6). In our study, the prevalence of CVV in An. punctipennis and An. quadrimaculatus mosquitoes during 2010-2016 (0.91) was higher than that during 2000-2009 (0.21). Although many mosquito species are apparently infected with CVV, our data and previous surveillance data for Connecticut (6) (18). However, only o'nyong nyong virus, which is closely related to chikungunya virus, is known to be consistently transmitted to vertebrates by Anopheles mosquitoes (19). Other studies supported potential roles of Anopheles mosquito species in the transmission of Rift Valley fever virus, Mayaro virus, Eastern equine encephalitis virus, and CVV (20)(21)(22)(23)(24). These data and our results confirmed that Anopheles mosquitoes have the potential to sustain transmission cycles of arboviruses. Additional studies are needed to elucidate their role in these cycles.
An. quadrimaculatus and An. punctipennis mosquitoes are mainly mammalian feeders in the northeastern United States, and white-tailed deer is the most commonly identified vertebrate host (25). Both mosquito species bite outdoors throughout the night and show higher activity at dusk and dawn and resting outdoors (26,27). In New York, white-tailed deer tested for CVV antibodies showed infection rates of 25.7% (28). White-tailed deer have been identified as the principal reservoir and amplification hosts for CVV, and their overabundance and availability for both Anopheles mosquitoes species that are frequently infected by the virus in nature (6,27-30) could partially explain the increase of CVV activity in Anopheles spp. observed in our study.
Early phylogenetic analysis of CVV strains from  A. Drebot, unpub. data). Furthermore, we demonstrated that An. quadrimaculatus mosquitoes are a competent vector for both CVV lineages and reassortants. The differential susceptibility between lineage 1 and lineage 2 suggest that An. quadrimaculatus mosquitoes might be actively involved in lineage 1 displacement in the northeast United States and can potentially increase the risk for spillover to humans in the region because lineage 2 is more infectious and more readily transmitted.
We isolated 4 reassortant strains that contained lineage 1 L segments and lineage 2 S and M RNA segments, and all came from counties in western New York. Reassortment is an evolutionary mechanism of segmented RNA viruses to exchange genetic information during co-infection of cells, which generates new genotypes and phenotypes (33,34). During reassortment, entire genes are exchanged among different viral strains or species by the swapping of segments, which confer major fitness advantages or disadvantages to the progeny virus (34). In the family Peribunyaviridae, reassortment events have occurred between virus lineages. Intraspecies, interlineage reassortment events were reported for Rift Valley fever virus, a phlebovirus and a mosquitoborne zoonotic virus that affects domestic animals and humans (35), and also for Crimean-Congo hemorrhagic fever virus (33,36,37), a highly infectious orthonairovirus transmitted by Hyalomma spp. ticks. Furthermore, interspecies reassortment also occurs. For example, reassortment among Bunymawera serogroup viruses has been documented with Ngari virus and Potosi virus (38)(39)(40)(41), among others. In addition, although segment reassortment among California serogroup viruses is infrequent (42), evidence of reassortment has been documented (43,44).
Earlier studies had demonstrated that genetic reassortment between members of the family Peribunyaviridae can occur in vitro in mosquito and mammal cells and in vivo in mosquitoes during a mixed infection and can produce viable new strains with major phenotypic changes in terms of infectivity and pathogenicity (38)(39)(40)42,(45)(46)(47). Furthermore, the phenomenon of superinfection resistance might promote opportunities for segment reassortment between more distantly related viruses. However, coinfection by closely related viruses can occur only in cases in which the second virus infects rapidly after the first virus and before superinfection resistance becomes effective (38). In our study, 3 CVV reassortants were isolated from mosquitoes and 1 was  (48,49). Among the reassortant strains tested in our study, only the human reassortant strain was not transmitted by An. quadrimaculatus mosquitoes despite persistent infection. This difference in phenotype was probably not caused by the viral titer in the infectious blood meal because the titer was only ≈0.5 log 10 lower for the human reassortant strain. We suspect that difference might be caused by other factors involving the virus strain and mosquito species used in our study. Addressing the potential mechanisms involved in differential vector competence phenotypes observed in An. quadrimaculatus mosquitoes and evaluating the role of strain variation in host competence and pathogenicity will help to clarify the consequences of genetic variation and displacement of CVV.