West Nile virus (WNV) is now endemic in California, with annual transmission documented by the statewide surveillance system. Although much is known about the horizontal avian-mosquito transmission cycle, less is known about vertical transmission under field conditions, which may supplement virus amplification during summer and provide a mechanism to infect overwintering female mosquitoes during fall. The current study identified clusters of WNV-infected mosquitoes in Sacramento and Yolo Counties, CA, during late summer 2011 and tested field-captured ovipositing female mosquitoes and their progeny for WNV RNA to estimate the frequency of vertical transmission. Space-time clustering of WNV-positive Culex pipiens complex pools was detected in the northern Elk Grove area of Sacramento County between July 18 and September 18, 2011 (5.22 km radius; p<0.001 and RR=7.80). Vertical transmission by WNV-infected females to egg rafts was 50% and to larvae was 40%. The estimated minimal filial infection rate from WNV-positive, ovipositing females was 2.0 infected females/1,000. The potential contribution of vertical transmission to WNV maintenance and amplification are discussed.
West Nile virus (WNV) is transmitted horizontally between ornithophilic Culex mosquitoes and passeriform birds, occasionally causing disease in “dead-end” hosts such as humans and horses (Komar 2003, Weaver and Reisen 2010). WNV invaded southern California in 2003 (Reisen et al. 2004) and by 2004 was detected throughout the state, including Sacramento and Yolo counties in northern California (Hom et al. 2005). The Sacramento-Yolo Mosquito and Vector Control District (SYMVCD) conducts mosquito-borne virus surveillance by monitoring mosquito abundance and WNV infection prevalence as well as human and equine WNV cases, sentinel chicken seroconversions, dead bird WNV testing results, and environmental conditions to direct physical, biological, and chemical vector control (SYMVCD 2009). Accurate identification of factors associated with high prevalence of WNV infection in mosquitoes is integral to planning and implementing effective mosquito control strategies (CDPH 2011).
Because vertically-infected mosquitoes presumably do not need to take a blood meal to acquire WNV, this transmission mechanism may contribute to local WNV maintenance by: 1) creating a population of infected overwintering mosquitoes that survive the winter months as non-blood-fed, inseminated females that could reinitiate enzootic WNV transmission in the spring or 2) supplementing virus amplification during summer when horizontal transmission is at its peak (Goddard et al. 2003). Virus has been detected in overwintering female Culex (Nasci et al. 2001, Bugbee and Forte 2004, Farajollahi et al. 2005, Andreadis et al. 2010) during investigations to understand the potential for overwintering mosquitoes to harbor WNV. However, there is a paucity of research regarding what role vertical transmission may play during the summer peak in horizontal WNV transmission when the number of infected females is greatest. If frequent at these times, vertical transmission could rapidly amplify infection rates in the mosquito population as the season progresses.
The objectives of the current study were (1) to identify clusters of WNV-infected mosquitoes in space and time in Sacramento and Yolo Counties, CA, using surveillance data from pools of adult female Cx. pipiens complex mosquitoes collected using gravid traps during 2011 and (2) to estimate the frequency of vertical transmission of WNV through testing the progeny of ovipositing Cx. pipiens complex females trapped within these clusters during the summer months of 2011.
MATERIALS AND METHODS
Space-time cluster analysis
Adult female Cx. pipiens complex mosquitoes were collected by SYMVCD personnel from 9 March to 2 November, 2011 in Sacramento and Yolo Counties, CA, using Reiter-Cummings gravid mosquito traps baited with hog and rabbit chow-yeast infusion media (Cummings 1992). Pools of ≤50 females were tested for WNV RNA by RT-PCR using previously described methods (Lanciotti et al. 2000).
A space-time scan statistic (Kulldorff 1997, Kulldorff and Information Management Services, Inc. 2009) was used to identify places and times with high WNV mosquito infection prevalence. Following SaTScan's procedures, WNV-positive pools were assigned as cases and WNV-negative pools as controls, with the latitude and longitude of trap locations as the spatial coordinates. A Bernoulli model was chosen due to the binary nature of the pool test results. To ensure that areas with the highest prevalence of infection were identified, maximum temporal and spatial cluster sizes were 50% of the population at risk and 50% of the study period, respectively; if these values had been set at more than 50%, the clusters would have been defined based on lower-than-expected prevalence outside of the cluster, rather than our intended definition of elevated prevalence inside the cluster (Kulldorff et al. 1998). Data were aggregated at a seven-day time step (Monday through Sunday) that corresponded to the frequency of mosquito trapping and avoided bias associated with the particular weekdays that SYMVCD sets mosquito traps. The maximum cluster radius was set at 11 km, which encompassed the typical scale of WNV foci, as indicated by past aerial adulticide treatments to suppress epidemic WNV transmission in Sacramento County (Carney et al. 2008), maximum flight distance of Cx. pipiens complex mosquitoes as found in mark-release studies in the Central Valley and southern California (Reisen et al. 1991, Reisen et al. 1992), and the approximate range of daily movement of avian species with high WNV host competence (Reisen et al. 2006b). The test statistic was computed using a maximum likelihood ratio function, and statistical significance was tested by 999 Monte Carlo replications, with significance of a cluster identified by a p-value < 0.05, rejecting the null hypothesis of random dispersion of WNV-positive mosquitoes. Secondary clusters were required to have no geographical overlap with the primary cluster; secondary clusters overlapping the most likely cluster would only marginally increase cluster size, and these are generally of less interest than the primary focus of virus activity (Kulldorff 1997). Mapping of significant clusters was performed using ArcMap version 10.0 software (Environmental Systems Research Institute, Redlands, CA). The relative risk (RR) was defined as the ratio of the risk within the cluster to the risk outside of the cluster, and risk was calculated as observed divided by expected number of WNV-positive mosquito pools (Kulldorff 2010). Mosquito infection prevalence per 1,000 mosquitoes tested was calculated as maximum likelihood estimates (MLEs) at one-week intervals for the spatial area defined by the cluster results (Biggerstaff 2009).
To maximize the probability of finding infected Cx. pipiens complex females to measure vertical transmission rates, gravid traps were placed near sites with recent WNV-positive mosquitoes detected by SYMVCD's surveillance program and delineated by the SaTScan analysis. A total of nine locations were sampled by one to three traps operated for a single night. Mosquitoes were anesthetized with CO2, identified, and gravid Cx. pipiens complex females transferred into 50 ml conical, screen-topped vials containing 5 ml of water for oviposition. Females were held at 26° C under 14:10 (L:D) h photoperiod and offered dried cranberries and water-moistened cotton balls as sugar and hydration sources.
Vials with egg rafts were monitored for the presence of first-instar larvae for 24–48 h. Following eclosion, the empty egg rafts from each vial were placed in 1 ml of virus diluent and frozen at –80° C for later testing. First instar larvae kept separate by family were collected using vacuum filtration, placed in 1 ml of mosquito diluent and frozen at –80° C for later virus testing. Adult females then were anesthetized using triethylamine and placed in 1 ml of mosquito diluent and also frozen at –80° C for WNV testing.
Female bodies were homogenized by mixer mill (MM300, Retsch; Haan, Germany), RNA extracted using an ABI MagMax system, and tested for WNV RNA using a singleplex real-time quantitative reverse transcription-polymerase chain reaction assay (qRT-PCR) with an ABI 7900HT Thermo Cycler (Applied Biosystems, Foster City, CA) and previously described probe/primers from the envelope gene (Lanciotti et al. 2000). Empty egg rafts and larvae from WNV-positive adult females also were tested for WNV by qRT-PCR as above. When possible, positive specimens were confirmed using primers from the nonstructural region (NS1 gene) (Shi et al. 2001). Analysis was done using data from the envelope gene primers/probe set. The vertical transmission rate, defined as the percentage of females transmitting virus to their progeny (regardless of the proportion of progeny infected) (Turell 1988, Rosen et al. 1989), was determined by dividing the total number of WNV-infected females that transmitted the virus to their eggs or progeny by the total number of WNV-infected females that laid egg rafts or that laid egg rafts that produced larvae, respectively.
Space-time cluster analysis
During 9 March – 2 November, 2011, 90 pools tested positive for WNV RNA of 1,495 pools tested from Cx. pipiens complex females collected by gravid traps from Sacramento and Yolo counties, yielding an overall maximum likelihood estimated infection prevalence of 9.1 females per 1,000 (95% CL = 7.4 – 11.1). A space-time cluster of WNV-positive Cx. pipiens complex mosquito pools was delineated in northern Elk Grove, Sacramento County, CA and surrounding areas between 18 July and 18 September, 2011 (p<0.001; Figure 1). The cluster radius was 5.22 km and consisted of 45 positive mosquito pools against an expected value of 10.22 pools, and the relative risk (RR) was 7.80. Weekly prevalence of WNV infection within the space-time cluster ranged from 15.7 to 79.5 per 1,000 Cx. pipiens complex mosquitoes tested (Figure 2). The actual prevalence of infection during the sampling period fell within these MLE infection limit estimates, with 11 positive of 304 individual females tested, giving a true prevalence of 36 per 1,000 gravid female Cx. pipiens complex mosquitoes tested.
Trapping of gravid females to measure vertical transmission rates occurred on 15 and 19 August, 14 September, and 3 October, 2011 (Figure 2). Of the nine trap sites, gravid female Cx. pipiens complex mosquitoes were captured at six sites during 18 of the 22 total trap nights. Three hundred and four (90.2%) of 337 field-collected mosquitoes collected were gravid female Cx. pipiens complex females, of which 297 (97.7%) laid eggs and were tested for WNV. West Nile virus RNA was detected in 11 (3.6%, n = 304) gravid females collected on 19 August and 14 September from four of the six sites where gravid Cx. pipiens complex mosquitoes were captured. Of these, seven were confirmed by probe/primers from the NS1 region or had Ct (cycle threshold) scores <30; of the remaining four, three with Ct scores >35 did not confirm and one was unavailable for testing. One of the infected adult females failed to lay eggs and therefore was not included in the vertical transmission rate calculations. Evidence of vertical transmission occurred in four females from two of the four trap sites where WNV-positive females were caught; the overall vertical transmission rate to eggs was 50% (five WNV-positive egg rafts from ten positive females) and to larvae was 40% (4 WNV-positive families of 1st instars). Interestingly, the five ovipositing females with negative egg rafts and larvae had Ct scores >30, indicating an infection of <100 plaque forming units (pfu) based on standard curves derived from WNV grown in Vero cell culture. Conversely, the five egg-laying females that passed virus vertically to either their eggs or larvae had Ct scores <20, indicating an infection of >106 pfu. Assuming an average number of 100 adult females/egg raft (Vinogradova 2000, Anderson and Main 2006) and an approximately equal distribution of vertical infection in male and female progeny (Baqar et al. 1993, Reisen et al. 2006a), the estimated minimal filial infection rate (MFIR), assuming a single infected larva per positive larval pool from all female progeny from the ten WNV-positive, egg-laying females was 2.0 infected females/1,000 females. If only the WNV-positive, egg-laying females with Ct scores <20 suggestive of disseminated infections were considered, the MFIR was 4.0 infected females/1,000 females.
Testing gravid female Cx. pipiens complex mosquitoes collected from a focus of virus activity maximized the probability of finding evidence of vertical transmission. Culex pipiens complex mosquitoes are competent WNV vectors in California (Goddard et al. 2002, Vaidyanathan and Scott 2007, Reisen et al. 2008), and gravid traps attract mosquitoes that are likely to have already taken and digested at least one blood meal and are seeking to oviposit, thereby increasing the chances of capturing WNV-infected mosquitoes (Cummings 1992, Moore et al. 1993, Reisen et al. 2009, Kwan et al. 2010). The weekly estimates of mosquito infection prevalence indicated that the time of the first two trapping events was during the second of two infection peaks, making these dates optimal for finding infected females. Interestingly, the females collected after the space-time cluster subsided were WNV RNA-negative, even though they likely were older. The space-time analysis also showed that the choice of trapping locations was consistent with the spatial clustering of high WNV activity. Of the six sites where gravid female Cx. pipiens complex mosquitoes were collected, four were located within the WNV mosquito infection cluster, and West Nile virus-positive mosquitoes were collected at all four of these sites. None of the mosquitoes caught at the two trap sites adjacent to the cluster tested WNV-positive during the vertical transmission study.
The overall vertical transmission rates to egg rafts (50%) and to larvae (40%) by field-collected WNV-infected gravid females are noteworthy because: 1. vertical transmission rates were essentially 100% by females with Ct values <20, and 2. both estimates are considerably greater than previously reported in studies that tested reared F1 adult progeny. It has been suggested that vertical transmission of flaviviruses occurs by infection not of the ovary itself but of the fully formed egg during oviposition (Rosen 1988), with WNV possibly existing on the exterior of the egg raft and infecting larvae as they eclose from the egg. Interestingly, all females that vertically transmitted to eggs or larvae had Ct values <20 suggestive of more intense infections than the females that had Ct values >30 and did not vertically transmit to either eggs or larvae. The number of days post-infection and the number of ovarian cycles post-infection have been associated with increases in the vertical transmission rate (Anderson et al. 2008, Anderson et al. 2012). In a previous experiment using per os infected colony female Cx. pipiens, the overall vertical transmission rate detected in reared F1 adult progeny was 4.7%. However, vertical transmission did not occur until after 13 days post-infection and the second blood meal; the vertical transmission rate by females taking more than one blood meal was 13.6% (Anderson et al. 2008). Because the current study used field-captured, naturally-infected gravid females of unknown age, it is possible the five egg-laying females that vertically-transmitted to either their eggs or larvae were reproductively older than those that failed to transmit and resulted in the high estimate of the vertical transmission rate. Regardless of the explanation for these differences, results from the current study support the conclusion that WNV vertical transmission occurs frequently in infected gravid female Cx. pipiens complex mosquitoes in Sacramento County during late summer and fall. While sites where vertical transmission occurred were within the WNV mosquito infection cluster, this does not necessarily mean the individuals within that cluster had an increased probability of vertically transmission; days post-infection and ovarian cycle number were unknown in this study, and these individual mosquito factors would be interesting to consider in any future, larger-scale investigation.
Third and 4th instar mosquito larvae have been tested in previous Flavivirus vertical transmission investigations (Rosen et al. 1978, Rosen et al. 1983, Hardy et al. 1984, Rosen et al. 1989) and WNV studies (Baqar et al. 1993, Turell et al. 2001), and the testing of 1st instar larvae for WNV using this method was validated experimentally before beginning this project. Other studies using larval testing have demonstrated WNV MFIRs between 0.62/1,000 and 4.8/1,000 in Culex species (Baqar et al. 1993, Turell et al. 2001); the MFIRs obtained in the current study of 2.0 infected females/1,000 females and 4.0 infected females/1,000 females (overall and from WNV-positive, egg laying females with low Ct scores suggestive of disseminated infections, respectively) are comparable to these as well as to two other studies that tested reared F1 progeny and demonstrated MFIRs of 3/1,000 and 2.8/1,000 in Cx. pipiens complex mosquitoes (Goddard et al. 2003, Anderson and Main 2006). One additional laboratory study of WNV vertical transmission in Cx. pipiens resulted in an MFIR of 0.52/1000 specimens (Anderson et al. 2008), and the unknown but theorized possible failure of transstadial transmission of WN and related viruses may account for this lower estimation (Rosen et al. 1989, Baqar et al. 1993, Turell et al. 2001). It is possible that not all larvae vertically infected will retain infection until adulthood, and the efficiency of transstadial transmission remains an important area for future research.
Although the vertical transmission rates to eggs and larvae at the time and place of peak horizontal transmission appear exceptionally high, the minimal filial infection rate may give a more realistic impression than the vertical transmission rate and this method's contribution to the virus cycle, because the MFIR quantifies the infection rate in the next generation progeny arising from infected females. Although the MFIR estimates were low, even very small vertical transmission and filial infection rates could enhance the overall field infection rates in very large mosquito populations (Dohm et al. 2002, Anderson et al. 2012) and contribute to epizootic summertime transmission (Anderson and Main 2006, Anderson et al. 2008, Anderson et al. 2012).
Perhaps the most important contribution of WNV vertical transmission is the potential to create an infected overwintering population that reinitiates enzootic transmission the following spring. Because the chance of a mosquito becoming infected increases as a female mosquito ages and the number of total blood meals increases, late-summer horizontal transmission combined with an aging mosquito population that has the potential to vertically transmit WNV may give rise to infected mosquitoes bound for diapause (Rosen et al. 1989, Goddard et al. 2003, Anderson and Main 2006, Anderson et al. 2008, Barker et al. 2010). Decreasing day length signals Cx. pipiens complex 4th instar mosquitoes to prepare for overwintering; females destined for overwintering do not take a blood meal, but rather utilize carbohydrate meals to form a fat body to survive winter (Moore et al. 1993, Vinogradova 2000, Komar 2003). Previously proposed gonotrophic dissociation (blood feeding without egg development) as a source of infected overwintering mosquitoes (Eldridge 1968, Eldridge and Bailey 1979, Bailey et al. 1982) has not been reported for Cx. pipiens complex mosquitoes in the wild (Mitchell 1981, 1983, Mitchell and Briegel 1989). These investigations combined with the lack of evidence for venereal WNV transmission (Reisen et al. 2006a) has led to the conclusion that vertical transmission is likely the only mechanism leading to WNV-infected overwintering adult mosquitoes. The MFIRs found in this study were consistent with previously reported WNV infection prevalence of overwintering adult mosquitoes, that ranged from 0.7/1000 to 2.7/1000 specimens tested (Nasci et al. 2001, Bugbee and Forte 2004, Farajollahi et al. 2005, Andreadis et al. 2010), Mosquitoes infected via vertical transmission are indeed able to initiate horizontal transmission; as proof of principle a vertically-infected female Cx. pipiens (descendant of a naturally-infected female) survived a simulated winter and transferred WNV to a hamster that succumbed to the disease eight days later (Anderson and Main 2006).
Strengths of our study include efficient sampling for vertical transmission within and during an identified space-time WNV-infected mosquito cluster. By purposefully choosing areas known to have high WNV activity, we documented that WNV vertical transmission does indeed occur under field conditions in Sacramento County and that these areas are correlated in space and time with known WNV prevalence. This study also tested individual adult mosquitoes, providing estimates of the true infection rate during peak transmission, which allowed a useful comparison to estimates of minimal infection rates from testing pools of mosquitoes. However, the sample being limited in size, location, and time may have precluded a general comparison to the maximum likelihood estimates over a broad range of values. Strategies to promote larger sample sizes for future investigations could include setting a greater number of traps at more sites near recent WNV-positive locations and/or performing a space-time analysis over multiple years to help identify areas that are repeated “hotspots” to be watched closely for heightened WNV activity. Other future research interests include WNV transstadial transmission rates and the interplay between vertical transmission and environmental conditions, landscape ecology, or urbanization that may affect enzootic WNV cycles in Sacramento County, CA.
We especially thank Dr. P. Macedo and the staff at the Sacramento-Yolo Mosquito and Vector Control District who provided the surveillance data used to identify the “hot spot.” This research was funded, in part, by Research Grant AI55607 from the National Institute of Allergy and Infectious Diseases, NIH.