Vector competence of Aedes aegypti, Culex tarsalis, and Culex quinquefasciatus from California for Zika virus

Zika virus (ZIKV) has emerged since 2013 as a significant global human health threat following outbreaks in the Pacific Islands and rapid spread throughout South and Central America. Severe congenital and neurological sequelae have been linked to ZIKV infections. Assessing the ability of common mosquito species to transmit ZIKV and characterizing variation in mosquito transmission of different ZIKV strains is important for estimating regional outbreak potential and for prioritizing local mosquito control strategies for Aedes and Culex species. In this study, we evaluated the laboratory vector competence of Aedes aegypti, Culex quinquefasciatus, and Culex tarsalis that originated in areas of California where ZIKV cases in travelers since 2015 were frequent. We compared infection, dissemination, and transmission rates by measuring ZIKV RNA levels in cohorts of mosquitoes that ingested blood meals from type I interferon-deficient mice infected with either a Puerto Rican ZIKV strain from 2015 (PR15), a Brazilian ZIKV strain from 2015 (BR15), or an ancestral Asian-lineage Malaysian ZIKV strain from 1966 (MA66). With PR15, Cx. quinquefasciatus was refractory to infection (0%, N = 42) and Cx. tarsalis was infected at 4% (N = 46). No ZIKV RNA was detected in saliva from either Culex species 14 or 21 days post feeding (dpf). In contrast, Ae. aegypti developed infection rates of 85% (PR15; N = 46), 90% (BR15; N = 20), and 81% (MA66; N = 85) 14 or 15 dpf. Although MA66-infected Ae. aegypti showed higher levels of ZIKV RNA in mosquito bodies and legs, transmission rates were not significantly different across virus strains (P = 0.13, Fisher’s exact test). To confirm infectivity and measure the transmitted ZIKV dose, we enumerated infectious ZIKV in Ae. aegypti saliva using Vero cell plaque assays. The expectorated plaque forming units PFU varied by viral strain: MA66-infected expectorated 13±4 PFU (mean±SE, N = 13) compared to 29±6 PFU for PR15-infected (N = 13) and 35±8 PFU for BR15-infected (N = 6; ANOVA, df = 2, F = 3.8, P = 0.035). These laboratory vector competence results support an emerging consensus that Cx. tarsalis and Cx. quinquefasciatus are not vectors of ZIKV. These results also indicate that Ae. aegypti from California are efficient laboratory vectors of ancestral and contemporary Asian lineage ZIKV.

Introduction Zika virus (ZIKV) is a mosquito-transmitted flavivirus that was first isolated in 1947 in the Zika forest of Uganda from a sentinel rhesus macaque [1]. Since its discovery, human ZIKV cases have been reported across Africa and Asia, but until 2007 the virus received little attention from researchers as it was thought to cause only mild disease. Following epidemics in Micronesia in 2007, French Polynesia in 2013, and Brazil in 2015 [2], ZIKV has now been confirmed as a cause of the neurological disease Guillain-Barre syndrome and congenital disorders, including microcephaly in infants [3]. Despite a dramatic decline in Brazilian cases since 2016, ZIKV remains a significant global human health threat [4], as other countries including Argentina, Bolivia, Peru, and Ecuador reported an increase in cases in 2017 [5].
California (CA) vector control districts have been combating stable Ae. aegypti populations in the state since 2013 [37], including in many counties in Southern CA. In addition, between 2015 and March 2018, 640 travel-associated ZIKV infections were reported in CA [38], 137 (21% of cases in state) of which were in Los Angeles County where the Ae. aegypti used for vector competence experiments here were collected. Due to the presence of Ae. aegypti and numerous travel-associated ZIKV infections, there is a risk of the establishment of local mosquito-borne ZIKV transmission. Additionally, genetic variation between Central Valley and Southern CA Ae. aegypti populations has been observed [39], even between populations in neighboring cities, such as Fresno and Clovis [40]. These findings indicate that gene flow is limited between Ae. aegypti populations and leave open the possibility that important traits, such as vector competence, may also vary among Ae. aegypti throughout the state. To better assess local ZIKV transmission risk, we evaluated the laboratory vector competence of Ae. aegypti from Los Angeles, CA, for ZIKV isolates from Puerto Rico (2015), Brazil (2015), and Malaysia (1966). We also evaluated the laboratory vector competence of two highly abundant Culex species, Cx. quinquefasciatus from Orange County, CA, and Cx. tarsalis from Kern County, CA, with a Puerto Rico (2015) ZIKV strain.

Sources of ZIKV strains, mosquitoes, and mice
Three Asian-lineage strains of ZIKV were used in our experiments. A 2015 Puerto Rican strain was isolated from human serum in 2015 (PR15, PRVABC59), passaged 4 times in Vero cells, and sequenced. The coding sequence for the complete genome of the passaged we used was identical to GenBank accession number KX601168. An Asian-lineage Malaysian ZIKV strain isolated from Ae. aegypti mosquitoes in 1966 (MA66, P6-740 [41]) that had been passaged in suckling mouse brains 6 times and once in Vero cells before it was received from the Centers for Disease Control was passaged once more in Vero cells. The complete coding genome sequence of our passage of MA66 was 100% identical to GenBank accession number KX601167. The Ae. aegypti mosquitoes used in this study were field-collected as larvae in Los Angeles, CA, in 2016 and morphologically identified. The F6 generation was used for this study. Adult Cx. quinquefasciatus mosquitoes were field-collected as adults in Orange County, California in 2016 and morphologically identified. The F5 generation was used for this study. The Cx. tarsalis mosquitoes were field-collected in the Kern National Wildlife Refuge, Kern County, CA in 2002, morphologically identified, and have been maintained continuously in colony since.

ZIKV vector competence experiments
Mice were inoculated with 5 log 10 Vero plaque forming units (PFU) of ZIKV via subcutaneous injection. ZIKV-infected mice were presented to mosquitoes 2 days post-inoculation, at peak viremia [33]. Mice were anesthetized prior to mosquito exposure with a ketamine (VETone Zetamine CIII, 75 mg/kg), xylazine (AnaSed, 10 mg/kg), and acepromazine (AceproJect, 1 mg/kg) solution administered intraperitoneally. The ZIKV viremia in each mouse was determined by Vero cell plaque assay from 30 μL of whole blood collected immediately prior to the mosquito feed. Viremic mice were presented for two cohorts of adult female mosquitoes 30-60 minutes on one of three arrangements depending on species: (1) 25 Cx. tarsalis in pint cartons (amazon.com), (2) 50 Ae. aegypti in pint cartons, or (3) >100 Cx. quinquefasciatus in a 1 ft 3 mesh cage (BugDorm, MegaView Science, Taiwan). Engorged females were sorted from non-fed individuals by vacuum aspiration. Mosquito ages at the time of blood-feeding were 4-14 days post eclosion (dpe) for Cx. tarsalis, 14-21 dpe for Cx. quinquefasciatus, and 4-12 dpe for Ae. aegypti. Cx. tarsalis and Ae. aegypti were held at 26˚C, 80% relative humidity, and 12:12 h light:dark cycle. Cx. quinquefasciatus were maintained at room temperature (22˚C and 33% relative humidity) to ensure survival. All mosquitoes had constant access to 10% sucrose before and after blood-feeding, except during a 24-hour starvation period prior to presentation of the viremic mice. At days 14 and 21 post bloodfeed, mosquitoes were cold-anesthetized at -20˚C for 5 minutes and then legs and wings were removed with forceps while immobilized on ice. Saliva was collected by inserting the proboscis into a capillary tube containing fetal bovine serum (FBS, GenClone) for 20 minutes. Individual bodies, legs+wings, and the saliva sample from each mosquito were stored separately in 2 mL tubes containing a 5 mm glass bead and 250 μL Dulbecco's modified eagle medium (DMEM, Gibco) supplemented with 50μg/mL of penicillin/streptomycin and 20% FBS. All samples were stored at -80˚C until further processing.

ZIKV RNA extraction
Mosquito tissues and glass capillary tubes containing saliva samples were homogenized in DMEM by shaking for 2 minutes at 30 shakes/second using a Tissuelyser (Qiagen, Hilden, Germany). Viral RNA was extracted using the MaxMax Viral RNA Extraction Kit (Thermo-Fisher, Waltham, MA). A total of 50 μL of homogenate for mosquito tissue and 100 μL of saliva samples were extracted. All RNA extracts were eluted in 50 μL of elution buffer (Buffer EB, Qiagen) and stored at -80˚C until further testing.

ZIKV RT-qPCR
ZIKV RNA titers were determined for each body, legs+wings, and saliva sample using the Taqman Fast Virus One-Step Master Mix (ThermoFisher) reverse transcription RT-qPCR kit with a previously described ZIKV-specific assay (primers: ZIKV 1086, ZIKV 1162c, and ZIKV 1107-FAM; [42]). At least two technical replicates were performed for all samples. Samples with a mean cycle threshold (Ct) value of 38 or below were considered positive for ZIKV RNA. This limit of detection was determined from prior testing of serially diluted samples of known ZIKV RNA concentrations with the same extraction and RT-qPCR reagents and protocols and equipment [43].

Infectivity of mosquito saliva
To estimate infectious ZIKV in expectorated Ae. aegypti saliva, viral titrations were performed on a random sample of RT-qPCR-positive saliva samples at the second or third thaw in Vero cell culture by plaque assay. In brief, cell monolayers were inoculated with 110 μL of undilute saliva from individual mosquitoes mixed with DMEM containing 2% (vol/vol) FBS, and 100 U/mL penicillin/streptomycin. After a one hour incubation period to allow for viral infection of cells, 0.8% agarose/DMEM was added to cover the cells. The plates were incubated at 37˚C in 5% CO 2 for 8 days. The cells were then fixed with 4% formaldehyde and stained with 0.05% crystal violet. Plaques were visualized as holes in the Vero cell monolayer and counted to determine PFU values. The limit of detection of the assay was 2.3 PFU where 110 μL of the total saliva sample (250 μL) was inoculated directly onto the cells. Since the volume of saliva was limited, each sample was tested in just 1 replicate.

Infection definitions, cohort grouping, and statistical analyses
In this study, we calculated infection rates as the number of RT-qPCR positive individual bodies divided by the number of individuals that ingested blood and were tested, dissemination rates as the number of RT-qPCR positive pooled leg & wing sets from each individual divided by the number of individuals that ingested blood and were tested, and transmission rates as the number of RT-qPCR positive saliva samples divided by the total number of individuals that ingested blood and were tested. For Ae. aegypti and Cx. tarsalis, multiple cohorts of the same species fed on different mice infected with the same ZIKV strain with slight ( 1 log 10 ) variations in viremias. Preliminary analysis across same-species cohorts that fed on different mice infected with the same ZIKV strain revealed no significant differences (Fisher's exact test, P>0.05) in infection, dissemination and transmission rates. We therefore combined the data presented for each ZIKV strain for Ae. aegypti and Cx. tarsalis, while also reporting the magnitudes of viremia in all mice ( Table 1). Comparisons of ZIKV RNA levels and PFU in saliva samples between ZIKV strains was performed using a one-way ANOVA with Tukey's correction for pairwise comparisons (reported as P adj ) and ZIKV RNA detection rates were compared using two-tailed Fisher's exact tests (scipy.stats). Data were plotted using matplotlib (Python).

Ethics statement
All procedures involving mice were performed in accordance with IACUC protocol #19404 that

Cx. tarsalis and Cx. quinquefasciatus from California were incapable of transmitting Puerto Rican ZIKV in laboratory vector competence experiments
Cx. tarsalis and Cx. quinquefasciatus mosquitoes were tested 14 or 21 days after ingesting blood from ZIKV-infected interferon receptor deficient mice. Two Cx. tarsalis bodies out of the 46 individuals tested (4%) had low levels of ZIKV RNA at 14 dpf (Ct < 38; 48 ZIKV genomes/body). Both infected individuals also had detectable ZIKV in their legs and wings, indicating disseminated infections. Neither of the ZIKV-infected Cx. tarsalis contained detectable ZIKV RNA in their saliva samples ( Table 1). The Cx. tarsalis infection rate significantly increased from 4% to 30% (2/46 to 6/20, P<0.01 Fisher exact test) from 14 to 21 dpf. Among the 6 infected Cx. tarsalis at 21 dpf, ZIKV RNA was detected in 1 leg and wing sample but, consistent with a lack of transmission 14 dpf, no ZIKV RNA was detected in the saliva (Table 1). We did not detect ZIKV RNA in any Cx. quinquefasciatus mosquito tissues 14 (N = 42) or 21 dpf (N = 37; Table 1).
To confirm infectivity and measure the transmitted ZIKV dose, plaque assays were performed on Ae. aegypti saliva collected 14 or 15 dpf to enumerate infectious ZIKV in Vero cell plaque forming units (PFU). Out of 45 RTq-PCR positive saliva samples that were tested by plaque assay, 32 (71%) yielded at least 1 detectable plaque. The expectorated PFU varied by  Fig 3).

Discussion
Understanding the mosquito species that vector ZIKV is important for estimating regional outbreak potential and for informing local mosquito control strategies, especially since Aedes and Culex species differ in life history traits and host-seeking behaviors that could impact control efforts. For example, oviposition traps bias towards Ae. aegypti that lay in artificial containers [44] while Culex typically prefer natural pools [45]. For Cx. tarsalis, we detected an overall ZIKV infection rate of 12% (8/66) in mosquitoes tested 14 and 21 dpf. Disseminated infections in Cx. tarsalis were detected at <5% on both 14 and 21 dpf, with high Ct values indicating low ZIKV RNA levels. We postulate that the disseminated infections detected in Cx. tarsalis may reflect false positives given that mosquitoes with true disseminated infections typically achieve very high viral RNA titers due to prolonged infection of multiple tissues. The absence of detectable ZIKV RNA in saliva at 14 or 21 dpf is evidence that Cx. tarsalis from CA is not capable of transmitting ZIKV in laboratory experiments. Furthermore, Cx. tarsalis feeds less often on human hosts compared to the highly anthropophilic Ae. aegypti [45][46][47], making humanmosquito-human ZIKV transmission by Cx. tarsalis unlikely. We also found no evidence for ZIKV infection of Cx. quinquefasciatus from California, with no ZIKV RNA detected in bodies, legs/wings or saliva from nearly 80 individuals. This is the first data showing ZIKV vector competence for California mosquitoes, and it supports results from many other studies which demonstrate that Cx. quinquefasciatus is not a competent laboratory vector of ZIKV. By contrast, Ae. aegypti mosquitoes exhibited infection rates of 85-90% and transmission rates of 53-80% at 14 dpf. The transmitted dose of infectious ZIKV by Californian Ae. aegypti is consistent with the range of doses observed in similar studies with Brazilian Ae. aegypti [48,49].  Ae. aegypti that ingested ZIKV MA66 in our laboratory vector competence studies developed higher ZIKV RNA levels than PR15-or BR15-infected mosquitoes. This pattern contrasted with the lower transmission rate and lower expectorated PFU of MA66-infected Ae. aegypti at 14 dpf.
A possible explanation for the lower transmissibility of MA66 at 14 dpf is that it lacks a A188V mutation in the NS1 gene that both PR15 and BR15 possess, which has been linked to higher infectivity (where infectivity can influence transmissibility) in mosquitoes when interferon-deficient mice are used for blood-feeding [7]. ZIKV strains from recent American outbreaks have also been shown to exhibit higher infection and transmission rates than historic Asian-lineage strains [8]. Additional vector competence studies involving region-specific Ae. aegypti and Ae. albopictus mosquito populations with sequenced genomes and multiple distinct ZIKV isolates will improve our understanding of the both mosquito and virus genetics involved in ZIKV vector competence, which could inform our ability to accurately estimate regional outbreak potential.
Among ZIKV MA66-infected Ae. aegypti, we observed that mosquitoes with low RNA copy numbers in bodies were less likely to transmit than those with infections that exceeded 6 log 10 genomes per body. This pattern is consistent with the presence of a midgut barrier to infection [50]. In that case, the mosquitoes with low body RNA levels represent infections that have not escaped the midgut while mosquitoes with high body RNA levels correspond to individuals with ZIKV that has disseminated to secondary amplification tissues.
This laboratory vector competence study confirmed that Ae. aegypti from Los Angeles, California, USA, can transmit Asian lineage ZIKV and that Cx. tarsalis and Cx. quinquefasciatus are inefficient ZIKV vectors. Given that Culex mosquitoes are poor ZIKV vectors and seek primarily non-human hosts, they are unlikely to facilitate a ZIKV outbreak. Thus, vector control efforts targeting ZIKV should remain focused on reducing urban Aedes populations.