Vector competence of selected North American Culex and Coquillettidia mosquitoes for West Nile virus.

To control West Nile virus (WNV), it is necessary to know which mosquitoes are able to transmit this virus. Therefore, we evaluated the WNV vector potential of several North American mosquito species. Culex restuans and Cx. salinarius, two species from which WNV was isolated in New York in 2000, were efficient laboratory vectors. Cx. quinquefasciatus and Cx. nigripalpus from Florida were competent but only moderately efficient vectors. Coquillettidia perturbans was an inefficient laboratory vector. As WNV extends its range, exposure of additional mosquito species may alter its epidemiology.

are competent but only moderately efficient laboratory vectors (7). The vector competence of Ae. aegypti, Ae. albopictus, Ochlerotatus atropalpus, Oc. j. japonicus, Oc. sollicitans, and Oc. taeniorhynchus for WNV has since been evaluated (8,9). WNV was isolated from Cx. restuans and Cx. salinarius caught during the 2000 outbreak in New York (4); however, the ability of these species to transmit WNV by bite is unknown. Other viruses circulating in the eastern United States have a similar epidemiology (e.g., St. Louis encephalitis [SLE] and eastern equine encephalomyelitis [EEE] viruses): they are maintained in an enzootic cycle involving birds as amplifying hosts and ornithophilic mosquitoes as enzootic vectors. Based on their association with these other arboviruses, several mosquito species should be considered potential vectors of WNV, although it has not yet spread to areas where these mosquitoes are found.
To assist public health personnel in assessing the risk that a potential mosquito vector represents for transmission of WNV, we conducted laboratory studies to evaluate the vector competence of Cx. nigripalpus, Cx. quinquefasciatus, Cx. restuans, Cx. salinarius, and Coquillettidia perturbans.

Mosquitoes
We tested five mosquito species for susceptibility to WNV (Table 1). Cx. nigripalpus was tested because it is the primary vector of SLE virus in Florida (10,11). Cq. perturbans is a potential epizootic vector of EEE virus in the eastern United States (12). Cx. salinarius has been found naturally infected with WNV (4) and has been implicated as a potential epizootic vector of EEE virus (12). Cx. quinquefasciatus has been implicated as a potential enzootic and epizootic vector of SLE virus (13). Cx. restuans has been found naturally infected with WNV (4) and may play a secondary role in the transmission and maintenance of SLE virus (14).

Virus and Virus Assay
The WNV strain (Crow 397-99) used was isolated from a dead crow found in the Bronx, New York, during an epizootic in 1999 (7); it had been passaged once in Vero cell culture. Stocks of virus at a concentration of 10 4.2 PFU/mL were prepared in a standard diluent (10% heat-inactivated fetal bovine serum in Medium 199 with Earle's salts [GIBCO-BRL, Gaithersburg, MD] NaHCO 3 , and antibiotics). Viral stocks, triturated mosquito suspensions, and chicken blood samples were tested for infectious virus by plaque assay on Vero cells as described (15), except that the second overlay, containing neutral red stain, was added 2 days after the first overlay.

Vector Competence Studies
Mosquitoes were allowed to feed on 2-to 3-day-old leghorn chickens (Gallus gallus) that had been inoculated with approximately 10 3 PFU of WNV 1 to 2 days earlier. Immediately after the mosquitoes fed, blood was drawn from the jugular vein of each chicken (0.1 mL of blood into 0.9 mL of heparinized diluent), and the blood suspensions were frozen at -70°C until assayed for virus to determine viremias at the time of mosquito feeding. After feeding on viremic chickens, engorged mosquitoes were transferred to 3.8-L screen-topped cardboard cages and held at 26°C with a 16:8(L:D)-hour photoperiod. After an incubation period of 12 to 14 days, the mosquitoes were allowed to feed again on 1-to 2-day-old chickens, either individually or in small groups, to determine if they could transmit virus by bite. Immediately after the transmission attempt, the mosquitoes were killed by freezing, their feeding status was determined, and their legs and bodies were triturated separately in 1 mL of diluent.
Infection was determined by recovery of virus from the mosquito tissue suspension. If virus was recovered from its body but not its legs, the mosquito was considered to have a nondisseminated infection limited to its midgut. If virus was recovered from both the body and leg suspensions, the mosquito was considered to have a disseminated infection (16). We defined the infection and dissemination rates as the percentages of mosquitoes tested that contained virus in their body or legs, respectively. Chickens used in the transmission attempts were bled from the jugular vein 2 days after mosquito feeding, and the blood was handled as described above. Recovery of virus from this blood indicated transmission (9).
To examine viral transmission more efficiently, some of the unfed mosquitoes were inoculated intrathoracically (17) with 0.3 µL of a viral suspension containing 10 4.2 PFU of WNV/mL (10 0.7 PFU/mosquito), held 7 to 14 days, and allowed to feed on 1-to 2-day-old chickens. Mosquitoes and blood specimens from these chicks were processed as described for the orally exposed mosquitoes.
To estimate transmission rates by species, we determined the percentage of mosquitoes with disseminated infection (after either oral exposure or by intrathoracic inoculation) that transmitted virus by bite. We then multiplied that percentage times the percentage of mosquitoes that developed a disseminated infection after feeding on a host with a particular viremia. The result is the estimated transmission rate for those mosquitoes.

Statistical Analysis
Confidence intervals (95%) for infection and dissemination rates were calculated by SAS 8.0 (18). We used Fisher exact test to compare transmission rates among disseminated mosquitoes in each species. Significance was tested at a level of alpha = 0.05.

Results
All mosquito species examined in this study were susceptible to infection with WNV and developed disseminated infections ( Table 2). Infection rates were >84% in all the Culex species when the viral titer in the donor chicken was >10 6.3 PFU/mL of blood. In contrast, the infection rate was 18% in Cq. perturbans fed on a chicken with a similar level of viremia. For most mosquito species tested, dissemination rates were approximately one fourth the infection rates.
None of the Culex species tested differed significantly in the percentages of mosquitoes with disseminated infection that transmitted virus (Table 3). However, the percentage of Cq. perturbans with disseminated infection that transmitted WNV was significantly lower than that for Cx. nigripalpus and Cx. quinquefasciatus (Fisher exact test, p <0.01).
We used the percentage of mosquitoes with disseminated infection that transmitted virus from Table 3 and the dissemination rates at 14 days after the infectious blood meal from Table 2 to estimate the transmission rate for each species. Under laboratory conditions and at the highest viral dose tested, the Culex species tested were moderately efficient vectors (estimated transmission rates 10% to 55%). In contrast, Cq. perturbans was an inefficient vector (estimated transmission rate <2%) ( Table 2).

Conclusions
Previous laboratory studies indicate that a number of North American mosquito species could serve as vectors of WNV (7-9). Our study indicated that several additional Culex species and Cq. perturbans are potential vectors of WNV. The viremias used in our study, 10 5.5-7.5 PFU/mL of blood, are consistent with levels considered to be low to moderate viremias for hooded crows and house sparrows in Egypt (19) and experimentally infected North American house sparrows and other passerine birds (N. Komar, pers. The Culex species tested in this study were moderately efficient vectors, with estimated transmission rates from 10% to 55% after exposure to viremias >10 6.3 . For comparison, the estimated transmission rate for Cx. pipiens held under conditions similar to those of our study is 20% (9). Although the Culex species tested were readily susceptible to oral infection, most infections were limited to the midgut and did not disseminate to the hemocoel. This finding is similar to results reported for Cx. pipiens, in which 81% became infected but only 23% developed disseminated infection (9). Compared with the moderately efficient Culex mosquito vectors of WNV, selected container-breeding Aedes and Ochlerotatus species are highly efficient vectors, and selected floodwater Aedes and Ochlerotatus mosquitoes are inefficient laboratory vectors (7-9). The Cq. perturbans in our study fell into the inefficient vector category.
Cx. nigripalpus has not been found naturally infected with WNV. However, the distribution of WNV in the United States is just beginning to reach the southern half of North Carolina, the northernmost limit of these mosquitoes' geographic distribution. Cx. nigripalpus is likely to become involved in WNV transmission because it is a primary vector of SLE in Florida (10,11) and is a competent laboratory vector of WNV. Furthermore, Cx. nigripalpus is an opportunistic feeder (20,21) and shifts host selection based on the season, feeding on avian hosts in the winter and spring and on mammalian hosts in the summer and fall (22). These factors, coupled with the vector competence data, suggest that Cx. nigripalpus could serve as an epizootic as well as an enzootic vector for WNV.
Our study showed that Cx. quinquefasciatus can transmit WNV by bite. WNV has not been isolated from wildcaught Cx. quinquefasciatus. However, the current distribution of WNV is just beginning to overlap the geographic range of this species (generally the southern United States).

Cx. quinquefasciatus has been implicated (through virus isolation and abundance during outbreaks) in the rural transmission of SLE virus in the western United States (23) and in urban transmission of SLE virus in the southern United
States (24). In contrast to Cx. pipiens, which primarily feeds on birds, Cx. quinquefasciatus shows a preference for avian blood but will feed readily on mammals, including humans (25). The data from this study, the bionomics of Cx. quinquefasciatus, and the mosquitoes' association with an arbovirus with similar epidemiology to WNV, suggest that Cx. quinquefasciatus may play a role in WNV transmission if-or more likely when-the distribution of the mosquito and the virus overlap to a sufficient degree.
Cx. restuans, which has been found naturally infected with WNV (4), transmitted WNV by bite in our study. Similarly, this species has been implicated as a vector of SLE

Emerging Infectious Diseases
virus by virus isolations from field-collected specimens (26,27), and its role is supported by laboratory transmission studies (13). Cx. restuans breeds in ground pools or container habitats, is widespread in its distribution in the United States, and adults are active early (by mid-May) in the eastern United States (28). This early season abundance, along with coinciding isolations of SLE virus from this species in early summer, implies that it may be involved in the overwintering or amplification of SLE virus (26). The isolation of WNV from Cx. restuans in July in Connecticut (29), relatively early in the WNV transmission season, raises concern that the role of Cx. restuans in WNV transmission may be similar to the one suggested for SLE virus. Cx. restuans feeds primarily on avian hosts (30), but whether it feeds on humans remains unclear (31). Given the lack of firm data on host preference, the role of this species as an enzootic or epizootic vector of WNV is still uncertain.
Our study indicated that Cx. salinarius transmits WNV efficiently by bite. During 2000, evidence of WNV infection was reported in 35 pools of this species, second in number only to the number of positive pools (126) of Cx. pipiens (4). To date, no summary of the data (e.g., minimum infection rates) from the 2000 season has been published, so the relative importance of these isolates cannot be compared. In general, Cx. salinarius appears to be mammalophagic in studies of blood meals, but its host feeding pattern is thought to be opportunistic, depending on host availability, innate host preference, or combination of these factors (20,25,32,33). Given the number of WNV-positive pools, its vector competence for WNV, and its feeding behavior, Cx. salinarius may be an ideal bridge vector between the enzootic avian cycle of WN and mammalian hosts.
Cq. perturbans was the least efficient WNV vector of those we tested. Contributing heavily to this finding was the presence of a salivary gland barrier. Less than one fourth of Cq. perturbans with disseminated infection transmitted WNV by bite (Table 3). Furthermore, this is the only North American species tested so far that exhibits a substantial salivary gland barrier. Cq. perturbans is generally regarded as mammalophagic (30,34); however, there are reports of its feeding on wading birds and passerines (34)(35)(36) and of numerous EEE virus isolates from field-collected specimens (37)(38)(39)(40). Despite the low transmission rate, the role of Cq. perturbans as a potential epizootic vector of WNV should not be totally discounted.
Our study extended the list of potential North American mosquito vectors of WNV. None of the North American species tested in this study or others (7-9) was refractory to WNV. However, there is a wide range in vector competence in these species, ranging from nearly incompetent (e.g., Cq. perturbans) to highly efficient (e.g., Oc. j. japonicus). These data are similar to those for Old World mosquito vectors of WNV, in which all Aedes and Culex species tested were competent vectors (5,6). Vector competence studies indicate that North American mosquitoes fall into three general categories depending on genera and, in some instances, breeding habitat: highly efficient, container-breeding Aedes and Ochlerotatus species; moderately efficient, Culex species; and inefficient, floodwater-breeding Aedes and Ochlerotatus and Cq. perturbans.
As WNV extends its range southward and westward, additional mosquito species (e.g., Cx. nigripalpus, Cx. quin-quefasciatus, Cx. tarsalis, and Ae. albopictus) will have greater exposure to this virus. Involvement of some of the species, particularly container-breeding Aedes and Ochlerotatus, may alter the epidemiology of WNV and present additional control problems for mosquito abatement personnel. In addition, mosquitoes are more efficient vectors at warmer temperatures (41,42;Dohm,unpub. data), a factor that will further change the epidemiology of WNV as its range extends southward.