West Nile Virus Genetic Diversity is Maintained during
Transmission by Culex pipiens quinquefasciatus
Doug E. Brackney, Kendra N. Pesko, Ivy K. Brown, Eleanor R. Deardorff, Jon Kawatachi, Gregory D. Ebel*
Department of Pathology, University of New Mexico School of Medicine, Albuquerque, New Mexico, United States of America
Due to error-prone replication, RNA viruses exist within hosts as a heterogeneous population of non-identical, but related
viral variants. These populations may undergo bottlenecks during transmission that stochastically reduce variability leading
to fitness declines. Such bottlenecks have been documented for several single-host RNA viruses, but their role in the
population biology of obligate two-host viruses such as arthropod-borne viruses (arboviruses) in vivo is unclear, but of
central importance in understanding arbovirus persistence and emergence. Therefore, we tracked the composition of West
Nile virus (WNV; Flaviviridae, Flavivirus) populations during infection of the vector mosquito, Culex pipiens quinquefasciatus
to determine whether WNV populations undergo bottlenecks during transmission by this host. Quantitative, qualitative and
phylogenetic analyses of WNV sequences in mosquito midguts, hemolymph and saliva failed to document reductions in
genetic diversity during mosquito infection. Further, migration analysis of individual viral variants revealed that while there
was some evidence of compartmentalization, anatomical barriers do not impose genetic bottlenecks on WNV populations.
Together, these data suggest that the complexity of WNV populations are not significantly diminished during the extrinsic
incubation period of mosquitoes.
Citation: Brackney DE, Pesko KN, Brown IK, Deardorff ER, Kawatachi J, et al. (2011) West Nile Virus Genetic Diversity is Maintained during Transmission by Culex
pipiens quinquefasciatus Mosquitoes. PLoS ONE 6(9): e24466. doi:10.1371/journal.pone.0024466
Editor: Patricia V. Aguilar, University of Texas Medical Branch, United States of America
Received May 13, 2011; Accepted August 10, 2011; Published September 12, 2011
Copyright: ? 2011 Brackney et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by the National Institute of Allergy and Infectious Disease, National Institutes of Health (NIH) under grant AI067380. DEB was
supported by 5F32AI084432-02. ERD was supported by Institutional Research and Academic Career Development Awards via Award K12GM088021 from the
National Institute of General Medical Sciences and the University of New Mexico Academic Science Education and Research Training fellowship. KP was supported
by National Research Service Award Institutional Training Grant 5T32-AI07538-13. IKB and JK were supported by the University of New Mexico Initiative to
Maximize Student Diversity, which is funded by the National Institute of General Medical Sciences, NIH under grant GM060201. The funders had no role in the
study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: email@example.com
West Nile virus (WNV; Flaviviridae, Flavivirus) was introduced into
North America in 1999 and has since spread across the continental
United States and into Canada, Mexico, the Carribean, and South
America . Molecular epidemiologic studies of WNV in the US
revealed that minor changes at the genomic level were associated
with a dramatic shift in the genotypic composition of WNV
circulating in North America [2–6]. Specifically, the introduced
genotype, termed NY99, was displaced by a new variant, WN02.
The WN02 genotype differs from NY99 by only a few nucleotide
and/or amino acid changes, but is more efficiently transmitted by
native Culex mosquitoes [5,7,8]. It was determined that the WN02
genotype requires a shorter extrinsic incubation period in
mosquitoes (EIP, time from vectorinfection to transmission) thereby
resulting in an increased vectorial capacity of local mosquitoes.
Similarly, the emergence of Chikungunya virus (CHIKV; Togavir-
idae, Alphavirus) seems to have been facilitated by analogous
mutations that result in increased transmission efficiency by the
vector Aedes albopictus [9,10]. Thus, relatively minor consensus
genetic changes can significantly influence arbovirus transmission
patterns and disease emergence. Determining the mechanistic
underpinnings of genetic change in arboviruses is therefore critical
to understanding their persistence and emergence.
RNA viruses exist within hosts as a dynamic distribution of non-
identical, but related viral variants [11–14]. High genetic diversity
profoundly influences the population biology of RNA viruses,
including WNV, polio, mumps and hepatitis C viruses [15–18]. In
the case of WNV, high genetic diversity is associated with
increased fitness in mosquitoes . Population bottlenecks may
reduce fitness by stochastically reducing the genetic diversity of the
virus population. In vitro studies of vesicular stomatitis virus, an
RNA virus, have demonstrated that repeated bottlenecks can lead
to fitness loss through the action of Muller’s ratchet . The
extent to which mosquitoes impose such population bottlenecks on
arthropod-borne viruses (arboviruses) is unclear. Analysis of WNV
populations from naturally infected birds revealed that non-
consensus, minority genotypes were shared among samples
collected from multiple birds, suggesting that WNV populations
may not be subject to bottlenecks during the natural transmission
cycle . Similarly, it was suggested that dengue virus type 1
(DENV1; Flaviviridae, Flavivirus) is not subject to widespread
population bottlenecks during the natural transmission cycle
because putatively defective genomes persist through complemen-
tation, requiring frequent coinfection of cells in both mosquitoes
and humans . Similarly, a high frequency of coinfection of
midgut cells has been reported for Venezuelan equine encephalitis
virus (VEEV; Togaviridae, Alphavirus) in Aedes taeniorhynchus .
PLoS ONE | www.plosone.org1September 2011 | Volume 6 | Issue 9 | e24466
Conversely, studies examining early infection of mosquitoes by
WNV and VEEV demonstrated that only a few (,15) midgut cells
are susceptible to arbovirus infection [22,23]. These findings
suggest that anatomical barriers, specifically cells of the midgut,
may act as genetic bottlenecks by restricting the population of
infecting virions thereby diminishing the genetic diversity of the
population. Importantly, these observations are not mutually
exclusive as several viral genomes may coinfect a single midgut
cell. Importantly, population bottlenecks associated with mosquito
transmission have not been assessed from a virus genetics
Therefore, we determined whether WNV experiences genetic
bottlenecks during the EIP in the vector mosquito Culex pipiens
quinquefasciatus. We hypothesized that WNV experiences genetic
bottlenecks during the EIP in mosquitoes, and reasoned that
sequential reductions in viral genetic diversity would occur as
infection progressed throughout the mosquito. To assess this,
WNV genetic diversity was quantified in mosquito midguts,
hemolymph, and salivary secretions, compartments that represent
three well-characterized infection stages (midgut colonization,
dissemination, and transmission). Three mosquitoes at three time
points (7, 14, and 21 days post infection (dpi)) were sampled. Our
data suggest that stochastic reduction of genetic diversity in
mosquitoes is at most a minor component of WNV population
biology during horizontal transmission.
Mosquito Infection Rates
Mosquito tissues were screened for the presence of WNV RNA
by one-step RT-PCR. All freshly fed mosquitoes were positive for
WNV RNA representing ‘input’, blood-meal associated virus. The
infection rates for midguts at 7, 14, and 21 days post infection
(dpi), reflecting viral populations able to overcome the midgut
infection barrier, were 88% (21/24), 86% (19/22), and 70% (14/
20), respectively. The percentage of mosquitoes positive for WNV
RNA in the legs, indicating virus dissemination from the midgut
and into surrounding hemolymph, was 58% (14/24), 36% (8/22),
and 55% (11/20) at 7, 14, and 21 dpi, respectively. In order for
mosquitoes to transmit WNV, the virus must be able to overcome
the salivary gland infection and escape barriers. The percentage of
mosquitoes with WNV in salivary secretions was 25% (6/24), 14%
(3/22), and 35% (7/20) at 7, 14, and 21 dpi, respectively. Three
mosquitoes per time point with WNV RNA in midgut,
hemolymph and saliva were selected for further analysis and
WNV genome equivalents quantified (Text S1). Genome equiv-
alents were highest in midguts and progressively decreased in the
hemolymph and saliva. Further, genome equivalents increased
with time post infection (21 dpi.14 dpi.7 dpi) (Figure S1). In
addition, three mosquitoes, representing the ‘input’ group, were
collected immediately post-engorgment. WNV genome equiva-
lents determined for each of these individuals were 3.2, 4.1 and
5.96105genome equivalents/ml (Figure S1). The bloodmeal
contained 66106pfu/ml, assuming 10–100 genomes per infec-
tious particle and an engorgment volume of ,3 ml, engorged
mosquitoes would be expected to contain ,1.46105 or 6genome
equivalents. The concentrations for the three individuals in the
‘input’ group are in agreement with these calculations and thus
faithfully represent the population of the bloodmeal as a whole.
Some arboviruses may enter mosquito hemolymph directly,
bypassing midgut infection via a ‘leaky midgut’ [24,25]. In order
to determine whether this occurred in the WNV-Cx. quinquefasciatus
system, hemolymph was removed from mosquitoes at 1, 3, 24, and
48 hpi as well as 8 and 16 dpi and tested for WNV by plaque assay
(Text S1). Hemolymph collected at 8 and 16 dpi commonly held
high titers of WNV. In contrast, hemolymph collected at early
timepoints after feeding almost never contained infectious WNV
WNV Genetic Diversity
The percent nucleotide diversity and proportion of unique viral
variants were used as indicators of viral genetic diversity in each of
the samples. The percent nucleotide diversity was determined by
calculating the total number of nucleotide changes for all clones
within a given sample divided by the total number of nucleotides
sequenced per sample. The data was grouped either by days post
infection (Figure 1 A, B, and C) or by tissue type (Figure 1 D, E,
and F). Analysis of the data set by days post infection revealed that
there was no significant difference in the percent nucleotide
diversity among the viral populations sequenced at 7 and 14 dpi
between ‘input’, midgut, legs or saliva (p=0.2739 and p=0.2662,
respectively) (Figures 1 A & B). Interestingly, genetic diversity
seemed to decrease with time post infection as there was a
significant reduction in diversity from the ‘input’ to the three tissue
types at 21 dpi (ANOVA p=0.0015; Tukey’s HSD post test,
‘input’ vs midgut q=7.262 p,0.05, ‘input’ vs legs q=8.493
p,0.05, and ‘input’ vs saliva q=5.293 p,0.05), but no difference
between tissue types (Figure 1C). Analyzing the data by tissue type
revealed that there was a significant reduction in diversity from the
‘input’ to midguts at 14 and 21 dpi (ANOVA p=0.0125, Tukey’s
HSD post test, ‘input’ vs 14 dpi q=5.404 p,0.05 and ‘input’ vs
21 dpi q=5.694 p,0.05), but no significant difference between
midguts at 7 dpi and ‘input’ (Figure 1D). There was no statistical
differences between any of the leg or saliva samples at 7, 14 or
21 dpi (legs p=0.0996, saliva p=0.3563) (Figure 1E & 1F).
The second indicator of genetic diversity used in these studies
was the proportion of unique viral variants. This was determined
by calculating the number of unique clones per sample and
dividing by the total number of clones sequenced per sample.
Again the data was grouped either by days post infection (Figure 2
A, B, & C) or tissue type (Figure 2 D, E, & F). At 7 dpi, the midguts
and saliva were significantly lower than the ‘input’ (ANOVA
p=0.0052, Tukey’s HSD post test, ‘input’ vs midguts q=7.038
p,0.05, ‘input’ vs saliva q=5.841 p,0.05), but the tissues were
not significantly different from one another (Figure 2A). Interest-
ingly, by 14 dpi the proportion of unique viral variants between
the tissues and ‘input’ was not significant (ANOVA p=0.0517),
but at 21 dpi each of three tissue types were significantly lower
than the ‘input’ (ANOVA p=0.0021, Tukey’s HSD post test,
‘input’ vs midguts q=6.898 p,0.05, ‘input’ vs legs q=8.028
p,0.05, and ‘input’ vs saliva q=5.306 p,0.05) (Figure 2B & 2C).
Analysis by tissue type revealed that midguts from all three time
points were significantly lower than the ‘input’, but not different
between time points (ANOVA p=0.0014, Tukey’s HSD post test,
‘input’ vs 7 dpi q=7.298 p,0.05, ‘input’ vs 14 dpi q=7.838
p,0.05, and ‘input’ vs 21 dpi q=7.576 p,0.05) (Figure 2D). Like
the midguts, the legs at 14 and 21 dpi contained significantly less
diversity than the ‘input’ (ANOVA p=0.0018, Tukey’s HSD post
test, ‘input’ vs 14 dpi q=6.597 p,0.05 and ‘input’ vs 21 dpi
q=8.458 p,0.05), but was not different from the 7 dpi time point.
Further, there was no difference between the time points
(Figure 2E). Finally, comparison of the saliva samples at each of
the time points revealed no significant difference between the three
time points and the ‘input’ or between time points (ANOVA
p=0.1431) (Figure 2F).
Because our frequency and location analysis of viral variants
revealed that numerous variants were found in both the ‘input’
and saliva, but not the midgut or legs, we performed a correlation
WNV Diversity is Maintained in Mosquitoes
PLoS ONE | www.plosone.org2September 2011 | Volume 6 | Issue 9 | e24466
analysis between the genetic diversity metrics and log transformed
viral genome equivalents. The Pearson correlation analysis
revealed that percent nucleotide diversity is significantly inversely
correlated to viral genome equivalents (p=0.003; Pearson
r2=20.2747) (Figure 3A). Similarly, viral genome equivalents
are inversely correlated to the proportion of unique viral variants
(p=0.0205; Pearson r2=0.1772) (Figure 3B).
Frequency and Migration Analysis
Analysis of the frequency and location of viral variants revealed
that 78 of 883 sequences sampled were unique. These variants were
found in all three tissue types and at all three time points. There
were 16 variants unique to the ‘input’ mosquitoes, 6 were found in
all four categories (input, midgut, legs, saliva), 19 were unique to
saliva, 15 unique to legs, and 9 unique to midguts, and the
remaining variants were found in multiple tissues. The 14 most
common variants were then plotted to display their relative
proportion in each mosquito sample (Figure 4). By analyzing the
data by this approach we were able to track individual variants from
‘input’ through infection (midguts), dissemination (legs), and
transmission (saliva). The ‘input’ set is a combination of all three
0 hpi mosquitoes and as expected represents a complex population
of multiple variants. Generally, the midgut populations, at all three
time points, are composed of only a few variants with no one variant
Figure 1. Percent nucleotide diversity by time and tissue. The percent nucleotide diversity was determined for each sample and ploted by
either days post infection (7 dpi (A), 14 dpi (B), & 21 dpi (C)) or by tissue type (midguts (D), legs (E), & saliva (F)). Dotted lines connect the means for
each sample set. P-values were determined by ANOVA using Tukey’s multiple comparison post test. Letters above sample sets represent statistically
significant groupings (p-value,0.05). Figures without letters denote that samples were not significantly different from one another.
WNV Diversity is Maintained in Mosquitoes
PLoS ONE | www.plosone.org3 September 2011 | Volume 6 | Issue 9 | e24466
dominating in all samples. Likewise, the WNV populations
recovered from the legs had, in general, low intrahost variability,
but with no overrepresentation of any one variant between
mosquitoes. Interestingly, there was an expansion in the total
number of variants identified in the saliva compared to the midguts
and legs. These findings are supported by the proportion of unique
viral variants analysis (Figure 2). Further, many of the variants that
were present in the input samples and subsequently undetected in
the midguts and legs were recovered from the saliva (Figure 4).
Included inthisdatawas a variantthatcontained a single nucleotide
deletion at nucleotide 2194 in the E-glycoprotein. This deletion
mutant was found in the legs or saliva of three different mosquitoes
at 7 and 14 dpi (black colored sections of Figures 4A & 4B).
Migration analyses were performed in order to more closely look
for evidence of genetic bottlenecks and test for tissue compartmen-
talization (Figure 5). A hypothetical tree was generated to represent
what would be expected if strong genetic bottlenecks were
influencing WNV populations during the EIP (Figure 5A). Under
this scenario, a single ‘input’ variant initiatesinfectioninthe midgut.
Subsequently, a single midgut variant establishes an infection in the
hemolymph from which a single variant invades the salivary glands
and is transmitted. However, this is not what we observed. Three
Figure 2. Proportion unique viral variants by time and tissue. The proportion of unique viral variants was determined for each sample and
plotted by either days post infection (7 dpi (A), 14 dpi (B), & 21 dpi (C)) or by tissue type (midguts (D), legs (E), & saliva (F)). Dotted lines connect
means for each sample set. P-values were determined by ANOVA using Tukey’s multiple comparison post test. Letters above sample sets represent
statistically significant groupings (p-value,0.05). Figures without letters denote that samples were not significantly different from one another.
WNV Diversity is Maintained in Mosquitoes
PLoS ONE | www.plosone.org4 September 2011 | Volume 6 | Issue 9 | e24466
representative mosquitoes,one fromeach timepoint,areshownand
the remaining six trees are provided in a supplement (Figure S3).
The migration analysis indicates some evidence for compartmen-
talization based on the p-values for ordered character states as
estimated from 10,000 randomly sampled trees. A cluster of closely
related variants isolated from legs was identified in our represen-
tative mosquito at 7 dpi (Figure 5B p,0.0001). Similarly, clusters of
saliva variants were identified at 14 dpi and 21 dpi (p,0.0001 and
p,0.0001, respectively) (Figures 5C & 5D).
In vivo Competition Assay
To test for the presence of genetic bottlenecks using a less
geneticallycomplex virus population, mosquitoes werefed on chicks
infected with a known mixture of marked-reference and wild-type
WNV (Text S1). The mean proportion of wild-type to reference
virus in chick viremia was 0.73 (n=4, SEM 0.011) (Figure S4).
Subsequent analysisofthemosquitosamplesrevealedthat therewas
no change in the proportion of wild-type to reference WNV in any
mosquito tissue compared to chick viremia (bodies 0.8 SEM 0.031,
legs 0.73 SEM 0.07, and saliva 0.75 SEM 0.087).
Population bottlenecks during transmission may profoundly
influence the evolution of arboviruses by stochastically reducing
population variation, thereby selecting random genomes that may
be less fit than the overall population. Currently, it is unclear
Figure 3. Viral genome equivalents and genetic diversity are
inversely correlated. Black=‘input’, Orange=midguts, Green=legs,
and Purple=saliva. (A) Log transformed genome equivalents for each
sample plotted against percent nucleotide diversity, n=30, p=0.003,
Pearson r2=0.2747. (B) Log transformed genome equivalents for each
sample plotted against the proportion of unique viral variants, n=30,
p=0.0205, Pearson r2=0.1772.
Figure 4. Frequency and location of unique viral variants. There
were 883 clones sequenced, of which 78 sequences were unique. The
frequency of the fourteen most common viral variants was mapped
back to each sample. The column labeled input combines the data from
all three 0 hours post infection mosquitoes. Samples are broken down
by days post infection (7 dpi (A), 14 dpi (B), & 21 dpi (C)). Each time
point includes three mosquitoes (denoted mosquitoes 1–3) and further
broken down by tissue (M (midguts), L (legs), & S (saliva)). The white
sections of the histograms represent the remaining 64 uncommon
variants and the black sections represent a single nucleotide deletion
mutant found in multiple samples.
WNV Diversity is Maintained in Mosquitoes
PLoS ONE | www.plosone.org5 September 2011 | Volume 6 | Issue 9 | e24466
whether arboviruses experience genetic bottlenecks during infec-
tion of the mosquito vector. Therefore we tracked the composition
of WNV populations during mosquito infection to quantify genetic
bottlenecks associated with infection of these hosts.
Cx. p. quinquefasciatus mosquitoes were exposed to a bloodmeal
containing a highly fit, genetically diverse WNV population (M24)
that has been described in detail elsewhere . Most mosquitoes
exposed to M24 had WNV RNA in midgut tissues after either 7,
14 or 21 days EI. Although fewer mosquitoes had WNV in
hemolymph and salivary secretions, at all timepoints at least three
individual mosquitoes had WNV in midgut, hemolymph and
salivary secretions. WNV from these mosquitoes was used to assess
population bottlenecks associated with mosquito transmission.
Extreme care was taken to minimize the possibility that WNV
RNA from one tissue would contaminate other tissues. Hemo-
lymph was first sampled from newly anesthetized mosquitoes by
gently removing their legs. Second, the mosquito mouthparts were
inserted into a pulled capillary tube charged with buffer and the
mosquito was allowed to salivate for approximately 30 minutes.
Finally, the midgut was removed from the mosquito and washed
three times in PBS to remove hemolymph-associated WNV. This
approach was validated by intrathoracically inoculating adult
female mosquitoes with 26104pfu/ml of WNV. Subsequently,
five whole mosquitoes, five unwashed midguts and three times
washed midguts were collected 45 minutes post inoculation. The
presence of WNV RNA was determined by one-step RT-PCR.
Expectedly, all five whole body mosquitoes were positive for WNV
RNA along with two of five unwashed midguts. All five of the
washed midguts were negative for WNV RNA (data not shown).
Since the greatest concentration of WNV RNA tended to be in the
mosquito midguts, handling this tissue last minimized the
possibility of contaminating samples from the same mosquito.
Several mosquitoes were detected that had midgut-limited
infections, or WNV in hemolymph but not salivary secretions
(data not shown). These results indicate that our efforts to
minimize contamination were effective and that the samples
selected for this study were not compromized by contaminating
Quantitative analysis of viral genetic diversity in mosquito
midguts compared with the highly genetically diverse ‘input’
WNV M24 clearly demonstrated that virus population diversity is
restricted in this tissue. Both the percent nucleotide diversity and
Figure 5. Bayesian trees with the most parsimonious reconstruction of tissue character states for three mosquitoes. Blue=input (i),
pink=midgut (m), gold=legs (l), black=salivary secretions (s), dotted=multiple tissues, with specific tissues indicated by abbreviations. Hatched
branches indicate equivocal reconstruction of character states. Numbers above nodes are the posterior probabilities inferred for each clade. A)
Hypothetical tree with predicted outcome assuming the presence of genetic bottlenecks. The most parsimonious reconstruction of ordered character
steps for each tree for three representative mosquitoes is as follows, B) 7 dpi mosquito 2 (22 steps, p,0.0001), C) 14 dpi mosquito 1 (30 steps,
p,0.0001), D) 21 dpi mosquito 3 (23 steps, p=0.0001).
WNV Diversity is Maintained in Mosquitoes
PLoS ONE | www.plosone.org6 September 2011 | Volume 6 | Issue 9 | e24466
proportion of unique viral variants in midguts are significantly
lower than in the ‘input’ population (Figures 1D & 2D).
Surprisingly, however, the genetic diversity of peripheral WNV
was not significantly different from virus in the bloodmeal, with
saliva-associated WNV (i.e. the WNV that would be transmitted
by a feeding mosquito) tending to be the most genetically diverse
of the three tissue types sampled (Figures 1 & 2). The mechanism(s)
that lead to increased genetic diversity outside of the midgut are
not clear. The presence of a ‘leaky midgut’ may explain this
discrepency between the percent nucleotide diversity in legs and
saliva compared to that of the midguts [24,25]. However, we
found little evidence of WNV bypassing the midgut and directly
infecting secondary tissues (Figure S2). In addition, we observed a
significant inverse correlation between viral genome equivalents
and the genetic diversity metrics (Figure 3). Taken together, these
findings suggest that although WNV populations appear to be
restricted in the midguts, and to a lesser extent in hemolymph, the
genetic diversity of transmitted WNV was similar to that of the
ingested virus population, and that variables other than tissue of
origin determine viral genetic diversity in mosquitoes.
It may be that WNV accumulates mutations during the course
of mosquito infection: relaxation of purifying selection on WNV
sequences has been associated with mosquito infection . To
assess this possibility, we compared the frequency and location of
viral variants present in our mosquito tissue samples to those of the
‘input’ population. Not surprisingly, the majority of the viral
variants identified in the midguts were also present in the ‘input’
population (Figure 4). Interestingly, however, variants found in
legs and saliva were also represented in our ‘input’ dataset, without
being present in the mosquito midguts. These findings support our
quantitative analysis of genetic diversity, in these tissues, and
indicate that the increased variation observed in peripheral WNV
populations was more attributable to genetic diversity in the
‘input’ WNV population than to the generation of novel mutants
during mosquito infection.
Additional evidence supporting the infection of a single cell by
multiple WNV variants was obtained through examination of a
defective WNV sequence in our dataset. Specifically, we
identified a single nucleotide deletion mutant that was found in
multiple mosquito samples, including peripheral compartments,
but not in the ‘input’ (Figure 4). Although it is possible that these
mutants arose independently, it seems more likely that an
ancestral mutant was present but undetected in the M24
population and was maintained in mosquitoes by complementa-
tion. Numerous studies have observed complementation of
defective Flavivirus genomes in cell passage experiments [26,27].
Typically, these studies have found large, ,2 kb, in-frame
deletions at the 59-end of the genome in the structural genes.
Interestingly, one study found long-term transmission of a
defective DENV-1 virus with a premature stop codon in the E
gene . This data suggests that defective WNV particles can
infect mosquitoes, propagate through complementation and
ultimately be transmitted (Figure 4; mosquito 3 saliva 7 dpi).
This implies that multiple WNV virions may frequently infect a
single midgut cell, providing a mechanism by which WNV
genetic diversity may be maintained in mosquitoes despite
limitations in the number of susceptible midgut cells [22,23].
Finally, we performed a migration analysis to formally test for
the presence of bottlenecks and compartmentalization. We
detected compartmentalization in legs and saliva, but found no
evidence of genetic bottlenecks (Figure 5). In this analysis, if
genetic bottlenecks exist, viral variants from the tissue samples
would originate from a single ‘input’ variant as demonstrated in
our hypothetical tree. Rather, variants identified in the saliva were
found to originate from multiple ‘input’ variants indicating the
ability of numerous ‘input’ variants to overcome multiple mosquito
barriers to infection (i.e. midgut infection, midgut escape, and
salivary gland escape barriers). The artificial nature of this
experimental system may explain the discrepancies between our
tests. Mosquitoes were offered a bloodmeal containing WNV M24
which contains an approximately 10 fold increase in the genetic
diversity compared to natural WNV populations [14,19]. This
approach was implemented as a means to more easily track
variation in our populations. It may be that the perceived
bottlenecks were artificial due to saturating the system. As a more
realistic approach to testing for bottlenecks we performed an in vivo
competition assay in which infectious clone-derived wild-type
WNV was competed against a marked reference virus . It was
observed that the proportion of marked refernce virus to wild-type
WNV remained unchanged from ‘input’ to bodies, legs or saliva
(Figure S4). Together, these data suggest that genetic bottlenecks
do not significantly influence WNV populations during the EIP in
Cx. p. quinquefasciatus.
Our genetic approach to transmission bottlenecks provides an
intersting contrast to previous studies of bottlenecks in arbovirus
transmission cycles [22,23]. Using virus-like particles to track
binding and internalization, one study demonstrated that WNV
infects only a few midgut epithelial cells during infection of Cx.
quinquefasciatus . Similar results were found during VEEV
infection of Aedes taeniorhynchus . By virtue of the small number
of infected cells it was concluded that arbovirus populations may
be stochastically reduced at the point of infection. Our genetics
studies of WNV do not support this obsevation. Notably, these
conclusions are not necessarily mutually exclusive: It may be that
the small proportion of susceptible midgut cells are infected with
more than one virus particle or that an undetectable level of
infection occurred in a higher proportion of cells. In fact, a high
frequency of dual infections were observed in the VEEV-Aedes
system . Essentially, only a few susceptible midgut cells may be
needed to propagate a diverse arbovirus population. Our
observation of a deletion mutant persisting, apparently through
complementation, during mosquito infection supports this possi-
The literature regarding the role of bottlenecks in natural
transmission cycles of RNA viruses is currently ambiguous.
Bottlenecks are seemingly unimportant for Cauliflower mosaic
virus in plants, but may exist for other RNA viruses [28–31].
Numerous factors may contribute to this discrepency such as
virus species, single vs two-host systems, mode of transmission
and/or site of inoculation. Further, environmental or host genetic
factors may influence differences between individual hosts within
a given population and likely explain the high variablity observed
between individual mosquitoes in this experiment . Never-
theless, our data establish that transmitted WNV populations are
at least as diverse as those of the imbibed population and
therefore suggests that genetic bottlenecks are unlikely to
significantly influence WNV population biology during horizon-
Materials and Methods
Virus and Mosquito Infections
The highly genetically diverse WNV population, WNV M24,
used for these studies has been previously described . Briefly,
24 WNV isolates from naturally infected mosquitoes and birds
were passaged once on C6/36 Aedes albopictus cells . Titers
were determined by plaque assay on Vero cells and mixed at a
1:1:1… ratio. This mixture was amplified once on C6/36 cells at
WNV Diversity is Maintained in Mosquitoes
PLoS ONE | www.plosone.org7 September 2011 | Volume 6 | Issue 9 | e24466
an MOI of 0.1 and the resultant population titered and genetically
To infect mosquitoes, WNV M24 was mixed 1:1 with
defibrinated goose blood and offered to adult female Culex pipiens
quinquefasciatus 7–8 days post emergence. The virus titer in the
bloodmeal was 66106pfu/ml. Fully engorged females were
separated from the remaining unfed mosquitoes and housed in
an environmental chamber (27uC, 16:8 L:D photoperiod) for the
remainder of the experiment.
Sample Collection and Virus Detection
To quantify viral genetic diversity of the ‘input’ virus
population, three fully engorged mosquitoes were placed in
RNA extraction buffer immediately after feeding and homoge-
nized using a mixer mill. Viral genetic diversity was quantified as
described below. At 7, 14, and 21 days post-infection paired tissues
samples (midguts, legs, and saliva) were collected from 20–25
mosquitoes. To ensure that contaminating WNV from the
hemoceol was not introduced into our midgut samples, dissected
midguts were washed three times in PBS before placing the
samples in RNA extraction buffer. Dissecting forceps were flame
sterilized between dissections to avoid cross contamination
between samples. Total RNA was extracted from mosquito
hemolymph and tissues using the RNeasy Mini Protect kit
(Qiagen, Valencia, CA) and screened for the presence of WNV
RNA by one-step RT-PCR using the Superscript III kit with
platinum Taq (Invitrogen, Carlsbad, CA). WNV specific primers
used in this study spanned a 934 nt. region corresponding to the
E-NS1 junction (1971 nt–2928 nt). Three mosquitoes with
detectable WNV RNA in all three tissue types were selected from
each time point for further analyses.
Quantification of Viral Genetic Diversity
Viral genetic diversity was determined according to methods
previously described . Briefly, cDNA was generated from 5 ml
of total RNA using the High Fidelity Reverse Transcription kit
(Stratagene, Cedar Creek, TX) according to the manufacturers
specifications and WNV specific primers, WNV 1971 F and WNV
2928 R. Subsequently, the cDNA served as template for high
fidelity Pfu Ultra polymerase amplification (Stratagene). Amplicons
were PCR purified and cloned into the pCR Script Amp(+)vector
(Stratagene). Between 21–30 individual clones from each of the
samples were sequenced using the M13F, M13R, WNV 2369 F,
and WNV 2768 R primers. DNAStar’s SeqMan module
(DNAStar Inc., Madison, WI) was used for sequence alignment
and analysis of genetic diversity. Only clones with two-fold
sequencing coverage were considered complete. As a means to
estimate genetic diversity, consensus sequences for each sample
were determined and individual clones within that sample were
then compared to the specimen-specific consensus sequence. The
percent nucleotide diversity (total number of mutations from all
clones within a sample divided by the total number of nulcoetides
sequenced per sample) and the proportion of unique viral variants
(the number of unique clones differing from the consensus divided
by the total number of clones sequenced per sample) were
calculated and used as indicators of genetic diversity.
Quantification of Viral Genome Equivalents
WNV genome equivalents were determined by quantitative-RT-
PCR (Q-RT-PCR). As a standard control for this assay a ,2 kb
fragment from the WNV E gene was amplified using the WNV
1031 F and WNV3430 R primers. The resultant amplicon was
cloned into the pCR2.1-TOPO vector (Invitrogen) downstream of
the T7 promoter. The recombinant vector was linearized with Kpn
I, purified and used as template for in vitro transcription using the T7
Megascript kit according the manufacturer’s instructions (Ambion,
Austin, TX). The resultant RNA was quantified and aliquoted in
serial ten-fold dilutions. Using a probe specific for the E gene, the
WNV 1160 F and WNV 1229 R primers, and the TaqMan H One-
Step RT-PCR Master Mix Reagent (Applied Biosystems, Foster
City, CA) viral RNA copy numbers were determined . Samples
were run on the ABI Prism 7000 Sequence Detection System
Frequency and Migration Analysis
The presence of genetic bottlenecks and/or compartmentaliza-
tion was further assessed by migration analyses. To determine the
frequency and location of viral variants, sequences from each
sample were aligned in DNAStar’s SeqMan module, exported as
FASTA files and duplicates removed using BioEdit .
Alignments were generated for each mosquito and tested for
recombination using the Genetic Algorithm for Recombination
detection program implemented on the datamonkey.org website
. Evidence of recombination was not detected, so the
alignments were used to perform a migration analysis. To test
the null hypothesis of panmixis versus the alternative that there are
distinct WNV sub-populations within different mosquito tissues,
we used the Slatkin-Maddison test for gene flow in MacClade
version 4 (Sinauer Associates, Sunderland, MA) . Tissue of
origin was assigned to each taxon in a one-character data matrix.
‘Input’ sequences from freshly-fed mosquitoes were included as an
estimate of the population of variants present in the infectious
bloodmeal. In total there were four character states (input, midgut,
legs, and saliva). The Slatkin-Maddison test was performed
independently for each mosquito. This analysis was performed
on Bayesian phylogenies, generated with MrBayes 3.1.2 .
These were run with a general time reversible (GTR) model with
invariable rates with substitution rates following a gamma plus
invariants distribution. Two Markov Chains Monte Carlo
(MCMC) tree searches of 5 million generations each were run in
parallel with sampling one in every 250 trees. 50% majority-rule
consensus trees are shown based on the last 19,000 trees. Briefly,
the phylogenetic tree resulting from the nucleotide data was
loaded into MacClade and the most parsimonious reconstruction
of this ancestral character inferred with the Fitch algorithm  in
order to estimate the minimum number of steps required to
explain the distribution of tissue states on the tree of interest. We
then generated 10,000 random trees by random joining and
splitting of the input tree and compared the number of steps on
our input tree to those calculated in the random trees, as described
previously for HIV-1, using ordered tissue states .
Statistical analyses were completed in Microsoft Excel and
GraphPad Prism. A one-way analysis of variance (ANOVA) with
the Tukey’s multiple comparison post-test with a significance level
of a=0.05 was used for analysis of the percent nucleotide diversity
and proportion of unique viral variants. A Pearson correlation
analyses was perfomed on log transformed viral genome
equivalents versus percent nucleotide diversity and proportion of
unique viral variants. Figures were generated in GraphPad.
WNV genome equivalents were determined by Q-RT-PCR for
each sample characterized.
WNV genome equivalents per tissue sample.
WNV Diversity is Maintained in Mosquitoes
PLoS ONE | www.plosone.org8 September 2011 | Volume 6 | Issue 9 | e24466
hemolymph at early time points. Mosquitoes were offered a
WNV Mix24 infectious bloodmeal and hemolymph extracted at
multiple time points. WNV titers were determined by plaque
WNV titers in Culex pipiens quinquefasciatus
reconstruction of tissue character states from the six
remaining mosquitoes. Blue=input (i), pink=midgut (m),
gold=legs (l), black=salivary secretions (s), dotted=multiple
tissues, with specific tissues indicated by abbreviations. Hatched
branches indicate equivocal reconstruction of character states.
Numbers above the nodes are the posterior probabilities inferred
for each clade. Mosquito analyzed and most parsimonious
reconstruction of ordered character steps for each tree is as
follows A) 7 dpi mosquito 1 (21 steps, p=0.0002) B) 7 dpi
mosquito 3 (10 steps, p=0.0057), C)14 dpi mosquito 2 (8 steps,
p=0.087), D) 14 dpi mosquito 3 (23 steps, p=0.0002), E) 21 dpi
mosquito 1 (12 steps, p=0.0772), F) 21 dpi mosquito 2 (16 steps,
Bayesian trees with the most parsimonious
competed against a marked reference virus does not
The proportion of wild-type WNV when
change as the virus disseminates through the mosquito.
Culex pipiens quinquefasciatus mosquitoes were fed on live chicks
circulating a mixed population of WNV comprised of wild-type
(WT) and reference viruses. Tissues were harvested 7 dpi from 20
mosquitoes and the proportion of WT-WNV was determined by
RT-PCR followed by SNPS analysis. Samples negative for WNV
RNA by RT-PCR were omitted.
Materials and Methods.
The authors thank Kelly Fitzpatrick for technical assistance, and Marcelo
Jacobs-Lorena for helpful discussions regarding leaky midgut experiments.
In addition we would like to thank the reviewers and editors for their
thoughful and insightful suggestions.
Conceived and designed the experiments: GDE DEB. Performed the
experiments: DEB IKB ERD JK. Analyzed the data: KP DEB ERD.
Contributed reagents/materials/analysis tools: ERD GDE. Wrote the
paper: DEB GDE KP.
1. Randolph SE, Rogers DJ (2010) The arrival, establishment and spread of exotic
diseases: patterns and predictions. Nat Rev Microbiol 8: 361–371.
2. Anderson JF, Vossbrinck CR, Andreadis TG, Iton A, Beckwith WH, et al.
(2001) A phylogenetic approach to following West Nile virus in Connecticut.
Proc Natl Acad Sci USA 98: 12885–12889.
3. Beasley DWC, Davis CT, Guzman H, Vanlandingham DL, da Rosa A, et al.
(2003) Limited evolution of West Nile virus has occurred during its southwesterly
spread in the United States. Virology 309: 190–195.
4. Davis CT, Beasley DWC, Guzman H, Raj P, D’Anton M, et al. (2003) Genetic
variation among temporally and geographically distinct West Nile virus isolates,
United States, 2001, 2002. Emerg Infect Dis 9: 1423–1429.
5. Ebel GD, Carricaburu J, Young D, Bernard KA, Kramer LD (2004) Genetic
and phenotypic variation of West Nile virus in New York, 2000–2003. Am J Trop
Med Hyg 71: 493–500.
6. Ebel GD, Dupuis AP, Ngo K, Nicholas D, Kauffman E, et al. (2001) Partial
genetic characterization of West Nile Virus strains, New York State, 2000.
Emerg Infect Dis 7: 650–653.
7. Moudy RM, Dupuis A, Ebel GD, Kramer LD (2006) Displacement of the
introduced genotype of West Nile virus in New York state. Am J Trop Med Hyg
8. Moudy RM, Meola MA, Morin LLL, Ebel GD, Kramer LD (2007) A newly
emergent genotype of west Nile virus is transmitted earlier and more efficiently
by Culex mosquitoes. Am J Trop Med Hyg 77: 365–370.
9. Tsetsarkin KA, McGee CE, Volk SM, Vanlandingham DL, Weaver SC, et al.
(2009) Epistatic roles of E2 glycoprotein mutations in adaption of chikungunya
virus to Aedes albopictus and Ae. aegypti mosquitoes. PLoS One 4: 12.
10. Tsetsarkin KA, Vanlandingham DL, McGee CE, Higgs S (2007) A single
mutation in chikungunya virus affects vector specificity and epidemic potential.
PLoS Path 3: 1895–1906.
11. Domingo E, Escarmis C, Sevilla N, Moya A, Elena SF, et al. (1996) Basic
concepts in RNA virus evolution. Faseb J 10: 859–864.
12. Duarte EA, Novella IS, Weaver SC, Domingo E, Wainhobson S, et al. (1994)
RNA virus quasispecies - Significance for viral disease and epidemiology. Infect
Agent Dis 3: 201–214.
13. Eigen M, McCaskill J, Schuster P (1988) Molecular quasispecies. J Phys Chem
14. Jerzak G, Bernard KA, Kramer LD, Ebel GD (2005) Genetic variation in West
Nile virus from naturally infected mosquitoes and birds suggests quasispecies
structure and strong purifying selection. J Gen Virol 86: 2175–2183.
15. Jerzak GVS, Bernard K, Kramer LD, Shi PY, Ebel GD (2007) The West Nile
virus-mutant spectrum is host-dependant and a determinant of mortality in
mice. Virology 360: 469–476.
16. Sauder CJ, Vandenburgh KM, Iskow RC, Malik T, Carbone KM, et al. (2006)
Changes in mumps virus neurovirulence phenotype associated with quasispecies
heterogeneity. Virology 350: 48–57.
17. Sullivan DG, Bruden D, Deubner H, McArdle S, Chung M, et al. (2007)
Hepatitis C virus dynamics during natural infection are associated with long-
term histological outcome of chronic hepatitis C disease. The Journal of
Infectious Diseases 196: 239–248.
18. Vignuzzi M, Stone JK, Arnold JJ, Cameron CE, Andino R (2006) Quasispecies
diversity determines pathogenesis through cooperative interactions in a viral
population. Nature 439: 344–348.
19. Fitzpatrick KA, Deardorff ER, Pesko K, Brackney DE, Zhang B, et al. (2010)
Population variation of West Nile virus confers a host-specific fitness benefit in
mosquitoes. Virology 404: 89–95.
20. Duarte E, Clarke D, Moya A, Domingo E, Holland J (1992) Rapid fitness losses
in mammalian RNA virus clones due to Muller’s ratchet. Proc Nat Acad Sci
USA 89: 6015–6019.
21. Aaskov J, Buzacott K, Thu HM, Lowry K, Holmes EC (2006) Long-term
transmission of defective RNA viruses in humans and Aedes mosquitoes. Science
22. Smith DR, Adams AP, Kenney JL, Wang E, Weaver SC (2008) Venezuelan
equine encephalitis virus in the mosquito vector Aedes taeniorhynchus: Infection
initiated by a small number of susceptible epithelial cells and a population
bottleneck. Virology 372: 176–186.
23. Scholle F, Girard YA, Zhao QZ, Higgs S, Mason PW (2004) trans-packaged
West Nile virus-like particles: Infectious properties in vitro and in infected
mosquito vectors. J Virol 78: 11605–11614.
24. Houk EJ, Hardy JL (1979) In vivo negative staining of the midgut continuous
junction in the mosquito, Culex tarsalis (Diptera, Culicidae). Acta Trop 36: 267–275.
25. Weaver SC, Scott TW, Lorenz LH, Repik PM (1991) Detection of eastern equine
encephalomyeltis virus deposition in Culiseta melanura following ingestion of
radiolabeled virus in blood meals. Am J Trop Med Hyg 44: 250–259.
26. Tsai KN, Tsang SF, Huang CH, Chang RY (2007) Defective interfering RNAs
of Japanese encephalitis virus found in mosquito cells and correlation with
persistent infection. Virus Res 124: 139–150.
27. Yoon SW, Lee SY, Won SY, Park SH, Park SY, et al. (2006) Characterization of
homologous defective interfering RNA during persistent infection of Vero cells
with Japanese encephalitis virus. Mol Cells 21: 112–120.
28. Ali A, Roossinck MJ (2010) Genetic bottlenecks during systemic movement of
cucumber mosaic virus vary in different host plants. Virology 404: 279–283.
29. Kuss SK, Etheredge CA, Pfeiffer JK (2008) Multiple host barriers restrict
poliovirus trafficking in mice. PLoS Path 4: 13.
30. Monsion B, Froissart R, Michalakis Y, Blanc S (2008) Large bottleneck size in
cauliflower mosaic virus populations during host plant colonization. PLoS Path
31. Wang GP, Sherrill-Mix SA, Chang KM, Quince C, Bushman FD (2010)
Hepatitis C virus transmission bottlenecks analyzed by deep sequencing. J Virol
32. Cadwell K, Patel KK, Maloney NS, Liu TC, Ng ACY, et al. (2010) Virus-plus-
susceptibility gene interaction determines Crohn’s disease gene Atg16L1
phenotypes in intestine. Cell 141: 1135–U1164.
33. Lanciotti RS, Kerst AJ, Nasci RS, Godsey MS, Mitchell CJ, et al. (2000) Rapid
detection of West Nile virus from human clinical specimens, field-collected
mosquitoes, and avian samples by a TaqMan reverse transcriptase-PCR assay.
J Clin Microbiol 38: 4066–4071.
34. Hall TA (1999) BioEdit: a user-friendly biological sequence alignment editor and
analysis program for Windows 95/98/NT. Nucl Acids Symp Ser. pp 95–98.
WNV Diversity is Maintained in Mosquitoes
PLoS ONE | www.plosone.org9 September 2011 | Volume 6 | Issue 9 | e24466
35. Pond SLK, Posada D, Gravenor MB, Woelk CH, Frost SDW (2006) Automated Download full-text
phylogenetic detection of recombination using a genetic algorithm. Mol Biol
Evol 23: 1891–1901.
36. Slatkin M, Maddison WP (1989) A cladistic measure of gene flow inferred from
the phylogenies of alleles. Genetics 123: 603–613.
37. Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference
under mixed models. Bioinformatics 19: 1572–1574.
38. Fitch WM (1971) Toward defining the course of evolution: minimum change for
a specific tree topology. System Zoo 20: 406–416.
39. Salemi M, Burkhardt BR, Gray RR, Ghaffari G, Sleasman JW, et al. (2007)
Phylodynamics of HIV-1 in lymphoid and non-lymphoid tissues reveals a central
role for the thymus in emergence of CXCR4-using quasispecies. PLoS One 2:
WNV Diversity is Maintained in Mosquitoes
PLoS ONE | www.plosone.org10 September 2011 | Volume 6 | Issue 9 | e24466