Limited Dengue Virus Replication in Field-Collected Aedes aegypti Mosquitoes Infected with Wolbachia

Introduction Dengue is one of the most widespread mosquito-borne diseases in the world. The causative agent, dengue virus (DENV), is primarily transmitted by the mosquito Aedes aegypti, a species that has proved difficult to control using conventional methods. The discovery that A. aegypti transinfected with the wMel strain of Wolbachia showed limited DENV replication led to trial field releases of these mosquitoes in Cairns, Australia as a biocontrol strategy for the virus. Methodology/Principal Findings Field collected wMel mosquitoes that were challenged with three DENV serotypes displayed limited rates of body infection, viral replication and dissemination to the head compared to uninfected controls. Rates of dengue infection, replication and dissemination in field wMel mosquitoes were similar to those observed in the original transinfected wMel line that had been maintained in the laboratory. We found that wMel was distributed in similar body tissues in field mosquitoes as in laboratory ones, but, at seven days following blood-feeding, wMel densities increased to a greater extent in field mosquitoes. Conclusions/Significance Our results indicate that virus-blocking is likely to persist in Wolbachia-infected mosquitoes after their release and establishment in wild populations, suggesting that Wolbachia biocontrol may be a successful strategy for reducing dengue transmission in the field.


Introduction
Dengue is one of the most common and widespread vectorborne diseases in the world, with up to 380 million infections estimated to occur annually [1]. The causative agent, dengue virus (DENV), has expanded its geographic range in the last two decades, with more than 100 countries now affected. Infection with DENV leads primarily to self-limiting fevers but recent decades have seen a marked increase in severe dengue, with manifestations such as hypovolemic shock and hemorrhage [2]. DENV is transmitted primarily by the mosquito vector Aedes aegypti and, to a lesser extent, by its congener A. albopictus. In the absence of an effective vaccine [3] and/or antivirals, prevention of dengue transmission relies primarily on control of mosquito vectors. The failure to prevent the global spread of dengue, increasing insecticide resistance in mosquito populations and subsequent escalating costs of insecticide-based programs, as well as environmental concern over the impact of these chemicals, have spurred the development of novel, inexpensive and green vector control methods [4,5].
The transinfection of vector mosquitoes with the bacterium Wolbachia pipientis has emerged as a promising method for the control of dengue. Wolbachia is the most common endosymbiont of insects, thought to infect up to 40% of arthropod species [6]. A. aegypti stably transinfected with different strains of Wolbachia show reduced replication and transmission of DENV [7][8][9]. An additional advantage of using Wolbachia for biocontrol of DENV is the ability of the bacterium to propagate through a population by inducing cytoplasmic incompatibility (CI) in its host [10]. CI confers a fitness advantage to Wolbachia-infected females that allows these maternally transmitted bacteria to spread unaided through a population [10]. The use of Wolbachia provides a means of biocontrol that is both pesticide-free and poses minimal environmental safety concerns [11].
In laboratory trials, mosquitoes with the wMel strain of Wolbachia showed both blocking of DENV transmission and minimal fitness effects due to infection with the bacterium [9]. In addition, wMel rapidly invaded wildtype mosquito populations in semi-field cage experiments due to CI and minimal fitness costs [9]. The results facilitated the field release of wMel-infected mosquitoes in two suburbs of Cairns, Queensland, Australia [12]. Within a short period, the frequency of wMel reached fixation in the two suburbs [12] and has remained established at both sites.
The persistence of the viral-blocking phenotype in field populations is fundamental to the utility of releases of Wolbachia-infected mosquitoes. The mechanisms that underpin viral interference are poorly understood but may be related to the density of Wolbachia [13,14], immune pre-activation [7,8,15], intra-host competition for cellular resources [16,17] or suppression of host cellular factors that are upregulated during viral infection [18]. The density of Wolbachia may decrease after several generations, as happened following the transinfection of the virulent strain of wMelPop into the novel host Drosophila simulans [19]. Wolbachia infection frequencies and associated CI effects may also be significantly lower in nature than observed in the lab, as observed in Drosophila simulans [20]. However, the wMel strain is avirulent and has limited negative effects on mosquito fitness in the laboratory [9], suggesting that the density of the wMel strain may remain stable over time. Protection against RNA virus-induced mortality was in fact first observed in the long term, evolutionarily stable association between wMel and its Drosophila melanogaster host [21].
Here, we investigated the extent of virus blocking in field wMelinfected A. aegypti, one year following field release, using three serotypes of DENV. We found limited replication and dissemination of DENV in field wMel mosquitoes, indicating stability of the viral-blocking phenotype in wild Wolbachia-infected mosquitoes. The extent of virus blocking was similar in field mosquitoes compared to the original, wMel-infected, outcrossed lab line used for release. Interestingly, the density of Wolbachia increased following blood feeding and to a greater extent in field versus lab wMel-infected mosquitoes. We suggest that if the viral blocking effect of field wMel is dependent on Wolbachia density, repeated blood feeding on human hosts might amplify this effect. Our results reinforce the utility of Wolbachia-based technology for biocontrol of dengue.

Ethics statement
Blood feeding of mosquito colonies using human volunteers was performed in accordance to Monash University Human Research Ethics Committee permit CF11/0766-2011000387. Written informed consent was obtained from all volunteers who participated in the study. Dengue viremic plasma was obtained from patients enrolled in a prospective study at the Hospital for Tropical Diseases, Ho Chi Minh City, Vietnam. All patients provided written consent to participate in the study. The study protocols relevant to this work, including vector competence experiments, were reviewed and approved by the Scientific and Ethical Committee of the Hospital for Tropical Diseases (CS/ND/09/ 24) and the Oxford Tropical Research Ethical Committee (OxTREC 20-09). The inclusion criteria were: a) adult patients ($15 years of age), with #72 hours of fever and suspected of having dengue based on clinical symptoms, b) a positive NS1 Rapid test and c) written informed consent. All plasma samples were anonymized (samples were identified using numbers only) prior to experiments.

Mosquito colony establishment and maintenance
Mosquito eggs were collected in January 2012 from ovitraps placed inside the Wolbachia release zone in the Cairns suburbs of Yorkey's Knob and Gordonvale and outside, in Edge Hill, Whitfield, Edmonton and Bentley Park. Eggs collected from outside the Wolbachia release zone were Wolbachia-uninfected. Eggs on ovistrips were allowed to hatch and larvae reared in water supplemented with fish food pellets (Tetramin, Tetra). Fourth instar larvae were identified as A. aegypti based on specific morphological characters. Adults (F 0 ) emerged in cages of approximately 450 individuals and were allowed to feed on 10% sucrose ad libitum. Five to seven day old females were allowed to feed on human volunteers and eggs were collected from several gonotrophic cycles. F 1 adults hatched from eggs obtained in the first gonotrophic cycle were used in vector competence experiments. The wMel-infected field mosquito line and its uninfected counterpart (derived from Wolbachia-uninfected eggs) were denoted wMel.F and wildtype, respectively. The original laboratoryreared, outcrossed wMel-infected MGYP2.out line [9] was used in some experiments. All mosquito colonies were kept at 26uC under a 12L:12D light cycle and 60% relative humidity.

Virus strains
Mosquitoes were challenged in vector competence experiments with virus strains belonging to DENV serotypes 1-3, using virus grown in cell culture and viremic plasma from human patients. DENV-2 strain 92T and DENV-3 strain Cairns 2008 (both isolated from outbreaks in north Queensland, Australia in 1992 and 2008, respectively) were grown in C6/36 cells and harvested and titered as described previously [13]. Virus was aliquoted in single-use 1 mL lots and stored at 280uC.

Vector competence experiments
Two separate vector competence experiments were carried out to determine if DENV could replicate and disseminate in field wMel-infected mosquitoes. For both experiments, female mosquitoes (5-7 days old) were allowed to feed on viremic blood meals contained in a membrane feeder with sheep intestine as the membrane. Virus was mixed with defibrinated sheep blood to

Author Summary
Almost half of the world's population is at risk of contracting dengue virus, particularly in the tropics and sub-tropics. The virus is transmitted by the mosquito Aedes aegypti, a cosmopolitan species that has proved difficult to control using traditional methods. A new biocontrol strategy has been developed involving the release of mosquitoes infected with Wolbachia bacteria. Mosquitoes with the wMel strain of Wolbachia show dramatically reduced replication and transmission of dengue virus in laboratory trials. Although promising, the utility of Wolbachia biocontrol depends on field wMel-infected mosquitoes retaining the phenotype of reduced viral replication. Mosquitoes with wMel were released in the field in Cairns, Australia in early 2011. We provide evidence that, one year later, field collected wMel mosquitoes showed reduced dengue virus replication in the body and limited dissemination to the head compared to controls. Wolbachia numbers in mosquitoes increased following blood meals, which may further decrease viral replication if the insects feed frequently. Our results indicate that Wolbachia-mediated dengue interference is sustained in field populations and shows no sign of attenuation after one year of deployment.
obtain final bloodmeal titers (see below). Mosquitoes were allowed to feed for 1 hour, with engorged females separated from unfed ones the next day. Females were kept in plastic cups at a density of 10-12 individuals/cup and allowed access to 10% sucrose ad libitum. Females were killed under CO 2 at either 7 or 14 days post infection (p.i.), immediately frozen in dry ice and stored at 2 80uC until further processing.

RNA extraction and qRT-PCR for DENV
RNA was extracted from mosquito bodies using Trizol reagent (Invitrogen), and from heads using the QIAamp viral RNA mini kit (Qiagen), following homogenization of tissues with 3 mm glass beads in a Beadbeater. A higher yield of total RNA was obtained on average from head samples using the QIAamp viral RNA mini kit versus Trizol (F. Frentiu, unpublished data). For mosquitoes challenged with Vietnamese viremic plasmas, virus genome copies were estimated by qRT-PCR using FAM-labeled DENV-1 and DENV-2 hydrolysis probe sequences and standard curves from reference [22]. Virus copies in mosquitoes challenged with DENV-2-92T and DENV-3-Cairns 2008 were estimated by qRT-PCR, using hydrolysis probes specific to the 39UTR region. Primer sequences were F: 59-AAGGACTAGAGGTTAGAGGA-GACCC-39 and R: 59-CGTTCTGTGCCTGGAATGATG-39, with probe sequence: 59-FAM-AACAGCATATTGACGCTGG-GAGAGACCAGA-BHQ1-39. Reactions were performed with the SuperScriptH III PlatinumH One-Step qRT-PCR kit (Invitrogen) and contained 5 mL of RNA template, 5 mM each of probe and forward and reverse primers, buffer and enzyme as per kit instructions, in a total volume of 20 mL. For head qPCRs, 10 mL of RNA template was used, with water adjusted accordingly. The number of DENV copies was calculated following a standard curve for DENV 39UTR, constructed as in [8]. All reactions were performed using a LightCycler480 Instrument (Roche) with the following run conditions: 50uC for 15 min, 95uC for 2 min, followed by 45 amplification cycles of 95uC for 15 s, 60uC for 30 s and a final cooling step of 40uC for 10 s.
Reactions were run in duplicate and samples where DENV failed to amplify in at least one replicate were classified as zero. Only samples where DENV amplified in both technical replicates and the amount of copies extrapolated by the LightCycler software was above the lower bound of the standard curve (limit of detection) were included in the analysis. All mosquitoes from field and lab wMel-infected lines that showed DENV breakthrough were tested for the presence of Wolbachia using IS5 repeat primers specific to the wMel and wMelPop strains [23]. Only one sample each from the field and lab wMel-infected mosquitoes was negative for Wolbachia. These samples were excluded from further analysis.

DNA extraction and quantification of Wolbachia density
The densities of Wolbachia were compared between field and lab strains of wMel-infected mosquitoes in a separate experiment. Five to seven-day old females from each line were fed on a mix of DENV-3 -Cairns 2008 and sheep blood and collected at 7 and 14 days post infection (as detailed above) for genomic DNA extraction. Control non-blood fed females from each line were maintained in parallel and collected at the same time points. Genomic DNA was extracted using the DNAEasy Blood and Tissue kit (Qiagen) as per the manufacturer's instructions. A multiplex qPCR amplifying the target Wolbachia-specific wsp and mosquito housekeeping RpS17 [24] genes was performed (wsp F: 59-CATTG-GTGTTGGTGTTGGTG-39, R: 59-ACACCAGCTTTTACTT-GACCAG-39, probe: 59-HEX-TCCTTTGGAACCCGCTGTG-AATGA-BHQ1-39; RpS17 F: 59-TCCGTGGTATCTCCAT-CAAGC-39, R: 59-CACTTCCGGCACGTAGTTGTC-39, probe: 59-FAM-CAGGAGGAGGAACGTGAGCGCAG-BHQ1-39).The RpS17 housekeeping gene was used to normalize wsp gene copies. qPCR reactions were performed in 10 mL total volume containing 16 Lightcycler 480 Probes Master reaction mix, 5 mM each of wsp primers and probe, 2.5 mM each of RpS17 primers and probe and 1 mL of DNA template. Cycling was performed using a Light-Cycler480 Instrument (Roche), with 1 cycle at 95uC for 5 min, followed by 45 amplification cycles of 95uC for 10 s, 60uC for 15 s, 72uC for 1 s, and a final cooling cycle of 40uC for 10 s. Target to housekeeping gene ratios were calculated using the Relative Quantification algorithm in the Lightcycler 480 software (Roche).

Fluorescence in-situ hybridization (FISH)
Tissue localization of wMel in field wMel.F and lab MGYP2.out mosquitoes was visualized using FISH. Females were collected under CO 2 and immediately placed overnight in 4% paraformaldehyde at 4uC with their wings and legs removed. Paraffinembedded mosquitoes were sectioned in 8 mM thin slices. Slides were de-paraffinated in 100% xylene, rehydrated in an ethanol series and hybridized overnight at 37uC in a buffer containing Wolbachia-specific W2 and W3 probes [8]. Post-hybridization processing followed [8]. Slides were mounted using an antifade reagent (Prolong Gold, Invitrogen) and viewed with a Zeiss Axio Imager II epifluorescence microscope equipped with an Axiocam camera, using the same exposure conditions for each filter channel.

Statistical analysis
Differences between mosquito lines in DENV infection rates for both vector competence experiments were analyzed using pairwise Fisher's exact tests. P-values were adjusted for multiple comparisons for each day of sampling within each experiment using the Holm method [25], with values ,0.05 considered significant. In experiment 1, differences in median DENV copy numbers between lines were analyzed using Mann-Whitney U tests. In experiment 2, differences among the three lines in copies of each virus were analyzed using Kruskal-Wallis tests, with Dunn's posthoc multiple comparison tests. Last, we tested for significant  differences in Wolbachia density between MGYP2.out and wMel.F mosquitoes using Mann-Whitney U tests. All analyses were performed in R [26] and GraphPad Prism v. 6 (GraphPad Software, San Diego, California USA).

Limited DENV infection and replication in field wMelinfected mosquitoes
We conducted two independent experiments to assess rates of DENV infection and replication in wildtype and wMel-infected field release mosquitoes. In experiment 1, at day 7 p.i., lower rates of body and head infection were detected in field wMel mosquitoes compared to wildtype for the two DENV-1 and DENV-2 viremic plasma samples and cell culture DENV-2-92T virus strains ( Table 1). However, only for DENV-2 strain P410, a viremic plasma sample, was there a statistically significant difference between the two mosquito lines ( Table 1). At day 14 p.i., rates of body and head infection were significantly lower in field wMel compared to wildtype mosquitoes for all three DENV strains, with a stronger effect in dissemination to heads ( Table 1). The highest observed dissemination rate in wMel.F heads was a low 6%, compared to 62% in wildtype heads. DENV genome copy titers in heads and bodies were uniformly higher for all strains in wildtype mosquitoes compared to respective wMel.F samples at day 14 p.i. (Figure 1). For example, titers in both bodies and heads typically reached 1610 8 copies for all virus strains in wildtype individuals. By contrast, most wMel.F individuals showed an absence of DENV replication (Figure 1). A similar difference in virus titers was present at day 7 p.i., but to a lesser extent because of low infection rates ( Figure S1).
We next investigated whether vector competence was similar in field wMel-infected A. aegypti compared to the original wMelinfected line that had been maintained in the lab with recurrent outbreeding [9]. In experiment 2, we estimated DENV infection rates and replication titers for three virus strains in wildtype, wMel.F and MGYP2.out mosquitoes. We tested for statistically significant differences in infection rates only between wildtypes and wMel.F, and between wMel.F and MGYP2.out mosquitoes ( Table 2). At day 7 p.i., significantly lower body infection rates were found in wMel.F mosquitoes versus wildtypes for DENV-2-92T and DENV-3-Cairns08 strains ( Table 2). However, rates of infection across all mosquito lines and all viruses were low in general, resulting in limited power for robust statistical tests. At day 14 p.i., significantly different infection rates between wildtypes and wMel.F mosquitoes were found for both bodies and heads across all DENV strains ( Table 2). For both experiments 1 and 2, dissemination of all virus strains by day 14 p.i. was dramatically lower in field wMel mosquitoes compared to wildtypes. There were no significant differences in infection rates between wMel.F and MGYP2.out mosquitoes across either day post-infection.
DENV titers were significantly lower across all virus strains in both heads and bodies in field wMel mosquitoes compared to wildtypes, at day 14 post-infection ( Figure 2). A similar pattern was observed at day 7 post-infection, although only for bodies and the strains DENV-2-92T and DENV-2-Cairns08/09 ( Figure S2).  At day 14, virus titers in wildtype mosquitoes ranged from below the limit of detection to 10 8 copies/mg of RNA whereas virus was observed only in a few instances in field wMel. Only in one field wMel individual was the maximum number of DENV copies observed (Figure 2, strain 92T body and heads panels). Overall, the results indicate that when breakthrough virus occurs in wMel mosquitoes, viral titers are most likely to be lower than those observed in wildtypes.

Wolbachia tissue tropism and density in field mosquitoes
We next investigated whether Wolbachia tissue tropism and density had changed significantly in field wMel mosquitoes since release in 2011. Using FISH, we found that Wolbachia was distributed in the same tissues in field mosquitoes and in the original wMel-transinfected laboratory line, MGYP2.out (Figure 3). In both wMel-infected lines, Wolbachia was present in two tissues that are critical in viral infection and dissemination, namely midguts and salivary glands (Figure 3 A-B & G-H). Wolbachia was also present in brains, although not at high densities which was consistent with levels expected for the wMel strain [9]. Field wMel ovaries appeared highly infected with Wolbachia (Figure 3  D), indicating the potential for stable transmission of the bacteria to offspring in the wild.
We also examined whether Wolbachia densities change following blood-feeding in field wMel mosquitoes compared to the original MGYP2.out line. By initially looking at whole mosquitoes we found that, by day 7, the density of wMel had increased following blood-feeding in both lines (Figure 4 A). A much higher increase in Wolbachia density was observed in field wMel mosquitoes versus MGYP2.out (Figure 4 A). Median ratios of wsp to RpS17 gene copy numbers increased significantly from 0.714 and 0.702 in non-blood fed wMel.F and MGYP2.out, respectively, to 1.465 and 1.241 in blood-fed wMel.F and MGYP2.out, respectively (Figure 4  A). The difference in Wolbachia density between blood-fed and non-blood fed mosquitoes persisted at 14 days post feeding (Figure 4 B), in the absence of repeat feeds. Median ratios of wsp to RpS17 gene copy numbers were 0.649 and 0.733 in non-blood fed wMel.F and MGYP2.out, respectively, compared to 1.542 and 1.675 in blood-fed wMel.F and MGYP2.out, respectively (Figure 4  B). Interestingly, by day 14, Wolbachia density continued to increase in blood-fed MGYP2.out and field wMel mosquitoes compared to non-blood fed ones, as indicated by the slightly higher median values of normalized wsp/RpS17 ratios (Figure 4 B). Following blood-feeding, increases in Wolbachia density in both field and laboratory lines were primarily localized in the bodies rather than heads ( Figure 5), probably due to the bacteria replicating in ovaries.

Discussion
Infection of the vector A. aegypti with Wolbachia has been proposed as a dengue biocontrol method that is environmentally friendly and able to spread unassisted in wild mosquito populations. Release of wMel-infected mosquitoes in north Queensland has indicated that this Wolbachia strain can rapidly reach fixation in wild populations [12]. Key to the utility of this biocontrol method is the maintenance of DENV-blocking following mosquito release and in subsequent generations as Wolbachia invades wild populations.
Our results indicate that, one year post-release, field wMel mosquitoes show significantly reduced DENV infection and replication compared to wildtype mosquitoes. Strikingly, we found very low infection rates in mosquito heads, indicating that DENV is largely unable to disseminate to the heads in wMel mosquitoes, under the experimental conditions used here. By day 14, in both experiments, wMel mosquitoes displayed dramatically reduced infection rates and viral titers in heads compared to wildtype. Reduced DENV dissemination and transmission rates due to the  presence of native Wolbachia endosymbionts have also been found in the vector A. albopictus [27]. The pattern was observed with a range of virus titers and serotypes (DENV-1 to -3), and using both cell-cultured and viremic human plasma. We did not test for systematic differences in response to these variables here, but work with other viruses has indicated the extent of Wolbachia-mediated viral blocking is dependent on virus titer [28].
Our data suggest stability of viral blocking and Wolbachia tissue tropism since divergence of field mosquitoes from the parental wMel-transinfected laboratory line MGYP2.out. We did not find statistically significant differences in either dengue infection rates or virus titers between field wMel and MGYP2.out mosquitoes. However, field wMel mosquitoes may be somewhat better at blocking dissemination of DENV-1 but not DENV-2 and DENV-3 compared to MGYP2.out ( Figure 2). This is because the number of MGYP2.out individuals infected with virus is much higher for DENV-1 than DENV-2 and DENV-3 compared to field mosquitoes. Virus was detected in a higher number of MGYP2.out individuals for DENV-1 strain P307, compared to the other virus strains tested. Additional experiments are needed to determine whether this effect is due to the particular strain or a phenomenon general to the DENV-1 serotype. DENV-2-92T dissemination rates in MGYP2.out were 12.5% several generations after transinfection in earlier work [9] and have stayed a low 7% in our study, at least 10 generations later and with frequent outcrossing of this line (every three generations). This time frame is comparable with that experienced by field mosquitoes, with the maximum number of generations per year in Cairns being 15 and populations persisting throughout the year [29]. MGYP2.out and field wMel-infected mosquitoes have therefore retained the virus blocking phenotype described in [9] that led to the field release of Wolbachia-infected mosquitoes. Our results suggest that the virus blocking phenotype induced by wMel may be retained not just over the short term, but also over the medium to longer term.
Wolbachia tissue tropism was similar in field and laboratory wMel-infected mosquitoes, with high densities of the bacterium found in the midgut and ovaries. Wolbachia was also present in the salivary glands and brains of both mosquito lines, which may contribute to the limited dissemination and replication of DENV observed in heads from the wMel-infected lines. In Drosophila simulans, high Wolbachia densities in head and midgut have been correlated with interference against Drosophila C virus [30].
Wolbachia density is critical in modulating transmission fidelity of the bacterium across generations and pathogenicity [19]. Wolbachia density changes dynamically in response to environmental variables [31]. We also found that Wolbachia density increased following blood-feeding, consistent with other studies that have shown an increase in endosymbiont density in response to high nutrient conditions [32]. Wolbachia provides a fitness benefit by modulating iron levels in D. melanogaster [33] and responds transcriptionally to iron overload [34]. Increased Wolbachia replication is most likely localized to the ovaries, although further work is needed to confirm this. Our results differ, however, from those of [35], who showed a blood-feeding induced reduction in the native endosymbiont wFlu in the ovaries of the mosquito Aedes fluviatilis. Surprisingly, the increase in Wolbachia density was more pronounced in field wMel mosquitoes compared to the laboratory line, although only at day 7 post-infection. The reasons for this difference are unknown but may be related to poor nutrition in the field or other environmental effects. Although mosquitoes were reared in the same environment for one generation, maternal nutritional effects can be detected up to several generations later in insects [36,37]. Maternal effects due to poor nutrition in the field may influence offspring immune status and the ability to control infection levels, potentially resulting in higher Wolbachia densities.
Dynamic changes in Wolbachia density following blood-feeding may have implications for vector competence of wMel-infected mosquitoes. The precise mechanism by which Wolbachia mediates viral blocking is not known but is positively related to density of the bacterium [13,14,38]. If blood-feeding acts to increase Wolbachia density and A. aegypti feed frequently on human hosts, viral blocking may be greater in field populations than anticipated from laboratory experiments, although further studies are needed to test this hypothesis. In laboratory experiments involving Drosophila, the density of Wolbachia has been shown to evolve to a level that is nonpathogenic to the fly but the bacteria are still maintained [19,39]. Understanding selection pressures on wMel-infected mosquitoes in nature will be necessary to predict how Wolbachia may evolve over the long term in field-released mosquitoes.