Wolbachia and dengue virus infection in the mosquito Aedes fluviatilis (Diptera: Culicidae)

Dengue represents a serious threat to human health, with billions of people living at risk of the disease. Wolbachia pipientis is a bacterial endosymbiont common to many insect species. Wolbachia transinfections in mosquito disease vectors have great value for disease control given the bacterium’s ability to spread into wild mosquito populations, and to interfere with infections of pathogens, such as dengue virus. Aedes fluviatilis is a mosquito with a widespread distribution in Latin America, but its status as a dengue vector has not been clarified. Ae. fluviatilis is also naturally infected by the wFlu Wolbachia strain, which has been demonstrated to enhance infection with the avian malarial parasite Plasmodium gallinaceum. We performed experimental infections of Ae. fluviatilis with DENV-2 and DENV-3 isolates from Brazil via injection or oral feeding to provide insight into its competence for the virus. We also examined the effect of the native Wolbachia infection on the virus using a mosquito line where the wFlu infection had been cleared by antibiotic treatment. Through RT-qPCR, we observed that Ae. fluviatilis could become infected with both viruses via either method of infection, although at a lower rate than Aedes aegypti, the primary dengue vector. We then detected DENV-2 and DENV-3 in the saliva of injected mosquitoes, and observed that injection of DENV-3-infected saliva produced subsequent infections in naïve Ae. aegypti. However, across our data we observed no difference in prevalence of infection and viral load between Wolbachia-infected and -uninfected mosquitoes, suggesting that there is no effect of wFlu on dengue virus. Our results highlight that Ae. fluviatilis could potentially serve as a dengue vector under the right circumstances, although further testing is required to determine if this occurs in the field.


Introduction
Dengue virus (DENV) represents a serious threat to human health across the tropical regions of the world. There are four genetically distinct DENV serotypes, and infection can cause dengue, which has fever-like symptoms, or the comparatively rare severe dengue, which can lead a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 to haemorrhaging and death. Subsequent infection with a different serotype greatly increases the risk that severe dengue will develop (WHO). There are an estimated 96 million cases that manifest clinically, each year [1], and nearly 4 billion people living at risk of infection [2], with severe, endemic transmission occurring in Latin America, and South and Southeast Asia.
In the last 40 years there has been an expansion in the incidence and prevalence of dengue cases [3]. This has occurred in conjunction with an increase in the geographic distribution of the mosquito vectors of the disease; the primary vector, Aedes aegypti, and the secondary vector Aedes albopictus [4]. Changes in climate, increased travel to disease-endemic areas, and increased deforestation and urbanization have served to bring humans and mosquitoes into contact more frequently, increasing the chance of viral transmission [1,[5][6][7][8]. Over the same time period there has been an increase in the prevalence of genetic resistance to commonly used insecticidal chemicals in mosquito populations, which has led to a decline in the efficacy of historically successful mosquito control programs [9,10]. Urbanization and deforestation could also lead to a change in the dynamics of a pathogen transmission cycle, by increasing the frequency of contact between humans and other mosquito species that might become infected with the virus in the field [6,11,12].
Aedes (Ochlerotatus) fluviatilis (Lutz, 1904) is an anthropophilic species of mosquito native to Latin America, which lives in a wide variety of habitats, including urban and undeveloped areas [13]. Ae. fluviatilis has been described as being competent for infection with yellow fever virus, and was subsequently hypothesised as a potential vector for the disease in Latin America [14]. It has also been described as a good laboratory model for the avian malaria parasite Plasmodium gallinaceum [15]. However, little is known about the species' ability to maintain infections with or transmit DENV.
Wolbachia pipientis is a maternally transmitted, bacterial endosymbiont that commonly infects many insect species [16]. Wolbachia strains are genetically diverse [17,18], and maintain intricate relationships with their different host species [19,20]. Many of these relationships are characterised by symbiont-induced manipulation of host reproduction, which allow the bacterium to persist in insect populations even in the face of fitness costs, and can permit the bacterium to spread through uninfected populations [21][22][23][24].
Many mosquito species are naturally infected by Wolbachia, including Ae. fluviatilis, which is infected by the wFlu strain [25,26]. However, several key vector species, including most Ae. aegypti populations [27], are naturally uninfected. In these cases, stable infections have been generated by transinfection-the transfer of Wolbachia between species to form a stable infection [28]. Ae. aegypti has been independently transinfected with multiple Wolbachia strains [29][30][31], including the wMel strain, originally found in the fruit fly Drosophila melanogaster [32]. Wolbachia transinfections in mosquitoes are strongly associated with pathogen interference-the ability to interfere with the replication and infection of pathogens such as the dengue and Zika viruses, and the human malarial parasite Plasmodium falciparum in mosquito tissues and saliva [32][33][34][35].
Transinfections in mosquitoes typically have a more severe physiological effect on their host than native infections, likely due to relatively novelty of infection, and subsequent activation of the host immune response [36]. Native infections are capable of altering host biology at the physiological and transcriptional level [25,37], although wFlu has little apparent fitness cost [25]. They can also influence interactions between the host and pathogens, although not in the same way as transinfections. In Ae. fluviatilis, wFlu has been demonstrated to enhance infection with P. gallinaceum, increasing the intensity of infection at the oocyst stage [25]. A similar effect occurs for Culex pipiens and its native Wolbachia, wPip, during infection with Plasmodium relictum [38]. This enhancement effect has also been observed in some instances of transient Wolbachia infection in mosquitoes [39][40][41], where Wolbachia is injected into adult mosquitoes to produce a somatic infection, but not in all cases [42,43]. Enhancement has only been observed rarely, and never with a stable Wolbachia transinfection for a pathogen of humans. Unless this changes, the potential impact on mosquito control programs that involve Wolbachia is likely to be limited [44,45]. However, given the enhancement of P. gallinaceum in wFlu-infected Ae. fluviatilis, we sought to determine if a similar effect occurred for DENV.
We performed a series of experimental infections with 2 different DENV isolates in order to examine the ability of Ae. fluviatilis to harbour infection with the virus. These experiments were performed on mosquitoes with and without the native wFlu infection in order to determine if the native Wolbachia infection of Ae. fluviatilis influenced infection with the virus. We also examined the effect of wFlu on the production of reactive oxygen species (ROS), as this had been linked to pathogen interference in other Wolbachia-infected insects [46,47].

Mosquito rearing
Two lines of Aedes fluviatilis mosquitoes were involved in these experiments-one naturally infected with the wFlu Wolbachia strain (Flu), and the other where this infection had been cleared by treatment with tetracycline (Flu.Tet), as previously described [25]. Experiments were also performed on three lines of Aedes aegypti-one infected with the wMel Wolbachia strain (Mel), and a second where the wMel infection had been removed by treatment with tetracycline (Mel.Tet), as previously described [48]. These lines were previously backcrossed to a Brazilian genetic background [48], and were regularly outcrossed with Wolbachia-uninfected mosquitoes (WT) collected from Rio de Janeiro, RJ, Brazil, thereafter, in order to maintain genetic diversity [49]. All experiments took place at least two years after tetracycline treatments had finished. Saliva injection experiments were performed using the WT line. Microbiota recolonization was performed as previously described [25,49].
Flu and Flu.Tet larvae were reared at a density of approximately 400 larvae in 3L of distilled water and were fed 15-20 balls of Alcon Goldfish Colour fish food per day. Mel and Mel.Tet larvae were reared at a density of 150 larvae in 3L of distilled water, and were fed a ½ tetramin tropical tablet (Tetramin) per day. Pupae were moved to large cages in groups of 500. For experimental infections, adult females were transferred to small, cylindrical cages (diameter-16cm, height-18cm) containing approximately 50-70 individuals. Adults were maintained on 10% sucrose, which was changed daily for virus-infected mosquitoes, and three times per week for other cages. All mosquitoes were reared in a climate-controlled insectary (temperature-27 ± 1˚C, RH -70 ± 10%, photoperiod-12 hours light: dark).
For oral infection experiments, viral stocks were mixed 1:1 with freshly drawn human blood, and fed to mosquitoes via glass feeders on a waterbath system. For the dilution experiment, one tube of frozen DENV-3 was thawed and diluted in L15 media, supplemented with 10% sterile, inactivated foetal bovine serum, to produce additional viral stocks at concentrations of, 1.9x10 4 , 1.9x10 3 and 1.9x10 2 pfu/mL. These were mixed with human blood and fed, as above. Immediately post-feeding, mosquitoes were screened visually for the presence of blood in the abdomen, and non-fed mosquitoes were discarded. Mosquitoes were collected at 7 or 14 days post-infection and stored at -80˚C.

Saliva collection & injection
Saliva was collected from mosquitoes that were infected with either DENV-2 or DENV-3 via injection, at 5 dpi, as previously described [34,35]. Briefly, legs and wings were removed from anesthetized mosquitoes on ice, and mosquitoes were allowed 30 mins to expectorate into pipet tips containing 5μL of sterile, inactivated, foetal bovine serum and 30% sucrose (1:1), prepared fresh on the day of collection. These saliva samples were stored at -80˚C and either assayed directly for the presence of DENV, or injected into naïve, wildtype, female Aedes aegypti mosquitoes, collected near Rio de Janeiro in 2016, as previously described [34]. Ae. aegypti mosquitoes were selected for injections given the propensity of that species to harbour infection with DENV. Each saliva was injected into 12-15 mosquitoes (volume injected: 138-276nL), which were collected at 5 dpi, and stored at -80˚C. The presence or absence of DENV was determined by RT-qPCR for 7-8 injected mosquitoes per saliva sample.

DENV quantification
DENV levels in mosquitoes from the Ae. fluviatilis and Ae. aegypti oral infection experiments, the DENV-3 dilution experiment, and the DENV-2 and DENV-3 injection experiments were quantified using TaqMan-based RT-qPCR, using a previously described protocol, primers and TaqMan probe [49]. Briefly, RNA from whole mosquito samples was extracted using the TRIzol protocol (Invitrogen), first-strand cDNA synthesis was performed using the M-MLV RT protocol (Promega), and qPCR was performed using a DENV general primer set and using a Viia 7 Real-Time PCR System (ThermoFisher Scientific). Absolute DENV levels were obtained by comparison with a serially diluted (10 7 to 10 3 copies) fragment of the PCR product, previously described [49], and were normalized per 1μg of total RNA. A minimum threshold of infection of 100 copies was applied to data across all experiments, as the 10 2 standard did not amplify reliably.
Total RNA from saliva-injected mosquitoes was extracted using the High Pure Viral Nucleic Acid Kit (Roche), according to manufacturer's instructions. The presence or absence of DENV was then determined via a duplex assay for DENV and the host RpS17 gene using the TaqMan Fast Virus 1-Step Master Mix (ThermoFisher Scientific), and run on a Lightcycler 96 (Roche). The qPCR mastermix contained the following components per reaction: 4.25μL water, 2μL RT-qPCR Mix, 0.5μL DENV primers (10μM), 0.1μL DENV probe (10μM-Texas Red), 0.5μL RpS17 primers (10μM), 0.1μL RpS17 probe (10μM-Hex), 0.05μL 200 X Reverse transcriptase enzyme. The water, RT-qPCR Mix, and enzyme were provided in the Fast Virus kit. The run profile was as follows: (1) 50˚C for 10 mins, (2) 95˚C for 30  Three types of samples were collected across two experimental replicates: (1) whole mosquito, (2) midgut, (3) fat body, with the later two obtained by dissection using a light microscope in ice cold PBS. Total samples collected ranged from 12-22 per treatment. Samples were collected on ice, and then immediately processed according to manufacturer's instructions, as described elsewhere [50]. H 2 O 2 levels were quantified using a microplate reader (SpectraMax M5, Molecular Devices, Sunnyvale, CA, USA), with an excitation wavelength of 530nm and an emission wavelength of 590nm, and values compared against a standard curve generated from H 2 O 2 samples of known concentration. Mosquitoes in these assays were not infected by DENV.

Data analysis
DENV prevalence of infection was compared using Fisher's exact test. For saliva-injected mosquitoes, prevalence of infection was compared on the individual mosquito level. Comparisons of viral load were only made for samples that tested positive for DENV. These data were all non-parametrically distributed, and were thus compared using Mann-Whitney U tests. No viral load comparisons were made for treatments with fewer than 3 positive samples, in order to satisfy the requirements of the Mann-Whitney U test. For these pairwise comparisons, only statistics for prevalence are presented. H 2 O 2 data were all normally distributed, and were compared using unpaired t tests. All analyses were performed using Graphpad Prism V 6.0g.

DENV injections
We injected adult female Ae. fluviatilis and Ae. aegypti with either DENV-2 or DENV-3, collected these mosquitoes 5 days later, and then quantified levels of DENV using RT-qPCR. For Ae. fluviatilis mosquitoes injected with DENV-2 (Fig 1A), we observed that 15/22 Flu mosquitoes, and 18/22 Flu.Tet mosquitoes were positive for the virus, and that there was no effect of wFlu infection on prevalence of infection (Fisher's exact test; P = 0.4876). Likewise, we saw no difference in viral load between Flu and Flu.Tet mosquitoes (Mann-Whitney U test; U = 89, P = 0.5045). For Ae. aegypti, we observed a decrease in DENV-2 prevalence associated with wMel infection (Fig 1B), as 8/20 Mel mosquitoes were positive for the virus compared with 12/ 14 Mel.Tet mosquitoes (Fisher's exact test; P = 0.0128). Likewise, DENV-2 viral load was also decreased for Mel mosquitoes (Mann-Whitney U test; U = 16, P = 0.0283).

DENV oral infections
We then orally infected Flu and Flu.Tet mosquitoes with either DENV-2 or DENV-3. Infection with the DENV-2 isolate produced only a very small number of positives from either line. Accordingly, we examined data from both experimental infections together. At 7dpi (Fig 2A), we saw that 4/40 Flu and 2/42 Flu.Tet mosquitoes became infected, but this did not amount to  https://doi.org/10.1371/journal.pone.0181678.g002 a significant difference in prevalence (Fisher's exact test; P = 0.6763). At 14dpi (Fig 2B), 1/32 Flu mosquitoes, and 4/39 Flu.Tet mosquitoes became infected, and again these proportions were not significantly different (Fisher's exact test; P = 0.3806). Numbers of infected mosquitoes were not sufficient to compare viral load.
Oral infection with the DENV-3 isolate produced a slightly higher rate of infection. Across the two experiments, we saw prevalence of infection of 15/29 for Flu mosquitoes and 13/33 for Flu.Tet mosquitoes at 7dpi (Fig 2C; Fisher's exact test; P = 0.6503), and 11/28 for Flu and 13/ 25 for Flu.Tet mosquitoes at 14dpi (Fig 2D; Fisher's exact test; P = 0.6280). Critically, there was no difference in prevalence due to Wolbachia. Likewise, we saw no difference in viral load between the two mosquito lines at either time point (Mann-Whitney U test; 7dpi-U = 71, P = 0.2352: 14dpi-U = 66, P = 0.7648).
We also examined DENV infection in Ae. aegypti, as a point of comparison. As the DENV-2 isolate did not infect Ae. fluviatilis well, we only performed experiments with the DENV-3 isolate. At 7dpi in the first experiment (Fig 3A), 27/29 Mel.Tet became infected compared to 3/30 Mel mosquitoes (Fisher's exact test; P < 0.0001). Those Mel mosquitoes that became infected had a significantly lower viral load than Mel.Tet mosquitoes (Mann-Whitney U test; U = 4, P = 0.0054). At 14dpi (Fig 3B), 28/30 Mel.Tet were infected compared to 6/30 Mel mosquitoes (Fisher's exact test; P <0.0001), and viral load was again significantly lower in the infected Mel mosquitoes (Mann-Whitney U test; U = 14, P = 0.0006).

Salivation assays
We quantified levels of DENV directly in saliva collected from Flu and Flu.Tet mosquitoes that had been injected with either the DENV-2 (Fig 4A) or DENV-3 isolates (Fig 4B).  We then took 5 saliva samples collected from Flu (Fig 5A), Flu.Tet (Fig 5B), Mel (Fig 5C), and Mel.Tet (Fig 5D) mosquitoes that had been injected with DENV-3, and injected these into WT mosquitoes. For saliva from Ae. fluviatilis, 4/5 Flu and 4/5 Flu.Tet saliva produced subsequent infections, at an average infection rate of 65% and 45%, respectively. We saw no significant difference in the overall prevalence of saliva-injected mosquitoes (Fisher's exact test; P = 0.1716). For saliva from Ae. aegypti, 2/5 Mel and 4/5 Mel.Tet saliva produced subsequent infections, at average infection rates of 20% and 47.5%, respectively. In this instance there was a decrease in overall prevalence associated with wMel infection in the mosquitoes that produced the saliva (Fisher's exact test; P = 0.0093).

Reactive oxygen species
We compared levels of ROS (H 2 O 2 ) in Flu and Flu.Tet mosquitoes for the whole mosquito, midgut or fat bodies (Fig 6). We saw that H 2 O 2 levels were not significantly different in whole mosquitoes (Unpaired t test; t = 0.8635, P = 0.3972) or in the fat body (Unpaired t test; t = 0.0514, P = 0.9594), however, H 2 O 2 levels were slightly, but significantly elevated in the midguts of Flu mosquitoes compared to Flu.Tet (Unpaired t test; t = 2.739, P = 0.0090). Mean (± s.e.m.) H 2 O 2 levels were 0.7563 ± 0.027 for Flu mosquitoes, and 0.6222 ± 0.041 for Flu.Tet mosquitoes.

Discussion
Our results from the experimental infection of Ae. aegypti with DENV-3 showed high prevalence of infection and DENV load in Mel.Tet mosquitoes, and a strong pathogen interference effect in Mel mosquitoes. This persisted even when the mode of infection was via injection. Several recent papers have demonstrated that wMel interferes with Zika virus and chikungunya virus in Ae. aegypti populations from Brazil and Colombia [34,51,52]. Our results provide further evidence that the pathogen interference phenotype operates effectively in Latin American Ae. aegypti. These results also suggested that the DENV-3 isolate that we used was sufficiently infectious to serve as a good model strain for examining DENV infection in Ae. fluviatilis. It should be noted that the our results were obtained using whole mosquito samples, and thus likely underreport the blocking effects of wMel, as this strain is a strong blocker of disseminated and salivary infection [34].
We performed experimental infections of Ae. fluviatilis mosquitoes with DENV-2 and DENV-3 isolates via oral feeding and intrathoracic injection. With the oral feeding experiments we observed very low infection rates for the DENV-2 isolate (3-10%), with this likely being due to a low viral titre (2.0x10 4 pfu/mL). For the DENV-3 isolate, we observed combined infection rates of 27-44% at 7dpi, and 35-37% at 14dpi for both Flu and Flu.Tet mosquitoes, across 3 experiments. These infection rates were far lower than what we observed when the Mel.Tet line (Ae. aegypti without a Wolbachia infection) was infected with the same virus, where overall infection rates were 78% at 7dpi and 71% at 14dpi, across 2 experiments. We observed higher rates of infection after Ae. fluviatilis mosquitoes were infected by intrathoracic injection, with an overall infection rate of 68-82% for DENV-2, and 73-76% for DENV-3. These values were more similar to the infection rates we observed for the Mel.Tet line (DENV-2: 86%, DENV-3: 89%). These data suggest that Ae. fluviatilis can become infected by DENV after experimental infection in the laboratory. As the titre of the DENV isolates used in our experiments was somewhat low, it is possible that Ae. fluviatilis may prove to be more susceptible to infection with different DENV isolates, or those fed at a higher titre.
We also examined the saliva of Ae. fluviatilis mosquitoes for the presence of DENV using two different techniques, after the mosquitoes had been infected with either DENV-2 or DENV-3 by injection. When viral load was quantified directly in the saliva, we observed that 41-42% of saliva tested positive for DENV-2, and 17-25% were positive for DENV-3, for Flu and Flu.Tet mosquitoes. When these saliva were injected into WT Ae. aegypti we were also able to detect DENV infection in a high percentage of mosquitoes. As virus load in these mosquitoes was quantified 5 days post-injection, this may indicate that the virus was able to replicate, suggesting that infectious vius can be found in the saliva of Ae. fluviatilis.
Our results provide some evidence that Ae. fluviatilis could potentially transmit DENV under the right circumstances. However, we state this with the acknowledgment of several caveats. Firstly, all of our data showing DENV infection in saliva were obtained after intrathoracic injection and not oral feeding. Given the lower prevalence of infection associated with oral feeding this may suggest that that the midgut is an important barrier to DENV infection in Ae. fluviatilis, and it is possible that this species does not possess some of the receptors for DENV that are present in other mosquito species [53]. Secondly, the Ae. fluviatilis lines we used are not necessarily representative of the species in the field, given that the colony has been maintained in the laboratory since 1975 [25]. It is also unclear how likely it is that Ae. fluviatilis in the field would encounter and bite a human infected with dengue, or if a DENVinfected Ae. fluviatilis would be capable of transmitting the virus to a new human host. To that end, we suggest that future research on this topic should involve a survey of wild Ae. fluviatilis populations for the presence of DENV, and that saliva be collected from further experimental infections where field populations of Ae. fluviatilis are orally infected with fresh, recently circulating DENV strains.

Influence of Wolbachia
With respect to a potential effect of wFlu on DENV infection, we observed high infection rates in Flu mosquitoes across multiple experiments, for both modes of infection. However, we saw similar results for Flu.Tet mosquitoes. In one oral feeding experiment there was increased prevalence of infection associated with wFlu, and in one injection experiment, there was decreased viral load for DENV-3 associated with wFlu. For all other experiments we saw no effect of wFlu infection on prevalence or intensity of infection. At the saliva level, there was slightly higher prevalence of infection for mosquitoes injected with saliva from DENV-infected Flu mosquitoes than Flu.Tet mosquitoes, although this difference was not significant. Taken together these results provide no evidence of a consistent effect of wFlu on DENV, which stands in contrast to what was previously observed for P. gallinaceum [25].
The ability of a Wolbachia strain to interfere with pathogens has been strongly linked to high bacterial density, and high Wolbachia density in mosquitoes is typically associated with transinfections, rather than native associations [29,32]. In Ae. fluviatilis, wFlu is most abundant in the ovaries, which facilitates maternal transmission, but it is much less abundant in somatic tissues than transinfections, including that of wMel [25,26,32]. Given these observations it was not surprising that we observed no pathogen interference effect of wFlu.

ROS induction
ROS induction (higher levels of H 2 O 2 in the presence of Wolbachia infection) in mosquitoes is a trait that is most commonly associated with Wolbachia transinfections [19,46]. We recently demonstrated that this induction effect occurs for wMel-infected Ae. aegypti [49], while a similar effect has been observed for wAlbB-infected Ae. aegypti [46]. However, the effect has not been associated with native infections in mosquitoes. The reason for this is hypothesized to be due to a longer period of host-symbiont co-adaptation, and potentially restored redox homeostasis in native infections [54]. Our data suggest that wFlu does cause a moderate level of ROS induction in Ae. fluviatilis, but only in the midgut. Our recent transcriptomic profile of Ae. fluviatilis revealed wFlu-induced changes in the expression of several genes linked to redox process [37]. However, the fact that we observed no difference in whole mosquitoes, or in fat bodies, suggests that wFlu is unlikely to have a major impact on redox homeostasis, which could be indicative of extensive co-evolution between host and symbiont [54]. While the small scale ROS induction that we observed in the midguts was likely insufficient to have a significant effect on a pathogen. ROS induction has also been shown to occur with some native Wolbachia infections in Drosophila, and its presence correlates well with the presence of pathogen interference [47]. But, higher ROS levels are not found in all Wolbachia-host associations where pathogen interference has been observed [55], and it is still unclear how Wolbachia-induced changes in ROS contribute to the phenotype. Not all mosquito midgut cells become infected with DENV, and there are foci of infection [56,57]. It is unclear if the Wolbachia-induced increase in ROS is homogeneously or heterogeneously distributed through tissues and cells. If the latter were true, cells with higher ROS levels could become more resistant to the virus, leaving fewer "entry points" available.

Conclusions
We have demonstrated that the mosquito Ae. fluviatilis can become infected with DENV after intrathoracic injection or oral feeding, although at lower rates than what is observed with Ae. aegypti, a proven vector of DENV in the field. We observed that DENV can be detected in the saliva of Ae. fluviatilis after viral challenge via injection, and that this virus is likely infectious, given that it can be used to infect naïve Ae. aegypti. The vector status of this species is still uncertain, but our results suggest that it could potentially play a role in DENV transmission. Our results also indicate that the native Wolbachia infection of Ae. fluviatilis does not influence DENV infection, but does increase ROS levels in the host midgut, with neither of these observations being particularly unexpected for a native Wolbachia-host association.  at 7 (A) and 14dpi (B), and Ae. aegypti at 7 (C) and 14dpi (D), as determined via RT-qPCR with absolute quantification. The figure shows data for two DENV-3 concentrations that were fed to mosquitoes: 1.9 x10 6 pfu/ mL (1), and 1.9 x 10 4 pfu/mL (2). FT (orange)-Flu.Tet. F (black)-Flu. MT (red)-Mel.Tet. M (blue)-Mel. Prevalence data analysed by Fisher's exact test. DENV load data analysed by Mann-Whitney U test. ns = P > 0.05, Ã = P < 0.05, ÃÃÃÃ = P < 0.0001. Black lines represent treatment medians. (TIFF)