Our studies identified, total 13 bacterial isolates from the midguts of field collected and laboratory reared Ae. aegypti by traditional biochemical methods and 16s rDNA sequence analysis (Table 1). Based on the 16s rDNA sequence analysis, the isolated midgut microbes were classified into three major groups, viz., gamma-proteobacteria, beta-proteobacteria and fermicutes. Biochemical characterization based identification of the isolates corroborated with their 16s rDNA sequence based identification. The sequences obtained in this work are deposited in Genbank database (Table 1).

A large proportion of the isolates (64%) were identified as Gamma-proteobacteria, where dominant genera were Aeromonas, Enterobacter and Pseudomonas. For the mosquitoes reared in three different laboratories, we observed existence of ten different microbes in their midguts with S. odorifera biogroup II and Microbacterium oxydans as common gut inhabitants (Table 1). Additionally, in the field collected mosquitoes, 11 distinct isolates were identified with a dominance of Aeromonas media, Serratia odorifera biogroup II and Pseudomonas alcaligenes (Table 1). S. odorifera was the common gut microbe of all the mosquitoes analyzed in this study.

To know the source of these gut inhabitants, the bacterial consortium of the respective larval habitat was also analyzed. Our observations suggested that the larval surroundings influenced the type of gut microbes they harbored (Table 1). Conversely, it was also observed that the microbes like Xenorhybdus luminiscens and Escherichia coli; although present in abundance in the larval habitats, did not colonize their guts suggesting selective retention of specific microbes as midgut inhabitants (Table 1).

The gut microbial community was monitored during metamorphosis of mosquitoes to investigate the dynamics of this community during the transition from larvae to adult. S. odorifera was the only microbe commonly associated in the midguts of pupae, emerging imagoes and adults of all field collected and laboratory reared Ae. aegypti (Table 2). These observations suggested that S. odorifera was transstadially transmitted from larvae to adult.

During our study, it was observed that the status of the midgut flora changed significantly during the developmental stages of this vector as it transits from aquatic to terrestrial life. It was therefore important to understand if the higher ambient temperatures can affect the dynamics of gut microbiota in larval stage which leads mainly aquatic life and is predominantly exposed to severe temperatures than other stages. The ambient temperature in summer season in Pune and many other cities in India reaches up to 41°C or more, hence, to simulate this condition in the laboratory, a heat shock exposure (41°C) was given to fourth instar larvae and the survivors in the gut were analyzed. Consequently, the larvae were also collected from the natural habitats during summer season when ambient temperature was 39°C.

There was a significant reduction of about two logs in CFU of gut microbes after high temperature exposure of the fourth instar larvae. The average CFU of the gut microbes after heat shock exposure of larvae collected from AFMC showed a reduction from 2×10⁴ to 5×10² while it decreased to 4.8×10² from initial count of 3×10⁴ in case of the larvae reared in MCC. Incidentally, the percent distribution of S. odorifera in the midgut of these larvae was increased, probably due to the elimination of the most of the other gut flora after high temperature exposure. The observation of the natural habitat when ambient temperature was 39°C was analogous to the observations of heat shock treatment where the total bacterial population in the larval gut was reduced along with elevated relative presence of S. odorifera biogroup II.

Transstadial transmission and selective retention at high temperature qualified S. odorifera to be specially considered for the further studies. Understanding its interaction either with the midgut tissue of the adult Ae. aegypti or with the arbovirus, or both and its subsequent influence on the vector competence was the focus of further investigations.

Studies were designed to understand the effect of S. odorifera’s presence in the midgut of Ae. aegypti on its susceptibility to DENV-2. For an individual experiment, Ae. aegypti females were split into three groups based on the type of blood meal they received. Group one received DENV-2 with no bacteria in the blood meal whereas, groups two and three were co-fed with S. odorifera and M. oxydans respectively, along with DENV-2 via blood meal. The data obtained in three independent experiments showed similar pattern.

On the 14^th post infection day (PID), the DENV-2 infection in the infected females was demonstrated by immunofluorescence assay and DENV-2 titer was determined by plaque assay. Immunofluorescence assay revealed that at 14 PID, the group receiving S. odorifera along with DENV-2 showed significant increase in the susceptibility to DENV-2 (39.4%±8.4; data of four independent experiments) as compared to the groups that received only dengue-2 (21.6%±1.2; data of three independent experiments) and M. oxydans (18.9%±4.4; data of three independent experiments). An ANOVA-one way analysis showed that the means were significantly different for the dissemination rates among the three groups (P = 0.0016). The dissemination rates between dengue-2 and dengue-2+ S. odorifera groups (Mann-Whitney U test P<0.05) as well as dengue-2+ M. oxydans and dengue-2+ S. odorifera groups (Mann-Whitney U test P<0.05) were significantly different. Whereas the dissemination rate did not vary significantly between dengue-2 and dengue-2+ M. oxydans groups (Mann-Whitney U test P>0.05).

It was important to analyze the DENV-2 titers in all the groups immidiately after blood meal and on 14^th PID. The titers were determined in individual mosquito carcasses. Immediately after blood meal, Ae. aegypti females had an average dengue-2 titer of 8.83×10²±1.92×10² PFU/mosquito for the first group and 8.61×10²±1.56×10² PFU/mosquito and 7.62 ×10²±2.77×10² PFU/mosquito for second and third groups respectively, indicating almost equal number of viruses were ingested by all the groups (Mann-Whitney U test P>0.5). At 14 PID, no significant difference in the virus titers was observed amongst three groups as the titer for the first group was 7.13×10⁴±9.97×10³ PFU/mosquito whereas for second and third groups it retained at 6.86×10⁴±1.88×10⁴ PFU/mosquito and 6.51×10⁴±1.83 ×10⁴ PFU/mosquito (Mann-Whitney U test P>0.5) respectively. Virus titers in dengue-2 negative mosquitoes were undetectable. These observations suggested that the presence of S. odorifera influenced the susceptibility of Ae. aegypti to DENV-2 specifically as more number of mosquitoes were affected with DENV-2 in S. odorifera fed group.

This led to the speculation that S. odorifera modulated Ae. aegypti susceptibility either through its direct interaction with virus or virus-specific gut receptor component or both. Therefore, it was essential to identify the repertoire of polypeptides of either bacteria or mosquito, interacting with the virus to understand the causal mechanism of enhanced vector susceptibility.

Overlay assays were performed to investigate the interaction between S. odorifera and Ae. aegypti midgut BBMF. S. odorifera and M. oxydans cell lysates were transferred to hybond C membrane and overlaid with Ae. aegypti BBMF. Overlay assays revealed that 40 kDa polypeptide (P40) of S. odorifera interacted specifically with BBMF (Fig. 1b). However, none of the polypeptides of M. oxydans showed interaction with BBMF although it is a cohabitant with S. odorifera in most of the larval/adult midguts (Fig. 1b). P40 was also detected in the culture supernatant of S. odorifera (Fig. 1c). The expression of P40 in culture filtrate and cell lysate of S. odorifera was significantly enhanced after heat shock treatment (Fig. 1c). The peptide mass fingerprint of P40 by MALDI-TOF/TOF showed significant homology to the putative periplasmic protein of Salmonella typhi and exported protein of Serratia marcescens. Domain analysis of 65 amino acids showed the presence of N- myristoylation site, N-glycosylation site and casein kinase II phosphorylation site. Glycoprotein staining of S. odorifera cell extract revealed that P40 was a glycoprotein.

When SDS-PAGE separated polypeptides from Ae. aegypti BBMF were incubated with S. odorifera cell lysate and the interaction with P40 was detected by anti P40 antibody, two polypeptides of the midgut BBMF with molecular masses 32 kDa and 38 kDa were recognized as P40 binding proteins (Fig. 1d). The polypeptides interacting with P40 were further characterized using MALDI-TOF/TOF analysis. They were identified as prohibitin (ABF18314) and porin (ABF18270) respectively (For details see Table S2).

Earlier experiments using BBMF revealed that P40 interacts with midgut extract of Ae. aegypti; it was therefore essential to demonstrate that these observations also hold true for dissected midgut tissues. Immunofluorescence microscopy confirmed the P40 interactions predominantly with columnar epithelial cells in the posterior region of the dissected midgut tissue (Fig. 2a, 2b and 2c).

Virus overlay protein binding assays (VOPBA) were performed to investigate the repertoire of S. odorifera polypeptides interacting with DENV-2. When immobilized S. odorifera cell lysate was incubated with native DENV-2, three polypeptides of molecular masses 19 kDa, 29 kDa and 36 kDa of S. odorifera were recognized as DENV-2 binding proteins (Fig. 3a). However, P40 was not involved in this interaction. Putative identification of these proteins was carried out using MALDI-TOF/TOF analysis. A 36 kDa protein showed homology to DNA directed RNA polymerase subunit alpha of Enterobacter sp. (Strain 638) whereas 29 kDa polypeptide showed homology to Ribosomal RNA large subunit methyl transferase of Xanthomonas oryzae pv. Oryzae (For details see Table S3).

The third and fourth instar larvae of Ae. aegypti used in this study were obtained from three laboratory reared colonies and two natural habitats. The laboratories maintaining mosquito colonies were: 1. Microbial Containment Complex (MCC), NIV, Pune; 2. National Chemical Laboratories (NCL), Pune; 3. Entomology group at the Armed Force Medical College (AFMC), Pune, whereas, the household containers in the city of Pune and Ahemedabad were natural habitats. The fourth instar larvae collected from the fields were brought to the laboratory and their identity was confirmed by morphological analysis to be Ae. aegypti and were then used in the study. The larvae were surface sterilized for five seconds in 70% ethanol. The larval guts were dissected aseptically in laminar hood. The dissected midguts were pooled and transferred to 100 µl of sterile phosphate-buffered solution (PBS pH 7.4) and were homogenized. The homogenates of the midguts were serially diluted and were plated on the nutrient agar plates. All the plates were incubated at 30°C for 24 h. The midguts of the pupae and imagoes were also processed similarly for the isolation of gut microbes. Each pool of midguts had five guts of either larvae, pupae or imagoes.

A single colony isolation technique was used for all the morphologically distinct colonies visible on the nutrient agar plate. The bacterial identification was done by the conventional morphological and biochemical characterization, following the protocol in Bergy’s manual of determinative bacteriology [41]. These identifications were further confirmed by 16S rDNA sequence comparisons [42].

The higher ambient temperature during summer season, raises the temperature of the water body which is the natural habitat of the mosquito larvae. To understand the status of larval midgut inhabitants during this season, larval collection was made from water bodies in the city of Pune when the temperatures reached the range of 39–42°C. To simulate the similar effect in the laboratory, the mosquito larvae were exposed to high temperature (42°C) as described by Patil et al., [44]. At the end of exposure time, the larvae were dissected to isolate the midguts and the isolation and identification of the gut flora was carried out as mentioned in the earlier sections.

The aim of this exercise was to understand the status of gut microflora during larval metamorphosis into an adult. For this study, fourth instar larvae and pupae from other laboratories and fields were brought to our laboratory and were maintained under standard laboratory conditions in clean water surroundings and with a change of water every 24 hours till the larvae metamorphosed into adults. The developing pupae were transferred to new containers and emerging imagoes were not offered any food. Their midguts were dissected aseptically within 24 hours of emergence for the isolation and identification of bacterial flora and prior to dissection; the imagoes were surface sterilized with 70% alcohol to avoid any microbial carry over from their surroundings.

Dengue-2, Trinidad (TR1751) virus stock was prepared in infant Swiss albino mice (1–2 days old) as per the protocol described by Ilkal et al., [45]. In order to prepare the virus stock, DENV-2 suspension was initially reconstituted in 0.5 mL distilled water, and the dilutions were prepared in 1.25% bovine serum albumin phosphate saline (BAPS), pH 7.4. One to two day old mice were inoculated intra-cerebrally with 0.02 mL of the DENV-2 suspension (8×10⁷ PFU/mL). The mice were monitored for sickness. After about 8–10 days, when the mice showed dengue-2 sickness, they were sacrificed, and their brains were homogenized using 1.25% BAPS (20% w/v). The homogenates were centrifuged at 16,000 g for 60 minutes. The supernatants were distributed in glass vials, lyophilized and stored at −20°C. The virus titer (5×10⁷ PFU/mL) of one of the vials was determined by plaque assay.

The BBMFs from the midgut epithelial cells of Ae. aegypti larvae and adults were prepared as described by Mourya et al., [46]. The adult females were chilled on ice, the midguts were extruded by dissection under a binocular microscope, peritrophic membranes and malpighian tubules were removed, and the midguts were rinsed in ice-cold buffer A (0.3 M mannitol/5 mM EGTA, 20 mM Tris/HCl, 2 mM PMSF, pH 7.4). The midguts were then placed in 1.5-mL cryogenic vials and stored at −80°C until required. For the preparation of the BBMF, the midguts were homogenized in buffer B (0.3 M mannitol, 5 mM EGTA, 20 mM Tris/HCl, 2 mM PMSF, Triton X-100, pH 7.4). The samples were allowed to stand on ice for 20 minutes and then centrifuged at 2,000 g for 15 minutes at 4°C. The supernatants were collected and kept on ice. The pellets were re-extracted in buffer B. The supernatants from both the extractions were centrifuged at 23,200 g for 60 minutes at 4°C. The pellets were resuspended in 1× PBS with PMSF (2 mM) such that the final dilution is 1 midgut/µl.

After electrophoretic separation of S. odorifera cell lysate by SDS-PAGE, the gel piece corresponding to 40 kDa polypeptide was excised from the gel and washed with normal saline. The gel piece was then homogenized in 1 mL of normal saline for immunization, centrifuged and purity of the 40 kDa protein was reconfirmed by SDS-PAGE before injecting it into mice (Fig. S1).

Three-four weeks old mice were inoculated intra-peritonealy with dengue-2/BBMF/P40 and were maintained under standard laboratory conditions. Booster doses of the antigen along with Freund’s incomplete adjuvant (1∶1) were given (one dose/week) for two weeks. The mice with higher titers of the antibodies were injected with 10% ascitic tumor cells, intra-peritonealy. The intra-peritoneal fluid was collected on the 5^th PID and after removal of the debris by centrifugation, the supernatant was tested for the presence of antibodies. Anti DENV-2 and anti P40 serum were adsorbed with the mosquito midgut extract. Anti BBMF serum was adsorbed with Serratia cell lysate. The antibodies were then purified using protein A column. The pre-immune serum was collected one day prior to the immunizations.

The presence of the viral antigen was determined by the IFA test on the14^th post infection day (PID) by modifying the procedure described by Mourya and Mishra, [48]. The head squashes were prepared on glass slides. The slides were immersed in blocking buffer (0.1% Tween 20 and 2% BSA in PBS) for 1 hour at room temperature; the slides were then incubated with anti DENV-2 mouse antibodies, followed by FITC-conjugated goat anti-mouse IgG for 1 hour each at 37°C. These slides were mounted with ProLong Gold anti-fade reagent with Evans Blue, and were visualized under a fluorescent microscope. For each experiment, positive and negative controls were processed using the same protocol. These experiments were repeated three times and the statistical significance was determined by calculating ANOVA with Tukey’s HSD and further confirmed the statistical significance by analyzing the data with non-parametric test Mann Whitney U test.

Pure cultures of Serratia odorifera and Microbacterium oxydans were grown in the nutrient broth at 37°C with constant shaking for 6 hours. The cells were centrifuged and resuspended in phosphate buffer (pH 7) and were lysed by sonication. The whole cell lysates were used for overlay assays.

To understand the protein-protein interactions among virus or BBMF or S. odorifera cell lysates, overlay assays were performed. The protocol mentioned in this section is common for all the overlay assays. The brush border membrane proteins of Ae. aegypti or S. odorifera cell lysate (50 µg) were separated on SDS-polyacrylamide gel (SDS-PAGE) and were transferred to the nitrocellulose membrane Hybond C using a semi-dry blotting apparatus (Biorad Laboratories, USA) in 48 mmol Tris, 39 mmol glycine, and 20% (vol/vol) methanol. The membranes were blocked with 2% BSA (SIGMA) in PBST (phosphate buffered saline pH 7.4, 0.5% Tween 20) at 4°C overnight, and were washed three times 30 minutes each with PBST following incubation. Based on the type of study, the membranes were either incubated with the purified DENV-2 or BBMF or S. odorifera cell lysate in PBS at 37°C for 1 hour and were washed three times for 30 minutes each with PBST. The membranes were then incubated for one hour with respective antibodies (anti DENV-2 or anti BBMF or anti P40). After washing the membranes three times for 30 minutes each with PBST, the membranes were incubated for 1 hour at room temperature in secondary antibody conjugated with peroxidase/alkaline phosphatase (diluted 1∶3000 in PBS). Finally, the membranes were washed three times 30 minutes each with PBST and were developed with H₂O₂ and DABT or NBT-BCIP solution depending on the secondary antibody conjugate.

The spots corresponding to the protein of interest were excised from SDS-PAGE gels and subjected to alkylation followed by in-gel digestion with trypsin. The masses of resultant peptides were analyzed by matrix-assisted laser desorption/ionization time of flight (MALDI-TOF/TOF) on Ultraflex TOF/TOF (Bruker Daltonics, Germany). MALDI TOF/TOF analysis was carried out in Indian Institute of Science, Bangalore. The mass spectrum generated by each sample was searched against protein databases (NCBInr, MSDB, and Swissprot) using the MASCOT search engine, and the proteins were identified based on homology.

The midgut epithelium of Ae. aegypti adult female mosquito was exposed by dissecting the gut longitudinally, and the tissue was fixed in 4% formaldehyde. The fixative was removed by thorough washing and was incubated with S. odorifera cell extract for 30 minutes. After washing, the tissues were covered with mouse anti P40 antibody (1∶200) followed by incubation with cy3-conjugated rabbit anti-mouse (1∶500) antibody. The nuclei were stained with DAPI followed by washing. The midgut tissues were then mounted in 50% glycerol and visualized by fluorescence microscope (Zeiss Axioskop equipped with camera and AxioVision software). Same exposure time was used for the control and the experimental slides while capturing the images. The negative control was also processed using the same protocol as above, where the gut tissues were overlaid with PBS pH 7.4.

Analysis of variance (ANOVA) was used to evaluate the effect of presence of different midgut bacteria on DENV-2 dissemination rate. The groups were also compared by nonparametric Mann-Whitney U test for confirmation of the results. The viral loads were log-transformed for improvement of the normality and compared by non-parametric Mann-Whitney U test with the significance levels set at p<0.05.