Suppression of the pelo protein by Wolbachia and its effect on dengue virus in Aedes aegypti

The endosymbiont Wolbachia is known to block replication of several important arboviruses, including dengue virus (DENV), in the mosquito vector Aedes aegypti. So far, the exact mechanism of this viral inhibition is not fully understood. A recent study in Drosophila melanogaster has demonstrated an interaction between the pelo gene and Drosophila C virus. In this study, we explored the possible involvement of the pelo protein, that is involved in protein translation, in Wolbachia-mediated antiviral response and mosquito-DENV interaction. We found that pelo is upregulated during DENV replication and its silencing leads to reduced DENV virion production suggesting that it facilities DENV replication. However, in the presence of Wolbachia, specifically in female mosquitoes, the pelo protein is downregulated and its subcellular localization is altered, which could contribute to reduction in DENV replication in Ae. aegypti. In addition, we show that the microRNA aae-miR-2940-5p, whose abundance is highly enriched in Wolbachia-infected mosquitoes, might mediate regulation of pelo. Our data reveals identification of pelo as a host factor that is positively involved in DENV replication, and its suppression in the presence of Wolbachia may contribute to virus blocking exhibited by the endosymbiont.


Introduction
Dengue virus (DENV) is one of the major medically important arboviruses [1,2]. It belongs to the Flaviviridae family that comprises lipid-enveloped, positive-sense single stranded RNA viruses [3]. DENV is classified into four antigenically distinct but closely related serotypes, designated as DENV-1 to DENV-4. The bite of an infected female of either Aedes aegypti or Aedes albopictus is the common mode of DENV transmission to humans [4]. Infected humans may suffer from dengue fever (DF), dengue shock syndrome (DSS) or dengue haemorrhagic fever (DHF), leading to fatality [5,6]. Currently, there is no specific therapy or effective vaccine available and the treatments available are palliative in nature. Despite substantial efforts to control DENV through vector control, it is still geographically expanding and alternative vector control strategies and therapeutic options are urgently needed [7]. One such strategy involves the use of a bacterial endosymbiont Wolbachia in transinfected Ae. aegypti mosquitoes which limits DENV replication [8,9].
Wolbachia is an endosymbiotic, vertically transmitted bacterium that appears to have infected more than 40% of insect species in addition to other terrestrial arthropods [10]. However, it is not a natural symbiont of Ae. aegypti, which is the primary vector of DENV. Recently, several different types of Wolbachia strains have been successfully introduced into Ae. aegypti mosquitoes, among which Wolbachia wMel-Pop-CLA (Pop) and wMel strains are the most promising ones [11,12]. Both strains produce cytoplasmic incompatibility (CI), which facilitates replacement of the wild populations through this reproductive manipulation [13,14]. Furthermore, introduction of Wolbachia has induced viral resistance in Ae. aegypti mosquitoes to a variety of arboviruses including dengue, Zika, West Nile and chikungunya viruses [11,[15][16][17]. However, the exact mechanism which causes this antiviral effect is not fully understood. Despite a small number of studies that have elucidated the potential role of microRNAs (miR-NAs) [18,19], reactive oxygen species (ROS) [20,21], and competition for resources [22] in Wolbachia-mediated antiviral response, the fundamental molecular mechanism(s) underlying virus blocking are yet to be explored. One of the possible strategies to discover molecular interactions between DENV-Wolbachia-Ae. aegypti is to find host factors that facilitate replication of arboviruses and examine their relative abundance in the presence of Wolbachia infection. A recent study in Drosophila melanogaster has identified the protein pelo as an important host factor facilitating efficient replication of Drosophila C virus (DCV) by improving access of the viral genome to ribosomes leading to increased synthesis of viral structural proteins [23].
Pelo is an evolutionary conserved protein which plays an important role in the regulation of D. melanogaster germ cell meiosis. In the pelo mutant males of D. melanogaster, cell cycle in spermatogenesis is arrested at the late prophase stage [24,25]. In contrast, in mutant female flies, mitotic division during oogenesis is affected along with impaired development of the eyes [26]. Deletion of pelo's homolog, (Dom34, Duplication Of Multilocus Region) in yeast Saccharomyces cerevisiae led to the build-up of free ribosomes and decline in the number of polysomes suggesting its involvement in translational regulation [27].
In this study, we aimed to characterize the expression of Ae. aegypti's pelo protein in response to Wolbachia and DENV infection. We found that the pelo protein facilitates DENV maturation/secretion and Wolbachia suppresses the protein which may contribute to the inhibition of DENV in Wolbachia-infected cells. In addition, we explored the subcellular localization of the pelo protein in response to Wolbachia and DENV infection. Furthermore, we demonstrate an indirect involvement of the miRNA aae-miR-2940-5p in regulating pelo in response to Wolbachia infection.

Mosquitoes, flies and cell lines
All mosquitoes used in this study for Wolbachia comparisons were previously produced by McMeniman et al. (2008) by infecting Ae. aegypti mosquitoes with wMelPop-CLA strain of Wolbachia (Pop). Uninfected mosquitoes were generated by tetracycline (Tet) treatment to ensure the complete removal of Wolbachia infection [13]. After confirmation of Wolbachia removal, the Tet mosquitoes were no longer under tetracycline treatment. To ensure the availability of functional gut microflora the Tet mosquitoes were allowed to feed for two generations on water used to raise natural mosquito larvae [13]. Ae. aegypti mosquitoes used for dsRNA experiments were obtained from a colony established from adults collected from Innisfail, north Queensland, and maintained at the QIMR Berghofer Medical Research Institute. w1118 fly line stably infected with wMelPop (Pop) and the tetracycline cured line (Tet) were generated by Min et al. (1997) [14]. Ae. aegypti Aag2 and Pop cell lines previously described by Frentiu et al. (2010) were maintained in 1:1 Mitsuhashi-Maramorosch and Schneider's insect medium (Invitrogen) supplemented with 10% FBS (Bovogen Biologicals) [28]. For microRNA direct interaction analysis, Spodoptera frugiperda Sf9 cells were used which were maintained in Sf900II medium (Life Technologies). All the above-mentioned cell lines were kept at 28˚C and passaged after every 3-4 days.

RT-PCR and qPCR analyses
Total RNA was extracted from either cell lines, mosquitoes or flies using Qiazol (Qiagen) and subsequently treated with Turbo DNase (Ambion) according to the manufacturer's guidelines. Total RNA (1000ng or 200 ng for tissues with less RNA) was used to make cDNA using Superscript III (Invitrogen) using either oligo(dT) primer for cellular genes or DENV-qR primer to amplify the DENV genomic RNA. The pelo transcript sequence (XM_001658653.1) was retrieved from NCBI in order to design primers. The full-length pelo gene was amplified by Taq DNA polymerase (New England Biolabs) using 1μl of cDNA in 25μl reaction together with pelo-specific primers (all primer sequences are listed in S1 Table).
The cDNA reaction was diluted in 1:5 ratio with water, from which 2μl was used for qPCR reactions in which gene-specific primers were utilized to amplify the target genes, using Quan-tiFast SYBR Green (Qiagen) in a Rotorgene thermocycler (Qiagen). For mosquito samples, ribosomal protein S17 (RPS17; AY927787.2) transcripts were used for normalization of RNA templates, while ribosomal protein L32 (RPL32; NM_079843) was used for normalization in the case of D. melanogaster. Each qPCR reaction was performed in two technical replicates each with at least three biological replicates. All qPCR data were normalized with Qiagen analysis templates and were further analysed by Prism 6.0. The qPCR cycling profile included 95˚C for 5 min for the initial denaturation and hot start Taq activation, and 40 cycles of 95˚C for 10 sec, 60˚C for 15 sec and 72˚C for 20 sec. Melt curve analysis was carried out on default settings of the Rotorgene machine to check the specificity of the qPCR products. Unpaired ttest was used to compare differences between two individual groups while One-way ANOVA with Tukey's PostHoc test was performed to find differences between more than two groups of data.

Nuclear and cytoplasmic fractionation
Nuclear and cytoplasmic fractions of Aag2 and Pop cells were separated using the PARIS kit (Ambion) according to manufacturer's instructions. Briefly, cells were washed with 1×PBS, centrifuged at 200 ×g at 4˚C, followed by resuspension of the cell pellet in 300μl ice-cold cell fractionation buffer, incubated on ice for 10 min, centrifuged at 300 ×g at 4˚C for 5 min. The supernatant was collected as the cytoplasmic fraction, and the pellet was washed five times with 200μl ice-cold cell fractionation buffer to ensure removal of traces of the cytoplasmic fraction and considered as the nuclear fraction. The pellet containing the nuclear fraction was resuspended in 300μl of cell disruption buffer. 300μl of 4×SDS-PAGE buffer was added to both nuclear and cytoplasmic fractions.

Western blot analysis
RIPA buffer (Sigma) was used to get total cell lysates from cells and mosquitoes. Briefly, samples were homogenised with the help of 2mm glass beads using a TissueLyser II (Qiagen) at 30 frequency for 90 sec. The lysate was then centrifuged at 4˚C at full speed for 5 min and supernatant was collected for protein analysis. Relative protein concentrations were determined by measuring absorption at 280 nm (Epoch).
Protein samples were run on 10% polyacrylamide gels, transferred to nitrocellulose membrane and then blocked with 5% skimmed milk for 1 hr at room temperature with continuous shaking. After 3 times washing with 1×TBST, membrane was incubated with either anti-pelo (raised in rabbit) [23], anti-GAPDH (Sigma, raised in rabbit), anti-HSP70 (Sigma, raised in rabbit) or anti-DENV envelope protein antibody (Abcam, raised in rabbit) overnight at 4˚C on shaker. For secondary antibody incubation, the blot was washed three times with 1×TBST and then incubated with secondary antibodies either conjugated with alkaline phosphatase (Sigma, anti-rabbit raised in goat) or infrared detection probe (IRDye 800CW goat antihuman LI-COR). After 1 h of incubation, the blot was again washed three times with 1×TBST and then it was further incubated with 5-bromo-4-chloro-3-indolyl-phosphate/nitro blue tetrazolium (BCIP/NBT) ready to use solution (Thermoscientific) or scanned with an Odyssey imager (LI-COR Biosciences).

Oversupply and inhibition of aae-miR-2940-5p
Aag2 cells were plated in a 12-well plate and transfected with 100 μM of artificially synthesized mimic or inhibitor of either aae-miR-2940-5p or non-specific control mimic/inhibitor (Gene-Pharma) (sequences in S1 Table). Cellfectin was used as the transfection reagent (Invitrogen). Cells were collected 72 h post-transfection for downstream analyses.

Northern blot analysis
20 μg of total RNA was run on a 15% urea denaturing polyacrylamide gel, electro-blotted onto a nylon membrane by a semi-dry Western blotting apparatus (Bio-Rad), and UV cross-linked. The blot was blocked in ExpressHyb hybridization solution (Clonetech) for 1h in a rotary oven at 50˚C. Oligonucleotides (20-23 mer) reverse complementary to aae-miR-2940-5p and U6 sequences were labelled with [α-32P]-dATP using terminal nucleotide transferase kit (New England Biolabs). Probe hybridization was carried out in ExpressHyb hybridization buffer (Clonetech) and washes were done under stringent conditions (3× SSC + 5% SDS twice and 1× SSC + 1% SDS twice) at 50˚C. The blot was exposed to a phosphorimager screen overnight, and signals were detected using a Storm phosphorimager scanner (Amersham). To reprobe with U6, the blot was stripped to remove the aae-miR-2940-5p probe with boiling in 0.1% SDS twice for 30 min each time on a shaker. The stripped blot was checked for any radioactive signal by scanning it with Storm phosphorimager scanner.

RNAi-mediated gene silencing
To knockdown the transcript levels of the pelo gene for functional analyses during DENV replication, primers were designed to amplify a 552 bp fragment from the pelo gene containing the T7 promoter sequences at both ends. The MEGAscript T7 Transcription kit (Ambion) was used according to the manufacturer's instructions to produce dsRNA targeting the pelo transcripts. The same approach was used to synthesize dsRNA against Ae. aegypti Argonaut 1 (Ago1; XP_001662554), Ago2 (FJ979880.1), and Green fluorescent protein (GFP; X83960). To knockdown genes, cells in each well were double transfected with 5 and 2μg of dsRNA, respectively. As control, dsGFP RNA was used.
For in vivo silencing of pelo, 3-day old females were injected with 200 nL of dsRNA (dspelo or dsGFP as control) or filtered PBS. Capillary needles (Drummond) were pulled using a Narishige capillary puller using single step program at 86˚C. Mosquitoes were aspirated from the cups, anaesthetized with CO 2 , then placed onto a petri dish on ice under a stereo-dissecting microscope. Females were intra-thoracically injected with 200 nL of appropriate treatment solution (stock concentration 4μg/μl in the case of dsRNAs) using a Nanoject III (Drummond) microinjector at a flow rate of 100 mL/sec. Between 102-104 females were injected for each treatment. Injected mosquitoes were placed into 750ml plastic gauze top containers, provided with 10% sugar solution on cotton pledgets. Injected mosquitoes were maintained in an Environmental Growth Chamber (Panasonic) maintained at 28˚C, 75% RH and 12:12 hour day: night and fading regulated by a 6 step light cycle.

Virus infection and plaque assay
For in vitro virus infection experiments, Ae. aegypti Aa20 cells [29] were seeded at the density of 3×10 5 cells per well in 12-well plates. After settlement of cells, they were double transfected with dsRNA against the target gene or the control. After 6 h post-secondary transfection, they were infected with DENV-2 (New Guinea strain) at the multiplicity of infection (MOI) indicated in the text. Media were collected 72 h after infection for plaque assay and the cells were subjected to RNA extraction to check the DENV genomic RNA levels by RT-qPCR.
For in vivo infection, 48 h after injection of mosquitoes with dsRNAs, mosquitoes were offered the opportunity to feed on a mixture of DENV-2 (QML16) strain virus supernatant in defibrinated sheep blood at a titre of 10 7 pfu/ml using an artificial membrane feeding apparatus with dried bovine caecum lining as the membrane. A second group of the dsRNA injected mosquitoes was intra-thoracically microinjected with 50 nL of the DENV-2 stock (10 7 pfu/ml) using the Nanoject III microinjector at a flow rate of 25 nL/sec while maintained on ice. Following DENV injection, mosquitoes were returned to the 750 ml plastic container for maintenance in the environmental growth chamber and collected 7 days post virus injection. Mosquito lysates were subsequently prepared by homogenizing the body of each mosquito in 500 μl of OptiMEM medium (Sigma) containing 2.5% FBS followed by centrifugation at 5900 ×g for 5 min, at 4˚C. The supernatant was collected and used for virus titration and RNA extraction in order to confirm the pelo knock down.
For plaque assays, Vero cells monolayers previously seeded in a 96-well plate were inoculated with serial dilutions of media or mosquito lysates (10 0 ,10 −1 ,10 −2 , 10 −3 ) collected from experiments in duplicates. Plates were first incubated at room temperature on a rocker for 1 h and then incubated at 37˚C for an extra hour. Subsequently, media were removed and cells were overlayed with 1.5% carboxymethyl cellulose (CMC) and 2.5% FBS in OptiMEM medium. Cells were then incubated for 72 h at 37˚C and 5% CO 2 . Subsequently, the overlay was removed and the cells were fixed with 80% ice-cold acetone in 1×PBS for 20 min at -20˚C, and then air dried overnight. Then cells were blocked with 5% skimmed milk in 1×PBST at 37˚C for 30 min. This was followed by incubating cells with the primary antibody against DENV-2-Envelope (human) in 1:1000 dilution in 0.1% skimmed milk in 1×PBST for 2 h at 37˚C as described previously [30]. After that, plates were washed three times with 1×PBST and then incubated with the secondary antibody (IRDye 800CW goat anti-human LI-COR) for 1 h at 37˚C. Plates were washed three times with 1×PBST and dried as above and scanned on the Odyssey imager (LI-COR Biosciences) at 41μM resolution. Plaques were counted and viral titre was calculated accordingly. Plaque numbers obtained for compound titration were performed in triplicates.

Pelo is ubiquitously expressed in mosquito tissues
Pelo has been demonstrated to be ubiquitously expressed in human and D. melanogaster tissues [24,31]. This protein is also a highly conserved protein in mammals [24], and using multiple sequence alignments for pelo sequences from different insect species we found that the protein is also highly conserved among insect species (S1 Fig) with a distinct nuclear localization signal (PRKRK), and shows structural similarities with human pelo (S2 Fig). In order to investigate the tissue-specific expression of Ae. aegypti pelo, we dissected various tissues from 3-day-old Ae. aegypti female mosquitoes, including salivary glands, midgut, muscles, ovaries, and fat body. RT-qPCR results demonstrated that pelo is ubiquitously expressed in all the Ae. aegypti tissues tested. One-way ANOVA showed statistically significant differences between the different tissue groups (F(4, 10) = 5.786, p = 0.0112). Further Tukey post hoc test showed that significant differences were found in the transcript levels of pelo between salivary gland and fat body (p = 0.006), midgut (p = 0.0091) and ovary (p = 0.025), but not muscle (p = 0.094) (Fig 1). The difference in the levels of pelo was not significant among other tissues.

Wolbachia suppresses pelo in Ae. aegypti
In order to find the effect of Wolbachia on pelo gene expression, RT-qPCR was carried out using pelo gene-specific primers to examine its expression in tetracycline (Tet) treated and Wolbachia wMelPop-CLA infected (Pop) female mosquitoes at two different time points. Transcript levels of pelo were significantly reduced in Pop mosquitoes compared to Tet mosquitoes at 7-days (t = 6.932, df = 4, p<0.0023) and 12-days (t = 11.68, df = 4, p = 0.0003) postemergence (Fig 2A). This result was also cross-validated in Aag2 and Aag2 cells infected with wMelPop-CLA (Pop) cells (Fig 2B; t = 6.495, df = 4, p = 0.0029). To further investigate in which tissue(s) Wolbachia affect pelo, relative expression levels of the gene were assessed in different tissues of both uninfected and infected female mosquitoes at 4 and 12-days after emergence by dissecting salivary gland, midgut, muscle, ovary and fat body from a total of 30 mosquitoes from each time point divided into three biological replicates. We found significant decreases in the pelo transcript levels in both 4-day (except for fat body in which a decrease trend was not significant) and 12-day-old Pop mosquitoes in ovary (F

Wolbachia affects the subcellular localization of the pelo protein
Pelo has been reported to be localized to the cytoplasm in D. melanogaster [32]. Although the pelo protein does have a conserved nuclear localization signal (PRKRK), there is a lack of experimental evidence of its presence inside the nucleus [24]; therefore, we explored its subcellular localization in Ae. aegypti cells and whether Wolbachia affects the localization of pelo. Nuclear and cytoplasmic fractions of both Aag2 and Pop cells were separated. Lysates were run on a 5-15% Bis/Tris polyacrylamide gel, total proteins were transferred to nitrocellulose membrane and then probed with an anti-pelo antiserum along with other control antisera, including anti-histone H3 (to confirm the nuclear fraction), anti-GAPDH (to confirm the cytoplasmic fraction), and anti-WSP (to confirm Wolbachia infection). Here it is important to note that we detected the WSP protein in the nuclear fraction of Pop cells as well, although in smaller amount as compared to the cytoplasmic fraction, which is consistent with the earlier report demonstrating the presence of Wolbachia (wMelPop) in the nuclei of D. melanogaster cells with the help of electron microscopy [14]. Nevertheless, the control antibodies showed a very good fractionation of the two cellular compartments. Our results demonstrated that in Aag2 uninfected cells, pelo was only detectable in the cytoplasmic fraction, however, Wolbachia infection led to the nuclear import of the protein thus affecting its subcellular localization as compared to uninfected Aag2 cells (Fig 3F). Further investigation is required to elucidate if the pelo protein plays any role inside the nucleus of Wolbachia-infected cells.

Wolbachia-mediated downregulation of pelo is highly specific to female mosquitoes
We measured the levels of the pelo transcripts in both male and female 4-day-old Tet and Pop mosquitoes. RT-qPCR analysis showed that there was no significant change in the levels of pelo between male Tet and Pop mosquitoes (Fig 4A; t = 0.06021, df = 4, p = 0.9549). Conversely, the level of pelo transcripts was significantly reduced in females in the presence of Wolbachia (Fig 4A; t = 4.307, df = 4, p = 0.0126). Although there is no evidence in Ae. aegypti mosquitoes to show differential host response to bacterial infections in the two sexes, a study carried out in the parasitoid Asobara tabida has highlighted differential immune response in males as compared to females challenged with Wolbachia strains wAtab1, wAtab2, and wAtab3, where males showed higher levels of immune gene expression upon Wolbachia infection as compared to females [33]. Our results also point towards a sex-specific response in Ae. aegypti mosquitoes during Wolbachia infection.
As Wolbachia is not a natural symbiont of Ae. aegypti, we were interested to test whether Wolbachia has the same suppressive effect on pelo in its natural host D. melanogaster. By measuring the relative levels of the pelo transcripts in Tet and Wolbachia wMelPop infected 4-dayold flies, we found that there was no significant change in the levels of pelo in infected and uninfected male and female flies (Fig 4B). The finding that the Wolbachia wMelPop-CLA infection suppresses pelo in females of the recently transinfected mosquito species Ae. aegypti, and not in its natural host D. melanogaster suggests that this effect could be attributed to the relatively recent association between Wolbachia and a new host. microRNA aae-miR-2940-5p is involved in regulation of pelo microRNAs (miRNA) are known to regulate different target genes. To find out whether the suppression of pelo in Wolbachia-infected mosquitoes is mediated by miRNA, we knocked down Ago1 and Ago2 genes in Pop cells, as both are important components of the RISC complex involved in miRNA functions [34,35]. After confirming silencing of the Agos with dsRNA specific to the genes (Fig 5A and 5B), we employed RT-qPCR followed by the statistical analysis with One-way ANOVA (F(2, 6) = 25.1, p = 0.0012). Tukey post hoc test results showed that after Ago2 knockdown there was a significant increase in the expression of pelo as compared to both mock (p = 0.0014) and Ago1 knock down (p = 0.0035) (Fig 5C;), which suggested that miRNAs might be involved in regulating pelo levels. Interestingly, we found that when Ago1 was silenced, Ago2 levels increased (Fig 5A), and conversely when Ago2 was knocked down, Ago1 transcript levels increased (Fig 5B). This is consistent with previous studies carried out in D. melanogaster that Ago proteins may compensate for each other [36].
As Wolbachia-mediated pelo suppression was found to be specific to female Ae. aegypti mosquitoes, we aimed to determine if a mosquito-specific miRNA that was previously shown to be significantly up-regulated in Wolbachia wMelPop-CLA infected female mosquitoes, as compared to uninfected ones, could be involved in regulation of pelo [18]. Although three targets of this miRNA have already been identified [18,19,37], we tried to determine if it has any interaction with the transcripts of the pelo gene. RNAhybrid analysis showed potential aae-miR-2940-5p target sequences in the ORF of the pelo gene at positions 561-582 with significant seed region complementarity and high minimum free energy of -25.8 kcal/mol (S3 Fig). In order for a miRNA to regulate a target mRNA, it is important that the mature miRNA and its target transcript co-localize in the same tissue. As we found that pelo was expressed in all the mosquito tissues that we examined and Wolbachia supresses its expression in all the tested tissues ( Fig 3A-3E), we were interested to find out the tissue-specific expression of aae-miR-2940-5p in mosquito tissues in the presence and absence of Wolbachia. Northern blot analysis of RNA isolated from various tissues (muscle was not included) of Pop and Tet Ae. aegypti mosquitoes confirmed induction of aae-miR-2940-5p due to Wolbachia infection in all tissues, with substantial induction in the midgut and fat body (Fig 6A). The trend in these inductions was of an inverse association with reductions of the pelo transcript levels in the corresponding tissues examined (Fig 3A-3D).
To verify the interaction between aae-miR2940-5p and pelo, Aag2 cells were transfected with a synthetic aae-miR2940-5p mimic, control mimic with random sequences, and without a mimic (mock transfection) as a second control. RT-qPCR results subjected to One-way ANOVA (F(2, 6) = 33, p = 0.0006) followed by Tukey post hoc test showed that there were significant reductions in the transcript levels of pelo in Aag2 cells transfected with aae-miR2940-5p mimic as compared to mock (p = 0.0022), and control mimic transfected cells (p = 0.0006) (Fig 6B). Furthermore, the effect of synthetic mimic and inhibitor of aae-miR-2940-5p on Pelo protein expression was analysed 72 h post-transfection. Western blot results suggested downregulation of the pelo protein in the presence of the aae-miR-2940-5p mimic, while application of the inhibitor did not affect pelo levels as compared to the negative control inhibitor; it showed an increase when compared to the mock transfection or negative control mimic (Fig 6C).
In order to examine if there is direct interaction between aae-miR-2940-5p and the predicted target sequences in pelo, we cloned the target sequences downstream of GFP in the pIZ/ V5-His vector. The GFP-pelo target sequence clone was co-transfected with the control random sequence mimic and aae-miR-2940-5p mimic into Aag2 cells. RNA was collected after 72 h, and RT-qPCR was carried out to check the level of GFP transcripts. The results showed that there was no change in the levels of GFP in response to aae-miR-2940-5p mimic (Fig 6D; F(2,  6) = 0.9271, p = 0.4458), which suggests that likely there is no direct interaction between pelo and aae-miR-2940-5p. This result is not surprising as most miRNAs are reported to target Western blot analysis to examine the effect of aae-miR-2940-5p mimic and inhibitor on pelo protein levels using an anti-pelo specific antibody (upper panel), while anti-beta-actin antibody was used as control to confirm equal loading of samples. Aag2 M, mock transfection; NC mimic, negative control mimic; 2940 mimic, aae-miR-2940-5p mimic; NC inhibitor, negative control inhibitor; 2940 inhibitor, aae-miR-2940-5p inhibitor. The relative density of pelo to actin determined by densitometric analysis using Image J is shown in the lower panel. D) RT-qPCR analysis of the effect of aae-miR-2940-5p mimic on the level of GFP transcripts in Sf9 cells transfected with pIZ/GFP fused with the pelo target sequences along with the negative control (NC) or aae-miR-2940-5p mimic. Mock, no mimic with transfection reagent only. ns, not significant; One-way ANOVA.
https://doi.org/10.1371/journal.pntd.0006405.g006 transcription factors or other proteins that in return fine tune the abundance of different transcripts [38]. To rule out the possibility of another target site for the miRNA in the pelo transcript undetected by the prediction program, the full-length pelo, including its 3'UTR, was cloned into the pIZ/V5-His vector and co-transfected with aae-miR-2940-5p mimic in Sf9 cells. There was no downregulation of pelo in the presence of aae-miR-2940-5p mimic (S4 Fig), which confirmed our aforementioned observation.

Pelo plays a role in DENV virion production
Recently, it has been reported that pelo is required for efficient replication of Drosophila C virus (DCV) [23]. In order to examine if the pelo protein has a similar effect on DENV replication, we infected Ae. aegypti Aa20 cells with DENV-2 (NGC strain) at 1 multiplicity of infection (MOI; pfu/cell). Cells were collected at 3 and 5 days post-infection (dpi) from which RNA and proteins were extracted. RT-qPCR results showed no significant change in the pelo transcript levels during DENV infection at both 3 (t = 1.12, df = 4, p = 0.3254) and 5 dpi (t = 1.447, df = 4, p = 0.2214) (Fig 7A). However, western blot analysis of the cells using an antiserum against D. melanogaster pelo clearly showed an increase in the levels of the protein upon infection at both 3 and 5 dpi (Fig 7B). The same blot was re-probed with an anti-DENV-Envelope antiserum to confirm infection. This suggests that pelo could be regulated at the post-transcriptional level. To check the localization of the pelo protein during DENV infection, we fractionated mock and 5 dpi Aa20 cells (1 MOI) into nuclear and cytoplasmic fractions. Both nuclear and cytoplasmic lysates were run on a 5-15% Bis/Tris polyacrylamide gel. After transfer to nitrocellulose membrane, the blot was probed with the pelo antiserum. Results showed that unlike Wolbachia, DENV infection has no effect on the subcellular localization of pelo, which remains in the cytoplasm in both uninfected and DENV-2 infected cells (Fig 7C).
The above results suggest that DENV might increase levels of the host pelo protein in order to facilitate its replication, similar to DCV in D. melanogaster. In order to further confirm this link, the transcript levels of the pelo gene was knocked down in Aa20 cells using pelo specific dsRNAs (Fig 8A; about 70% silencing) and subsequently infected with DENV-2 (NGC strain) at 1 MOI. We carried out DENV infection experiments in Ae. aegypti cell line Aa20, because Aag2 cell line is persistently infected with insect-specific flavivirus cell fusing agent virus (CFAV) [39], but not Aa20 cells. At 3 dpi, cells were harvested and subjected to RT-qPCR using DENV-2 specific primers to the NS1 gene. Surprisingly, RT-qPCR results showed no change in the DENV-2 genomic RNA levels in pelo depleted Aa20 cells as compared to mock and dsGFP transfected cells (Fig 8B; F(2, 6) = 3.106, p = 0.1186). However, plaque assay displayed a significant reduction (F(2, 9) = 12.88, p = 0.0023) in the virus titre in the medium collected from Aa20 cells treated with dspelo and infected with DENV-2 compared to both mock (p = 0.0028) and dsGFP (p = 0.0084) treated Aa20 cells followed by DENV-2 infection ( Fig  8C). These results demonstrated that when pelo was knocked down the number of DENV infectious particles produced and secreted outside the cells significantly declined, but there was no effect on the total genomic RNA of DENV, implicating that pelo might play an important role in the translation of the viral genomic RNA and/or assembly and release of DENV-2 virions.
To show the effect of pelo silencing in vivo, we initially injected 4-day-old mosquitoes with dsPelo, dsGFP and PBS, and after two days fed them with sheep blood containing 10 7 pfu/ml DENV-2. Seven days following virus inoculation, mosquitoes were collected and tested for DENV genomic RNA levels in the treatments using RT-qPCR. This time point was selected to allow sufficient time for the virus to replicate for comparison between the treatments. However, we detected very little viral RNA in the samples (S5 Fig), and further testing of individual mosquitoes determined that only about 13% of the mosquitoes were infected with DENV. Although oral inoculation is the natural route of infection, since our main purpose was to find out if dsPelo treatment has the same effect on DENV replication in mosquitoes as in the cell line, we also injected mosquitoes with virus, which resulted in 100% infection. Silencing pelo in mosquitoes when injected with dspelo (Fig 8D; about 92%) followed by DENV injection also showed no significant change (F(2, 3) = 4.719, p = 0.118) in DENV gRNA levels (Fig 8E), but significant reduction (F(2, 8) = 22.74, p = 0.0005) in DENV titres as compared to mock (p = 0.0022) or dsGFP (p = 0.0006) injected mosquitoes at 7 dpi (Fig 8F). The sample size we used for the in vivo assay was not high, but it corroborated with the in vitro results.

Discussion
Despite numerous efforts to unveil the mechanism by which Wolbachia manipulates its host environment to restrict virus replication, the exact mechanism(s) that govern this antiviral property are largely unknown. In this study, we have found that the Wolbachia supresses the pelo protein which in turn may contribute towards restricting DENV virion production in Ae. aegypti. Wolbachia is a facultative endosymbiont in many insect species and other invertebrates, however, the wMelPop-CLA strain, which most effectively suppresses DENV replication, has no natural association with the primary DENV vector Ae. aegypti and has been artificially transinfected into Ae. aegypti [13]. Several reports have shown that Wolbachia-mediated activation of immune genes in the Imd and Toll pathways could be involved in host protection from various viruses in mosquitoes [11,40,41]. Other reports, however, have shown that Wolbachia does not elicit the host immune response, in particular in hosts that are naturally infected with Wolbachia [42][43][44]. A recent study investigating the role of Wolbachia-induced restriction of Semliki Forest virus within D. melanogaster cells (JW18) suggested that interference occurs at a very early stage of infection and at the level of viral RNA translation or host RNA transcription [45]. However, in some Wolbachia-host-virus associations viral replication is not inhibited but rather a tolerance to viral infection is conferred by Wolbachia [46,47].
Recently  [23,48]. However, the potential role of pelo has not been characterized yet in the case of the medically important DENV and its controlling agent Wolbachia in the mosquito vector. RT-qPCR analyses revealed or with dsRNA against GFP as control or with dsRNA against pelo to silence the gene. RPS17 was used to normalize the qPCR data, showing no effect at the DENV genomic RNA level. C) Viral titre determination by plaque assay conducted on the media collected from cells treated as in B. D) RT-qPCR based confirmation of knock down of pelo in Ae. aegypti mosquitoes. E) RT-qPCR analysis to examine the genomic RNA level of DENV in 7-day infected mosquitoes injected with either no dsRNA (Mock) or with dsRNA against GFP as control or with dsRNA against pelo to silence the gene. RPS17 was used to normalize the qPCR data. F) Viral titre determination by plaque assay conducted on mosquito lysates prepared from mosquitoes treated as in E. Error bars represent SEM from three biological replicates. Ã , p<0.05; ÃÃ , p<0.01; ÃÃÃ ; ns, not significant; One-way ANOVA. Mock, no dsRNA and transfection reagent only; dsgfp, dsRNA to green fluorescent protein gene, dspelo, dsRNA to the pelo gene.
https://doi.org/10.1371/journal.pntd.0006405.g008 that the pelo gene is expressed ubiquitously throughout all the main tissues of Ae. aegypti with higher transcript levels in the salivary glands, which is consistent with the studies conducted on tissue localization of pelo in the case of human and D. melanogaster [24,31]. Interestingly, we found that pelo is supressed in the presence of Wolbachia in cell lines, whole mosquitoes and a number of tissues such as the salivary gland, muscle, and ovary. However, this suppression is specific to female mosquitoes; it was not seen in the male mosquitoes or in either male or females of D. melanogaster infected with wMelPop. The difference in pelo regulation in Wolbachia-infected Ae. aegypti versus D. melanogaster could be due to the length of association between the bacteria and host; Wolbachia is a natural infection in the fly whereas its infection of Ae. aegypti mosquitoes was only recently established through transinfection [49].
DENV titre in Pop cells was below the detection limit of our plaque assay, and others [11,12] have also shown that barely any virion is produced in wMelPop-infected cells or mosquitoes that is detectable by plaque assay. While in the Aa20 cell line treated with dspelo, inhibition of DENV was not as dramatic as seen in wMelPop-infected mosquitoes or the cell line, silencing the gene in mosquitoes had very similar effects on DENV replication to wMelPopinfected mosquitoes, which could be due to higher efficiency of gene silencing. Nevertheless, as indicated above, downregulation of pelo in the presence of Wolbachia does not seem to be the only mechanism of virus blocking.
Pelo is a highly conserved protein in mammals [24], and using multiple sequence alignments for pelo sequences from different insect species we found that the protein is also highly conserved among insect species (S1 Fig). It has a distinct nuclear localization signal (PRKRK), and shows structural similarities with human pelo (S2 Fig). However, it has been demonstrated that pelo mostly resides in the cytoplasm and there is a lack of evidence for the presence of pelo in the nucleus [32]. This led us to investigate the localization of the pelo protein in mosquito cells and in particular in the instance of Wolbachia infection. Western blot results showed that the pelo protein is mainly found in the cytoplasm of mosquito cells when uninfected, however, Wolbachia infection leads to a change in the subcellular localization of pelo, which was found both in the cytoplasm and the nucleus. This change in the subcellular localization of pelo during Wolbachia infection may restrict availability of the protein in the cytoplasm, which is required for rapid translation of specific viral proteins. This would be in agreement with a previous study in which it was shown that pelo is required for the synthesis of the capsid protein of DCV, which they described as a quickly translated protein [23]. Alternatively, the translocation of pelo into the nucleus may have an inhibitory effect on a host gene (s) that normally facilitates viral replication. miRNAs are important regulators of different cellular processes including cell division, timing, differentiation, death, control of metabolism, transposon silencing and antiviral defence [50][51][52]. Recently, our group has shed light on the modulation of different cellular miRNAs in the case of infection with wMelPop strain of Wolbachia in female mosquitoes [18]. Along similar lines, in this study we have investigated whether pelo has an interaction with any miRNA by silencing both Argonaute 1 (Ago1) and Argonaute 2 (Ago2) genes, which are important components of the RISC complex and involved in miRNA function [36,53]. Interestingly, we found that silencing of one Ago led to the upregulation of the other Ago. One possible reason might be class switching between the Agos [36]. Increase in pelo levels in Ago2 silenced cells suggested that pelo downregulation in Wolbachia-infected cells could possibly be mediated by miRNA. Hussain et al (2011) provided evidence that in Wolbachia wMelPop-CLA infected mosquitoes the mosquito-specific aae-miR-2940-5p was significantly upregulated as compared to uninfected mosquitoes [18]. While three targets of this miRNA have already been identified [18,19,37], we explored whether it has any interaction with the transcripts of the pelo gene. The co-localization of both pelo transcripts and mature aae-miR-2940-5p highlights the possibility that in Wolbachia-infected cells aae-miR-2940-5p might be utilized to downregulate pelo transcripts. These results also further confirmed differential expression of this miRNA in Wolbachia-infected mosquitoes at the tissue level. Further investigation showed that pelo transcript levels were downregulated in the presence of the artificially synthesized mimic of aae-miR-2940-5p both at the transcript and the protein levels, while there was an increase in the pelo protein in the case of addition of an artificially synthesized inhibitor of aae-miR-2940-5p. However, target validation results using GFP as a reporter suggested that pelo may not be a direct target of aae-miR-2940. The action of aae-miR-2940-5p could be through regulation of a transcription factor(s) or other protein(s) by the miRNA that fine-tunes the abundance of the pelo transcript [38].
Viruses are well known to modulate transcripts of host cells for their own benefit. Recently, a research group examined D. melanogaster mutants that resist DCV replication through a forward genetic screen and demonstrated a DCV-resistant mutant to be deficient of the pelo gene [23]. Pelo protein has also been implicated in TYLCV resistance in tomato TY172 cultivar [48]. However, its potential role in the case of DENV has not yet been explored. Our results suggest an increase in the levels of pelo in DENV-infected cells at the protein level without a change in its subcellular localization, which was mostly cytoplasmic. Furthermore, pelo knockdown studies suggest it plays a role in DENV virion production/secretion. The outcome of this study is in agreement with the previous findings that showed pelo is vital for the translation of the capsid protein of DCV thus positively affecting replication of DCV. A similar role of pelo was found in other viruses including Cricket Paralysis virus, Double Drosophila X virus, and invertebrate iridescent virus 6 and TYLCV replication [23,48], suggesting that pelo might be an important host factor that is recruited by viruses to facilitate the translation of viral genome.
In summary, we have demonstrated that the pelo protein facilitates DENV production/ secretion, and in female Ae. aegypti mosquitoes, Wolbachia suppresses the pelo protein, which may consequently contribute to restriction of DENV in the mosquitoes. This effect could be due to relocalization of the pelo protein to the nucleus in Wolbachia-infected cells as compared to non-infected cells. In addition, regulation of pelo in Wolbachia-infected mosquito cells appears to be mediated by aae-miR-2940-5p. However, the regulation of pelo was found to be female mosquito specific and not observed in male mosquitoes or in D. melanogaster males or females. Therefore, while suppression of pelo in female mosquitoes may contribute to virus inhibition, this does not seem to be the universal mechanism of virus inhibition seen across different host-virus interactions. However, identification of an association between pelo, the miRNA aae-miR-2940-5p and virus inhibition in Wolbachia-infected Ae. aegypti provides new insights into the mechanism of Wolbachia induced virus inhibition, which is a central feature of a major new strategy against arboviral diseases.  Fig. Pelo transcript is not a direct target of aae-miR-2940-5p. The full-length pelo transcript, including the 3'UTR was cloned into the pIZ vector and transfected into Aag2 cells either without or with negative control (NC) or aae-2940-5p mimic. The results confirmed overexpression of pelo in cells transfected with pIZ/Pelo only. Application of the aae-miR-2940-5p did not significantly change the expression levels of pelo, indicating there is no direct interaction of the miRNA with the transcripts of pelo. Error bars represent SEM from three biological replicates ( ÃÃ , p<0.01; ÃÃÃ , p<0.001; ns, not significant; One-way ANOVA).