The Relative Importance of Innate Immune Priming in Wolbachia-Mediated Dengue Interference

The non-virulent Wolbachia strain wMel and the life-shortening strain wMelPop-CLA, both originally from Drosophila melanogaster, have been stably introduced into the mosquito vector of dengue fever, Aedes aegypti. Each of these Wolbachia strains interferes with viral pathogenicity and/or dissemination in both their natural Drosophila host and in their new mosquito host, and it has been suggested that this virus interference may be due to host immune priming by Wolbachia. In order to identify aspects of the mosquito immune response that might underpin virus interference, we used whole-genome microarrays to analyse the transcriptional response of A. aegypti to the wMel and wMelPop-CLA Wolbachia strains. While wMel affected the transcription of far fewer host genes than wMelPop-CLA, both strains activated the expression of some immune genes including anti-microbial peptides, Toll pathway genes and genes involved in melanization. Because the induction of these immune genes might be associated with the very recent introduction of Wolbachia into the mosquito, we also examined the same Wolbachia strains in their original host D. melanogaster. First we demonstrated that when dengue viruses were injected into D. melanogaster, virus accumulation was significantly reduced in the presence of Wolbachia, just as in A. aegypti. Second, when we carried out transcriptional analyses of the same immune genes up-regulated in the new heterologous mosquito host in response to Wolbachia we found no over-expression of these genes in D. melanogaster, infected with either wMel or wMelPop. These results reinforce the idea that the fundamental mechanism involved in viral interference in Drosophila and Aedes is not dependent on the up-regulation of the immune effectors examined, although it cannot be excluded that immune priming in the heterologous mosquito host might enhance the virus interference trait.


Introduction
Wolbachia is a vertically transmitted endosymbiont that infects up to 70% of all insect species. The association is usually not obligatory for the insect and many Wolbachia strains assure their maintenance in populations by manipulating the reproduction of their host [1]. Interestingly, some strains interfere only weakly with host reproduction but still spread and are maintained in insect populations [2]. Their success may be explained by an additional positive selective advantage associated with Wolbachia infection. One possible advantage is the recently described pathogen blocking that the bacterium confers upon its host. This phenotype was first demonstrated in Drosophila, where Wolbachia induces resistance to different types of RNA viruses by reducing viral titer and/or making the host resistant to virus pathogenicity [3][4][5]. The extent and nature of blocking vary according to the virus and the Wolbachia strains tested. For example, Wolbachia reduces the titer of the closely related DCV and Nora viruses in Drosophila melanogaster and D. simulans [4,5] and as a consequence, the pathology associated with those two viruses is less intense in Wolbachiainfected flies [3][4][5]. In contrast, the bacterium does not affect FHV titer in Drosophila but still reduces the pathogenicity of the virus [3][4][5]. In D. simulans, the wAu Wolbachia strain has a strong effect against DCV pathogenicity, whereas the strains wHa and wNo do not [5]. This observation is thought to be related to the low infection density of wHa and wNo in Drosophila compared to that of the wAu strain [5].
Wolbachia does not naturally infect the main mosquito vector of dengue viruses, Aedes aegypti. However, two Wolbachia strains originally isolated from D. melanogaster (wMelPop-CLA and wMel) and one strain originally from A. albopictus (wAlbB) have been successfully trans-infected into A. aegypti and subsequently stably maintained [6][7][8]. All of these strains express cytoplasmic incompatibility in A. aegypti as they do in their original hosts, D. melanogaster and A. albopictus [6][7][8]. In addition, the virulent wMelPop-CLA strain that lacks normal replication control and reduces lifespan in D. melanogaster also does so in A. aegypti [7]. As observed in Drosophila, Wolbachia-infected A. aegypti are more resistant to RNA virus infection, including dengue and chikungunya [8,9], as well as bacteria, nematodes and Plasmodium [9,10]. Transient somatic infection of the main African vector of human malaria, Anopheles gambiae, by wMelPop also significantly decreased Plasmodium infection intensity [11].
The molecular mechanisms involved in Wolbachia-mediated pathogen protection are still not clear. One plausible hypothesis is that Wolbachia interferes with pathogens by pre-activating the immune response of its host. The virulent strain wMelPop-CLA activates a wide range of immune processes in A. aegypti, including the Toll and Imd signaling pathways, anti-microbial peptide synthesis, melanization, RNA interference and opsonisation [9,10] and the somatic infection of An. gambiae by wMelPop caused an increase in expression of opsonisation genes [11]. Evidence for the role of opsonisation in protection against Plasmodium in this host was demonstrated by knocking down expression of the TEP1 gene [11]. Transcriptional analyses of A. aegypti immunity genes showed that wAlbB increases expression of genes in the Toll pathway and in particular the anti-microbial peptide gene, defensin [12]. Activation of the Toll pathway has been shown previously to suppress dengue infection in mosquitoes [13]. Each of these previous studies was limited in that they examined Wolbachia strains that were either virulent and/or recently introduced into naturally uninfected host species. To our knowledge, only two previous studies have examined expression of innate immune genes in insect species naturally infected by Wolbachia, including D. simulans, D. melanogaster and A. albopictus. In these cases no differences in gene regulation were observed between Wolbachiainfected insects and their uninfected counterparts [14,15].
Since all previous studies that have shown evidence of immune activation have been based on recently established heterologous infections, it is unclear how generalizable the Wolbachia activation of the mosquito immune system is for all insects. To determine whether immune up-regulation by the bacterium is a general mechanism underlying Wolbachia-induced dengue interference, we performed transcriptional analyses on the two heterologous associations, wMel and wMelPop-CLA infected A. aegypti, and the two native associations, wMel and wMelPop infected D. melanogaster. We also tested if the non-virulent strain wMel blocks dengue replication in Drosophila as it does in mosquitoes. If the same strain of Wolbachia blocks the replication of the same virus in different hosts, we can make the parsimonious assumption that virus interference is likely to have a common mechanistic basis across different hosts. This cross-comparison with the two Wolbachia strains and dengue virus in both native and heterologous hosts allows us to remove extraneous effects, such as recent transfer to a heterologous host or virulence associated with the wMelPop infection, that might confound an understanding of the underlying mechanistic basis of Wolbachia-induced viral interference.
This study also contributes to our understanding of the physiological impact of wMel infection on A. aegypti. This is of particular relevance because wMel-infected A. aegypti have been released in north Queensland, Australia, in a field trial using Wolbachia as a biocontrol mechanism for dengue [16]. In the near future, this biological tool is also likely to be applied in dengueendemic areas of Vietnam and Indonesia [17].

Transcriptional response of Aedes aegypti to Wolbachia infection
We examined the global transcriptional response of mosquitoes to Wolbachia infection using microarrays. We compared the responses of 8 day old, non blood-fed A. aegypti females stably transinfected with wMelPop-CLA (line PGYP1) or wMel (line MGYP2) to those of the corresponding tetracycline-cured lines PGYP1.tet and MGYP2.tet. The design of the microarray included 12,336 transcripts, which represented 12,270 of the 15,988 genes present in the A. aegypti genome. We considered a gene to be up-or down-regulated by wMelPop-CLA or wMel infection if the fold change in transcription relative to non-infected mosquitoes was significantly different from 1.0 and greater than 1.5. Because the Drosophila genome is better characterized, we identified Drosophila orthologs of each A. aegypti gene where possible to obtain additional functional annotations.
The wMelPop-CLA infection affected the transcription of far more genes (2723) than the wMel infection (327) (Figure 1). This is likely related to wMelPop-CLA's higher density in its host, broader cellular tropism and pathogenicity [8,9,18]. Based on Gene Ontology (GO) annotations, wMelPop-CLA has an impact on a broader range of A. aegypti biological and molecular functions than wMel ( Table 1, 2).
Many of the changes in gene regulation observed in mosquitoes infected with the virulent strain wMelPop-CLA are likely to be responses to the high physiological cost imposed by that strain. To identify mechanisms more likely to be involved in pathogen interference, we decided to focus on the 210 gene transcripts that showed significant changes in expression in both PGYP1 and MGYP2 compared to uninfected mosquitoes ( Figure 1). Among those genes, 138 gene transcripts had functional annotations (Table S1).
Most of the 210 transcripts were either up-regulated in both PGYP1 and MGYP2 or down-regulated in both lines (Table S1). However, the magnitude of response was typically greater to wMelPop-CLA infection (Table S1). One of the few genes differentially expressed between PGYP1 and MGYP2 is AAEL002487, which is up-regulated in MGYP2 and downregulated in PGYP1. This gene encodes the protein P53 regulated pa26 nuclear protein sestrin (dSesn in Drosophila) (Table S1). This protein is involved in the regulation of the target of rapamycin (TOR), a key protein in age-related pathologies like life-shortening or muscle degeneration [19], two phenotypes exclusively associated with wMelPop-CLA pathogenicity in A. aegypti [7,20]. Among the 210 genes, most of the genes showing the greatest upregulation are immune genes (Table S1). Gene Ontology (GO) annotations also revealed enrichment in genes related to immunity and proteolysis for MGYP2 and PGYP1 (  The virulent strain wMelPop-CLA significantly affected regulation of many characterized immune genes in the mosquito (Table S2, [10]). By comparison, many fewer of these genes were activated by wMel (Table 3, S1, S3). Those included genes

Author Summary
Wolbachia pipientis is an inherited intracellular bacterium that is widespread in insects. Because of its ability to interfere with various pathogens such as dengue viruses, nematodes and Plasmodium in insects, it has been proposed as a possible tool to control insect-transmitted disease. Recently, two strains of Wolbachia that interfere with RNA viruses in their natural host, Drosophila melanogaster, were introduced into the naturally uninfected mosquito vector of dengue fever, Aedes aegypti. As in their natural host, those two strains block the replication and the dissemination of viruses in the mosquito. Some studies suggest that pathogen blocking is due to Wolbachia priming the insect innate immune system. Here, we show that Wolbachia induces transcription of some immunity related genes only in its new host A. aegypti, and not in its natural host D. melanogaster, while Wolbachia reduces dengue replication in both hosts. These results suggest that immune priming by Wolbachia might not be the only mechanism responsible for viral interference.
encoding anti-microbial peptides, four cecropins (CECE, CECF, CECN, CECD), one defensin (DEFC) and one diptericin (DPT1). The magnitude of change in expression was substantial for some of these genes. The activation of these peptides is regulated by both Toll and Imd pathways, but we found up-regulation only of some Toll pathway genes, including the peptidoglycan recognition protein PGRP-SA and the Gram-negative binding proteins GNBPB4 and GNBPA1 (GNBP1 Drosophila homologs, Table 3). The Toll pathway effector defensin was the most highly upregulated immune gene in A. aegypti infected by wMel (Table 3). This is consistent with the results of Bian et al [12], who examined immune gene expression in heterologous wAlbB infection in A.
aegypti and found that among the immune genes tested defensin was also the most up-regulated.
Excluding anti-microbial peptides and the Toll pathway, the only other immune response activated by both wMel and wMelPop-CLA in A. aegypti was melanization. Four genes in this pathway were up-regulated: one pro-phenoloxidase (PPO4), one dopachrome-conversion enzyme (DCE) that converts dopachrome into 5,6-dihydroxyindole just before melanin production by phenoloxidase [21], one putative protease inducer sp7 and one protease inhibitor Srpn4 ( Table 3). The activation of these genes suggests that production of melanin is induced in Wolbachiainfected mosquitoes.

Effect of Wolbachia on dengue virus in Drosophila
Since a comparative approach between Drosophila and Aedes to examine the effect of immune activation on virus interference is predicated on an assumption that dengue virus interference also occurs in Wolbachia-infected Drosophila, we tested the ability of dengue virus serotype 2 (DENV-2) to grow in Drosophila carrying the wMel Wolbachia strain.
For both dengue virus strains, 92T and ET300, the total number of flies infected by dengue was lower in the presence of wMel, with only 40% of flies detected positive for the 92T strain compared with 93% for the Wolbachia-uninfected control. Similarly for the ET300 strain, 73% of Wolbachia-infected flies were positive for dengue compared to 93% for the Wolbachiauninfected control ( Figure 2). In addition, for the flies that did become infected with dengue the amount of DENV-2 RNA present was significantly reduced in the presence of wMel ( Figure 2). It was unsurprising to note that dengue grew to higher levels when injected into its natural mosquito host compared to Drosophila but regardless of absolute virus levels significant Wolbachia interference effects were detected in both insect species. Dengue injection in flies did not have an effect on insect life span nor increased mortality compared to controls (data not shown).

Effect of wMelPop and wMel on the Drosophila melanogaster immune system
Considering that the Wolbachia strains wMelPop [22] and wMel in their original host interfere with natural Drosophila RNA viruses and also with dengue virus replication, we then investigated the  possibility that both Wolbachia strains boost Drosophila immunity as seen in the heterologous mosquito host. We examined by quantitative real time PCR the expression of the Drosophila homologs of the mosquito immune genes identified through microarray analysis to be up-regulated in the presence of Wolbachia.
There have been multiple gene losses and gene duplications in immune gene families in both flies and mosquitoes [23], and we were therefore unable to reliably identify all orthologs for our antimicrobial peptide genes and pro-phenoloxidase genes of interest. Thus, we targeted all the cecropin, diptericin and pro-phenoloxidase  genes present in the genome of D. melanogaster. In total 13 immune genes were analyzed: seven anti-microbial peptide genes, two Toll pathway genes and four melanization genes (Table 4). No significant changes in the expression of anti-microbial peptide genes were observed for w 1118 wMelPop or w 1118 wMel, except for cecropin A1 ( Table 4). The expression of cecropin A1 was two-fold higher in the presence of wMelPop, whereas no change was observed in the presence of wMel (   Wolbachia in Drosophila. The expression patterns of two major genes in the Toll pathway, PGRP-SA and GNBP1, differed between flies infected by wMel and wMelPop. A slight inhibition of PGRP-SA was observed in flies infected by wMelPop, while in wMel-infected flies there was no effect. For GNBP1, a minor but significant difference, 1.29-fold change, was observed for w 1118 wMel but not for w 1118 wMelPop ( Table 4). The expression of only a single melanization gene was affected by wMel infection: proPO-A1 was down-regulated. In contrast, in flies infected with wMelPop, proPO-A1 was significantly up-regulated and another melanization gene, CG42640, was down-regulated (Table 4). An enrichment of gene transcripts encoding the iron binding proteins transferrin and ferritin was detected in the data obtained from the A. aegypti transcriptome analysis in response to wMel and wMelPop-CLA infections ( Table 1, 2, S1). These proteins have multiple functions in insects, including iron homeostasis and immunity [24], two potential mechanisms that could be involved in Wolbachia-mediated pathogen protection. The expression of the genes encoding transferrin 1 (Tsf1) and the light chain of ferritin (Fer2lch) was evaluated in w 1118 wMel and w 1118 wMelPop compared to w 1118 tet. However, no induction was found in Wolbachia-infected flies (Table 4) and wMelPop infection even significantly reduced the expression of transferrin.
The expression of immune genes was also tested in the same fly lines (w 1118 wMel and w 1118 tet) infected with DENV-2, strain 92T. Even in the presence of dengue, wMel infection did not increase the expression of anti-microbial peptides and pro-phenoloxidases ( Figure S1). No correlation was found between the amount of dengue detected and the level of expression for each of the antimicrobial peptide and pro-phenoloxidases genes tested in each fly line ( Figure S2).

Discussion
Host immune priming by Wolbachia offers an appealing mechanistic explanation for pathogen blocking as it is conceivable that this single effect could lead to protection against a diversity of pathogens. The objective of this study was to compare the effect of two closely-related strains of Wolbachia on the immune system of hosts where the age of the Wolbachia association differs. By comparing wMelPop-CLA and wMel we could exclude any potential immune activation that may simply be due to the virulence of the wMelPop-CLA infection. By examining both D. melanogaster and A. aegypti, we were able to dissect aspects of the immune response that may be attributed solely to a host's response to a recently acquired Wolbachia infection. This analysis depends on an assumption that the mechanism of virus interference is similar in the two insect hosts. Considering that Wolbachia infection in Drosophila interferes with dengue replication, as it does in A. aegypti, the assumption of a similar mechanism seems parsimonious. Moreover the success of maintaining dengue in Drosophila, even if viral replication is not as strong as in A. aegypti, provides a tractable genetic model for future studies into the mechanistic basis of Wolbachia-mediated dengue interference.
A previous analysis of A. aegypti whole genome transcription in response to wMelPop-CLA revealed strong immune induction by the bacterium [10]. In this present study, a similar approach was taken to analyze the impact of the non-virulent wMel strain on the immune system of A. aegypti, in comparison with the wMelPop-CLA strain. The results obtained revealed that wMel induces the activation of far fewer immunity genes in the mosquito. The comparative analysis between the different lines identified common responses only for genes encoding anti-microbial Transcripts are ranked by biological process and/or molecular function. Gene identifiers (Gene ID), Description and Symbol were compiled from Flybase. AFC, Absolute Fold Change, ND, No Detection. Asterisks indicate a statistically significant difference (n = 10, Mann-Whitney U test with q-value adjustment, *: q,0.05, **: q,0.01). doi:10.1371/journal.ppat.1002548.t004 peptides, the Toll pathway and melanization-associated proteins. Recent studies have provided important insights into A. aegypti immune response to dengue virus, showing that the Toll pathway and anti-microbial peptides are important for the mosquito's defense against dengue infection [13,25]. Melanization is also a prominent immune response in insects against parasites like malaria and nematodes [26] but as far as we know it has never been demonstrated for dengue. The main anti-viral pathway, RNA interference [27], seems to be activated exclusively by wMelPop-CLA. Several pieces of evidence also indicate that RNAi cannot explain virus blocking. First, Glaser et al [28] showed that even in Ago2 (a key gene in the RNAi pathway) mutant flies, Wolbachia infection increases resistance to viruses. Second, Frentiu et al [29] demonstrated that wMelPop-CLA induces complete inhibition of dengue virus replication in the C6/36 cell line that has been shown to be defective in the RNAi pathway [30].
This comparative analysis between wMel and wMelPop-CLA infection within A. aegypti supports the potential implication of anti-microbial peptides and Toll pathway activation in dengue virus interference by the bacterium. If we assume that the fundamental mechanism involved in Wolbachia-mediated dengue interference is the same in mosquitoes and flies, and this mechanism is immune-based, then the same constitutive immune induction should also be observed in D. melanogaster infected by wMel or wMelPop. We tested for transcriptional changes of the same immune genes identified through microarray analysis in D. melanogaster in response to Wolbachia infection, and identified a number of statistically significant changes. However, in no case were these changes consistent between wMel and wMelPop infection. Furthermore, if we employed the same threshold for biological significance we used for our microarray data, that a gene is significantly up-regulated by Wolbachia infection only when its level is changed at least 1.5-fold compared with non-infected flies, we would conclude that wMel did not constitutively prime any of the different immune genes tested in its natural host D. melanogaster. Those results are in accordance with previous data showing no pre-activation of different immune genes in D. melanogaster, D. simulans and A. albopictus by Wolbachia [14,15].
In summary, the only immune genes up-regulated by wMelPop-CLA and wMel in A. aegypti are anti-microbial peptides, Toll pathway and melanization genes. However, the same Wolbachia strains did not up-regulate these genes in Drosophila, and yet dengue interference occurs in this host. This indicates that the up-regulation of these immune effector genes is not required to interfere with dengue virus replication, although it is likely that the immune up-regulation that occurs in mosquitoes, presumably due to the recent association with Wolbachia, might enhance this effect.

Insect rearing
All the mosquito strains used in this study were laboratory lines of A. aegypti infected with wMel (MGYP2) or wMelPop-CLA (PGYP1), and their tetracycline-treated uninfected counterparts, MGYP2.tet and PGYP1.tet [7,8]. Adult mosquitoes were kept on 10% sucrose solution at 25uC and 60% humidity with a 12-h light/dark cycle. Larvae were maintained with fish food pellets (Tetramin, Tetra).
The fly experiments were performed with w 1118 fly lines stably infected with wMel (w 1118 wMel) [31] and wMelPop (w 1118 wMel-Pop) [18] compared to the tetracycline-cured lines derived by the addition of tetracycline (0.3 mg/ml) to the adult diet for two generations. Those lines were confirmed to be free of Wolbachia by PCR, using primers specific for the wMel and wMelPop IS5 repeat [22]. Females were kept under controlled conditions, low-density (30 females per vial), at 25uC with 60% relative humidity and a 12-h light/dark cycle.

Sample collection and hybridization
Three replicate pools of 20 female mosquitoes, 8 days posteclosion were collected from each of the four lines (PGYP1, MGYP2, PGYP1.tet and MGYP2.tet), snap frozen in liquid nitrogen and extracted for total RNA using Trizol (Invitrogen). RNA was then purified using RNeasy kits (Qiagen) according to manufacturer's instructions. Whole-genome microarrays were then used to compare gene expression in the Wolbachia-infected lines relative to uninfected controls, using a dual-color reference design. All sample preparations and hybridizations were then carried out by the IMB Microarray Facility at the University of Queensland. Briefly, sample quality was examined using the Agilent 2100 Bioanalyzer (Agilent Technologies) and fluorescent cDNA was synthesized using Agilent Low RNA Input Linear Amplification Kit with Cy3 or Cy5. Each infected line and respective paired tetracycline-treated line was represented by 3 biological replicates (3 pools above). A total of 6 hybridizations were then carried out for each biological replicate, 3 labeled with cy3 and three with cy5 (dye swaps).

Microarray design
Microarrays were of the 4644 K format (Agilent) each containing standard control features and 3 technical replicates of each 60 mer feature randomly distributed across the layout. The A. aegypti genomic sequence (Vectorbase genome build 1.1) was used for construction of oligonucleotide microarrays using eArray Version 5.0 (Agilent Technologies). After removing probes that cross hybridized, a total of 12,336 transcripts that represented 12,270 genes were spotted onto each microarray [32].

Microarray data analyses
For each transcript, raw data was extracted and analyzed using Genespring v.9.0 (Agilent Technologies). An intensity dependent (Lowess) normalization (Per Spot and Per Chip) was used to correct for non-linear rates of dye incorporation as well as irregularities in the relative fluorescence intensity between the dyes. Hybridizations from each mosquito line were used as replicate data to test for significance of expression changes using the cross-gene error model. The occurrence of false positives was corrected using the q-value [33,34]. All array data have been deposited in ArrayExpress (http://www.ebi.ac.uk/microarray-as/ ae/) under the accession number E-MEXP-2931.
Functional annotations of A. aegypti genes were retrieved from Biomart [35] in Vectorbase [36] and analyzed using the Ontologizer software with the parent child intersection method [37,38]. The over-expression of particular GO categories in the microarray data set was tested against the distribution of GO categories for the A. aegypti genome.

Virus injection
Dengue virus serotype 2 (DENV-2), strains 92T [9] and ET300 were isolated from human serum collected from patients from Townsville, Australia, in 1992 and East Timor in 2000, respectively. DENV-2 (strains 92T and ET300) was propagated and quantified as described by Frentiu et al [29]. For virus injection, 8 day old D. melanogaster females (w 1118 wMel and w 1118 tet) and A. aegypti females (MGYP2 and MGYP2tet) were briefly anesthetized with CO 2 and injected under a dissecting scope into their thorax with a pulled glass capillary and a handheld microinjector (Nanoject II, Drummond Sci.). 69 ml of virus stock (10 7 pfu/ml) or sterile PBS 1X were injected. After injection flies and mosquitoes were maintained under identical controlled conditions, low-density (10 females per vial or cup), at 25uC with 60% relative humidity and 12-h light/dark cycle. Insects were collected 8 days post-injection and kept at 280uC for RNA extraction.
Quantitative DENV-2 PCR analysis RNA extraction was done on 15 individual 16 day old females per condition using Trizol (Invitrogen). 1 mg of total RNA was kept to quantify DENV-2 while the rest was used for immune gene expression analysis as described below.
Accumulation of genomic (+RNA) RNA strands was assessed by quantitative real time PCR using hydrolysis probes specific to the 39 UTR region of the four dengue serotypes [39] with modifications (A.T. Pyke, unpublished data). The sequences of the primers were FWD: 59-AAGGACTAGAGGTTAGAGGAGACCC-39 and RWD: 59-CGTTCTGTGCCTGGAATGATG-39 and the sequence of the probe was 59-AACAGCATATTGACGCTGGGA-GAGACCAGA-39. 1 mg of total RNA for each sample was mixed with 0.625 mM of the reverse primer plus 0.2 mM dNTPs. Samples were incubated at 86uC for 15 minutes and 5 minutes on ice, then 5X first strand buffer and 100 U of Superscript III (Invitrogen) was added to a total volume of 20 ml. Samples were incubated at 25uC for 10 minutes, followed by 42uC for 50 minutes and 10 minutes at 95uC to inactivate the transcriptase.
The qPCR reaction consisted of 2 ml of the synthesized cDNAs, 5 ml of 2X LightCycler 480 Probes Master (Roche), 0.5 mM of each primer (see above) and 0.5 mM of the probe (see above) in 10 ml total volume. Reactions were performed in duplicate in a LightCycler 480 Instrument (Roche) with the following conditions: 95uC for 5 minutes, and 45 cycles of 95uC for 10 s, 60uC for 15 s, 72uC for 1 s. A standard curve was created by cloning the DENV-2 39UTR region fragment into pGEMH T-Easy (Promega). After linearization with Pst I the plasmid was serially diluted into known concentrations and run in parallel, in order to determine the absolute number of DENV-2 copies in each 1 mg of total RNA. First, percentages of individuals infected with dengue were calculated for each treatment. Then only individuals with dengue infection (non zero quantification) were used to examine the effect of wMel on dengue titer using Mann-Whitney U tests (Graph Pad Prism 5).

Quantitative PCR analysis of immune genes
RNA extraction from flies was done using between 10 to 15 individual 8 day old females per condition using Trizol reagent (Invitrogen). To eliminate any contamination by DNA, samples were treated with DNase I recombinant (Roche), in accordance with the manufacturer's instructions. cDNAs were synthesized from 1 mg of total RNA, using oligodT primers and the SuperScript III enzyme (Invitrogen), in accordance with manufacturer's instructions. For each sample qRT-PCR was performed in triplicate on a 10 times dilution of the cDNAs using Platinum SYBR Green (Invitrogen) according to the manufacturer's protocol. Primers are listed in Table S4. The temperature profile of the qPCR was 50uC for 2 minutes (UDG incubation), 95uC for 2 minutes, 45 cycles of 95uC for 5 s, 60uC for 5 s, 72uC for 10 s with fluorescence acquisition of 78uC at the end of each cycle, then a melting curve analysis after the final cycle. The housekeeping gene rpS17 was used to normalize expression. Target gene to housekeeping gene ratios were obtained for each biological replicate using Q-Gene software [40]. Raw data were graphed as median 6 interquartile range (IQR) and outliers beyond 1.5 IQR excluded. Treatment effects on expression ratios were then examined using the Mann-Whitney U tests (Graph Pad Prism 5). The occurrence of false positives was corrected using the q-value [33,34]. Figure S1 Immune gene expression in Drosophila melanogaster in response to wMel and DENV-2. The expression of immune genes was analyzed by qRT-PCR on individual females injected either with DENV-2 strain 92T (w 1118 wMel D+, w 1118 tet D+) or PBS (w 1118 wMel PBS, w 1118 tet PBS) in presence/absence of Wolbachia strain wMel. Flies were collected 8 days post-injection. Graphs show the target gene to house-keeping gene expression ratio (n = 15, Mann-Whitney U test with q-value adjustment, *: q,0.05, **: q,0.01, ***,0.001). (TIF) Figure S2 Correlation analysis between dengue titer and immune gene expression in Drosophila melanogaster in presence/ absence of Wolbachia strain wMel (w 1118 wMel, w 1118 tet). The values were compared using Spearman correlation coefficients. (TIF)

Supporting Information
Table S1 Aedes aegypti transcriptional responses common to wMel and wMelPop-CLA infections. Transcripts are ranked by the magnitude of Absolute Fold Change (AFC). Transcript identifiers (Transcript ID) and Description were compiled from Vectorbase.