Wolbachia Infections in Anopheles gambiae Cells: Transcriptomic Characterization of a Novel Host-Symbiont Interaction

The endosymbiotic bacterium Wolbachia is being investigated as a potential control agent in several important vector insect species. Recent studies have shown that Wolbachia can protect the insect host against a wide variety of pathogens, resulting in reduced transmission of parasites and viruses. It has been proposed that compromised vector competence of Wolbachia-infected insects is due to up-regulation of the host innate immune system or metabolic competition. Anopheles mosquitoes, which transmit human malaria parasites, have never been found to harbor Wolbachia in nature. While transient somatic infections can be established in Anopheles, no stable artificially-transinfected Anopheles line has been developed despite numerous attempts. However, cultured Anopheles cells can be stably infected with multiple Wolbachia strains such as wAlbB from Aedes albopictus, wRi from Drosophila simulans and wMelPop from Drosophila melanogaster. Infected cell lines provide an amenable system to investigate Wolbachia-Anopheles interactions in the absence of an infected mosquito strain. We used Affymetrix GeneChip microarrays to investigate the effect of wAlbB and wRi infection on the transcriptome of cultured Anopheles Sua5B cells, and for a subset of genes used quantitative PCR to validate results in somatically-infected Anopheles mosquitoes. Wolbachia infection had a dramatic strain-specific effect on gene expression in this cell line, with almost 700 genes in total regulated representing a diverse array of functional classes. Very strikingly, infection resulted in a significant down-regulation of many immune, stress and detoxification-related transcripts. This is in stark contrast to the induction of immune genes observed in other insect hosts. We also identified genes that may be potentially involved in Wolbachia-induced reproductive and pathogenic phenotypes. Somatically-infected mosquitoes had similar responses to cultured cells. The data show that Wolbachia has a profound and unique effect on Anopheles gene expression in cultured cells, and has important implications for mechanistic understanding of Wolbachia-induced phenotypes and potential novel strategies to control malaria.


Introduction
Wolbachia are alpha-proteobacteria that infect a range of arthropods and nematodes, and are possibly the most common endosymbiotic bacteria on the planet. In their arthropod hosts, Wolbachia induce a variety of reproductive manipulations that enhance the fitness of infected females compared to their uninfected counterparts [1]. Wolbachia have recently been shown to interfere with pathogen infection and transmission in both naturally-infected and artificially-transinfected insects [2,3,4,5,6,7]. These phenotypes make Wolbachia-based control strategies an attractive option to minimize the impact of arthropod-borne diseases and insect pests [8,9].
Anopheles mosquitoes transmit human malaria, a devastating disease that kills approximately 2 million people per year, and are naturally uninfected with Wolbachia [10,11,12]. Transfer of Wolbachia into cultured Anopheles gambiae cells and transient somatic infection of adult female mosquitoes demonstrates that the bacteria can survive in this species, suggesting that the Anopheles genus may be amenable to stable infection [13,14]. Although several novel Wolbachia-mosquito associations have been created using a variety of transinfection techniques, no stable Wolbachiainfected Anopheles line has been developed [15,16,17,18,19,20]. The development of such a strain may open the possibility for Wolbachia-based control strategies for malaria. Indeed somatic infections of the wMelPop strain reduce oocyst levels in the murine malaria model, Plasmodium berghei [7]. However the global effects of Wolbachia on Anopheles and the interplay within the tripartite association of the human malaria Plasmodium parasites and the mosquito host are currently unknown.
Novel phenotypes are sometimes observed upon transinfection of Wolbachia into novel insect hosts [15,21,22]. In the artificially infected wMelPop-Aedes aegypti strain (wMelPop CLA), Wolbachia limits infection by a broad range of pathogens including dengue virus, filarial nematodes and Plasmodium [2,3]. The mode of action for pathogen resistance is uncertain, however two mechanisms have been postulated; immune activation of the host by Wolbachia and/or metabolic competition between the bacteria and the pathogen. Evidence for both hypotheses was observed with a range of immune genes up-regulated in wMelPop-infected Ae. aegypti [2,3] and the finding that dengue virus only persisted in Wolbachia-uninfected cells of the insect [3]. A similar phenotype was observed in some infected Drosophila strains where Wolbachia infection induced refractoriness to multiple RNA viruses [4,5]. Interestingly, a previous study using naturally infected hosts found that Wolbachia seems to be able to evade the host immune response in Drosophila and Aedes albopictus [23], suggesting Wolbachia-induced immune activation may be more likely in novel rather than coevolved Wolbachia-host associations.
Within Anopheles mosquitoes, there is a conserved immune response towards foreign bacteria and Plasmodium [24]. By using multiple methods such as co-feeding, injection or removal of microflora, bacteria have been seen to mediate Plasmodium infection levels in the Anopheles host [25,26,27,28], which is thought to be due to the bacteria priming the host immune response. Interestingly, Gram-negative bacteria elicit a greater response compared to Gram positive, although there are speciesspecific differences [25,27]. If Wolbachia (a Gram-negative bacterium) evokes a similar response and up-regulates the basal immunity in infected Anopheles, infection may confer an anti-Plasmodium phenotype. Some evidence for this has been shown in somatically-infected mosquitoes infected with rodent malaria [7].
The generation of Wolbachia-infected Anopheles cell lines allows the investigation of Wolbachia-Anopheles interactions in the absence of a stably-infected mosquito strain [14]. Cell lines provide a platform whereby Wolbachia host lineages can be generated with relative ease, and allow the exploration of both natural and artificial Wolbachia host interactions [29,30,31]. To investigate the effect of Wolbachia infection on global patterns of Anopheles gene expression we performed microarray analysis on both wAlbB (from Ae. albopictus) and wRi (from Drosophila simulans) infected Anopheles gambiae Sua5B cells compared to uninfected cells. We validated microarray results in vitro, and in vivo for a subset of differentially expressed genes in somatically-infected adult female mosquitoes.

Results/Discussion
Wolbachia infection of Anopheles cells resulted in the regulation of 690 genes relative to uninfected Sua5B cells (False discovery rate (FDR) P,0.05, $ 2.0 fold-change (FC)) (Table S1). When comparing Wolbachia strains, 255 genes were uniquely regulated by wAlbB infection, while 331 were regulated specifically by wRi infection ( Figure 1A). Of the 104 genes regulated by both strains, the majority (74 genes) were down-regulated, 11 were similarly upregulated and the remainder had alternating regulation patterns between the two Wolbachia strains ( Figure 1B). Interestingly, we observed a greater number of genes regulated by wRi compared to wAlbB even though the cell infection density of wRi was much less than wAlbB (wRi,10% cells infected, wAlbB .90% of cells infected) [14]. It is possible that since wRi was purified from live flies, it has a greater impact that wAlbB which was purified from another cell line [14]. Of the regulated genes, a diverse range of functional classes was represented with a large proportion being genes of unknown or diverse function, which was consistent for both Wolbachia strains. Among the genes assigned to specific known functional classes, immune-, transport-and metabolismrelated transcripts were the most abundant categories regulated by Wolbachia ( Figure 1C). Strikingly, over 75% of the immune related transcripts were down-regulated, which was consistent between both strains. Overall, down-regulation was a common theme, with only redox/stess/mitochondrial (RSM) and replication/transcription/translation (RTT) classes not down-regulated in wRi infected cells and RTT in wAlbB. Microarray data is available at gene expression omnibus (accession number GSE23215) [32].

qPCR validation of microarray genes and comparison to whole mosquitoes
To gauge the accuracy of the microarray data, we selected a subset of genes to validate by quantitative real-time PCR from cell culture. Eight genes, (HSP20, HSP90, HSPDnaJ, cold-shock protein, cecropin, Serpin11, Filamin, TEP3) with varying expression profiles, regulated by both Wolbachia strains were evaluated. These genes spanned a variety of functional classes including defensive and immune genes that may be relevant to Plasmodium infection and potential Wolbachia-mediated reproductive phenotypes. qPCR results corroborated the array data and had a positive linear correlation (R 2 = 0.9595) when comparing the log2 values using both gene expression techniques (Figure 2A).
Wolbachia has been shown to persist, disseminate, and replicate in injected adult Anopheles mosquitoes [13]. We injected live female mosquitoes with Wolbachia to determine if the effect of infection on gene regulation in vivo was consistent with results observed from infected cell cultures. Several immune related transcripts and other genes, which potentially convey interesting phenotypes and had varying expression profiles identified in cell culture, were assessed. When comparing wAlbB regulation in cells and mosquitoes, the direction of regulated expression was similar ( Figure 2B), although, not surprisingly, the intensity of expression varied leading to a lack of significant correlation (data not shown). The loss of the wRiinfected cell line prevented a direct comparison to somatically infected mosquitoes, however, this array data was compared to wMelPop-infected mosquitoes. wMelpop and wRi both infect Drosophila and are classed in supergroup A. When making this comparison, again we observed that the direction of gene regulation was similar ( Figure 2C), but the intensity of expression varied. Notably, the intensity of two genes, the LRR-like transcript

Author Summary
Wolbachia are bacteria that infect many insect species, but do not infect Anopheles mosquitoes. These mosquitoes transmit Plasmodium parasites, which cause malaria in humans. Wolbachia infection in Aedes aegypti mosquitoes reduces their ability to transmit diverse pathogens including viruses, nematodes and bird malaria parasites. Wolbachia-induced stimulation of the mosquito's innate immune system has been suggested as a mechanism conferring this pathogen interference. Since no Wolbachiainfected Anopheles mosquito strain exists, we used infected cultured Anopheles cells to examine the effect of infection on Anopheles gene expression. Wolbachia had a profound influence on Anopheles gene expression. Many of the genes regulated by Wolbachia have been seen in other studies to influence Plasmodium levels in mosquitoes, but interestingly and in contrast to other mosquitoes, many of the host genes were suppressed rather than induced. and FBN9 is greater in the cells compared to the whole mosquito. This may be explained by the hemocyte-like character of the cell line or Wolbachia strain-specific variation. Nevertheless, the similarity in the direction of gene regulation in vivo and in vitro suggests that the effect of Wolbachia in the cell line may be applicable to whole mosquitoes.

Comparison to other systems
We compared Wolbachia-regulated Anopheles transcripts identified in this study to genes regulated by wMelPop in Aedes aegypti [2] and by bacterial infection in A. gambiae [26]. Fourteen A. aegypti homologues were identified from differentially expressed Anopheles transcripts in response to Wolbachia infection, with five having an immune related function (Table S2). When comparing these results, 75% of both wRi and wAlbB regulated homologs displayed a similar direction of expression. Similarly, when comparing Wolbachia-regulated transcripts to those of regulated by bacterial infection in A. gambiae, 15 homologs were regulated by Wolbachia. (Table S3). Most of these homologs were of unknown function.
In comparison to other studies using Drosophila cell culture systems to examine the influence of Wolbachia on host gene expression, we find a dramatically elevated number of identified regulated Anopheles genes compared to Drosophila. In Drosophila S2 cells, 263 genes had a 1.2 fold change due to Wolbachia infection, however when the more common $2 fold criteria was used, very few regulated Drosophila genes were identified [33]. At the proteomic level, only four proteins, all host antioxidant proteins, were elevated in Wolbachia infected Ae. albopictus Aa23 cells [34]. A lower Wolbachia titer may account for the subtle gene regulation in wRi infected Drosophila cells [33], although the infection density of wRi in infected Anopheles was similarly sparse [14]. Alternatively, the mild effect of Wolbachia on gene regulation in Drosophila and protein expression in Ae. albopictus could be due to previous coevolution between the Wolbachia strains and their naturally infected hosts.

Effect of Wolbachia on transcription of Anopheles genes potentially affecting pathogen transmission
Stress-response. The most striking effect observed for both wRi and wAlbB infections was the general suppression of heat shock protein transcripts (HSP20, HSP70, HSP90, HSP-DNAJ). Cells infected with wAlbB had a dramatic suppression of these genes with 5 out of the top 6 most down-regulated genes (FC 231 to 216). Similarly, these genes were down-regulated by wRi, albeit to a lesser extent (to 25.3). Presenting a similar pattern of regulation, multiple HSPs were down-regulated by wRi infected Drosophila S2 cells [33]. In vivo, it has been shown that Wolbachiainfected flies have altered expression of HSP, which in turn affects Wolbachia-induced reproductive phenotypes [35]. HSPs have also been implicated in Anopheles-pathogen interactions. Elevated levels of HSP20 were identified in An. gambiae heads after infection with P. berghei [36]. If this protein assists transmission, either directly or indirectly, the antagonistic actions of may potentially reduce P. berghei sporozoite infection. Additionally, knockdown of a heat shock proteins (HSC70B) via injection of dsRNAi in conjunction with O'nyong nyong virus (ONNV) significantly reduced the lifespan of adult mosquitoes as compared with the control [37]. We speculate that if this expression pattern translates to in vivo Anopheles infections, Wolbachia-induced down-regulation of HSPs may modulate vector competence of ONNV or shorten mosquito lifespan.
Metabolic and other genes. Wolbachia regulates a suite of genes involved in Anopheles metabolism, with most of these transcripts being down-regulated by infection. Although the heterotrophic needs of Plasmodium and mosquito growth factors required for parasite development are not well understood in the insect, changes in transcription of metabolism genes which alter the mosquito environment may affect Plasmodium growth. Infection of Sua5B cells with wAlbB drastically reduces phosphoenolpyruvate carboxykinase (PEPCK) transcripts 26 fold. In response to P. falciparum, PEPCK is up-regulated in the mosquito [38,39]. Carbonic anhydrase, which catalyses the reversible hydration of carbon dioxide to bicarbonate, is down-regulated in wAlbB-infected cells by 2.6 fold. In many mosquitoes, inhibition of this enzyme results in a reduction in pH of the mosquito midgut [40]. Moreover, carbonic anhydrase inhibitors in P. falciparum reduced parasite survival in the human blood stages and have been suggested as targets of anti-malarial drug design [41,42]. The effect of these host derived enzymes on parasite development is unknown, however changes in regulation between mosquito and Plasmodium suggest that further examination of these genes is warranted to determine their affects on parasite development. Although not strictly metabolism related, laminin and collagen are components of the basal lamina, which are interrelated with parasite invasion [43,44,45]. Both laminin (FC 22.1, 23.8) and collagen (FC 24.4) are downregulated by wRI infection. RNAi knock down of laminin lead to a substantial reduction of oocysts in mosquito midguts [44] possibly due to laminin inhibiting the melanotic encapsulation of oocysts [46].
Immunity-related transcripts. Many Anopheles genes associated with arthropod immunity were regulated by Wolbachia infection. Genes within all the broad categories of immunity (pathogen recognition receptors, signaling amplification cascades, immune signaling pathways, and effector molecules) were regulated. Immune genes up-regulated by both infections included CLIPs and antimicrobial peptides (AMP), while serpins (SRPN), and a leucine rich repeat (LRR) were induced by wRi and fibrinogens (FBN) and thioester-containing protein (TEP) were induced by wAlbB ( Figure 3). More striking were those immune genes down-regulated by infection. wRi significantly suppressed expression of class C scavenger receptors, Gram-negative binding proteins (GNBP), FBN, CLIP, SRPN, LRR-containing genes, a TEP, effector proteins involved in phagocytosis and a lysozyme ( Figure 3). The wAlbB strain down-regulated genes of similar functions, however in the class of effector molecules, this strain had more of an influence on peroxidases rather than AMPs (Figure 3).
In addition, other immune-associated apoptosis and detoxification transcripts were regulated by infection. Brennan et al. [34] identified Wolbachia-induced host antioxidant proteins in cell culture. In contrast to the enrichment of these genes at the protein level, a peroxiredoxin transcript was down-regulated 2.1 times by wAlbB and 11 times by wRi. Likewise, superoxide dismutase was down-regulated in wRi-infected cells (FC 22). Additionally, eight glutathione S transferases were regulated. Two of these were co-regulated by both strains, while 3 were induced and 3 suppressed in wRi. The level of regulation for these genes was approximately 2-3 fold, however one transcript was suppressed 19-fold by wRi compared to uninfected cells. Taken in total, these data suggest that Wolbachia can significantly affect cellular defense, detoxification and immunity in An. gambiae cells, and that expression of many of these defensive genes is suppressed rather than induced. These results contrast with observations of up-regulation of the majority of immunerelated transcripts in stably-infected Ae. aegypti mosquitoes, which have reduced capacity to transmit pathogens [2,3]. Gene expression of a small subset of immune genes were characterized in response to wMelPop infection of a different An. gambiae cell line (Mos55), where they were up-regulated, suggesting a potential difference between Anopheles cell lines or Wolbachia strain-specific variation [7].
Although pathogen interference occurs in naturally infected hosts, there is evidence that the transfer of Wolbachia to a new host is a catalyst for pathogen interference, illustrated by wAlbB inducing dengue resistance in a novel host, Ae. aegypti, yet not conferring interference in it's native host, Ae. albopictus [6]. The effects of tripartite relationship of Wolbachia-Anopheles-Plasmodium are relatively unknown, however, recently wMelPop somatically infected into Anopheles was seen to decrease P. berghei oocyst levels, with evidence that TEP1 may involved in the process [7]. Many of the regulated defensive genes we identified have been shown to directly or indirectly affect Plasmodium infection in Anopheles, either positively or negatively. TEP3 was dramatically up-regulated (FC 7.6) in response to wAlbB. Similar up-regulation is observed when mosquitoes are fed a blood meal, either uninfected or infected (P. berghei), or challenged with bacteria [47,48]. TEP1, a protein similar to TEP3, has been shown to be an important molecule involved in the melanization and anti-Plasmodium response across the Anopheles genus [49,50]. Looking at genes involved in the immune signaling cascade, CLIP7A, a suppressor of melanization, was suppressed by both wAlbB (FC 25.2) and wRi (FC 22.6), which may confer an anti-Plasmodium phenotype as seen in knockdown experiments of this gene [51]. In contrast, the gene galectin, which is up-regulated in response to P. berghei infection and immune challenge by Micrococcus luteus, had conflicting strainspecific responses: up-regulated by wAlbB (FC 9.1) but downregulated by wRi (FC 23.5) [52].
In contrast to genes that may abate Plasmodium infection, a suite of genes were also regulated in ways that may elevate parasite levels in infected mosquitoes. For example we observed downregulation of many CLIPs. Reverse genetic techniques have shown that both CLIPB4 and CLIPB8 are involved in the melanization process, where knock-down of these genes ablates melanization [53]. In double knock-down (KD) experiments, reducing transcripts of both CLIPB4 and CLIPB8 in tandem with CTL4 partially interferes with P. berghei ookinete melanization [51]. Using over-expression, up-regulation of cecropin was shown to decrease Plasmodium levels in Anopheles [46]. Expression of both SRPN18 (FC wAlbB 23.2, wRi 23.6) and TEP15 (FC wAlbB 23.5, wRi 22.1) is suppressed by both Wolbachia strains and although the specific function of these molecules has not been identified, these classes of molecules are associated with immunity [48,54]. In Ae. aegypti, TEP15 is one of the most strongly induced genes in response to KD of Cactus, the negative regulator of the Toll pathway [55]. In addition, GNBPB1, which was also down-regulated by both strains (FC wAlbB 25.2, wRi 26.0), is strongly induced by parasite invasion of the midgut and bacterial challenge [52,56,57]. In contrast to our study, GNBP was induced in Aedes mosquitoes infected with wAlbB and wMelpop [2,6].
In terms of a general response to bacterial infection, we see the regulatory transcriptional factor for the Toll pathway (Rel1) downregulated 2.3 times by wRi infection. We observed an upregulation of caspar (FC 2.2), the negative regulator of the IMD pathway in response to wRi. PGRP-LA expression was suppressed 3.2 times by wRi. In Drosophila, PGRP-LA is likely to be a hemocyte transmembrane protein [58], while other PGRPs activate negative feedback loops in the IMD pathway [59,60]. A similar long transcript PGRP (PGRP-LC) in An. gambiae controlled proliferation of gut microbiota, which subsequently influenced Plasmodium infection [61]. When all three PGRP-LC isoforms were silenced simultaneously, mosquitoes challenged with Staphylococcus aureus had induced expression of cecropin and defensin. In Drosophila, silencing of PGRP-LC by RNAi induced expression of diptericin, cecropin A1, and attacin A, but these effector molecules were not regulated due to depletion of PGRP-LA [62].  [81] and components of the melanization regulatory module [51] divided into the 4 broad categories of immune molecules. Blue color represents induction, while yellow color represents suppression. The intensity of coloring is proportional to the intensity of expression. Regulation is depicted to a maximum fold change of 64. Some transcripts were greater than 64 Here we see similar independent regulation of attacin which was up-regulated 3.3 times in wRi infected cells, while defensin is also up-regulated by wAlbB (FC 2.3). Interestingly, attacin was found to inhibit the outer membrane synthesis of Escherichia coli in the giant silk moth, Hyalophora cecropia [63]. Thus, we may be observing an active defensive response from Anopheles to prevent Wolbachia infection.
The general pattern of immune gene down-regulation appears to be a Wolbachia-specific phenomenon in this cell line. In addition to Wolbachia, Sua5B cells can support infection of additional intracellular bacteria such as Rickettsia [64]. We used qPCR to test selected immune-related genes (cecropin1, defensin1, gambicin and immune-responsive serpin-related protein [IserpF1]) in Sua5B cells that had been infected with a taxonomically and phenotypically diverse array of Rickettsia species: R. typhi (typhus group), R. felis (transitional group), R. montenensis and R. peakockii (both in the spotted fever group). R. typhi and R. felis are human pathogens, while R. montenensis is non-pathogenic. R. peakockii is a nonpathogenic vertically-transmitted tick endosymbiont. While there was variation between bacterial species and the gene tested, all four Rickettsia induced expression of most tested immune genes (up to 12-fold induction), including the endosymbiont R. peakockii ( Figure S1). These results suggest that the natural response of Sua5B cells to intracellular bacterial infection is immune upregulation, and that Wolbachia is suppressing this response. It should be noted however that Wolbachia exist in a potentially protective host vacuole, while Rickettsia are free in the cytoplasm.

Wolbachia influence on reproduction-related genes
Wolbachia-induced CI expression is associated with abnormal decondensation of the paternal pronucleus during fertilization, epigenetic factors, and/or problems during embryogenesis. Xi et al. [33] observed that in wRi-infected Drosophila cells, the gene angiotensin converting enzyme (Ance), which is involved in spermatogenesis, was up-regulated by Wolbachia infection in cells and flies, and was potentially involved in the CI phenotype. In our study, the six Anopheles homologues of Ance on the microarray were not affected by Wolbachia infection. We screened our data for other significantly regulated genes associated with cytoskeleton formation/function, epigenetic modification, gametogenesis or embryonic development. Multiple cytoskeleton-associated genes, genes associated with chromatin formation and remodeling and genes associated with embryogenesis and cell division were regulated by both infections.
We identified multiple genes that may be linked to the CI phenotype. Transcription of a Kazal-like serpin was enhanced dramatically due to Wolbachia infection (FC wAlbB 13.1, wRi 5.3). Kazal domain-containing proteins identified in animals have a diverse array of functions. A Kazal-like serpin was found to inhibit both gelatinolytic activity of sperm and the proteolytic activity of sperm extracts to vitelline coat in prawns [65], while in mice, a serine protease inhibitor Kazal-type-like protein bound to sperm, enhancing motility and suppressing sperm capacitation [66]. Although in these two species the function of the Kazel-like serpin is varied, it has the commonality that it interferes with sperm-oocyte interactions. Up-regulated (FC 2.3) in wRi-infected cells, crooked neck (crn) transcripts are involved in embryogenesis. In it's recessive form, crn is lethal to embryos, while heterozygotes display a crooked phenotype [67]. In both Drosophila and humans, crn has been implicated in the mRNA splicing process and is thought to be a pre-mRNA splicing factor [68,69]. Another gene induced by wRi (FC 3.1), otefin, codes for a nuclear laminin which is essential for germ cell maintenance in Drosophila [70]. A further candidate protein, Dumpy-30 (Dpy-30) is expressed in spermatids in Drosophila, and mutations or knockout of the male-specific dpy-30L2 gene results in male sterility as mutant sperm have impaired motility and fail to accumulate in sperm storage organs of females [71]. In Anopheles cells, wAlbB up-regulates (FC 2.0; significant at unadjusted P,0.05) Dpy-30, and although the effect of over-expression is unknown, this could potentially have a role in the CI phenotype. Serine active site containing (Serac1) mediates sterility in mice [69] and is up regulated (FC 2.4) by wRi infection. TEP15, suppressed by both strains (FC wAlbB 23.5, wRi 22.2), may influence reproduction. TEP15 is a male accessory glands protein and is transferred to female in the mating plug [72]. It would be interesting to determine if Wolbachia-induced regulation of these transcripts is Anopheles specific or common to other insect species infected with Wolbachia.
In addition to these genes, heat shock proteins were dramatically down regulated by both bacterial strains, but the effect was most dramatic by wAlbB. HSPs have been associated with sperm production and are inferred to be involved in CI [35,73]. A range of chaperone proteins were also up-regulated by wRi, including a cold shock protein (FC 4.8) multiple DNAJ heat shock proteins (FC 3.3, 2.1), GrpE protein (FC 2.7), and a ubiquilin-1 gene (FC 2.3).

Pathogen related phenotypes
Other identified regulated genes may have behavioral implications for infected Anopheles. It has been reported that some older wMelPop-infected Ae. aegypti mosquitoes have ''bendy'' and ''shaky'' phenotypes [74,75]. The proboscis of ''bendy'' individuals is flexible and unable to penetrate the skin [74]. Mosquitoes with the ''shaky'' phenotype have a jittering action of the insect body [75]. Here, we have identified genes that may elucidate these phenotypes at the molecular level. Both Wolbachia strains suppress the defective proboscis extension response (dpr) gene (wAlbB 23.3, wRi 22.6). Moreover, this gene is also down regulated in Wolbachia-injected mosquitoes (wAlbB 23.0, wMelPop 22.4; Figure 2B & 2C). In Drosophila, dpr is part of a gene family encoding predicted cell adhesion molecules that contain two Ig domains [76]. It is possible that a reduction in cell adhesion causes plasticity in the proboscis leading to the ''bendy'' phenotype. In addition to reduced dpr transcripts, Wolbachia down-regulated numerous other cell adhesion genes. Interestingly, dpr also has been shown to be required for the proper timing of male courtship [76], and given that Anopheles have elaborate swarming courtship behaviors in the wild, Wolbachia infection may have the potential to alter reproductive success.
Sestrins (Sesn), a family of conserved proteins, accumulate in cells in response to stress and are inhibitors of target of rampamycin (TOR) that prevent age-related pathologies [77,78]. In wAlbB-infected cells, we see a down regulation of Sesn (FC 23.5). In Drosophila dSesn-null mutants, age related degeneration of muscle was observed in the form of cardiac malfunction and abnormal skeletal muscle [78]. Possibly, suppression of Sesn in wMelpop infected Ae. aegypti is related to the ''shaky'' phenotype [75]. Moreover, it would be interesting to correlate Sesn levels in both Drosophila and Ae. aegypti infected with wMelpop, which display life shortening and age related pathologies [17,79], to determine if Sesn plays a role in life shortening from this strain of regulated. Abbreviations: LLR leucine rich repeats; FBNs fibrinogens; TEPs thioester containing proteins; GNBPs Gram-negative binding proteins; CTLs C type lectins; CLIPs clip-domain serine protease; PGRPs peptidoglycan recognition proteins; SRPNs serpins; CEC cecropins; Def defensins; PPO Prophenoloxidase; PO phenoloxidase; LYS lysozmyes. doi:10.1371/journal.ppat.1001296.g003 Wolbachia. The ''shaky'' and ''bendy'' phenotypes are more prevalent in older Wolbachia infected Aedes mosquitoes [75]. If the genes identified here confer the ''bendy'' and ''shaky phenotypes in a Wolbachia-infected Anopheles mosquito, these effects could be more influential on malaria transmission compared to direct pathogen interference.

Conclusion
Wolbachia-infected mosquito cells provide a tractable platform to characterize Wolbachia-Anopheles transcriptomic interactions in the absence of a stably-infected mosquito strain. Using this system, we identified a suite of Anopheles genes regulated by two divergent Wolbachia strains. As a general theme, Wolbachia have a profound effect on transcription of many host defensive genes, possibly to facilitate and maintain intracellular infection. These data may give insights into the transfer of Wolbachia into novel hosts, Anopheles-Wolbachia interplay, interaction with pathogens transmitted by Anopheles and other Wolbachia-induced phenotypes such as reproductive manipulations.

Cell culture
Wolbachia-infected (wRi and wAlbB) and uninfected Sua5B cells were generated and cultured as previously described [14]. Both cell lines were .30 passages post-infection at the time of experiments. Cell line transcriptome expression was assessed using the Affymetrix Anopheles/Plasmodium GeneChip. Processing of samples for microarray analysis was performed by the Johns Hopkins Malaria Research Institute Gene Array Core Facility (JHMRI-GACF), using standard Core protocols as described below.

RNA extraction
Cells were harvested, washed, resuspended in PBS, flash frozen in liquid nitrogen, and stored at 280uC. Homogenization and lysis of cells was performed with Lysing Matrix D (Qbiogene) in Trizol LS reagent (Invitrogen) by rapid agitation in a FastPrep 120 Instrument (Qbiogene) for 15 seconds at speed setting 5. Homogenates were subsequently processed according to the manufacturer's (Invitrogen) protocol with the following minor modifications. Two microliters of 5 mg/ml glycogen was used as a carrier for overnight isopropanol precipitation, and all centrifugation times were increased to 15 minutes. RNA pellets were resuspended in Nuclease-free water. Further purification was performed using the Qiagen RNeasy Mini kit, according to manufacturer's recommended protocol. Quantitation of RNA was performed using a NanoDrop spectrophotometer, and quality assessment determined by RNA Nano LabChip analysis on an Agilent BioAnalyzer 2100.

Affymetrix GeneChip protocols
Processing of templates for GeneChip Analysis was in accordance with methods described in the Affymetrix GeneChip Expression Analysis Technical Manual, Revision 5. Double stranded cDNA was synthesized from 5 micrograms of total RNA using the GeneChip Expression 39 amplification reagents one-cycle cDNA synthesis kit (Affymetrix), and subsequently column-purified using the GeneChip Sample Cleanup Module. Biotinylated cRNA was synthesized from the double-stranded cDNA by in vitro transcription (IVT) using the GeneChip Expression 39 amplification reagents for IVT labeling (Affymetrix), according to the manufacturer's recommended protocol. Resultant cRNAs were purified by column purification with the GeneChip Sample Cleanup Module (Affymetrix), and quantified. 15 micrograms of cRNA were fragmented by metal-induced hydrolysis in fragmentation buffer (250 mM Tris acetate pH 8.1, 150 mM MgOAc, 500 mM KOAc) at 94uC for 35 minutes. Quality of preand post-fragmentation cRNAs was assessed by RNA Nano LabChip analysis on an Agilent Bioanalyzer 2100. Hybridization cocktails were prepared as recommended for arrays of ''Standard'' format including incubation at 94uC for 5 minutes and 45uC for 5 minutes, and centrifugation at maximum speed for 5 minutes prior to pipetting into the GeneChips (Affymetrix Plasmodium/Anopheles). Hybridization was performed at 45uC for 16 hours at 60 rpm in the Affymetrix rotisserie hybridization oven. The signal amplification protocol for washing and staining of eukaryotic targets was performed in an automated fluidics station (Affymetrix FS450). Arrays were scanned in a GeneChip 3000 7G laser scanner with autoloader (Affymetrix) at an emission wavelength of 570 nm and 2.5 mm resolution. Intensity of hybridization for each probe pair was computed by GCOS software.

Data analysis
Detailed analysis was performed with Genomics Suite Software, version 6.4 (Partek). GC-RMA algorithm defaults were used for background correction (GC-RMA), normalization (Quantile), and summarization (median polish) of probesets. Analysis of variance (ANOVA) was performed with linear contrasts for each Wolbachia treatment (strain) vs. control. Gene lists were developed based on 2 fold or greater gene expression and a False Discovery rate P,0.05 criteria. Lists were annotated manually. Immune gene networks were developed using Pathvisio2 [69].

qPCR verification of expression analysis
Using qPCR, microarray data were validated using infected cell cultures and also somatically-infected mosquitoes. Live female mosquitoes (2 days post emergence) were immobilized on ice and transferred to an electronic cold plate. Mosquitoes were injected with Wolbachia (wMelpop or wAlbB) or Schneider's medium as described previously [13]. Although a standard protocol was followed for Wolbachia preparations, titers were not explicitly standardized. Injected mosquitoes were incubated at 19uC for 2 days before transfer to 28uC (80% humidity) insectary and were provided with access to a 10% sucrose solution through a cotton wick. After 15 days, mosquitoes were collected and RNA was extracted using TriReagent (Ambion) following manufactures guidelines. For verification of microarray data, total RNA was extracted from Sua5B cell lines (uninfected, wAlbB-infected, or wRi-infected) using the RiboPureTM kit (Ambion) following the manufacturer's instructions. RNA from cells or mosquitoes was DNase treated (Ambion) and cDNA synthesized using superscript III (Invitrogen) following manufactures guidelines. qPCR was performed in triplicate on an AB 7300 Sequence Detection System using the QuantiTect SYBR Green PCR Kit (Qiagen). Analysis was performed using Sequence Detection Software v.1.3 (ABI). Relative quantitation was completed by normalizing gene of interest to the ribosomal protein S7 gene (primers listed in Table 1) and data analyzed using the comparative Ct method (DDCt method) [80].