The Transcriptional Response of Drosophila melanogaster to Infection with the Sigma Virus (Rhabdoviridae)

Background Bacterial and fungal infections induce a potent immune response in Drosophila melanogaster, but it is unclear whether viral infections induce an antiviral immune response. Using microarrays, we examined the changes in gene expression in Drosophila that occur in response to infection with the sigma virus, a negative-stranded RNA virus (Rhabdoviridae) that occurs in wild populations of D. melanogaster. Principal Findings We detected many changes in gene expression in infected flies, but found no evidence for the activation of the Toll, IMD or Jak-STAT pathways, which control immune responses against bacteria and fungi. We identified a number of functional categories of genes, including serine proteases, ribosomal proteins and chorion proteins that were overrepresented among the differentially expressed genes. We also found that the sigma virus alters the expression of many more genes in males than in females. Conclusions These data suggest that either Drosophila do not mount an immune response against the sigma virus, or that the immune response is not controlled by known immune pathways. If the latter is true, the genes that we identified as differentially expressed after infection are promising candidates for controlling the host's response to the sigma virus.


Introduction
Viral infections in arthropods are widespread and are of considerable economic and medical importance. For example, viruses have had devastating economic consequences on honeybee and shrimp populations [1,2], and many viral pathogens in humans, crops and livestock are vectored by insects. Carefully chosen model systems could provide great insight into how arthropods combat viral infections. The principal model organism used to study invertebrate immune defenses is the fruit fly Drosophila melanogaster [3]. Although many studies have examined Drosophila's defenses against bacterial and fungal infections, relatively little is known about antiviral defenses. This is despite D. melanogaster being infected with a diverse range of viruses, including positive-stranded RNA viruses (several picornaviruses, including the Drosophila C virus), a negative-stranded RNA virus (sigma virus; Rhabdoviridae), and a double-stranded RNA virus (DFV; Reoviridae) [4].
The viruses that infect Drosophila have very different lifecycles and biology, which may have important implications for immune recognition and the immune response. The picornaviruses are released by lysing host cells, and the viral particles are non-enveloped [5]. In contrast, the Rhabdovirus sigma is released from host cells by budding, and the viral particles are enclosed in a lipid envelope with surface-exposed glycoproteins [4,6]. Furthermore, the picornavirus DCV can cause severe pathology in infected flies, while the sigma virus is relatively benign [4]. Antiviral immune responses often recognize RNA viruses by the presence of dsRNA. Typically, positive-sense RNA viruses produce much more double stranded RNA than negative sense viruses, probably because the nucleocapsid protein of negatively sensed RNA viruses can prevent the two strands from annealing to produce dsRNA [7]. All of these factors may mean that the mechanisms by which flies can recognize viruses and protect themselves against infection may be differ between different viruses.
The only immune effector that has been found to target viruses in Drosophila is RNAi [8][9][10]. RNAi can distinguish self from non-self because RNA viruses often produce double-stranded RNA, while host cells typically do not. Some RNA viruses have a double-stranded genome, and even single stranded RNA viruses can produce doublestranded RNA during replication, gene expression or as a consequence of RNA secondary structure. RNAi processes this double-stranded RNA into short fragments, which are passed to an effector complex that recognizes and degrades viral RNA with the complimentary sequence. RNAi pathway genes are constitutively expressed, and are not known to be up-regulated by viral infection [11].
Other studies have examined the changes in gene expression that occur when flies are infected by viruses [11][12][13]. Microarray analyses of flies injected with DCV identified many up-regulated genes, raising the possibility that there is an induced immune response to viruses in addition to RNAi [11]. Detailed studies of one of the up-regulated genes identified in this study, vir-1, revealed that it was under the control of the Jak-STAT pathway, which is an important component of the antiviral response of vertebrates. Not only does DCV activate this pathway, but flies that are deficient for the Jak-kinase Hopscotch have a higher viral load and lower survival than wild-type flies [11]. It is currently unknown how the Jak-STAT pathway detects viral infection, or how it protects flies against DCV (knocking down vir-1 expression does not make flies more susceptible to DCV).
In addition to the Jak-STAT pathway, the Toll pathway, an immune signaling pathway activated by bacteria and fungi, has been implicated in antiviral immunity. A previous study has shown that the Drosophila X virus (DXV) activates the Toll pathway, and flies that are deficient for the Toll pathway transcription factor Dif are more susceptibility to DXV infection [13]. Furthermore, a gene called ref(2)P, which is required by the Toll immune response, has a naturally occurring polymorphism that reduces the rate at which sigma virus replicates within the fly [14,15]. However, the role of the Toll pathway as a antiviral response remains uncertain because previous studies have shown that Toll pathway genes are not up-regulated by DCV infection [11], and not all Toll pathway mutants alter the flies' susceptibility to DXV [13].
In this study we have used microarrays to see which immune pathways, if any, are up-regulated when Drosophila is infected with the sigma virus-a naturally occurring pathogen that infects about 4% of D. melanogaster in the wild [16]. The sigma virus is transmitted only vertically, from parent to offspring through both eggs and sperm [4]. Typical for a vertically transmitted pathogen, the sigma virus is a fairly benign infection that causes a slight reduction in the survival and fecundity of infected flies [17]. There is however a great deal of genetic variation in the susceptibility of flies to sigma infection in natural populations [18,19]. Several major effect resistance genes have been mapped [4], but only one of these has been identified (ref(2)P; see above). It is currently unclear whether these resistance genes are part of an antiviral immune response mounted by the flies or are host molecules that the virus exploits for its own replication and transmission. The only study which looked at the transcriptional response to sigma virus infection measured the expression of 15 immunity-related genes using quantitative realtime PCR (qPCR), and found that several antimicrobial peptides and the peptidoglycan recognition proteins (PGRP-SB1 and PGRP-SD) are up-regulated in infected flies [20]. However, it is unclear whether this transcriptional response has any effect on the replication or transmission of the virus.

Drosophila stocks and hybridizations
We compared patterns of gene expression in genetically identical flies that were either infected or uninfected with the sigma virus. The flies were an isogenic stock SM5/Pm;spa pol that had been infected with the sigma virus isolate AP30 several generations before (see [19] for details). These flies have the susceptible allele of the ref(2)P gene which controls sigma virus replication. Four replicates of both the infected and uninfected flies were reared on Lewis medium [21] at a constant density of approximately 220 flies per bottle at 25uC on a 12 hour light-dark cycle for a minimum of three generations before the experiment. The flies were aged and allowed to mate for six days before being sexed on ice (they were not exposed to CO 2 ), and RNA was extracted from pools of 180 males or 60 females using Trizol (Invitrogen, Carlsbad, CA, USA). We confirmed the infection status of a sample of flies from the same bottles using a standard CO 2 test [19].
We performed 5 dye swap replicates (10 arrays) on the males and 4 dye swap replicates (8 arrays) on the females. Each dyeswap compared a different pair of RNA extractions from sigmainfected and -uninfected flies (i.e., biological replicates). A more detailed description of the hybridizations and statistical analysis is given by Hutter et al. [22], who used many of the same methods. We used a D. melanogaster microarray obtained from the Drosophila Genomics Resource Center (DGRC; Bloomington, IN, USA) known as DGRC-1. This consists of 13,921 exonic PCR amplicons (100-600 bp in length) representing 11,895 unique genes (,88% of the genome).
The RNA was reverse transcribed and labeled with the SuperScript Plus Indirect cDNA Labeling System and Alexa Fluor 555 and 647 dyes (Invitrogen, Carlsbad, CA, USA). Hybridizations were performed following DGRC protocols and arrays were scanned using an aQuire microarray scanner (Genetix, New Milton, UK). All array data have been submitted to the Gene Expression Omnibus database (http://www.ncbi.nlm. nih.gov/geo) under series XXX.

Data analysis
To correct each spot on our arrays for local background effects, within-array variation and between-array variation, we normalized the signal intensity of the two dye channels using the three-step procedure described by [22] and implemented in CARMAweb [23]. In short, the relative expression level and probability of differential expression between infected and uninfected flies for each gene was estimated using the Markov Chain Monte Carlo algorithm implemented in BAGEL [24]. The results of the BAGEL runs are provided as Supporting Files S1 and S2. As some genes are represented by multiple probes, we defined a gene as differentially expressed if at least one of its probes displayed a significant difference. For each slide, only those spots displaying a signal greater than 95% of the negative control probes (182 probes from other species) in each dye channel were considered 'expressed', and 'non-expressed' data points were excluded from the analysis. To determine the experimentwide false discovery rate (FDR), we repeated the BAGEL analysis on a randomized version of our final data-set [22]. To estimate the power of our experiment to detect expression differences between infected and uninfected flies, we calculated the GEL 50 statistic [25].
To test if any gene ontology (GO) categories were overrepresented in our list of differentially expressed genes, we used the web-based tool g:Profiler [26] that corrects for multiple testing while taking the hierarchical nature of GO terms into account. The analysis of gene ontology was based on the annotation of the D. melanogaster genome included in release v49 of ENSEMBL [27]. CG numbers were updated to match this version and these lists are provided separately in the Supporting File S3.

Quantitative real-time PCR
To confirm the results of the microarray analyses, we measured the expression of several genes using qPCR. These genes included PGRP-SC2 and Tudor-SN, whose expression differed between sigma-infected and -uninfected flies in our microarray experiment, and Attacin-A and Drosocin, which were found to be up-regulated in sigma-infected by Tsai et al. [20], but not in our experiments. As an endogenous control, we also measured the expression of RpL32 (Rp49).
To check the infection status of our flies and estimate their viral loads we also amplified viral genomic RNA by qPCR. The primers were designed to amplify a fragment spanning the sigma N and P genes to ensure that they amplified genomic RNA rather than mRNA. To allow our data on data to be compared to the results of Tsai et al. [20], we used Act88F as an endogenous control.
For the qPCR, 1.1 mg of total RNA was reverse transcribed using Superscript II reverse transcriptase (Invitrogen) and random hexamer primers. The resulting cDNA was used at 1:10 dilution for qPCR using TaqMan probes and a 7500 Fast Real-Time PCR System (Applied Biosciences, Foster City, CA, USA). Pre-designed probe IDs were as follows: PGRP-SC2: Dm01818611_s1, Tudor-SN: Dm01834411_g1, Attacin-A: Dm02362218_s1, Drosocin: Dm01821449_s1, RpL32: Dm02151827_g1 and Act88F: Dm02362815_s1. Probes for quantify-ing viral loads were designed using the Custom TaqMan Assay Design Tool provided by Applied Biosciences. The region included in this assay corresponds to positions 3127 to 3239 of the AP30 isolate sequence (EMBL Accession AM689309) with the following primer and probe sequences: 59-GCTCACAGTGAAGATCCATTACATG-39 (forward), 59-GCGGCTTCACAGAGAATTTGTC-39 (reverse) and 59-ACGAGATCTTAGTCAGCACCCT-39 (probe). Seven replicate assays were performed for each of the 4 treatments: male, female, infected and uninfected and the threshold cycle value (Ct) was averaged across these replicates.

Data quality and identification of differentially expressed genes
In total we performed 10 microarray hybridizations on males and 8 on females.
After removing probes that had no signal in either the sigmainfected or -uninfected flies, we were left with 4301 probes representing 3923 genes in females, and 5532 probes representing 4996 genes in males. The complete list of these genes is given in Supporting File S1. The higher number of genes detected as expressed in males may be explained by sex-related differences in gene expression, as we detected expression of a higher proportion of sex-biased genes in males than in females (see Discussion).
To assess the statistical power of the experiment, we calculated the GEL 50 statistic, which is the fold-change in gene expression at which we have a 50% chance of detecting a difference (P,0.05) between the infected and uninfected flies. The GEL 50 was 1.37 for the males and 1.52 for the females, which is similar to other microarray studies [22].
In order to define a significance threshold we calculated the FDRs for both the male and female datasets at different P-values (Figure 1a). We used a cut-off of P,0.02 for the male dataset, which corresponds to an FDR of 8.6%. Using this threshold we found that a total of 629 genes showed expression differences between sigma-infected and -uninfected flies (in infected flies 293 genes were up-regulated and 336 genes were down-regulated). Adjusting the P-value to produce a comparable FDR for the female dataset resulted in a very short list of only ,30 differentially expressed genes. For further analysis we therefore decided to also use a cut-off of P,0.02 for females. This corresponds to an FDR of 21% (Figure 1a). At this threshold, we detected 134 differentially expressed genes (in infected flies 46 genes were up-regulated and 88 genes down-regulated). The excess of down-regulated genes in females was statistically significant (x 2 -test, P = 0.0003).
We found that the transcriptional response to sigma infection was very different between males and females. Of the 2862 genes detected as expressed in both males and females, only 41 showed a significant difference in expression in both sexes, and of these, 35 showed a consistent pattern of up-or down-regulation in both sexes (Table 1). Overall the magnitude of these differences was small; for both males and females, the maximum difference in expression was 2.5-fold ( Figure 2) and the median difference was 1.28. Despite this, there seems to be a striking difference between males and females in the number of genes affected by infection. In order to investigate if this was due to differences between our replicates we removed the replicate hybridizations with the best quality signal from the male dataset and repeated the calculations. Removing the four bestquality hybridizations lowered the detection sensitivity to below that of the female dataset, and yet, males still showed consistently greater differences in expression (Figure 1b, see Discussion).

Expression of known immunity genes
Previous studies have shown that a large number of genes are induced when flies are infected with bacteria or fungi, and that many of these genes are under the control of the Toll and IMD immune signaling pathways. To investigate whether these pathways might be activated in response to sigma virus infection, we have compared our results to previous studies that examined the fly's immune response to different pathogens. First, we compared data from a previous microarray study, which examined gene expression in flies infected with bacteria and fungi and identified some 400 upor down-regulated genes [28], to our list of genes that showed a significant change in regulated in response to sigma-infection. From this comparison, it is clear that there is little overlap between the transcriptional response to bacteria and fungal infections and that of the sigma virus (Table 2). Of the genes that did overlap between the two studies, seven genes were up-regulated in both cases (Table 2). These include the antimicrobial peptide Metchnikowin, a translational regulator that is important in immune defense (Thor) and five other genes (CG13323, CG10912, CG16743, CG9928, CG15293). Seventeen genes were down-regulated in both studies including four Jonah proteases (Jonah 25Bii, Jonah 25Biii, Jonah 65Ai and Jonah 25Bi), two related calcium binding proteins that play an important role in many cellular processes (regucalcin and Smp-30) and 11 other genes (fit, CG18594, CG18179, CG12813, CG9090, CG4019, CG13947, CG7322, CG9672, CG9914 and CG3699).
Next, we investigated whether sigma activates specific immune signaling pathways (Figure 3). Initially, we investigated how the expression of immunity genes that are known to be induced by either the Toll or IMD pathway changes in response to sigma infection. These genes were also selected to overlap with Figure 1 of Dostert et al. [11], who performed a similar analysis to this study, but investigated changes in gene expression in response to DCV-infection. The genes we selected fall into three groups-genes involved in the Toll pathway, the IMD pathway and the Jnk pathway. We looked at the regulation of IM2 and Drosomycin that are up-regulated by the Toll pathway [29,30], and three antimicrobial peptides -Cecropin A1, Diptericin and Drosocinthat are up-regulated by the transcription factor Relish within the IMD pathway [29]. It should be noted that some of these antimicrobial peptides might also be under the control of the Toll pathway [31]. Finally, we looked at Act88F, fln, Mlc1 and TpnC41C, which are upregulated by the Jnk pathway [29]. It is clear from Figure 3 that there is no evidence that any of these groups of genes have been up-regulated in response to sigma-infection. In a similar analysis, we compared our data to lists of genes under the control of the Toll and IMD-Relish pathways that were identified by De Gregorio et al. [28], and again there is no evidence that these pathways are activated (Table 3). We confirmed that there was no significant difference in the expression of Attacin A, Drosocin or Act88F by qPCR (Table 4). There has only been one published study on the transcriptional response of Drosophila to sigma virus infection, which used qPCR to measure the expression of a selection of immunity genes [20]. This study found that four antimicrobial peptides and two PGRPs were up-regulated in infected flies. In our study none of these genes were significantly up-regulated in infected flies, despite all six being detected as expressed in both males and females. For two of the genes (Drosocin and Attacin A), we confirmed this result using qPCR on both the male and female samples ( Table 4). The difference between our results and those of Tsai et al. [20] is not due to our flies having lower viral titers, as when we measured the copy number of sigma virus genomic RNA relative to Act88F by qPCR, we found that our flies had higher viral titers than in this previous study (males Finally, we compared our data with a microarray analysis of DCV-infected flies (Dostert et al., 2005[32]; Jean-Luc Imler, personal communication). Few genes were up-regulated by both DCV and the sigma virus-of the 85 genes that were significantly up-regulated in response to DCV infection, only nine were upregulated in response to the sigma virus, and seven were instead down-regulated in sigma-infected flies. There was a greater The X-axis defines the magnitude of expression difference between the infected and uninfected state, the Y-axis the corresponding P-value of the BAGEL analysis. Probes for which BAGEL assigned a P-value of 0 (i.e., P,0.0001), were set to P = 0.0001. Black dots represent probes up-regulated (log 2 infected/uninfected.0) or down-regulated (log 2 infected/ uninfected,0) in the infected state at P,0.02. doi:10.1371/journal.pone.0006838.g002 overlap in the genes that were down-regulated in both DCV and sigma-infected flies-of the 200 genes that were significantly down-regulated in response to DCV infection, 30 were downregulated in response to the sigma virus, and 11 were up-regulated in sigma-infected flies. Overall, the association between the two studies is marginally non-significant (Fisher Exact Test on 262 contingency table of differentially expressed genes: P = 0.06).
There were however a few notable immune-related genes that were differentially expressed. The genes PGRP-SC2, which is important in dampening the immune response [33], was one of the few genes to be down-regulated in both our male and female datasets (females: ratio infected/uninfected = 0.67, P = 0.005; males: ratio = 0.81, P = 0.015). We replicated this result using qPCR (females: ratio = 0.66, P = 0.011; males: ratio = 0.93, P = 0.95). Similarly, Tudor-SN, which is involved in RNAi, was down regulated in females, but this was not repeatable using qPCR (Table 4).

Up-and down-regulated genes
To investigate which biological processes are affected by sigma virus infection, we identified gene ontology (GO) terms that were overrepresented among our up-and down-regulated genes ( Table 5). A selection of the genes with these GO terms is listed in Table 6. Among the genes that were down-regulated in sigmainfected males, ribosomal proteins were overrepresented (29 of the 93 genes in one of the two ribosomal subunits). Virtually all of the other GO categories overrepresented among the genes downregulated in infected males are related to mitochondria. In sigmainfected females, serine proteases were overrepresented among the down-regulated genes, including seven genes from the chymotrypsin superfamily, and six from the Jonah family (note that these categories overlap in the GO annotations). Six chymotrypsin genes were also down-regulated in infected males, including three of the same genes as were detected in infected females. Among the genes  that were up-regulated in infected males, genes related to DNA binding and regulating transcription were overrepresented. While in infected females, chorion structural proteins are overrepresented, with three of the nine genes in the genome being significantly up-regulated. Furthermore, three additional chorion proteins (CP18, CP19 and CP36) showed up-regulation with P-values very close to our detection threshold (P = 0.023, P = 0.024 and P = 0.039 respectively).

Immune pathway activation
When Drosophila is infected by bacteria or fungi, the Toll and IMD pathways are activated, leading to the up-regulation of large numbers of genes. These genes include immune effectors such as antimicrobial peptides that are secreted into the hemolymph and defend the flies against the invading pathogens. It is currently unclear whether there is a comparable induced immune defense against viruses. We found that neither genes up-regulated by bacterial or fungal infection, nor the subset of these controlled by the Toll and IMD pathways, are induced in sigma-infected flies. As these pathways control the majority of the genes up-regulated by fungal and bacteria infections [31], any induced immune response to the sigma virus must be controlled by distinct regulatory mechanisms. And although it has been reported that DXV activates the Toll pathway [13] and sigma-virus activates the IMD pathway [20], our results are more similar to with microarray analyses of DCV that found no evidence for the Toll or IMD pathways being activated [11].
An additional pathway implicated in viral infection in Drosophila is the Jak-STAT pathway-genes within the Jak-STAT pathway have been shown to be up-regulated in response to DCV infection, and flies deficient in this pathway are more susceptible to DCV. However, we found that there is little overlap between the genes induced by DCV and the sigma virus, suggesting that the Jak-STAT pathway is not activated by sigma virus. Therefore, we conclude that it is unlikely that Drosophila mounts a general immune response to all viruses.
Why is there little overlap between the genes induced by sigma virus and DCV? One possibility is that DCV, unlike the sigma virus, causes cells lysis and it is this rupturing of cells, and the ensuing tissue damage, that induces an immune response [3]. This hypothesis is consistent with the observation that when flies are fed DCV that results in a relatively benign infection [4] and far fewer genes are induced compared to when flies are injected with the virus [11,12]. Alternatively, the sigma virus may avoid recognition by the immune system for other reasons. As the sigma virus is only transmitted vertically, it must establish a persistent infection and therefore it must avoid being cleared by the immune response. Avoiding inducing an immune response may be essential for persistent vertically transmitted infections, as the vertically transmitted bacteria Spiroplasma poulsonii and Wolbachia do not induce an immune response in D. melanogaster either [34,35]. Vertically transmitted infections will also be selected to minimize the harm that the cause to the host, as they rely on their host surviving and reproducing to be transmitted. This may also select for viruses that do not induce a costly transcriptional response in their host.
A previous study by Tsai et al. [20] found that four antimicrobial peptides and two PGRPs were strongly induced by the sigma virus. Despite our flies having a higher viral titer, none of these genes were induced in our study. Therefore, the Table 3. The effect of the sigma virus on the expression of genes controlled by the Toll and IMD pathways (identified by [31]

Up-and down-regulated genes
What genes changed in expression in sigma infected flies? In males, the down-regulated genes were strongly enriched for proteins involved in translation, especially ribosomal proteins. Viruses rely on the host's translational machinery to produce proteins, and many viruses-including related virus vesicular stomatitis virus (Rhabdoviridae)-both inhibit host translation and cause viral mRNAs to be preferentially translated over host mRNAs [36]. Furthermore, depleting ribosomal proteins inhibits the replication of DCV in Drosophila cells [37]. The down-regulation of genes involved in translation by the sigma virus probably reflects this interaction, but the biological importance is unclear. We found that serine proteases were down-regulated in infected flies. Serine proteases have a wide range of functions, including key roles in the regulation of the immune system. The Toll and IMD responses to bacteria and fungi induce and repress a number of serine proteases, some of which have important roles in immune regulation [31,38]. In females there was an overrepresentation of serine proteases that were down-regulated. Of particular note was an excess of chymotrypsin superfamily serine protease, and several Jonah serine proteases. Several of these genes were also downregulated in sigma-infected males, making these genes interesting candidates for controlling the fly's response to viruses.
We found that in infected females, genes encoding for chorion proteins were up-regulated. During oogenesis, the chromosomal copies of these genes are amplified up to 10 times, allowing high levels of gene expression [39]. The sigma virus is transmitted through the fly's eggs and it is possible that this relates to the upregulation of these genes.

Sex differences in gene expression
Sigma virus infection appears to alter the expression of many more genes in males than in females. For example, with an FDR of 10% we detect over 10 times as many significant genes in males as in females (Figure 1b). Several factors could contribute to this difference. First, the two experiments differ slightly in their replication schemes (10 arrays for males versus 8 arrays for females) and statistical power (GEL 50 = 1.37 for males and 1.52 for females). Thus, we have more power to detect expression differences in the male experiment. This alone, however, cannot explain the large discrepancy in the number of differentially expressed genes. If we exclude two (GEL 50 = 1.43) or even four (GEL 50 = 1.57) of the male replicates, we still detect a larger fraction of significant genes in the male experiment (Figure 1b). A second factor could be sex-biased gene expression, as the genes that we detected differ between the two experiments. Using the sex-bias classifications of Gnad and Parsch [40], we find that 59% of the male-biased genes on the array are detected in males, while only 16% are detected in females. In contrast, 43% of femalebiased genes on the array are detected in females, while only 28% are detected in males. Sex-biased genes, however, are not overrepresented among those whose expression was significantly altered by the sigma virus. Male-biased genes comprise 18% of the genes detected in males, but only 14% of the genes significantly affected by sigma infection. Similarly, female-biased genes comprise 18% of the genes detected in females, but only 11% of the genes significantly affected by the sigma infection.
It is also possible that the difference between males and females results from intersexual differences in the mechanism of viral transmission or host defense. Sigma virus is transmitted along with sperm and there appear to be specific barriers to it entering the male germline [4]. Furthermore, genes that cause variation in the transmission of the virus often affect transmission through sperm, but not eggs [19]. This may reflect a qualitative difference in the nature of the infection in males, or indicate that specific resistance responses are triggered in the male germline, leading to a greater transcriptional response to infection. Alternatively, males and females may invest differently in their immune defenses [41], and this may be reflected in sex differences in the transcriptional response to sigma infection.
A final possibility is that, in general, male gene expression is more sensitive to genetic and/or environmental perturbations than female gene expression. Some support for this comes from the observation that, among laboratory strains, greater gene expression variation is observed among males than among females [42]. However, more experimental comparisons of male and female transcriptional responses to a common treatment are necessary to determine the generality of this observation.

Supporting Information
File S1 The relative expression level, 95% confidence limits of expression levels in males and females, and probability of differential expression of each gene between infected and uninfected males Found at: doi:10.1371/journal.pone.0006838.s001 (1.28 MB XLS) File S2 The relative expression level, 95% confidence limits of expression levels in males and females, and probability of differential expression of each gene between infected and uninfected females Found at: doi:10.1371/journal.pone.0006838.s002 (1.00 MB XLS) File S3 The list of significantly differentially expressed genes (P,0.02) in which the CG numbers have been updated to match v49 of ENSEMBL Found at: doi:10.1371/journal.pone.0006838.s003 (2.20 MB XLS)