Abstract
Free full text
Pseudomonas aeruginosa Suppresses Host Immunity by Activating the DAF-2 Insulin-Like Signaling Pathway in Caenorhabditis elegans
Abstract
Some pathogens have evolved mechanisms to overcome host immune defenses by inhibiting host defense signaling pathways and suppressing the expression of host defense effectors. We present evidence that Pseudomonas aeruginosa is able to suppress the expression of a subset of immune defense genes in the animal host Caenorhabditis elegans by activating the DAF-2/DAF-16 insulin-like signaling pathway. The DAF-2/DAF-16 pathway is important for the regulation of many aspects of organismal physiology, including metabolism, stress response, longevity, and immune function. We show that intestinal expression of DAF-16 is required for resistance to P. aeruginosa and that the suppression of immune defense genes is dependent on the insulin-like receptor DAF-2 and the FOXO transcription factor DAF-16. By visualizing the subcellular localization of DAF-16::GFP fusion protein in live animals during infection, we show that P. aeruginosa–mediated downregulation of a subset of immune genes is associated with the ability to translocate DAF-16 from the nuclei of intestinal cells. Suppression of DAF-16 is mediated by an insulin-like peptide, INS-7, which functions upstream of DAF-2. Both the inhibition of DAF-16 and downregulation of DAF-16–regulated genes, such as thn-2, lys-7, and spp-1, require the P. aeruginosa two-component response regulator GacA and the quorum-sensing regulators LasR and RhlR and are not observed during infection with Salmonella typhimurium or Enterococcus faecalis. Our results reveal a new mechanism by which P. aeruginosa suppresses host immune defense.
Author Summary
Bacterial pathogens have evolved mechanisms to overcome the immune defenses that animals and plants deploy against them. In some cases, this involves directly interfering with the proper functioning of the immune system. Because pathogens that employ these strategies are often the most deadly and difficult to treat, it is important to understand how they are able to suppress the immune system in the context of the whole organism. In this paper, we show that Pseudomonas aeruginosa, a bacterial pathogen that is a major contributor to hospital-borne infections such as pneumonia, suppresses an immune defense pathway during infection of the simple animal host Caenorhabditis elegans. Using genetic modifications of both the pathogen and host, we identify components of the signaling pathways required to suppress host immune defenses. We find that P. aeruginosa employs the cell-to-cell communication system known as quorum sensing, which coordinates the expression of virulence factors to suppress host immune defense. In the host, an evolutionarily conserved insulin-like signaling pathway is affected by P. aeruginosa, resulting in the suppression of genes that are required for defense against infection in the intestinal epithelial cells. These findings suggest the possibility that P. aeruginosa may exploit similar mechanisms when causing infections of human epithelium, such as the epithelial lining of the lungs.
Introduction
The innate immune system is a genetically-encoded host defense mechanism that constitutes the first line of defense against pathogens in plants and animals. Innate immunity in animals is evolutionarily ancient, and the molecular components of mammalian innate immunity are partly conserved in invertebrates such as Drosophila and C. elegans that lack a somatic recombination-based adaptive immunity. Innate immunity comprises several functions, including the production of defense proteins, such as antimicrobial peptides, lysozymes, and other immune modulators [1]. A common theme emerging from the studies of host-pathogen interactions is that recognition of pathogen-associated molecular patterns (PAMPs) or of infection byproducts by host receptors, such as the Toll-like receptors (TLRs), trigger highly regulated immune responses, including the induction of antimicrobial effectors [2],[3]. For example, infection by Gram-positive bacteria triggers activation of the Toll pathway in Drosophila. Activation of Toll signaling in turn induces expression of specific antimicrobial genes through the activation of a Rel/NF-κB transcription factor [4],[5]. Despite lacking a functionally conserved TLR/NF-κB pathway [6], C. elegans is able to mount robust immune responses against a variety of pathogens (see [7]–[10] for examples). This underscores the importance of other pathways in C. elegans innate immunity. Recently, through genetic studies, several conserved signal transduction pathways that are required for innate immunity in C. elegans have been identified. They include the p38 MAPK, the Sma/TGF-β, and the DAF-2/DAF-16 insulin-like signaling pathways (reviewed in [11],[12]). For example, mutants in sek-1, which encodes a p38 MAPK kinase, and pmk-1, which encodes a p38 MAPK, are sensitive to killing by bacterial pathogens [13]. Mutants in sma-6, which encodes a TGF-β receptor, are also sensitive to killing by bacterial pathogens [10],[14]. In contrast, mutants in daf-2, which encodes an insulin-like receptor, are resistant to killing by bacterial pathogens. The resistance of daf-2 mutants is completely dependent on DAF-16, a FOXO transcription factor [15]. Microarray studies suggest that each of these immune signaling pathways regulates the expression of host effector genes, which may account for the altered pathogen susceptibility of pathway mutants [8],[16],[17].
Host defense effectors include diverse classes of small molecules and antimicrobial peptides. Lysozyme, a bacteriolytic enzyme, is ubiquitously expressed in mammalian secretions. The C. elegans genome encodes ten lysozyme-like proteins (lys-1 to lys-10), some of which have been directly implicated in host defense [7],[10]. The C. elegans genome also encodes many amoebapore or saposin-like proteins (e.g., spp-1) that are members of a large and diverse class of antimicrobial peptides and a number of defensin-like molecules homologous to ASABF (Ascaris suum antibacterial factor), including abf-2. In mammals and Drosophila, the expression of defensin-family proteins contributes to antibacterial, antifungal, and antiviral defenses. Recombinant SPP-1 and ABF-2 have antimicrobial activity [18],[19], and endogenous expression contributes to antibacterial defense in C. elegans [20]. The C. elegans genome also encodes homologs of the thaumatin family of plant antifungal proteins (thn-1 to thn-8). While it is not known whether C. elegans thaumatins have antifungal activity, RNAi knockdown of thn-family genes results in worms that are more sensitive to killing by P. aeruginosa [7]. Host effector genes are differentially regulated during infection of C. elegans [7]–[10]. Microarray studies also suggest that different pathogens elicit specific transcriptional responses [21]. This is presumably due to host recognition of different PAMPs followed by induction of specific host defense pathways. For example, during fungal infection, TIR-1 (Toll and IL-1 receptor) activates an antifungal defense through p38 MAPK signaling [22]. The mechanisms by which specific transcriptional responses are elicited are still largely unknown. However, the intestine-specific GATA transcription factor ELT-2 is required for the regulation of host immune defense effectors and resistance to bacterial pathogens [7],[23].
P. aeruginosa is an important Gram-negative human pathogen that is associated with infection of immunocompromised patients [24], including cystic fibrosis (CF) patients and individuals with burn wounds [25],[26]. P. aeruginosa has evolved at least three strategies to combat the vast repertoire of host defenses. First, P. aeruginosa can produce an extensive array of cell-associated and secreted virulence factors that are deleterious or damaging to the host. A substantial number of these virulence determinants are regulated by the two-component regulator encoded by gacA and the quorum-sensing regulators encoded by lasR and rhlR [27],[28]. For example, P. aeruginosa secretes several phenazines, including pyocyanin [29], that have tissue-damaging properties attributed to the induction of free radical production in host cells [30]. Second, P. aeruginosa can evade detection by the host, either by directly destroying host molecules that are involved in pathogen detection or by downregulating PAMP expression. For example, an important component of host defense is the deposition of a complement component C3b on the bacterial surface, leading to the induction of host responses and pathogen clearance [31]. To counter complement activation, P. aeruginosa produces alginate to limit accessibility of complement and secretes proteases, including alkaline protease and elastase that degrade C3b [32],[33]. The flagellum, an important virulence determinant required for motility and attachment, is also a PAMP that is detected by the host through the interaction of monomeric flagellin with TLR5, resulting in NF-κB and p38 MAPK activation; this innate immune response is intended to protect the host. Upon growth on purulent mucus from CF and non-CF patients, P. aeruginosa downregulates flagellin synthesis, thereby blunting the host protective immune response [34],[35]. A third strategy is to compromise the host by suppressing the host defense responses. It has been suggested that the ability of P. aeruginosa to rapidly kill Drosophila is associated with downregulation of antimicrobial peptide expression by a yet-unknown mechanism that requires a putative S-adenosyl-methionine-dependent methyltransfrease [36].
We found that an unexpected feature of the transcriptional response to P. aeruginosa is the downregulation of several intestinally-expressed host defense effectors [7],[37]. We hypothesized that repression of immune effector expression, such as thn-2, spp-1, and lys-7, may represent a virulence mechanism used by P. aeruginosa to suppress host defenses. In this report, we identify both host and pathogen factors required for the downregulation of immune effectors during P. aeruginosa infection of C. elegans. We present data indicating that P. aeruginosa infection causes the activation of DAF-2 insulin-like signaling, which leads to translocation of DAF-16 protein from nuclei of intestinal cells and downregulation of DAF-16 transcriptional targets. Delocalization of DAF-16 requires DAF-2 and an upstream neuroendocrine signaling pathway, including the DAF-2 agonist INS-7. Each of these effects is dependent on the P. aeruginosa two-component regulator GacA and the quorum-sensing regulators LasR and RhlR. Our results demonstrate that P. aeruginosa infection of C. elegans results in the suppression of intestinal immune defense through virulence factor mediated effects on the activity of insulin-like signaling in the intestine.
Results
P. aeruginosa downregulates the expression of host effector molecules
An important component of the innate immune response in plants and animals is the induced expression of host defense effectors following pathogenic challenge [38]. Several genes that are induced following infection and are important to protect C. elegans from pathogenic challenge have been identified. These host defense effectors include defensin (abf-2), saposin (spp-1), and several genes with homology to antimicrobial proteins: lysozymes (lys-2 and lys-7), thaumatin (thn-2), and a CUB-domain containing protein, F08G5.6. The expression of abf-2 and spp-1 was induced following infection by the Gram-negative bacterium Salmonella typhimurium, and both genes were required to protect against S. typhimurium infection [20]. Independent microarray studies showed that lys-2 and F08G5.6 were induced following infection by P. aeruginosa and that thn-2 and lys-7 were induced by the Gram-positive pathogens E. faecalis and M. nematophilum [7]–[9],[21]. Interestingly, spp-1, thn-2, and lys-7 expression was repressed during infection by P. aeruginosa [7],[8],[39]. This expression pattern is unexpected given that these genes are required for optimal survival on P. aeruginosa (Figure S1). To confirm the expression data, we used quantitative RT-PCR (qRT-PCR) to measure the expression of these genes in worms exposed to P. aeruginosa (PA14), E. faecalis (V583), and S. typhimurium (SL1344) for 12 hours at 25°C (Figure 1A–B). Three of the genes (thn-2, lys-7, and spp-1) were significantly repressed by exposure to P. aeruginosa compared to the laboratory food source E. coli (OP50-1) (Figure 1A). Three of the genes (abf-2, F08G5.6, and lys-2) were significantly induced by exposure to P. aeruginosa (Figure 1B). In contrast to P. aeruginosa, none of the six genes tested were significantly repressed following exposure to either S. typhimurium or E. faecalis. Among the genes that were significantly repressed by exposure to P. aeruginosa, lys-7 was induced by exposure to S. typhimurium and E. faecalis, while thn-2 and spp-1 were induced by exposure to E. faecalis (Figure 1A). Though we did not detect a significant induction of spp-1 by S. typhimurium following 12-hour exposure (Figure 1A), spp-1 has been previously observed to be induced after 48-hour exposure to S. typhimurium [20]. Thus, the downregulation of known immune effectors appears to be an atypical response to infection that is characteristic of infection by P. aeruginosa because similar responses are not observed by exposure to S. typhimurium or E. faecalis.
Downregulation of effector genes requires P. aeruginosa virulence regulators
P. aeruginosa GacA is a two-component response regulator that is required for full virulence in C. elegans and mammals [40]–[42]. GacA regulates the production of molecules that are detrimental to the host, including pyocyanin, elastase, and exotoxin A [27],[28]. We wondered whether the suppression of host defense effectors is an active outcome of infection: specifically, whether GacA regulates factors that are important in downregulating host defense gene expression in C. elegans. We therefore compared the expression of thn-2, lys-7, and spp-1 in worms exposed to the P. aeruginosa gacA mutant for 12 hours (Figure 1C). In contrast to wildtype P. aeruginosa (hereafter referred to as PA14), none of the genes were repressed by exposure to an in-frame deletion mutant of gacA (PA14 gacA). This indicates that downregulation of immune genes by PA14 requires gacA. The quorum-sensing regulators encoded by lasR and rhlR function downstream of GacA [27]. lasR and rhlR are also required for PA14 virulence in C. elegans ([42] and Table S1). As with exposure to PA14 gacA, thn-2, lys-7, and spp-1 were not repressed in worms exposed to a lasR transposon insertion mutant (PA14lasR) compared to OP50-1 (Figure 1C). In worms exposed to a rhlR transposon-insertion mutant (PA14rhlR), neither thn-2 nor lys-7 was downregulated, but spp-1 remained significantly repressed (Figure 1C). Thus, the ability of PA14 infection to downregulate the expression of thn-2 and lys-7 requires factors that are dependent on gacA, lasR, and rhlR. By contrast, while the downregulation of spp-1 requires gacA and lasR, it is independent of rhlR.
The PA14 gacA and PA14 lasR mutant strains are unable to colonize the worm intestine [42]. To determine whether the failure of PA14 gacA, PA14 lasR, and PA14 rhlR mutants to downregulate immune gene expression is a consequence of a low inoculum in the intestine, we compared the expression of thn-2, lys-7, and spp-1 in the C. elegans tnt-3(aj3) mutant following infection with the wildtype PA14, PA14 gacA, PA14 lasR, and PA14 rhlR strains. The tnt-3(aj3) mutant is unable to grind bacteria, thus a large inoculum of live bacteria, including the nonpathogenic OP50-1, accumulate in the intestinal lumen [13],[20]. First, we observed that although the tnt-3(aj3) animals accumulate OP50-1 in the intestine, expression of thn-2, lys-7, and spp-1 was indistinguishable from expression in wildtype worms which do not accumulate OP50-1 (t-test, p=0.46, 0.43, and 0.07, respectively). Importantly, despite the accumulation of intact PA14 gacA, PA14 lasR, or PA14 rhlR mutant bacteria within the intestinal lumen of tnt-3(aj3) animals, the expression of thn-2, lys-7, and spp-1 was not repressed (Figure S2). Thus, the lack of immune suppression by the PA14 gacA, PA14 lasR, and PA14 rhlR mutants is not due simply to limited intestinal colonization of wildtype worms. These results, coupled with the observation that S. typhimurium and E. faecalis infections do not downregulate thn-2, lys-7, and spp-1 expression (Figure 1A), indicate that the mere presence of live bacteria in the intestinal lumen is insufficient to suppress the expression of a subset of immune genes. We conclude that a genetically regulated aspect of PA14 virulence, which is absent in S. typhimurium or E. faecalis, causes downregulation of a subset of immune effectors in C. elegans.
Several Gram-negative bacterial pathogens suppress host immunity by employing the type III secretion system (T3SS), which injects virulence effectors directly into host cytoplasm to inhibit or limit the duration of NF-κB and MAP kinase activation (reviewed in [43],[44]). In P. aeruginosa, T3SS is an important virulence determinant in pathogenesis in insects and mammals [45]–[48]. Type III secretion genes are under the control of the GacS/GacA two-component regulator and the quorum-sensing system [49],[50]. To date, four T3SS effector proteins have been identified for P. aeruginosa: ExoS, ExoT, ExoU, and ExoY. In PA14, only ExoT, ExoU, and ExoY are encoded in the genome and their expression requires PscD. Although the T3SS is induced during infection of the C. elegans intestine, it appears to be dispensable for C. elegans killing [46],[51]. This led to the proposal that other virulence factors that play a predominant role in C. elegans pathogenesis could mask the effect of type III secretion effectors when death was used as the metric for the pathogenesis assay. We therefore determined whether the T3SS contributes to the suppression of C. elegans immune effectors, an arguably more sensitive assay. As previously observed, neither loss of any of the Type III effectors nor the entire repertoire of effectors in the ΔpscD mutant significantly affects the ability of these strains to kill C. elegans ([46] and data not shown). We found that the ΔpscD mutant was still able to significantly suppress the expression of lys-7 and spp-1, as measured by qRT-PCR (Figure S3A). We further confirmed that the P. aeruginosa T3SS is not required for immune suppression by showing that neither ΔexoT, ΔexoU, ΔexoY, nor ΔpscD mutant significantly affected the expression of a lys-7 GFP-reporter (Figure S3B).
In Drosophila, downregulation of NF-κB-regulated antimicrobial peptide expression by P. aeruginosa is mediated by an unknown mechanism that requires an S-adenosyl-methionine-dependent methyltransferase domain-containing protein encoded by PA14_41070 [36]. A PA14_41070 transposon insertion mutant is not defective in the ability to downregulate lys-7 or spp-1 expression, as measured by qRT-PCR (Figure S3A) and by GFP-reporter gene expression of lys-7 (Figure S3B). Thus, PA14_41070 is dispensable for immune suppression in C. elegans.
To determine whether suppression of host defense gene expression is strictly associated with P. aeruginosa virulence, we tested several mutants that are impaired in C. elegans killing. First, we tested two genes that are part of the GacA-LasR-RhlR regulon: dsbA and pqsA. Expression of dsbA, which encodes a periplasmic dithiol:disulfide oxidoreductase, requires GacA [52]. DsbA is required for the formation of disulfide bonds in periplasmic proteins and important for proper folding of multiple virulence factors exported by Type II secretion, including elastase and lipase. Pseudomonas quinolone signal (PQS) system is the third component of the quorum-sensing signaling system, and it is regulated by Las and Rhl quorum-sensing systems [53],[54]. Production of all known quinolone/quinolines in P. aeruginosa requires PqsA, an anthranilate-coenzyme A ligase that is the product of the pqsABCDE operon [55]–[57]. Both the dsbA and pqsA mutant strains are attenuated for killing C. elegans ([42] and Table S1). Yet, neither the dsbA nor the pqsA mutants is defective in the ability to downregulate lys-7 expression, as measured by a GFP-reporter (Figure S3B). Thus, PQS and DsbA, despite their requirement for virulence are not necessary for host immune suppression in the C. elegans model. Because dsbA and pqsA are part of the GacA-LasR-RhlR regulon, these results suggest that only a subset of the genes regulated by the GacA two-component and acyl-homoserine lactone quorum-sensing systems are required for immune suppression.
To determine the specificity of the GacA-LasR-RhlR regulon in immune suppression, we tested three additional genes that are not known to be part of this regulon: PA14_23420, PA14_23430, and PA14_59010 [58],[59]. We found that while these mutants are attenuated for killing C. elegans (Table S1), they had no detectable defect in the ability to downregulate lys-7 expression as measured by a GFP-reporter (Figure S3B). Overall these results indicate that virulence in P. aeruginosa is not strictly associated with immune suppression and that a subset of the GacA-, LasR- and RhlR-dependent factors are required for the downregulation of immune genes by P. aeruginosa.
P. aeruginosa downregulates a subset of DAF-2–dependent immune genes
In C. elegans, at least three conserved signaling pathways contribute to host defense: p38 MAPK signaling, Sma/TGF-β signaling, and DAF-2 insulin-like signaling (reviewed in [11],[12]). We had previously noted that the transcriptional profiles of PA14-infected and daf-2 loss-of-function worms were overlapping, but the genes regulated in common tend to be regulated in opposite directions [7]. Here, we compared whole-genome transcriptional profiles of uninfected daf-2 mutants [16],[60] to PA14-infected wildtype animals [7],[8] and found that daf-2 mutants and PA14 infection have predominately discordant effects on gene expression across a range of data sets using a variety of criteria for comparison (Table S6, Table S7, Figure S10A). This suggests that the gene expression pattern in PA14-infected animals is opposite of the pattern produced by reducing the activity of DAF-2, such as occurs in loss-of-function daf-2 mutants. Importantly, genes that were inversely regulated by daf-2 and PA14 infection were enriched for immune effector genes, thus raising the possibility that PA14 infection activates DAF-2 insulin-like signaling (Table S8, Figure S10B; for details, see Text S1).
If indeed activation of DAF-2 insulin-like signaling mediates PA14 suppression of host defense effectors, then the ability of PA14 to suppress host defense effectors would be attenuated or eliminated when loss-of-function daf-2(e1370) mutants are infected with PA14. Alternatively, if the transcriptional effect of daf-2 loss of function and PA14 infection operate in parallel, then the transcriptional response to PA14 will be unaffected in daf-2(e1370) animals. We therefore compared the change in gene expression following PA14 infection between wildtype and daf-2(e1370) animals using whole-genome microarrays. First, using the criteria of t-test p-values<0.05 and 2-fold up- or down-regulation, we identified 247 induced and 137 repressed genes in wildtype worms infected with PA14 for 24 hours compared to uninfected controls. We confirmed that the selection of these criteria did not affect our conclusions by repeating the analysis with a range of p-value and fold-change thresholds (data not shown). Next, using this list of PA14-induced and -repressed genes, we compared the change in gene expression in response to PA14 in daf-2(e1370) and wildtype (N2) animals (Figure 2A). Separate linear regressions of induced and repressed genes revealed that the induction response to PA14 was largely intact in daf-2(e1370) animals (r2=0.4280, p<0.0001), but the repression response to PA14 was substantially attenuated to the extent that the repression response observed in N2 and daf-2(e1370) was not significantly associated (r2=0.001, p=0.3). Comparison of induced and repressed genes provides an internal control for this analysis and indicates that attenuation of the downregulation of genes in response to PA14 infection does not simply reflect a failure of daf-2 animals to become infected by PA14 or to respond transcriptionally to PA14 infection. These results are consistent with the model that PA14 infection suppresses host defense genes through activation of the DAF-2 insulin-like signaling pathway. It also indicates that the induction of genes in response to PA14 infection is largely independent of DAF-2.
To examine whether the Sma/TGF-β signaling pathway could also contribute to the suppression of host defense genes following PA14 infection, we compared whole-genome expression data from infected and uninfected TGF-β receptor null mutant sma-6(wk7) and wildtype worms. Using the same parameters as the daf-2(e1370) analysis, we note that both the induction and repression of genes in response to PA14 infection were largely intact in sma-6(wk7) (r2=0.5231 and 0.4051, respectively, p<0.0001 each; Figure 2B). The correlation coefficients for induced and repressed genes were not significantly different. Comparison of the regression results from daf-2(e1370) with sma-6(wk7) is revealing. The correlations for induced genes between the daf-2(e1370) and sma-6(wk7) analyses (r2=0.4280 and r2=0.5231, respectively) were not significantly different (p=0.09, one-tailed test). Among the repressed genes, however, correlations with daf-2(e1370) were significantly less than with sma-6(wk7) (p=10−5, one-tailed test). Together, the data indicate that the Sma/TGF-β pathway is unlikely to contribute substantially to the suppression of host defense genes by PA14 infection.
To corroborate the results obtained from whole-genome microarray analysis, and also to analyze the involvement of p38 MAPK signaling, we repeated the transcriptional profile analysis using qRT-PCR measurement of a panel of 146 infection and stress response genes. We designed gene-specific qRT-PCR primers to a panel of 146 genes selected on the basis of their likely involvement in the response to infection and stress. Many immune and stress response genes are members of large gene families [7],[61]. The use of gene-specific qRT-PCR primers overcomes the problem of cross hybridization of gene families in microarray studies and provides more precise measures of mRNA levels. Moreover, the use of genes with known or putative function in immune and stress response supports the inference that observed transcriptional effects are functionally important. We measured gene expression in young adult worms exposed to OP50-1 or PA14 for 12 hours for N2, daf-2(e1370), sma-6(wk7), and sek-1(km4), a p38 MAPKK null mutant, to obtain the change in gene expression following infection for each worm strain (Table S2). We repeated the linear regression analysis of induced and repressed genes, comparing the transcriptional response to PA14 in each mutant to N2 (Figure 2C–2E). Transcriptional analysis of daf-2(e1370) using this targeted gene set indicated that the induction of immune and stress response genes in response to PA14 was largely intact in daf-2(e1370) animals (r2=0.62, p<0.0001), but the repression of genes in response to PA14 was substantially attenuated to the extent that the correlation between N2 and daf-2(e1370) was not statistically significant (r2=0.061, p=0.11; Figure 2C). This pattern of gene expression mirrors the result of the whole-genome analysis (Figure 2A) and provides further support that a substantial subset of the normal downregulation of gene expression in response to PA14 infection requires daf-2, but the upregulation of gene expression is largely independent of daf-2. In contrast to daf-2(e1370), both induced and repressed genes in sek-1(km4) and sma-6(wk7) mutants correlated significantly with N2 (Figure 2D–2E), thus corroborating the results of our whole-genome analysis with sma-6(wk7). Visually, the induced and repressed linear regression lines are concordant for sek-1(km4) (Figure 2D) and sma-6(wk7) (Figure 2E), but highly discordant in daf-2(e1370) (Figure 2C). Analysis with a p38 MAPK mutant pmk-1(km25) yielded nearly identical results to sek-1(km4): the response to PA14 infection in pmk-1(km4) correlated significantly (r2=0.80, p<0.0001) with the response in sek-1(km4) (data not shown).
To detect more subtle effects of sma-6, sek-1 and daf-2 mutations on transcriptional responses to PA14 we compared the average differential expression of the immune and stress response genes under normal (OP50-1 exposure) and infection (PA14) conditions between mutant (Xmutant) and wildtype (Xwildtype) worms using a paired t-test (see Materials and Methods). The average difference in induction or repression (XΔ=Xmutant−Xwildtype) was calculated for each mutant-wildtype pair. This approach accounts for both the magnitude and direction of attenuation of the response to infection, with positive values of XΔ indicating attenuated repression and negative values of XΔ indicating attenuated induction.
In sma-6(wk7) mutants, the correlation analysis indicated that both induction and repression were largely intact. Consistent with this analysis, the average difference in either induction or repression (XΔ) of immune and stress response genes in sma-6(wk7) mutants was not significantly different from wildtype (induction: XΔ=0.09, p=0.38; repression: XΔ=0.00, p=0.99). It remains possible that the increased susceptibility of sma-6(wk) to PA14 may be a consequence of deregulation of immune gene expression that could not be detected by this analysis.
In sek-1(km4) mutants, the average induction in response to infection is significantly less than that of wildtype (induction: XΔ=−0.61, p<10−4), indicating that the induction response is attenuated when p38 signaling is abrogated. Although there is a trend towards attenuation of host gene repression in sek-1(km4), the overall effect is not statistically significant for this set of 42 genes (XΔ=0.38, p=0.16). Thus, a definitive conclusion regarding the requirement of p38 in repression of immune genes following PA14 infection awaits a whole-genome analysis. By this analysis we are able to show that p38 MAPK signaling is required for the induction of many genes during PA14 infection, consistent with a previous report that p38 MAPK is an important regulator of the transcriptional response to PA14 [8]. Immune and stress response genes that require sek-1 for induction by PA14 include clec-85, lys-1, lys-8, F35E12.5, Y40D12A.2 and gst-38. Interestingly, with the exception of gst-38, the basal expression of these genes under normal growth conditions on OP50-1 also requires sek-1 (Figure S4A–S4F). We also identified a class of genes whose expression levels during normal growth and following infection require sek-1 but whose induction or repression in response to PA14 do not require sek-1; they include lys-2, cpr-3, spp-18, F55G11.2, and T10D3.6 (Figure S4G–S4K). For example, expression levels of lys-2 in sek-1(km4) animals are 0.5% of levels in wildtype worms both in worms exposed to OP50-1 and worms exposed to PA14, yet expression of lys-2 is induced by a similar ratio in wildtype and sek-1 mutant worms (Figure S4G).
The average differences in induction and repression of immune and stress response genes in daf-2(e1370) mutants were both significantly different from wildtype (induction: XΔ=−0.36, p=0.0004; repression: XΔ=1.32, p<10−5). Thus, by this analysis we are able to detect a small but significant attenuation in gene induction following infection in daf-2(e1370) mutants, suggesting that daf-2 activity is required for the induction of a small subset of infection-responsive genes, which include abf-2, lys-2, F08G5.6, ZK6.11, and C17H12.8 (Table S2). The effect of daf-2(e1370) on repression of gene expression in response to infection was the strongest observed effect, consistent with the regression analysis. The immune and stress response genes that require daf-2 for repression include thn-2, lys-7 spp-1, and gst-4 (Table S2). Importantly, among sma-6(wk7), sek-1(km4), and daf-2(e1370), only daf-2(e1370) significantly attenuated repression of infection response genes. Together, the transcriptional analyses confirm that of the three immune signaling pathways, DAF-2 insulin-like signaling is required for the downregulation of many genes during PA14 infection.
Suppression of some host defense genes is DAF-16–dependent
Loss-of-function daf-2 mutants are resistant to bacterial pathogens, and this resistance is dependent on the FOXO transcription factor DAF-16 [15]. DAF-2 regulates DAF-16 at least in part through the activation of phosphoinositide 3-kinase (PI3-kinase), which is encoded by age-1 [62],[63]. PI3-kinase potentiates the activity of several serine threonine kinases, including homologs of mammalian PDK1, AKT, and SGK [64]–[66]. These kinases phosphorylate DAF-16, retaining it in the cytoplasm and suppressing DAF-16 transcriptional activity [67],[68]. To determine whether DAF-16 is required for the suppression of host defense genes, we quantified the mRNA levels of thn-2, lys-7, and spp-1 by qRT-PCR in N2, daf-2(e1370), daf-16(mu86), and double mutant daf-16(mu86);daf-2(e1370) worms under normal growth conditions (on OP50-1) or following infection with PA14. Downregulation of thn-2, lys-7, and spp-1 by PA14 exposure was abolished in daf-2(e1370) (Figure 3A–3C), confirming that daf-2 activity is required for the downregulation of a number of functionally important immune genes. Next, we inactivated each of these genes by RNAi in daf-2(e1370) and compared the degree of colonization by a PA14 strain that expresses GFP (PA14-GFP) in these animals to daf-2(e1370) animals exposed to vector control. Knockdown of thn-2, lys-7, and spp-1 individually resulted in a significant increased in colonization (Figure S5), indicating that the ability to prevent downregulation of immune genes contributes to the resistance of daf-2(e1370) to PA14. Under normal growth conditions, both thn-2 and lys-7 were expressed at lower levels in daf-16(mu86) compared to N2, indicating that daf-16 is required for basal expression of these immune effectors (Figure 3A–3B). Following infection with PA14, the levels of thn-2 and lys-7 mRNA were not reduced further in daf-16(mu86), indicating that suppression of thn-2 and lys-7 by PA14 requires daf-16 (Figure 3A–3B). Curiously, the expression of lys-7 in daf-16(mu86);daf-2(e1370) double mutants was intermediate to the expression in either single mutant. Nonetheless, lys-7 expression was not repressed by PA14 infection in either single or double mutants. These results suggest that basal levels of lys-7 expression are regulated by daf-2-dependent factors in addition to DAF-16. By contrast, under normal growth conditions, spp-1 expression was not significantly different between N2 and daf-16(mu86), indicating that the basal expression of spp-1 is independent of daf-16 (Figure 3C), in contrast to previous reports [39]. Curiously, as in wildtype animals, spp-1 expression was significantly downregulated in daf-16(mu86) and daf-16(mu86);daf-2(e1370) mutants, indicating that while the downregulation of spp-1 is daf-2-dependent, it does not require daf-16. Overall, these results implicate DAF-2 and DAF-16 in the downregulation of thn-2, lys-7, and spp-1 during PA14 infection. DAF-16 appears to be required for the expression of thn-2 and lys-7, but the role of DAF-16 in the regulation of spp-1 expression is more complex.
Next, we determined whether the induction of abf-2, F08G5.6, and lys-2 by PA14 infection is mediated by daf-2 and daf-16 (Figure 3D–3F). In contrast to wildtype, the expression of abf-2 was not significantly induced in daf-2(e1370), daf-16(mu86), and daf-16(mu86);daf-2(e1370) mutants following infection, suggesting that the induction of abf-2 is modulated by the DAF-2/DAF-16 signaling pathway (Figure 3D). The induction of F08G5.6 was not statistically significant only in daf-2(e1370), suggesting that F08G5.6 induction requires daf-2 but not daf-16 (Figure 3E). The expression of F08G5.6 in daf-2 and daf-16 single and double mutants mirrors the expression pattern of spp-1. Thus, the induction of some genes is dependent on daf-2 as suggested by the paired t-test analysis performed on the immune and stress response gene set. Expression of lys-2 was largely unaffected in the insulin-like signaling mutants. However, expression of lys-2 was higher in daf-2(e1370) under basal conditions (on OP50-1). This difference was not observed in worms exposed to PA14, resulting in net a reduced induction of lys-2 in daf-2(e1370) (Figure 3F). Overall, the gene expression analyses suggest that while DAF-2 and DAF-16 are largely dispensable for the induced response to PA14 infection, some genes do require DAF-2 for induction.
Pseudomonas infection suppresses DAF-2–regulated stress response genes
In addition to immune effectors, the insulin-like signaling pathway also regulates stress response genes [16],[69]. For example, sod-3 encodes a well-characterized DAF-16-regulated superoxide dismutase that is associated with oxidative stress resistance and longevity in C. elegans, presumably through its reactive oxygen species detoxification activity [16],[70],[71]. We showed by qRT-PCR that the expression of sod-3 was significantly repressed following 24-hour exposure to PA14 (Figure S6), consistent with previous microarray studies [8]. This further supports the model that PA14 activates DAF-2 and inhibits DAF-16. Suppression of an antioxidant may be beneficial to P. aeruginosa, which produces pyocyanin as a virulence determinant that causes oxidative damage to host tissues [30]. However, knockdown of sod-3 by RNAi in daf-2 animals did not significantly affect the ability of daf-2 mutants to resist colonization by PA14 (Figure S5), perhaps because of the expression of other genes that contributes to oxidative stress resistance, such as gst-4, mtl-1, and ctl-1, are also increased in daf-2 mutants [16]. We note that the expression of gst-4, which encodes a glutathione-S transferase [72] is also downregulated during PA14 infection, as determined by reporter gene expression (Figure S7A) and qRT-PCR analyses (Figure S7B). Downregulation of gst-4 by PA14 requires gacA, lasR, and rhlR (Figure S7A), and is suppressed by daf-2(e1370) but not daf-16(mu86) (Figure S7B). Thus, activation of daf-2 signaling during P. aeruginosa infection results in the simultaneous downregulation of stress response and immune effectors genes that together could aid in the pathogenesis of these bacteria.
P. aeruginosa infection causes delocalization of nuclear DAF-16
The transcriptional effects of daf-2 mutants are largely dependent on DAF-16 [16]. Under normal growth conditions, the DAF-2 pathway is active and DAF-16 protein is distributed predominately in the cytoplasm of every tissue. Conditions that reduce signaling in the DAF-2 pathway, including heat stress, ablation of the germline, and loss of daf-2 function, cause DAF-16 protein to be localized in the nucleus [68],[73]. DAF-16 translocates from the nucleus to the cytoplasm when DAF-2 signaling is increased [73]. We determined the effect of PA14 infection on DAF-16 localization using transgenic worms that express a functional DAF-16::GFP fusion protein [73]. We confirmed that exposure to PA14 does not cause increased DAF-16 nuclear localization [7],[8]. Our gene expression data suggested that PA14 infection activates DAF-2 signaling and could consequently result in delocalization of DAF-16 from the nucleus. To measure this effect, it was necessary to first localize DAF-16 to the nucleus and then compare the reversal of that localization between infected and uninfected animals. We used two independent means to drive DAF-16::GFP to the nucleus: brief heat shock and removal of the germline, both of which are reported to induce DAF-16 nuclear localization in a DAF-2-independent manner [74],[75]. Heat shock causes transient nuclear localization that reverses over time, whereas the degree of nuclear localization caused by loss of germline proliferation is relatively stable in young worms. For the heat-shock approach, worms exposed to 37°C dry heat for 70 to 90 minutes were shifted onto plates containing OP50-1, PA14, or PA14 gacA. After 16 hours at 25°C, almost all of the worms exposed to OP50-1 (Figure 4A) or PA14 gacA (Figure 4B) retained nuclear DAF-16::GFP. By contrast, in a majority of the PA14-infected population, DAF-16::GFP was delocalized from the nuclei of intestinal cells (Figure 4C–4D). We quantified this effect by counting the number of intestinal nuclei in which DAF-16::GFP fluorescence was apparent. As shown in Figure 4E, the number of nuclei with visible DAF-16::GFP nuclear localization was significantly reduced in PA14-infected worms compared to OP50-fed worms (p<0.001). Because the distribution of nuclei containing DAF-16::GFP per worm was distinctly bimodal (representative examples shown in Figures 4C and 4D), individual worms could be classified as having predominately nuclear-localized or predominately nuclear-delocalized (i.e., cytoplasmic) DAF-16::GFP in subsequent assays. Following 16 hours recovery from acute heat shock, approximately 80% of PA14-infected worms no longer retained the DAF-16::GFP fusion protein in the nucleus, whereas worms exposed to OP50-1 or PA14 gacA were not significantly affected (Figure 4F, p<0.0001). DAF-16::GFP remains nuclear localized in heat-shocked worms exposed to OP50-1 or PA14 gacA for 48 to 60 hours.
To confirm that the rapid delocalization of DAF-16 is a general physiological response to PA14 infection and not an artifact specific to the use of heat shock stimulus to nuclear localize DAF-16, we examined the effect of PA14 infection on nuclear localized DAF-16 using worms in which germline proliferation was eliminated. Loss of germline proliferation causes nuclear localization of DAF-16 in the intestinal cells of adult worms [68]. RNAi knockdown of cdc-25.1 produces worms that lack germline proliferation [76] and localizes DAF-16::GFP to intestinal nuclei. Nuclear localization of DAF-16 due to loss of germline proliferation requires kri-1 [77]. We confirmed that knockdown of cdc-25.1 by RNAi affects DAF-16::GFP nuclear localization through its effect on germline proliferation by showing that nuclear localization of DAF-16::GFP could be suppressed when kri-1 was knocked down by RNAi (Figure S8). In adult worms without proliferating germline, PA14 infection caused delocalization of nuclear DAF-16 in approximately 75% of worms after 16 hours (p<0.0001), whereas worms exposed to OP50-1 or PA14 gacA were not significantly affected (Figure 4G). Delocalization of DAF-16::GFP upon PA14 infection was not due to a generalized loss of nuclear localization because we failed to observe a loss of nuclear integrity. Nor was it due to a loss of the ability to retain transcription factors in the nucleus because other nuclear localized GFP constructs, including a translational fusion of GFP to the GATA transcription factor ELT-2, remained nuclear over the course of the experiment (data not shown). In addition, DAF-16 nuclear delocalization caused by PA14 infection was reversible by subsequent heat shock (data not shown), indicating that PA14 infection does not simply render DAF-16 incapable of nuclear translocation or retention. Notably, the nuclear delocalization of DAF-16 occurred early during PA14 infection and affected DAF-16 in intestinal nuclei. Thus, delocalization of DAF-16 occurred in the appropriate time and location to account for the transcriptional patterns observed during PA14 infection, providing an independent category of evidence for the model that PA14 infection activates the DAF-2 insulin-like signaling pathway.
DAF-16 target genes were downregulated following infection by P. aeruginosa (PA14), but not by E. faecalis (V583) or S. typhimurium (SL1344) (Figure 1A). We therefore hypothesized that infection with V583 and SL1344 would not result in the translocation of DAF-16 from the nucleus. Because survival of worms on V583 and SL1344 is extended compared to PA14, delocalization of DAF-16 might occur later in worms exposed to V583 or SL1344 than in worms exposed to PA14. We therefore used worms lacking a proliferating germline to cause DAF-16 nuclear localization to ensure that nuclear localization of DAF-16 is distinguishable for at least 96 hours. Consistent with the failure of V583 and SL1344 to downregulate DAF-16 target genes (Figure 1A), there was no significant decrease in DAF-16 nuclear localization in worms exposed to these pathogens (Figure 4G). Finally, consistent with the requirement for gacA, lasR, and rhlR for the downregulation of immune gene expression during PA14 infection (Figure 1C), similar to the PA14 gacA mutant, the rhlR and lasR mutants also failed to cause significant delocalization of DAF-16::GFP compared to uninfected controls (Figure 4H). Thus, we can conclude that activation of DAF-2 affects the nuclear localization of DAF-16 and the transcription of DAF-2-dependent immune genes, such as thn-2, lys-7, and spp-1. This effect is specific to infection by P. aeruginosa and requires GacA-, LasR-, and RhlR-regulated virulence factors.
DAF-2 and a DAF-2 agonist, INS-7, are required for P. aeruginosa to affect DAF-16 nuclear localization
The results presented thus far indicate that PA14 infection activates DAF-2, resulting in the translocation of DAF-16 from the nucleus and the DAF-2-dependent repression of immune gene expression. They further suggest that loss-of-function mutations in daf-2 would suppress the delocalization of DAF-16. Indeed, when we compared DAF-16::GFP localization in daf-2(e1370); DAF-16::GFP animals that were exposed to either OP50-1 or PA14, no significant difference in DAF-16::GFP localization was observed between PA14-infected and uninfected controls over the course of 4 days (Figure 5A). This indicates that daf-2 is required for the delocalization of DAF-16 during PA14 infection and is consistent with the failure of PA14 to downregulate host effector genes in daf-2(e1370). It further suggests that PA14 infection suppresses C. elegans immune gene expression by affecting host components that are upstream of daf-2.
daf-2 encodes the only homolog of a mammalian insulin/IGF-1-family receptor in the C. elegans genome [78], and its activity is affected by insulin-like molecules [79],[80]. Therefore, insulin-like peptides are attractive candidates to be subverted by P. aeruginosa to activate DAF-2. As shown in Figure 5B, two insulin-like peptides, ins-7 and ins-11, were upregulated in worms exposed to PA14. Several lines of evidence implicate ins-7—and not ins-11—as contributing to the activation of DAF-2 during PA14 infection. First, ins-7 expression is not induced by exposure to either E. faecalis or S. typhimurium, but ins-11 expression is induced by exposure to E. faecalis (Figure 5B). Also, ins-7 is not induced in worms exposed to PA14 gacA, lasR, or rhlR mutants (Figure 5B and unpublished data). Moreover, ins-7 is known to be an insulin agonist [16],[80]. Thus, the induction of ins-7, but not ins-11, is concordant with the activation of DAF-2 by PA14 but not E. faecalis infection. Second, the ins-7 deletion mutant ins-7(tm1907) is resistant to PA14 infection, but the ins-11 deletion mutant ins-11(tm1053) is not distinguishable from wildtype worms with respect to susceptibility to PA14 (Figure S9, Table S3). Lastly, RNAi knockdown of ins-7 suppresses the effect of PA14 infection on DAF-16 nuclear delocalization, whereas ins-11 RNAi knockdown has no distinguishable effect (Figure 5C). To confirm that ins-7 is required for the effect of PA14 on DAF-16 localization, we examined DAF-16::GFP localization in the deletion mutant ins-7(tm1907). Loss of ins-7 suppressed the effect of PA14 infection on DAF-16 nuclear delocalization in worms lacking germline proliferation (Figure 5D), indicating that ins-7 is required for the effect of PA14 on DAF-16 nuclear localization.
DAF-16 is required in the intestine for defense against PA14
daf-16 null mutants are indistinguishable from wildtype in their ability to survive infection by a variety of pathogens [8],[15],[23],[81]. However, the phenotype that results from the loss of gene function in an entire organism is the combination of the effects on various tissues; and the tissue required for daf-16 in immune function has not been investigated. The profound effects of PA14 infection on insulin-like signaling, and the central role for DAF-16 in that interaction, led us to reevaluate the role of DAF-16 in defense against bacterial pathogens. Several lines of evidence suggested an important role for DAF-16 in the intestine. First, the intestine is the site of PA14 infection [82], and the major site of expression of host defense genes including spp-1 and lys-7 [7],[39]. Second, resistance to PA14 is associated with nuclear localization of DAF-16 specifically in the intestine. Loss of germline proliferation, which causes DAF-16-dependent resistance to PA14 (unpublished data), causes DAF-16 nuclear localization predominately in the intestine [68]. A forward genetic screen for enhanced resistance to PA14 identified mutants with enhanced nuclear localization of DAF-16 in the intestine [81]. Similarly, we have observed that loss of ins-7 caused DAF-16::GFP nuclear localization primarily in intestinal cells (unpublished data). Third, DAF-16 nuclear delocalization during PA14 infection is most evident in intestinal cells. Thus, we sought to examine the function of intestinal DAF-16 by knocking down the expression of daf-16 only in the intestine. Tissue-specific knockdown of gene expression can be achieved in C. elegans using strains in which the RNAi-deficient rde-1 mutant is rescued by expressing a transgene carrying the rde-1 gene in a specific tissue, such as intestine [83], hypodermis or muscle [84]. The strain VP303 expresses rde-1 in the intestine of an rde-1(ne219) mutant background, allowing for intestine-restricted RNAi knockdown [83],[85]. VP303 and N2 worms are indistinguishable for resistance to PA14 (Table S4). As expected from previous reports, [8],[15],[23],[81], RNAi knockdown of daf-16 in wildtype N2 worms had no effect on susceptibility to PA14 (Figure 6A). RNAi knockdown of daf-16 in VP303 worms, however, caused enhanced susceptibility to PA14 (Figure 6B). To confirm the specificity of VP303 for intestine-specific knockdown, we exposed N2 and VP303 strains to bacteria that express double-stranded RNA corresponding to a muscle-specific gene unc-22, which encodes for twitchin that functions in the muscles to regulate the actomyosin contraction-relaxation cycle and to maintain normal muscle morphology [86]. As expected, RNAi against unc-22 resulted in the canonical twitching phenotype in N2 but not in the VP303 strain. unc-22 RNAi also has no effect on pathogen resistance in both the N2 and VP303 strains (Table S4). Recently, the hypodermal tissue has also been shown to contribute to immune response to a fungal pathogen [87]. Using the NR222 strain, in which RNAi knockdown is restricted to the hypodermis, we showed that loss of daf-16 in the hypodermis is not required for pathogen resistance (Table S4). These results indicate that DAF-16 is required in the intestine for resistance to PA14, whereas loss of DAF-16 in the intestine, in combination with loss of DAF-16 in non-intestinal tissues, has an overall neutral effect on the ability of worms to defend against PA14 infection. Intestinally but not hypodermally expressed DAF-16 is essential for resistance to PA14, a function that was previously masked in studies examining the requirement for DAF-16 in the whole worm.
Discussion
Pseudomonas infection suppresses host defense genes
Models of host-pathogen interactions predict that pathogen challenge will result in the induction of host defense genes. Repression of host defense genes is often associated with suppression of host defense pathways by the pathogen. Using quantitative RT-PCR, we confirmed that known host defense effectors, including thn-2, lys-7, and spp-1 are downregulated during PA14 infection (Figure 1A). Because knockdown of basal expression of thn-2, lys-7, and spp-1 by RNAi resulted in enhanced susceptibility to PA14 (Figure S1), suppression of their expression during infection should compromise host defense. The observation that knockdown of thn-2, lys-7, and spp-1 by RNAi causes enhanced sensitivity to PA14 despite the downregulation of their expression in PA14-infected worms is explained in part by the fact that expression levels drop following infection whereas expression levels are already reduced to low levels when RNAi-treated worms are exposed to PA14. The downregulation of these host defense effectors is not a typical response of C. elegans to pathogen challenge; the expression of these host effectors was induced or unaffected in response to E. faecalis and S. typhimurium (Figure 1A). A large number of host defense effectors are efficiently induced in response to PA14 infection, suggesting that the downregulation of a particular set of host defense genes is a specific effect (Figure 1B and Table S2). Also, repression of a specific subset of host defense effectors by PA14 is likely to represent a specific interaction between P. aeruginosa and C. elegans because PA14 requires the virulence regulatory genes gacA, lasR and rhlR to suppress host defense genes (Figure 1C). gacA, lasR and rhlR mutants are each attenuated in virulence (Table S1), suggesting a link between virulence and the downregulation of host defense effectors. The specific requirement of GacA-, LasR- and RhlR-regulated virulence factors in immune suppression is supported by the observation that the PA14_23420, PA14_23430, PA14_59010 mutants that are attenuated in virulence are not defective in immune suppression (Figure S3B). As an initial effort to define the GacA-, LasR- and RhlR-regulated virulence factors that are required for immune suppression, we showed that dsbA, pqsA and the T3SS, all of which have been shown to be under the control of the GacS/GacA two component regulator and the quorum-sensing system [49], [50], [52]–[54] are not required for the downregulation of lys-7 expression (Figure S3B). The hypothetical protein, PA14_41070 that contains the S-adenosyl-methionine (SAM)-dependent methyltransferase domain that is required for downregulation of Drosophila NF-κB-regulated antimicrobial peptides expression [36] is also dispensable for immune suppression in C. elegans (Figure S3B). Overall these results suggest that the ability of P. aeruginosa to suppress C. elegans host defense effectors requires a subset of GacA-, LasR- and RhlR-regulated virulence factors and are not merely reflections of attenuation of virulence.
We have thus provided evidence that we have identified a new mechanism of host immune suppression. This mechanism does not require the SAM-dependent methyltransfrease or the type III secretion system, which have been previously implicated in host immune suppression through their effects on NF-κB and MAPK signaling [36],[43],[44]. Instead, it requires the GacA two-component regulator and the LasR and RhlR quorum-sensing regulators to regulate factors necessary to subvert host defense. These factors, which we showed are likely to be independent of DsbA and quinolone signaling, could potentially be identified by screening for P. aeruginosa mutants that fail to downregulate the expression of host defense effectors, a strategy that we are actively pursuing. Finally, as discussed below, this immune suppression is mediated by the host insulin-like signaling pathway, instead of the MAPK and NF-κB pathways.
Pseudomonas infection activates insulin-like signaling
The specific downregulation of the host defense effectors thn-2, lys-7, and spp-1 during P. aeruginosa infection (Figure 1A) suggested that a host defense signaling pathway was being subverted by P. aeruginosa to repress host defenses. Gene expression and host protein localization studies support a model that PA14 infection activates the DAF-2/DAF-16 insulin-like signaling pathway, resulting in the downregulation of DAF-16-regulated immune genes (Figure 7). The downregulation of thn-2, lys-7, and spp-1 during PA14 infection was abolished in daf-2 loss-of-function mutants (Figure 3A–3C). The requirement for daf-2 for gene downregulation during infection held true when the gene expression analysis was extended to include the entire genome by microarray (Figure 2A) or to a set of 146 candidate immune and stress genes that were measured more precisely by qRT-PCR (Figure 2C). Comparison of the transcriptional response to infection in mutants of each of the p38 MAPK, Sma/TGF-β, and insulin-like signaling pathways to the response in wildtype N2 worms confirms that the insulin-like signaling mutant, daf-2(e1370), suppresses a portion of the wildtype response to PA14 infection in a fashion distinct from the effects of loss of p38 MAPK or TGF-β signaling (Figure 2C–2E).
The model we propose in Figure 7 predicts that during PA14 infection, activation of insulin-like signaling will result in the inhibition of DAF-16. Inhibition of DAF-16 by insulin-like signaling is mediated by phosophorylation of DAF-16 by serine threonine kinases that are homologous to mammalian AKT and SGK, and results in the cytoplasmic retention of DAF-16 [68],[73]. We used several assay conditions to examine the nuclear localization of DAF-16 during PA14 infection, providing evidence that PA14 infection causes DAF-16 nuclear delocalization (Figure 4). Based on the transcriptional response data (Figure 1), we further predicted that nuclear delocalization would occur in response to wildtype PA14, but would not occur in response to infection by E. faecalis or S. typhimurium, and that PA14 mutants of gacA, lasR and rhlR would also be defective for the ability to cause delocalization of nuclear DAF-16. Each of these predictions was confirmed when tested (Figure 4G–4H), suggesting that DAF-16 delocalization is a PA14-infection specific effect. The model also predicted that while stimuli that function in parallel to insulin-like signaling to cause DAF-16 nuclear localization can be reversed by PA14 infection, such as heat shock and loss of germline proliferation, mutations that disrupt insulin-like signaling would also disrupt the DAF-16 nuclear delocalization caused by PA14. We confirmed that nuclear localization of DAF-16 in the insulin-like receptor mutant daf-2(e1370) was not affected by PA14 infection (Figure 5A). Finally, we showed that PA14 infection, but not infection by E. faecalis, S. typhimurium, or PA14 gacA caused increased expression of the insulin-like peptide ins-7, suggesting that ins-7 may participate in signaling upstream of daf-2 in response to PA14 infection (Figure 5B). Consistent with this hypothesis, we found that a deletion mutant of ins-7 attenuated the ability of PA14 to cause nuclear delocalization of DAF-16 (Figure 5C). Collectively, the expression profile and DAF-16 localization data indicate that PA14 infection induces the expression of ins-7 to activate insulin signaling that leads to inhibition of DAF-16 and downregulation of immune gene expression.
In a separate report, we showed that a neuroendrocine signaling axis that functions upstream of the DAF-2/DAF-16 pathway to regulate lifespan [88],[89] and dauer formation [90] also regulates innate immunity. Regulation of innate immunity is mediated by the expression of ins-7 in the neurons (unpublished data). We showed that mutants with decreased neurosecretion, such as unc-64(e246) and unc-31(e928), are more resistant to PA14 because they express significantly higher levels of immune effectors due to constitutive nuclear localization of intestinal DAF-16. In that report, we tested whether neuronal function was required for activation of DAF-2 by PA14 infection. We found that constitutive nuclear localization of DAF-16::GFP in neurosecretion defective mutants, such as unc-64(e246) is not reversed by exposure to PA14 (data not shown). Taken together, the data indicate neuronal secretion of an insulin-like peptide, INS-7, contributes to activation of DAF-2 insulin-like signaling during PA14 infection. It will be interesting to determine how PA14 infection induces the expression of ins-7 in neurons and thereby suppress immune gene expression.
Other contributors to the transcriptional response to Pseudomonas
Activation of DAF-2 appears to account for a substantial proportion of the downregulation of host defense genes during PA14 infection, but is less important for the upregulation of host defense genes (Figure 2A and 2C). To understand the mechanisms contributing to gene upregulation during infection, we also examined the contributions of the Sma/TGF-β and p38 MAPK pathways to the transcriptional response to PA14 (Figure 2B,2D,2E). For many genes, the transcriptional response to PA14 is intact in mutants of all three pathways tested (Table S2). This indicates that the regulation of the transcriptional response to PA14 is redundant and/or that additional pathways contribute substantially to the transcriptional response to PA14. Functional redundancy in the regulation of host immune effectors is a conserved feature of innate immunity [91]. In light of this observation, it is especially notable that the downregulation of a substantial number of genes is dependent on DAF-2 insulin-like signaling.
The suppression of spp-1 by PA14 (Figure 3C), unlike the suppression of thn-2 and lys-7 (Figure 3A and 3B), is not entirely DAF-16 dependent, suggesting additional layers of complexity in the regulation of DAF-2 target genes. Neither p38 MAPK nor Sma/TGF-β is required for the downregulation of spp-1 (Figure S4 and Table S2). The factor that contributes to the downregulation of spp-1 remains unknown, but may function downstream of DAF-2 in cooperation with DAF-16 (Figure 7).
A number of transcriptional co-regulators have been identified which regulate lifespan and stress resistance in conjunction with DAF-16 [75],[92],[93]. We considered two transcription factors, ELT-2 and SKN-1, as likely candidates because they are also expressed in the intestine. elt-2 encodes a GATA transcription factor that is required for a substantial portion of the transcriptional response to PA14 [7], and is required to protect C. elegans from infection by pathogens [7],[23]. However, the pathways that act upstream of ELT-2 in the regulation of immune defense has not be elucidated. Intriguingly, a CTTATCA (reverse complement: TGATAAG) DNA motif that is enriched in the 5-prime flanking regions (5′-flank) of DAF-2/DAF-16-regulated genes [16] is essentially identical to a GATA-like motif (TGATAAGA) that is enriched in the 5′-flank of genes that are significantly up- or down-regulated in worms exposed to P. aeruginosa [7], and each is a derivative of the consensus GATA motif (WGATAR). For example, the 5′-flank of spp-1, thn-2, gst-4, lys-7, and sod-3 each contain GATA motifs. This motif is distinct from the canonical FOXO binding site (TRTTTAG), which has been shown to bind DAF-16 in vitro [70]. The FOXO-binding DNA motif is present in the 5′-flank of lys-7 and sod-3, but not spp-1, thn-2, or gst-4. Moreover, knockdown of elt-2 by RNAi reduces the expression of lys-7 and spp-1 under uninfected conditions (data not shown) and enhances the relative downregulation of thn-2 during PA14 infection [7]. Overall, these findings suggest that ELT-2 or another GATA-motif-binding transcription factor may function downstream of DAF-2 together with DAF-16 to regulate the expression of immune-response gene. One instantiation of the consensus-binding motif of the transcription factor SKN-1 (ATGATAAT) is remarkably similar to the WGATAR motifs found upstream of genes regulated by DAF-2 and PA14 infection. Recently, it has recently been shown that for lifespan SKN-1 is regulated by insulin-like signaling [93]. We observed that knockdown of skn-1 by RNAi throughout the life of the animal caused enhanced sensitivity to PA14 (unpublished data). Thus, we wondered whether SKN-1 might contribute to DAF-2-dependent regulation of gene expression following PA14 infection. Knockdown of skn-1 by RNAi does not affect the downregulation of thn-2, spp-1, or lys-7 (data not shown). gst-4 is known to be SKN-1-regulated. However, like thn-2, spp-1, and lys-7, knockdown of skn-1 expression by RNAi does not abolish the downregulation of gst-4 following PA14 infection, despite causing a reduction in the basal level of gst-4 expression (Figure S7C). Future work must focus on determining transcription factors in addition to DAF-16 that accounts for the DAF-2-dependent downregulation of host defense genes.
We also note that not every downregulated gene is dependent on DAF-2 insulin-like signaling. For example, following PA14 infection, acdh-1 is strongly repressed in wildtype and all of the mutants tested, including daf-2(e1370) (Table S2). This suggests that factors in addition to DAF-2 insulin-like, Sma/TGF-β, and p38 MAPK signaling may contribute to the repression of host defense genes during PA14 infection. However, acdh-1 expression is also downregulated in worms exposed to E. faecalis, E. carotovora, P. luminescens, S. marcescens, S. typhimurium and PA14 gacA (M. Nandakumar and M.-W. Tan, unpublished data, and [21]), suggesting that the downregulation of acdh-1 may represent a different phenomena than the repression of other host defense genes.
Insulin-like signaling regulates intestinal immune defense
Given the dynamic regulation of insulin-like signaling during infection, and particularly the complex role for DAF-16 in this interaction, we wondered why null mutants of daf-16 are nonetheless indistinguishable from wildtype animals for their ability to survive infection by a variety of bacterial pathogens [8],[15],[23],[81]. Expression of daf-16 in the intestine partially enhanced pathogen resistance of a daf-16(mu86);daf-2(e1370) worm (unpublished data), consistent with the intestine being the site of infection, and the major site of immune gene expression. Additional observations that germline signaling, neuroendocrine signaling, and PA14 infection primarily modulate DAF-16 nuclear localization in the intestine highlighted the importance of intestinal DAF-16. A compensatory mechanism or other complex interaction acting across tissues could potentially mask the function of daf-16 in the intestine. Using of intestine-specific daf-16 knockdown by RNAi we show that DAF-16 is required for a protective response against pathogen. Supporting this finding further is the observation that enhanced resistance of ins-7 to PA14 is associated with constitutive nuclear localization of DAF-16 specifically in the intestine (unpublished data).
The most parsimonious explanation for the failure to observe enhanced sensitivity to pathogens when daf-16 is knocked down or knocked out in the whole organism is that intestinal and non-intestinal DAF-16 have antagonistic effects on host defense. Given the known complex regulation of DAF-16, this scenario is plausible. Balanced antagonistic effects on DAF-16 are observed in the context of signaling from the C. elegans germline and somatic gonad that regulate aging. Signals from the germline that normally suppress DAF-16 activity are balanced by distinct signals from the somatic gonad that normally enhance DAF-16 activity. The gonad signal appears to operate via the neurons and DAF-2 insulin-like signaling, while the germline signal appears to operate through a parallel pathway that does not require DAF-2 [94]. In this context, it is most appropriate to describe the effects of genetic manipulations in terms of net aggregate effects. Thus, the net aggregate effect of loss of daf-16 in the entire organism is neutral while the net aggregate effect of loss of daf-16 in the intestine results in pathogen sensitivity.
Endodermal GATA transcription factors have been shown to have a conserved role in regulating epithelial innate immune responses in the C. elegans intestine and human lung epithelial cells [7]. Here we find that P. aeruginosa can suppress epithelial immunity by affecting DAF-16 activity in the intestinal cells. The regulation of FOXO/DAF-16 by insulin-like signaling is highly conserved across organisms (reviewed in [95]). Given the medical importance of P. aeruginosa, for example in cystic fibrosis patients where it chronically infects lung epithelial cells [25],[26], it will be interesting in the future to investigate whether insulin-like signaling regulates an epithelial immune response and whether P. aeruginosa suppresses this immune response in human patients.
Materials and Methods
Worm and bacterial strains
Caenorhabditis elegans and bacterial strains are described in Table S5. Double mutants were constructed and verified using standard genetic and molecular methods [96]. Bacteria expressing dsRNA directed against daf-16, lys-7, thn-2, and spp-1 was part of a C. elegans RNAi library expressed in E. coli (Geneservice, Cambridge, U.K.). Bacteria expressing dsRNA directed against sod-3, ins-7, ins-11, kri-1, and skn-1 were part of a C. elegans RNAi library expressed in E. coli (Open Biosystems, Huntsville, Alabama). All bacterial strains were cultured under standard conditions.
Microarray analysis
Growing worms for microarrays, RNA extraction, microarray hybridization, and data processing were performed as previously described [7]. Reference RNA and microarray hardware were matched to allow direct comparison of transcriptional response to PA14 in all worm strains. In particular, sma-6(wk7) experiments were performed in parallel with previously reported N2 experiments [7]. daf-2(e1370) animals were grown at 20°C until young adulthood and then placed for 24 hr at 25°C on lawns of either OP50-1 or PA14 grown on modified NGM. Microarray production, hybridization, and scanning were performed at the Stanford Functional Genomics Facility. cDNA was generated from total RNA. Experimental cDNA was labeled with Cy3 and reference cDNA was labeled with Cy5. Four replicates of each condition were examined. Microarray data was deposited in the Stanford Microarray Database [97]. The default background subtraction and normalization settings were used. Spots were filtered based on foreground to background ratio; values less than 1.2 were flagged. Log (base 2) values were exported. Gene expression on OP50-1 and PA14 were compared by t-test. Differences of 2 fold with p-values<0.05 were considered significant. Sensitivity analysis was conducted with a range of significance levels and fold-difference values to confirm that the choice of thresholds did not affect our conclusions.
Quantitative RT-PCR
Age-matched young adult worms were exposed to bacterial lawns of OP50-1 or PA14 on NGM for 12 hours. RNA extraction and quantitative RT-PCR (qRT-PCR) was performed as previously described [7]. Briefly, 25 µl reactions were performed using the iScript One-Step RT-PCR kit with SYBR green according to the manufacturer's instructions (BioRad Laboratories, Hercules, CA), primers at a final concentration of 1 µM, and a data acquisition temperature of 76°C. In order to control for variation in RNA loading concentration, cycle threshold (Ct) values were normalized to three primer pairs (ama-1, F44B9.5, and pan-actin (act-1,3,4)) that were found to not change with infection. Summary statistics and statistical tests were calculated from N2-normalized cycle threshold values prior to conversion to relative fold change. Calculations were performed with a custom Perl script, Excel and R.
A panel of 146 infection and stress response genes queried by qRT-PCR included representatives of the antibacterial factor related (abf), lysozymes (lys), saposin-like proteins (spp), or thaumatin family (thn) gene classes, as well as genes identified to be transcriptionally regulated by PA14 [7] or by the DAF-2/DAF-16 pathway [16]. Whenever possible, primers were designed to amplify a sequence found in spliced cDNA but not genomic DNA by having one of the primer pairs overlap an exon junction. To this end, primer design was aided with the program AutoPrime [98]. Primer sequences are available from the authors upon request.
Analysis of induction and repression responses to PA14 infection
To compare the response to infection in wildtype and mutant worms, RNA concentrations were measured in matched samples of worms exposed to OP50-1 and PA14. The RNA concentration of gene g in worms exposed to OP50-1 (designated OP50g) indicates the basal level of expression. The RNA concentration of gene g in worms exposed to PA14 (designated PA14g) indicates the level of expression that follows infection. The response to infection for gene g is the difference in RNA levels between OP50-1 and PA14 samples. This was quantified as xg=PA14g – OP50g for qRT-PCR cycle threshold measurements and log2-converted microarray measurements. These log2-scale values were interpreted as fold changes using the conversion fold change=2x. Positive values of xg (PA14g – OP50g>0) indicate induction of the expression of g in response to infection. Negative values of xg (PA14g – OP50g<0) indicate repression of the expression of g in response to infection.
An ordered set of xg values for a set of genes (such as the set of induced or the set of repressed genes) from a particular worm strain forms a vector designated Xstrain. Correlation analysis can be used to determine the effect of a mutation on the induction or repression response to PA14 infection. The squared correlation between Xwildtype and Xmutant, r2, indicates the degree to which induction or repression is conserved in the mutant strain. Values of r2 that are statistically indistinguishable from 0 indicate that induction or repression is strongly attenuated.
For a more sensitive analysis, Xwildtype and Xmutant were compared by paired t-test. The summary statistic XΔ was calculated as the average of Xmutant – Xwildtype. XΔ indicates both the magnitude and direction of the difference in induction or repression in response to infection between wildtype and mutant strains. For induced genes, XΔ<0 indicates attenuation of induction. For repressed genes, XΔ>0 indicates attenuation of repression. The significance of XΔ values was determined by paired t-test.
DAF-16 nuclear localization assay
Acute heat shock was performed at 37°C for 70 to 90 minutes. Precise exposure times were determined empirically during each run by qualitative assessment of worms after heat shock. Germline proliferation was disrupted by RNAi knockdown of cdc-25.1 as previously described [76]. Worms were then exposed to PA14 or controls for approximately 16 hours; times varied from 14 to 18 hours as needed. Individual worms were classified as exhibiting predominantly nuclear DAF-16::GFP or predominantly delocalization DAF-16::GFP under 160× total magnification using an Olympus SZX12 dissecting microscope with a 1.6× objective and an EN GFP-LP filter cube (catalog no. 41018, Chroma Technology, Brattleboro, VT) which has the following specification: excitation band-pass filter 470±20 nm, emission filters >495 nm long pass followed by another >500 nm long pass. Fluorescence micrographs were collected at 200× magnification by fluorescence microscopy using a Leica DMRXA2 microscope with the Leica I3 filter set (Excitation 420 nm, Emission 525 nm, 30 nm band pass) for GFP. The criteria for predominately nuclear was the greater than 4 intestinal nuclei with nuclear localized DAF-16::GFP. The proportion of worms in each category is a metric for comparing the extent of nuclear localization of DAF-16 between populations, and have previously been used to examine the genetic control of DAF-16 nuclear localization in other contexts [99]. Assays scored in this fashion are highly reproducible between experiments and across experimenters.
C. elegans survival assays
Assays to determine the ability of C. elegans to survive PA14 infection were performed as described [76]. Briefly, to avoid the confounding effects of progeny production and internal hatching on survival, sterile worms were used in survival assays. To sterilize worms, rrf-3(pk1426);glp-4(bn2) and pha-1(e2123) were raised at 25°C. ins-7(tm1907), ins-11(tm1053), and wildtype controls were sterilized using RNAi knockdown of cdc-25.1. VP303 and NR222 worms are resistant to RNAi in the germline. Thus, VP303, NR222 and N2 worms were grown at 20°C and shifted to 27.5°C for 24 hours at the L3/L4 molt, which causes sterility. PA14 was grown overnight in King's Broth containing 100 mg/ml rifampicin at 37°C. 10 µl was spread on modified NGM and grown for 24 hours at 37°C. Worms were infected at 25°C by feeding on PA14 lawns. Kaplan-Meier survival analysis was performed using StatView 5.0.1. The Mantel-Cox logrank test was used to assess statistical significance of differences in survival. Only p-values<0.01 were considered significant. Mean time to death and standard error of the mean was calculated in StatView and then normalized to N2 for graphical comparison (Tables S1, S3, and S4).
Supporting Information
Figure S1
Knockdown of thn-2, lys-7, and spp-1 by RNAi enhances the susceptibility of C. elegans to P. aeruginosa infection. Survival of worms in which (A) spp-1, (B) lys-7, or (C) thn-2 was knocked down by RNAi followed by exposure to wildtype PA14 at 25°C was monitored over time. rrf-3(pk1426);glp-4(bn2) worms were used to enhance sensitivity to RNAi and to prevent progeny production, which can confound pathogen survival assays.
(417 KB TIF)
Figure S2
P. aeruginosa gacA, lasR, and rhlR are required immune suppression in C. elegans mutants that are unable to limit bacterial accumulation. Expression of downregulated host defense effectors genes in tnt-3(aj3) worms exposed to the PA14 and PA14 gacA, lasR, and rhlR mutants. Mean transcript levels are plotted relative to matched controls exposed to OP50-1. Error bars indicated SEM. At least 3 replicates of each condition were examined. * t-test, p<0.05 comparison to tnt-3(aj3) exposed to OP50-1.
(166 KB TIF)
Figure S3
Many P. aeruginosa virulence factors are not required for immune suppression in C. elegans. (A) Expression of the host defense effector lys-7 and spp-1 in wildtype worms exposed to wildtype PA14 and the PA14 mutant pscD and PA14_41070 measured by qRT-PCR. Mean transcript levels are plotted relative to matched controls exposed to OP50-1. Error bars indicated SEM. 3 replicates of each condition were examined. * t-test, p<0.05 (B) Expression of the host defense effector lys-7 in wildtype worms exposed to wildtype PA14 and the PA14 mutants gacA, lasR, rhlR, exoU, exoT, exoY, pscD, PA14_41070, dsbA, pqsA, PA14_23420, PA14_23430, and PA14_59010 measured by visual classification of lys-7::GFP fluorescence as high or low intensity at 100× total magnification. Expression of lys-7::GFP in uninfected worms was the standard for high intensity fluorescence. Data is also shown in Table S1. * t-test, p<0.05 (C–D) Fluorescence micrographs of lys-7::GFP worms categorized as (C) high or (D) low expression.
(442 KB TIF)
Figure S4
2×2 factorial interaction plots depicting the contribution of p38 MAPK to the transcriptional response to P. aeruginosa. Both p38-dependent (A–F) and p38-independent (G–K) induction and repression of gene expression in response to PA14 infection are observed. Mean (and SEM) of log2 scale transcript levels relative to N2 OP50-1 were plotted for (A) clec-85, (B) lys-1, (C) lys-8, (D) F35E12.8, (E) Y40D12A.2, (F) gst-38, (G) lys-2, (H) cpr-3, (I) spp-18, (J) F55G11.2, (K) and T10D3.6 in N2 (blue) and sek-1(km4) (red) exposed to OP50-1 and PA14.
(525 KB TIF)
Figure S5
Knockdown of thn-2, lys-7, and spp-1 by RNAi decreases the resistance of daf-2(e1370) mutants to colonization by P. aeruginosa. RNAi knockdown of thn-2, lys-7, and spp-1 resulted in increased P. aeruginosa accumulation in daf-2(e1370) worms compared to control RNAi (Fisher's exact test; p=0.0035, p=0.0054, and p<0.0001, respectively). RNAi knockdown of sod-3 did not significantly affect colonization (Fisher's exact test, p>0.9999). RNAi or control treated daf-2(e1370) were exposed to a PA14 strain that expresses GFP (PA14-GFP) as young adults for 108 hr. Individual worms were classified as having detectable or undetectable GFP fluorescence in the intestinal lumen by visual inspection at 200× total magnification. For each RNAi treatment, a total of 65 worms were assayed blind.
(193 KB TIF)
Figure S6
sod-3 expression is repressed following P. aeruginosa infection. qRT-PCR measurement of sod-3 transcript levels in N2 worms exposed to OP50-1 and PA14 is plotted relative to OP50-1 levels. * t-test, p<0.05.
(734 KB TIF)
Figure S7
The expression of gst-4 is repressed in worms exposed to P. aeruginosa in a skn-1-independent manner. (A) Fluorescence intensity of gst-4::GFP in worms exposed to wildtype PA14 and the PA14 mutants gacA, lasR and rhlR. Worms of uniformly bright GFP intensity were placed on PA14 for 24 hours. Individual worms were scored for GFP intensity in three categories: low, medium, and high. Population proportions are significantly different in all pairwise comparisons with the exception of lasR and rhlR, which are statistically indistinguishable (Chi-square test, p<0.05). (B) Expression of gst-4 measured in N2, daf-2(e1370), daf-16(mu86), and daf-16(mu86);daf-2(e1370) exposed to OP50-1 and PA14. Mean transcript levels were plotted relative to N2 OP50-1. Error bars represent SEM. t-test ^ p<0.05 comparison to N2 PA14, + p<0.05 comparing OP50-1 to PA14. (C) Mean (and SEM) of log2 scale transcript levels relative to N2 OP50-1 for gst-4 in skn-1 RNAi and vector control worms exposed to OP50-1 and PA14.
(437 KB TIF)
Figure S8
kri-1 is required for nuclear localization in cdc-25.1 RNAi-treated animals without proliferating germlines. The localization of DAF-16 in animals exposed first to cdc-25.1 RNAi and then kri-1 RNAi was assayed. *** Fisher's exact test, p<0.0001.
(721 KB TIF)
Figure S9
ins-7 but not ins-11 is required for resistance to P. aeruginosa. Survival of N2, ins-7(tm1907) and ins-11(tm1053) worms was monitored on PA14 over time at 25°C. ins-7(tm1907) is significantly more resistant to PA14 than N2 or ins-11(tm1053) (logrank, p<0.0001 and p=0.0002, respectively). ins-11(tm1053) is statistically indistinguishable from N2 (logrank, p=0.0902).
(164 KB TIF)
Figure S10
P. aeruginosa exposure negatively regulates host defense daf-2 target genes. (A) Venn diagram representing the intersection of the “broad” daf-2 and PA14 consensus datasets. Values are fold enrichment over chance. The greatest enrichment of genes that are affected in both daf-2 mutants and during PA14 infection was found in the inversely regulated categories. (B) Venn diagram representing the intersections of the “broad” daf-2 and PA14 datasets among immune genes. All immune genes that are affected in both daf-2 mutants and during PA14 infection are found in the inversely regulated categories.
(236 KB TIF)
Table S1
Survival of pha-1(e2123) worms and lys-7::GFP expression in transgenic worms following exposure to PA14 and PA14 mutants.
(6 KB PDF)
Table S2
Relative magnitude of induction or repression following 12 hr exposure to PA14 (PA14 - OP50) in log base-2 scale.
(40 KB XLS)
Table S3
ins-7, but not ins-11, is required for resistance to PA14. Worms were sterilized by RNAi knockdown of cdc-25.1. Glp worms were exposed to PA14 and survival was monitored over time.
(5 KB PDF)
Table S4
Intestinal-specific knockdown of daf-16 causes enhanced susceptibility to PA14. Worms were sterilized by 27.5°C treatment.
(6 KB PDF)
Table S6
Meta-analysis of DAF-2 pathway and PA14 microarray datasets.
(761 KB XLS)
Table S7
DAF-2-regulated genes are enriched for discordantly regulated infection-response genes.
(6 KB PDF)
Table S8
List of immune effector genes: genes encoding proteins with antimicrobial activity and/or required to protect C. elegans from P. aeruginosa infection.
(17 KB PDF)
Acknowledgments
We thank Madhumitha Nandakumar for sharing unpublished data, Andrew Fire for use of his laboratory microscope, and all the members of the Tan Laboratory for critical review and discussion of this work. Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center. ins-7(tm1907) and ins-11(tm1053) mutants were provided by the Japanese Bioresource Project. VP303 and NR222 were a kind gift from Kevin Strange, Vanderbilt University. lys-7::GFP was a kind gift from Scott Alper, National Heart Lung and Blood Institute.
Footnotes
The authors have declared that no competing interests exist.
This work was supported by a grant from the National Institute of Health to M.-W.T., and E.A.E. was supported by a Stanford Graduate Fellowship and a National Science Foundation Graduate Research Fellowship.
References
Articles from PLOS Pathogens are provided here courtesy of PLOS
Full text links
Read article at publisher's site: https://doi.org/10.1371/journal.ppat.1000175
Read article for free, from open access legal sources, via Unpaywall: https://journals.plos.org/plospathogens/article/file?id=10.1371/journal.ppat.1000175&type=printable
Citations & impact
Impact metrics
Citations of article over time
Article citations
Modeling Host-Pathogen Interactions in C. elegans: Lessons Learned from Pseudomonas aeruginosa Infection.
Int J Mol Sci, 25(13):7034, 27 Jun 2024
Cited by: 1 article | PMID: 39000143
Review
SPP-5 affects larval arrest via insulin signaling pathway in Caenorhabditis elegans.
J Mol Histol, 55(4):491-502, 13 Jun 2024
Cited by: 0 articles | PMID: 38869752
Uncovering the function of insulin receptor substrate in termites' immunity through active immunization.
J Insect Sci, 24(4):1, 01 Jul 2024
Cited by: 0 articles | PMID: 38958928 | PMCID: PMC11221318
Procatechuic acid and protocatechuic aldehyde increase survival of Caenorhabditis elegans after fungal infection and inhibit fungal virulence.
Front Pharmacol, 15:1396733, 22 May 2024
Cited by: 2 articles | PMID: 38841375
TGF-β pathways in aging and immunity: lessons from <i>Caenorhabditis elegans</i>.
Front Genet, 14:1220068, 05 Sep 2023
Cited by: 3 articles | PMID: 37732316 | PMCID: PMC10507863
Review Free full text in Europe PMC
Go to all (133) article citations
Data
Data behind the article
This data has been text mined from the article, or deposited into data resources.
BioStudies: supplemental material and supporting data
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
The DAF-2 insulin-like signaling pathway independently regulates aging and immunity in C. elegans.
Aging Cell, 7(6):879-893, 08 Sep 2008
Cited by: 89 articles | PMID: 18782349 | PMCID: PMC2630471
GATA transcription factor required for immunity to bacterial and fungal pathogens.
PLoS One, 1:e77, 20 Dec 2006
Cited by: 65 articles | PMID: 17183709 | PMCID: PMC1762309
Luteolin promotes pathogen resistance in Caenorhabditis elegans via DAF-2/DAF-16 insulin-like signaling pathway.
Int Immunopharmacol, 115:109679, 12 Jan 2023
Cited by: 4 articles | PMID: 36640711
Regulation of aging and innate immunity in C. elegans.
Aging Cell, 3(4):185-193, 01 Aug 2004
Cited by: 86 articles | PMID: 15268752
Review