|Home | About | Journals | Submit | Contact Us | Français|
Gene regulation by microRNAs (miRNAs) under specific physiological conditions often involves complex interactions between multiple miRNAs and a large number of their targets, as well as coordination with other regulatory mechanisms, limiting the effectiveness of classical genetic methods to identify miRNA functions. We took a systematic approach to analyze the miRNA-induced silencing complex (miRISC) in individual tissues of C. elegans and found that mRNAs encoded by pathogen-responsive genes were dramatically over-represented in the intestinal miRISC, and that multiple miRNAs accumulated in the intestinal miRISCs upon infection. Inactivation of the miRISC or ablation of miRNAs from different families, resulted in over-expression of several pathogen-responsive genes under basal conditions and, surprisingly, enhanced worm survival on pathogenic Pseudomonas aeruginosa. These results indicate that much of the miRNA activity in the gut is dedicated to attenuating the activity of the pathogen response system, uncovering a complex physiological function of the miRNA network.
Although classical genetic studies were essential to the discovery of miRNAs (Brennecke et al., 2003; Johnston and Hobert, 2003; Lee et al., 1993; Reinhart et al., 2000), it is now clear that the vast majority these small regulatory RNAs, due to redundancy and complex interactions with many targets, cannot be fully analyzed with straightforward genetic methods. Indeed, it has now been shown that deletion of individual miRNAs or even deletion of complete miRNA families rarely results in a discernable phenotype (Alvarez-Saavedra and Horvitz; Miska et al., 2007), indicating that miRNAs such as lin-4, let-7, and lsy-6 standout as anomalous examples of this regulatory class. These observations suggest that the majority of animal miRNAs may function in specific non-developmental events or have extensive inter-family redundancies. Furthermore, miRNA-mediated regulation of specific physiological processes is likely coordinated with other regulatory mechanisms such as transcriptional control and protein degradation (Leung and Sharp, 2010). Therefore, functional studies of miRNAs under specific physiological conditions demand effective methods to simultaneously study multiple miRNA families and to identify relevant in vivo targets.
Rather than focusing on individual miRNA-target interactions, we have designed a study to identify the major physiological activities of the miRNA-mediated regulatory network, as a whole, in specific tissues. To this end, we have developed a tissue-specific miRNA-induced silencing complex (miRISC) immunoprecipitation (IP) method to systematically identify miRNA activities and their targets in two tissues. This approach has permitted the identification of hundreds of miRISC-associated mRNAs in both the muscle and intestine of C. elegans, and, when coupled with genetic analysis of overall miRNA function, revealed a complex role of miRNAs in limiting the basal activity animals’ pathogen response system. The data presented in the paper also yield a valuable resource for physiological studies of miRNA-mediated gene regulation in these tissues
We have previously developed an effective biochemical method to systematically identify miRISC-associated miRNAs and their targets in C. elegans by IP of RNAs and proteins associated with AIN-1 or AIN-2 followed by microarray analysis and deep sequencing (Zhang et al., 2007; Zhang et al., 2009). AIN-1 and AIN-2 are GW182-family proteins that are essential components of miRISCs and, to a large extent, redundant in their functions (Ding et al., 2005; Zhang et al., 2007). To investigate miRNA activities in specific tissues, we further developed this system to permit the selective IP of the miRISC from specific tissues. Intestine- and body-wall muscle-specific expression of GFP-tagged AIN-2 protein was directed by single-copy, integrated transgenes driven by the ges-1 and myo-3 promoters, respectively (Supplemental Figure S1). Lysates of mixed-stage worms were subjected to GFP IPs as previously described (Zhang et al., 2007; Zhang et al., 2009). The associated mRNAs from four separate biological replicates were identified by microarray.
We identified 516 mRNAs, encoded by 502 genes, and 569 mRNAs, encoded by 551 genes, significantly enriched (p < 0.01) in the intestinal and body-wall muscle AIN-2 miRISCs, respectively (Figure 1A). Because the intestine is intimately involved in organismal homeostasis and interactions with the environment, we also performed intestine-specific miRISC IPs in worms synchronized at the fourth larval stage (L4) to potentially identify miRNAs involved in non-developmental processes. These IPs identified 463 transcripts, encoded by 446 unique genes, enriched in the L4 intestinal miRISC. The intestinal IPs from asynchronous and L4 cultures showed substantial overlap (242 mRNAs, p < 1 × 10−100) (Figure 1B); however, many transcripts were identified in the L4 IPs, but not in the asynchronous IPs, suggesting that the stage-restricted IP yields greater sensitivity to stage-specific miRNA targets due to reduced heterogeneity of the input RNA. There was minimal overlap between the GFP control IP and the AIN-2 IPs (Figure 1B).
More than 60% of the mRNAs indentified in each AIN-2 IP were also found in our previous AIN-1 or AIN-2 IPs from whole worms (Zhang et al., 2007), (Supplemental Figure S1C), suggesting that the tissue-restricted AIN-2 IPs partition the mRNAs that we have previously identified as miRISC-associated, and also identify likely tissue-specific miRISC-associated mRNAs that eluded previous non-tissue specific IPs. We matched AIN-2-associated mRNAs from each IP to 3’UTR sequences identified by Jan et al. (Jan et al., 2011) (Supplemental Table S2). 3’UTRs from AIN-2-associated mRNAs were significantly enriched, compared to their respective sets of testable mRNAs, for perfect 7-mer (nucleotides 2–8) (Supplemental Figure S2A) and perfect 8-mer (nucleotides 2–9) (Supplemental Figure S2B) seed matches to at least one annotated worm miRNA. Consistent with the notion that miRNA-regulated mRNAs on average have longer 3’UTRs than unregulated mRNAs (Cheng et al., 2009; Stark et al., 2005), we found that the median length of AIN-2 associated mRNAs in each IP was significantly longer than that of the corresponding set of testable mRNAs (Supplemental Figure S2C). Importantly, the longer 3’UTRs do not account for the enhanced frequency of miRNA seed matches in 3’UTRs of AIN-2-associated mRNAs, as these mRNAs show more perfect 7-mer seed matches per 1000nt of annotated 3’UTR than the corresponding sets of testable mRNAs (Figure 2A). We next restricted our seed match analysis to miRNAs present in high abundance in each IP sample (those in the top 50th, 25th, and 12.5th percentile of all detectable miRNAs by raw read count in each IP RNA, see analysis below) and found that restricting the seed match analysis to increasingly abundant miRNAs lead to significantly more robust seed match enrichment in IP mRNAs compared to testable mRNAs (Figure 2B and Supplemental Table S3). Together, these data strongly suggest that mRNAs enriched in the tissue-specific AIN-2 IPs are targets of miRNAs present in the corresponding tissue.
Finally, we examined known intestine-enriched mRNAs in each IP. We found that previously identified intestine-enriched mRNAs (>1500) were not systematically enriched in the intestinal AIN-2-IPs, whereas they were, as a population, markedly depleted from the muscle AIN-2 IP (see Supplemental Experimental Procedures and Supplemental Table S1). This indicates that the AIN-2 IP has no systematic bias towards highly expressed mRNAs in the intestine, consistent with a previous analysis of the method (Zhang et al., 2009). However, as expected, a subpopulation of these intestine-enriched mRNAs were found to be reproducibly enriched in the intestine IPs, which as a whole, contained 27.7% (11.3% expected from chance, p < 1.0 × 10−19) intestine-enriched mRNAs in asynchronous cultures and 17.5% (12% expected, p = 3.5 × 10−4) in L4 cultures, compared to 12.5% (11% expected p = 0.14) in the muscle IPs (Supplemental Figure S1D). Together, these data indicate that the AIN-2 IP from intestine and muscle is highly tissue-specific and enriches for a select subset of mRNAs in each tissue. As indicated in previous studies, these AIN-2-associated mRNAs are high-confidence miRNA targets (Zhang et al., 2007; Zhang et al., 2009; Zisoulis et al.).
We cloned the set of miRNAs that co-IPed with AIN-2 in each tissue. Deep sequencing analysis of these clones (four biological replicates per sample) showed that, on average, AIN-2 IPs yielded ~100-fold more miRNA reads than IPs from the GFP control strain (Supplemental Figure S3). Most miRNAs were detectable in IPs from both asynchronous intestine and muscle samples; however, some miRNAs showed specific enrichment in either the intestine or muscle IPs compared to their relative abundance in total worm lysate (Figure 3A).
Seventeen miRNAs showed significant (p < 0.01) enrichment in intestinal AIN-2 IP samples compared with their relative abundance in Total RNA (Figure 3A). This included all members of the lin-4 and miR-72 families, and two members of the miR-58 and miR-44 families. This is consistent with the observation that members of the same miRNA family are often co-expressed (Martinez et al., 2008). Notably, among the 17 intestine-enriched miRNAs, only miR-70, miR-238, and miR-248 showed no enrichment in muscle-specific miRISCs, suggesting that many miRNAs have activity in both muscle and intestine.
miRNA activities specific to intestine or muscle tissue of worms have been poorly characterized with the exception of the phylogenetically conserved, muscle-specific miRNA miR-1 (Simon et al., 2008). We detected an approximately 20-fold enrichment of miR-1, as well as an enrichment of its validated targets unc-29 and mef-2 in the AIN-2 IPs from muscle but not from asynchronous intestine (Figures 3A and 3B). These data indicate that the tissue-specific miRISC IPs can specifically enrich for both tissue-restricted miRNAs and their targets.
We next examined the frequency of perfect 7-mer seed matches to tissue-enriched miRNAs in muscle and intestine. We could detect a modest enrichment of seed matches to some tissue-restricted miRNAs compared to related scrambled sequences (see Supplemental Experimental Procedures). For example, miR-1 seed matches are enriched in the muscle IP, in contrast to seed matches to the neuron-specific miRNA lsy-6 in both muscle and intestine. However, the strongest seed match enrichment appeared to come from broadly expressed miRNAs such as those from the miR-51 family, miR-80/81/82, and the let-7 family (Figures 3C and 3D and Supplemental Figure S3).
To identify the functions of miRNAs in each tissue, we compared the sets of AIN-2-associated mRNAs to a variety of microarray expression data sets generated by other labs. Most notably, we found a prominent and statistically significant over-representation of pathogen-responsive genes in the sets of AIN-2-associated mRNAs from both the asynchronous and L4-synchronized intestine, but not muscle (Table 1A). The over-representation of pathogen responsive genes was statistically significant under several pathogenic conditions, including exposure to the Pseudomonas aeruginosa strain PA14 compared to OP50 (E. coli) for 4, 8, or 24 hours (Shapira et al., 2006; Troemel et al., 2006); PA14 compared to non-pathogenic PA14 gacA (Troemel et al., 2006); as well as exposure to a variety of other pathogenic bacteria that infect the worm via the intestine (Bolz et al.; Wong et al., 2007). In total, approximately 20% of the mRNAs associated with the AIN-miRISC in the intestine are encoded by known pathogen-responsive genes, whereas only 6% of mRNAs in muscle miRISCs are encoded by pathogen-responsive genes (Table 1B).
Dicer is an RNAse III enzyme required for the biogenesis of many small RNAs including miRNAs (Bernstein et al., 2001). Welker et al. previously reported a set of genes that are deregulated in dcr-1(loss-of-function) (lf) worms (Welker et al., 2007). We therefore examined the sets of mRNAs associated with AIN-2 in intestine and muscle for genes that were previously shown to be up-regulated in dcr-1(lf) worms. We found that both the intestine (3.0-fold higher than expected, p < 1 × 10−20) and muscle (2.2-fold higher than expected, p = 4.5 × 10−11) data sets were enriched for genes up-regulated in dcr-1(lf) (Figure 4A). Remarkably, when we examined the set of intestinal AIN-2-associated mRNAs that are encoded by known pathogen-induced genes, we found an even more striking enrichment for dcr-1-responsive genes (7.0-fold above what is expected by chance, p = 1.0 × 10−15). Indeed, nearly 50% of genes in this class are up-regulated in dcr-1-deficient animals. This observation not only supports the hypothesis that some pathogen-responsive mRNAs are targeted to the AIN-miRISCs by DCR-1-produced miRNAs, but also suggests that this miRISC targeting results in mRNA degradation.
To further test this hypothesis, we examined mRNA levels of a subset of these genes in ain-2(reduction-of-function)(rf);ain-1(lf) double mutants in which miRISC activity is strongly and specifically reduced (Zhang et al., 2007). We found a significant up-regulation in 8 of the 9 AIN-2-associated, dcr-1-and pathogen-responsive genes we tested (we refer to these genes as miRISC-regulated pathogen-response [MRPR] genes hereafter) (Figure 4B). As controls, we examined the expression of ges-1 (an intestine-specific, pathogen non-responsive gene), F08G5.6 (a pathogen-responsive gene encoding mRNAs that were not detected in the miRISC), and sod-3 (a general stress-induced gene) and found no change in the expression of these genes in ain-2(rf);ain-1(lf) double mutants. Therefore, up-regulation of AIN-associated, pathogen-responsive mRNAs is not the result of a gross deregulation of the pathogen response, a stress response, or the intestinal transcription program as a whole. Together, these data suggest that dcr-1 and ain-1/2 cooperate in the intestine to regulate basal mRNA levels of a subset of pathogen-responsive genes.
The direct association of the MRPR mRNAs with the miRISC suggests that miRNA-mediated targeting of mRNAs to the miRISC may result in their degradation. However, we were surprised by the magnitude of the up-regulation of some of these genes in the ain-2(rf);ain-1(lf) double mutant, since miRNA-mediated mRNA degradation is generally believed to result in small changes in mRNA levels. We noticed that one of the MRPR genes, tsp-1, resides in an operon with a second gene, tsp-2. Both of these genes are regulated by the pathogen response (Troemel et al., 2006), but only tsp-1 shows a statistically robust association with the miRISC (Figure 4C). Consistent with our observed association of tsp-1 mRNA, but not tsp-2 mRNA, with the intestinal miRISC, annotation of tsp-1 by previous analysis reveals that it has a relatively long 3’UTR compared to tsp-2 (146 nucleotides for tsp-1 compared to 69 nucleotides, including several low complexity poly-A tracts, for tsp-2; (Jan et al., 2011). Furthermore, both mirWIP and TargetScan predict tsp-1 but not tsp-2 to be a miRNA target (Supplemental Table S6). Since these genes are transcriptionally coupled, differential regulation of their relative mRNA levels likely reflects post-transcriptional regulation. We therefore examined the induction of both mRNAs in the ain-2(rf);ain-1(lf) double mutant and unexpectedly found that both tsp-1 and tsp-2 were markedly up-regulated in these mutants. However, the up-regulation of tsp-1 in the ain-2(rf);ain-1(lf) double mutant was consistently over three-fold higher than tsp-2. These data suggest that the miRISC may indirectly regulate the transcription of the tsp-1/tsp-2 operon, as well as directly regulate the post-transcriptional degradation of tsp-1 (but not tsp-2). Collectively, our data suggest that miRNAs target many pathogen-responsive mRNAs to the AIN-miRISC where they are degraded; however, miRNAs likely also indirectly regulate the transcription of some pathogen response genes.
We next examined the functional consequences, with respect to the organismal pathogen response, of altered miRISC activity. Although ain-1 and ain-2 are apparently biochemically redundant with largely overlapping functions (Zhang et al., 2007), a single mutant of either gene is expected to compromise, to some extent, the overall miRISC function, possibly severely disrupting specific miRNA-target interactions that are normally mediated predominately by one of the two GW182 proteins (Zhang et al., 2007). We therefore tested if loss-of-function alleles in either ain-1 or ain-2 result in a pathogen response phenotype. We found that, whereas loss of function of ain-2 yielded no discernable phenotype, three ain-1(lf) alleles conferred an enhanced resistance to P. aeruginosa strain PA14 (Figure 5A). Similarly, wild-type worms treated with ain-1 RNAi showed a significantly enhanced resistance to P. aeruginosa compared to wild-type worms treated with a control gfp RNAi (Supplemental Figure S4B). Thus reducing overall miRNA functions enhances the ability of worms to respond to pathogenic insult by P. aeruginosa. This physiological consequence of ain-1 mutations likely reflects the collective effect of increased expression of many pathogen-responsive mRNAs that are normally regulated by miRNAs. However, it should be noted that the functional contributions of these genes to pathogen responses have not been directly determined by genetic analysis.
Consistent with a previous observation that pathogen resistance is often correlated with reduced fecundity (Miyata et al., 2008), we observed a modestly reduced brood size in each of the three ain-1(lf) alleles (Figure 5B). However, reduced matricidal bagging in ain-1(lf) animals is not likely to account for the pathogen resistance phenotype since fer-15;ain-1(lf) animals also showed resistance to PA14 relative to fer-15 animals under conditions in which both strains were sterile (Supplemental Figure S4C).
The p38 kinase PMK-1-mediated signaling pathway has been shown to play a crucial role in regulating the pathogen response in C. elegans (Garsin et al., 2003; Kim et al., 2002). In addition, activation of the DAF-16 FOXO transcription factor acting downstream of the DAF-2 insulin receptor was shown to confer pathogen resistance (Garsin et al., 2003). To determine how the ain-1 activity within the pathogen response relates to these pathways, we first examined if any known PMK-1 (Troemel et al., 2006) or DAF-2/DAF-16 (Murphy et al., 2003) pathway target genes were detectable in the intestinal AIN IPs. Target genes from both pathways were modestly over-represented in the intestinal AIN IPs from both asynchronous and L4 worms (Figure 5C), suggesting that the miRISC may function as a secondary, post-transcriptional regulator of some targets of these pathways. We next tested the ability of the MRPR genes to be induced in the absence of either pmk-1 or daf-16. We found that nearly all of the genes could be robustly induced by exposure to P. aeruginosa, even in the absence of pmk-1 or daf-16 (Figure 5D). These data suggest that ain-1/2 may function, at least in part, in parallel to these pathways to regulate the expression of genes involved in the pathogen response.
We next used a genetic approach to further characterize the relationship of ain-1 to pmk-1 and daf-16 in the pathogen response. We found that while daf-16(lf) mutants showed largely wild-type survival rates on P. aeruginosa, daf-16(lf);ain-1(lf) were nearly identical to ain-1(lf) single mutants in their resistance to P. aeruginosa (Figure 5E). This indicates that ain-1(lf) does not confer resistance to P. aeruginosa by activating daf-16. In contrast, a pmk-1(lf); ain-1(lf) double mutant is hypersensitive to P. aeruginosa infection, showing a sensitivity similar to that of a pmk-1(lf) mutant alone. This indicates that ain-1-mediated pathogen resistance requires pmk-1 activity. Therefore we conclude that the major miRISC activity likely represents a regulatory pathway that functions in parallel to pmk-1.
To identify miRNAs that function in the pathogen response, we extended upon our identification of intestine-enriched miRNAs by identifying miRNAs that are enriched in the L4 intestine and miRNAs that show altered association with the intestinal miRISC following exposure to P. aeruginosa. miRNAs were cloned and deep sequenced from IP samples from Pges-1::ain-2::gfp transgenic worms synchronized at L4 and treated with non-infectious OP50 E. coli or pathogenic PA14 P. aeruginosa. This approach identified six other miRNAs showing statistically significant (p < 0.01) enrichment in L4 intestinal miRISCs (Figure 6A) under non-pathogenic conditions. The specificity of these IPs was highlighted by the significant depletion of the muscle-specific miRNA miR-1 (p = 0.007) and the neuron-specific miRNA lsy-6 (p = 0.02) from the intestine-specific miRISC IPs. When we compared the relative abundance of miRNAs in both the intestinal miRISCs of OP50- and PA14-fed worms we detected highly reproducible changes in the relative abundance of several miRNAs between the two populations of worms (Figure 6B). These data indicate that the pathogen response is accompanied by dynamic changes in miRNA activities within the intestinal miRISC.
We identified 22 families of miRNAs (including single-member families) that could potentially be involved in the pathogen response owing to their intestinal enrichment or dynamic behavior upon infection. We focused on miRNA families for which exist viable deletions of the majority or entirety of the family. We first surveyed these miRNA deletion strains for de-regulation of the MRPR genes we identified. Whereas the majority the miRNA deletion strains showed wild-type expression levels of MRPR genes, separate loss of the miR-58 family, lin-4 family, miR-251 family, or miR-253 each resulted in robust and significant up-regulation specific subsets of the MRPR genes (Figure 6C). Notably, we observed several instances of regulation of MRPR genes by miRNAs that are not predicted to target their 3’ UTRs (Supplemental Table S6). Therefore ablation of several miRNA families can result in the direct or indirect up-regulation of subsets of pathogen-responsive genes.
We next examined the survival of select miRNA deletion mutants on P. aeruginosa. miRNA families were chosen for this analysis on the basis of marked changes in miRISC-association on PA14 (miR-70, miR-252, miR-253), dramatic enrichment in the intestine (miR-58 family, miR-243), or significant changes in MRPR gene expression (miR-58 family, miR-251 family, miR 253). Only overtly healthy strains were tested. Although deletions of miR-58 family members (data not shown), miR-243 and miR-253 showed no significant difference from wild-type, deletions of either miR-70 or miR-251 and miR-252 showed modestly enhanced survival on P. aeruginosa compared to wild-type (Figure 6D). Taken together, our data demonstrate that miRNAs from several distinct families respond to pathogenic insult, and that miRNAs from distinct families are involved in the regulation of pathogen-responsive genes, as well as the organismal response to pathogen exposure.
The complexity of regulatory miRNA-target interactions in non-developmental processes presents a major obstacle for analyzing miRNA functions with straightforward genetics on individual miRNAs or their targets. To circumvent this difficulty, we have used two broad approaches to identify important functions for miRNAs in C. elegans. First, our tissue-specific miRISC IPs allowed the systematic identification of hundreds of miRNA targets in muscle and intestine. Second, by studying worms in which miRNA activity, as a whole, is compromised, we were able to identify a requirement for miRNAs in the regulation of the worm pathogen response.
C. elegans, like most metazoans, appears to have a specialized, inducible innate immune system that protects the animal from microbial pathogens (Mallo et al., 2002; Tan et al., 1999). In worms, the innate immune system functions predominantly in the intestine, the primary site of attack by ingested pathogens. Multiple signaling pathways regulating pathogen-inducible gene expression have been identified; however, understanding of the mechanisms by which the pathogen response is attenuated when not needed is limited. We show in this study that compromising the regulation of pathogen-responsive genes by miRNAs is generally beneficial to the animal under pathogenic conditions. However, this mis-regulation is expected to be deleterious to the animals under normal conditions. For example, consistent with our observation of reduced fecundity in ain-1(lf) animals, many pathogen-resistant mutants show compromised reproduction even in the absence of pathogenic stimuli (Miyata et al., 2008), potentially highlighting the importance of the mechanisms of silencing the pathogen response. Chronic innate immune activation has been linked to several pathological conditions, including inflammatory bowel disease in humans (Cario, 2010), and in both insects and nematodes the innate immune system appears to compete with the reproductive system for limited resources (Fedorka et al., 2004; McKean and Nunney, 2001; Miyata et al., 2008). Additionally, activation of the innate immune system in worms has been shown to induce potentially lethal endoplasmic reticulum stress (Bischof et al., 2008; Richardson et al.). It is therefore critical to organismal physiology that the innate immune response is completely squelched in the absence of pathogenic stimuli.
Our data suggest that miRNA activity is required to suppress the basal activity of the pathogen response and the expression of its target genes. We found that the collective miRNA activity likely acts in parallel to the major pathogen-response signaling pathways defined by PMK-1 and DAF-2/DAF-16. This suggests the miRISC may function as a secondary, post-transcriptional regulator of many pathogen-response mRNAs, thereby enforcing robust inactivation of the innate immune system under basal conditions. Nonetheless, we also found evidence that miRNAs may indirectly regulate the transcription of some pathogen-response genes (e.g. tsp-2). Thus the miRISC-mediated regulatory network functioning to limit the activity of the pathogen response appears to be complex, potentially acting via multiple direct and indirect mechanisms. The results of this study present an example of how miRNAs function to influence a complex physiological process, similar to what has been hypothesized for the regulation of aging (Ibanez-Ventoso and Driscoll, 2009).
The complexity of this regulatory network is further illustrated by the observation that several miRNA families appear to regulate many of the same MRPR genes. For example, lys-1 is significantly up-regulated in the absence of the miR-58 family, the lin-4 family, and miR-253, yet computational prediction by mirWIP (Hammell et al., 2008) and TargetScan (Lewis et al., 2005) identify among these only the miR-58 family as having a seed match in the lys-1 3’UTR. Similarly, T27F2.4 is significantly up-regulated in the absence of the lin-4 family, as well as other miRNA families, yet it is predicted to have no seed matches for any of these miRNAs. We therefore argue that, regardless of their direct association with the miRISC, target mRNA regulation by miRNAs within the pathogen response may be both direct and indirect.
The precise role of individual miRNAs in the pathogen response is unclear. Given that many pathogen response genes are up-regulated in the absence of functional miRISCs, and that reduction of total miRISC activity enhances worm resistance to P. aeruginosa, we conclude that the collective action of miRNAs suppresses the activity of the innate immune system. Notably, many miRNA showed enhanced association with the intestinal miRISC upon exposure to P. aeruginosa. This suggests that the activity of many miRNAs, for example miR-253, may be induced in concert with the pathogen response in order to reverse the transcriptional up-regulation of several targets upon clearance of the infection. Alternatively, it has been observed that miRNAs are stabilized by the presence of their targets (Chatterjee et al., 2011). It is therefore possible that miRNAs such as miR-253 are constitutively transcribed, but degraded in the absence of the pathogen response and their cognate targets. Upon infection, these miRNAs target a subset of pathogen-induced mRNAs to ensure the transient nature of their expression, and consequently the miRNAs are themselves transiently stabilized.
Together, our systematic approach not only advances our understanding of miRNA functions in regulating a specific, non-developmental event, but also provides important insights into the multi-tiered mechanisms by which tight control of stress-responsive genes is maintained. The precise nature of miRNA regulatory network controlling aspects of the pathogen response remains to be fully explored.
Worm strains were maintained according to the methods of Brenner. Strains were maintained at 20°C on OP50 E. coli. Methods for synchronization, brood size measurement and alleles used can be found in the Supplemental Experimental Procedures.
Transgenes were constructed by fusing a single FLAG tag to the 3’ end of the ain-2::gfp fusion and Pain-2::gfp construct described previously(Zhang et al., 2007). For intestine- and muscle-specific expression 3.1 Kbp of the ges-1 and 2.5 Kbp of the myo-3 promoters, respectively, were cloned in place of the ain-2 promoter. Single-copy integrated transgenics were established by ballistic transformation(Wilm et al., 1999).
Was performed essentially as described (Zhang et al., 2007; Zhang et al., 2009), with the exception that the worms were cross-linked with formaldehyde prior to lysis according to the method of Roy et al. (Roy et al., 2002). Detailed methods of IP and data analysis can be found in the Supplemental Experimental Procedures.
Worms survival on lawns of P. aeruginosa isolate PA14 was performed according to the method of Tan et al. (Tan et al., 1999).
All quantitative RT-PCR data was normalized to the average Ct of three housekeeping genes, rpl-26, act-1, eft-2. Reverse transcription reactions were primed with oligo-dT. qRT-PCR data for additional control genes can be found in Supplemental Figure S4A). See Supplemental Experimental Procedures for detailed methods.
We thank S. Mitani, R. Horvitz, and the Caenorhabditis Genetics Center for providing strains. M. Heinz at GTAC of Washington University and J. Dover at University of Colorado, Denver for assistance. We thank M. Cui, M. Kniazeva, J. Cavaleri, M. Than, C. Johnson, D. Ronai, B. Weaver, A. Sewell, Y. Teng, R. Yi, T. Blumenthal, B. Wood, A. Weiner, B. Kennedy, E. Smith, and E. Spindler for invaluable discussions and feedback. This study was supported by NIH grants F32-GM087902 to BAK and R01-GM47869 to MH.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Supplemental material includes four data sets containing enrichment data for each IP presented, four figures, six tables, Supplemental Experimental Procedures, and Supplemental References.
The Gene Expression Omnibus accession number for microarray analysis of AIN-2-IP mRNAs is pending.