HIF-1 and SKN-1 Coordinate the Transcriptional Response to Hydrogen Sulfide in Caenorhabditis elegans

Hydrogen sulfide (H2S) has dramatic physiological effects on animals that are associated with improved survival. C. elegans grown in H2S are long-lived and thermotolerant. To identify mechanisms by which adaptation to H2S effects physiological functions, we have measured transcriptional responses to H2S exposure. Using microarray analysis we observe rapid changes in the abundance of specific mRNAs. The number and magnitude of transcriptional changes increased with the duration of H2S exposure. Functional annotation suggests that genes associated with protein homeostasis are upregulated upon prolonged exposure to H2S. Previous work has shown that the hypoxia-inducible transcription factor, HIF-1, is required for survival in H2S. In fact, we show that hif-1 is required for most, if not all, early transcriptional changes in H2S. Moreover, our data demonstrate that SKN-1, the C. elegans homologue of NRF2, also contributes to H2S-dependent changes in transcription. We show that these results are functionally important, as skn-1 is essential to survive exposure to H2S. Our results suggest a model in which HIF-1 and SKN-1 coordinate a broad transcriptional response to H2S that culminates in a global reorganization of protein homeostasis networks.


Introduction
Exogenous H 2 S has dramatic effects on mammalian physiology that can improve survival in changing environmental conditions.Mice exposed to H 2 S enter into a hibernation-like state that allows them to endure periods of low metabolic rate and decreased core body temperature without apparent ill effects [1].The H 2 Sinduced state enables mice to survive exposure to otherwise lethal hypoxic conditions [2], and improves outcome in rodent models of severe blood loss [3], myocardial infarction [4], aortic occlusion [5] and hepatic ischemia/reperfusion [6].
C. elegans grown in H 2 S have increased thermotolerance and lifespan [7].Increased lifespan and thermotolerance require the conserved sirtuin homologue sir-2.1,though mutant animals with deletions in sir-2.1 grow normally in H 2 S. In contrast, the hif-1 transcription factor is required to survive exposure to H 2 S [8].hif-1 is a highly conserved bHLH transcription factor that is wellknown for its role in coordinating the transcriptional response to hypoxia in all animals, including C. elegans [9,10].Sirtuins have been shown to modulate lifespan in yeast, worms, flies and mice [11].Recent work has demonstrated that HIF-1 activity can influence lifespan in C. elegans [12,13,14].Thus, the response to H 2 S involves at least two genes, hif-1 and sir-2.1, which influence lifespan.
In this study, we investigated the transcriptional response to H 2 S in C. elegans.Using an unbiased microarray approach, we show that there are rapid and progressive changes in mRNA abundance associated with exposure to H 2 S. Functional genomic analysis suggests that adaptation to H 2 S results in significant changes to protein homeostasis pathways.We found that hif-1 is required for nearly all of the early changes observed, though there is little overlap between genes regulated in response to H 2 S and those that have been reported to change in hypoxia.Moreover, our data show that other factors contribute to coordinate the response to H 2 S, as we found that some H 2 S-induced transcriptional changes require the skn-1 transcription factor.We demonstrate that, like hif-1, skn-1 is required to survive in low concentrations of H 2 S.These data suggest a possible model in which HIF-1 and SKN-1 act together to coordinate a transcriptional response to H 2 S that ultimately leads to alterations in the expression of genes involved in protein homeostasis.

H 2 S exposure leads to rapid and progressive changes in mRNA abundance
To investigate the transcriptional responses to H 2 S in C. elegans, we employed a microarray approach to identify mRNAs that were altered in abundance by exposure to H 2 S. In these experiments, we exposed synchronized cultures of C. elegans to 50 ppm H 2 S for 1, 12 or 48 h (schematized in Fig. 1A).We measured the response to 50 ppm H 2 S, as this is the same concentration of H 2 S that increases lifespan and thermotolerance [7].Exposure to H 2 S was immediately prior to harvest.We previously showed that developmental rate is not affected by this concentration of H 2 S [7], ensuring that all animals were first-day adults when RNA was harvested.This experimental design enabled us to compare transcript abundance without confounding effects from comparing different developmental stages.
The number of transcriptional changes we observed in animals exposed to H 2 S increased with the duration of exposure (Figure 1B).After 1 h exposure we detected significantly altered mRNA levels of 17 transcripts (adjusted p-value,0.05,see Experimental Procedures).All but one of these transcripts was more abundant in the animals exposed to H 2 S compared to untreated controls (Table 1), suggesting that the transcription of these genes was increased upon exposure to H 2 S. The effect of H 2 S on transcript abundance progressed with increased duration of exposure to H 2 S, both in magnitude of effect on mRNA level and number of gene products affected.After 12 h exposure to H 2 S, we observed 445 mRNA that were significantly changed, 259 (58%) of which were more abundant after exposure to H 2 S (Table S1).Nine of the 16 gene products that were increased after 1 h exposure to H 2 S were still significantly increased after 12 h in H 2 S. The scope and magnitude of transcript alterations was even more pronounced after 48 h in H 2 S, the time required to observe an increase in lifespan and thermotolerance [7].We observed 5,089 genes that had significantly altered mRNA levels after 48 h exposure to H 2 S (Table S2), which represents 44% of the gene products included in the analysis.Of the significantly altered transcripts, 143 were increased by at least 5.5fold and 126 were decreased by at least 5.5-fold.Together, these data indicate that there is a rapid and progressive induction of transcriptional activity upon exposure to H 2 S.
We used quantitative reverse transcript PCR (qRT-PCR) to validate changes in mRNA levels that we observed by microarray.We focused on transcripts that changed after brief exposure to H 2 S, reasoning that these early transcriptional changes represent the initial response to H 2 S and may be important to set up later, more progressive changes.When wild-type animals were grown on E. coli strain OP50, 6 of the 11 gene products that were predicted to be increased based on the microarray data were more abundant after 1 h exposure to H 2 S (Table 2).Several of the gene products trended toward higher expression, but did not reach significance in this assay.For many of these, we noticed that the level of transcript measured in the untreated sample was near background, which may have increased the variance in the measurements.These changes did reach significance when animals were grown on HT115(DE3) RNAi control food.In these conditions, 10 of the 11 gene products tested were upregulated after 1 h exposure to H 2 S (Table 2).In general, the magnitude of H 2 S-induced changes in transcript abundance was greater on RNAi food than on OP50.The source of this variation is unclear, but may hint at an effect of nutritional status on adaptation to H 2 S. Indeed, the HT115(DE3) bacterial food used for feeding RNAi has previously been shown to affect survival in hypoxia (DLM and MBR, unpublished observation and [15]), consistent with the idea that nutritional status can influence responses to environmental changes.The one gene product that was less abundant after exposure to H 2 S by microarray, C18H7.1, was not significantly altered after exposure to H 2 S in either nutrient condition in qRT-PCR measurements.Further validating these results, our microarray data corroborate previous studies that showed T05B4.2 (nhr-57) and K10H10.2(cysl-2) are upregulated after short exposure to H 2 S [8].We conclude that our microarray experiments identified transcripts that change in mRNA abundance associated with exposure to H 2 S.

Function annotation of genes induced by exposure to H 2 S
Only 16 gene products were more abundant after exposure to H 2 S for 1 h, which precluded the use of bioinformatic analysis to measure enrichment of functional classes.However, we noticed that many of the genes on this list are annotated to be involved in cellular metabolic processes (Table 1).The most-highly expressed gene in response to H 2 S is a glutathione-S transferase, F37B1.8 (gst-19).Two other up-regulated genes are predicted to have a role in amino acid metabolism, including the rate-limiting enzyme in serine production, phosphoglycerate dehydrogenase (C31C9.2),and an enzyme with homology to cysteine synthase (cysl-2, K10H10.2).We also observed upregulation of nit-1 (ZK1058.6), a predicted carbon-nitrogen hydrolase.In addition to metabolic enzymes, exposure to H 2 S also resulted in the upregulation of 6 of the 8 nspe (nematode specific peptide, class E) genes.There is little known about these genes, other than they code for short peptides, 70-75 amino acids long, that are annotated to be integral to the membrane.Although the nspe transcripts were greatly increased in abundance after 1 h exposure to H 2 S, they were not significantly changed after 12 h exposure to H 2 S, suggesting that they were only transiently upregulated.Instead, after 12 h in H 2 S we found that 8 of the 12 nspa (nematode specific peptide, class A) transcripts were significantly upregulated (Table S1).
To evaluate functional categories of genes over-represented in the microarray dataset of transcripts changed after longer exposure to H 2 S, we employed the online Database for Annotation, Visualization and Integrated Discovery (DAVID, v6.7) functional annotation clustering tool [16,17].We focused on gene products that were most increased in response to H 2 S, as we Plots show magnitude of change in transcript level (log 2 FC) as a function of adjusted p-value (log 10 p-value).Each point is data from one gene product.Significant changes (adj.p-value,0.05)are red.After 1 h exposure to H 2 S (left), 16 genes were significantly up-regulated and one was down-regulated (Table 1).After 12 h exposure to H 2 S (middle), 445 transcripts were significantly changed, with 259 up-regulated (Tables 1  and S1).After 48 h in H 2 S (right), 5089 transcripts were significantly altered relative to untreated controls (Table S2).doi:10.1371/journal.pone.0025476.g001hypothesize that these might be important for the phenotypic changes that we observe and likely to be most robust.This analysis showed that there were two highly enriched functional clusters in the genes increased by at least 3-fold after 12 h exposure to H 2 S (Table 3, Table S3).The most enriched cluster included gene ontology terms related to aging and stress resistance.This result is consistent with our previous observation that adaptation to H 2 S increases lifespan and thermotolerance [7].
We observed that proteins with the F-box motif were enriched in the genes upregulated by H 2 S. F-box proteins were first identified as members of the SCF (Skp-cullin-F-box) ubiquitin ligase complex that targets proteins for ubiquitination and eventual degradation [18,19].F-box proteins were significantly enriched in genes upregulated after 12 h exposure to H 2 S (Table 3).These data raise the possibility that the response to H 2 S alters the stability of proteins regulated by the ubiquitinproteosome system.When we extended this analysis to define the functional clusters in gene products upregulated by at least 5-fold after 48 h exposure to H 2 S, we observed an even greater enrichment of F-box containing proteins (Table 3).Moreover, the second most enriched cluster included genes associated with the BTB/POZ domain, another protein-protein interaction motif that has been associated with SCF function [20].We did not observe an enrichment of genes involved in aging or stress resistance after 48 h exposure to H 2 S. Together, these data show that the transcriptional response to H 2 S includes many gene products involved in protein turnover mediated by the ubiquitin proteosome system.
The hif-1 transcription factor is required for H 2 S-induced transcriptional changes HIF-1 is the C. elegans homologue of the highly conserved hypoxia inducible transcription factor, best known for its role in coordinating the transcriptional response to decreased O 2 [9,10,21].Recently, it has been demonstrated that C. elegans require hif-1 to survive exposure to H 2 S [8].On exposure to H 2 S, the HIF-1 protein accumulates and is localized to the nucleus.At least some transcriptional responses to H 2 S result from the activation of HIF-1, including K10H10.2 (cysl-2), nhr-57, and sqrd-1 [8,22].These observations motivated us to consider the possibility that our microarray data might reveal other hif-1dependent transcriptional responses to H 2 S.
We measured the abundance of mRNA in hif-1(ia04) mutant animals exposed to H 2 S for 1 h by qRT-PCR to evaluate if hif-1 is required for transcriptional changes that occur upon exposure to H 2 S.This short exposure to H 2 S was sufficient to induce transcriptional changes in wild-type, but was not lethal to the hif-1(ia04) mutant animals.In fact, after 1 h in H 2 S the hif-1(ia04) mutant animals were still moving normally [8].We found that for 5 of 11 mRNAs tested, transcript levels in hif-1(ia04) mutant animals were significantly lower than wild-type (Figure 2).This includes the most highly-induced mRNAs.In the hif-1(ia04) mutant animals we observed very little change in any transcript abundance after exposure to H 2 S, both for messages that were highly induced as well as the lower-expressed transcripts where statistical significance of p,0.05 was not achieved.We conclude that hif-1 has a centrally important function in inducing transcriptional changes associated with exposure to H 2 S, particularly those changes that occur immediately upon exposure to H 2 S.
Although the response to hypoxia and H 2 S both require the hif-1 transcription factor, there is little overlap between genes regulated in these two conditions.A previous report identified 68 hypoxia-induced, HIF-1-dependent transcriptional changes in C. elegans [15].Of the 16 genes upregulated after 1 h exposure to H 2 S, only 3 (19%) are also regulated by hif-1 in hypoxia: nhr-57, rhy-1 and K10H10.2(cysl-2) (Figure 2B, Table S4).The slight overlap between these data sets is greater than would be expected by chance (hypergeometric probability 0.001), consistent with our observation that that hif-1 is required for both responses.Similarly, of the 298 transcripts that are more abundant after 12 h exposure to H 2 S, 8 (3%) are also regulated by hypoxia (hypergeometric probability 0.006; Figure 2C, Table S4).Since hif-1(ia04) mutant animals die after 12 h exposure to H 2 S we could not determine which of these changes were hif-1-dependent.Thus, we included all hypoxia-induced genes, regardless of whether they require hif-1.

Role of SKN-1 in response to H 2 S
We noted many Sdz genes were upregulated after 12 or 48 h exposure to H 2 S, including several with F-box domains (Tables S1  and S2).Sdz genes were named for their skn-1 dependent zygotic expression during embryogenesis between the 4-and 12-cell stage [23].Several Sdz transcripts that were more abundant after exposure to H 2 S contained F-box domains (Table S3).skn-1 is a maternally-supplied factor required early in embryogenesis for specification of the EMS blastomere that gives rise to mesendodermal lineages [24], and acts postembryonically in the intestine to control the phase II response to oxidative stress [25] and in the two ASI neurons to control the effects of dietary restriction on lifespan [26].The abundance of Sdz genes on the list led us to consider the possibility that skn-1 is also involved in the response to H 2 S.
To test the possibility that H 2 S-induced transcriptional changes depended on skn-1, we measured mRNA abundance by qRT-PCR from N2 wild-type animals raised on skn-1(RNAi) and exposed to H 2 S for 1 h as adults.These animals laid only dead eggs, demonstrating that SKN-1 levels were reduced below those required for early embryonic development.We found that 7 of the gene products showed skn-1-dependent changes in abundance after 1 h exposure to H 2 S (Figure 3).Three genes regulated in a skn-1-dependent manner in response to H 2 S, C31C9.2,K10H10.2 and ZK1058.6 (nit-1), were previously shown to be regulated by skn-1 postembryonically [27], as was embryonic expression of ZK1058.6 in the EMS lineage after the 4-cell stage and in E descendants after the 300-cell stage [23].In addition to previously-identified skn-1 dependent transcripts, H 2 Sinduced upregulation of Y38D10A.12 and T05B4.1 also required skn-1.The promoter region of all these genes contain core skn-1 consensus binding sequences, RTACT [27] (Figure 3A).We further observed that three gene products up-regulated by 1 h exposure to H 2 S were changed even more dramatically in the skn-1(RNAi) animals: K01H10.2,F37B1.8 (gst-19) and W07A12.7 (rhy-1).These data are consistent with reports that skn-1 can act to negatively regulate the expression of genes involved in the response to some oxidative stresses [27].Our data show that skn-1 acts to both up and down regulate genes in response to H 2 S.
There is minimal overlap between transcriptional responses to H 2 S and previously characterized, post-embryonic skn-1 dependent transcripts (Figure 3B, Table S5).We observed significant overlap between the set of genes changed in response to H 2 S and transcripts that require skn-1 for normal expression in unstressed conditions (16 genes in common, hypergeometric probability 0.006) [27], though 4 of these 16 genes were less abundant after exposure to H 2 S. Similarly, 10 H 2 S-regulated genes were identified as skn-1 depenendent targets in response to arsenic stress (hypergeometric probability 0.01) [27].In contrast, we did not observe significant overlap with skn-1-dependent targets induced by exposure to t-butyl hydroperoxide (hypergeometric probability 0.06) [27], those downregulated in unstressed conditions (hypergeometric probability 0.13) [27], or messages regulated in response to hyperoxia (hypergeometric probability 0.15) [28].Our observations suggest that skn-1-dependent transcriptional responses to H 2 S are somewhat distinct from skn-1 mediated responses to oxidative and xenobiotic stress.This is consistent with accumulating evidence that skn-1 transcriptional outputs are context dependent [27].
We monitored the viability of skn-1 mutant animals exposed to H 2 S to evaluate the functional significance of the skn-1-dependent transcriptional response (Figure 3C).Like wild-type animals, all skn-1(zu169)/nT1 control animals survived exposure to 50 ppm H 2 S (n = 24, 2 independent experiments).In contrast, none of the skn-1(zu169) homozygous mutant animals tested survived (n = 28 in same 2 experiments).To rule out the possibility that H 2 S-induced death was a result of the nT1 balancer chromosome, we crossed the balancer away from the zu169 allele.24% (16/68) of the selfprogeny from zu169/+ heterozygotes died when exposed to 50 ppm H 2 S.This is not statistically different the expected frequency of skn-1(zu169) homozygotes (25%; x 2 = 0.078, df = 1, p.0.05).Of the survivors, 14/15 were fertile, indicating that these animals were not homozygous for the skn-1 allele, which confers a maternal-effect lethal phenotype.We conclude that the skn-1(zu169) homozygous animals died when exposed to H 2 S.These data demonstrate that SKN-1 activity is essential to appropriately respond to H 2 S.

Discussion
Our results indicate that hif-1 and skn-1 cooperate to orchestrate a progressive transcriptional response to H 2 S. Previous studies have demonstrated hif-1 dependent responses to H 2 S [8,22].We have extended this observation using an unbiased microarray approach that identified several new hif-1-dependent transcriptional changes upon exposure to H 2 S. In addition, we have identified skn-1 as another essential factor during exposure to H 2 S.
H 2 S protects mice from otherwise lethal whole-body hypoxia [2] and improves outcome in a variety of rodent models of ischemia/reperfusion injury [4,29,30].The mammalian orthologue of HIF-1 has been implicated in protection against ischemia/ reperfusion in mammals [31].Thus, the observation that HIF-1 is activated by H 2 S suggests a mechanistic basis for the beneficial effects of H 2 S [8].Curiously, our data suggest that there is little Our finding that SKN-1 plays a role in the response to H 2 S is consistent with a recent report that nuclear accumulation of NRF2, a mammalian homologue of SKN-1, is correlated with H 2 S-induced protection against from ischemia-induced heart failure [30].SKN-1 controls the Phase II response to toxins and oxidative stress [25].However, we do not favour the hypothesis that H 2 S activates the Phase II response.The canonical Phase II targets gst-4 nor gcs-1 were not dramatically induced by H 2 S, and the H 2 S-induced upregulation of another glutathione S-transferase, gst-19, was exaggerated in skn-1(RNAi) animals.Moreover, we observed little overlap in genes regulated by skn-1 in response to xenobiotic or oxidative stress and those that are changed in H 2 S. Instead, we found at least 7 of the gene products included in the F-box and BTB/POZ clusters (Table 3) were identified as skn-1-dependent zygotic transcripts [23].Finally, we did not observe obvious accumulation of skn-1::GFP in the intestinal nuclei of animals exposed to H 2 S using epifluoresence microscopy (not shown), although we cannot rule out the possibility that the nuclear enrichment of GFP was below the detection limit in this experiment.We suggest that, during adaptation to H 2 S, skn-1 may play a role in remodeling the protein turnover machinery.
Protein homeostasis is increasingly appreciated for its importance to aging and age-associated decline [32].We propose that one consequence of adaptation to H 2 S is to increase transcription of genes related to protein turnover by the ubiquitin ligase and proteasome system, including F-box and BTB/POZ domain proteins.In this model, the effect of H 2 S to increase lifespan and thermotolerance may be attributed, at least in part, from effects on protein homeostasis.Further understanding the mechanisms by which adaptation to H 2 S can improve homeostasis and influence lifespan may provide novel insights into the mechanisms that mediate the beneficial effects of H 2 S in mammals.
C. elegans were grown on nematode growth medium plates seeded with live Escherichia coli OP50 food (NGM/OP50 plates) as described previously [34].All experiments and worm culturing were conducted at room temperature to avoid effects resulting from changing temperature.Exposure to H 2 S was in continuous flow H 2 S chambers that were created as previously described [7], by continuously diluting 5000 ppm H 2 S/balance N 2 (Airgas, Seattle, WA USA) with house air to a final concentration of 50 ppm H 2 S. For viability assays, worms were exposed to H 2 S as fourth-stage larvae (L4) and scored for survival after 18-24 h.The effect of skn-1(zu169) mutations on viability in H 2 S was determined by picking sterile, non-Unc progeny of skn-1/nT1 animals, and compared to Unc heterozygous siblings.To cross the zu169 allele away from the nT1 balancer, N2 males were crossed with skn-1/nT1 mutant hermaphrodites.Non-Unc heterozygous F1 were allowed to produce F2 progeny, which were scored for sensitivity to H 2 S and the skn-1 maternal effect lethal phenotype.
skn-1(RNAi) animals were generated by feeding N2 from starved L1 on HT115(DE3) bacteria carrying either the skn-1 clone or empty vector control (L4440) from the Ahringer library [35].exposure to H 2 S overlap slightly with hif-1-dependent changes in response to hypoxia. 3 of 16 transcripts upregulated in response to 1 h exposure to H 2 S were identified as hif-1-dependent targets in hypoxia (n = 68) [15].The probability of observing this overlap randomly is 0.001.C.There is minimal overlap between the transcriptional responses to hydrogen sulfide and hypoxia.Venn diagram shows overlap between genes induced by exposure to 12 h H 2 S (n = 298) and all genes products that are altered by hypoxia (n = 654) [15].The probability of randomly observing an overlap of 8 genes between these datasets is 0.006.doi:10.1371/journal.pone.0025476.g002 RNAi by feeding was essentially as described [35].Bacteria expressing the dsRNA was diluted from an overight culture grown in LB containing 25 mM carbenicillin and 10 mM tetracycline, regrown to OD 600 ,0.6 in LB with 25 mM carbenicillin and then seeded onto NGM-lite plates that contained 3 mM isopropyl b-D-1-thiogalactopyranoside and 25 mM carbenicillin.RNAi plates were allowed to dry overnight, stored at 4 C, and used within 5 days of being seeded.skn-1(RNAi) adults laid only dead embryos.

RNA sample isolation
For microarray and qRT-PCR experiments, animals were synchronized as starved first-stage larvae (L1) after isolating embryos by hypochlorite treatment.For microarray analysis, 3000 starved L1 were distributed onto 15 cm NGM/OP50 plates, with each independent replicate performed on a different day.For quantitative RT-PCR (qRT-PCR), 1000 L1 larvae were distributed onto 10 cm NGM/OP50 plates.C. elegans were exposed to  shows the frequency that core skn-1 consensus sites (RTACT, [27]) are found within the upstream 2 kb flanking region of each transcript whose regulation in response to H 2 S was altered by skn-1(RNAi).{ genes reported to have SKN-1 bound in the promoter in the ModENCODE database [38].B. There is little similarity between response to H 2 S and other skn-1-dependent transcriptional responses.The overlap between the H 2 Sregulated genes after 12 h (n = 445) was greater than chance when compared with skn-1-dependent gene products in unstressed conditions (n = 233, 16 common transcripts, hypergeometric probability 0.006) and for genes that require skn-1 for arsenic-induced upregulation (n = 118, 10 common transcripts, hypergeometric probability 0.01) [27].There was not significant overlap between transcripts altered by exposure to H 2 S and skn-1 dependent transcripts that are downregulated in unstressed conditions (n = 63, hypergeometric probability 0.13), upregulated by tert-butyl hydroperoxide (n = 64, hypergeometric probability 0.06) or hyperoxia (n = 68, hypergeometric probability 0.15).C. skn-1 is required to survive exposure to H 2 S. Unc animals (skn-1/nT1 heterozygotes) were compared to non-Unc, skn-1 homozygotes for sensitivity to H 2 S (#animals alive/total after exposure to 50 ppm H 2 S).doi:10.1371/journal.pone.0025476.g003H 2 S on plates for the amount of time indicated immediately prior to harvest.All animals were harvested as first-day gravid adults (schematized in Figure 1A).For nematode harvest, plates were removed from H 2 S, the worms were immediately rinsed off the plates with distilled water, caught on a 43-micron nylon filter and collected by centrifugation.100 mL of sedimented worms were added to 900 mL Trizol (Invitrogen, Carlsbad, CA), frozen in liquid nitrogen and stored at 270 C. Less than 2 min elapsed from when plates were removed from H 2 S until samples were frozen.Frozen samples were thawed, vortexed for 30 s, and the RNA was isolated following the protocol included with the Trizol manual, followed by isopropanol precipitation.

Microarray expression profiling and analysis
Each RNA sample was labelled, hybridized to a single-channel Nimblegen 4672 K (build 160) expression array, and scanned following manufacturer's suggested protocols by the Fred Hutchinson Cancer Research Center's DNA Array Facility.Three biological replicates for each H 2 S-treated sample (1, 12 and 48 h exposure) and 5 biological replicates from untreated controls were used.Data were RMA normalized and probe-level data were summarized with the NimbleScan software.Genes with weak signal intensity across all groups and those with low variability across samples were excluded from further analysis.Each H 2 S-treated sample was statistically compared to a matched untreated control using the Bioconductor package limma [36].The false discovery rate (FDR) method of Benjamini and Hochberg [37] was used to adjust p-values for multiple testing.An adjusted p-value#0.05was used to define differential expression.Results were annotated using WormBase WS190 (www.wormbase.org).Expression results and microarray raw intensity files, in compliance with MIAME guidelines, can be accessed through the Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO series accession number GSE25199.Functional annotation clustering analysis was performed using the Database for Annotation, Visualization and Integrated Discovery (DAVID) v6.7 (http:// david.abcc.ncifcrf.gov/home.jsp).Gene list submitted for 12 h exposure to H 2 S included only the 91 gene products with logFC.1.6(fold change.3).For the 48 h dataset, the 95 gene products with logFC.2.5 (fold change.5.5) were included.In both cases, the C. elegans background list in the database was used with default analysis parameters.Using only gene products included in the analysis after filtering did not alter the results.Annotation clusters that included at least one term with p,0.05 were considered to be functionally enriched clusters.
Hypergeometric probabilities were calculated including all 11,522 features included in the microarray analysis as the population, with successes and sample size as defined in the text (http://stattrek.com/tables/hypergeometric.aspx).The number of successes in each sample (overlap) was determined by manually comparing data from H 2 S-induced changes measured by our microarray experiments and hypoxia-induced genes [15] or skn-1dependent transcripts [27,28].Probabilities less than 0.05 were considered significant.Core skn-1 consensus sites in the promoter region of candidate transcripts were defined manually, based on the published consensus RTCAT [27].The promoter region was defined as 2 kb upstream of the start site.The ModENCODE database [38] was searched to determine if any of the 7 transcripts changed in a skn-1-dependent changes in response to H 2 S were shown to have SKN-1::GFP bound in the promoter.

Real-time RT-PCR (qRT-PCR)
Quantitative RT-PCR (qRT-PCR) was used to validate microarray measurements and determine if H 2 S-dependent changes occurred in hif-1(ia04) or skn-1(RNAi) animals.mRNA was isolated as described above, and cDNA was synthesized from 300 ng of RNA using the included random primers using the ProtoScript M-MuLV First Strand cDNA Synthesis Kit (New England Biolabs) according to manufacturer's suggested protocol.Primers to amplify cDNA targets were designed using Primer3 (http://frodo.wi.mit.edu/primer3/).When possible, primer pairs spanned a small intron so that genomic and cDNA amplification products could be distinguished by agarose gel electrophoresis.Primer sequences are available upon request.Primers were tested to ensure amplification of the correct size genomic target, and then calibrated against serial dilutions of genomic DNA.qRT-PCR reactions were performed using an ep realplex2 S (Eppendorf).Each 10 mL reaction contained 5 mL 2X KAPA SYBR green Master Mix (Kapa Biosystems), 0.45 mL cDNA and 3 mL primers (10 mM each primer).Reactions were performed in duplicate and at least three independent biological replicates were included for each condition.Each experiment included primers that amplified only genomic DNA (negative controls to identify background signal levels) as well as 4 control targets (sir-2.1,tba-1, irs-2, and hil-1) that are not affected by H 2 S exposure for normalization.DC t for each gene product was calculated by subtracting C t values from the geometric mean of the control targets [39].DC t were averaged across the three experiments.Student's t-test was used to evaluate differences between DC t values of treated samples and untreated controls (EXCEL).For differences between genotypes (Figures 2 and 3), p-values were calculated by oneway Anova from summary statistics (mean, standard deviation, n) (www.statpages.org).Reported fold-changes were calculated as

Figure 1 .
Figure 1.Exposure to H 2 S induces rapid and progressive changes in mRNA abundance.A. Experimental design schematic.C. elegans were grown from synchronized first-stage larvae (L1) for 48 h to young adult before being collected for RNA extraction.Each bar represents 48 h from L1 to first-day adult for one experimental group.Time in room air is indicated in white and time in H 2 S indicated in red.Exposure to H 2 S (50 ppm in room air) was always immediately prior to isolating RNA.B. Changes in mRNA abundance measured by microarray.Plots show magnitude of change in transcript level (log 2 FC) as a function of adjusted p-value (log 10 p-value).Each point is data from one gene product.Significant changes (adj.p-value,0.05)are red.After 1 h exposure to H 2 S (left), 16 genes were significantly up-regulated and one was down-regulated (Table1).After 12 h exposure to H 2 S (middle), 445 transcripts were significantly changed, with 259 up-regulated (Tables1 and S1).After 48 h in H 2 S (right), 5089 transcripts were significantly altered relative to untreated controls (TableS2).doi:10.1371/journal.pone.0025476.g001

Figure 2 .
Figure 2. HIF-1 is required for early transcriptional responses to H 2 S. A. H 2 S-induced transcriptional changes require HIF-1.Changes in mRNA abundance after 1 h exposure to H 2 S were measured by qRT-PCR in wild-type (N2, open bars) and hif-1(ia04) mutant animals (filled bars).Three biological replicates for each group were performed, and each PCR reaction was run in duplicate.Error bars represent the standard deviation of the biological replicates, as propagated through the DDC t and fold-change calculations.*Difference between induction in wild-type (N2) is statistically different than in hif-1(ia04) mutant animals, p,0.05.Red dashed line demarks where transcript levels in H 2 S are the same as in room air.B. Transcriptional changes after 1 hexposure to H 2 S overlap slightly with hif-1-dependent changes in response to hypoxia. 3 of 16 transcripts upregulated in response to 1 h exposure to H 2 S were identified as hif-1-dependent targets in hypoxia (n = 68)[15].The probability of observing this overlap randomly is 0.001.C.There is minimal overlap between the transcriptional responses to hydrogen sulfide and hypoxia.Venn diagram shows overlap between genes induced by exposure to 12 h H 2 S (n = 298) and all genes products that are altered by hypoxia (n = 654)[15].The probability of randomly observing an overlap of 8 genes between these datasets is 0.006.doi:10.1371/journal.pone.0025476.g002

Figure 3 .
Figure 3. SKN-1 is essential for appropriate response to H 2 S. A. Some H 2 S-induced transcriptional changes require skn-1.Changes in mRNA abundance after 1 h exposure to H 2 S were measured by qRT-PCR in N2 animals grown on control RNAi food L4440 (open bars) or on skn-1(RNAi) (filled bars).Three biological replicates for each group were performed, and each PCR reaction was run in duplicate.Error bars represent the standard deviation of the biological replicates, propagated through the DDC t and fold-change calculations.*Difference between induction in control is significantly different than skn-1(RNAi) p,0.05.Tableshowsthe frequency that core skn-1 consensus sites (RTACT,[27]) are found within the upstream 2 kb flanking region of each transcript whose regulation in response to H 2 S was altered by skn-1(RNAi).{ genes reported to have SKN-1 bound in the promoter in the ModENCODE database[38].B. There is little similarity between response to H 2 S and other skn-1-dependent transcriptional responses.The overlap between the H 2 Sregulated genes after 12 h (n = 445) was greater than chance when compared with skn-1-dependent gene products in unstressed conditions (n = 233, 16 common transcripts, hypergeometric probability 0.006) and for genes that require skn-1 for arsenic-induced upregulation (n = 118, 10 common transcripts, hypergeometric probability 0.01)[27].There was not significant overlap between transcripts altered by exposure to H 2 S and skn-1 dependent transcripts that are downregulated in unstressed conditions (n = 63, hypergeometric probability 0.13), upregulated by tert-butyl hydroperoxide (n = 64, hypergeometric probability 0.06) or hyperoxia (n = 68, hypergeometric probability 0.15).C. skn-1 is required to survive exposure to H 2 S. Unc animals (skn-1/nT1 heterozygotes) were compared to non-Unc, skn-1 homozygotes for sensitivity to H 2 S (#animals alive/total after exposure to 50 ppm H 2 S).doi:10.1371/journal.pone.0025476.g003

Table 1 .
Changes in mRNA abundance associated with exposure to H 2 S.

Table 2 .
qRT-PCR Validation of changes in transcript abundance after 1 h H 2 S.Most increased transcripts in abundance after 12 h exposure to 50 ppm H 2 S a Fold-change of transcript as measured by microarray, as in Table 1.b Fold-change of transcript as measured by qRT-PCR.Animals were grown on E. coli OP50 strain or the HT115(DE3) strain containing the control RNAi plasmid L440.*p,0.05.doi:10.1371/journal.pone.0025476.t002

Table 3 .
Functional annotation of gene products increased after exposure to H 2 S.
[8].Consistent with this view, HIF-1 activity, as measured by an nhr-57::GFP transcriptional reporter, is in different tissues of animals exposed to hypoxia as compared to H 2 S[8].
overlap between transcriptional targets of HIF-1 in hypoxia and H 2 S.These results may indicate that H 2 S does not mediate protection against ischemia simply by inducing a standard hypoxia response.Further understanding this conserved adaptive response to H 2 S will provide new insight into mechanisms that can improve homeostasis in changing conditions.