Introduction

The detection and response to chemical signals are crucial to interact effectively with the environment in various organisms, spanning from bacteria to humans. During the late Proterozoic era, the rise in oxygen (O₂) levels eliminated hydrogen sulfide (H₂S) from most of the habitats. Yet, low O₂ levels together with high concentrations of H₂S and carbon dioxide (CO₂) can still persist in enclosed environments where bacteria actively decompose organic matter (Olson & Straub, 2016). H₂S can easily penetrate biological membranes and interfere with various cellular processes, leading to toxicity. One of its detrimental effects is the disruption of cellular respiration by binding to cytochrome c oxidase (COX) and inhibiting the mitochondrial electron transport chain (Cooper & Brown, 2008; Khan et al., 1990; Nicholls & Kim, 1982b).

Animals inhabiting environments rich in H₂S have evolved mechanisms to either avoid or adapt to high concentrations of H₂S, providing them with survival advantages. Species that encounter periodic increases in H₂S often possess sensing abilities and exhibit avoidance behaviors. For example, benthic species undertake daily vertical migrations between sulfidic and non-sulfidic waters (Abel, Koenig, & Davis, 1987; Salvanes, Utne-Palm, Currie, & Braithwaite, 2011). Certain metazoan species have seized the ecological opportunities presented by H₂S-rich environments through the development of cellular adaptations, allowing them to occupy a unique ecological niche that is inaccessible to the other organisms. Livebearing fish species, for instance, have evolved COX subunits that are more resistant to H₂S and are more efficient in H₂S detoxification (Kelley et al., 2016; Pfenninger et al., 2014).

Free living nematodes thrive in decaying matter where they are exposed to significant fluctuations in O₂, CO₂, and H₂S levels over short distances (Rodriguez-Kabana, Jordan, & Hollis, 1965). Nematodes like C. elegans have evolved the ability to utilize these gasous stimuli to navigate their environment, locate bacterial food sources, or escape the dangers (Bretscher, Busch, & de Bono, 2008; Carrillo, Guillermin, Rengarajan, Okubo, & Hallem, 2013). Acute exposure to H₂S triggers transiently increased locomotion in C. elegans (Budde & Roth, 2011). However, prolonged exposure to H₂S induces cellular adaptation to survive the toxic environment. C. elegans can tolerate H₂S levels up to 50 ppm, but the concentrations exceeding 150 ppm prove lethal (Budde & Roth, 2010, 2011; Fawcett, Hoyt, Johnson, & Miller, 2015; Horsman, Heinis, & Miller, 2019; Miller, Budde, & Roth, 2011a; Miller & Roth, 2007). Extended periods of low O₂ and/or high H₂S exposure stabilize the hypoxia-inducible factor HIF-1, leading to transcriptional reprogramming (Budde & Roth, 2010). Through HIF-1-dependent and independent mechanisms, acclimation of C. elegans to 50 ppm H₂S extends lifespan and improves the response to hypoxia-induced protein aggregation (Budde & Roth, 2010; Fawcett et al., 2015; Horsman et al., 2019; Miller & Roth, 2007; Qabazard et al., 2014).

In this study, we investigated how the nematode C. elegans responds acutely to avoid an increase of H₂S levels, as well as how it adapts at the molecular and physiological levels if they fail to escape the H₂S exposure.

Results

cGMP signaling in ASJ cilia contributes to H₂S avoidance

When encountering aversive cues such as noxious gases, C. elegans initiates a pirouette followed by an increase of its locomotory activity to escape the noxious stimuli. When acutely exposed to 150 ppm H₂S, C. elegans exhibited an increased locomotory speed and enhanced turning behavior, while decreasing the frequency of reversals, as a strategy to escape the noxious stimuli (Figure 1—figure supplement 1A–C). The maximum speed was reached after 6–8 minutes in H₂S, and returned to the baseline over a period of approximately 1 hour (Figure 1—figure supplement 1D). The reduced locomotory activity after 2 hours in H₂S was fully reversible if these animals were immediately challenged by a different stimulus such as hypoxia (Figure 1—figure supplement 1E). The magnitude of the avoidance responses appeared to be concentration dependent. Under our assay conditions, acute behavioral response to 50 ppm H₂S was minimal and lacked consistency, whereas the response to 150 ppm H₂S was robust and reproducible (Figure 1—figure supplement 1F).

To gain molecular insight into H₂S evoked locomotory response, we performed a candidate screen for mutants that failed to increase their locomotory speed when H₂S levels acutely increased to 150 ppm (Supplementary file 1). We discovered that mutants defective in ciliogenesis exhibited diminished H₂S responses (Figure 1A–C). These included daf-19 mutants lacking cilia, dyf-3 mutants with reduced cilia length, and dyf-7 mutants where ciliogenesis occurs but the cilia are not anchored to their sensilla and are not exposed to the external environment (Heiman & Shaham, 2009; Murayama, Toh, Ohshima, & Koga, 2005; Starich et al., 1995; Swoboda, Adler, & Thomas, 2000). Our observations suggest that cilia exposure to the external environment is necessary for H₂S detection. Cilia-mediated sensory responses often involve the activation of guanylate cyclases and the opening of cGMP-gated channels (Ferkey, Sengupta, & L’Etoile, 2021). We observed attenuated H₂S responses in tax-4 or tax-2 mutants, lacking the key subunits of cilia-enriched cGMP-gated channels (Figure 1D and E). Comparable observations have also been reported in acute response to CO₂, suggesting that locomotory responses to CO₂ and H₂S are similar (Bretscher et al., 2008; Hallem et al., 2011; Hallem & Sternberg, 2008). Consistent with this, mutations in genes encoding the calcineurin subunits TAX-6 or CNB-1, or nuclear hormone receptor NHR-49, impaired both H₂S and CO₂ responses (Hallem & Sternberg, 2008) (Figure 1—figure supplement 2A–C). Additionally, locomotory response to H₂S was also repressed following a 24-hour period of starvation, as observed for CO₂ (Figure 1—figure supplement 2D) (Bretscher et al., 2008; Hallem & Sternberg, 2008). In C. elegans, the regulation of starvation-responses involves the reduced activity of two major pathways: insulin and TGF-β signaling. Disrupting the genes encoding the insulin receptor DAF-2 or the TGF-β ligand DAF-7 eliminated locomotory activity in response to acute H₂S (Figure 1F and G, Figure 1—figure supplement 2E). Insulin and TGF-β signaling involve the inhibition of downstream transcription factors DAF-16 or DAF-3 respectively. H₂S avoidance defects of daf-2 and daf-7 mutants were fully suppressed in daf-2; daf-16 or in daf-7; daf-3 double mutants (Figure 1F and G).

Figure 1.

These observations prompted us to further investigate if the same molecular and circuit mechanisms underlay the acute avoidance to CO₂ and H₂S. The response to CO₂ requires receptor guanylate cyclase GCY-9 (Hallem et al., 2011). Interestingly, GCY-9 was dispensable for the locomotory responses to H₂S (Figure 1H). In addition, animals responded more rapidly to CO₂ than to H₂S, but with a reduced magnitude (Figure 1I), implying that distinct mechanisms likely drive these two responses. In a comprehensive survey of all guanylate cyclase mutants, the daf-11 strain emerged as the only mutant showing defects in the avoidance of H₂S (Figure 1D, Supplementary file 1). Selective expression of daf-11 in ASJ neurons, but not in other neurons, partially restored the locomotory response to acute H₂S (Figures 1J, Figure 1—figure supplement 2F–I). Consistently, the H₂S response defect of tax-4 mutants was also rescued by selective expression of its cDNA in ASJ neurons (Figure 1K), suggesting that the receptor guanylate cyclase DAF-11 and the CNG channel TAX-4 partially act in ASJ neurons to promote acute response to H₂S. This differed from the site of action of the CO₂ sensor GCY-9, which primarily operates in BAG neurons (Hallem et al., 2011). Taken together, these observations suggest that common signaling pathways modulate the locomotory responses to both H₂S and CO₂, although the molecular and circuit underpinning of H₂S and CO₂ sensing are different.

Activation of the O₂ sensing circuit antagonizes H₂S avoidance

The sensory responses of wild C. elegans are adapted to their living environment, which is characterized by low O₂ levels and high concentrations of CO₂. In comparison to the domesticated lab strain N2, wild isolates of C. elegans exhibit a strong response to high O₂ but are relatively indifferent to CO₂ (Beets et al., 2020; Bretscher et al., 2008; Hallem & Sternberg, 2008). Local H₂S concentrations could also be significantly higher in decomposing substances where wild C. elegans thrives. We suspected that wild strains could display a modified response to H₂S in comparison to standard laboratory N2 animals. Indeed, wild isolates CB4855, CB4856, and CB4858 exhibited attenuated responses to H₂S (Figure 2A, Figure 2—figure supplement 1A). One predominant genetic trait in these strains is a naturally occurring variation of neuropeptide receptor NPR-1 215F, which is less active than NPR-1 215V in N2 animals (de Bono & Bargmann, 1998). This led us to investigate whether the NPR-1 signaling plays a role in modulating acute avoidance of H₂S. Mutating npr-1 in N2 strain background was sufficient to abolish locomotory responses to H₂S, as observed in the wild isolates (Figure 2B). NPR-1 primarily acts within the RMG interneurons to modulate responses to environmental stimuli by reducing signal output from these neurons (Laurent et al., 2015; Macosko et al., 2009). The selective expression of the highly active npr-1(215V) variant in RMG neurons effectively restored H₂S evoked locomotory response in npr-1 mutant worms (Figure 2C). Furthermore, optogenetic stimulation of RMG interneurons using channelrhodopsin-2 (ChR2) significantly dampened locomotory response to H₂S of N2 animals, suggesting high RMG activity inhibits H₂S responses (Figure 2—figure supplement 1B).

Figure 2.

The sensation of 21% O₂ involves the soluble guanylate cyclases GCY-35/GCY-36 and the downstream cyclic nucleotide gated (CNG) channels TAX-4/TAX-2 (Busch et al., 2012; Cheung, Cohen, Rogers, Albayram, & de Bono, 2005; Couto, Oda, Nikolaev, Soltesz, & de Bono, 2013; Gray et al., 2004; Laurent et al., 2015; Persson et al., 2009; Zimmer et al., 2009). We examined if disrupting O₂ sensing machinery had the same effect as the inhibition of RMG signaling by NPR-1(215V). We found that silencing O₂ sensing neurons by mutating gcy-35 and tax-4, or ablating these neurons by cell-specific expression of the pro-apoptotic gene egl-1 restored H₂S evoked behavioral response in npr-1 null mutants (Figure 2D–F). Specific expression of gcy-35 cDNA in URX, AQR, PQR was sufficient to repress acute response to H₂S in npr-1; gcy-35 double mutants (Figure 2G). Finally, enhancing the presynaptic activities of URX, AQR, PQR neurons by the expression of a gain-of-function PKC-1(A160E) (Dekker, Mcintyre, & Parker, 1993; Hiroki et al., 2022) significantly dampened H₂S regulated locomotory activity (Figure 2—figure supplement 1C). Taken together, these observations indicate that activation of O₂ sensing circuit is inhibitory to H₂S evoked locomotory response.

H₂S exposure reprograms gene expression in C. elegans

Prolonged exposure to H₂S has a dramatic impact on the physiology of C. elegans: 50 ppm H₂S exposure extends its lifespan, while 150 ppm H₂S is lethal (Budde & Roth, 2010, 2011; Fawcett et al., 2015; Horsman et al., 2019; Miller & Roth, 2007 Miller, 2011 #5). We conducted a comparative analysis of the transcriptome profiles in animals exposed to either 50 ppm or 150 ppm H₂S for the durations of 1 hour, 2 hours, or 12 hours. Our RNA-seq analysis revealed that a one-hour exposure to either 50 ppm or 150 ppm H₂S was sufficient to trigger significant changes in gene expression, with 518 or 304 genes showing differential expression, respectively (Figure 3A, Supplementary file 2, p<1e-10). Genes induced by 1-hour and 2-hour H₂S exposure displayed a considerable overlap (Figure 3B, Figure 3— figure supplement 1A). Gene Ontology (GO) analysis revealed that biological processes such as defense against bacteria and cysteine biosynthesis from L-serine were immediately activated upon H₂S exposure (Figure 3C and D, Figure 3—figure supplement 1B and C). As expected, we observed a robust induction of genes involved in H₂S detoxification (Figure 3E and F) (Miller, Budde, & Roth, 2011b; Niu et al., 2011; Vora et al., 2022). H₂S is detoxified in mitochondria by sulfide:quinone oxidoreductase SQRD-1 to persulfide, which is then metabolized by sulfur dioxygenase ETHE-1 to generate sulfite. Thiosulfate transferase further oxidizes sulfite to produce thiosulfate (Hildebrandt & Grieshaber, 2008). Glutathione S-transferases (GSTs) remove accumulated sulfur molecules during H₂S oxidation (Jackson, Melideo, & Jorns, 2012). Notably, gst-19 and sqrd-1 were among the most significantly upregulated genes after one or two hours of exposure to either 50 ppm or 150 ppm H₂S, while ethe-1 showed a weak increase (Figure 3E and F, Figure 3—figure supplement 1D and E, Supplementary file 2). H₂S activates the HIF-1 pathway through the interaction of sulhydrylase/cysteine synthase CYSL-1 with the HIF-1 proline hydroxylase EGL-9 (Ma, Vozdek, Bhatla, & Horvitz, 2012). We found that a proportion of genes with HIF-1 regulated promoters displayed increased expression after 1 or 2 hours of H₂S exposure (Figure 3— figure supplement 1F) (Miller et al., 2011b; Vora et al., 2022). These genes included the above-mentioned detoxifying enzymes GST-19, SQRD-1, and ETHE-1, as well as enzymes involved in cysteine metabolism, CYSL-1, CYSL-2, and CDO-1 (Supplementary file 2). Mutations increasing SKN-1 activity bypass the requirement for HIF-1 to survive in high H₂S (Horsman et al., 2019). Consistent with early observations (Miller et al., 2011a), we also observed that a set of SKN-1 targets were differentially expressed in H₂S (Figure 3—figure supplement 1G) (Niu et al., 2011). This included the heat shock protein encoding genes hsp-16.2 and hsp-16.41 (Figure 3E and F, Figure 3—figure supplement 1D and E).

Figure 3.

Surprisingly, we also observed a rapid increase in the expression of genes associated with intracellular H₂S production, such as semo-1 and mpst-3 genes (Figure 3E and F, Figure 3—figure supplement 1D and E) (Philipp et al., 2022; Qabazard et al., 2014). However, these data aligned with earlier observations made in sulfide spring fish (Kelley et al., 2016; Mathew, Schlipalius, & Ebert, 2011; Qabazard et al., 2014).

When exposed to the H₂S for 12 hours, the differentially expressed genes showed reduced overlap with those identified at earlier time points (Figure 3B, Figure 3—figure supplement 1A). This suggests that different defense strategies against H₂S might be employed at various stages of exposure. Although similar sets of genes were induced after 1h or 2h at 50 ppm and 150 ppm H₂S, the overlap between 50ppm and 150ppm decreased after 12 hours exposure (Figure 3—figure supplement 1H–J). This observation suggests that defense mechanisms might also differ for prolonged exposure at low and high levels of H₂S. In particular, we noted that 150 ppm H₂S specifically triggered the expression of the HIF-1 regulator rhy-1 and the HIF-1 target nhr-57 at all exposure time points, which was less pronounced in response to 50 ppm H₂S (Figures 3E and F, Figure 3—figure supplement 1D, E, K and L).

Acute response to H₂S is modulated by HIF-1 signaling

Consistent with early observations (Budde & Roth, 2010), our RNA-seq data suggested that H₂S exposure activates HIF-1 pathway (Figure 3—figure supplement 1F). Stabilization of HIF-1 by H₂S is critical for animals to survive H₂S exposure by the induction of detoxification genes as well as other adaptative responses (Ma et al., 2012). Such an adaptation is usually accompanied with the reduction in sensory sensitivity to this stressor. Therefore, we suspected that HIF-1 stabilization might dampen the acute response to H₂S. Under normoxia conditions, the conserved proline-4-hydroxylase PHD/EGL-9 hydroxylates HIF-1, which is subsequently recognized by the von Hippel-Lindau (VHL) tumor suppressor protein for degradation (Kaelin & Ratcliffe, 2008; Semenza, 2010). We found that disrupting hif-1 resulted in extremely transient behavioral response to H₂S characterized by an initial bout of activity and a rapid decline of locomotory speed to the basal level (Figure 4A). While eliminating PHD/EGL-9 or VHL activity, which stabilizes HIF-1, the locomotory response to H₂S was completely inhibited (Figure 4B). egl-9; hif-1 or vhl-1; hif-1 double mutants had the identical response as hif-1 mutants (Figure 4C and D), confirming that mutating egl-9 or vhl-1 inhibits acute response to H₂S by activating HIF-1. Acclimating animals in 1% O₂ for more than 12 hours also significantly decreased acute response to H₂S (Figure 4E). In addition, animals carrying the non-degradable form of HIF-1 in all neurons were defective in H₂S evoked locomotory responses (Figure 4F). Taken together, these observations suggest that constitutive activation of HIF-1 promotes the adaption to H₂S and inhibits the behavioral responses to H₂S stimulation.

Figure 4.

The transient behavioral response to H₂S in hif-1 mutants suggested that animals deficient of HIF-1 can still detect the presence of H₂S. This observation together with the importance of HIF-1 signaling in H₂S detoxification provoked us to explore whether detoxification enzymes were also required for the sustained locomotory activity in H₂S. Similar to hif-1 mutants, both sqrd-1 and ethe-1 mutants displayed a transient response to H₂S (Figure 4G and H). We next examined the contribution of other upregulated genes in response to H₂S, including the genes encoding sulfhydrylase/cysteine synthases CYSL-1, CYSL-2, and CYSL-3 as well as SEMO-1. Disrupting CYSLs and SEMO-1 encoding genes phenocopied hif-1 or sqrd-1 mutants (Figure 4I and J). Our data suggest that while H₂S detoxification system is not essential for H₂S sensing, it is required for preserving animals’ locomotory activity in the presence of H₂S.

Labile iron pool is critical to sustain the locomotory activity in H₂S

C. elegans tightly regulates iron homeostasis through the control of iron transporter SMF-3, as well as the regulation of iron storage protein Ferritins FTN-1 and FTN-2 (Cha’on et al., 2007; Gourley, Parker, Jones, Zumbrennen, & Leibold, 2003; Kim, Cho, Yoo, & Ahnn, 2004; Pekec et al., 2022). Among the most differentially regulated genes, ftn-1 appeared to be consistently downregulated by H₂S (Figure 5A and B, Figure 5—figure supplement 1A–D). The inhibited expression of ftn-1 and ftn-2 as well as the elevated expression of smf-3 (Figure 5A and B, Figure 5—figure supplement 1A–D) prompted us to explore if labile iron levels play a role in behavioral response to H₂S exposure. The labile iron pool is chelatable and redox-active. In the presence of iron chelator 2,2’-dipyridyl (BP), H₂S evoked locomotory response was fully inhibited (Figure 5C). This was not caused by the general sickness of BP-exposed animals as they responded robustly to acute hypoxia (Figure 5—figure supplement 1E). By contrast, the supplementation of ferric ammonium citrate (FAC) significantly prolonged animals’ locomotory activity in H₂S (Figure 5C). In addition, disrupting ftn-1, which was expected to increase labile iron levels (Anderson & Leibold, 2014; Romney, Newman, Thacker, & Leibold, 2011), promoted the locomotory activity of animals in H₂S (Figure 5D), Conversely, mutating smf-3, which impaired iron uptake (Anderson & Leibold, 2014; Romney et al., 2011), led to a faster decline of locomotory activity in H₂S (Figure 5D). Furthermore, depleting labile iron pool by ftn-1 overexpression resulted in an attenuated response to H₂S (Figure 5E), while it did not affect acute response to hypoxia (Figure 5—figure supplement 1F). These data suggests that high levels of free iron sustain locomotory activity in H₂S, whereas low iron levels inhibit behavioral responses to H₂S. It has been reported that hypoxia inhibits the expression of ftn-1 and ftn-2, while upregulating smf-3 expression (Ackerman & Gems, 2012; Gourley et al., 2003; Romney et al., 2011). Additionally, hif-1 mutants have been reported to display reduced intracellular iron content and increased susceptibility to iron starvation (Rajan et al., 2019; Romney et al., 2011). We explored if the locomotory defect of hif-1 mutants in H₂S could be explained by iron deficiency (Rajan et al., 2019; Romney et al., 2011). The presence of FAC significantly improved H₂S evoked locomotory activity of hif-1 mutants (Figure 5F). In addition, disrupting ftn-1 partially suppressed the locomotory defect of hif-1 mutants in H₂S (Figure 5G). These data confirm that labile iron pool is critical to maintain the locomotory activity of animals in H₂S.

Figure 5.

Iron deficiency causes mitochondrial dysfunction including decreased cytochrome concentration and respiratory efficiency (Dallman, 1986; Masini, Salvioli, et al., 1994; Masini, Trenti, et al., 1994; Walter et al., 2002). H₂S is known to exert both positive and negative effects on the mitochondria, which are not only the primary targets of H₂S but also represent one of the main cellular hubs of H₂S detoxification (Cooper & Brown, 2008; Khan et al., 1990; Nicholls & Kim, 1982a). We probed the possibility that impairing mitochondrial function might alter locomotory response to H₂S. To this end, we tested mutants with the disruption of gas-1, clk-1, mev-1, or isp-1 genes. Similar to iron depletion, these mutants abrogated locomotory response to H₂S (Figure 6A–D). Additionally, exposing animals to the mitochondrial complex I inhibitor rotenone fully repressed locomotory response to H₂S (Figure 6E). While rotenone exposure inhibits H₂S response, it evoked increased locomotory activity over 2 hours (Figure 6E). These observations confirm the critical role of mitochondria in supporting locomotory response in exposure to H₂S in C. elegans.

Figure 6.

Discussion

Animals’ behavior and physiology are profoundly influenced by the surrounding environment, in which they reside. Wild C. elegans isolates thrive in the decomposing matters, where the local concentrations of O₂ are low while the levels of CO₂ and H₂S could be high. These animals have adapted their behavior in such an environment, displaying increased sensitivity to high O₂ exposure but dampened responses to CO₂. In contrast, the standard laboratory strain N2, which is constantly exposed to high O₂ but low CO₂, is relatively insensitive to elevated levels of O₂ but robustly avoids CO₂ exposure (Beets et al., 2020; Bretscher et al., 2008; Carrillo et al., 2013; Hallem & Sternberg, 2008). As observed in response to CO₂, N2 animals exhibit strong reorientation and acceleration to escape the H₂S exposure, while wild isolates display a significantly diminished response to H₂S.

The responses to CO₂ and H₂S in N2 animals are modulated by common signaling pathways, which include the ciliogenesis, the cGMP-gated channels TAX-2 and TAX-4, calcineurin subunit TAX-6/CNB-1, and nuclear hormone receptor NHR-49. Additionally, these two responses are affected by the nutritional state of the animals. In C. elegans, starvation is signaled by reduced activity in two major signaling pathways: the insulin and TGF-β pathways. H₂S and CO₂ responses are abolished in starved animals or mutants that disrupt insulin or TGF-β signaling. Therefore, the modulation of behavioral responses to H₂S and CO₂ by external and internal factors appears highly similar. However, CO₂ response is mediated by the guanylate cyclase GCY-9 in BAG neurons (Hallem et al., 2011), which appears to be dispensable for H₂S sensing. We observed a prominent role for the guanylate cyclase DAF-11 expressed in the chemosensory neurons ASJ for H₂S responses. daf-11 signaling in ASJ also mediates chemosensory responses to volatile chemicals and to NO (Birnby et al., 2000; Hao et al., 2018). However, DAF-11 controls the normal larval development of C. elegans and may have pleiotropic indirect effects on sensory functions (Murakami, Koga, & Ohshima, 2001). In absence of more direct evidence, its role as H₂S sensor in ASJ remain uncertain.

Long-term exposure to H₂S can elicit adaptive responses including changes in gene expression patterns to mitigate the toxic effects of H₂S. Within just one hour of H₂S exposure, we observed the upregulation of genes associated with H₂S detoxification. This response enables animals to detoxify H₂S and maintain cellular homeostasis under sulfidic conditions. This adaptive strategy is accompanied by the reduced locomotory response to H₂S exposure. The HIF-1 (hypoxia-inducible factor) signaling is a major pathway involved in the adaption during prolonged H₂S exposure (Budde & Roth, 2011; Miller et al., 2011b). When the HIF-1 pathway is constitutively induced in the nervous system, the behavioral responses to H₂S are eliminated. Similarly, reduced behavioral responses to H₂S are observed after 12 hours of exposure to 1% O₂. It has been proposed that persulfides, polysulfides, and thiosulfate, which are the by-products during sulfide detoxification, play a role in H₂S signaling (Mishanina, Libiad, & Banerjee, 2015). However, in the case of locomotion responses to H₂S in C. elegans, this is unlikely to be the mechanism for the initial response, as mutants lacking detoxification enzymes still exhibit an acute and transient response to H₂S. Interestingly, the sustained locomotory activity was abolished in these mutants, suggesting that either the by-products of H₂S detoxification contribute to maintain the behavioral response or the lack of detoxification enzyme results in rapid cellular toxicity, preventing sustained locomotory activity in H₂S (Budde & Roth, 2011).

Interestingly, we provide evidence that labile iron pool is critical for H₂S responses. Removal of labile iron pool by the chelator BP or ftn-1 overexpression attenuated both acute and sustained responses to H₂S, suggesting that intracellular iron is required for the sensation of H₂S. While increasing labile iron levels sustains the locomotory activity in H₂S. In C. elegans, labile iron homeostasis is predominantly regulated by the HIF-1 signaling, the activation of which decreases the transcription of ftn-1 and increases the expression of smf-3 (Ackerman & Gems, 2012; Romney et al., 2011). Consistent with this, we have observed a downregulation of the ferritins FTN-1 and FTN-2, as well as an upregulation of the iron transporter SMF-3 during exposure to H₂S. This suggests that H₂S may control the expression of FTN-1 and SMF-3, and consequently the labile iron pool, through the HIF-1 pathway.

Although the exact mechanisms are not yet fully understood, several possibilities shed light on the importance of labile iron for H₂S responses. Fe²⁺ acts as a critical cofactor for the PHD/EGL-9 enzymes, which are involved in the hydroxylation-dependent destabilization of HIF-1 (Myllyharju & Kivirikko, 1997). PHD inactivation by iron sequestration by ferritin can increase HIF-1 signaling, suggesting a potential role of labile iron as a negative feedback loop to limit HIF-1 activity (Siegert et al., 2015). In addition, iron is necessary for the biosynthesis of iron-sulfur (Fe-S) clusters, critical cofactors involved in redox reactions and mitochondrial respiration (Read, Bentley, Archer, & Dunham-Snary, 2021; Xu & Moller, 2011). Labile iron is beneficial for the biogenesis of Fe-S clusters despite its oxidative properties under normal O₂ conditions (Liochev, 1996). Interestingly, hypoxia has been observed to enhance Fe-S cluster biogenesis by increasing the availability of bioactive iron (Ast et al., 2019). Therefore, low oxidative stress may alleviate the need to maintain a low labile iron pool. In this context, the antioxidant properties of H₂S or the interruption of mitochondrial respiration by H₂S might provide condition favorable to high labile iron pool (Shefa, Kim, Jeong, & Jung, 2018; Sunda, Kieber, Kiene, & Huntsman, 2002).

The primary mechanism underlying the toxic effects of H₂S has been suggested to be the direct inhibition of cytochrome c oxidase, a critical enzyme involved in mitochondrial respiration (Cooper & Brown, 2008; Khan et al., 1990; Nicholls & Kim, 1982b). Intriguingly, we have observed that genetic and pharmacological inhibition of mitochondrial respiration prevents the locomotory response of C. elegans to H₂S. This effect may be mediated by the induction of the HIF-1 pathway, which has been observed in mutants with impaired mitochondrial respiration (Lee, Hwang, & Kenyon, 2010). Furthermore, acute inhibition of mitochondrial respiration by rotenone also enhances locomotion while suppressing the response to H₂S. Reducing mitochondrial respiration has been found to stimulate aversive behavior in C. elegans (Melo & Ruvkun, 2012). Therefore, the acute inhibition of mitochondrial respiration by H₂S could transiently promote aversive behavior in C. elegans, while long-term inhibition of mitochondrial respiration may contribute to the induction of the HIF-1 pathway and adaptation to H₂S.

In summary, this study unveils a novel behavior of C. elegans in their avoidance to H₂S. The response to H₂S is intricately shaped by their environment, where it integrates with other sensory signals, including food availability, CO₂ and O₂ levels. We emphasize the vital role of H₂S detoxification system, particularly the involvement of mitochondria and labile irons, in sustaining the locomotory activity in H₂S. In the future, it will be exciting to elucidate the neuronal mechanisms driving H₂S avoidance and to explore how the changes of labile iron pools modulate cellular and organismal behavior responses to noxious sensory stimuli.

Materials and Methods

Strains

C. elegans were maintained using standard protocols (Brenner, 1974). Strains used in this study are listed in Supplementary file 1 and Supplementary file 3.

Preparation of H₂S gas mixture

Different H₂S concentrations were created as previously described (Miller & Roth, 2007). Briefly, the H₂S-containing gas mixture were prepared by diluting 5000 ppm H₂S in nitrogen (N₂) with 7% O₂ balanced with N₂. The gas flow was tuned by Sierra Smart-Trak 100 mass flow controllers. H₂S concentration in the mixture was measured using two H₂S detectors (MSA ALTAIR 2X gas detector for H₂S and Clip Single Gas Detector, SDG, CROWCON). The pre-defined gas mixtures of 7% O₂, 1% O₂ and 5% CO₂ with N₂ were purchased from Air Liquide Gas AB. 5000 ppm H₂S stock in N₂ was obtained from Linde Gas AB. Gas mixtures in all experiments were hydrated before use.

Molecular Biology

The Multisite Gateway system (Thermo Fisher Scientific, United States) was used to generate expression vectors. Promoters, including gcy-37 (2.7kb), trx-1 (1 kb), ftn-1(2 kb), daf-11(3 kb), odr-3(2.7 kb), gpa-11(3 kb), sra-6(3 kb), odr-1(2.4 kb), flp-21(4.1kb), and ocr-2(2.4 kb) were amplified from N2 genomic DNA and cloned into pDONR P4P1 using BP clonase. gcy-35, tax-4 and pkc-1 cDNAs were amplified using the first strand cDNA library as the template, while daf-11 and ftn-1 genomic sequences were amplified using genomic DNA as the template. The entry clones of these genes pDONR 221 were generated using BP reaction. To generate the gain of function mutation of pkc-1 (A160E), Q5 Site-Directed Mutagenesis Kit (NEB) was used according to manufacturer instructions. All expression vectors were generated using LR reaction. The primer sequences were displayed in Supplementary file 4.

To generate transgenic animals, the daf-11 expression vectors were injected at the concentration of 20 ng/μl supplemented with 50 ng/μl of a coelomocyte co-injection marker (unc-122p::GFP) and 30 ng/μl of 1kb DNA ladder. For tax-4 gene, the injection mixtures were prepared using 40 ng/μl of tax-4 expression vectors, 50 ng/μl of a coelomocyte co-injection marker and 10 ng/μl of 1kb DNA ladder. The rest of expression constructs were injected at 50 ng/μl together with 50 ng/μl of a coelomocyte marker.

CRISPR/Cas9 genome editing

Genes were disrupted using CRISPR/Cas-9 mediated genome editing as described (Dokshin, Ghanta, Piscopo, & Mello, 2018; Ghanta & Mello, 2020). The strategy involved the utilization of homology-directed insertion of custom designed single strand DNA template (ssODN), which had two homology arms of 35bp flanking the targeted PAM site. Between two homology arms, a short sequence containing a unique restriction enzyme cutting site and in frame as well as out of frame stop codons was included. The insertion of ssODN template would delete 16 bases of coding sequence, ensuring the proper gene disruption. To prepare the injection cocktail, 0.5 μl of Cas-9 protein (IDT) was mixed with 5 μl of 0.4μg/μl tracrRNA (IDT, United States) and 2.8 μl of 0.4 μg/μl crRNA (IDT). The mixture was incubated at 37°C for at least 10 minutes before 2.2 μl of 1μg/μl ssODN (or 500 ng dsDNA) and 2 μl of 0.6 μg/μl rol-6 co-injection marker were added. Nuclease-free water was used to bring the final volume to 20 μl. The injection mixture was centrifuged for 2 minutes before use.

Behavioral assays

H₂S evoked locomotion activity was monitored as described previously (Laurent et al., 2015; Zhao et al., 2022). Briefly, OP50 bacteria were seeded on the assay plates 16 hours before use. The border of bacterial lawn was removed using a PDMS stamp. For each assay, 25 to 30 day-one adult animals were picked onto assay plates, allowed to settle down for 15 minutes, and subsequently sealed within microfluidic chambers. A syringe pump (PHD 2000, Harvard Apparatus) was employed to deliver gas mixtures into the microfluidic chamber at a constant flow of 3 ml/min. The rapid switch between different gas mixtures were controlled by Telfon valves coupled with ValveBank Perfusion Controller (AutoMate Scientific). The locomotory activity at different gas mixtures were monitored using a high-resolution camera (FLIR) mounted on a Zeiss Stemi 508 microscope. Videos were captured at a rate of 2 frames per second, starting with a 2-minute recording in 7% O₂, followed by 11 minutes in H₂S, and ending with a final 5-minute period in 7% O₂. For the long-term recording in H₂S, video capture commenced with a 2-minute period in 7% O₂, followed by a duration of 148 minutes in 150 ppm H₂S, and concluded with a final 2-minute interval in 7% O₂.

To assess the effects of rotenone on H₂S evoked locomotory response, day-one adult animals were subjected to 10 μM rotenone for 10min, 30min, 1 hour and 2 hours on the drug containing plates. Subsequently, the rotenone-treated animals were assayed on the standard assay plates without the drug. To explore the impact of ferric ammonium citrate (FAC) (F5879, Sigma-Aldrich) and 2,2′-Bipyridyl (BP) (D216305, Sigma-Aldrich) on behavioral response to H₂S, L4 animals were exposed to 100 μM FAC or 5mg/ml BP for 16 hours in the presence of bacterial food. The FAC or BP treated animals were assayed on FAC or BP containing plates.

Optogenetic experiments were performed as previously outlined (Zhao et al., 2022). Transgenic L4 animals were exposed to 100μM all-trans retinal (ATR) (R2500, Sigma) in the dark for 16 hours. The ATR-fed animals were assayed with the illumination of 70 μW/mm² blue light during the whole assay period. Blue light was emitted from an ultra-high-power LED lamp (UHP-MIC-LED-460, Prizmatix), and the light intensity was determined using a PM50 Optical Power Meter (ThorLabs). To minimize the effect of transmitted light on the transgenic animals, a long-pass optical filter was used to eliminate the lights with short wavelengths during picking. All videos of behavioral analysis were analyzed using a home-made MATLAB program Zentracker (https://github.com/wormtracker/zentracker). For all behavior analyses, at least three assays were performed for each strain with more than 80 worms in total.

RNA extraction and sequencing

To obtain RNA samples for RNA-seq, thirty young adult animals were picked to each fresh plate and allowed to lay eggs for 2 hours, after which the adult animals were removed and eggs were allowed to develop into day one adults. Day one animals were subsequently challenged with H₂S for three different time points (1 hour, 2 hours and 12 hours). For each time point, five plates of day one animals were exposed to 50 ppm H₂S, 150 ppm H₂S or 7% O₂ as a control group. Subsequently, the animals were immediately collected and rinsed three times with M9 buffer. After washing, the worm pellet was frozen in liquid nitrogen. Animals were homogenized using Bullet Blender (Next Advance) in the presence of Qiazol Lysis Reagent and 0.5mm Zirconia beads at 4°C. RNA was prepared using RNeasy Plus Universal Mini Kit (Qiagen). Six independent RNA samples were prepared for all different treatments. The 2100 Bioanalyzer instrument (Agilent) was used for the RNA quality control with a microfluidic chip specific for RNA (Agilent RNA 6000 Nano Kit). Then 1 μg of qualified RNA samples in RNase-free ddH2O was used for library preparation. Library was constructed and sequenced by Novogene.

RNA-seq analysis

The RNA-seq data was aligned using STAR v2.7.9a (Dobin et al., 2013) and gene expression counts extracted by featureCounts v2.0.3 (Liao, Smyth, & Shi, 2014). Differential expression was calculated using DESeq2 (Love, Huber, & Anders, 2014). GO analysis was performed using the EnrichR R package (Kuleshov et al., 2016) on genes having p.adj < 1e-10. For brevity, several highly similar GO categories were manually omitted (full EnrichR output available at Github repository).

Data availability

Raw sequencing data has been deposited at ArrayExpress # E-MTAB-13296. The R code is available at https://github.com/henriksson-lab/ce_h2s_tc.

Acknowledgements

We thank the Caenorhabditis Genetics Center (funded by NIH Office of Research Infrastructure Programs P40 OD010440) and the National BioResources Project Japan for strains.

Funding

The computations were enabled by resources provided by the National Academic Infrastructure for Supercomputing in Sweden (NAISS) and the Swedish National Infrastructure for Computing (SNIC) at UPPMAX partially funded by the Swedish Research Council through grant agreements no. 2022-06725 and no. 2018-05973. This work is supported by the Swedish VR Research Council grant MIMS (2021-06602) to J.H., the Belgian National Fund for Scientific Research (FRS-FNRS) to P.L., the ERC starting grant (802653 OXYGEN SENSING), the Swedish Research Council VR starting grant (2018-02216), and the Wallenberg Centre for Molecular Medicine (Umeå) to C.C.

Figure legends

Figure 1—figure supplement 1.

Figure 1—figure supplement 2.

Figure 2—figure supplement 1.

Figure 3—figure supplement 1.

Figure 5—figure supplement 1.