Introduction

During the late Proterozoic era, the rise in oxygen (O₂) levels eliminated hydrogen sulfide (H₂S) from most habitats. Yet, low O₂ levels, along with high concentrations of hydrogen sulfide (H₂S) and carbon dioxide (CO₂), can persist in enclosed environments where bacteria actively decompose organic matter (Olson & Straub, 2016). H₂S can easily permeate biological membranes and interfere with various cellular processes. One of the most detrimental effects of H₂S is the disruption of cellular respiration by remodeling the mitochondrial electron transport chain and inhibiting cytochrome c oxidase (COX) (Cooper & Brown, 2008; Khan et al., 1990; Nicholls & Kim, 1982; Romanelli-Cedrez, Vairoletti, & Salinas, 2024).

The detection and response to chemical signals are crucial for various organisms to interact effectively with their environment. Animals living in H₂S-rich environments have evolved mechanisms to either avoid or adapt to these conditions, conferring them 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. Livebearing fish species, for instance, have evolved COX subunits that are more resistant to H₂S and more efficient at H₂S detoxification (Kelley et al., 2016; Pfenninger et al., 2014).

Free-living nematodes feed on bacteria that thrive in decaying organic matter, such as compost. This complex and dynamic environment, where species like C. elegans are commonly found, may exhibit significant fluctuations in H₂S levels over short distances (Adams, Farwell, Pack, & Bamesberger, 1979; Budde & Roth, 2011; Morra & Dick, 1991; Patange, Breen, Arsuffi, & Ruvkun, 2025; Rodriguez-Kabana, Jordan, & Hollis, 1965; Romanelli-Cedrez et al., 2024). Evidence suggests that C. elegans has evolved tolerance to prolonged H₂S exposure by reprogramming gene expression through H₂S-induced stabilization of the hypoxia-inducible factor HIF-1 (Budde & Roth, 2010; Powell-Coffman, 2010), and by developing an H₂S-resistant electron transport chain (Romanelli-Cedrez et al., 2024). In particular, low concentrations of H₂S are not only well tolerated but can also be beneficial. For instance, exposure to 50 ppm H₂S has been shown to extend the lifespan and enhance thermotolerance (Budde & Roth, 2010; Fawcett, Hoyt, Johnson, & Miller, 2015; Horsman, Heinis, & Miller, 2019; Miller & Roth, 2007; Qabazard et al., 2014). Moreover, endogenous H₂S production in C. elegans modulates its interactions with actinobacteria (Patange et al., 2025). However, exposure to concentrations exceeding 150 ppm proved lethal (Budde & Roth, 2010, 2011; Fawcett et al., 2015; Horsman et al., 2019; Miller, Budde, & Roth, 2011a; Miller & Roth, 2007). In addition to long-term adaptive mechanisms, C. elegans exhibits an acute behavioral response to H₂S exposure (Budde & Roth, 2011). In this study, we investigated whether and how the nematode C. elegans acutely avoids environments rich in H₂S, how it adapts at the molecular and physiological levels to H₂S exposure, and how these adaptations alter the avoidance response.

Results

C. elegans increases locomotory activity in response to acute H₂S exposure

In response to aversive cues, C. elegans initiates a pirouette followed by an increase of its locomotory speed to escape the noxious stimuli. When acutely exposed to 150 ppm H₂S, the laboratory N2 strain exhibited an increased locomotory speed and enhanced turning behavior, while generally decreasing the frequency of reversals, as a strategy to escape the noxious stimuli (Figure 1A–C, Figure 1—Video 1). The maximum speed was reached after 6–8 minutes in H₂S and returned to the baseline over a period of approximately 1 hour (Figure 1D). The reduced locomotory speed after 30 minutes in 150 ppm H₂S was reversible when these animals were immediately challenged with another stimulus, such as near-UV light (Figure 1E), suggesting that neuromuscular function remains preserved. The magnitude and dynamics of the H₂S-induced avoidance responses were concentration-dependent. Under our assay conditions, 50 ppm H₂S did not elicit an acute speed increase, and responses to 75 ppm were variable. In contrast, exposure to 150 ppm H₂S consistently triggered a robust and reproducible escape response (Figure 1F), consistent with earlier reports that H₂S at 50 ppm exerts beneficial effects, while H₂S at 150 ppm is toxic to C. elegans (Budde & Roth, 2010; Fawcett et al., 2015; Horsman et al., 2019; Miller & Roth, 2007; Qabazard et al., 2014). To investigate whether low levels of H₂S might be preferable or even attractive to the animals, we exposed wild type animals to a linear gradient of H₂S (Figure 1G; See methods for details) (Bretscher, Busch, & de Bono, 2008; Chang, Chronis, Karow, Marletta, & Bargmann, 2006; Gray et al., 2004). Animals were allowed to move freely on a thin layer of bacteria within a microfluidic chamber containing an H₂S gradient ranging from 150 ppm at one end to 0 ppm at the other. As expected, C. elegans robustly avoids 150 ppm H₂S but does not accumulate at the extreme end of 0 ppm (Figure 1G, H). The highest proportion of animals was found in the region with approximately 40 ppm H₂S. These data suggest that H₂S acts as a potent repellent for C. elegans at high concentrations but as an attractant at low levels.

Figure 1.

Candidate gene screen identifies signaling pathways required for H₂S-evoked locomotion

To investigate the molecular mechanisms underlying the H₂S-evoked locomotory response in C. elegans, we conducted a candidate gene screen for mutants that failed to increase their locomotory speed upon exposure to an acute rise of H₂S level to 150 ppm. Our analysis prioritized genes that are either known to be, or potentially involved in, sensory responses to gaseous stimuli in both C. elegans and mammals (Supplementary file 1). We began by examining mutants affecting globins, guanylate cyclases, and cyclic nucleotide-gated (CNG) channels, which are involved in O₂ or CO₂ sensing in C. elegans, as well as wild isolates and other previously characterized mutants with defective responses to gas stimuli (Supplementary file 1) (Bretscher et al., 2008; Gray et al., 2004; Hallem & Sternberg, 2008; Persson et al., 2009). Given the essential role of K⁺ channels in acute hypoxia sensing in mammals, we also screened all potassium (K⁺) channel mutants (Gao, Gonzalez-Rodriguez, Ortega-Saenz, & Lopez-Barneo, 2017; Weir, Lopez-Barneo, Buckler, & Archer, 2005). To explore how the H₂S sensory signals were transmitted within the neural circuit, we assayed mutants with defects in ciliogenesis or cilia-mediated signaling, as well as animals deficient in neurotransmission involving classical neurotransmitters, neuropeptides, or biogenic amines. Finally, considering that mitochondrial function is central to H₂S detoxification, we included mitochondrial electron transport chain (ETC) mutants and key factors implicated in H₂S and superoxide clearance. Together, these functionally diverse categories allowed us to survey a wide spectrum of potential mechanisms contributing to the behavioral response to H₂S.

In the screen, a collection of mutants displayed reduced speed response, whereas only a few exhibited a lack of omega-turn and/or reversal responses to H₂S (Supplementary file 1). The reversal inhibition upon H₂S exposure and its rebound following H₂S withdrawal appeared to vary substantially across experiments for N2, suggesting that differences in population density, food availability, and animals’ physiological state prior to the assay may matter. Nevertheless, our screen revealed that signaling pathways, including cGMP, insulin and TGF-β signaling, contributed to H₂S-evoked avoidance (Supplementary file 1). A naturally occurring variant of the neuropeptide receptor gene npr-1, previously implicated in the modulation of O₂ and CO₂ sensing, also affected locomotory speed in H₂S (Supplementary file 1). In addition, mitochondrial components, including those involved in respiration and H₂S detoxification, played critical roles in triggering and maintaining escape behavior in H₂S. In contrast, we did not find evidence that globins, K⁺ channels, or biogenic amine signaling contributed significantly to the behavioral avoidance to H₂S (Supplementary file 1).

cGMP signaling in ASJ cilia contributes to H₂S avoidance

One group of mutants exhibiting reduced locomotory speed in H₂S had defects in ciliogenesis (Figure 2A–C). This group included daf-19 mutants lacking cilia, dyf-3 mutants with reduced cilia length, and dyf-7 mutants in which 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). These findings suggest that ciliogenesis and exposure of cilia to the external environment are essential for animals to escape H₂S exposure. In addition, these mutants exhibited consistently high reversal rate, likely associated with their reduced locomotory speed (Figure 2 —figure supplement 1A–F). 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 significantly attenuated locomotory speed in response to H₂S in tax-4 or tax-2 mutants, which lack key subunits of cilia-enriched cGMP-gated channels (Figure 2D, E). In a comprehensive survey of all guanylate cyclase mutants, the daf-11 strain emerged as the only mutant showing speed response defects in the avoidance of H₂S (Figure 2D, supplementary file 1). Similar to the cilia mutants, animals deficient in daf-11 had low speed, attenuated omega turns, and constantly high reversal rate (Figure 2D, Figure 2—figure supplement 1G, H). Selective expression of daf-11 in ASJ neurons, but not in other neurons, partially restored the speed response to acute H₂S (Figures 2F, Figure 2—figure supplement 2A–D). In addition, the H₂S response defect of tax-4 mutants was also rescued by selective expression of its cDNA in ASJ neurons (Figure 2G), suggesting that the receptor guanylate cyclase DAF-11 and the CNG channel TAX-4 act at least partially in ASJ neurons to promote acute avoidance to H₂S. To validate the role of ASJ neurons in H₂S responses, we blocked neurotransmission from ASJ using the catalytic domain of Tetanus Toxin (TeTx), which impairs neurosecretion by specifically cleaving synaptobrevin (Schiavo et al., 1992). Cell-specific expression of TeTx in ASJ significantly inhibited the speed response to H₂S (Figure 2H). Surprisingly, using the Ca²⁺ sensor GCaMP6s to monitor a direct response of ASJ neurons to H₂S, we observed no H₂S-induced Ca²⁺ transients under any of the tested conditions, whereas CO₂-evoked Ca²⁺ increases were readily detected (Figure 2—figure supplement 2E, F). These findings suggest that although ASJ neuronal activity and neurosecretion contribute to the H₂S responses, ASJ neurons are unlikely to serve as the primary H₂S sensor.

Figure 2.

Starvation modulates H₂S avoidance

In the candidate gene screen, another set of mutants with reduced speed responses to H₂S had previously been shown to be defective in response to CO₂ (Hallem & Sternberg, 2008). Specifically, the H₂S-evoked speed response was reduced in mutants deficient in nutrient-sensitive signaling, including the insulin receptor DAF-2, the TGF-β ligand DAF-7, and the nuclear hormone receptor NHR-49 (Figure 3A–C, Figure 3—figure supplement 1A). Similarly, mutants lacking other CO₂ response modulators, such as the calcineurin subunits TAX-6 and CNB-1, also failed to respond to H₂S (Figure 3—figure supplement 1B, C). In addition to the abolished speed response, we also observed that the H₂S-evoked omega-turn response was nearly abolished in daf-2 mutants, and high proportion of animals exhibited persistently high reversal rate regardless of H₂S stimulation (Figure 3—figure supplement 1D– F). The H₂S avoidance defects observed in daf-2 and daf-7 single mutants were fully suppressed in daf-2; daf-16 and daf-7; daf-3 double mutants (Figure 3A, B). Therefore, it is likely that insulin and TGF-β signaling modulate H₂S responses by regulating the expression of relevant genes via DAF-16 and DAF-3 transcription factors, respectively. Given that H₂S-evoked speed response was modulated by the nutrient-sensitive pathways, we further explored whether the locomotory response to H₂S was sensitive to the nutrient state. Similar to the response to CO₂ (Bretscher et al., 2008; Hallem & Sternberg, 2008), we observed that a 24-hour period of starvation suppressed the speed response to H₂S (Figure 3D). However, the CO₂ sensor GCY-9 was dispensable for H₂S-evoked avoidance (Figure 3—figure supplement 1G– I). These findings suggest that even though acute avoidance of CO₂ and H₂S are modulated through shared mechanisms, distinct molecular sensors are used.

Figure 3.

The O₂ sensing circuit antagonizes H₂S avoidance in wild isolates

Wild strains of C. elegans thrive in environmental niches characterized by variable O₂, CO₂ and H₂S levels (Adams et al., 1979; Bretscher et al., 2008; Budde & Roth, 2011; Gea, Barrena, Artola, & Sanchez, 2004; Hallem & Sternberg, 2008; Morra & Dick, 1991; Oshins, Michel, Louis, Richard, & Rynk, 2022; Patange et al., 2025; Rodriguez-Kabana et al., 1965). Compared to the laboratory reference strain N2, wild isolates display robust responses to high O₂ but reduced sensitivity to CO₂ stimulation (Beets et al., 2020; Bretscher et al., 2008; Carrillo, Guillermin, Rengarajan, Okubo, & Hallem, 2013; Hallem & Sternberg, 2008; Kodama-Namba et al., 2013; McGrath et al., 2009). It is thought that the unique O₂ and CO₂ responses of N2 strain evolved during its domestication on solid media at atmospheric gas concentrations (Sterken, Snoek, Kammenga, & Andersen, 2015). We hypothesized that wild C. elegans strains might also exhibit an attenuated response to H₂S, and therefore included a set of wild strains in our candidate gene screen. Consistent with this hypothesis, wild isolates including CB4855, CB4856, and CB4858 exhibited reduced locomotory speed in H₂S (Figure 3E, Figure 3— figure supplement 1J).

A naturally occurring variation in the neuropeptide receptor NPR-1 predominantly determines the differences in locomotory responses to O₂ and CO₂ between the laboratory N2 strain and wild isolates. The NPR-1(215F) variant found in wild isolates is less active than the NPR-1(215V) found in N2 animals (de Bono & Bargmann, 1998). We wondered whether the variations in the npr-1 gene also contribute to differences in the acute response to H₂S. The absence of npr-1 was sufficient to reduce the speed response to H₂S, suggesting that active NPR-1 signaling promotes H₂S avoidance (Figure 3F). NPR-1(215V) primarily acts within the RMG interneurons to modulate responses to environmental stimuli by inhibiting signal output from these neurons (Figure 3G) (Laurent et al., 2015; Macosko et al., 2009). Cell-specific expression of the highly active npr-1(215V) variant in RMG neurons effectively restored the H₂S-evoked locomotory speed in npr-1 null mutant worms (Figure 3H). Furthermore, optogenetic stimulation of RMG interneurons using channelrhodopsin-2 (ChR2) significantly dampened the locomotory response to H₂S in N2 animals, suggesting that high RMG activity inhibits H₂S responses (Figure 3—figure supplement 1K).

The perception of 21% O₂ involves the soluble guanylate cyclases GCY-35/GCY-36 and the downstream cyclic nucleotide-gated (CNG) channels TAX-4/TAX-2 in the URX, AQR, and PQR O₂-sensing neurons, which relay the signals to RMG (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 investigated whether disrupting the O₂ sensing machinery would have the same effect as inhibiting RMG signaling pathway by NPR-1(215V). Mutating gcy-35 and tax-4, or genetically ablating the O₂ sensing neurons URX, AQR, and PQR by cell-specific expression of the pro-apoptotic gene egl-1 restored H₂S-evoked speed response in npr-1 null mutants (Figure 3I–K). Specific expression of gcy-35 cDNA in O₂ sensing neurons was sufficient to repress this response to H₂S in npr-1; gcy-35 double mutants (Figure 3L). Finally, enhancing the presynaptic activity 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-induced locomotory speed (Figure 3—figure supplement 1L). Taken together, these observations indicate that activation of the O₂-sensing circuit inhibits the H₂S-evoked avoidance response.

H₂S exposure reprograms gene expression in C. elegans

Wild isolates that were relatively insensitive to H₂S exposure (Figure 3E, Figure 3—figure supplement 1J) likely developed sensory and physiological adaptations to persist in environments where H₂S may transiently increase (Budde & Roth, 2011; Patange et al., 2025; Rodriguez-Kabana et al., 1965; Romanelli-Cedrez et al., 2024). These adaptations to H₂S are mediated, at least in part, by the reprogramming of gene expression (Miller et al., 2011a). To identify the genes whose expression was induced by prolonged H₂S exposure, we conducted a comparative analysis of transcriptome profiles in animals exposed to 50 ppm or 150 ppm H₂S for 1 hour, 2 hours, or 12 hours. RNA-seq analysis revealed that exposure to either 50 ppm or 150 ppm H₂S for one hour was sufficient to trigger significant changes in gene expression, with 518 or 304 genes showing differential expression, respectively (Figure 4A, Supplementary file 2). Genes induced by H₂S exposure for 1 or 2 hours displayed considerable overlap (Figure 4B, Figure 4—figure supplement 1A). Gene Ontology (GO) analysis revealed that biological processes such as defense against bacteria and cysteine biosynthesis from L-serine were significantly enriched after H₂S exposure (Figure 4C, D, Figure 4—figure supplement 1B, C). As expected, we observed a robust induction of genes involved in H₂S detoxification (Figure 4E, 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 produce 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 4E, F, Figure 4—figure supplement 1D, E, Supplementary file 2).

Figure 4.

H₂S exposure is known to activate the HIF-1 pathway (Budde & Roth, 2010, 2011; Ma, Vozdek, Bhatla, & Horvitz, 2012; Miller et al., 2011a; Miller & Roth, 2007; Vora et al., 2022). In line with this, a proportion of genes with HIF-1 regulated promoters displayed increased expression after 1 or 2 hours of H₂S exposure (Figure 4—figure supplement 1F). Among these were the aforementioned detoxifying enzymes GST-19, SQRD-1, and ETHE-1, as well as CYSL-1, CYSL-2, and CDO-1, which are involved in cysteine metabolism (Supplementary file 2). Consistent with earlier studies, we also observed differential expression of a set of SKN-1 targets in response to chronic H₂S exposure (Figure 4—figure supplement 1G) (Horsman et al., 2019; Miller et al., 2011a; Niu et al., 2011). This included the heat shock protein genes hsp-16.2 and hsp-16.41 (Figure 4E, F, Figure 4—figure supplement 1D, E). Surprisingly, we detected a rapid increase in the expression of genes associated with intracellular H₂S production, including the methanethiol oxidase-encoding gene semo-1 and the mercaptopyruvate sulfurtransferase-encoding gene mpst-3 (Figure 4E, F, Figure 4—figure supplement 1D, E) (Philipp et al., 2022; Qabazard et al., 2014). Elevated expression of 3-mercaptopyruvate sulfurtransferase has similarly been reported in sulfide spring fish, suggesting a conserved response mechanism to H₂S (Kelley et al., 2016; Mathew, Schlipalius, & Ebert, 2011; Qabazard et al., 2014). However, the functional implications of increased expression of H₂S synthesis genes remains unclear.

When exposed to H₂S for 12 hours, the differentially expressed genes showed reduced overlap with those identified at earlier time points (Figure 4B, Figure 4—figure supplement 1A), suggesting that distinct defense strategies may be employed at various stages of H₂S exposure. Furthermore, substantial overlap was observed between the sets of genes induced by 50 ppm and 150 ppm H₂S after 1h or 2h exposure. However, this overlap decreased significantly after 12 hours exposure (Figure 4—figure supplement 1H–J). Additionally, we noticed 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, a response that was less pronounced at 50 ppm H₂S (Figures 4E, F, Figure 4—figure supplement 1D, E, K and L). Together, these findings suggest that multiple defense mechanisms are induced to adapt to the stress imposed by H₂S, which can diverge over time and at different H₂S concentrations.

Locomotory speed response to H₂S is modulated by HIF-1-induced detoxification

Given that HIF-1 signaling is robustly activated shortly after H₂S exposure (Figure 4—figure supplement 1F) (Budde & Roth, 2010, 2011; Miller et al., 2011a), we next sought to determine how HIF-1 signaling might contribute to H₂S-evoked avoidance response, and began by examining the effects of HIF-1 stabilization. Acclimating animals in 1% O₂ for 12 hours or longer significantly decreased the speed response to H₂S, indicating that HIF-1 stabilization inhibits the H₂S-evoked avoidance behavior (Figure 5A). To confirm this observation, we examined several mutants with stabilized HIF-1. The conserved proline-4-hydroxylase PHD/EGL-9 and the von Hippel-Lindau (VHL) tumor suppressor protein are required for HIF-1 degradation (Kaelin & Ratcliffe, 2008; Semenza, 2010). Disruption of either egl-9 or vhl-1 leads to HIF-1 stabilization under normoxic conditions. Similar to the effect of prolonged hypoxia, both mutants exhibited a markedly reduced locomotory speed in H₂S (Figure 5B). Interestingly, the omega-turn and reversal responses to H₂S were also substantially inhibited in egl-9 mutants (Figure 5—figure supplement 1A, B). In contrast, the locomotory speed response to 1% O₂ remained largely unaffected in egl-9 deficient animals (Figure 5—figure supplement 1C), suggesting a selective impairment of the H₂S response. In addition, expressing the non-degradable forms of HIF-1, hif-1(P621A or P621G), under the pan-neuronal unc-14 promoter or under the endogenous hif-1 promoter, was sufficient to inhibit H₂S evoked speed responses (Figure 5C). Together, these findings indicate that the activation of HIF-1 signaling during prolonged H₂S or hypoxia exposure mediates an adaptive response associated with reduced behavioral responses to H₂S, and that HIF-1 stabilization in neurons alone is sufficient to promote this behavioral adaptation.

Figure 5.

On the other hand, animals lacking HIF-1 signaling have previously been shown to display increased sensitivity to H₂S and become paralyzed rapidly upon exposure (Budde & Roth, 2010; Horsman et al., 2019). Consistent with this, hif-1 mutants exhibited a brief increase in speed upon H₂S exposure, which rapidly declined to baseline levels (Figure 5D). They also displayed a stronger initial omega-turn response and greater inhibition of the reversal rate than WT (Figure 5E, F), confirming that animals deficient in HIF-1 signaling are sensitized to H₂S exposure. This transient avoidance behavior was also observed in egl-9; hif-1 and vhl-1; hif-1 double mutants (Figure 5G, H), suggesting that this phenotype is linked to the absence of HIF-1 signaling. Importantly, hif-1 mutants displayed a robust speed response to other stimuli, such as near-UV light and 1% O₂ stimulation (Figure 5I, Figure5—figure supplement 1D), suggesting that the brief avoidance response is specific to H₂S rather than reflecting a general locomotory defect. Further supporting increased sensitivity to H₂S in hif-1 mutants, the locomotory speed response of hif-1 mutants to subsequent near-UV light stimulation was nearly abolished after 30 minutes of H₂S exposure, whereas egl-9 mutants and WT animals remained fully responsive (Figure 5I). The reduced movement and lack of responsiveness to other stressors in hif-1 mutants resemble the stress-induced sleep-like behavior previously observed in response to noxious heat (Byrne Rodgers & Ryu, 2020). Prompted by this observation, we next explored whether HIF-1-induced detoxification genes were also required to maintain H₂S-evoked behavioral responses (Figure 5—figure supplement 1E). Similar to hif-1 mutants, sqrd-1 and ethe-1 mutants displayed brief responses to H₂S, with locomotory activity rapidly returning to baseline levels (Figure 5J, K, Figure 5—figure supplement 1F–I). However, both mutants responded normally to 1% O₂, suggesting a specific hypersensitivity to H₂S toxicity (Figure 5—figure supplement 1J). Furthermore, we examined the contribution of additional genes whose expression was upregulated in response to H₂S, including the sulfhydrylase/cysteine synthase-encoding genes cysl-1, cysl-2, and cysl-3, as well as semo-1. Disrupting any of these genes phenocopied the responses observed in hif-1, ethe-1, or sqrd-1 mutants, while they displayed normal responses to 1% O₂ (Figure 5L, Figure 5—figure supplement 1D, K–P). These observations suggest that while the HIF-1-induced H₂S detoxification system is not essential for the initial sensing of H₂S, it is required to maintain animals’ locomotory activity in the presence of H₂S.

Labile iron pool sustains the locomotory activity in H₂S

Among the most differentially regulated genes upon exposure to H₂S, we observed consistent downregulation of ftn-1 and to a lesser extent ftn-2, whereas smf-3 expression was increased under specific H₂S conditions (Figure 6A, B, Figure 6—figure supplement 1A–D). The ferritin-encoding genes ftn-1 and ftn-2, along with the iron transporter smf-3, are critical for maintaining intracellular labile iron homeostasis in C. elegans (Ackerman & Gems, 2012; Gourley, Parker, Jones, Zumbrennen, & Leibold, 2003; Rajan et al., 2019; Romney, Newman, Thacker, & Leibold, 2011), suggesting that H₂S exposure may interfere with cellular iron storage. It has been shown that disrupting ftn-1 increases labile iron pools in the cytoplasm, while its overexpression reduces these levels (Anderson & Leibold, 2014; Romney et al., 2011). We found that ftn-1 mutants had enhanced speed response to H₂S, whereas ftn-1 overexpression attenuated H₂S-induced speed increase without damaging acute response to 1% O₂ (Figure 6C, D, Figure 6—figure supplement 1E). In addition, smf-3 mutants, which impair iron uptake, exhibited a reduced speed response to H₂S (Figure 6D). These results support the idea that H₂S exposure disrupts iron homeostasis.

Figure 6.

To complement our genetic analysis, we pharmacologically depleted labile irons. In the presence of the iron chelator 2,2’-dipyridyl (BP), the H₂S-evoked speed response was fully lost (Figure 6E), whereas locomotory speed in response to hypoxia was maintained (Figure 6—figure supplement 1F). We noticed that, similar to hif-1 mutants, both smf-3 mutants and BP-treated animals exhibited a strong initial omega-turn response to H₂S (Figure 6—figure supplement 1G, I), suggesting that these animals are hypersensitive to H₂S toxicity. One possibility is that low iron levels compromise H₂S detoxification capacity, as the detoxification enzyme ETHE-1 requires iron for its activity (Kabil & Banerjee, 2012; Pettinati, Brem, McDonough, & Schofield, 2015). By contrast, increasing iron levels by feeding ferric ammonium citrate (FAC) or by disrupting ftn-1 significantly sustained animals’ locomotory speed in H₂S (Figure 6D, E), likely by delaying H₂S-mediated iron depletion. However, the reorientation and reversals were not significantly affected (Figure 6—figure supplement 1G– J). This impact of FAC supplementation and ftn-1 disruption on speed response was largely lost in the absence of hif-1 (Figure 6F, G), suggesting that the effects of iron require HIF-1-dependent detoxification (Figure 6H). Yet, increased iron availability modestly improved the locomotory activity of hif-1 mutants in H₂S (Figure 6F, G), presumably by compensating for the low iron levels in hif-1 mutant animals (Rajan et al., 2019; Romney et al., 2011).

Mitochondrial function modulates H₂S-evoked locomotion responses

The mitochondrial electron transport chain (ETC) is not only the primary target of H₂S toxicity, but is also an essential component of the H₂S detoxification pathways (Cooper & Brown, 2008; Khan et al., 1990; Nicholls & Kim, 1982; Romanelli-Cedrez et al., 2024). Based on this, we hypothesized that ETC plays a central role in the locomotory response to H₂S. Indeed, mutating the genes critical for ETC function, including gas-1, clk-1, mev-1, and isp-1, impaired the H₂S-evocked speed and reversal responses (Figure 7A–D, Figure 7— figure supplement 1A–D). These data suggest that a functional ETC is required to trigger and support H₂S-evoked avoidance behavior or that dysfunctional ETC induces an adaptive response to cellular stressors. Consistently, exposing animals to rotenone, an ETC complex I inhibitor, robustly inhibited the speed responses to H₂S at all time points we tested (Figure 7E). Interestingly, we noticed that transient rotenone exposure substantially increased animals’ basal locomotory speed (Figure 7E) (Onukwufor et al., 2022), suggesting that acute ETC interruption evokes behavioral avoidance, in a manner similar to that observed with other noxious stimuli. However, a 2-hour exposure to rotenone abolished the H₂S-evocked speed response without increasing the basal locomotory speed (Figure 7E), suggesting that prolonged ETC inhibition may either impair the cellular capacity to support H₂S-evoked behavioral responses or drive animals’ adaptation to cellular stress and consequently reduce their locomotory response to H₂S.

Figure 7.

Given that rotenone exposure and mutations in ETC genes increase mitochondrial ROS generation (Ishii et al., 2013; Ochi et al., 2016; Ramsay & Singer, 1992; Yang & Hekimi, 2010; Zorov, Juhaszova, & Sollott, 2014), and that high levels of H₂S also induce ROS production (Jia et al., 2020), we sought to determine how mitochondrial ROS contributes to speed response to high H₂S. We focused on superoxide dismutases (SODs), key enzymes in superoxide detoxification. The C. elegans genome encodes five sod genes: sod-1 and sod-5 encode cytosolic Cu/ZnSODs, sod-2 and sod-3 encode mitochondrial MnSODs, and sod-4 encodes the extracellular Cu/ZnSOD isoforms (Doonan et al., 2008; Zubovych, Straud, & Roth, 2010). All sod single mutants exhibited a robust initial response to 150 ppm H₂S (Figure 7F, G). Interestingly, the mitochondrial SOD mutants sod-2 and sod-3, as well as sod-5, showed an initial burst of locomotory activity followed by a rapid decline in speed in the presence of H₂S (Figure 7F, G), but maintained a high locomotory speed in response to 1% O₂ (Figure 7—figure supplement 1E). In the quintuple sod-1; sod-2; sod-3; sod-4; sod-5 mutant, which lacks all superoxide dismutases and presumably accumulates higher ROS levels, the H₂S-evoked speed and omega-turn responses were nearly abolished, and the proportion of animals exhibiting reversals remained consistently high, a pattern resembling that of egl-9 and ETC mutants (Figure 7G, Figure 7—figure supplement 1F, G). Similar to sod-2 and sod-3, the quintuple sod mutant exhibited robust speed response to 1% O₂ (Zhao et al., 2022). In addition, the quintuple mutants responded normally to near-UV light after 30 minutes of H₂S exposure (Figure 7—figure supplement 1H), indicating that their neuromuscular function remains largely preserved. Therefore, high ROS levels in sod mutants do not substantially compromise the detoxification capacity required for survival in H₂S. Instead, high ROS appears to promote adaptation to H₂S-induced stress, presumably by activating several stress-responsive pathways such as HIF-1, NRF2/SKN-1, and DAF-16/FOXO (Figure 7H) (Lennicke & Cocheme, 2021; Patten, Germain, Kelly, & Slack, 2010).

Discussion

Animals’ behavior and physiology are profoundly influenced by the environments in which they evolved. The laboratory strain N2, which is adapted to low atmospheric CO₂ and H₂S concentrations, robustly avoids both gases (Beets et al., 2020; Bretscher et al., 2008; Carrillo et al., 2013; Hallem & Sternberg, 2008; Kodama-Namba et al., 2013; McGrath et al., 2009). Upon acute exposure to H₂S above 75 ppm, N2 exhibits an avoidance behavior characterized by reorientation and increased locomotion. This response is significantly reduced in wild isolates. These divergent responses between the laboratory strain and wild isolates are primarily driven by variations in the npr-1 gene, which encodes a neuropeptide receptor, suggesting that C. elegans has undergone rapid evolutionary adaptation to its environment.

The behavioral response to a specific sensory stimulus is usually shaped by an interplay of multiple environmental and physiological cues. We observe that conditions dampening C. elegans’ response to CO₂ also impair the response to H₂S. Specifically, the speed response to H₂S is significantly inhibited at high O₂ levels, and in mutants with defective insulin or TGF-β pathway that signal starvation. However, despite these shared modulation, CO₂ and H₂S sensing involve distinct molecular mechanisms. For example, CO₂ responses are mediated by the guanylate cyclase GCY-9 in BAG neurons (Hallem et al., 2011), which is dispensable for H₂S sensing. In addition, acute exposure to H₂S induces stronger reorientation but delayed acceleration compared to CO₂, suggesting different sensing and information processing mechanisms (Bretscher et al., 2008). Therefore, even though responses to CO₂ and H₂S are modulated by overlapping regulatory mechanisms, the two gases are detected by distinct sensors and processed differently.

In a candidate gene survey, we excluded the potential involvement of globins, K⁺ channels, biogenic amines, and most guanylate cyclases in H₂S responses. However, we observed that H₂S avoidance requires the activity of guanylate cyclase DAF-11 in ASJ neurons, as well as neurosecretion from these neurons. Since the DAF-11 pathway and neurosecretion from ASJ neurons regulate developmental programs that modify sensory functions in C. elegans (Murakami, Koga, & Ohshima, 2001), it is not surprising that daf-11 mutants display pleiotropic phenotypes including impaired H₂S and CO₂ responses (Hallem & Sternberg, 2008). In addition, we did not detect an H₂S-evoked calcium transient in ASJ neurons. Therefore, although ASJ neurons and DAF-11 activity are clearly required, the exact sensory neurons, circuits, and molecular triggers mediating acute H₂S avoidance remain to be elucidated.

Consistent with previous reports (Horsman et al., 2019; Miller et al., 2011a), H₂S exposure substantially alters the gene expression in C. elegans, including those involved in H₂S detoxification and iron homeostasis pathways. The repression of ftn-1 and induction of smf-3 observed during H₂S exposure suggest that H₂S depletes intracellular labile iron. Iron is a critical cofactor required for the activity of many enzymes and supports a wide range of cellular processes (Dev & Babitt, 2017; Hentze, Muckenthaler, & Andrews, 2004). One important enzyme that requires labile iron as cofactor is PHD/EGL-9, which targets HIF-1 for degradation. Changes in iron availability are therefore expected to modulate EGL-9 activity, which in turn alter the expression of HIF-1-regulated genes (Liochev, 1996; Myllyharju & Kivirikko, 1997; Read, Bentley, Archer, & Dunham-Snary, 2021; Xu & Moller, 2011). However, ETHE-1, a key H₂S detoxification enzyme upregulated upon HIF-1 stabilization, also requires iron for its persulfide dioxygenase activity (Kabil & Banerjee, 2012; Pettinati et al., 2015). Therefore, H₂S-evoked responses under varying iron availability are determined by the combined effects of labile iron on HIF-1 signaling and on the H₂S detoxification enzymes, among other factors. For instance, although reduced iron availability in smf-3 or BP-treated animals may promote the HIF-1 induced detoxification pathways by inactivating EGL-9, the actual enzymatic clearance of H₂S is impaired due to low ETHE-1 activity. Supporting this, smf-3 mutants and BP-treated animals display increased sensitivity to H₂S, including enhanced initial omega-turn responses and rapid inhibition of locomotion. In contrast, increased iron availability in ftn-1 mutants or FAC-treated animals likely delays iron depletion during H₂S exposure, preserving ETHE-1 detoxification capacity and postponing the onset of HIF-1-mediated adaptations that are associated with reduced locomotion. This may explain the sustained high locomotory speed observed in ftn-1 mutants or FAC-treated animals. In the absence of HIF-1, iron supplementation only partially improves the locomotory response to H₂S, likely because the H₂S detoxification system including ETHE-1 cannot be transcriptionally induced. The modest effect observed may instead reflect correction of iron deficiency in hif-1 mutants.

At concentrations below 50 ppm, H₂S is well tolerated or even preferred, with beneficial effects such as improved thermotolerance (Budde & Roth, 2010; Fawcett et al., 2015; Horsman et al., 2019; Miller & Roth, 2007; Qabazard et al., 2014). Notably, behavioral avoidance is absent below 50 ppm H₂S, suggesting that escape behavior is triggered only when H₂S is not efficiently detoxified. This led us to hypothesize that animals with enhanced detoxification capacity would show reduced avoidance of otherwise toxic H₂S levels. Indeed, the speed response to H₂S is attenuated when the detoxification program is upregulated under conditions of constitutive activation of HIF-1, such as in egl-9 or vhl-1 mutants, after prolonged exposure to 1% O₂, or when HIF-1 signaling is specifically activated in neurons. These observations support the idea that the initial avoidance response to H₂S is likely driven by acute toxicity, while the subsequent decline in speed reflects a rapid activation of adaptive mechanisms, particularly through stabilization of HIF-1. The fact that wild type animals remain responsive to other stimuli after prolonged H₂S exposure suggests that reduced locomotory speed in H₂S reflects active desensitization rather than general paralysis. However, persistently high H₂S exposure is likely to exhaust cellular defense systems, leading to toxicity and paralysis. By contrast, the rapid loss of locomotion observed in hif-1 mutants and detoxification-defective mutants is mediated by a mechanism distinct from adaptation. Their loss of responsiveness to other stimuli after H₂S exposure suggests that these animals likely become sensitized and rapidly intoxicated by H₂S due to impaired detoxification. Therefore, H₂S exposure promotes a behavioral program that includes an initial reorientation and acceleration responses followed by a progressive adaptation driven by cellular detoxification processes. A similar behavioral program has been observed in C. elegans during noxious heat exposure, which induces short-term heat avoidance followed by long-lasting cytoprotective adaptation and a gradual reduction in avoidance (Byrne Rodgers & Ryu, 2020).

The interaction of H₂S with mitochondrial ETC is multifaceted, acting as an electron donor at low concentrations and becoming a potent toxicant at high levels (Szabo et al., 2014), in part by promoting superoxide generation through complex IV inhibition and reverse electron transport at complex I (Cooper & Brown, 2008; Jia et al., 2020; Khan et al., 1990; Nicholls & Kim, 1982; Romanelli-Cedrez et al., 2024). We propose that mitochondria play a dual role in H₂S-evoked locomotory avoidance. On one hand, the mitochondrial ETC contributes to H₂S detoxification and promotes adaptation. On the other hand, toxic levels of H₂S remodel the ETC, leading to increased ROS production, which may serve as a trigger for the avoidance response. Supporting the idea that acute mitochondrial ROS generation initiates avoidance of high H₂S levels, short-term rotenone exposure, known to promote mitochondrial ROS formation (Ochi et al., 2016; Ramsay & Singer, 1992; Zorov et al., 2014), substantially increases locomotory speed (Onukwufor et al., 2022). Meanwhile, the speed response to high H₂S is fully suppressed by rotenone. This inhibition could result either from excessive mitochondrial ROS generated by rotenone, which may dampen the H₂S-triggered ROS spike, or from direct complex I inhibition, which may disrupt other H₂S signaling processes required to initiate avoidance. However, persistent mitochondrial ROS production appears to suppress high locomotory speed and inhibit responsiveness to H₂S, as observed after 2-hour rotenone exposure, in mitochondrial ETC mutants, and in animals lacking all superoxide dismutases. One likely explanation is that mitochondrial ROS can activate a variety of stress-responsive pathways, including HIF-1, NRF2/SKN-1, and DAF-16/ FOXO signaling (Lennicke & Cocheme, 2021; Patten et al., 2010), priming animals’ adaptation to prolonged stress rather than causing toxicity. This is supported by the observation that even though SOD-deficient animals do not display strong initial locomotory responses to H₂S, they remain responsive to other stimuli after 30 minutes of H₂S exposure, suggesting that high ROS levels do not compromise general viability or the H₂S detoxification capacity. Therefore, we favor a model in which mitochondrial ROS exert a biphasic effect on H₂S-induced avoidance, facilitating H₂S detection and avoidance under acute conditions, and contributing to locomotory inhibition when it is chronically elevated.

In summary, this study unveils a novel behavior of C. elegans in their avoidance to H₂S. The speed response to H₂S is intricately shaped by environmental context, integrating inputs from other sensory cues such as food availability and O₂ levels. Although we did not identify the specific neural circuit mediating H₂S signaling, our findings suggest that H₂S-induced locomotion arises from acute remodeling of mitochondrial ETC and mitochondrial ROS production. We emphasize the vital role of the HIF-1-induced detoxification pathways and iron homeostasis in protecting against H₂S-induced mitochondrial toxicity. In future work, it will be important to delineate the neural circuits underlying H₂S-evoked locomotory avoidance and to clarify how H₂S-induced changes in mitochondrial respiration and labile iron pools shape these responses.

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/Cas9 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 as well as in-frame and 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 Cas9 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 10 or 11 minutes in H₂S, and ending with 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. For each strain or condition, 3 to 4 replicates were performed.

Near-UV light experiments were conducted to assess whether animals remained responsive following a 30-minute incubation in 150 ppm H₂S balanced with 7% O₂. For each experiment, 6-minute videos were recorded. The first 2 minutes captured the baseline locomotion of the animals under white light, followed by 2 minutes of exposure to near-UV light (435 nm, 0.7 mW/mm²), and concluded with 2 minutes of white light. 3 to 4 replicates were performed, with approximately 15 animals recorded per replicate.

The H₂S gradient experiment was performed in a PDMS chamber connected to a pump delivering gas at 0.5 ml/min. To establish a gradient ranging from 150 ppm to 0 ppm H₂S, we pumped 150 ppm H₂S balanced with 7% O₂ into one side and 7% O₂ into the opposite side. After 25 minutes, the number of animals in each of the 5 sections was counted (Figure 1G). Four replicates were performed, each using 30 day-one adult animals.

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 5 mg/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, respectively.

Optogenetic experiments were performed as previously outlined (Zhao et al., 2022). Transgenic L4 animals were exposed to 100 μM 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.

Ca²⁺ Imaging

L4 animals expressing the GCaMP6s sensor in ASJ neurons were picked 24 hours before imaging. The agarose pads were prepared 1 hour before the experiment. Animals were incubated in 15 µl of 10 mM levamisole for 40 minutes prior to imaging. Between 15 and 20 animals were immobilized using 10 mM levamisole diluted in a concentrated OP50 bacterial suspension and mounted on a 2% agarose pad in M9 buffer on a glass slide. The animals were then covered with a microfluidic chamber. Imaging lasted for 16 minutes. O₂ (7%) was pumped into the microfluidic chamber during the first and last 4 minutes. From minute 4 to 12, a mixture of 7% O₂ and 150 ppm H₂S was pumped.

Imaging was performed on a Nikon AZ100 microscope equipped with a TwinCam adapter (Cairn Research, UK), fitted with two DMK 33U monochrome cameras (The Imaging Source, Germany). A 2x AZ Plan Fluor objective with 1x zoom was used, with an exposure time of 300 ms. Excitation of GCaMP and RFP was provided by a CoolLED pE-300 (CoolLED, UK). The TwinCam adapter was mounted with emission filters for green fluorescent protein (510/20 nm) and red fluorescent protein (590/35 nm), along with a DC/T510LPXRXTUf2 dichroic mirror. Imaging data were analyzed using Neuron Analyzer, a custom-written Matlab program for processing image stacks (code available at https://github.com/neuronanalyser/neuronanalyser).

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 periods (1 hour, 2 hours and 12 hours). For each time period, 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 each condition. 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 were 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).

Figure supplement legends

Figure 2—figure supplement 1.

Figure 2—figure supplement 2.

Figure 3—figure supplement 1.

Figure 4—figure supplement 1.

Figure 5—figure supplement 1.

Figure 6—figure supplement 1.

Figure 7—figure supplement 1.

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.

Additional information

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.

Funding

European Research Council (802653)

Swedish Research Council (2018-02216)

Additional files

Candidate gene screen

RNA-seq analysis

Strains

Reagents

Figure 1 - video 1. The response of WT animals to H2S An 8x accelerated video clip (original length: 16 minutes) showing WT (N2) animals exposed to 7% O₂ for 2 minutes, followed by 11 minutes in 150 ppm H₂S, and then returned to 7% O₂. The H₂S exposure segment is highlighted and begins at the 15-second mark, corresponding to 2 minutes in the original recording.