Introduction

Cysteine is a sulfur-containing amino acid that mediates many oxidation/reduction reactions of proteins, is a substrate in the production of iron-sulfur clusters and hydrogen sulfide (H₂S), a volatile signaling molecule, and a precursor to the antioxidant glutathione (1–3). Cysteine residues in close proximity in the primary or folded protein sequence are oxidized in the endoplasmic reticulum to form intra- and interprotein disulfide linkages, most commonly in secreted proteins which mediate intercellular signaling and defense (4–6). In many enzymes, the reactivity of the cysteine sulfur is key for the coordination of metals such as zinc or iron, which support protein structure and catalytic activity (7–9). In contrast to its essential function, excess cysteine is also toxic. High cysteine impairs mitochondrial respiration by disrupting iron homeostasis (10), acts as a neural excitotoxin (11), and promotes the formation of toxic levels of hydrogen sulfide gas (2, 12, 13). Given this balance between essential and toxic, cysteine homeostasis is key for the health of cells and organisms.

Cysteine is oxidized to cysteinesulfinate by cysteine dioxygenase (CDO-1 in C. elegans, CDO1 in mammals) (Fig. 1A) (14). CDO1-mediated oxidation is the primary pathway of cysteine catabolism when sulfur amino acid (methionine or cysteine) availability is normal or high (15). As a critical player in cysteine homeostasis, CDO1 is a highly regulated enzyme: the activity and abundance of CDO1 increase dramatically in cells and animals fed excess cysteine and methionine (16). CDO1 activation is governed both transcriptionally and post-translationally via the 26S proteasome (17–21). However, the molecular players that sense high cysteine and promote CDO1 activation remain incompletely defined.

Figure 1:

CDO1 dysfunction is also implicated in disease. CDO1 is a tumor suppressor whose activity is silenced in diverse cancers via promoter methylation, a potential biomarker of tumor grade and progression (22–24). Decreased CDO1 activity may support tumor cell growth by reducing reactive oxygen species and decreasing drug susceptibility (25–28).

Using the nematode C. elegans, we have shown that CDO-1 is a key player in the pathophysiology of 2 fatal inborn errors of metabolism; molybdenum cofactor deficiency (MoCD) and isolated sulfite oxidase deficiency (ISOD) (29, 30). Molybdenum cofactor (Moco) is an essential 520 Dalton prosthetic group synthesized from GTP by a conserved multistep biosynthetic pathway that is present in about 2/3 of bacterial genomes and nearly all eukaryotic genomes (31, 32). C. elegans can either retrieve Moco synthesized by the bacteria it consumes or can synthesize Moco de novo using its own Moco biosynthetic pathway (33). C. elegans strains carrying mutations in genes encoding Moco biosynthetic enzymes (moc) and feeding on wild-type E. coli develop normally. Yet, these same moc-mutant C. elegans fed Moco-deficient E. coli are completely inviable. A genetic selection for mutations that suppress this inviability identified mutations in cth-2 (cystathionase) and cdo-1 (cysteine dioxygenase), enzymes in the cysteine biosynthetic and degradation pathway. Using the dipeptide cystathionine as a substrate, CTH-2 produces α-ketobutyrate and cysteine. Cysteine is further oxidized by CDO-1, generating highly toxic sulfites. These sulfites are normally oxidized to more benign sulfate by Moco-requiring sulfite oxidase, SUOX-1, an essential enzyme in C. elegans and humans (Fig. 1A) (30, 33). Therefore, loss of cdo-1 or cth-2 suppresses the lethality caused by Moco deficiency in C. elegans by not producing sulfites which are toxic when SUOX-1 is inactive due to Moco deficiency (33–35). Thus, understanding the fundamental mechanisms that govern the levels and activity of CDO-1 is critical to generating new therapeutic hypotheses to treat these diseases.

To define genes that regulate cdo-1 levels and activity, we performed an unbiased genetic screen in the nematode C. elegans to identify mutations that increase the expression or abundance of CDO-1. We identified loss-of-function mutations in 2 genes, egl-9 and rhy-1, that dramatically increase expression of a Pcdo-1::CDO-1::GFP reporter transgene. These enzymes act in the hypoxia and H₂S-sensing pathway, and we demonstrate that the conserved hypoxia-inducible transcription factor (HIF-1) activates cdo-1 transcription in this pathway (36–38). We further show that high cysteine promotes cdo-1 transcription, and that hif-1 and cysl-1 (another component of the H₂S-sensing pathway) are required for viability under high cysteine conditions. We demonstrate that transcriptional activation of cdo-1 via HIF-1 promotes CDO-1 activity and establish the C. elegans hypodermis as a key tissue of CDO-1 activation and function. Unexpectedly, we find that cdo-1 regulation is governed by a HIF-1 pathway largely independent of EGL-9 prolyl hydroxylase activity and von Hippel-Lindau (VHL-1), the canonical oxygen-sensing pathway (39, 40). These data establish a new connection between the HIF-1/H₂S-sensing pathway and sulfur amino acid catabolism governed by CDO-1.

Results

egl-9 and rhy-1 negatively regulate cdo-1 transcription

To identify regulators of CDO-1 levels, we engineered a transgene expressing a C-terminal green fluorescent protein (GFP) tagged version of full-length CDO-1 protein driven by the native cdo-1 promoter (Pcdo-1::CDO-1::GFP, Fig. 1B). Transgenic animals were generated by integrating the Pcdo-1::CDO-1::GFP construct into the C. elegans genome (41). This transgene was functional and rescued a cdo-1 loss of function mutation (Fig. S1). The Pcdo-1::CDO-1::GFP transgene reverses the suppression of Moco-deficient lethality caused by cdo-1 loss of function (Fig. S1). The reanimation of Moco-deficient lethality by the transgene depends on CDO-1 enzymatic activity because a transgene expressing an active-site mutant Pcdo-1::CDO-1[C85Y]::GFP does not rescue the cdo-1 mutant phenotype (Fig. S1) (33, 42). These data demonstrate that the Pcdo-1::CDO-1::GFP reporter transgene is functional and suggests its expression pattern reflects endogenous protein expression, localization, and levels.

We performed a mutagenesis screen to identify genes that control the expression or accumulation of CDO-1 protein. Specifically, we performed an EMS chemical mutagenesis of C. elegans and screened for mutations that caused increased GFP accumulation by the Pcdo-1::CDO-1::GFP reporter transgene. New mutants were isolated two generations post mutagenesis to allow novel mutations to become homozygous (43). Using whole-genome sequencing, we determined that 2 independently isolated mutants carried distinct mutations in egl-9 and four other independently isolated mutants had unique mutations in rhy-1 (Fig. 1C). The presence of multiple independent alleles suggests these mutations in egl-9 or rhy-1 are causative for the increased Pcdo-1::CDO-1::GFP expression or accumulation phenotype. egl-9 encodes the oxygen-sensing prolyl hydroxylase and orthologue of mammalian EGLN1. rhy-1 encodes the regulator of hypoxia inducible transcription factor, an enzyme with homology to membrane-bound O-acyltransferases (36, 39). The genetic screen produced nonsense alleles of both egl-9 and rhy-1 suggesting the increased Pcdo-1::CDO-1::GFP expression or accumulation phenotype is caused by loss of egl-9 or rhy-1 function (Fig. 1C). Inactivation of egl-9 or rhy-1 activates a transcriptional program mediated by the hypoxia inducible transcription factor, HIF-1 (36, 39). To determine if egl-9 or rhy-1 regulate cdo-1 transcription, we engineered a separate reporter construct where only GFP is transcribed by the cdo-1 promoter (Pcdo-1::GFP, Fig. 1B). This Pcdo-1::GFP transgene was introduced into strains with independently isolated egl-9(sa307) and rhy-1(ok1402) null reference alleles (36, 44). The egl-9(sa307) and rhy-1(ok1402) mutations dramatically induce GFP driven by the Pcdo-1::GFP transgene (Fig. 1D,E). The activation of cdo-1 transcription by independently isolated egl-9 or rhy-1 mutations demonstrates that the egl-9 and rhy-1 mutations isolated in our screen for the induction or accumulation of Pcdo-1::CDO-1::GFP are the causative genetic lesions. Furthermore, these data demonstrate that egl-9 and rhy-1 are necessary for the normal transcriptional repression of cdo-1.

HIF-1 directly activates cdo-1 transcription downstream of the rhy-1, cysl-1, egl-9 genetic pathway

rhy-1 and egl-9 act in a pathway that regulates the abundance and activity of the HIF-1 transcription factor (36, 38). The activity of rhy-1 is most upstream in the pathway and negatively regulates the activity of cysl-1, which encodes a cysteine synthase-like enzyme likely of algal origin (45). CYSL-1 directly binds to and inhibits EGL-9 in an H₂S-modulated manner (38). EGL-9 uses molecular oxygen as well as an α-ketoglutarate cofactor to directly inhibit HIF-1 via prolyl hydroxylation, which recruits the VHL-1 ubiquitin ligase to ubiquitinate HIF-1, targeting it for degradation by the proteasome (46–48). Given that loss-of-function mutations in rhy-1 or egl-9 activate cdo-1 transcription, we tested if cdo-1 transcription is activated by HIF-1 as an output of this hypoxia/H₂S-sensing pathway. We performed epistasis studies using null mutations that inhibit the activity of hif-1 (cysl-1(ok762) and hif-1(ia4)) or activate hif-1 (rhy-1(ok1402) and egl-9(sa307)). The induction of Pcdo-1::GFP by egl-9 inactivation was dependent upon the activity of hif-1, but not on the activity of cysl-1 (Fig. 2A,B). In contrast, induction of Pcdo-1::GFP by rhy-1 inactivation was dependent upon the activity of both hif-1 and cysl-1 (Fig. 2A,C). These results reveal a genetic pathway whereby rhy-1, cysl-1, and egl-9 function in a negative-regulatory cascade to control the activity of HIF-1 which transcriptionally activates cdo-1. Our epistasis studies of cdo-1 transcriptional regulation by HIF-1 align well with previous analyses of this genetic pathway in the context of transcription of cysl-2 (a paralog of cysl-1) and the “O₂-ON response” (38).

Figure 2:

To demonstrate that HIF-1 activates transcription of endogenous cdo-1, we explored published RNA-sequencing data of wild-type, egl-9(-), and egl-9(-) hif-1(-) mutant animals (49). egl-9(-) mutant C. elegans display an 8-fold increase in cdo-1 mRNA compared to wild type. This induction was dependent on hif-1; a hif-1(-) mutation completely suppressed the induction of cdo-1 mRNA caused by an egl-9(-) mutation (49). These RNA-seq data confirm our findings using the Pcdo-1::GFP transcriptional reporter that HIF-1 is a transcriptional activator of cdo-1. ChIP-seq data of HIF-1 performed by the modERN project show that HIF-1 directly binds the cdo-1 promoter (peak from −1,165 to −714 base pairs 5’ to the cdo-1 ATG start codon) (50). This HIF-1 binding site contains 3 copies of the HIF-binding motif (5’-RCGTG-3’) (51). We conclude that cdo-1 is a downstream effector of HIF-1 and is likely a direct transcriptional target of HIF-1.

High cysteine promotes cdo-1 transcription and cysl-1 and hif-1 are necessary for survival on high cysteine

Mammalian CDO1 levels and activity are highly induced by dietary cysteine (14–16). To determine if this homeostatic response is conserved in C. elegans, we exposed transgenic C. elegans carrying the Pcdo-1::GFP transcriptional reporter to high supplemental cysteine (100μM). Like our egl-9 and rhy-1 loss-of-function mutations, high cysteine promoted cdo-1 transcription (Fig. 3).

Figure 3:

We hypothesized that cysteine might activate cdo-1 transcription through the RHY-1/CYSL-1/EGL-9/HIF-1 pathway. In this pathway, cysl-1 and hif-1 act to promote cdo-1 transcription. We tested whether cysl-1 or hif-1 were necessary for the induction of Pcdo-1::GFP by high cysteine. However, this experiment was not possible as we observed 100% lethality in cysl-1(-); Pcdo-1::GFP (n=174 individuals) and hif-1(-); Pcdo-1::GFP (n=139 individuals) animals when exposed to 100μM supplemental cysteine. In contrast, 100% of cysl-1(-); Pcdo-1::GFP (n=143 individuals) and 99% hif-1(-); Pcdo-1::GFP (n=110 individuals) animals were alive under control conditions without supplemental cysteine. Wild-type C. elegans carrying the Pcdo-1::GFP reporter transgene were healthy under control (100% alive, n=134 individuals) and high-cysteine (100% alive, n=140 individuals) conditions. These data demonstrate that cysl-1 and hif-1 are necessary for survival under high cysteine conditions.

Given the established role of CDO-1 in cysteine catabolism, we tested whether cdo-1(-) mutants were also sensitive to high cysteine. 98% of cdo-1(-) mutants were alive after exposure to 100μM supplemental cysteine (n=141 individuals). 100% of cdo-1(-) mutant animals were alive in the control group not exposed to high cysteine (n=142 individuals). Similarly, wild-type C. elegans were healthy under control (100% alive, n=131 individuals) and high-cysteine (100% alive, n=153 individuals) conditions. Thus, cdo-1 is not necessary for survival under high cysteine conditions. This suggests the existence of alternate pathways that promote cysteine homeostasis. Given the requirement of cysl-1 and hif-1 for survival in high cysteine, we propose that HIF-1 activates pathways (in addition to cdo-1) that promote survival under high cysteine conditions.

Activated CDO-1 accumulates and is functional in the hypodermis

We sought to identify the site of action of CDO-1. To observe CDO-1 localization, we used CRISPR/Cas9 to insert the GFP open reading frame into the endogenous cdo-1 locus, fused in place of the native cdo-1 stop codon. This new cdo-1(rae273) allele encodes a C-terminal tagged “CDO-1::GFP” protein from the native cdo-1 genomic locus (Fig. 4A). To determine if CDO-1::GFP was functional, we combined CDO-1::GFP with a loss-of-function mutation in moc-1, a gene that is essential for C. elegans Moco biosynthesis (33). We then observed the growth of moc-1(-) CDO-1::GFP C. elegans on wild-type and Moco-deficient E. coli. moc-1(-) mutant C. elegans expressing CDO-1::GFP from the native cdo-1 locus arrest during larval development when fed Moco-deficient E. coli, but not when fed Moco-producing E.coli (Fig. 4C). This lethality is caused by the CDO-1-mediated production of sulfites which are only toxic when C. elegans is Moco deficient, and demonstrates that the CDO-1::GFP fusion protein is functional (33). These data suggest that the CDO-1::GFP expression we observe is physiologically relevant.

Figure 4:

When wild-type animals expressing CDO-1::GFP were grown under standard culture conditions, we observed CDO-1::GFP expression in multiple tissues including prominent expression in the hypodermis (Fig. 4B, Fig. S2). We tested if CDO-1::GFP levels would be affected by inactivation of egl-9 or rhy-1. We generated egl-9(-); CDO-1::GFP and rhy-1(-); CDO-1::GFP animals and assayed expression of CDO-1::GFP. Consistent with our studies using Pcdo-1::CDO-1::GFP and Pcdo-1::GFP transgenes, we found that CDO-1::GFP levels, encoded by cdo-1(rae273), were increased by egl-9(-) or rhy-1(-) mutations (Fig. 4B, Fig. S2). Furthermore, high cysteine also promoted accumulation of CDO-1::GFP (Fig. S3). In all scenarios, the hypodermis was the most prominent site of CDO-1::GFP accumulation. Specifically, we found that CDO-1::GFP was expressed in the cytoplasm of Hyp7, the major C. elegans hypodermal cell (Fig. 4B).

Based on the expression pattern of CDO-1::GFP encoded by cdo-1(rae273), we hypothesized that CDO-1 acts in the hypodermis to promote sulfur amino acid metabolism. To test this hypothesis, we engineered a cdo-1 rescue construct in which cdo-1 is expressed exclusively in the hypodermis under the control of the col-10 promoter (Pcol-10::CDO-1::GFP) (52). Multiple independent transgenic C. elegans strains were generated by integrating the Pcol-10::CDO-1::GFP construct into the C. elegans genome (41). We tested the ability of the Pcol-10::CDO-1::GFP transgene to rescue the cdo-1(-) mutant suppression of Moco-deficient larval arrest. We found that multiple independent transgenic strains of cdo-1(-) moc-1(-) double mutant animals carrying the Pcol-10::CDO-1::GFP transgene displayed a larval arrest phenotype when fed a Moco-deficient diet (Fig. 4D). These data demonstrated that hypodermal-specific expression of cdo-1 is sufficient to rescue the cdo-1(-) mutant suppression of Moco-deficient lethality. This rescue was dependent upon the enzymatic activity of CDO-1 as an active site variant of this transgene (Pcol-10::CDO-1[C85Y]::GFP) did not rescue the suppressed larval arrest of cdo-1(-) moc-1(-) double mutant animals fed Moco-deficient diets (Fig. 4D). Taken together, our analyses of CDO-1::GFP expression demonstrate that CDO-1 is expressed, and that expression is regulated, in multiple tissues, principal among them being the hypodermis and Hyp7 cell. Our tissue-specific rescue data demonstrate that hypodermal expression of cdo-1 is sufficient to promote cysteine catabolism and suggest that the hypodermis is a critical tissue for sulfur amino acid metabolism. However, we cannot exclude the possibility that CDO-1 also acts in other cells and tissues.

HIF-1 promotes CDO-1 activity downstream of the H₂S-sensing pathway

We sought to determine the physiological impact of cdo-1 transcriptional activation by HIF-1. We reasoned mutations that activate HIF-1 and increase cdo-1 transcription may cause increased CDO-1 activity. CDO-1 sits at a critical metabolic node in the degradation of the sulfur amino acids cysteine and methionine (Fig 1A). A key byproduct of sulfur amino acid metabolism and CDO-1 is sulfite, a reactive toxin that is detoxified by the Moco-requiring sulfite oxidase (SUOX-1). Null mutations in suox-1 cause larval lethality. However, animals carrying the suox-1(gk738847) hypomorphic allele are healthy under standard culture conditions (33, 34). suox-1(gk738847) mutant animals display only 4% SUOX-1 activity compared to wild type, and are exquisitely sensitive to sulfite stress (53). Thus, the suox-1(gk738847) mutation creates a sensitized genetic background to probe for increases in endogenous sulfite production. To test if increased cdo-1 transcription would impact the growth of suox-1-comprimised animals, we combined the egl-9 null mutation, which promotes HIF-1 activity and cdo-1 transcription, with the suox-1(gk738847) allele. While egl-9(-) and suox-1(gk738847) single mutant animals are healthy under standard culture conditions, the egl-9(-); suox-1(gk738847) double mutant animals are extremely sick and slow growing, establishing a synthetic genetic interaction between these loci (Table 1). To determine the role of sulfur amino acid metabolism in the egl-9(-); suox-1(gk738847) synthetic sick/slow growth phenotype, we engineered egl-9(-); cdo-1(-) suox-1(gk738847) and cth-2(-); egl-9(-); suox-1(gk738847) triple mutant animals. The egl-9; suox-1 synthetic sick/slow growth phenotype was suppressed by inactivating mutations in cdo-1 or cth-2 which block the endogenous production of sulfite (Table 1). These data demonstrate that the deleterious activity of the egl-9(-) mutation in a suox-1(gk738847) background requires functional sulfur amino acid metabolism.

Table 1:

To determine the role of the H₂S-sensing pathway in the synthetic sick/slow growth phenotype displayed by egl-9(-); suox-1(gk738847) double mutant animals, we introduced null alleles of cysl-1 or hif-1 into the egl-9(-); suox-1(gk738847) double mutant. The egl-9; suox-1 synthetic sick/slow growth phenotype was dependent upon hif-1 but not cysl-1 (Table 1). These results are consistent with our proposed genetic pathway and support the model that transcriptional activation of cdo-1 by HIF-1 causes increased CDO-1 activity and increased flux of sulfur amino acids through their catabolic pathway.

Loss of rhy-1 also strongly activates cdo-1 transcription. We hypothesized that rhy-1(-); suox-1(gk738847) double mutant animals would display a synthetic sick/slow growth phenotype like egl-9(-); suox-1(gk738847) double mutant animals. However, rhy-1(-); suox-1(gk738847) double mutant animals were just as healthy as either rhy-1(-) or suox-1(gk738847) single mutant C. elegans (Table 1). These data suggest that increasing cdo-1 transcription alone is not sufficient to promote sulfite production via CDO-1. In addition to the role played by rhy-1 in the regulation of HIF-1 activity, rhy-1 itself is a transcriptional target of HIF-1. Loss of egl-9 activity induces rhy-1 mRNA ∼50-fold in a hif-1-dependent manner (49). Given this potent transcriptional activation, we wondered if rhy-1 might play an additional role downstream of HIF-1 in the regulation of sulfur amino acid metabolism and sulfite production. To test this hypothesis, we engineered rhy-1(-); egl-9(-); suox-1(gk738847) triple mutant animals. To our surprise, the rhy-1(-); egl-9(-); suox-1(gk738847) triple mutant animals were healthy, demonstrating that rhy-1 was necessary for the deleterious activity of the egl-9(-) mutation in a suox-1(gk738847) background. These genetic data suggest a dual role for rhy-1 in the control of sulfur amino acid metabolism; first as a component of a regulatory cascade that controls the activity of HIF-1 and second as a functional downstream effector of HIF-1 that is required for sulfur amino acid metabolism.

This is not the first description of a rhy-1 role downstream of hif-1. Overexpression of a rhy-1-encoding transgene suppresses the lethality of a hif-1(-) mutant during H_SS stress (54). These data establish RHY-1 as both a regulator and effector of HIF-1. How RHY-1, a predicted membrane-bound O-acyltransferase, molecularly executes these dual roles remains to be explored.

EGL-9 prolyl hydroxylase activity and VHL-1 are largely dispensable in the regulation of CDO-1

EGL-9 inhibits HIF-1 through its prolyl hydroxylase domain (PHD) (Fig. 5A) (39). To evaluate the impact of the EGL-9 PHD on the regulation of cdo-1, we generated a PHD-inactive egl-9 mutation using CRISPR/Cas9. We engineered an egl-9 mutation that substitutes an alanine in place of histidine 487 (H487A) (Fig. 1C). Histidine 487 of EGL-9 is highly conserved and catalytically essential in the PHD active site (Fig. 5B) (55, 56). To evaluate the impact of an inactive EGL-9 PHD on the transcription of cdo-1, we engineered a C. elegans strain carrying the egl-9(H487A) mutation with the Pcdo-1::GFP transcriptional reporter. egl-9(H487A) caused a modest increase in Pcdo-1::GFP accumulation in the C. elegans intestine, suggesting that the EGL-9 PHD is necessary to repress cdo-1 transcription in the intestine (Fig. 5C,D). However, the activation of Pcdo-1::GFP by the egl-9(H487A) mutation was markedly less when compared to Pcdo-1::GFP activation caused by an egl-9(-) null mutation (Fig. 5C,D). These data suggest that EGL-9 has a PHD-independent activity that is responsible for repressing cdo-1 transcription.

Figure 5:

EGL-9 PHD hydroxylates specific proline residues on HIF-1. These hydroxylated proline residues are recognized by the von Hippel-Lindau E3 ubiquitin ligase (VHL-1). VHL-1-mediated ubiquitination promotes degradation of HIF-1 by the proteasome (Fig. 5A) (40). Thus, the EGL-9 PHD and VHL-1 act in a pathway to regulate HIF-1. To determine the role of VHL-1 in regulating cdo-1, we engineered a C. elegans strain carrying the vhl-1(-) null mutation with our Pcdo-1::GFP reporter. vhl-1 inactivation also caused a modest increase in Pcdo-1::GFP expression in the C. elegans intestine, suggesting that vhl-1 is necessary to repress cdo-1 transcription (Fig. 5C,D). However, activation of Pcdo-1::GFP by the vhl-1(-) mutation was less than activation caused by an egl-9(-) null mutation (Fig. 5C,D). These data suggest that EGL-9 has a VHL-1-independent activity that is responsible for repressing cdo-1 transcription.

To evaluate the impact of inactivating the EGL-9 PHD or VHL-1 on cysteine metabolism, we again employed the suox-1(gk738847) hypomorphic mutation that sensitizes animals to increases in sulfite. We engineered egl-9(H487A); suox-1(gk738847) and vhl-1(-) suox-1(gk738847) double mutant animals and evaluated the health of those strains. In contrast to egl-9(-); suox-1(gk738847) double mutant animals which are extremely sick/slow growing, egl-9(H487A); suox-1(gk738847) and vhl-1(-) suox-1(gk738847) double mutant animals are healthy (Table 1). These genetic data suggest that neither the EGL-9 PHD nor VHL-1 are necessary to repress cysteine catabolism/sulfite production. However, it is plausible that the egl-9(H487) or vhl-1(-) mutations modestly activate cysteine metabolism, likely proportional to their activation of the Pcdo-1::GFP transgene (Fig. 5C,D), and that this activation is not sufficient to produce enough sulfites to negatively impact the growth of suox-1(gk738847) mutant animals.

Discussion

CDO-1 is a physiologically relevant effector of HIF-1

The hypoxia-inducible factor HIF-1 is a master regulator of the cellular response to hypoxia. It activates the transcription of many genes and pathways that are critical to maintain metabolic homeostasis in the face of low oxygen. For example, mammalian HIF1A induces the hematopoietic growth hormone erythropoietin, glucose transport and glycolysis, as well as lactate dehydrogenase (57–60). The nematode C. elegans encounters a range of oxygen tensions in its natural habitat of rotting material: as microbial abundance increases, oxygen levels decrease from atmospheric levels (∼21% O₂). In fact, C. elegans prefers 5-12% O₂, perhaps because hypoxia predicts abundant bacterial food sources (61). Elements of the HIF-1 pathway have emerged from C. elegans genetics as have critical targets of HIF-1 (39). For instance, C. elegans HIF-1 promotes H₂S homeostasis by inducing transcription of the mitochondrial sulfide quinone oxidoreductase (sqrd-1), detoxifying H₂S (37).

We sought to define genes that regulate the levels and activity of cysteine dioxygenase (CDO-1), a critical regulator of cysteine homeostasis (14–16). Taking an unbiased genetic approach in C. elegans, we found that cdo-1 was highly regulated by HIF-1 downstream of a signaling pathway that includes rhy-1, cysl-1, and egl-9. We demonstrated that HIF-1 promotes cdo-1 transcription, accumulation of CDO-1 protein, and increased CDO-1 activity. Loss of rhy-1 or egl-9 activate hif-1 to in turn strongly induce the cdo-1 promoter. The pathway for activation of cdo-1 also requires cysl-1, which functions downstream of rhy-1 and upstream of egl-9. Based on ChIP-Seq studies, HIF-1 directly binds to the cdo-1 promoter (50).

We also show that the transcriptional activation of cdo-1 by HIF-1 promotes CDO-1 enzymatic activity. Genetic activation of cdo-1 by loss of egl-9 causes severe sickness in a mutant with reduced sulfite oxidase activity, an activity required to cope with the toxic products of CDO-1. This synthetic sickness is dependent upon an intact HIF-1 signaling pathway and a functioning sulfur amino acid metabolism pathway, as the dramatic sickness of an egl-9; suox-1 double mutant is suppressed by loss of hif-1, cth-2, or cdo-1. These striking genetic results validate the intersection of HIF-1 and CDO-1 as converging biological regulatory pathways and encourage further exploration of this regulatory node. The potential physiological relevance of the connection between HIF-1 signaling and cysteine metabolism is discussed below.

Evidence for distinct pathways of HIF-1 activation by hypoxia and cysteine/H₂S

The hypoxia-signaling pathway is defined by EGL-9-dependent prolyl hydroxylation of HIF-1. Hydroxylated HIF-1 is then targeted for degradation via VHL-1-mediated ubiquitination (39, 40, 51). However, multiple lines of evidence, reinforced by our work, demonstrate VHL-1- and prolyl hydroxylase-independent activity of EGL-9. egl-9(-) null mutant C. elegans accumulate HIF-1 protein and display increased transcription of many genes, including nhr-57, an established target of HIF-1 (36). In rescue experiments of an egl-9(-) null mutant, Shao et al. (2009) demonstrate that a wild-type egl-9 transgene restores normal HIF-1 protein levels and HIF-1 transcription. However, rescue experiments with a PHD-inactive egl-9(H487A) transgene do not correct the accumulation of HIF-1 protein and only partially reduce the HIF-1 transcriptional output (56). Through our studies of cdo-1 transcription, we demonstrate that an egl-9(H487A) mutant incompletely activates HIF-1 transcription when compared to an egl-9(-) null mutation (Fig. 5C,D). Taken together, we conclude EGL-9 has activity independent of its PHD domain, mirroring and supporting previous work (56).

In studies of the HIF-1-dependent Pnhr-57::GFP transcriptional reporter, Shen et al. (2006) observed that egl-9(-) null mutants promote HIF-1 transcription more than a vhl-1(-) mutant (36). We observe this same distinction between egl-9 and vhl-1 mutations with our Pcdo-1::GFP reporter (Fig. 5C,D). This difference in transcriptional activity is not explained by HIF-1 protein levels as HIF-1 protein accumulated equally in egl-9(-) and vhl-1(-) null mutants (36). This observation suggests that EGL-9 represses both HIF-1 levels and activity. This study also notes that vhl-1 represses HIF-1 transcription in the intestine while egl-9 acts in a wider array of tissues including the intestine and the hypodermis (36). Budde and Roth (2010) also demonstrate that loss of vhl-1 promotes HIF-1 transcription in the C. elegans intestine while total loss of egl-9 promotes HIF-1 transcription in multiple tissues including the intestine and the hypodermis (62). Our data expand upon these observations by demonstrating that loss of the EGL-9 PHD promotes cdo-1 transcription in the intestine alone, mirroring the loss of vhl-1 (Fig. 5C). Budde and Roth (2010) additionally demonstrate that physiologically relevant stimuli also elicit a tissue-specific transcriptional response: hypoxia promotes HIF-1 transcription in the intestine while H₂S promotes HIF-1 transcription in the hypodermis. Importantly, H₂S promotes hypodermal HIF-1 transcription in a vhl-1(-) mutant, demonstrating a VHL-1-independent pathway for H₂S activation of HIF-1 (62). We demonstrate that high cysteine promotes cdo-1 transcription in the hypodermis. Taken together with our data, these studies suggest 2 distinct pathways for activating HIF-1 transcription: i) a hypoxia-sensing pathway that is dependent upon vhl-1 and the EGL-9 PHD and promotes HIF-1 activity in the intestine and ii) an H₂S-sensing pathway that is independent of vhl-1 and the EGL-9 PHD and promotes HIF-1 activity in the hypodermis.

The genetic details of the H₂S-sensing pathway were solidified through studies of rhy-1 in C. elegans. Ma et al. (2012) demonstrate that hif-1 repression via rhy-1 requires the activity of cysl-1, a gene encoding a cysteine synthase-like protein (38). They further demonstrate that high H₂S promotes a physical interaction between CYSL-1 and EGL-9, resulting in the inactivation of EGL-9 and increased HIF-1 activity. Together, these studies suggest distinct genetic regulators of EGL-9/HIF-1 signaling: rhy-1 and cysl-1 govern the H₂S-sensing pathway while vhl-1 mediates the hypoxia-sensing pathway. These pathways are distinct in their requirement for the EGL-9 PHD.

A negative feedback loop senses high cysteine/H₂S, promotes CDO-1 activity, and maintains cysteine homeostasis

Why would the RHY-1/CYSL-1/EGL-9/HIF-1 H₂S-sensing pathway control the levels and activity of cysteine dioxygenase? We speculate this intersection facilitates a homeostatic pathway allowing C. elegans to sense and respond to cysteine level. We propose that H₂S acts as a gaseous signaling molecule to promote cysteine catabolism. H₂S activates HIF-1 in the hypodermis by promoting the CYSL-1-mediated inactivation of EGL-9 (38). We show that high cysteine similarly induces cdo-1 transcription in the hypodermis. Our genetic data demonstrate that cdo-1 is induced by the same genetic pathway that senses H₂S in C. elegans and CDO-1 acts in the hypodermis, the major site of H₂S-induced transcription. Furthermore, H₂S induces endogenous cdo-1 transcription >3-fold but cdo-1 mRNA levels do not change when C. elegans are exposed to hypoxia (63, 64). Thus, it is likely that H₂S promotes cdo-1 transcription through RHY-1, CYSL-1, EGL-9, and HIF-1. H₂S is a reasonable small molecule signal to alert cells to high cysteine stress. Excess cysteine results in the production of H₂S mediated by multiple enzymes including cystathionase (CTH), cystathionine ý-synthase (CBS), and 3-mercaptopyruvate sulfurtransferase (MST) (2, 65). We speculate that excess cysteine in C. elegans promotes the enzymatic production of H₂S which would activate HIF-1 via the RHY-1/CYSL-1/EGL-9 signaling pathway. In this homeostatic model, H₂S-activated HIF-1 would then induce cdo-1 transcription, promoting CDO-1 activity and the catabolism of the high-cysteine trigger (Fig. 6). Supporting this model, cysl-1(-) and hif-1(-) mutant C. elegans cannot survive in a high cysteine environment, demonstrating their central role in promoting cysteine homeostasis.

Figure 6:

Members of the H₂S-sensing pathway have also been implicated in the C. elegans response to infection (44, 66). Many secreted proteins that mediate intercellular signaling or innate immune responses to pathogens are cysteine-rich and form disulfide bonds in the endoplasmic reticulum before protein secretion (67, 68). Levels of free cysteine may fall during the massive inductions of cysteine-rich secreted proteins in development or immune defense (69). The oxidation of so many cysteines to disulfides in the endoplasmic reticulum might locally lower oxygen levels preventing EGLN1 hydroxylation of HIF1A. Thus, the intersection between oxygen sensing and cysteine homeostasis we have uncovered may contribute to a regulatory axis in cell-cell signaling during development and in immune function.

Methods

General methods and strains

C. elegans were cultured using established protocols (43). Briefly, animals were cultured at 20°C on nematode growth media (NGM) seeded with wild-type E. coli (OP50). The wild-type strain of C. elegans was Bristol N2. Additional E. coli strains used in this work were B2W5113 (Wild type, Moco+) and JW0764-2 (τιmoaA753::kan, Moco-) (70).

C. elegans mutant and transgenic strains used in this work are listed here. When previously published, sources of strains are referenced. Unless a reference is provided, all strains were generated in this study.

Non-transgenic C. elegans

JT307, egl-9(sa307) (44)

GR2254, moc-1(ok366) X (33)

GR2260, cdo-1(mg622) X (33)

GR2261, cdo-1(mg622) moc-1(ok366) X (33)

GR2269, suox-1(gk738847) X (33)

CB5602, vhl-1(ok161) X (39)

USD414, rhy-1(ok1402) II; suox-1(gk738847) X

USD421, egl-9(sa307) V; suox-1(gk738847) X

USD422, vhl-1(ok161) suox-1(gk738847) X

USD430, egl-9(sa307) V; cdo-1(mg622) suox-1(gk738847) X

USD431, hif-1(ia4) egl-9(sa307) V; suox-1(gk738847) X

USD432, rhy-1(ok1402) II; egl-9(sa307) V; suox-1(gk738847) X

USD433, cth-2(mg599) II; egl-9(sa307) V; suox-1(gk738847) X

USD434, egl-9(sa307) X; cysl-1(ok762) suox-1(gk738847) X

USD512, rhy-1(ok1402) II Outcrossed 4x for this work

USD706, unc-119(ed3) III; cdo-1(mg622) moc-1(ok366) X

USD920, cdo-1(rae273) moc-1(ok366) X

USD921, egl-9(sa307) V; cdo-1(rae273) X

USD922, rhy-1(ok1402) II; cdo-1(rae273) X

USD937, egl-9(rae276) V; suox-1(gk738847) X

MiniMos transgenic lines

USD531, unc-119(ed3) III; raeTi1 [pcdo-1::CDO-1::GFP unc-119(+)]

USD719, unc-119(ed3) III; cdo-1(mg622) moc-1(ok366); raeTi14 [pcdo-1::CDO-1(C85Y)::GFP unc-119(+)]

USD720, unc-119(ed3) III; raeTi15 [Pcdo-1::GFP unc-119(+)]

USD730, rhy-1(ok1402) II; unc-119(ed3) III; raeTi15 [Pcdo-1::GFP unc-119(+)]

USD733, unc-119(ed3) III; egl-9(sa307) V; raeTi15 [Pcdo-1::GFP unc-119(+)]

USD739, unc-119(ed3) III; cdo-1(mg622) moc-1(ok366) X; raeTi1 [pcdo-1::CDO-1::GFP unc-119(+)]

USD766, unc-119(ed3) III; cdo-1(mg622) moc-1(ok366) X; raeTi32 [Pcol-10::CDO-1::GFP unc-119(+)]

USD767, unc-119(ed3) III; cdo-1(mg622) moc-1(ok366) X; raeTi33 [Pcol-10::CDO-1::GFP unc-119(+)]

USD776, rhy-1(ok1402) II; unc-119(ed3) III; hif-1(ia4) V; raeTi15 [Pcdo-1::GFP unc-119(+)]

USD777, unc-119(ed3) III; egl-9(sa307) hif-1(ia4) V; raeTi15 [Pcdo-1::GFP unc-119(+)]

USD780, rhy-1(ok1402) II; unc-119(ed3) III; cysl-1(ok762) X; raeTi15 [Pcdo-1::GFP unc-119(+)]

USD787, unc-119(ed3) III; egl-9(sa307) V; cysl-1(ok762) X; raeTi15 [Pcdo-1::GFP unc-119(+)]

USD808, unc-119(ed3) III; cdo-1(mg622) moc-1(ok366) X; raeTi40 [Pcol-10::CDO-1[C85Y]::GFP unc-119(+)]

USD810, unc-119(ed3) III; cdo-1(mg622) moc-1(ok366) X; raeTi41 [Pcol-10::CDO-1[C85Y]::GFP unc-119(+)]

USD940, unc-119(ed3) III; vhl-1(ok161) X; raeTi15 [Pcdo-1::GFP unc-119(+)]

USD1160, unc-119(ed3) III; cysl-1(ok762) X; raeTi15 [Pcdo-1::GFP unc-119(+)]

USD1161, unc-119(ed3) III; hif-1(ia4) V; raeTi15 [Pcdo-1::GFP unc-119(+)]

EMS-derived strains

USD659, unc-119(ed3) III; egl-9(rae213) V; raeTi1 [pcdo-1::CDO-1::GFP unc-119(+)]

USD674, unc-119(ed3) III; egl-9(rae227) V; raeTi1 [pcdo-1::CDO-1::GFP unc-119(+)]

USD655, rhy-1(rae209) II; unc-119(ed3) III; raeTi1 [pcdo-1::CDO-1::GFP unc-119(+)]

USD656, rhy-1(rae210) II; unc-119(ed3) III; raeTi1 [pcdo-1::CDO-1::GFP unc-119(+)]

USD657, rhy-1(rae211) II; unc-119(ed3) III; raeTi1 [pcdo-1::CDO-1::GFP unc-119(+)]

USD658, rhy-1(rae212) II; unc-119(ed3) III; raeTi1 [pcdo-1::CDO-1::GFP unc-119(+)]

CRISPR/Cas9-derived strains

USD914, cdo-1(rae273) X, CDO-1::GFP

USD926, egl-9(rae276) V, EGL-9[H487A]

USD928, unc-119(ed3) III; egl-9(rae278) V; raeTi15 [Pcdo-1::GFP unc-119(+)]

MiniMos transgenesis

Cloning of original plasmid constructs was performed using isothermal/Gibson assembly (71). All MiniMos constructs were assembled in pNL43, which is derived from pCFJ909, a gift from Erik Jorgensen (Addgene plasmid #44480) (72). Details about plasmid construction are described below. MiniMos transgenic animals were generated using established protocols that rescue the unc-119(ed3) Unc phenotype (41).

To generate a construct that expressed CDO-1 under the control of its native promoter, we cloned the wild-type cdo-1 genomic locus from 1,335 base pairs upstream of the cdo-1 ATG start codon to (and including) codon 190 encoding the final CDO-1 amino acid prior to the TAA stop codon. This wild-type genomic sequence was fused in frame with a C-terminal GFP and tbb-2 3’UTR (376 bp downstream of the tbb-2 stop codon) in the pNL43 plasmid backbone. This plasmid is called pKW24 (Pcdo-1::CDO-1::GFP).

To generate pKW44, a construct encoding the active site mutant transgene Pcdo-1::CDO-1(C85Y)::GFP, we performed Q5 site-directed mutagenesis on pKW24, following manufacturer’s instructions (New England Biolabs). In pKW44, codon 85 was mutated from TGC (cysteine) to TAC (tyrosine).

To generate pKW45 (Pcdo-1::GFP), the 1,335 base pair cdo-1 promoter was amplified and fused directly to the GFP coding sequence. Both fragments were derived from pKW24, excluding the cdo-1 coding sequence and introns.

pKW49 is a construct driving cdo-1 expression from the hypodermal-specific col-10 promoter (Pcol-10::CDO-1::GFP) (52). The col-10 promoter (1,126 base pairs upstream of the col-10 start codon) was amplified and fused upstream of the cdo-1 ATG start codon in pKW24, replacing the native cdo-1 promoter. pKW53 [Pcol-10::CDO-1(C85Y)::GFP] was engineered using the same Q5-site-directed mutagenesis strategy as was described for pKW44. However, this mutagenesis used pKW49 as the template plasmid.

Chemical mutagenesis and whole genome sequencing

To define C. elegans gene activities that were necessary for the control of cdo-1 levels, we carried out a chemical mutagenesis screen for animals that accumulate CDO-1 protein. To visualize CDO-1 levels, we engineered USD531, a transgenic C. elegans strain carrying the raeTi1 [pKW24, Pcdo-1::CDO-1::GFP] transgene. USD531 transgenic C. elegans were mutagenized with ethyl methanesulfonate (EMS) using established protocols (43). F2 generation animals were manually screened, and mutant isolates were collected that displayed high accumulation of Pcdo-1::CDO-1::GFP. We demanded that mutant strains of interest were viable and fertile.

We followed established protocols to identify EMS-induced mutations in our strains of interest (73). Briefly, whole genomic DNA was prepared from C. elegans using the Gentra Puregene Tissue Kit (Qiagen) and genomic DNA libraries were prepared using the NEBNext genomic DNA library construction kit (New England Biolabs). DNA libraries were sequenced on an Illumina Hi-Seq and deep sequencing reads were analyzed using standard methods on Galaxy, a web-based platform for computational analyses (74). Briefly, sequencing reads were trimmed and aligned to the WBcel235 C. elegans reference genome (75, 76). Variations from the reference genome and the putative impact of those variations were annotated and extracted for analysis (77–79). Here we report the analysis of 6 new mutant strains (USD655, USD656, USD657, USD658, USD659, and USD674). Among these mutant strains, we found 2 unique mutations in egl-9 (USD659 and USD674) and 4 unique mutations in rhy-1 (USD655, USD656, USD657, USD658). The allele names and molecular identity of these new egl-9 and rhy-1 mutations are specified in Fig. 1C. These genes were prioritized based on the isolation of multiple independent alleles and their established functions in a common pathway, the hypoxia and H₂S-sensing pathway (36–38).

Genome engineering by CRISPR/Cas9

We followed standard protocols to perform CRISPR/Cas9 genome engineering of cdo-1 and egl-9 genomic loci in C. elegans (80–83). Essential details of the CRISPR/Cas9-generated reagents in this work are described below.

We used homology-directed repair to generate cdo-1(rae273) [CDO-1::GFP]. The guide RNA was 5’-gactacagaggatctaagaa-3’ (crRNA, IDT). The GFP donor double-stranded DNA (dsDNA) was amplified from pCFJ2249 using primers that contained roughly 40bp of homology to cdo-1 flanking the desired insertion site (84). The primers used to generate the donor dsDNA were: 5’-gtacggcaagaaagttgactacagaggatctaagaataatagtactagcggtggcagtgg-3’ and 5’-agaatcaacacgttattacattgagggatatgttgtttacttgtagagctcgtccattcc-3’. Glycine 189 of CDO-1 was removed by design to eliminate the PAM site and prevent cleavage of the donor dsDNA.

The egl-9(rae276) and egl-9(rae278) [EGL-9(H487A)] alleles were also generated by homology-directed repair using the same combination of guide RNA and single-stranded oligodeoxynucleotide (ssODN) donor. The guide RNA was 5’-tgtgaagcatgtagataatc-3’ (crRNA, IDT) The ssODN donor was 5’-gcttgccatctatcctggaaatggaactcgttatgtgaaggctgtagacaatccagtaaaagatggaagatgtataaccactatttattactg-3’ (Ultramer, IDT). Successful editing resulted in altering the coding sequence of EGL-9 to encode for an alanine rather than the catalytically essential histidine at position 487. We also used synonymous mutations to introduce an AccI restriction site that is helpful for genotyping the rae276 and rae278 mutant alleles.

C. elegans growth assays

To assay developmental rates, C. elegans were synchronized at the first stage of larval development. To synchronize animals, embryos were harvested from gravid adult animals via treatment with a bleach and sodium hydroxide solution. Embryos were then incubated overnight in M9 solution causing them to hatch and arrest development at the L1 stage (85). Synchronized L1 animals were cultured on NGM seeded with B2W5113 (Wild type, Moco+) or JW0764-2 (τ..moaA753::kan, Moco-) E. coli. Animals were cultured for 48 or 72 hours (specified in the appropriate figure legends), and live animals were imaged as described below. Animal length was measured from tip of head to the end of the tail.

To determine qualitative ‘health’ of various C. elegans strains, we assayed the ability of these strains to consume all E. coli food provided on an NGM petri dish. For this experiment, dietary E. coli was produced via overnight culture in liquid LB in a 37°C shaking incubator. 200μl of this E. coli was seeded onto NGM petri dishes and allowed to dry, producing nearly identical lawns and growth environments. Then, 5 L4 animals of a strain of interest were introduced onto these NGM petri dishes seeded with OP50. For each experiment, petri dishes were monitored daily and scored when all E. coli was consumed by the population of animals. This assay is beneficial because it integrates many life-history measures (i.e. developmental rate, brood size, embryonic viability, etc.) into a single simple assay that can be scaled and applied to many C. elegans strains in parallel.

Cysteine exposure

To determine the impact of supplemental cysteine on expression of cdo-1 and animal viability, we exposed various C. elegans strains to 0 or 100μM supplemental cysteine. Wild-type and mutant animals carrying the Pcdo-1::GFP or CDO-1::GFP transgenes were synchronously grown on NGM media supplemented with E. coli OP50 at 20°C until reaching the L4 stage of development. Live L4 animals were then collected from the petri dishes and cultured in liquid M9 media containing 4X concentrated E. coli OP50 with or without 100μM supplemental cysteine. These liquid cultures were gently rocked at 20°C overnight. Post exposure, GFP imaging was performed as described in the Microscopy section of the materials and methods. A subset of the assayed animals were also scored for viability after being seeded onto NGM petri dishes. Animals were determined to be alive if they responded to mechanical stimulus.

Microscopy

Low magnification bright field and fluorescence images (imaging GFP simultaneously in multiple animals) were collected using a Zeiss AxioZoom V16 microscope equipped with a Hamamatsu Orca flash 4.0 digital camera using Zen software (Zeiss). For experiments with supplemental cysteine, low magnification bright field and fluorescence images were collected using a Nikon SMZ25 microscope equipped with a Hamamatsu Orca flash 4.0 digital camera using NIS-Elements software (Nikon). High magnification differential interference contrast (DIC) and GFP fluorescence images (imaging CDO-1::GFP encoded by cdo-1(rae273)) were collected using Zeiss AxioImager Z1 microscope equipped with a Zeiss AxioCam HRc digital camera using Zen software (Zeiss). All images were processed and analyzed using ImageJ software (NIH). All imaging was performed on live animals paralyzed using 20mM sodium azide. For all fluorescence images shown within the same figure panel, images were collected using the same exposure time and processed identically. To quantify GFP expression, the average pixel intensity was determined within a set transverse section immediately anterior to the developing vulva. Background pixel intensity was determined in a set region of interest distinct from the C. elegans samples and was subtracted from the sample measurements.

Acknowledgements

Some C. elegans strains were provided by the CGC, which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440). Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under award numbers R01 GM044619 (to G.R.) and R35 GM146871 (to K.W.).

Supplemental Figures

Figure S1:

Figure S2:

Figure S3: