The transsulfuration pathway suppresses the embryonic lethal phenotype of glutathione reductase mutants in Caenorhabditis elegans

TL;DR

This study shows that a specific metabolic pathway can overcome the lethal effects of a gene mutation in C. elegans embryos.

Contribution

The study identifies the transsulfuration pathway as a suppressor of glutathione reductase mutant lethality in C. elegans.

Findings

Impairing the NMD pathway suppresses the embryonic lethality of gsr-1 mutants.

The transsulfuration pathway, via cth-1 and cth-2, is essential for this suppression.

The thioredoxin system is not required for this compensatory mechanism.

Abstract

The gsr-1 gene encodes the only glutathione reductase in Caenorhabditis elegans and gsr-1 loss-of-function alleles have a fully penetrant embryonic lethal phenotype. Therefore, maintenance of glutathione redox homeostasis is essential for nematode survival. We report here that impairment of the nonsense-mediated mRNA decay (NMD) pathway suppresses the embryonic lethality of gsr-1 mutants, allowing their normal development and growth. This NMD pathway dependent suppression requires cth-1 and cth-2 that encode 2 isoforms of cystathionine-γ-lyase that catalyze the conversion of cystathionine to cysteine through the transsulfuration pathway. In contrast, the thioredoxin system that can also provide cysteine through the cystine reduction pathway appears to be dispensable for the suppression of the lethal phenotype of gsr-1 embryos when the NMD pathway is inactivated. Together, our data indicate that increasing the activity of the reverse transsulfuration pathway can compensate the detrimental effect of the gsr-1 mutation, raising the interesting question of why C. elegans has not preserved such compensatory mechanism to avoid the embryonic lethality of these mutants.

Figures

mRNA and protein domain organization of wild-type and mutant gsr-1 and identification of the molecular lesion segregating with the gnaIs2 transgene. a) Schematic representation of the 2 gsr-1 mRNA variants. Boxes represent exons and lines show spliced introns. White boxes indicate 5′-UTR and 3′-UTR, respectively, and gray boxes indicate the ORF. Boundaries of gsr-1(tm3574) and gsr-1(syb2343) deletions are shown as black lines. b) Schematic representation of GSR-1 proteins. Protein domain organization of GSR-1a and GSR-1b isoforms as well as those of the shorter proteins resulting from translation of the gsr-1(tm3574) and gsr-1(syb2343) deletion alleles. c) The gnaIs2 transgene suppresses gsr-1 mutants embryonic lethality. gsr-1(tm3574) embryos carrying the gnaIs2 transgene (expressing YFP in pharynx and Aβ in neurons) in homozygosis develop normally. The gnaIs2 transgene in heterozygosity and control transgenes gnaIs1 (expressing YFP in pharynx) and dvIs50 (expressing Aβ in neurons) do not suppress gsr-1 mutants embryonic lethality. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red. d) Diagram of the deletion boundaries at 6.36 MB position of LG IV identified in the strain harboring the gnaIs2 integrated transgene. Genes within this region are represented in boxes where ORF exons are in blue and 3′-UTR exons are in gray, connected by lines representing the introns.

Impairment of the NMD pathway suppresses the embryonic lethal phenotype of gsr-1 mutants. a) Representative micrographs of first day adult gsr-1(tm3574), gnaIs2, or smg-3(ma117) mutants. All mutants develop similar to wild-type control, except that those carrying the gnaIs2 transgene or smg-3 mutation have a protruding vulva phenotype (Pvu). Insets show the magnification of the vulva area. b–d) smg-1(e1228), smg-2(e2008), and smg-3(ma117) mutations allow normal embryonic development of gsr-1(tm3574) and gsr-1(syb2363) animals. The variable number of arrested embryos in the strains carrying the smg-1, smg-2, and smg-3 alleles is due to the him-2(e1065) or him-5(e1490) (high incidence of males) mutations present in their respective backgrounds (Hodgkin et al. 1979) (See Supplementary Table 1 for complete genotype). In all cases, the number of arrested embryos substantially decreased upon crossing with the gsr-1 mutation, including the smg-1(e1228) allele that is linked to the him-2(e1065) mutation at LG I. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red.

The smg-2(e2008) mutation fails to suppress the molting phenotype of gsr-1; trxr-1 double mutants. All triple mutants gsr-1; smg-2; trxr-1 display a highly penetrant molting larval arrest phenotype, similar to that of the control strain gsr-1; trxr-1. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red.

The members of the thioredoxin system with GSSG-reducing activity are not required for the suppression of gsr-1 embryonic lethality upon impairment of the NMD pathway. a) C. elegans  TRXR-1 is devoid of GSSG-reducing activity in vitro. NADPH consumption in the presence of 1 mM GSSG of recombinantly expressed human glutathione reductase (Hs GR) and C. elegans thioredoxin reductase 1 (Ce  TRXR-1). b, c) The thioredoxins DPY-11, TRX-1, TXDC-17, and TXL-1 are not required for the embryonic development of gsr-1; smg-3 animals. All triple mutant combinations of genes encoding thioredoxin proteins with GSSG-reducing activity in a gsr-1; smg-3 double mutant background develop similarly as gsr-1; smg-3 control animals. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red.

Inactivation of the transsulfuration pathway impairs gsr-1; smg-3 double mutants development. a) Reverse transsulfuration pathway. Human proteins are depicted in red and C. elegans proteins in blue. CBS/CBS-1/CBS-2, cystathionine-β-synthase; CGL/CTH-1/CTH-2, cystathionine-γ-lyase. b) Simultaneous inactivation of cth-1 and cth-2 genes restores the embryonic lethal phenotype in gsr-1; smg-3 double mutants. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red. c) cth-1 and cth-2 mRNA levels in gsr-1 and smg-3 mutants. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates). Data were not significantly different by one-way ANOVA multiple comparisons test. d, e) CTH-1 and CTH-2 protein levels in gsr-1 and smg-3 mutants. Differential interference contrast (left panel) and fluorescence (right panel) microscopy images and quantification of CTH-1::GFP and CTH-2::GFP endogenous reporters in gsr-1 and smg-3 mutants. Graphs represent the data of 3 independent experiments with at least 30 embryos. * P < 0.001 by Kruskal–Willis with Dunn’s multiple comparisons test. Error bars are SEM.

Full text

Introduction

The tripeptide glutathione (GSH: L-γ-glutamyl-L-cysteinyl-glycine) is the most abundant low molecular weight thiol in the vast majority of organisms and it plays key roles in many cellular processes including cell signaling, DNA synthesis and repair, regulation of protein function, detoxification, or antioxidant defense. GSH performs its diverse functions either directly through nonenzymatic reactions or as cosubstrate in enzyme-catalyzed reactions (Wang et al. 2020). Glutathione is synthesized in the cytoplasm of eukaryotic cells by the sequential action of 2 ATP-consuming enzymes: γ-glutamylcysteine synthetase, which catalyzes the condensation of glutamate and cysteine, the rate limiting reaction of GSH biosynthesis, and glutathione synthetase, which catalyzes the conjugation of γ-glutamylcysteine with glycine (Lu 2013). The relevance of GSH as an essential metabolite is illustrated by the fact that mutations that impair γ-glutamylcysteine synthetase function causes lethality in all organisms studied from yeast to mammals (Wu and Moye-Rowley 1994; Dalton et al. 2000; Shi et al. 2000; Romero-Aristizabal et al. 2014; Enya et al. 2017). Similarly, genetic inactivation of glutathione synthetase is also lethal in Caenorhabditis elegans and mice (Winkler et al. 2011; Mora-Lorca et al. 2016). On the other hand, no overt phenotype in yeast under normal growth conditions is observed, probably due to the increased levels of γ-glutamylcysteine that may partially compensate the GSH requirement for growth (Grant et al. 1997). In humans, the very few cases of patients with glutathione synthetase deficiency have been found to have increased levels of γ-glutamylcysteine, supporting this compensatory role on GSH requirement (Ristoff et al. 2002; Njalsson 2005).

Once GSH serves as an electron donor it becomes oxidized (GSSG) and is then recycled to its reduced form by the flavoenzyme glutathione reductase (GR), which is not needed for yeast, zebrafish, or mice survival (Muller 1996; Rogers et al. 2004; Zhao et al. 2020). The dispensability of GR in these organisms has been explained by the thioredoxin system working as an alternative pathway to regenerate reduced glutathione (Muller 1996; Sun et al. 2001; Tan et al. 2010), although the exact mechanism needs yet to be fully elucidated. In contrast to yeast, zebrafish and mice, GR is essential for C. elegans development: homozygous gsr-1 animals segregating from heterozygous parents (hereafter referred as gsr-1(m+,z−); m, maternal and z, zygotic) have a normal embryonic and postembryonic development and reach adulthood indistinguishably of wild-type controls thanks to the maternally contributed gsr-1 mRNA and/or GSR-1 protein. In turn, the gsr-1(m−, z−) embryos generated by these gsr-1(m+,z−) adults arrest at the pregastrula stage due to lack of maternal contribution (Mora-Lorca et al. 2016). This embryonic lethal phenotype is intriguing as it has been shown that cytoplasmic thioredoxin reductase TRXR-1 and glutathione reductase GSR-1 have redundant functions in the nematode molting cycle, probably by TRXR-1 mediating a thioredoxin-dependent reduction of GSSG in the hypodermis in the absence of GSR-1 (Stenvall et al. 2011). Why TRXR-1 can functionally substitute GSR-1 to allow molting of gsr-1(m+,z−) larvae but not to allow gsr-1(m−, z−) embryos’ development is currently unknown.

When GSSG reduction is compromised in mammalian cells by simultaneous blockage of GR and cytoplasmic thioredoxin reductase 1, cells rely on de novo glutathione synthesis for survival using the reverse transsulfuration pathway to provide cysteine as a GSH precursor (Eriksson et al. 2015). This pathway employs homocysteine (generated from methionine by the S-adenosylmethionine cycle) to synthesize cysteine in 2 sequential enzymatic steps catalyzed by cystathionine β-synthase and cystathionine γ-lyase, respectively (Sbodio et al. 2019). Although mammals have a second independent pathway able to supply cysteine from extracellular cystine (oxidized form of cysteine) in a reaction catalyzed by the cystine reductase TRP14 (Pader et al. 2014), this pathway is not operative in TRXR1 knockout cells because TRP14 is a member of the thioredoxin family that requires TRXR1 and NADPH to be maintained in its reduced active conformation (Jeong et al. 2004). Of note, while the transsulfuration and cystine reduction pathways have been relatively well characterized in mammals (Sbodio et al. 2019; Marti-Andres et al. 2024), very little is known on their functional roles in other model organisms like Drosophila melanogaster or C. elegans.

The nonsense-mediated mRNA decay pathway (NMD) is a surveillance mRNA system that was initially discovered as a mechanism to degrade transcripts that contain premature stop codons but was later found to eliminate aberrant mRNAs that cannot be translated for other different reasons, like unproductively spliced mRNAs or mRNAs with very long 3′-untranslated regions (reviewed in He and Jacobson 2015). Importantly, whole transcriptomic analyses have widened this view by showing that NMD regulates the dynamics and stability of a large proportion of the eukaryotic transcriptome in all organisms (Lelivelt and Culbertson 1999; Mendell et al. 2004; Rehwinkel et al. 2005; Ramani et al. 2009). The core components of the NMD pathway were first identified in yeast and named UPFs (up frameshift proteins) (Leeds et al. 1992), but have been shown to be conserved across metazoan evolution: UPF1 is an ATP-dependent RNA helicase; UPF2 acts as a molecular bridge between UPF1 and UPF3, promoting UPF1 RNA unwinding and helicase activity while UPF3 binds to exon junction complexes (Yi et al. 2021; Monaghan et al. 2023). In C. elegans, the molecular constituents of the NMD pathway were originally discovered as a class of extragenic suppressor genes that, when mutated, share a common phenotype of abnormal morphogenesis of the male bursa and the hermaphrodite vulva, thus named smg (suppressor with morphogenetic effect on genitalia) genes (Hodgkin et al. 1989). This gene class was later found to encode the orthologues of yeast UPF genes: SMG-2/UPF1, SMG-3/UPF2, and SMG-4/UPF3 (Pulak and Anderson 1993; Page et al. 1999; Grimson et al. 2004; Johns et al. 2007). In addition to these core components, there are other proteins that assist for the correct function and regulation of the NMD pathway in all organisms (Chang et al. 2007; Casadio et al. 2015; Monaghan et al. 2023).

In this work, we report that impairment of the NMD pathway suppresses the embryonic lethal phenotype of C. elegans  gsr-1 loss-of-function mutants and that this suppression requires the 2 cystathionine γ-lyase orthologues CTH-1 and CTH-2.

Materials and methods

C. elegans strains

The standard methods used for culturing and maintenance of C. elegans were as previously described (Stiernagle 2006). A list of all strains used and generated in this study is provided in Supplementary Table 1. The alleles gsr-1(syb2363), dpy-11(syb4162), cth-1(syb7115), and cth-2(syb7086) were generated at SunyBiotech (http://www.sunybiotech.com) by CRISPR-Cas9 editing. All VZ strains were 6× outcrossed with N2 wild-type, except those strains generated by CRISPR-Cas9, which were 2× outcrossed. Worm reagents and details on the protocols used for genotyping the different alleles reported in this work can be provided upon request.

Whole genome sequencing single-nucleotide polymorphism mapping

Using the Hawaiian single-nucleotide polymorphism (SNP) mapping method, we backcrossed the GRU102 strain carrying the gnaIs2 transgene (Fong et al. 2016) with the polymorphic C. elegans Hawaiian strain CB4856 (Davis et al. 2005). Next, we isolated the newly generated F2 recombinants homozygous for the gnaIs2 transgene. Total DNA extraction was performed using the Plant/Fungi DNA Isolation Kit (Norgen Biotek Corp). Sequencing libraries were constructed using the NEXTflex Rapid DNA-Seq Kit according to manufacturer’s instructions (Bioo Scientific). DNA quality and integrity were evaluated by Experion Automated Electrophoresis System (Bio-Rad) and the concentration was calculated using qPCR. Libraries were prepared at the Genomic Platform at CIBIR (http://cibir.es/es/plataformas-tecnologicas-y-servicios/genomica-y-bioinformatica) and sequenced on an Illumina HiSeq15000. The quality of DNAseq results was assessed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The FastQ files were analyzed using a Cloud-Based Pipeline for Analysis of Mutant Genome Sequences (Cloudmap tool, https://usegalaxy.org/cloudmap) with standard parameters following Cloudmap workflow (Minevich et al. 2012).

Embryonic arrest phenotype

All experiments were performed on synchronized embryos generated by allowing 10 to 15 gravid hermaphrodites to lay eggs during 2.5 h on seeded plates at 20°C. After parent removal, laid embryos were counted and plates were further incubated at 20°C. During the next 2 days, arrested embryos were counted and the number of arrested embryos at day 2 were used for quantification. Viable progeny were quantified on the plates, which were incubated for 1 or 2 more days (depending on the strain) at days at 20°C.

Larval arrest phenotype

Synchronized animals were generated by allowing 10 to 15 gravid hermaphrodites to lay eggs during 2.5 h on seeded plates at 20°C. After parent removal, laid embryos were counted and further incubated for 4 days at 20°C. For quantification, all larvae were censored and only animals that passed the L4 molt and became young adults were counted.

Recombinant expression and enzymatic characterization of C. elegans TRXR-1

The amino acid sequence encoding the C. elegans thioredoxin reductase 1 (TRXR-1) was retrieved from the GenBank database (Accession No. AAD41826.1). The open reading frame (ORF) was subsequently synthesized by Integrated DNA Technologies with codon optimization for enhanced expression in Escherichia coli. Additionally, an N-terminal hexahistidine-tagged Small Ubiquitin-like Modifier (H6SUMO) fusion was engineered upstream of TRXR-1, and a selenocysteine insertion sequence element was placed downstream. This construct was cloned into the pABC2 vector as previously described (Cheng and Arnér 2017), yielding the plasmid pABC2a-CeTRXR-1, which was subsequently transformed into the E. coli C321.ΔA strain (Lajoie et al. 2013).

For the protein expression, a single bacterial colony was cultured in 10 mL of Terrific Broth (TB) at 30°C with constant shaking at 250rpm overnight. The resultant culture was then scaled up to 2 L of TB, and incubation was continued under the same conditions until the optical density at 600 nm (OD600) reached 1–1.5. At this point, 5 µM sodium selenite and 0.5 mM IPTG were added into the culture, which was then incubated at a reduced temperature of 24°C overnight.

The bacterial cells were harvested by centrifugation at 5,000 × g for 15 min and resuspended in IMAC binding buffer (50 mM Tris-HCl, 250 mM NaCl, and 20 mM imidazole). The cells were then lysed, and the lysate was centrifuged at 30,000 × g for 30 min at 4°C. The resulting supernatant was subjected to affinity chromatography purification. The CeTRXR-1 protein was engineered with an N-terminal His-tagged SUMO tag (H6SUMO). The H6SUMO-CeTRXR-1 was initially purified using IMAC (ÄKTAExplorer 10 FPLC equipped with a 5  mL HisTrap FF column, GE Healthcare). To the eluted fraction containing H6SUMO-CeTRXR-1, 15 µg/mL His-tagged ULP1 (SUMO protease) was added, and the protein mixture was incubated at 4°C overnight with gentle shaking to cleave the CeTRXR-1. Excess imidazole in the digestion mixture was removed using a NAP-25 desalting column (GE Healthcare), and the protein mixture was subjected to a second round of IMAC. During this step, noncleaved H6SUMO- CeTRXR-1, the cleaved H6SUMO tag, His-tagged ULP1, and any impurities from the first round of IMAC bound to the nickel column, while the nontagged CeTRXR-1 was collected in the flow-through. The purified protein was concentrated and stored in TE buffer containing 30% glycerol for long-term storage at −20°C. Protein concentration was determined spectrophotometrically by measuring FAD absorption at 463 nm (ε = 13,600 M⁻¹cm⁻¹, with 1 FAD corresponding to 1 CeTRXR-1 subunit).

Enzyme activity assays were conducted using previously established protocols (Arnér and Holmgren 2005; Cheng and Arnér 2022). Specifically, the NADPH-dependent reduction of 5,5′-dithiobis(2-nitrobenzoic) acid (DTNB) was employed to determine the specific activities of CeTRXR-1. This assay involves the reduction of 1 DTNB molecule to two 5-thio-2-nitrobenzoic acid (TNB) molecules by CeTRXR-1. TNB exhibits a strong absorbance at 412 nm, allowing the reaction velocity to be quantified as µmol DTNB reduced per minute, calculated using the extinction coefficient for TNB (13,600 M⁻¹cm⁻¹). To further verify Sec-dependent activities of CeTRXR-1, the alternative insulin-coupled Trx reduction assay was used. Insulin is a substrate of human Trx1 that can be recycled by CeTRXR-1 in the consumption of NADPH. This assay therefore monitored the consumption of NADPH through the decreased absorbance at 340 nm, using NADPH's extinction coefficient of 6,200 M⁻¹cm⁻¹. For measuring human GR activity, the assay utilized GSSG as the substrate, which was reduced by glutathione reductase using NADPH, that can be monitored similarly.

Microscopy

For protruding vulva phenotype analysis (Fig. 2a), animals were evaluated on the second day of adulthood. For CTH-1::GFP and CTH-2::GFP fluorescence determination in embryos (Fig. 5c and d) an 1 h egg lay was performed and the embryos were allowed to develop 2 more hours at 20°C on the plate to match the stage at which gsr-1 embryos arrest. Next, embryos are transferred to a 3% agarose pad on a microscope slide and imaged. Differential interference contrast and fluorescence imaging was performed in an Olympus BX61 microscope equipped with a DP72 digital camera coupled to CellSens Software for image acquisition and analysis. Photoshop CC 2018 and Adobe Illustrator software were used to produce figures. ImageJ Software was used to quantify the fluorescence of the embryos.

qPCR analysis

To determine the transcriptional activity of cth-1 and cth-2 genes in embryos, young adult worms were collected and bleached, to use only embryos at the early stage of division. This way they can be compared with gsr-1 embryos that arrest between 15 and 120 cells stage (Mora-Lorca et al. 2016). For total RNA extraction, eggs were lysed using mirVana Kit, following manufacturer’s instructions (Ambion). Homogenization of the lysate was performed using a conventional rotor-stator homogenizer polytron, prechilled with liquid nitrogen. For cDNA synthesis, RNA was extracted with DNAse to eliminate any DNA contamination. In each sample, a total reaction of 10 µL contained: 500 ng RNA, 1 µL RQ1 RNase-Free DNAse (Promega), 1 × RQ1 DNAse 10 × Reaction Buffer, and DEPC water. Reactions were incubated for 30 min at 37°C and then stopped by adding 1 µL STOP solution (Promega) and further incubation for 15 min at 65°C. cDNA synthesis was performed using SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen) following instructions for random hexamers primed. cDNA was eluted in TE buffer.

For qRT-PCR analysis Power SYBR Green Master Mix (ThermoFisher Scientific) and specific primers were used in a QuantStudio 5 Real Time PCR System (Applied Biosystems, ThermoFisher). Normalization to actin act-1 expression was used to calculate relative expression. The experiments were carried out in 3 independent replicas. Primers pairs used were (5′->3-): cth-1Fw: ATGACTCCGTACTTCCAGCG; cth-1Rv: TAGATGCTGCTGGAACTCGT; cth-2Fw: GCCGTGTTGCTGTTCCTAAT; cth-2Rv: CCACTGCGGCAATATCAACA; act-1Fw: ACGCCAACACTGTTCTTTCC; act-1Rv: GATGATCTTGATCTTCATGGTTGA.

Western blot analysis

Young adult worms were collected and bleached to obtain embryos at the early stage of cell divisions. Samples were boiled for 10 min in Sample buffer, Laemmli 2 × Concentrate (S3401 Sigma-Aldrich, Merck) to a 500 embryos/µL concentration. Mini-PROTEAN TGX Stain-Free Gels 4–20% (Ref. 4568094, Bio-Rad) were used to separate proteins based on their molecular weight. After electrophoresis, proteins were transfer to a PVDF membrane (Merck Millipore Immobilon®-E PVDF Transfer Membrane, Ref. IEVH85R) for immunodetection. CTH-1::3xFLAG::eGFP and CTH-2::3xFLAG::eGFP proteins were detected using a FLAG antibody (Monoclonal ANTIFLAG M2 Clone M2, F1804, Sigma-Aldrich) while β-actin, used as cargo control, was detected with the actin antibody ((C4): sc-47778, Santa Cruz).

For chemiluminescent detection, ECL Western Blotting Analysis System (Ref. RPN2109, Amersham^TM^) was used according to the manufacturer's instructions. Relative protein abundance was quantified using Image Lab software (version 6.1). Immunodetection images of the blots were uploaded into the software and the intensity of each band was measured, with values assigned proportionally to the signal intensity of the corresponding bands.

MosSCI transgenesis

The eft-3 promoter was PCR amplified from genomic N2 DNA with primers eft-3Fw: 5′-gacctagGCACCTTTGGTCTTTTATTGTC-3′ and eft-3Rv: 5′-ctcctgcaggTGAGCAAAGTGTTTCCCAACTG-3′, digested with AvrII and SbfI and used to replace the hsp-16.41 promoter of pBN209 (Cabianca et al. 2019). In parallel, the sequences of cth-1 and cth-2 were PCR amplified from N2 genomic DNA with the following primers pairs (5′->3′): cth-1Fw: 5′-CTATATCTCCTCTCGTGCAGG-3′; cth-1Rv: 5′-GTTCTCATTTCACCATCGATG-3′; cth-2Fw: 5′-TCTCTCCTCTTTCTCTCACAC-3′; cth-2Rv: 5′-CTCTATGATTCCTTCTACGC-3′ and inserted into the vector pSpark I (Canvax) according to the manufacturer's instructions. Next, the cth-1 and cth-2 fragments were excised with StuI and SbfI and inserted into pBN209-eft-3p digested with NruI and SbfI to generate plasmids pBN570 eft-3p::cth-1 and pBN571 eft-3p::cth-2. Inserts were verified by Sanger sequencing. The transgenes were integrated as single copies into the ttTi5605 (chrII) and oxTi365 (chrV) loci in strains EG4322 and EG8082 by MosSCI (Frokjaer-Jensen et al. 2014).

Eggshell and permeability barrier removal

For eggshell removal, embryos were collected from dissected adult gsr-1 (m+,z−) worms in egg buffer (118 mM NaCl, 48 mM KCl, 2 mM CaCl_2_, 2 mM MgCl_2_, and 25 μM HEPES (pH 7.4)). Developmental stages of collected embryos were not sorted out. Embryos of 2-cell-stage and later-stages were included. The embryos were incubated in 1% NaOCl for 2 min followed by the incubation in chitinase–chymotrypsine solution (2 U/mL chitinase in egg buffer pH 6.0) for 4 min at 22˚C. It is noted that the permeability barrier inside the eggshell still remains around the embryos after this procedure.

For lethality measurements, the eggshell-removed embryos as well as intact gsr-1 (m−, z−) embryos were incubated in 100 μL of egg buffer in a chamber slide. After overnight incubation at 22˚C, unhatched embryos were counted under a stereomicroscope (Leica, M205C). DIC imaging was performed on an optical microscope (Olympus, BX51).

For the permeability barrier integrity assay, embryos were placed into a solution of 5 μg/mL FM4-64 in egg buffer and were mounted on a microscope slide. Fluorescent images were obtained with a confocal microscope (Nikon Ti-E with the spinning disk confocal unit CSU-X1 (Yokogawa electric Corp.)) with a solid-state laser line (561 nm, Andor Technology).

Graphical and statistical analysis

Data were processed in Excel (Microsoft Corporation) then Prism (GraphPad Software) was used to generate bar charts and perform the statistical analyses described in the Figure Legends.

Results

Isolation of a gsr-1 embryonic lethality suppressor

The C. elegans  gsr-1(tm3574) deletion allele encodes truncated versions of mitochondrial (GSR-1a) and cytosolic glutathione reductase (GSR-1b) isoforms that lack part of their respective NADPH and FAD binding domains (Fig. 1a and b) and gsr-1(tm3574) mutants exhibit a fully penetrant embryonic lethal phenotype (Fig. 1c) (Mora-Lorca et al. 2016). When generating a strain combining the gsr-1(tm3574) allele with the integrated transgene gnaIs2 [Pmyo-2::yfp; Punc-119::Aβ1-42] (which expresses human Aβ peptide in all worm neurons along with YFP in pharynx as fluorescence coinjection marker) (Fong et al. 2016), we were surprised to isolate double gsr-1(tm3574); gnaIs2 homozygous animals that were viable and grossly wild-type (Fig. 1c). Since gsr-1; gnaIs2/+ embryos arrested development (Fig. 1c), we concluded that the causative suppressor mutation is recessive. The rescue of the gsr-1(tm3574) embryonic lethal phenotype by the gnaIs2 transgene is not due to YFP or Aβ as the control transgenes gnaIs1 [Pmyo-2::yfp], expressing YFP in pharynx, or dvIs50 [Psnb-1:: Aβ1-42], expressing Aβ in neurons, were unable to restore viability of gsr-1(tm3574) mutants (Fig. 1c). Ten times outcrossing of the gnaIs2 transgene with the wild-type strain maintained the viability of gsr-1(tm3574) mutants, suggesting that the transgene insertion locus or a closely linked mutation was responsible for the suppression of the embryonic lethal phenotype.

mRNA and protein domain organization of wild-type and mutant gsr-1 and identification of the molecular lesion segregating with the gnaIs2 transgene. a) Schematic representation of the 2 gsr-1 mRNA variants. Boxes represent exons and lines show spliced introns. White boxes indicate 5′-UTR and 3′-UTR, respectively, and gray boxes indicate the ORF. Boundaries of gsr-1(tm3574) and gsr-1(syb2343) deletions are shown as black lines. b) Schematic representation of GSR-1 proteins. Protein domain organization of GSR-1a and GSR-1b isoforms as well as those of the shorter proteins resulting from translation of the gsr-1(tm3574) and gsr-1(syb2343) deletion alleles. c) The gnaIs2 transgene suppresses gsr-1 mutants embryonic lethality. gsr-1(tm3574) embryos carrying the gnaIs2 transgene (expressing YFP in pharynx and Aβ in neurons) in homozygosis develop normally. The gnaIs2 transgene in heterozygosity and control transgenes gnaIs1 (expressing YFP in pharynx) and dvIs50 (expressing Aβ in neurons) do not suppress gsr-1 mutants embryonic lethality. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red. d) Diagram of the deletion boundaries at 6.36 MB position of LG IV identified in the strain harboring the gnaIs2 integrated transgene. Genes within this region are represented in boxes where ORF exons are in blue and 3′-UTR exons are in gray, connected by lines representing the introns.

Inactivation of the NMD pathway suppresses the embryonic lethality of C. elegans gsr-1 mutants

Next, to identify the suppressor locus, we performed whole genome sequencing (WGS) using a SNP mapping strategy (Doitsidou et al. 2010). Briefly, the gnaIs2 harboring strain was crossed with the highly polymorphic Hawaiian C. elegans strain CB4856 and the resulting F2 progeny carrying the gnaIs2 transgene in homozygosis was pooled for WGS-SNP analysis. The loss of Hawaiian SNPs identified the chromosomal interval containing the suppressor locus within a wide region of LG IV (Supplementary Fig. 1). A detailed inspection of this interval identified a 4.5 kb deletion, centered at 6,364 MB position of LG IV (Fig. 1d), which harbors 3 genes: Y73B6BL.47 and Y73B6BL.270 of unknown function and smg-3, which encodes the C. elegans orthologue of human UPF2, a core regulator of the NMD pathway (Johns et al. 2007).

C. elegans

• hermaphrodites with mutations in smg genes display a protruding vulva (Pvu) phenotype (Hodgkin et al. 1989). Interestingly, the original strain carrying the gnaIs2 transgene as well as the viable gsr-1(tm3574); gnaIs2 animals exhibit this Pvu phenotype, similar to that of smg-3(ma117) mutants used as control (Fig. 2a), suggesting that animals bearing the gnaIs2 transgene may have impaired smg-3 function. To determine whether defective SMG-3 function allows gsr-1 embryos’ development, we generated double mutants of the gsr-1(tm3574) deletion with 3 independent smg-3 alleles: ma117 whose molecular lesion is unknown and the deletions tm5719 and tm5906. We found that all smg-3 alleles rescued the gsr-1 embryonic lethal phenotype (Fig. 2b).

Impairment of the NMD pathway suppresses the embryonic lethal phenotype of gsr-1 mutants. a) Representative micrographs of first day adult gsr-1(tm3574), gnaIs2, or smg-3(ma117) mutants. All mutants develop similar to wild-type control, except that those carrying the gnaIs2 transgene or smg-3 mutation have a protruding vulva phenotype (Pvu). Insets show the magnification of the vulva area. b–d) smg-1(e1228), smg-2(e2008), and smg-3(ma117) mutations allow normal embryonic development of gsr-1(tm3574) and gsr-1(syb2363) animals. The variable number of arrested embryos in the strains carrying the smg-1, smg-2, and smg-3 alleles is due to the him-2(e1065) or him-5(e1490) (high incidence of males) mutations present in their respective backgrounds (Hodgkin et al. 1979) (See Supplementary Table 1 for complete genotype). In all cases, the number of arrested embryos substantially decreased upon crossing with the gsr-1 mutation, including the smg-1(e1228) allele that is linked to the him-2(e1065) mutation at LG I. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red.

To confirm that impairment of NMD pathway activity is the underlying cause of the gsr-1 embryonic lethal phenotype suppression, we combined the gsr-1(tm3574) deletion with mutations in smg-1 and smg-2 genes that encode other components of the NMD pathway acting upstream smg-3 (Page et al. 1999; Grimson et al. 2004; Johns et al. 2007). As shown in Fig. 2c, both smg-1(e1228) and smg-2(e2008) mutants also restored the viability of gsr-1 mutants. As described above, the gsr-1(tm3574) deletion allele encodes shorter GSR-1 isoforms (both cytoplasmic and mitochondrial) that lack part of the NADPH and FAD binding domains (Fig. 1a and b) and we have shown that this shorter isoforms are devoid of GR enzymatic activity in vitro (Mora-Lorca et al. 2016). Although the smg genes were initially defined as a class of extragenic suppressors (Hodgkin et al. 1989), given the role of NMD in the regulation of the dynamics and stability of many transcripts (Lelivelt and Culbertson 1999; Mendell et al. 2004; Rehwinkel et al. 2005; Ramani et al. 2009), we first set to rule out that the suppression phenotype would arise from a cryptic expression of the gsr-1(tm3574) transcript, producing a shorter GSR-1 protein with residual enzymatic activity in vivo. For this purpose, we generated by CRISPR-Cas9 a new deletion allele, gsr-1(syb2363) that almost completely eliminates the gsr-1 ORF, thus being a putative null allele (Fig. 1a and b), Similar to gsr-1(tm3574) allele, homozygous gsr-1(syb2363) animals also display a fully penetrant embryonic lethal phenotype which is suppressed by mutations in the genes encoding the different components of C. elegans NMD pathway (Fig. 2d), indicating that the suppressor phenotype is extragenic to gsr-1. Collectively, these data demonstrate that impairment of the NMD pathway function suppresses gsr-1 mutant’s embryonic lethality. This is most likely achieved by allowing the stability of 1 or more mRNAs (normally degraded by a functional NMD pathway) encoding components of an alternative system capable of generating enough GSH to sustain gsr-1 embryos development.

The inactivation of the NMD pathway does not suppress the synthetic phenotype of gsr-1(m+,z−); trxr-1 double mutants

Aiming to identify the alternative system that allows the development of gsr-1 embryos in an smg-3 (NMD deficient) background, we first focused on the cytoplasmic thioredoxin reductase TRXR-1 that we have previously shown to function redundantly with GSR-1 in the worm molting cycle (Stenvall et al. 2011). For this purpose, we used 3 different trxr-1 alleles: trxr-1(sv47) is a deletion that removes most of trxr-1 ORF and is probably a null allele, whereas trxr-1(cer34[Sec666Cys]) and trxr-1(cer55[Sec666])* are 2 different point mutation alleles that eliminate the selenocysteine residue of TRXR-1, required for its enzymatic activity (Stenvall et al. 2011; Garcia-Rodriguez et al. 2018). Because smg-3 maps to LG IV, as does trxr-1, we decided to investigate the role of trxr-1 in gsr-1 mutants development when the NMD pathway is impaired using instead the smg-2(e2008) allele, which is equally effective in inactivating the pathway (Fig. 2c). In all cases, the smg-2 mutation did not suppress the synthetic defect of gsr-1(m+,z−); trxr-1 double mutants that produce sick animals with a pale, sluggish phenotype and defective ecdysis (Fig. 3). Importantly, the very few gsr-1(m+,z−); trxr-1; smg-2 survivors that managed to overcome the 4 larval molts became adults that did not differ from gsr-1(m+,z−); trxr-1 controls, all showing a severe sick appearance and producing none or very few embryos (1 to 10 per parent) that did not develop. Taken together, the fact that the highly detrimental phenotypes of gsr-1(m+,z−); trxr-1 animals cannot be suppressed by impairment of the NMD pathway, precludes any conclusion on the possible role of TRXR-1 on the rescue of gsr-1(m−, z−) embryonic arrest phenotype.

The smg-2(e2008) mutation fails to suppress the molting phenotype of gsr-1; trxr-1 double mutants. All triple mutants gsr-1; smg-2; trxr-1 display a highly penetrant molting larval arrest phenotype, similar to that of the control strain gsr-1; trxr-1. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red.

The GSSG-reducing activity of the thioredoxin system is dispensable for the growth of gsr-1; smg-3 embryos

Because of the limitations imposed by the deleterious phenotype of gsr-1(m+,z−); trxr-1 double and gsr-1(m+,z−); trxr-1; smg-2 triple mutants and to elucidate whether the thioredoxin system is involved in the suppression of the gsr-1(m−, z−) embryonic arrest exerted by impairment of the NMD pathway, we next aimed to determine if C. elegans  TRXR-1 exhibits GR enzymatic activity in vitro. Despite metazoan thioredoxin reductases and glutathione reductases share domain organization and functional groups (Arner 2022) (except the additional C-terminal active site motif present in thioredoxin reductases mentioned above; Zhong et al. 1998), no metazoan thioredoxin reductase has been shown to reduce GSSG in vitro. Consistently, this is also the case for C. elegans  TRXR-1, which we found unable to reduce GSSG in the presence of NADPH (Fig. 4a and Supplementary Fig. 2). Having ruled out direct GSSG reduction by TRXR-1, we next addressed whether development of gsr-1; smg-3 embryos could be rescued by one of the nematode thioredoxin family members that have previously been shown to interact with the glutathione system in other organisms. We first focused on the thioredoxin member DPY-11 because is expressed in the hypodermis (where GSR-1 and TRXR-1 display redundant functions in the molting cycle) (Ko and Chow 2002) and its mammalian orthologue TMX1 has been reported to use glutathione for its catalytic activity (Matsuo and Hirota 2017). However, 2 different putative dpy-11 null alleles (e207 and sy748) (Ko and Chow 2002; Chiu et al. 2013) and a newly generated CRISPR-Cas9 allele (syb4162[C51S;C54S]) that encodes a redox-dead DPY-11 variant, all failed to restore embryonic lethality in gsr-1; smg-3 mutants (Fig. 4b). To our surprise, the redox-dead allele dpy-11(syb4162) does not cause any dumpy phenotype (Supplementary Fig. 3), indicating that DPY-11 enzymatic redox activity is dispensable for its biological function in maintaining body morphology, in contrast to what has been previously proposed (Ko and Chow 2002). Like dpy-11 mutants, null or loss-of-function alleles of trx-1, txdc-17, and txl-1 genes, whose mammalian orthologs encode the thioredoxin proteins with GSSG-reducing activity in vitro TRX-1 (Gromer et al. 2002), TRP14/TXNDC17 (Pader et al. 2014) and TXNL1/TRP32 (Andor et al. 2023) had not effect on the development of gsr-1; smg-3 mutants (Fig. 4c). Together, these data suggest that the thioredoxin system is not directly involved in the mechanism by which impairment of the NMD pathway suppresses the embryonic lethality of gsr-1 mutants. However, given the high number of thioredoxin family members in C. elegans (Johnston and Ebert 2012) and the possibility of a functional redundancy in redox-dependent reactions, our data cannot completely exclude that 1 or more yet unidentified TRXR-1-dependent thioredoxins may play a role in the viability of gsr-1; smg-3 animals.

The members of the thioredoxin system with GSSG-reducing activity are not required for the suppression of gsr-1 embryonic lethality upon impairment of the NMD pathway. a) C. elegans  TRXR-1 is devoid of GSSG-reducing activity in vitro. NADPH consumption in the presence of 1 mM GSSG of recombinantly expressed human glutathione reductase (Hs GR) and C. elegans thioredoxin reductase 1 (Ce  TRXR-1). b, c) The thioredoxins DPY-11, TRX-1, TXDC-17, and TXL-1 are not required for the embryonic development of gsr-1; smg-3 animals. All triple mutant combinations of genes encoding thioredoxin proteins with GSSG-reducing activity in a gsr-1; smg-3 double mutant background develop similarly as gsr-1; smg-3 control animals. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red.

The transsulfuration pathway supports development of gsr-1; smg-3 double mutants

We have previously shown that C. elegans expressing aggregation-prone proteins in muscle cells are extremely sensitive to GSH depletion (Guerrero-Gomez et al. 2019) and that these animals rely on functional transsulfuration and cystine reduction pathways for survival, most likely by provisioning cysteine for de novo GSH synthesis (Marti-Andres et al. 2024). Cystine reduction is performed by TRXR-1 and TRP14/TXDC-17 in both mammals and C. elegans (Marti-Andres et al. 2024). However, our data indicate that TRXR-1 and TXDC-17 are not required for viability of gsr-1; smg-2 or gsr-1; smg-3 mutants (Figs. 3 and 4c). Hence, we asked whether the transsulfuration pathway (Fig. 5a) would suffice to supply the necessary cysteine for de novo GSH synthesis to allow development of gsr-1; smg-3 embryos. To test this hypothesis, we employed animals that carry loss-of-function mutations in the cth-1 and cth-2 genes, encoding 2 paralogues of the enzyme cystathionine γ-lyase that catalyzes the last step in the transsulfuration pathway, converting cystathionine into cysteine and α-ketobutyrate (Fig. 5a) (Sbodio et al. 2019). Similar to txdc-17 mutants, worms bearing mutations either in the cth-1 or cth-2 genes did not impair the development of gsr-1; smg-3 embryos (Fig. 5b). However, gsr-1; smg-3; cth-1; cth-2 quadruple mutant worms produced about 80% of arrested embryos (Fig. 5b), demonstrating a functional redundancy of CTH-1 and CTH-2 in allowing gsr-1; smg-3 embryos growth and suggesting that cystathionine to cysteine conversion through transsulfuration is the main pathway responsible for sustaining development of gsr-1; smg-3 embryos. To investigate the molecular mechanism by which 20% of gsr-1; smg-3; cth-1; cth-2 quadruple mutant embryos were still viable, we generated a gsr-1; smg-3; txdc-17; cth-1; cth-2 quintuple mutant that increased the number of arrested embryos to 93% as compared to the 80% of the gsr-1; smg-3; cth-1; cth-2 quadruple mutant (Fig. 5b), indicating that some cysteine provisioning through the cystine reduction pathway can occur in the absence of a functional transsulfuration pathway, but that it is not enough on its own to allow gsr-1 embryos survival. The fact that a 7% of gsr-1; smg-3; txdc-17; cth-1; cth-2 quintuple mutant embryos are able to develop suggests that a residual cystine reductase activity independent of txdc-17 may exist.

Inactivation of the transsulfuration pathway impairs gsr-1; smg-3 double mutants development. a) Reverse transsulfuration pathway. Human proteins are depicted in red and C. elegans proteins in blue. CBS/CBS-1/CBS-2, cystathionine-β-synthase; CGL/CTH-1/CTH-2, cystathionine-γ-lyase. b) Simultaneous inactivation of cth-1 and cth-2 genes restores the embryonic lethal phenotype in gsr-1; smg-3 double mutants. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates) with at least 100 embryos laid per plate. Strains with the wild-type gsr-1 allele are depicted in blue and strains with the gsr-1 mutant allele are depicted in red. c) cth-1 and cth-2 mRNA levels in gsr-1 and smg-3 mutants. Data are the mean +/− SEM of 3 independent experiments (3 biological replicates). Data were not significantly different by one-way ANOVA multiple comparisons test. d, e) CTH-1 and CTH-2 protein levels in gsr-1 and smg-3 mutants. Differential interference contrast (left panel) and fluorescence (right panel) microscopy images and quantification of CTH-1::GFP and CTH-2::GFP endogenous reporters in gsr-1 and smg-3 mutants. Graphs represent the data of 3 independent experiments with at least 30 embryos. *** P < 0.001 by Kruskal–Willis with Dunn’s multiple comparisons test. Error bars are SEM.

As the NMD pathway regulates the dynamics and stability of part of the eukaryotic transcriptome, we reasoned that the smg-3 mutation may stabilize or increase the levels of cth-1 and cth-2 mRNAs, probably resulting in higher levels of protein expression, thus allowing the development of gsr-1 embryos. To test this hypothesis, we first quantified cth-1 and cth-2 transcript levels in embryos of the different genetic backgrounds by qPCR. As shown in Fig. 5c, the smg-3 mutation did not significantly increase the amount of cth-1 and cth-2 mRNAs.

Next, to address if the smg-3 mutation would increase CTH-1 and CTH-2 proteins amount without altering their corresponding mRNA levels, we generated by CRISPR-Cas9 transgenic strains that produce CTH-1::GFP and CTH-2::GFP fusion proteins, expressed from their respective endogenous promoters (Supplementary Table 1). Using these endogenous GFP reporters we found that the smg-3 mutation only slightly increased the levels of CTH-1::GFP and had no effect on CTH-2::GFP (Fig. 5d and e). In contrast, in a gsr-1; smg-3 double mutant background CTH-1::GFP levels remained similar to those found in smg-3 single mutants while CTH-2::GFP levels were significantly increased (Fig. 5d and e). Finally, gsr-1 mutant embryos slightly decrease CTH-1::GFP while increased CTH-2::GFP levels (Fig. 5d and e). To rule out that changes of redox potential in gsr-1 mutant backgrounds could interfere on GFP fluorescence emission (Inouye and Tsuji 1994) we performed western blot analysis, which confirmed that CTH-1::GFP and CTH-2::GFP protein levels roughly correlate with fluorescence quantifications (Supplementary Fig. 4).

Taken together, these data indicate that gsr-1 embryos do not substantially modify cth-1 and cth-2 mRNA levels but induce an increase of mainly CTH-2 protein to compensate for the lack of GSSG-reducing capacity, which could explain how gsr-1 embryos bypass the pregastrula stage checkpoint in a NMD-defective genetic background (Mora-Lorca et al. 2016).

Discussion

Glutathione is the most prevalent thiol-based redox metabolite in virtually all organisms. Few exceptions are the low molecular weight thiols bacillithiol or mycothiol that serve similar functions in some Gram-positive bacteria, ergothioneine in some fungi or trypanothionine in euglenozoa (Poole 2015). Inactivating mutations in genes encoding the first step in glutathione synthesis enzymes are lethal in all those organisms that use glutathione (Wu and Moye-Rowley 1994; Dalton et al. 2000; Shi et al. 2000; Romero-Aristizabal et al. 2014; Enya et al. 2017). In contrast, mutations in the gene encoding glutathione reductase, the enzyme that recycles oxidized glutathione to its reduced form, do not always result in lethal phenotypes. For instance, human patients with mutations in the GSR gene can reach old age, although they suffer hemolytic anemia, cataracts, deafness, and hyperbilirubinemia (Loos et al. 1976; Kamerbeek et al. 2007). GSR knockout mice do not show any phenotype under stabulary conditions (Rogers et al. 2004) but are highly sensitive to bacterial and fungal infections (Yan et al. 2013; Kim et al. 2019). Similarly, putative null GSR zebrafish develop similarly to wild-type control animals (Zhao et al. 2020). The viability of vertebrate GSR mutants has been explained by the thioredoxin system acting as a backup pathway to reduce GSSG (Sun et al. 2001; Tan et al. 2010), although the embryonic lethality of TXNRD1 and TXN1 knockout mice (Matsui et al. 1996; Jakupoglu et al. 2005) has hampered the experimental demonstration with double knockout mice combinations. Consistent with the possibility of the thioredoxin system acting as an alternative mechanism to maintain the GSH pool, Drosophila melanogaster and probably other insects do not possess glutathione reductase and GSSG recycling is carried out by dedicated thioredoxin reductase and thioredoxin proteins (Kanzok et al. 2001). Also, yeast lacking glr1, which encodes glutathione reductase, have an absolute requirement of the thioredoxin system for survival (Muller 1996).

Contrary to the examples above, C. elegans  gsr-1 mutants are embryonic lethal (Mora-Lorca et al. 2016), despite having a complete thioredoxin system (Johnston and Ebert 2012). Why C. elegans  gsr-1 mutants are the only eukaryotes that have an embryonic lethal phenotype even when TRXR-1 can functionally substitute GSR-1 in the worm molting cycle (Stenvall et al. 2011)? One possibility is that the trxr-1 gene is not expressed at enough levels to compensate for the lack of gsr-1 during the first embryonic divisions, as the gsr-1 embryos arrest at the pregastrula stage (Mora-Lorca et al. 2016). This is consistent with the fact that trxr-1 gene expression gradually increases during the first 200 min of embryonic development (Hutter and Suh 2016). Alternatively, mammalian cells with strong impairment of GR activity excrete GSSG to the extracellular medium to alleviate its toxic intracellular built-up (Eklow et al. 1984). Should this mechanism also operate in C. elegans embryonic cells, the constrain imposed by the embryo eggshell could hamper the survival of the gsr-1 embryos by avoiding GSSG dilution in the extracellular milieu. To test this hypothesis, we eliminated the eggshell of wild-type and gsr-1 embryos by chitinase treatment and found that gsr-1 embryos lacking eggshell still displayed an embryonic lethal phenotype while wild-type embryos develop normally. Because gsr-1 embryos lacking eggshell maintained the integrity of the permeability barrier, GSSG could still accumulate within the peri-embryonic. However, removing the embryo permeability barrier to allow GSSG diffusion provokes wild-type embryos death, thus precluding any conclusion in this direction (Supplementary Fig. 5).

The serendipitous finding of the inactivation of the NMD pathway allowing the development of gsr-1(m−, z−) embryos suggested that expression or stabilization of 1 or more unknown mRNAs as the cause of the suppression of the embryonic lethal phenotype. Given that the thioredoxin system has been shown to act as backup system to reduce GSSG in the absence of glutathione reductase (Muller 1996; Sun et al. 2001; Tan et al. 2010), and that C. elegans  TRXR-1 cooperates with GSR-1 in the worm molting cycle (Stenvall et al. 2011), we first approached to evaluate if the thioredoxin system was behind the suppression of the gsr-1 embryos lethal phenotype by smg-1, smg-2 or smg-3 mutations. To our surprise, mutations in the trxr-1 gene encoding cytoplasmic thioredoxin reductase as well as in dpy-11, trx-1 and txdc-17 genes, encoding the orthologues of those thioredoxins that have been reported in other organisms to reduce GSSG, failed to restore embryonic arrest in gsr-1; smg-2 or gsr-1; smg-3 double mutant backgrounds. Thus, the worm thioredoxin system was not directly involved in the development of gsr-1 embryos with inactivated NMD pathway function.

Mouse hepatocytes that lack cytoplasmic thioredoxin reductase TRXR1 and glutathione reductase GSR can grow in vitro as long as methionine is supplied to allow cysteine production for de novo GSH synthesis by the reverse transsulfuration pathway (Eriksson et al. 2015). In this pathway, methionine enters the S-adenosylmethionine cycle to generate homocysteine, whose accumulation is toxic in both mammals and worms (Watanabe et al. 1995; Vozdek et al. 2012). Homocysteine is then converted into cysteine by a 2-step enzymatic process: first homocysteine is used by cystathionine β-synthase to generate cystathionine and subsequently, cystathionine γ-lyase converts cystathionine into cysteine and a- α-ketobutyrate (Sbodio et al. 2019). While in mammals there is only 1 enzyme of each class, C. elegans has 2 cystathionine β-synthases CBS-1 and CBS-2 (Vozdek et al. 2012; Santonicola et al. 2020) and 2 cystathionine γ-lyases CTH-1 and CTH-2 (Warnhoff and Ruvkun 2019). To evaluate whether the activity of the transsulfuration pathway underlays the viability of gsr-1; smg-3 embryos we combined this double mutant with cth-1 and/or cth-2 mutants and found that only the quadruple mutant gsr-1; smg-3  cth-1; cth-2 restored the embryonic lethality. This result implies that NMD pathway inactivation somehow stabilizes the activity of the reverse transsulfuration pathway to a level that counteracts the absence of GSR-1 to reduce GSSG, probably providing enough cysteine for de novo GSH synthesis. The dual requirement of cth-1 and cth-2 genes in the survival of gsr-1 embryos is in sharp contrast with previous data demonstrating nonredundant functions of C. elegans  cth-1 and cth-2 in molybdenum cofactor transfer from bacteria to worms (Warnhoff and Ruvkun 2019), nematode longevity (Statzer et al. 2022), or muscle proteostasis (Marti-Andres et al. 2024) where only cth-2 is required. In this last scenario, cystine reduction pathway via TRP14/TXDC-17 and transsulfuration pathway via CTH-2 cooperate to maintain worm viability when muscle proteostasis is compromised (Marti-Andres et al. 2024). To the contrary, the viability of gsr-1; smg-3 embryos is only dependent on the transsulfuration pathway that requires both CTH-1 and CTH-2. This difference can be explained by the fact that cystine is incorporated from the extracellular environment and therefore, given the impermeability of C. elegans embryos eggshell (Olson et al. 2012), cystine cannot be imported into the embryonic cells, thus relying exclusively on cysteine provision by the transsulfuration pathway.

In summary, we have found that in C. elegans, like in mammals, the reverse transsulfuration pathway is able, under certain circumstances, to allow survival of gsr-1(m−, z−) embryos, probably by supplying cysteine for de novo GSH synthesis when GSSG reduction is compromised. Why this mechanism is not operative in gsr-1 mutants is intriguing and may be related to low expression levels or tissue specificity of CTH-1 and/or CTH-2 proteins during the early embryo development, thus causing embryonic lethality by accumulation of the toxic metabolite homocysteine and GSSG.

Limitations of the study

To further support that increased levels of CTH-1 and CTH-2 can rescue the lethal phenotype of gsr-1(m−, z−) embryos, we overexpressed both proteins under the control of the eft-3 promoter using the Mos1-mediated single copy insertion (MosSCI) system (Frokjaer-Jensen et al. 2014). However, neither overexpression of CTH-1 or CTH-2 separately nor simultaneously allowed the development of gsr-1(m−, z−) embryos (See Supplementary Table 1 for strain information). The most likely explanation is that a functional NMD pathway in gsr-1(m−, z−) embryos inhibits the overexpression of these 2 proteins, probably at the translational level, and only when the NMD pathway is impaired, like in smg mutant backgrounds, the gsr-1(m−, z−) embryonic arrest is bypassed. Alternatively, we cannot rule out that the eft-3 promoter used for the overexpression strains may not provide the correct developmental timing or expression level required to overcome the embryonic arrest checkpoint.

Supplementary Material

jkaf102_Supplementary_Data