Development of a mouse model expressing a bifunctional glutathione-synthesizing enzyme to study glutathione limitation in vivo

Glutathione (GSH) is a highly abundant tripeptide thiol that performs diverse protective and biosynthetic functions in cells. While changes in GSH availability are associated with inborn errors of metabolism, cancer, and neurodegenerative disorders, studying the limiting role of GSH in physiology and disease has been challenging due to its tight regulation. To address this, we generated cell and mouse models that express a bifunctional glutathione-synthesizing enzyme from Streptococcus thermophilus (GshF), which possesses both glutamate-cysteine ligase and glutathione synthase activities. GshF expression allows efficient production of GSH in the cytosol and mitochondria and prevents cell death in response to GSH depletion, but not ferroptosis induction, indicating that GSH is not a limiting factor under lipid peroxidation. CRISPR screens using engineered enzymes further revealed genes required for cell proliferation under cellular and mitochondrial GSH depletion. Among these, we identified the glutamate-cysteine ligase modifier subunit, GCLM, as a requirement for cellular sensitivity to buthionine sulfoximine, a glutathione synthesis inhibitor. Finally, GshF expression in mice is embryonically lethal but sustains postnatal viability when restricted to adulthood. Overall, our work identifies a conditional mouse model to investigate the limiting role of GSH in physiology and disease.

Glutathione (GSH) is a highly abundant tripeptide thiol that performs diverse protective and biosynthetic functions in cells.While changes in GSH availability are associated with inborn errors of metabolism, cancer, and neurodegenerative disorders, studying the limiting role of GSH in physiology and disease has been challenging due to its tight regulation.To address this, we generated cell and mouse models that express a bifunctional glutathione-synthesizing enzyme from Streptococcus thermophilus (GshF), which possesses both glutamate-cysteine ligase and glutathione synthase activities.GshF expression allows efficient production of GSH in the cytosol and mitochondria and prevents cell death in response to GSH depletion, but not ferroptosis induction, indicating that GSH is not a limiting factor under lipid peroxidation.CRISPR screens using engineered enzymes further revealed genes required for cell proliferation under cellular and mitochondrial GSH depletion.Among these, we identified the glutamate-cysteine ligase modifier subunit, GCLM, as a requirement for cellular sensitivity to buthionine sulfoximine, a glutathione synthesis inhibitor.Finally, GshF expression in mice is embryonically lethal but sustains postnatal viability when restricted to adulthood.Overall, our work identifies a conditional mouse model to investigate the limiting role of GSH in physiology and disease.
Glutathione (GSH) is the predominant antioxidant in most aerobic species (1,2).In mammals, it is synthesized in the cytosol in a two-step process involving glutamate and cysteine ligation followed by condensation with glycine.In addition to its antioxidant function, GSH also plays key roles in disulfide bond formation, protein regulation, and iron-sulfur cluster biosynthesis, among others (3)(4)(5)(6).Consistent with these essential functions, GSH synthesis is required for mammalian embryonic development (4, 7).While GSH depletion is associated with diseases such as type II diabetes, non-alcoholic fatty liver disease, and Parkinson's disease, GSH levels are frequently upregulated in cancer (4, 8-13).However, whether GSH dysregulation is causative or symptomatic of disease progression in these diverse settings remains poorly understood.
Genetic and pharmacological inhibition of GSH synthesis has provided insight into the consequences of GSH depletion, but it has been challenging to study the limiting role of GSH abundance in disease models (14,15).Current models are limited by the complex regulation of GSH abundance at the cellular level.The rate-limiting enzyme, glutamate-cysteine ligase (GCL), is subject to feedback inhibition by GSH itself as well as cysteine availability.Additionally, both steps of GSH synthesis are regulated transcriptionally and posttranscriptionally.GSH is not readily imported into cells and rapidly cleared from the serum (16)(17)(18).As a result, methods to supplement GSH availability in disease settings have been limited to global provision of cell-permeable GSH esters or GSH precursors, which require functional GSH synthesis and can act as antioxidants on their own (4, 19,20).Activation of the nuclear factor erythroid 2-related factor 2-Kelch-like ECH-associated protein 1 axis to upregulate expression of enzymes in the GSH synthesis pathway also impacts other downstream effectors involved in the cellular antioxidant response (21)(22)(23).Finally, none of these methods can discern whether subcellular GSH pools have differential roles in disease progression.In recent years, engineered bacterial enzymes such as NDI1 and LbNOX have increasingly been used to interrogate the role of redox metabolism in complex biological systems (24,25).Here, we demonstrate that bacterial enzymes can similarly be used to manipulate compartment-specific GSH pools in cells and mouse models.

An engineered bacterial enzyme to study oxidative cell death
Mammalian GSH synthesis is a two-step process regulated transcriptionally, posttranscriptionally, and by feedback inhibition on the rate-limiting enzyme, GCL (Fig. 1A).Conversely, some bacterial species express a single bifunctional enzyme, GshF, that possesses both GCL and glutathione synthetase activities (Fig. 1A).In particular, the GshF enzyme expressed in Streptococcus thermophilus is not subject to feedback inhibition, allowing rapid accumulation of GSH when expressed in bacterial and mammalian systems (26,27).To evaluate the use of this enzyme in manipulating compartment-specific GSH pools, we transduced HEK-293T cells with either untargeted or mitochondria-targeted, human codon-optimized GshF, henceforth referred to respectively as GshF and mito-GshF.We confirmed mito-GshF localization in the mitochondria by cell fractionation (Fig. 1B).
To determine the efficacy of GshF-mediated GSH synthesis, we quantified GSH abundance from HEK-293T cells expressing either GshF or mito-GshF in the presence or absence of the mammalian GSH synthesis inhibitor buthionine sulfoximine (BSO) (Fig. S1A).Expression of either GshF or mito-GshF increased GSH abundance in cells grown in standard medium and maintained cellular GSH levels in the presence of BSO (Figs. 1, C and D and S1, B and C).Mito-GshF expression was also sufficient to rescue mitochondrial GSH abundance after treatment with BSO (Fig. 1D).Both GshF-and mito-GshF expression prevented BSO-induced cell death up to millimolar amounts of drug treatment in a panel of human and mouse cell lines (Fig. 1E).
The central line of defense against lipid peroxide accumulation is GPX4, which converts lipid peroxides to lipid alcohols at the expense of GSH (Fig. S1A) (28).Consequently, GSH depletion can induce ferroptosis, a programmed cell death pathway caused by accumulation of lipid peroxides (28)(29)(30)(31)(32).We asked whether GSH availability is limiting during ferroptosis in mammalian cells.To address this, we treated GshFand mito-GshF-expressing cells with oxidizers and ferroptosis inducers (Fig. S1A).Surprisingly, GshF and mito-GshF expression failed to protect cells from hydrogen peroxide or paraquat exposure, which both induce reactive oxygen species (Figs.1F and S1, A and D).GshF-and mito-GshF-expressing cells also remained sensitive to RSL3-mediated inhibition of GPX4, which repairs lipid peroxide to prevent ferroptosis (Figs.1G and S1A).However, GshF expression protected cells from iron overload induced by ferric ammonium chloride (Figs.1H and S1A).To further interrogate GSH antioxidant activity, we inhibited the thioredoxin pathway with auranofin.Interestingly, GshF expression enabled auranofin resistance, but mito-GshF expression did not, emphasizing the segregation between the two glutathione pools (Fig. S1E).These data suggest that increased GSH availability is not sufficient to prevent ferroptosis and is not a limiting factor in this process.

Targeted CRISPR-Cas9 screens upon compartment-specific GSH limitation
To investigate compartment-specific vulnerabilities in cells with uncoupled mitochondrial and cytosolic GSH pools, we performed a metabolism-targeted CRISPR-Cas9 screen in Jurkat cells expressing GshF or mito-GshF in which the recently identified mitochondrial GSH transporter, SLC25A39, was knocked out (Figs. 2A and S2, A and B) (26).Cells were grown in standard RPMI or treated with a dose of BSO lethal to SLC25A39-knockout cells complemented with a guideresistant SLC25A39 complementary DNA (cDNA) (Fig. 2B).Both GshF-and mito-GshF-expressing cells were sensitive to depletion of genes involved in ferroptosis and iron metabolism, confirming that GSH is not limiting for ferroptosis (Fig. 2C).Additionally, neither GshF nor mito-GshF expression was sufficient to rescue superoxide dismutase (SOD1) depletion or defects in iron sulfur cluster biosynthesis resulting from cysteine desulfurase (NFS1) knockout (Fig. 2C).As expected, SLC25A40, a paralog mitochondrial GSH transporter, was the top hit in GshF-expressing SLC25A39-KO cells compared to those expressing mito-GshF irrespective of BSO treatment (Figs. 2C and S2C) (26).Under BSO treatment, SLC25A39-KO cells expressing GshF or mito-GshF should experience opposite phenomena: the GSH content in GshF-expressing cells should be unchanged throughout the cell, but mito-GshFexpressing cells likely only retain GSH abundance in the mitochondria.Consistent with this hypothesis, the genes essential in GshF-expressing cells were largely related to lipid metabolism and ferroptosis (Fig. S2D).However, the genes most essential in cells expressing mito-GshF were largely related to oxidative phosphorylation, indicating that cells with high mitochondrial GSH are still sensitive to mitochondrial dysfunction (Fig. S2E).Together, these data indicate that cytosolic and mitochondrial GSH fulfill distinct essential roles.
Surprisingly, loss of the nonessential subunit of GCL, GCLM, enabled both cell lines to proliferate in the presence of BSO, evidenced by an enrichment of sgRNAs targeting GCLM (Fig. 2D).GCLM lowers the K m of GCL for glutamate and increases the K i of GSH but is not required for catalytic function or survival, despite resulting in the production of 90% less glutathione when knocked out (23,33,34).To determine whether loss of GCLM could be a generalizable mechanism of BSO resistance, we generated GCLM-knockouts in WT HEK-293T and Jurkat cells.Indeed, loss of GCLM conferred resistance to BSO, and complementation of GCLM-KO cells with a guide-resistant GCLM cDNA attenuated BSO resistance (Figs.2E and S3A).We performed metabolomics analysis on GCLM-KO cells treated with BSO and confirmed that Gclmknockouts, but not those complemented with GCLM cDNA, maintained their GSH abundance in the presence of BSO (Fig. 2F).To determine whether GCLM-KO cells continued to synthesize glutathione in the presence of BSO, we measured cystine incorporation into GSH after pretreating cells with BSO for 24 h (Fig. 2G).GCLM-KO cells, but not those complemented with GCLM cDNA, were able to incorporate 13 C 6 , 15 N 2 -L-Cystine into glutathione even after pretreatment with BSO for 24 h (Figs.2H and S3B).Altogether, these data show that loss of GCLM enables resistance to BSO.

Increased GSH availability is incompatible with early mouse development
Increasing glutathione pools in vivo has been challenging due to the complex regulation of GSH synthesis.GSH is not orally available, and supplemented GSH is cleared from the bloodstream within min (18,35,36).Additionally, genetic Complementation of glutathione synthesis methods to increase GSH are limited to manipulation of multifaceted antioxidant response pathways, further complicating interrogation of GSH sufficiency (21)(22)(23).To determine whether GshF expression could increase GSH abundance in vivo, we generated genetically engineered mice by inserting either GshF or mito-GshF under control of a STOP codon flanked by two LoxP sites at the Rosa26 locus (Figs.3A and  S4A).To increase GSH availability in all tissues, we mated heterozygous floxed mice to a CMV-Cre model.CMV-mito-GshF Δ/0 pups were born at expected ratios and appeared  GshF Δ/0 of the total litter at indicated developmental stages.Mendelian genetics predict 50%.C, immunoblots of indicated tissues from littermates treated with 75 mg/kg of tamoxifen for five consecutive days.D, abundance of indicated metabolites in each tissue from male ER T2 -GshF Δ/Δ or GshF fl/fl littermates treated with 75 mg/kg of tamoxifen for five consecutive days.For clarity, only p values < 0.05 are displayed on the charts.E, metabolomics analysis of indicated plasma metabolites from male ER T2 -GshF Δ/Δ or GshF fl/fl littermates before and after tamoxifen induction.F, GSH abundance in livers from ER T2 -GshF Δ/0 or ER T2 -GshF 0/0 mice treated with 300 mg/kg acetaminophen (APAP) for 3 h.healthy but did not exhibit increased GSH concentration in any tissues despite expressing mito-GshF (Fig. S4, B-D).Conversely, CMV-GshF Δ/0 mice were greatly underrepresented at P0, and the few CMV-GshF Δ/0 pups born on P0 died within 12 h of birth (Figs.3B and S5A).We assessed the presence of CMV-GshF Δ/0 embryos at several stages and found a general trend of decreasing numbers of CMV-GshF Δ/0 embryos as development proceeded (Fig. 3B).We examined CMV-GshF Δ/0 and their CMV-GshF 0/0 littermates at E17.5 and found that CMV-GshF Δ/0 embryos were slightly shorter than their CMV-GshF 0/0 littermates but did not observe histologic differences in any of the tissues (Fig. S5, B-D).These results indicate that GSH homeostasis is essential in early development and that increased GSH levels during embryogenesis are not compatible with life.

Mice with constitutively high glutathione levels are viable
To examine the impact of GSH accumulation in adult mice, we mated GshF fl/0 mice to a tamoxifen-inducible ER T2 -Cre model and induced GshF expression once the mice were 4 weeks old.The body and tissue weights of ER T2 -GshF Δ/0 mice were comparable to their ER T2 -GshF fl/0 littermates, and we did not observe any signs of distress or death for 4 months (Fig. S6, A and B).We confirmed that GshF was indeed expressed in the tissues of ER T2 -GshF Δ/0 mice by Western blot (Fig. 3C).To determine the effect of GshF expression on tissue metabolism, we analyzed plasma and tissue metabolites from ER T2 -GshF Δ/Δ and their GshF fl/fl littermates and found that in every tissue except the liver and plasma, GSH was more abundant in mice expressing GshF (Figs. 3, D and E and S6C).The extent to which GSH was increased in each tissue varied, with the kidney and the lung exhibiting the greatest increase in GSH abundance compared to those from GshF-null mice.Importantly, the abundance of amino acids in GshF-expressing tissues was largely comparable to those in GshF-null tissue, including the precursors of GSH synthesis, glutamate and glycine (Fig. 3D).In humans, liver GSH is rapidly depleted in acetaminophen (APAP) overdose, and provision of the GSH precursor N-acetylcysteine is the current standard of care (37).To determine whether GshF was active in the liver, we injected mice with 300 mg/kg APAP and quantified liver GSH from GshF-expressing and GshF-null littermates (37,38).GshFexpressing mice had about three times as much GSH in their liver 3 h after APAP treatment (Fig. 3F).These results indicate that increased GSH availability is compatible with life and opens up the potential to examine GSH sufficiency in a variety of contexts.
Initially, we characterized the effects of high GSH on cell populations where GSH synthesis is required for cell development or activity, such as blood cells and T cells.GSH is required for iron-sulfur cluster biosynthesis and defects in GSH synthesis or GSH import into mitochondria can cause anemia (6).To determine whether high GSH could impact blood cell development, we performed complete blood counts on GshF-expressing mice.There were very few differences between the GshF-expressing mice and their GshF-null littermates, with the only significant difference being in the mean corpuscular hemoglobin concentration, a measure of hemoglobin relative to the size of the cell (Fig. S6D).GSH synthesis is also required for T cell activation (39,40).To determine whether GshF expression impacted immune cell development, we analyzed the T cell populations in the spleen, thymus, lungs, mesenteric lymph nodes, and colon of ER T2 -GshF Δ/Δ and GshF fl/fl littermates (Fig. S7, A-J).There were no significant differences in any of the immune populations between ER T2 -GshF Δ/Δ and GshF fl/fl mice, although we observed several nonsignificant trends in the colon, particularly in the lamina propria (Fig. S7, A-J).The only trend we observed across tissues was a slight increase in TCR γδ cells in ER T2 -GshF Δ/Δ mice (Fig. S7J).
To unbiasedly investigate the effect of high GSH on tissue homeostasis, we next performed RNAseq analysis on the tissues with the greatest increase in GSH with GshF expression, the kidney and lung (Figs.3D and 4, A and B).Gene ontology (GO) analysis revealed that the only common pathway differentially expressed in both the kidney and the lung was response to steroid hormones, which was downregulated in both tissues (Fig. 4, A-D).Several significant GO terms in lung were related to an increased immune response and to an increase in genes contributing to the extracellular matrix (Fig. 4, A and B).In the kidney, differential gene expression corresponded with cell differentiation and lipid metabolism, as well as membrane transport (Fig. 4, C and D).Some of the downregulated lipid metabolism genes are involved in steroid biosynthesis, which may contribute to the downregulation of the steroid hormone response in both tissues.Future work will need to determine whether GSH has differential functions in individual tissues.

Discussion
Glutathione is an evolutionarily ancient antioxidant that arose around the same time as photosynthetic cyanobacteria (41)(42)(43).Millions of years of evolution have extended GSH's role as a central reducing agent, and systems developed coupling GSH to critical cell processes including iron homeostasis, electron transport, and reducing power in the form of NADPH (43)(44)(45).Interestingly, GSH dysfunction is associated with a wide range of pathologies.GSH depletion is observed in neurodegenerative diseases, metabolic diseases like type II diabetes, chronic immunodeficiencies like AIDS, anemia, liver diseases, and lung disease, among others (3,46,47).Similarly, GSH abundance increases in many cancers and correlates with disease progression and mortality (8,9,13,(48)(49)(50).
Genetically encoded tools to manipulate cellular metabolites in vivo have been useful in determining the role of specific molecules, particularly in oxidative metabolism (24,25,51).We have described here the design and validation of GshF to manipulate compartmentalized GSH concentration.Previous work from our and other labs has demonstrated that mitochondrial GSH is critical for cell viability, and indeed we find that cells with uncoupled mitochondrial and cytosolic GSH remain sensitive to the inhibition of mitochondrial GSH transport or disruption of mitochondrial functionality (26,52).Surprisingly, increasing total GSH did not protect cells from ferroptosis.We interpret this to mean that GSH abundance is not limiting for GPX4 activity.The exception seems to be in iron overload, where increased cytosolic GSH is protective.While this could be related to GSH-mediated repair systems, it is also possible that GSH actively complexes free iron to prevent the Fenton, which can damage proteins, lipids, and nucleic acids (53)(54)(55).Further work will need to be done to determine the precise mechanism.
In mammals, GSH availability is tightly controlled though transcriptional, posttranscriptional, and posttranslational mechanisms.While GSH depletion is detrimental for cell and organismal viability, excess GSH can also cause reductive stress and lethally disrupt iron metabolism (6,7,56).Our work expands upon this knowledge in demonstrating that, like in yeast, there is a toxic level of GSH incompatible with mammalian development, emphasizing that homeostatic mechanisms to limit GSH production are critical for embryonic viability.The viability of CMV-mito-GshF Δ/0 underscores that unregulated GSH production alone may not be sufficient to disrupt homeostasis, but rather there is a critical range of GSH that must be maintained during embryonic development.While it was surprising that mito-GshF expression had little impact on tissue GSH levels in vivo, there are several possible contributing factors.First, endogenous GSH synthesis could be downregulated in response to mito-GshF expression in vivo.
Second, mitochondrial amino acid resources may be more limited in some cell types.Consistent with the second possibility, we observed variable GSH production by mito-GshF across different cell lines, although even cell lines with little to no increase in mitochondrial GSH abundance due to mito-GshF expression were protected from BSO-induced cell death (Fig. S1B).Determining the processes most impacted by unregulated GSH production during embryonic development will require extensive work.This phenotype indicates that unregulated GSH synthesis or high levels of GSH may disrupt certain critical aspects of development.Perinatal lethality can be caused by a myriad of deficiencies, and determining the precise cause of death in these mice will require many tissuespecific models.
In the course of this work, we also identified a possible mechanism of BSO resistance.BSO is a glutamate analog originally described by Alton Meister to inhibit the ratelimiting step of GSH synthesis, GCL (57).In the past, several clinical trials have queried the efficacy of using BSO to sensitize tumor cells to oxidative stress to little success (58,59).GCL is a homodimer comprised of a catalytic subunit, GCLC, and a modifying subunit GCLM that is dispensable for GSH synthesis (34).Our work indicates that BSO may only inhibit the heterodimeric GCL complex, not GCLC alone.We speculate that downregulation of or loss of GCLM might enable resistance to BSO in vivo.Future efforts to develop GSH inhibitors specific for GCLC may still prove clinically A, RNAseq of lung tissues from male ER T2 -GshF Δ/Δ versus GshF fl/fl littermates 2 weeks after completion of tamoxifen injections.B, GO analysis of top differentially expressed genes in lung.The GO analysis was performed by using the hypergeometric test followed by Benjamini-Hochberg multiple test correction.C, RNAseq of kidney tissues from male ER T2 -GshF Δ/Δ versus GshF fl/fl littermates 2 weeks after completion of tamoxifen injections.D, GO analysis of top differentially expressed genes in kidney.The GO analysis was performed by using the hypergeometric test followed by Benjamini-Hochberg multiple test correction.
useful in some contexts, particularly in combination with thioredoxin inhibitors in cancer settings and in combination with anti-parasitic agents in African sleeping sickness and other Trypanosome diseases, which BSO has shown some promise in treating (60,61).
Intriguingly, increased GSH availability has minimal impact on metabolism in adult animals, which provides a unique opportunity to restore GSH concentration without otherwise perturbing disease states.While there were no observable differences at baseline in ER T2 -GshF Δ/Δ mice, it is possible that there will be differential effects in specific immune challenges and disease models, although this will require a battery of tests to determine.We anticipate that complementary experiments with tumor cells expressing GshF alongside the GshF mouse models will enable dissection of the role of glutathione in many settings including lung cancers, which upregulate the Kelch-like ECH-associated protein 1-nuclear factor erythroid 2-related factor 2 axis, resulting in pleiotropic effects that include an upregulation of glutathione synthesis (21,62,63).Similarly, establishing orthotopic tumors in models will enable investigation of glutathione limitation in cancer progression, which may be of particular interest because clinical trials have demonstrated that oral consumption of antioxidants increases risk of cancer (64,65).Finally, mating these mice to genetic models of disease will enable restoration of GSH abundance in a temporally controlled manner, enabling interrogation of the role of GSH sufficiency in disease onset and progression.In conclusion, our work provides a conditional mouse model that will enable dissection of the contribution of GSH sufficiency to disease progression in a variety of contexts.

Cell lines and reagents
All cell lines were purchased from the ATCC.Cell lines were regularly monitored to be free of mycoplasma contamination and the identities of all were verified by STR profiling.All cells except HepG2 were maintained in RPMI 1640 media (Gibco) containing 2 mM glutamine, 10% fetal bovine serum, 1% penicillin and streptomycin.HepG2 cells were maintained in DMEM media (Gibco) containing 4.5 g/l glucose, 110 mg/l pyruvate, 4 mM glutamine, 10% fetal bovine serum, 1% penicillin and streptomycin.All cells were maintained at 37 C, 21% oxygen, and 5% carbon dioxide.
Lentiviral particles were produced by transfecting HEK-293F cells with sgRNA-containing vector, viral packaging vector deltaVPR, and lentiviral envelope vector CMV VSV-G using XtremeGene 9 transfection reagent (Roche).Viral particles were collected 48 h after transfection and passed through a 0.45 μm filter to eliminate cells and cell debris.Virus was then delivered to target cell lines seeded in 6-well tissue culture plates in medium containing 8 μg/ml polybrene.Cells were centrifuged at 2200 rpm for 1 h to improve infection efficiency.Virus was removed 24 h after transduction, and top 1% of GFP+ cells were collected by fluorescence-activated cell sorting on a BD FACSAria II.Cells were validated for loss of the target protein by immunoblotting except Gclm-KO cells, which had an interfering band at the same molecular weight as Gclm.The genomic region targeted by sgGclm was PCRamplified and submitted to Sanger sequencing to ensure that the exon sequence was disrupted.

Generation of cDNA overexpression
As described in Wang et al. (28), S. thermophilus GshF cDNA was humanized using the IDT codon-optimization tool and custom synthesized by IDT.To target GshF to mitochondria, the Lactobacillus brevis COXIV mitochondrial targeting sequence and human ACO2 mitochondrial targeting sequence were added in tandem with a Flag-tag to the N terminus of the humanized GshF cDNA.
Codon-optimized cDNA was cloned into BamH1, Not1linearized pMXS-blasticidin, or pMXS-blasticidin by Gibson reaction.Retroviral particles were produced by transfecting HEK-293F cells with cDNA-containing vector, viral packaging vector deltaVPR, and viral envelope vector gag-pol using XtremeGene 9 transfection reagent (Roche).Collection of viral particles and transduction was performed as described for lentiviral particles.Cells were selected with blasticidin or puromycin after removal of virus.

cDNA sequences
Below are Homo sapiens codon-optimized sequences of S. thermophilus GshF with added mitochondrial targeting sequences (MTS) and FLAG-tags.To ensure mitochondrial localization of the mito-GshF construct, a tandem MTS consisting of the L. brevis COXIV MTS followed by the H. sapiens ACO2 MTS was added to the N terminus of the GshF sequence, along with an N-terminal FLAG-tag.For clarity, corresponding regions of each sequence are highlighted as follows: [FLAG tag], {L.brevis COXIV MTS}, <H.sapiens ACOII MTS>, ((GGS) 3 linker).

Cell fractionation
Cell fractionation was performed according to Cell Fractionation Kit (9038, Cell Signaling Technology) instructions, with the following modifications: 2.5e6 cells were pelleted at 350g for 5 min, washed twice with ice-cold PBS, and resuspended in 500 μl cold PBS.Hundred microliters of cell suspension was collected for the whole cell lysate.Remaining cells were pelleted at 500g at 4 C and resuspended in 250 μl cytoplasm isolation buffer containing protease inhibitors (Sigma-Aldrich, 11836170001).After saving the supernatant, the pellet was washed 3× with 100 μl cytoplasm isolation buffer before resuspension in 250 μl membrane isolation buffer containing protease inhibitor.After removing the membrane and organelle fraction, the pellet was washed 3× with 100 μl membrane isolation buffer before resuspending the cytoskeletal and nuclear fraction in 125 μl cytoskeletal and nuclear isolation buffer.Samples were boiled at 95 C and centrifuged for 3 min at 15,000g before resolution on 10 to 20% SDS-PAGE gel.

Immunoblotting
Cell pellets were washed twice with ice-cold PBS before lysis in ice-cold RIPA buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% SDS) supplemented with protease inhibitors (Sigma-Aldrich, 11836170001) and phosphatase inhibitors (Roche, 04906837001).Lysis was allowed to proceed on ice for 30 min before centrifugation at 20,000g for 10 min at 4 C. Supernatant was collected and protein concentration was determined by BCA (Thermo Fisher Scientific, 23227).Except for cell fractionation, samples were diluted to equal concentrations and heated at 70 C for 10 min before resolution on 12% or 10 to 20% SDS-PAGE gels (Invitrogen).Resolved proteins were transferred to nitrocellulose membrane in CAPS transfer buffer.Membranes were blocked in 5% BSA in 1X TBST before incubation in primary antibodies at 4 C overnight with shaking.Membranes were incubated with HRP-linked antimouse IgG (Cell Signaling, 7076, 1:5000) or anti-rabbit IgG (Cell Signaling, 7074, 1:5000) for 1 h at room temperature while shaking.Membranes were incubated in PerkinElmer Enhanced Chemiluminescence Substrate (PerkinElmer, NEL105001EA) for 1 min.HyBlot CL Autoradiography Film (Thomas Scientific, 114J52) was exposed to membranes in the dark and developed on an SRX-101A Film Processor (Konica Minolta).
For tissue samples, a 4 mm biopsy punch from snap-frozen tissue was homogenized in 500 μl RIPA buffer with protease and phosphatase inhibitors in a 2 ml dounce homogenizer.Samples were rotated at 4 C for 10 min before centrifugation for 2 min at 1000g at 4 C to remove excess tissue.Ten microliters of the supernatant was reserved to determine protein concentration, and Laemmli dye was added to 150 μl of supernatant.Samples were diluted to equal concentrations and sonicated before Western blots were performed as described above.

Polar metabolite profiling
For whole cell samples, cells were washed twice with 1 ml ice-cold 0.9% NaCl.Polar metabolites were extracted in 500 μl ice-cold 80% methanol containing amino acid standards (Cambridge Isotope Labs, MSK-A2-1.2).Samples were vortexed at 4 C for 10 min and centrifuged for 10 min at 20,000g at 4 C before supernatant was collected and dried via nitrogen evaporation.Total protein in the pellet was determined by BCA to normalize metabolite levels after quantification by mass spectrometry.Dried samples were stored at −80 C until resuspension in 60 μl of 50% acetonitrile for metabolomics analysis.
Mitochondria were purified from HEK-293T cells expressing 3xHA-OMP25-mCherry or a negative control expressing 3xMyc-OMP25-mCherry according to the protocol described by Chen et al. (66).In brief, triplicate samples of 25 to 30 million cells were collected per condition.Cells were washed twice with ice-cold 0.9% NaCl and scraped into 1 ml of icecold KPBS.Cells were spun at 1000g for 1.5 min at 4 C and resuspended in 1 ml of cold KPBS.Ten microliters of resuspended cells were transferred for input protein and whole cell metabolites were extracted from an additional 10 μl transferred into 40 μl of 80% methanol with amino acid standards.The remaining sample was transferred to a 2 ml dounce homogenizer and homogenized with two sets of 30 strokes.The homogenate was centrifuged and supernatant was incubated with 200 μl of 50% anti-HA magnetic beads (Thermo Fisher Scientific Pierce 88837) in KPBS for 5 min while rotating at 4 C. Beads were washed three times in 1 ml cold KPBS.Ten percent of bead volume was taken for protein and metabolites were extracted from the remaining bead volume with 50 μl of 80% methanol containing amino acid standards.Samples were rotated at 4 C for 10 min before being spun at 20,000g for 10 min at 4 C. Input and mitochondrial samples were subjected to LC-MS profiling without drying.Data were normalized to citrate synthase or GAPDH protein level as quantified from Western blot or NAD + abundance.
Tissue samples were immediately flash frozen in liquid nitrogen after rapid isolation from mice euthanized by cervical dislocation.Samples were stored at −80 C until metabolite extraction.To extract metabolites, 2× 4 mm biopsy punches were taken from snap-frozen tissue and homogenized in 1 ml ice cold 0.9% NaCl in a 2 ml dounce homogenizer with 2 rounds of 35 strokes.Excess tissue was removed by centrifugation at 1000g for 2 min at 4 C. Metabolites were extracted from 10 μl of supernatant with 40 μl of 1:1 acetonitrile/ methanol containing amino acid standards.Samples were vortexed for 10 min at 4 C before being centrifuged at 20,000g for 10 min at 4 C.The supernatant was dried and stored at −80 C until resuspension in 2:2:1 acetonitrile/methanol/ water for mass spectrometry analysis.
Metabolomics analysis was performed on a QExactive benchtop orbitrap mass spectrometer equipped with an Ion-Max source and a HESI II probe coupled to a Vanquish UPLC system (Thermo Fisher Scientific).External mass calibration was performed using the standard calibration mixture every 3 days.Two microliters of sample were injected into a ZIC-pHILIC 150 × 2.1 mm (5 μm particle size) column (EMD Millipore).Chromatographic separation was achieved using the following conditions: buffer A was 20 mM ammonium carbonate, 0.1% ammonium hydroxide; buffer B was acetonitrile.The column oven and autosampler tray were held at 40 C and 4 C, respectively.The chromatographic gradient was run at a flow rate of 0.15 ml/min as follows: 0 to 22 min linear gradient from 90% to 40% B; 22 to 24 min held at 40% B; 24 to 24.1 min returned to 90% B; 24.1 to 30 min equilibrated at 90% B. The mass spectrometer was operated in full-scan, polarity switching mode with the spray voltage set to 3.0 kV, the heated capillary held at 275 C, and the HESI probe held at 350 C. The sheath gas flow was set to 40 units, the auxiliary gas flow was set to 15 units, and the sweep gas flow was set to 1 unit.The MS data acquisition was performed in a range of 55 to 825 m/z, with the resolution set at 70,000, the AGC target at 1e6, and the maximum injection time at 20 ms.Relative quantification of polar metabolites was performed with Skyline using a 2 ppm mass tolerance and referencing an in-house library of chemical standards.Metabolite levels were normalized to the total protein amount for each sample, determined by BCA on the pellet of each sample.

Plasma profiling and GSH derivatization
GSH is difficult to reliably detect in serum and whole blood due to its high sensitivity to oxidation.To minimize oxidation during sample collection, 100 μl whole blood was collected into EDTA-coated tubes containing 20 μl of N-ethylmaleimide (NEM), amino acid standards, and 5 μl of 1 mM [ 13 C 3 , 15 N]-GSH standard.Samples were vortexed briefly on ice before centrifugation twice at 400g for 5 min to obtain plasma.DTT (Sigma) was then added to equal the final concentration of NEM in plasma.Proteins were precipitated with 80% methanol for 30 min at −20 C and removed by centrifugation at 20,000g for 10 min.The supernatant was dried under nitrogen and resuspended in 100 μl of cold water, and the pellets were reserved to determine total protein by BCA.The samples were vortexed for 10 s before addition of 300 μl cold dichloromethane to remove excess NEM and DTT.Samples were vortexed vigorously for 10 min at 4 C prior to centrifugation at 20,000g for 10 min.The aqueous phase was dried and stored at −80 C for no more than 48 h before resuspension in 50 μl of 75:25 acetonitrile/methanol for metabolomics analysis.Metabolomics analysis was performed as described above.

Metabolite tracing
For cystine tracing experiment, HEK-293T Gclm-KO cells expressing indicated cDNAs were seeded in triplicate in 6 well plates.Cells were allowed to attach for 24 h prior to addition of BSO.After 24 h of BSO treatment, media was replaced with RPMI + 10% dialyzed FBS (Gibco) lacking cystine and containing [ 13 C 6 , 15 N 2 ]-L-Cystine (207.67 μM).After 24 h of incubation with [ 13 C 6 , 15 N 2 ]-L-Cystine in the presence or absence of BSO, metabolites were extracted and quantified as described above.Fractional labeling was corrected for natural abundance with IsoCorrectoR (67) using RStudio version (2022.12.0 + 353).

Proliferation assays
Cells were seeded in triplicate in 96-well plates as follows: Jurkat cells, 2000 cells/well; PANC-1 and HepG2, 500 cells/ well; all other cell lines 1000 cells/well.Cells were allowed to attach to plates for 6 h prior to addition of compounds described in each experiment.An additional plate was seeded without treatment and taken for an initial time point upon addition of compounds to the experimental plates.Cells were allowed to proliferate for 5 days.At either the initial timepoint or the endpoint, 40 μl of Cell Titer Glo reagent (Promega) was added to each well.Plates were incubated in the dark on a rocker for 10 min prior to quantification of luminescence per well on a SpectraMax M3 plate reader (Molecular Devices).The fold change in luminescence relative to day 0 was determined and reported on a log 2 scale.

CRISPR screens
The metabolism-focused human sgRNA library used was first described by Birsoy et al., and screens were performed as described previously (68)(69)(70).Oligonucleotides for sgRNAs were synthesized by CustomArray and amplified by PCR.In brief, a lentiviral library was generated from the plasmid pool and transfected into HEK-293F cells to produce viral particles as described above.SLC25A39-KO Jurkat cells expressing either GshF or mito-GshF were infected at a multiplicity of infection of 0.7 and selected with puromycin for 3 days.An initial sample of 30 million cells was taken for each cell line, and cells were seeded into 500 ml spinning flasks in RPMI alone or RPMI with 200 μM BSO (Sigma).Cells were split every 3 days and grown under indicated conditions for approximately 14 population doublings.Final samples of 30 million cells were collected, and DNA was extracted according to DNeasy Blood and Tissue Kit instructions (Qiagen 69506).sgRNA inserts were amplified and barcoded by PCR with primers unique to each condition.PCR amplicons were gelpurified and sequenced on a NextSeq500 (Illumina).Screens were analyzed using Python (v.2.7.13),R (v.3.3.2), and Unix (v.4.10.0-37-genericx86_64).The gene score for each gene was defined as the median log2 fold change in the abundance of each sgRNA targeting that gene.A complete list of differential gene scores for each screen is provided in Table S1.

Generation of transgenic mouse strains
All animal studies and procedures were conducted according to a protocol approved by the Institutional Animal Care and use Committee at the Rockefeller University.All mice were maintained on a standard light:dark cycle with food and water provided ad libitum.
GshF cDNA was cloned into FseI-linearized Ai9 vector, which was previously reported to result in broad expression patterns when inserted at the Rosa26 locus (71).mito-GshF cDNA was cloned into a CTV vector.Each of these targeting constructs contains 5 0 and 3 0 Rosa26 homology arms and a LoxP-STOP-LoxP cassette upstream of the inserted cDNA.Each targeting construct was electroporated into Cy2.4 [B6(Cg)-Tyr <c2J> genetic background] embryonic stem cells.cDNA-containing embryonic stem clones were selected with neomycin, and presence of the cDNA was validated by PCR and with anti-Flag immunoblots from Cre-transfected clones.Two single clones for each construct were injected into blastocysts to produce GshF or mito-GshF germlinetransformed heterozygous chimera mice.Chimeras with a high percentage of albino coat color were selected to breed with WT C57BL6/J mice (The Jackson Laboratory, 000664), and the germline-transmitted offspring were confirmed by genotyping.All mice were backcrossed for at least four generations prior to conducting any experiments shown here.GshF and mito-GshF mice were generated in the Birsoy laboratory with the assistance of the CRISPR and Genome Editing Center at the Rockefeller University.

Mouse strain accessibility
GshF fl/fl and mito-GshF fl/fl have been deposited to the Jackson Laboratory under stock numbers 038275 and 038277, respectively.

Mouse studies
All mice were intraperitonially injected with 100 μl tamoxifen (Cayman Chemical, 13258) on alternating abdominal sides for five consecutive days.Tamoxifen was prepared at 20 mg/ml in sterile corn oil (Sigma C8267) and shaken at 32 C until in solution.All experiments were performed at least 7 days after the final tamoxifen injection.All mice used in these studies were 8 to 14 weeks old, and mice of both sexes were used except when specified in figure legends.Both ER T2 -Cre -null, GshF-positive and ER T2 -Cre-positive, GshF-null mice were used as GshF-null controls.

Genotyping
Mito-GshF fl/fl and CMV-mito-GshF mice were genotyped by Transnetyx.GshF fl/fl , CMV-GshF, and ERT2-GshF mice were genotyped by Transnetyx and/or using the following primers: GshF primers result in a 900 bp band when floxed and a 370 bp band when induced by Cre.Cre primers only produce bands when Cre is present; Rosa26 primers only produce bands when the locus is WT.

Embryo imaging
Timed pregnancies were determined by confirmation of a visible vaginal plug the morning after mating pairs were combined.After confirmation of vaginal plug, males were removed from cages to prevent additional copulation.The day that vaginal plugs are identified is considered gestational day E0.5.Females were euthanized by CO 2 exposure at indicated timepoints and embryos were rapidly collected.For embryos younger than E15.5, egg sacs were collected for genotyping; for E15.5 and E17.5, tail clippings were collected for genotyping.Brightfield images were taken with a Zeiss Axiocam 506 mono camera on an AxioZoom V.16 microscope.Embryos were fixed in formalin for 48 h.Sectioning and H&E staining was performed by Histoserv.Sagittal sections were taken at the midline, at the kidney, and at the eye level.Histopathology evaluation was performed by Ileana Miranda at the Laboratory of Comparative Pathology at Memorial Sloan Kettering.

Complete blood counts
Whole blood was collected in EDTA-coated tubes.Complete blood counts were performed within 30 min of blood sampling on a Heska Element HT5 hematology analyzer.
Colons were processed as previously described for the small intestine (72).Briefly, colons were filet-opened, washed with cold PBS, chopped into 50 ml conical tubes filled with 10 ml of PBS, and supplemented with 1 μM DTT (Sigma).Tubes were shaken at room temperature for 10 min, then intensely shaken by hand for 2 min before tissue was filtered through a metal strainer into a new 50 ml conical tube containing with 25 ml of cRPMI (RPMI Media 1640 supplemented with 1% Hepes and 2% FCS).Flowthrough was considered fraction 1 of intraepithelial lymphocytes (IEL).Tissues remaining in the strainer were transferred to 10 ml of PBS containing 30 mM EDTA (Thermo Fisher Scientific) and 10 mM HEPES, and tubes were shaken in a 37 C shaker at 180 rpm for 10 min.Tubes were then shaken vigorously by hand for 2 min and supernatant filtered into IEL fraction 1.The combined IEL fractions were filtered in a 40% to 80% Percoll gradient as previously described.Remaining tissues were transferred to 10 ml of PBS containing 30 mM EDTA and 10 mM Hepes and shaken at 180 rpm for 10 min at 37 C followed by 2 min of vigorous hand shaking.The supernatant was discarded and tissues were washed several times with cold PBS before being finely minced with scissors in 6-well plates.The tissue slurry was resuspended in 6 ml of collagenase mix (collagenase 8 and DNase in TCM) and while shaken for 45 min at 37 C, 80 rpm before filtration through 70 μm mesh.Samples were then washed and filtered through a Percoll gradient as previously described.All centrifugations were done at 400g for 5 min at 4 C, except for the Percoll gradient.
For flow cytometry analysis, cells were stained in a surface antibody mix in PBS for 25 min at 4 C, washed in PBS, then fixed/permeabilized for 1 h at 4 C using eBioscience Foxp3/ Transcription Factor Staining Buffer Set (Thermo Fisher Scientific).Cells were then washed with 1X permeabilization buffer (Thermo Fisher Scientific) and stained in intracellular antibody mix for 45 to 90 min at 4 C and washed with PBS before flow cytometry analysis.

RNAseq
Tissues were rapidly isolated and snap-frozen in liquid nitrogen from mice euthanized by cervical dislocation.RNA was extracted from whole kidney or single lung using TRIzol reagent (Thermo Fisher Scientific) according to manufacturer's manual.After DNase I treatment (New England Biolabs), RNA concentrations were determined using a Qubit 2.0 Fluorometer (Life Technologies) and RNA integrity was confirmed by Agilent TapeStation 4200 (Agilent Technologies).RNA-sequencing libraries were prepared with unique barcodes using the NEB-Next Ultra RNA library kit for Illumina (New England Biolabs) according to manufacturer's instructions.The sequencing libraries were pooled at equimolar ratios and sequenced on an Illumina NextSeq 500 using 2 × 150 bp paired-end reads.Read quality was confirmed with Rfastp (https://bioconductor.org/packages/Rfastp).The reads alignment and gene models were based on Mus_musculus.GRCm39.107.gtf.gz.Transcript expression was computed using Salmon (73), followed by processing with DESeq2 (74).Significant differentially expressed genes between conditions were identified by DESeq2 with a Benjamini-Hochberg adjusted p-value < 0.05 and absolute fold change >0.5.All RNAseq results are available in Tables S2 and  S3.GO terms were retrieved from Mouse MSigDB Collections Gene Ontology Analysis (75,76).Terms consisting of fewer than 10 or more than 500 were excluded.The GO analysis was performed by using the hypergeometric test followed by Benjamini-Hochberg multiple test correction.

Figure 1 .
Figure 1.Expression of an engineered bacterial enzyme protects cells from death due to inhibition of GSH synthesis, but not ferroptosis.A, schematic of glutathione (GSH) synthesis and regulation in mammals compared to Streptococcus thermophilus.B, the localization of engineered GshF proteins was determined by immunoblotting subcellular fractions from HEK-293T cells expressing cDNA constructs of either GshF or mito-GshF.The cytosolic, membrane-bound organelle, and nuclear and cytoskeletal fractions are denoted by C, M, and N, respectively.C, total GSH abundance in HEK-293T cells expressing GshF or a vector control and treated with indicated buthionine sulfoximine (BSO) doses for 48 h, normalized to total protein abundance.Each point represents a technical replicate; error bars represent SD. p values were calculated by two-tailed unpaired t test with Bonferroni-Dunn multiple test correction.D, total (left) and mitochondrial (right) GSH abundance in HEK-293T cells expressing indicated mito-GshF or a vector control and treated with indicated BSO doses for 48 h, normalized by GAPDH levels (total) or NAD + levels (mitochondria).Each point represents a technical replicate; error bars represent SD. p values were calculated by two-tailed unpaired t test with Bonferroni-Dunn multiple test correction.E, fold change in total ATP (log 2 ) of indicated cell lines expressing a vector control or specified cDNAs and treated for 5 days with the denoted concentrations of buthionine sulfoximine (BSO).Fold change in ATP is a proxy for cell doublings.Error bars are mean ± SD; points are biological replicates.p < 0.05 are indicated by *p ≤ 0.01are indicated by **p ≤ 0.001 are indicated by ***and p ≤ 0.0001 are indicated by ****p values were calculated by two-tailed unpaired t test with Bonferroni-Dunn multiple test correction.F, fold change in total ATP (log 2 ) of HEK-293T cells expressing a vector control or indicated cDNAs and treated for 5 days with the indicated concentration of paraquat.Fold change in ATP is a proxy for cell doublings.Error bars are mean ± SD; points are biological replicates.p values were calculated by two-tailed unpaired t test with Bonferroni-Dunn multiple test correction.G, fold change in total ATP (log 2 ) of HEK-293T cells expressing a vector control or indicated cDNAs and treated for 5 days with the indicated concentration of RSL3.Fold change in ATP is a proxy for cell doublings.Error bars are mean ± SD; points are biological replicates.p < 0.05 are indicated by *p ≤ 0.01 are indicated by **p ≤ 0.001 are indicated by ***, and p ≤ 0.0001 are indicated by ****.p values were calculated by two-tailed unpaired t test with Bonferroni-Dunn multiple test correction.H, fold change in total ATP (log 2 ) of HEK-293T cells expressing a vector control or specified cDNAs and treated for 5 days with the indicated concentrations of ferric ammonium citrate (FAC).Fold change in ATP is a proxy for cell doublings.Error bars are mean ± SD; points are biological replicates.p < 0.05 are indicated by *p ≤ 0.01are indicated by **p ≤ 0.001 are indicated by ***, and p ≤ 0.0001 are indicated by ****.p values were calculated by two-tailed unpaired t test with Bonferroni-Dunn multiple test correction.cDNA, complementary DNA; CS, Citrate Synthase; GCLC, glutamate cysteine ligase catalytic subunit; GCLM, glutamate cysteine ligase modifying subunit; GR, glutathione reductase; KEAP1, Kelch-like ECH-associated protein 1; NRF2, Nuclear factor erythroid 2-related factor 2.

Figure 2 .Figure 3 .
Figure 2. A metabolism-focused CRISPR-Cas9 screen identifies differential roles of subcellular GSH pools and mechanisms of BSO resistance.A, schematic of metabolism-targeted CRISPR-Cas9 screen in Jurkat SLC25A39-KO cells.B, fold change in total ATP (log 2 ) of Jurkat SLC25A39-KO cells expressing a vector control or indicated cDNAs and treated for 5 days with the indicated concentrations of BSO.Error bars are mean ± SD; points are biological replicates.p < 0.05 are indicated by *p ≤ 0.01are indicated by **p ≤ 0.001 are indicated by ***, and p ≤ 0.0001 are indicated by ****.p values were calculated by two-tailed unpaired t test with Bonferroni-Dunn multiple test correction.C, left, plot of gene score ranks from Jurkat SLC25A39-KO screens treated with 200 μM BSO, where genes are ranked from those with the most enriched sgRNAs to the most depleted.Genes essential in SLC25A39-KO cells expressing both GshF and mito-GshF are in the lower left quartile, those specifically essential in GshF-expressing cells are in the top left quartile, and those specifically essential in mito-GshF-expressing cells are in the bottom right quartile.Right, top 10 genes scoring as differentially required upon BSO treatment in cells expressing GshF compared to mito-GshF, colored by association with indicated pathways.D, gene scores of Jurkat SLC25A39-KO cells expressing GshF (left) or mito-GshF (right) and treated with 200 μM BSO for 14 doublings.The gene score is calculated as the median log 2 fold change in all sgRNAs targeting a specific gene during the course of the culture period, as compared to a sample taken immediately before treatment began.E, relative fold change in the total ATP (log 2 ) of Jurkat (left) and HEK-293T cells (right) transduced with indicated sgRNAs and either complemented with sgRNA-resistant Gclm cDNA or an empty vector control.Cells were treated for 5 days with the indicated BSO doses.Comparisons of sgGclm + Vector Control versus Gclm cDNA are colored red and comparisons of sgGclm versus sgCtrl are colored black.p < 0.05 are indicated by *p ≤ 0.01are indicated by **p ≤ 0.001 are indicated by ***, and p ≤ 0.0001 are indicated by ****.p values were calculated by two-tailed unpaired t test with Bonferroni-Dunn multiple test correction.F, left, relative abundance of GSH in Gclm-knockout HEK-293T cells complemented with sgRNA-resistant Gclm cDNA or an empty vector control and treated with indicated BSO concentrations for 24 h.Data is normalized to the total protein abundance.Each point represents a technical replicate; error bars represent standard deviation.p values were calculated by two-tailed unpaired t test with Bonferroni-Dunn multiple test correction.G, schematic describing incorporation of

Figure 4 .
Figure4.RNAseq of GshF mice reveals common and tissue-specific pathways impacted by high GSH levels.A, RNAseq of lung tissues from male ER T2 -GshF Δ/Δ versus GshF fl/fl littermates 2 weeks after completion of tamoxifen injections.B, GO analysis of top differentially expressed genes in lung.The GO analysis was performed by using the hypergeometric test followed by Benjamini-Hochberg multiple test correction.C, RNAseq of kidney tissues from male ER T2 -GshF Δ/Δ versus GshF fl/fl littermates 2 weeks after completion of tamoxifen injections.D, GO analysis of top differentially expressed genes in kidney.The GO analysis was performed by using the hypergeometric test followed by Benjamini-Hochberg multiple test correction.

GraphPad PRISM 9
and 10 and Microsoft Excel Software (version 16.77.1)were used for statistical analysis.Skyline-daily 64 bit version 21.2.1455 was used for metabolomics analysis, and FIJI (ImageJ2, NIH, version 2.14.0/1.54f) was used for image analysis including measuring embryo length.Error bars and statistical tests are reported in the figure captions.For all figures, only p values less than or equal to 0.05 are reported unless otherwise noted.All experiments except RNAseq and CRISPR screens were performed at least twice with similar results.Both technical and biological replicates were reliably reproduced.Comparison of two mean values was evaluated by two-tailed unpaired t test.Comparison of multiple mean values was evaluated by one-way ANOVA followed by post hoc Bonferroni test.Comparison of multiple mean values under different conditions was evaluated by two-way ANOVA.