Abstract

Restricting amino acids from tumours is an emerging therapeutic strategy with substantial promise 1 . Although typically considered an intracellular antioxidant with tumour-promoting capabilities 2 , glutathione (GSH), as a tripeptide of cysteine, glutamate and glycine, can be catabolized to release amino acids. The extent to which GSH-derived amino acids are essential to cancers is unclear. Here we show that depletion of intracellular GSH does not alter tumour growth and extracellular GSH is highly abundant in the tumour microenvironment, highlighting the potential importance of GSH outside tumours. Supplementation with GSH rescues cancer cell survival and growth in cystine-deficient conditions, and this rescue depends on the catabolic activity of γ-glutamyltransferases. Finally, pharmacological targeting of the activity of γ-glutamyltransferases prevents the breakdown of circulating GSH, reduces tumour cysteine levels and slows tumour growth. Our findings indicate a non-canonical role for GSH in supporting tumours by acting as a reservoir of amino acids. Depriving tumours of extracellular GSH or inhibiting its breakdown is potentially a therapeutically tractable approach for patients with cancer. Furthermore, these findings change our view of GSH and how amino acids, including cysteine, are supplied to cells.

Main

Amino acids are crucial for cancer initiation, progression and drug resistance 3 . Despite their indispensability, amino acids are often scarce in the tumour microenvironment 4 , which drives tumours to develop strategies to import and synthesize them 5 , 6 , 7 . Thus, interfering with these mechanisms and restricting amino acid access holds potential as an anticancer strategy 1 . One amino acid that has garnered substantial attention as an anticancer target is cysteine 8 , 9 . Beyond being a building block for proteins, cysteine has roles that include supporting the generation of antioxidants (for example, glutathione and persulfide species) 10 , metabolites (for example, H 2 S, CoA and hypotaurine) 11 , 12 and iron–sulfur clusters for mitochondria 13 . The intracellular cysteine pool is thought to be maintained by at least three sources: import through alanine, serine, cysteine, threonine transporter 1 (ASCT1); uptake of cystine through system x c − (xCT and CD98) and subsequent reduction by thioredoxin reductase (TXNRD1); and de novo synthesis from methionine through the transsulfuration pathway. Cystine uptake through system x c − is thought to be the predominant source of cysteine in cancer cells 14 . Notably, the deletion of Slc7a11 (which encodes xCT) in animals results in viable offspring 15 , which suggests that tissues (and potentially tumours) can obtain cysteine from another origin. The generation of cysteine from the transsulfuration pathway is a potential source. However, this pathway is reported to be inactive in most tissues, and in the few tissues in which it is active, its function is reduced in tumours from the corresponding tissue 16 . Together, these results suggest that tumours have an alternative mechanism to acquire cysteine.

GSH is an antioxidant that regulates oxidative stress, drug detoxification and post-translational modifications 2 . Disruption of GSH production can impair tumorigenesis 17 , 18 , 19 , but the exact mechanisms involved are unclear. GSH is a tripeptide of glutamate, cysteine and glycine, and it can be broken down into its individual constituent amino acids. The rate-limiting step in extracellular GSH catabolism is controlled by γ-glutamyltransferases (GGTs), which cleave the γ-glutamyl bond in GSH to release glutamate and dipeptide cysteinylglycine. After its release, cysteinylglycine is further broken down by peptidases to produce cysteine and glycine. GSH catabolism is proposed as a source of amino acids for cells 20 , 21 . This idea dates back to seminal work from Eagle, who showed that supplementation with extracellular GSH supports cell growth in cystine-free conditions 22 . Indeed, catabolism of mitochondrial GSH supplies cysteine for iron–sulfur cluster synthesis 23 . In animals, unlike the deletion of Slc7a11, the deletion of the GGT family member Ggt1 results in increased GSH levels in urine, reduced cysteine levels in tissues and perinatal lethality 24 . Notably, patients with mutations in GGT1 often present with developmental disorders, accumulation of GSH (glutathionaemia and glutathionuria) and reduced circulating levels of cystine 25 . Together, these findings indicate that GSH catabolism by GGTs is an important source of amino acids, including cysteine. Whether tumours can co-opt GSH catabolism to fuel their growth is poorly understood.

Here we identify an underappreciated role for GSH as a cysteine reservoir for tumours. Compared with cystine, GSH is highly abundant in the microenvironment of tumours, and supplementation with GSH or its product cysteinylglycine rescues cancer cell growth in cystine-depleted conditions. Mechanistically, GGT activity is necessary and sufficient to promote survival by catabolizing GSH and supplying amino acids to surrounding cells. Furthermore, supplementation with extracellular GSH renders cancer cells resistant to drugs that block cystine uptake or its reduction into cysteine. Finally, inhibition of GGT activity deprives tumours of cysteine and slows their growth. These results reveal an actionable pathway of nutrient acquisition in cancer with direct therapeutic implications. Furthermore, these findings change our perspectives about GSH biology and the acquisition of amino acids by cells.

GSH production in tumours is dispensable

GSH facilitates tumorigenesis 17 , 18 , 19 ; however, it is unclear whether intracellular GSH production by tumours themselves is required. To explore this idea, we bred the MMTV-PyMT transgenic mouse strain (which spontaneously develops breast tumours) 26 with an uninduced Gclc f/f Rosa26 creERT2 mouse 27 in which the rate-limiting step in GSH synthesis, the glutamate–cysteine ligase catalytic subunit (GCLC), can be knocked out throughout the body after induction with tamoxifen. After allowing tumours to develop in these mice, the tumours were excised and orthotopically implanted into recipient immunocompetent wild-type (WT) mice (C57BL/6). The recipient mice were treated with tamoxifen, which leads to the activation of Cre recombinase activity and subsequent deletion of Gclc specifically in the implanted tumours (Fig. 1a). Gclc mRNA and GSH levels were reduced in mice with Gclc specifically knocked out in the tumour (Fig. 1b,c). However, GSH levels were not completely abolished, which may be due to the presence of GCLC-expressing cells in the tumour microenvironment (Extended Data Fig. 1a–c). Notably, the deletion of Gclc in tumours did not affect their growth (Fig. 1d,e), which suggests that the ability of the tumour to synthesize intracellular GSH is dispensable. Moreover, tumours did not show signs of oxidative stress or compensatory metabolic changes, such as accumulation of cysteine (Extended Data Fig. 1d–j). Similar results were observed when tumours were implanted into the flanks of mice, a result that suggests that this phenotype is not specific to the mammary gland environment (Extended Data Fig. 2). Similar phenotypes were also observed when GCLC was deleted in HCC-1806 cells (a human breast cancer cell line), which showed complete depletion of GSH levels (Fig. 1f–i). The lack of an antitumour effect from the loss of intracellular GSH suggested that extracellular GSH is instead crucial for tumour growth. To explore this idea, we examined total GSH (tGSH; calculated as the sum of reduced GSH and two times the amount of oxidized glutathione (GSH + 2 × GSSG)) levels in the serum and in the tumour interstitial fluid (TIF), the extracellular compartment surrounding tumours. Examination of mouse and human breast tumours revealed that tGSH levels were enriched in the TIF compared to serum (Fig. 1j,k). Similar trends were observed in a mouse model of pancreatic ductal adenocarcinoma (PDAC) (LSL-Kras G12D/+ Trp53 f/f Pdx1 cre (KP –/– C) mice 28 ) and in humans with renal cell carcinoma 29 (Extended Data Fig. 3a,b). Levels of tGSH in the TIF were much higher than reported amounts added to formulations of cell culture media (Extended Data Fig. 3c). Analyses of TIF from tumours with WT Gclc and with Gclc knocked out (KO) showed that intracellular GSH synthesis partially contributed to tGSH levels in the TIF but not serum (Extended Data Fig. 3d–h). There was no difference between tGSH and cystine levels in the serum of mice without tumours (Extended Data Fig. 3i). Notably, serum tGSH (but not serum cystine) levels were lower in mice with tumours than in mice without tumours (Extended Data Fig. 3j,k). Overall, these findings indicate that although intracellular GSH does not affect tumour growth and survival, extracellular GSH is highly abundant and potentially supports tumour growth. Furthermore, in the presence or absence of GSH synthesis, tumours have an adequate supply of upstream metabolites (that is, cysteine) to sustain growth.

a, Schematic of the tumour-specific Gclc knockout mouse model. Autochthonous tumours from Gclc f/f Rosa26 creERT2 MMTV-PyMT mice were excised and orthotopically transplanted into mammary fat pads of WT C57BL/6 mice. C57BL/6 mice were treated with vehicle (corn oil; WT) or 50 mg kg –1 tamoxifen for 5 days (KO). b, Relative Gclc mRNA levels of WT and KO tumours (n = 5 representative animals from 3 independent experiments). c, GSH levels of WT and KO tumours (n = 6 representative animals from 3 independent experiments). d, e, Volume over time (d) and mass at end point (e) of WT tumours (n = 11) and KO tumours (n = 12) from a representative of 3 independent experiments. f, g, Immunoblot of GCLC protein levels (f) and GSH levels (g) in HCC-1806 human breast cancer cells transduced with lentiCRISPRv2 containing non-targeting guides (sgNTC) and guides against GCLC (sgGCLC 1 and sgGCLC 2) (n = 3 representative samples from 2 independent experiments). For gel source data, see Supplementary Fig. 1. h, i, Volume over time (h) and mass at end point (i) of orthotopically implanted HCC-1806 breast cancer cell treated with sgNTC, sgGCLC 1 or sgGCLC 2 (n = 7 representative animals from 2 independent experiments). j, Concentration of tGSH (GSH + 2 × GSSG) in serum (n = 11) and the TIF (n = 4) from Gclc WT MMTV-PyMT autochthonous tumours from a. k, Concentration of tGSH in serum (n = 8) and the TIF (n = 6) from patients with breast cancer. Significance was assessed by two-tailed unpaired t-test (b, c, e), Mann–Whitney test (j, k), ordinary two-way analysis of variance (ANOVA) (d, h) or one-way ANOVA followed by Tukey’s multiple comparisons test (g, i). Data are the mean ± s.e.m.

Extracellular GSH supplies cysteine

The uptake of extracellular cystine through the cystine–glutamate antiporter xCT is thought to be a major source of cysteine for tumours (Fig. 2a). Extracellular glutamate levels, which can counteract xCT-mediated cystine uptake, were enriched in the TIF compared to serum (Extended Data Fig. 3l–n), whereas cystine levels were not (Fig. 2b,c). GSH is a cysteine-containing tripeptide that can be broken down by GGT enzymes into cysteinylglycine and glutamate. Cysteinylglycine can be further processed to produce intracellular cysteine. We sought to determine whether GSH or its catabolic product cysteinylglycine can replace cystine as a source of cysteine for cancer cells. Supplementation with GSH, at a concentration within the range found in the TIF, or with cysteinylglycine rescued cancer cell growth in cystine-free conditions (Fig. 2d and Extended Data Fig. 4a). Moreover, GSH or cysteinylglycine rescued both tumour survival and proliferation under cystine-free conditions (Fig. 2e,f).

a, Schematic of the different mechanisms of cysteine acquisition and use. b, Concentration of cystine in serum (n = 11) and the TIF (n = 4) from Gclc WT MMTV-PyMT autochthonous tumours from Fig. 1a. c, Concentration of cystine in serum (n = 8) and the TIF (n = 6) from patients with breast cancer. d–f, HCC-1806 breast cancer cells were grown in the following medium: control (208 µM cystine; Cys 2 ); cystine-free (Cys 2 -free); cystine-free with GSH supplemented (750 µM, Cys 2 -free + GSH); or cystine-free with cysteinylglycine (CysGly) supplemented (750 µM). Cell numbers (n = 3 technical replicates representative of 4 independent experiments) (d), percentages of proliferative cells (positive for bromodeoxyuridine (BrdU + ) at t = 48 h; n = 6 technical replicates from 3 independent experiments) (e) and apoptotic cells (Annexin V + at t = 72 h; n = 9 technical replicates from 3 independent experiments) (f) were determined at the indicated time points. g, Levels of extracellular CysGly in the medium at 72 h after HCC-1806 breast cancer cells were grown in the indicated medium (n = 3 representative technical replicates from 2 independent experiments). h, Schematic of the 13 C-cystine stable-isotope labelling approach and metabolomics. i, Per cent labelling of 13 C-cystine stable-isotope in the indicated species at 72 h (n = 5 technical replicates). j, Schematic of the 15 N 13 C-cystine stable-isotope labelling approach and proteomics. k, Per cent labelling of 15 N 13 C-cystine stable-isotope in the indicated species at 96 h (average of n = 3 technical replicates). TXN1, thioredoxin 1; TXN2, thioredoxin 2. Significance was assessed by unpaired two-tailed t-test (b, c), ordinary two-way ANOVA followed by Šídák’s multiple comparisons test (e, f), Dunnett’s multiple comparisons test (d) or one-way ANOVA followed by Tukey’s multiple comparisons test (g). Data are the mean ± s.e.m.

This rescue by GSH was not observed after deprivation of other amino acids (serine, glycine, glutamine or glutamate) or after inhibition of amino acid synthesis (via the glutaminase 1 inhibitor CB-839) (Extended Data Fig. 4b–e). Moreover, this rescue did not depend on intracellular re-synthesis of GSH, as pharmacological inhibition of GSH synthesis did not prevent GSH rescue in cystine-free conditions (Extended Data Fig. 4f,g). Furthermore, the l-isomer of N-acetylcysteine (NAC), which is both a cysteine source and an antioxidant, but not d-NAC, which is only an antioxidant 30 , rescued cancer cell growth in cystine-free conditions (Extended Data Fig. 4h). Cysteine deprivation is linked to ferroptosis, a non-apoptotic form of cell death that involves lipid peroxidation 31 . Notably, supplementation with ferrostatin-1 or trolox, both of which are radical-trapping antioxidants and potent inhibitors of ferroptosis, rescued survival but not proliferation under cystine-free conditions (Extended Data Fig. 5a–c). Unlike GSH, supplementation with these antioxidants did not supply cysteine, which suggests that preventing lipid peroxidation after cystine deprivation is not sufficient to sustain cancer cells. This idea was further supported when similar phenotypes were observed across ferroptosis-sensitive and ferroptosis-resistant cancer cell lines (Extended Data Fig. 5d–j). Next, metabolite levels were compared across cells cultured with the following medium: control (cystine supplemented); cystine-free; and cystine-free with GSH supplemented. Supplementation with GSH resulted in time-dependent accumulation of extracellular cysteinylglycine (Fig. 2g and Extended Data Fig. 6a), which suggests that there is catabolism of extracellular GSH by cancer cells. Levels of intracellular cysteinylglycine and cysteine were not rescued by GSH supplementation in cystine-free conditions (Extended Data Fig. 6b,c). However, this effect may be due to the large amount of cystine provided in the control conditions (208 µM), as reduction of cystine levels in the medium did not impair growth but abolished intracellular cysteine and cystine levels (Extended Data Fig. 6d–f). Notably, GSH supplementation rescued the levels of downstream cysteine-related products (Extended Data Fig. 6g–i) and prevented the accumulation of ophthalmic acid (Extended Data Fig. 6j), a metabolite that is produced by GCLC in the absence of cysteine 32 , 33 . To examine the incorporation of GSH-derived cysteine into downstream pathways, we used an inverse stable-isotope labelling approach 34 , 35 . Cancer cells were grown in medium with 13 C-cystine and then switched to cystine-free medium with unlabelled GSH (Fig. 2h). Cysteine and cysteine-dependent metabolites (glutamylcysteine, GSH and hypotaurine) were labelled when cells were grown in medium with 13 C-cystine and became unlabelled when cancer cells were grown in cystine-free medium with unlabelled GSH (Fig. 2i and Extended Data Fig. 6k,l). Similar findings were observed using 15 N 13 C-cystine and by performing high-resolution proteomics (Fig. 2j,k and Extended Data Fig. 6k,m). Moreover, the level of residual labelling of individual peptides positively correlated with their protein half-lives (measured independently in primary human fibroblasts 36 ) (Extended Data Fig. 6n,o). This result suggests that the residual cystine labelling reflects the slow turnover of these peptides. Finally, as intracellular GSH synthesis partially contributed to tGSH levels in the TIF, we explored the impact of blocking GSH export through drug transporters 37 . Pharmacological inhibition of drug transporters in breast cancer cells did not affect the ability of GSH to rescue tumour growth in cystine-free conditions (Extended Data Fig. 7a–d), even though it promoted the accumulation of GSH in cells (Extended Data Fig. 7e–h). However, further research is required to fully understand the interplay between GSH export from cancer cells and GSH levels in the TIF in vivo. Together, these findings indicate that extracellular GSH is broken down to fuel the intracellular metabolic and proteogenic processes in cancer cells.

GGT1 drives GSH catabolism and survival

Several enzymes have GGT activity 38 , and GGT1 is the most catalytically active isoform 39 . We used three separate approaches to examine the necessity of GGT1 to sustain GSH catabolism and cysteine supply to cancer cells. We generated GGT1 knockdown cell lines using CRISPRi approaches and GGT1 KO cell lines (polyclonal and single-cell clones) using CRISPR–Cas9 approaches. GGT1 knockdown cell lines had minimal GGT1 mRNA (Extended Data Fig. 8a) but still retained substantial GGT activity (Extended Data Fig. 8b) and grew in a similar manner to control cell lines when cultured under cystine-replete and cystine-depleted with GSH-supplemented conditions (Extended Data Fig. 8c). GGT1 KO cell lines showed decreased GGT activity (especially in GGT1 KO single-cell clones). However, in both models, cancer cell growth was rescued under cystine-free with GSH-supplemented conditions (Extended Data Fig. 8d–g). Cancer cells express multiple GGT isoforms (Extended Data Fig. 9a–d), but none of these isoforms displayed a dependency in cancer cell line genetic screens (Extended Data Fig. 9e). Expression of GGT isoforms varied across tumour subtypes, but in most cases, there was increased expression in tumours compared with normal tissue (Extended Data Fig. 9f,g). However, GGT isoforms were rarely found mutated in cancers (Extended Data Fig. 9h). Moreover, expression of Ggt1 in mouse tumours did not depend on Gclc expression or the site of tumour growth (Extended Data Fig. 9i). Further studies are required to better understand how GGT1, other GGT family members and uncharacterized proteins with GGT activity contribute to extracellular GSH catabolism. Separately, we speculated that GGT1 is sufficient to support the breakdown of GSH in the extracellular environment. Overexpression of GGT1 (GGT1 + ) in cells 40 resulted in increased GGT1 protein levels (Fig. 3a) and GGT activity (Fig. 3b,c). Lower levels of GSH were required to rescue GGT1 + cell growth in cystine-free conditions than for control cells (Fig. 3d), which suggests that catabolism by GGT is a rate-limiting step in GSH-dependent rescue of cancer cell growth. We proposed that GSH catabolism by cells with high GGT activity can support surrounding cells in a paracrine fashion. To test this hypothesis, we co-cultured GGT1 + cells with WT cells using a Transwell assay (Fig. 3e). Even though GSH levels were below the threshold for rescuing WT cell growth in cystine-free conditions, co-culturing with GGT1 + cells led to complete rescue of WT cell growth (Fig. 3f). Moreover, GGT1 + cells grew faster in cystine-free with GSH-supplemented conditions but not in cystine-replete conditions (Fig. 3g). Finally, this growth advantage translated to faster growth of GGT1 + cells in vivo (Fig. 3h,i). These findings demonstrate that GGT activity is sufficient to support GSH catabolism and survival of surrounding cells in cystine-depleted conditions. Furthermore, this result suggests that non-tumorigenic tissues or cells in the tumour microenvironment with high GGT activity can drive tumour growth and progression by catabolizing GSH and supplying amino acids in a paracrine (or endocrine) manner.

a, b, Immunoblot analyses of human GGT1 (a) and GGT activity (b) in WT and GGT1 + PC3 prostate cancer cells (n = 3 technical replicates from 2 independent experiments). For gel source data, see Supplementary Fig. 1. c, GGT activity measured by γ-4-methoxy-β-naphthylamide (GMNA) histochemical staining in WT and GGT1 + cells. Serine–borate was used to competitively inhibit GGT (negative control). Magnification of micrographs, ×10. d, WT and GGT1 + cells were grown for 72 h in control medium (208 µM cystine) or in cystine-free medium supplemented with the indicated GSH concentrations (n = 6 technical replicates from 3 independent experiments). e, Schematic of non-contact co-culture experiments using 0.4 μm PET membrane Transwell inserts in medium containing low concentrations of GSH (250 µM), which were insufficient to rescue the growth of WT cells in cystine-depleted conditions. f, Relative cell numbers of WT, GGT1 + and WT cells co-cultured with GGT1 + cells in control, cystine-depleted or cystine-depleted with GSH-supplemented (250 µM) conditions (n = 8 technical replicates for WT and GGT1 + , n = 6 for co-cultured cells, from 4 independent experiments). g, Relative cell numbers at the indicated time points for WT and GGT1 + cells grown in medium containing 200 µM cystine with 0 µM GSH (left) or 0 µM cystine with 250 µM GSH (right) (n = 3 technical replicates representative of 3 independent experiments). h, i, Tumour volume over time (h) and mass at end point (i) of xenograft tumours from WT (n = 26) and GGT1 + (n = 28) PC3 cells (n = 2 independent experiments). Significance was assessed by unpaired two-tailed t-test (b, i) or two-way ANOVA followed by Tukey’s multiple comparisons test (f) or Šídák’s multiple comparisons test (g, h). Data are the mean ± s.e.m.

GSH catabolism alters drug sensitivity

The metabolic environment can influence the sensitivity of cancer cells to anticancer compounds 41 , 42 . To investigate the impact of shifting from cystine-driven to GSH-driven cysteine supply in cancer cells in an unbiased manner, we performed a multifunctional approach to pharmacological screening (MAPS) with a library of 240 metabolic inhibitors (Fig. 4a,b). When cancer cells relied on GSH for cysteine acquisition, they were more sensitive to GGsTop 43 , a putative inhibitor of GGT activity (Fig. 4b,c). Conversely, cancer cells were less sensitive to inhibitors of cystine uptake (for example, erastin) and the thioredoxin pathway (for example, auranofin, aurothioglucose and PX-12) (Fig. 4b–d); this pathway mediates the reduction of cystine to cysteine after its import 44 . The decreased sensitivity of cancer cells to these inhibitors may be due to several reasons, including extracellular GSH binding to and inactivating the inhibitors. Alternatively, when cancer cells use GSH as a cysteine source, they rely less on these pathways (that is, xCT and TXNRD1). Moreover, the presence of GSH in the tumour microenvironment argues against the feasibility of targeting these pathways for cancer therapy. Notably, commonly used cell culture media (for example, DMEM, RPMI and F12) contain supraphysiological levels of cystine but lack GSH and cysteinylglycine. Consequently, the contribution of xCT to cancer cell growth may be overestimated, whereas the importance of GGTs may be undervalued.

a, Schematic of MAPS. HCC-1806 cells grown in control (208 µM cystine) or cysteine-free with GSH supplemented (500 µM) medium were treated with libraries of drugs, each arrayed at 10 dose points (20 µM–1 nM). After 96 h, cell numbers were determined, and dose–response curves were generated for each drug. b, MAPS results showing each drug ranked by the difference in the area under the curve (ΔAUC) obtained for each drug curve in the indicated conditions (n = 2 technical replicates). c, Schematic showing the targets of selected hits from b. d, HCC-1806 cells were grown in 6-well plates and treated with erastin and auranofin for 72 h in different culture media, and relative cell numbers were determined (n = 6 technical replicates from 3 independent experiments). Significance was evaluated by two-way ANOVA followed by Tukey’s multiple comparisons test. Data are the mean ± s.e.m.

GSH catabolism supports tumour growth

High-throughput screening of cancer cells identified a GGT inhibitor (GGsTop) with increased sensitivity when GSH was provided as a cysteine source. To further explore this effect, we examined additional GGT inhibitors (acivicin and OU749) 45 but found that GGsTop was the most potent at blocking GGT activity (Fig. 5a,b and Extended Data Fig. 10a). Cancer cells were sensitive to GGsTop when grown in cystine-free with GSH-supplemented conditions, but this increased sensitivity was reversed when cells were provided with cysteinylglycine, the product of GGT-mediated GSH catabolism (Fig. 5c,d). The kidney is suggested to have the highest levels of GGT1 (ref. 46 ), which was confirmed by an analysis of GGT activity in mouse tissues and expression of GGTs from public datasets (Extended Data Fig. 10b,c). Daily intraperitoneal injections of GGsTop, which has previously been shown not to impair tumour growth 47 , did not abolish kidney GGT activity (Extended Data Fig. 10d). However, intraperitoneal injections twice a day resulted in substantial inhibition of kidney GGT activity (Fig. 5e). Notably, GGsTop treatment blocked GGT activity and slowed tumour growth without causing any overt toxicity to animals (Fig. 5f–i and Extended Data Fig. 10e,f). Mechanistically, GGsTop treatment led to an accumulation of serum tGSH and a depletion of cysteine in tumours (Fig. 5j,k). Furthermore, it resulted in tumours with reduced levels of cysteine-dependent metabolites (hypotaurine) and increased levels of ophthalmic acid, which is synthesized in the absence of cysteine 32 , 33 (Fig. 5l,m). GGsTop treatment did not affect levels of GSH, GSSG or glutamate in tumours (Extended Data Fig. 10g–i). The impaired tumour growth induced by GGsTop treatment was rescued by supplementation with a cell-permeable source of cysteine (NAC) (Fig. 5n). Treatment of cells with both NAC and GGsTop did not rescue tumour cysteine levels but did rescue hypotaurine levels and partially suppressed ophthalmic acid levels (Fig. 5o–q and Extended Data Fig. 10j-o). GGsTop activity remained stable in water at room temperature (Extended Data Fig. 10p), and GGsTop delivery in drinking water led to inhibition of GGT activity in animals and suppressed tumour growth (Extended Data Fig. 10q–t). Together, these findings indicate that GGT activity maintains GSH catabolism to supply cysteine and support tumour growth. Furthermore, the data showed that blocking GGT is a potential therapeutic strategy for patients with cancer.

a, GGT activity in mouse kidney extracts treated with the indicated GGT inhibitors (n = 3 technical replicates from 3 independent experiments). b, GGT activity in HCC-1806 cells treated with GGsTop for 4 h (n = 2 technical replicates representative of 3 independent experiments). c, d, Relative cell numbers of HCC-1806 cells treated with GGsTop for 96 h in control or cystine-free medium supplemented with GSH (c) or CysGly (d) (n = 4 technical replicates from 2 independent experiments). e, GGT activity in kidney extracts of mice intraperitoneally (i.p.) injected with vehicle (n = 6 mice) or with 5 mg kg –1 of GGsTop every 12 h for 1 (n = 9), 2 (n = 9) or 3 days (n = 8). f, Tumour volume of orthotopic HCC-1806 xenografts in mice treated with vehicle (n = 15) or 5 mg kg –1 GGsTop (n = 18) every 12 h. Data are representative of two independent experiments. g, Tumour mass at the end point from f. h, i, GGT activity in the kidney (h) and tumours (i) from mice in at the end point (vehicle, n = 7; GGsTop, n = 9). j, Serum GSH (N-ethylmaleimide (NEM)-GSH) levels from mice in f (n = 7). k–m, Relative abundance of cysteine (k), hypotaurine (l) and ophthalmic acid (m) from tumours in f (vehicle, n = 9; GGsTop, n = 10). n, Tumour volume of orthotopic HCC-1806 xenografts treated with vehicle or 5 mg kg –1 GGsTop every 12 h alone or supplemented with NAC (30 mM) in drinking water (vehicle, n = 5; GGsTop, n = 8; vehicle with NAC, n = 5, GGsTop with NAC, n = 7, representative of 2 independent experiments). o–q, Relative abundance of cysteine (o), hypotaurine (p) and ophthalmic acid (q) from tumours in n (vehicle, n = 8; GGsTop, n = 7; vehicle with NAC, n = 8, GGsTop with NAC, n = 9, from 2 independent experiments). Significance was assessed by ordinary two-way ANOVA followed by Šídák’s multiple comparisons test (f), unpaired two-tailed t-test (g–m, o–q) or ordinary two-way ANOVA (n). Data are the mean ± s.e.m.

Discussion

A prevailing dogma is that a major fate of cysteine is its incorporation into the antioxidant GSH. Here we demonstrated that the opposite is also true: extracellular GSH is a precursor to intracellular cysteine. Moreover, we showed that the downstream fate of cysteine in tumours is not necessarily GSH. Notably, the affinity of cysteine for cysteinyl-tRNA synthetase is orders of magnitude higher than its affinity for the rate-limiting enzyme in GSH synthesis (GCLC) 48 , 49 . This result highlights that rather than synthesizing GSH, cysteine has other downstream fates of greater importance, such as protein translation or cysteine polysulfidation 50 .

GSH levels were enriched in the TIF compared with serum in mouse models of cancer and in human patients with cancer. Although tumours themselves contributed to GSH abundance in the TIF, additional cell types probably also contribute. For example, after activation, macrophages can efflux GSH 51 . Moreover, GSH levels and GSH to GSSG ratios can vary across mouse strains 52 , and therefore could also affect cysteine supply to tissues (and tumours) in these mice. Finally, we observed lower serum GSH levels in mice with tumours than in mice without. Past studies have shown that serum GSH levels are lower in patients with colorectal cancer than in healthy individuals 53 . Further research is required to determine the utility of serum GSH measurement as a predictor of cancer incidence.

Our findings demonstrated that GSH catabolism by GGTs supports cancer cell growth and survival. A question that remains to be answered is whether GSH catabolism occurs primarily in tumours or in non-tumorigenic tissue (for example, kidney). Furthermore, less is known about whether a specific subtype of cancer relies more on GSH catabolism. Renal cell carcinomas (papillary and clear cell) express some of the highest levels of GGT1. By contrast, another subtype of renal cell carcinoma (chromophobe) contains the lowest GGT1 expression levels among tumours. Indeed, GGT1 has been studied in the context of kidney cancer, with pro-tumour and antitumour effects being attributed 54 , 55 , 56 . Even less is known regarding its potential role in tumour metastasis and the potential side effects of GSH catabolism in specific tissues (for example, release of glutamate in brain metastasis). Numerous additional questions remain regarding the catabolism of metabolites and their impact on tissue homeostasis in disease.

Following its release from GSH, cysteinylglycine must be further broken down into the individual amino acids cysteine and glycine. Dipeptidases are predicted to control the breakdown of dipeptides, including cysteinylglycine; however, the exact enzymes involved are poorly understood. Carnosine dipeptidase II (CNDP2) has been described to support the catabolism not only of cysteinylglycine 57 but also of other dipeptides, such as those containing glutamine 35 . Future studies involving a systematic analysis of dipeptidase–dipeptides relationships, in malignant and non-malignant tissues, are required.

Targeting the supply of amino acids, including cysteine, to tumours is an emerging area of therapeutic research. Substantial attention has been paid to blocking cystine uptake 58 or cysteine production from methionine through the transsulfuration pathway 59 . Previous studies that have measured the incorporation of cysteine into metabolites suggested that additional sources of cysteine for tumours, beyond extracellular cystine or de novo synthesized cysteine, probably exist 16 . We examined the importance of a non-intuitive source of cysteine: that is, GSH catabolism by GGTs. We showed that blocking GGT activity slows tumour growth by depriving it of intracellular cysteine. The therapeutic window for inhibiting GGT as a cancer therapy is potentially large, as sustained inhibition of GGT activity did not induce any overt toxicities. Degradation of extracellular GSH, in a similar approach to degradation of extracellular cystine or cysteine 9 , is an additional potential approach. Beyond a target for therapeutic intervention, monitoring GSH catabolism holds value as a cancer diagnostic. Indeed, serum GGT activity is a routine clinical measurement and is a risk factor for cancer 60 . Further research is required to fully elucidate the therapeutic potential surrounding GSH catabolism and the supply of amino acids to cancers.

Methods

Animal studies

All animal studies were performed according to protocols approved by the University Committee on Animal Resources at the University of Rochester Medical Center. Gclc f/f mice 27 were crossed with the MMTV-PYMT (The Jackson Laboratory, 022974) 61 and Rosa26 creERT2 (The Jackson Laboratory, 008463) 62 mouse strains. For the tumour allograft model, Gclc f/f Rosa26 creERT2 MMTV-PYMT mice and control C57BL/6 mice (The Jackson Laboratory, 000664) were used (female, at least 12 weeks old). For tumour xenograft experiments, athymic nude NU/J mice (The Jackson Laboratory, 002019) were used (female, at least 8 weeks old).

For the tumour allograft model, Gclc f/f Rosa26 creERT2 MMTV-PYMT mice were euthanized and autochthonous tumours were collected, cut into 8 mm 3 fragments and cryopreserved in liquid nitrogen. These fragments were later thawed on ice and implanted into the right fourth mammary fat pad of immunocompetent C57BL/6 mice. Once the grafted fragments had formed fully established tumours, they were collected, cut into 8 mm 3 fragments and cryopreserved using the same procedure. Finally, these tumour fragments were thawed and grafted into recipient immunocompetent C57BL/6 mice and used for the designated experimental treatments. To activate Cre recombinase and induce Gclc deletion in tumour pieces from Gclc f/f Rosa26 creERT2 MMTV-PYMT mice, female mice were intraperitoneally injected with tamoxifen (Sigma-Aldrich, T5648) at 50–160 mg kg –1 once daily for 5 consecutive days. For tumour xenograft models, 1 × 10 6 HCC-1806 cells or 5 × 10 6 PC3 cells were injected into the right fourth mammary fat pad or hind flank of athymic nude NU/J mice (The Jackson Laboratory, 002019) in 100 µl sterile PBS or a 1:1 mixture of cell suspension in PBS and Matrigel (Corning, 356237) using a 29-gauge 1 ml insulin syringe. In all cases, animals received only one cell injection at a single location per experiment.

Tumour volume was estimated using the formula for an oblate spheroid v=(pi )/(6)* (l 2 )* w, where volume (v) is calculated using caliper measurements of the longest (l) and shortest (w) sides of the tumour. For pharmacological interventions, treatment of mice was initiated once the average tumour volume in the groups reached approximately 300 mm 3 . Mice were treated with pharmacological agents by either intraperitoneal (i.p.) injection or inclusion in the drinking water. For treatment via i.p. injection, GGsTop (Tocris, 4452; WuXi, 926281-37-0; or MCE, HY-108467) was diluted in sterile saline and injected at the indicated concentrations. For treatment via drinking water, GGsTop (WuXi, 926281-37-0) was diluted to 62.5 µg ml –1 (189 µM) and provided to mice ad libitum. Based on an average daily water consumption of 4 ml, the estimated daily dose of GGsTop for a 25 g mouse was 10 mg kg −1 per day (equivalent to 5 mg kg −1 twice a day i.p. injections). NAC (Sigma-Aldrich, A7250) was diluted in drinking water at a concentration of 30 mM, the pH was adjusted to 7.00 and it was provided to mice ad libitum.

Mice of desired strains were age-matched and assigned randomly to their treatment groups. For xenograft studies, animals were allocated into groups ensuring the mean, median and standard error of tumor size was similar across all groups. Investigators were not blinded to group allocation during experiments owing to technical limitations. Statistical methods were not used to chose sample sizes. The number of animals assigned per condition was selected to account for the variability of the examined phenotypes based on pilot experiment and past experience with the animal models (≥6 animals per experimental condition). Mice were euthanized at humane end points (when the tumour diameter exceeded a total of 20 mm in length, when ulceration occurred or when weight loss exceeded 20%). Tumours and tissues were snap-frozen on dry ice and stored at –80 °C or placed in 10% neutral buffered formalin (Fisher Scientific, 22-110-761) for 24 h and then stored in 100% methanol (Fisher Scientific, A412P-4).

Immunohistochemistry

Formalin-fixed, paraffin-embedded tissue sections (5 µm) were used for haematoxylin and eosin staining and immunohistochemical analyses. The tissues were dewaxed and rehydrated through a series of xylene and ethanol changes. For antibody staining, antigen retrieval was performed on the slides by incubating them in a steamer for 40 min in citrate buffer (Vector Labs, H-3300-250). Tissues were permeabilized with PBS supplemented with 0.1% Tween-20 for 10 min, followed by a peroxidase block with 3% H 2 O 2 (Sigma-Aldrich, 216763) for 20 min. The slides were washed in PBS and then blocked using 10% goat serum in PBS for 1 h at room temperature before adding primary antibodies diluted in the blocking buffer. Primary antibody (anti-GCSc 1:100 (SantaCruz, sc390811), anti-CD45 1:100 (Santa Cruz, sc1178), anti-F4/80 1:100 (Cell Signaling Technology, 70076), anti-DNA/RNA Damage 1:200 (Abcam, 62623), anti-NRF2 1:100 (Abcam, 31163) and anti-GGT1 1:200 (donated from M. H. Hannigan) 63 ) incubation was carried out overnight at 4 °C. Biotinylated goat anti-mouse IgG or goat anti-rabbit IgG (Vector Labs, BA-9200 1:200 and BA-1000 1:200, respectively) secondary antibodies were added in blocking buffer and incubated for 1 h at room temperature. Stains were developed using VectaStain Elite ABC-HRP Peroxidase and DAB Substrate kits (Vector Labs, PK-7100 and SK-4100, respectively). Tissues were counterstained with haematoxylin, mounted with Permount mounting medium (Fisher Scientific, SP15500) and coverslipped for imaging. Images were taken using an Olympus VS120 virtual slide microscope and Visiopharm image analysis system.

Immunofluorescence

Formalin-fixed, paraffin-embedded tissue sections (5 µm) were used for immunofluorescence. The tissues were dewaxed and rehydrated through a series of xylene and ethanol changes. For antibody staining, antigen retrieval was performed on the slides by incubating them in a steamer for 40 min in citrate buffer (Vector Labs, H-3300-250). The slides were washed in water and then blocked using 10% goat serum in PBS for 1 h at room temperature before adding primary antibodies diluted in the blocking buffer. Primary antibody (anti-GCSc 1:10 (SantaCruz, sc390811) and anti-CD45 1:200 (Proteintech, 31243-1-AP)) incubation was carried out overnight at 4 °C. Secondary antibodies (AlexaFluor 594 anti-mouse 1:500 (Invitrogen, A11005) and AlexaFluor 488 anti-rabbit 1:500 (Invitrogen, A11034)) were added to blocking buffer and incubated at room temperature for 1 h. For immunofluorescence, the second-to-last wash included 1 µg ml –1 4′,6′-diamidino-2-phenylindole dihydrochloride (DAPI; Sigma-Aldrich D9542) in PBS to stain cell nuclei. Immunofluorescence tissues were mounted with Immuno-Mount (Thermo Fisher, 9990402) and coverslipped for imaging. Immunofluorescence-stained slides were imaged using a Zeiss Axioimager.M2m with an AxioCam MRm camera and the AxioVision program for imaging.

Cell culture

Cell lines were maintained in RPMI 1640 (Thermo Fisher, 11875119) with 5% FBS (Sigma-Aldrich, SH30396.03) and 1% penicillin and streptomycin (Thermo Fisher, 15070063). Cell lines were obtained from the American Type Culture Collection (ATCC), the National Cancer Institute Division of Cancer Treatment and Diagnosis (NCI-DCTD) or donated from J. Brugge, S. Mello, B. Altman and P. Rappold. PC3 wild-type and PC3 GGT1 + cell lines were a gift from M. Hannigan 40 . The authenticity of the HCC-1806, MDAMB231, MDAMB468, JIMT-1 and A549 cell lines was confirmed by STR profiling. The authenticity of the remaining cell lines was not tested. All cell lines tested negative for mycoplasma contamination using a MycoAlert Mycoplasma Detection kit (Lonza, LT07-418). For experiments involving alterations to the culture medium, cells were seeded in regular RPMI 1640 with 5% dialysed FBS (Gibco, 26400044) and 1% penicillin and streptomycin. RPMI 1640 medium without l-glutamine, l-cysteine, l-cystine and l-methionine was used as the base medium for different medium conditions (MP Biomedicals, 1646454). l-glutamine (Fisher Scientific, 25030081), l-methionine (Sigma-Aldrich, M5308), l-cystine (Sigma-Aldrich, C7602), GSH (Sigma-Aldrich, G4251) or cysteinylglycine (Bachem, 4002969) was added to each condition at the indicated concentrations. GGsTop (Tocris, 4452; WuXi, 926281-37-0; or MCE, HY-108467), acivicin (MCE, HY-W016586), OU749 (Cayman, 13804), auranofin (MCE, HY-B1123), erastin (MCE, HY-15763), vincristine sulfate (MCE, HY-N0488), ferrostatin-1 (Fer-1; MCE, HY-100579), RSL3 (MCE, HY-10018A) and MK-571 (MCE, HY-19989) were diluted in dimethyl sulfoxide (DMSO; Sigma-Aldrich, 472301) at 10–20 mM and added at the indicated concentrations. dl-Buthionine-(S,R)-sulfoximine (BSO; MCE, HY-106376) and trolox (Sigma-Aldrich, 238813) were diluted in water at 100 mM and added at the indicated concentrations. Cells were seeded at 50,000 per well in 6-well plates and at 10,400 per well in 24-well plates. After allowing the cells to attach for 24 h, the medium was aspirated, the wells were washed once with PBS and the indicated medium was added. For experiments in 384-well plates, 500 cells were seeded per well, and after 24 h, the medium was switched using a series of 4 aspiration–wash steps with the desired final medium using Multidrop Combi (Thermo). Co-culture experiments with PC3 WT and GGT1 + cell lines were performed using 6-well (24 mm) PET inserts with 0.4 µm pores (VWR, 76313-902 or Corning, 354570). For these experiments, cells were seeded at the same density (12,992 cells per cm 2 ) in both the bottom well and the insert. For cell number measurements in 6-well plates, the medium was aspirated, the wells were rinsed with PBS and the cells were detached by incubating with 500 µl of 0.25% trypsin (Thermo Scientific, 25200056) for 20 min at 37 °C. Trypsin was neutralized with 500 µl complete medium, the cells were resuspended and 100 µl of the suspension was mixed with 6.9 ml Beckman Coulter Isoton II diluent (Fisher Scientific, NC2470899) for counting with a Beckman Coulter Z1 Coulter Cell Particle counter. For cell number measurements in 24-well plates, unstained plates were scanned in bright field using SparkCyto (Tecan) at 37 °C under 5% CO 2 . For experiments in 384-well plates, wells were aspirated, rinsed with PBS and incubated with a fixative and staining solution containing 3.7% formaldehyde (Fisher Scientific, F75P-1GAL) and 5 µg ml –1 Hoechst 33342, trihydrochloride, trihydrate (Invitrogen, H1399). After 30 min, wells were aspirated, washed with PBS and plates were sealed with adhesive foil (VWR, 60941-124). Plates were imaged on a CellInsight CX5 HCS platform (Thermo), and counts of nuclei were used as the readout.

Generation of KO and knockdown cell lines

Optimized sgRNA oligonucleotides 64 were cloned into lentiCRISPRv2 (ref. 65 ) (Addgene, 52961) for CRISPR–Cas9 KO or into pLV hU6-sgRNA hUbC-dCas9-KRAB-T2a-Puro 66 (Addgene, 71236) for CRISPRi knockdown, as previously described 41 . HEK293T cells were transfected with 1 µg lentiCRISPRv2 containing the sgRNA insert, 0.67 µg psPAX2 (Addgene, 12260) and 0.33 µg pVSVg (Addgene, 8454) using 6 µg polyethylenimine (Sigma-Aldrich, 919012) in Opti-MEM (Thermo Fisher, 31985062). Supernatants containing lentiviral particles were collected and replaced with fresh medium daily for 3 consecutive days. The pooled lentiviral supernatant was filtered through 0.45 µm PVDF syringe filters (Whatman, 6746-2504), aliquoted and stored at –80 °C. For transduction, 200 µl lentiviral supernatant was thawed on ice and added to HCC-1806 cells together with 8 mg ml –1 Polybrene (hexadimethrine bromide; Sigma-Aldrich, H9268) in 6-well plates. Cells were then allowed to recover for 24 h in fresh medium before antibiotic selection with 2 µg ml –1 puromycin for at least 4 days, after which they were expanded or used for subsequent experiments. To isolate single-cell clones, antibiotic-selected polyclonal GGT1 KO cells were seeded at low density (50 cells per 15 cm dish) and grown for 21–28 days. Then colonies were individually collected by trypsinization using 8 mm cloning cylinders (Pyrex, 09-552-21) and transferred to individual wells for expansion. The following sgRNAs for GGT1 CRISPRi were used: sgGGT1 1 (HL-ID: 3) forward: 5′-CACCGGTGCCTCCCACTGTCCGCCG-3′; sgGGT1 1 (HL-ID: 3) reverse: 5′-CACCGGTCTGGGCTCACCCGACGCC-3′; sgGGT1 2 (HL-ID: 5) forward: 5′-AAACCGGCGGACAGTGGGAGGCACC-3′; and sgGGT1 2 (HL-ID: 5) reverse: 5′-AAACGGCGTCGGGTGAGCCCAGACC-3′. The following sgRNAs for GGT1 CRISPR–Cas9 KO were used: sgGGT1 1 (HL-ID: 6) forward: 5′-CACCGTGCGGGACGGTGGCTCTG-3′; sgGGT1 1 (HL-ID: 6) reverse: 5′-AAACCAGAGCCACCGTCCCGCAC-3′; sgGGT1 2 (HL-ID: 10) forward: 5′-CACCGCCCAGTGCGCCGCTCAG-3′; and sgGGT1 2 (HL-ID: 10) reverse: 5′-AAACCTGAGCGGCGCACTGGGC-3′. The following sgRNAs for GCLC CRISPR–Cas9 KO were used: sgNTC forward: 5′-CACCGGAGGCTAAGCGTCGCAA-3′; sgNTC reverse: 5′-AAACTTGCGACGCTTAGCCTCc-3′; sgGCLC-a forward: 5′-CACCGCATACTCACCTGAAGCGA-3′; sgGCLC-a reverse: 5′-AAACTCGCTTCAGGTGAGTATGC-3′; sgGCLC-b forward: 5′-CACCGAAATATCCGACATAGGAG-3′; and sgGCLC-b reverse: 5′-AAACCTCCTATGTCGGATATTTC-3′.

Flow cytometry

To quantify cell proliferation, 300,000 cells were plated in 100 mm dishes and treated as specified. At the end point, attached cells were pulsed with 10 mM BrdU (Sigma-Aldrich, B9285) for 30 min in their culture medium. Labelled cells were then detached by trypsinization, pelleted, washed once with PBS, resuspended in 300 µl ice-cold PBS and fixed by adding 700 µl ice-cold absolute ethanol while gently vortexing. Fixed cells were stored at –20 °C. For labelling, the fixed cells were thawed and washed for 5 min with 1 ml of 1% BSA (Sigma-Aldrich, A7906) in PBS. The cells were then centrifuged at 10,000g for 2 min and incubated with 1 ml of 2 M HCl and 0.5% Triton X-100 (Sigma-Aldrich, 93443) for 30 min with gentle agitation. Then the cells were centrifuged and washed once with 1 ml of 0.1 M borax (Sigma-Aldrich, 71997), followed by another wash with 1% BSA in PBS. After centrifugation and thorough removal of the supernatant, the pellets were incubated for 1 h in 60 µl staining solution (1% BSA and 0.5% Tween-20 in PBS) containing 3 µl anti-BrdU–FITC monoclonal antibody (BioLegend, 364104, clone 3D4). The samples were subsequently washed with 1 ml of 1% BSA and 0.5% Tween-20 in PBS, centrifuged and the pellets were resuspended in 1 ml of PI solution (20 µg ml –1 propidium iodide (Sigma-Aldrich, P4170), 0.5 mg ml –1 RNase A (Sigma-Aldrich, R5503) and 0.1% Triton X-100 in PBS). The cells were incubated overnight (approximately 16 h), and then 10,000 events were analysed by flow cytometry using a BD Accuri C6 Plus flow cytometer (BD Biosciences) and FCS Express 7 Research (DeNovo Software).

For quantifying cell death, 50,000 HCC-1806 cells were seeded into 6-well plates and treated as indicated. At the end point, cells were collected by trypsinization (the culture medium and PBS washes were collected and re-added to each sample) and washed once with assay buffer (10 mM HEPES, 140 mM NaCl and 3.3 mM CaCl 2 , pH 7.4). Then the pellets were stained for 20 min in 100 µl assay buffer containing 5 µg ml –1 propidium iodide (Sigma-Aldrich, P4170) and 3 µl Annexin V–FITC (BioLegend, 0640906). Then cells were diluted with 0.5–1 ml assay buffer, and 10,000 events were analysed by flow cytometry using a BD Accuri C6 Plus flow cytometer (BD Biosciences) and FCS Express 7 Research (DeNovo Software).

For quantifying lipid peroxidation, 95,000 cells were seeded into 12-well plates and treated as indicated. At the end point, 2.5 µM Bodipy 581/591 C11 (Cell Signaling, 95978) was added directly to each well and incubated for 30 min. Wells were then washed with PBS and collected by trypsinization. Pellets were resuspended in 300 µl PBS, and 10,000 events were analysed by flow cytometry using a BD Accuri C6 Plus flow cytometer (BD Biosciences) and FCS Express 7 Research (DeNovo Software).

Immunoblot assays

Cell lysates from HCC-1806, PC3 WT and PC3 GGT1 + cells were obtained by placing PBS-washed cell plates on ice and scraping with RIPA buffer (Thermo Scientific, 89900) containing Halt Protease and Phosphatase Inhibitor cocktail (Thermo Scientific, 1861280). Cell lysates were incubated for 30 min on ice and then centrifuged at 12,000g for 15 min at 4 °C. The supernatant was collected and stored at –80 °C. Extracted proteins were quantified using a Pierce BCA Protein Assay kit (Thermo Scientific, 23225). Lysates were combined with 6× Laemmli SDS sample buffer (Boston BioProducts, BP-111R) containing 5% 2-mercaptoethanol (VWR Life Science, M131), then loaded onto 4–20% Criterion TGX pre-cast gels (Bio-Rad, 5671093). Each lane received 40 µg protein for HCC-1806 samples or 75 µg for PC3 samples. Separated proteins were transferred onto Immobilon-P Transfer membranes (MilliporeSigma, IPVH00010), blocked for 1 h using 5% BSA in TBS–0.5% Tween-20 and stained overnight with primary rabbit-anti-GCLC (Sigma-Aldrich, HPA036359) or affinity-purified rabbit-anti-GGT1 (GGT129, as previously described 63 ) 1:1,000 in BSA 2.5% overnight at 4 °C. Stained membranes were washed three times with TBS-T and stained with HRP-linked secondary antibody donkey-anti-rabbit (Amersham, NAV934) 1:5,000 in BSA 2.5% for 1 h. Membranes were washed three more times with TBS-T, and then the antibody-stained protein signal was amplified and visualized using SuperSignal (Thermo Fisher, 34577). Blots were imaged with a ChemiDoc MP Imaging system (Bio-Rad). Membranes were stripped for 20 min at room temperature using a Blot Restore Membrane Rejuvenation kit (Sigma-Aldrich, 2520-M), then washed twice with TBS-T, re-blocked with BSA 5% for 30 min and incubated with mouse-anti-β-actin (Sigma-Aldrich, A1978) 1:3,000 in BSA 2.5% for 4 h at room temperature and then incubated with a sheep-anti-mouse (Amersham, NA931) secondary antibody at 1:5,000 in BSA 2.5% for 2 h.

Drug screening

MAPS was performed as previously described 41 . In brief, 500 cells were seeded per well in 384-well plates, and after 24 h, the medium was switched using a series of 4 aspiration–wash steps with the desired final medium. Compound libraries were gifts from J. Brugge and the ICCB-L. Next, 100 nl of compounds were transferred using a JANUS MDT Workstation (Revvity) and a 384-well pin tool (V&P Scientific). After 96 h, wells were aspirated, washed with PBS, and a fixative/staining solution containing 3.7% formaldehyde (Fisher Scientific, F75P-1GAL) and 5 µg ml –1 Hoechst 33342, trihydrochloride, trihydrate (Invitrogen, H1399) was added. After 30 min, wells were aspirated, PBS was added and plates were sealed with adhesive foil (VWR, 60941-124). Plates were imaged using a CellInsight CX5 HCS platform (Thermo), and nuclei were identified as a readout of cell counts. Data post-processing was conducted using R scripts.

GGT activity assay

Cell cultures were washed and scraped with PBS. The use of trypsin for cell detachment was avoided to prevent proteolytic degradation of cell-surface GGT. Detached cells were collected and centrifuged at 300g for 5 min at room temperature. Then pellets were resuspended in lysis buffer (Tris-HCl 100 mM and 0.5% Triton X-100, pH 8) containing Halt Protease and Phosphatase Inhibitor cocktail (Thermo Scientific, 1861280), thoroughly vortexed for 1 min and incubated for 15 min on ice. Lysates were cleared by centrifugation at 12,000g for 15 min at 4 °C and the supernatants were stored at –80 °C until analysis. Tissues were lysed by transferring them to pre-filled bead mill tubes (Fisher Scientific, 15-340-154) containing lysis buffer (Tris-HCl 100 mM and 0.5% Triton X-100, pH 8) containing Halt Protease and Phosphatase Inhibitor cocktail and homogenized using a bead mill (VWR) for 10 s. Tissue lysates were then rotated for 15 min at 4 °C and centrifuged at 12,000g for 15 min at 4 °C. Supernatants of cells or tissues were stored at –80 °C until analysis. Extracted proteins were quantified using a Pierce BCA Protein Assay kit. GGT reaction assays were carried out by using the GpNA method 67 . In brief, cell or tissue protein lysates were assayed in transparent 96-well plates (Greiner, 655101) in the presence of 1 mM of the GGT substrate l-glutamic acid γ-p-nitroanilide (GpNA; Cayman, 36209) and 20 mM of the transpeptidation acceptor glycyl-glycine (Sigma-Aldrich, G1002) in Tris-HCl 100 mM, pH 8, buffer. GGT catalyses the breakdown of the γ-glutamyl bond in the substrate and generates p-nitroanilide (pNA), which is monitored kinetically by an increase in absorbance at 418 nm. P-nitroaniline (Sigma-Aldrich, 185310) was used to generate a standard curve (0–40 nmoles) for the calculation of GGT activity. Absorbance (418 nm) was assessed using a plate reader set at 37 °C with a total reading time of 45–60 min and 1–5 min intervals between reads. GGT activity was calculated by obtaining the slope of the linear range of each sample (Abs per min) and dividing by the slope of a 4-nitroaniline standard curve (Abs per nmoles pNA) and by the protein mass (mg) to determine the reaction rate of the sample (nmoles pNA min −1 mg −1 ). The optimal protein mass for the GGT assay was experimentally determined for each cell line or mouse tissue (HCC-1806 lysates or tumour xenografts: 200 µg; PC3 cell line: 100 µg; PC3 GGT1 + : 1 µg; kidney: 0.5 µg; spleen: 150 µg; liver: 250 µg; seminal vesicle: 10 µg; pancreas: 2.5 µg; lungs: 2.5 µg; muscle: 200 µg; testis: 50 µg; epididymis: 10 µg; PyMT breast tumour: 50 µg; heart: 150 µg; mammary fat pad: 150 µg; brain: 80 µg; large intestine: 80 µg; brown adipose tissue: 300 µg).

GGT histochemical assay

Histochemical staining of GGT was performed as previously described 68 , with modifications. Cells were seeded into 6-well plates and cultured in complete medium. At the end point, cells were washed 3 times with saline and stained for 20 min at room temperature using a saline solution containing 0.2 mM l-glutamic acid γ-(4-methoxy-β-naphthylamide) (GMNA; Sigma-Aldrich, G0141), 20 mM glycyl-glycine (Sigma-Aldrich, G1002), 25 mM Tris (Sigma-Aldrich, 93362) and 1.2 mM FastBlue (Sigma-Aldrich, F3378). For negative controls, the staining solution included 5 mM serine (Sigma-Aldrich, S4311) and 10 mM sodium borate (Sigma-Aldrich, 71997). After incubation, cells were rinsed with saline and incubated with 100 mM CuSO 4 (Sigma-Aldrich, C1297) for 2 min. Following a final saline wash, 50% glycerol was added to the wells, and bright-field images were taken using an inverted microscope.

GSH quantification

For some experiments, the GSH concentration was determined using a GSH-Glo Glutathione Assay (Promega, V6912). In brief, 6,250 cells were seeded into 96-well white cell culture-treated plates (Falcon, 353296), treated for 24 h as indicated and labelled following the manufacturer’s instructions. The luminescence of samples and of the GSH standard curve was detected using a Spark Multimode microplate reader (Tecan).

RNA analysis

mRNA was isolated from cells or tissues using an E.Z.N.A. total RNA kit I (Omega Bio-Tek, R6834-02). For gene expression analysis, 1 µg RNA was used for cDNA synthesis using a qScript cDNA Synthesis kit (Quanta Bio, 66196756). The expression of target genes was analysed via quantitative real-time PCR with a QuantaStudio 5 qPCR machine (Applied Biosystems, Thermo Fisher Scientific).

Mouse serum analysis

Mice were anaesthetized with isoflurane, after which blood was collected via the retro-orbital venous sinus into BD microtainer tubes (BD, 365967). Serum was isolated from the blood by centrifuging blood samples at 10,000g for 5 min and stored at –80 °C. VRL Animal Health Diagnostics carried out the analysis of biomarkers of liver damage.

Metabolite analysis

For metabolite analysis of cultured cancer cells, cells were seeded in 6-well plates and treated as specified. At the end point, cells were rapidly washed with ice-cold PBS and incubated at –80 °C for 30 min with extraction solvent at a ratio of 1 ml per 1 × 10 6 cells. The cells were then scraped on ice, transferred to pre-chilled microcentrifuge tubes, centrifuged at 15,000g for 20 min and the supernatants were collected. Conditioned medium was extracted by taking 10 µl of the medium supernatant and mixing it with 390 µl of the extraction solvent. Control medium samples were generated by incubating the treatment medium in cell-free 6-well plates under the same conditions and duration as the cell culture samples. Extracts of the medium were incubated at –80 °C for at least 15 min before being centrifuged at 15,000g for 20 min. All extracts were stored at –80 °C. Metabolite profiling of reduced thiol-containing metabolites was carried out in the presence of NEM. The data shown reflect the detection of their corresponding NEM adducts (for example, NEM-cystine, NEM-cysteinylglycine and NEM-GSH). tGSH was calculated as the sum of reduced GSH and two times the amount of oxidized glutathione (tGSH = GSH + 2 × GSSG). The extraction solvent consisted of 80% methanol and 20% aqueous solution (pH 7.00) containing 10 mM ammonium formate (Sigma-Aldrich, 70221-25G-F) and 25 mM NEM (Thermo Scientific, 040526.06). The final concentrations of ammonium formate and NEM in the extraction solvent were 2 mM and 5 mM, respectively. For inverse tracing experiments, HCC-1806 cells were seeded in T75 flasks and cultured for >7 days in cystine-free medium supplemented with heavy isotope-labelled 13 C 2 -cystine (Cambridge Isotope Laboratories, CLM-520-0). At the end of this >7-day labelling period (t = 0 h), samples were collected to verify complete incorporation of the heavy label into cellular metabolites. The remaining cells were collected by trypsinization, washed three times with PBS and seeded into 6-well plates using cystine-free medium containing unlabelled GSH. After 3 days in the presence of unlabelled GSH (t = 72 h), cells were extracted for metabolomic analysis as described above.

For quantification of sulfur-containing metabolites in cultured cancer cells and mouse serum, 10 µl samples were extracted with 90 µl of ice-cold extraction solvent (80% methanol and 20% H 2 O) containing isotope-labelled internal standards (36 µM 13 C 2 15 N-GSH, 0.92 µM 13 C 4 15 N 2 -GSSG and 10 µM D 4 -cystine). Following 30 min of incubation at 4 °C, extracts were cleared by centrifugation (17,000g, 20 min, 4 °C) and then supernatants were analysed by liquid chromatography and mass spectrometry (LC–MS). For the quantification of these metabolites in TIF samples, 5 µl samples were extracted in 45 µl ice-cold extraction solvent (75% acetonitrile, 25% methanol and 0.1% formic acid) containing isotope-labelled internal standards (36 µM 13 C 2 , 15 N-GSH, 0.92 µM 13 C 4 , 15 N 2 -GSSG and 10 µM D 4 -cystine). Extracted samples were vortexed at 4 °C for 10 min, cleared by centrifugation (17,000g, 10 min, 4 °C) and then supernatants were analysed by LC–MS. For chromatographic metabolite separation, a Vanquish ultra-performance liquid chromatography (UPLC) system was coupled to a Q Exactive HF (QE-HF) mass spectrometer equipped with heated electrospray ionization (HESI; Thermo Fisher Scientific). Samples were run on an Atlantis Premier BEH Z-HILIC VanGuard FIT column (2.5 µm, 2.1 mm × 150 mm, Waters). Mobile phase A was 10 mM (NH 4 ) 2 CO 3 and 0.05% NH 4 OH in H 2 O, whereas mobile phase B was 100% acetonitrile. The column chamber temperature was maintained at 30 °C. The mobile phase parameters were set according to the following gradient: 0–13 min of 80% to 20% of mobile phase B; 13–15 min of 20% of mobile phase B. The ESI ionization mode was negative, and the MS scan range (m/z) was set to 65–975. The mass resolution was 120,000 and the automatic gain control (AGC) target was 3 × 10 6 . The capillary temperature and capillary voltage were maintained at 320 °C and 3.5 kV, respectively. A volume of 5 μl of each sample was loaded for metabolite detection. The LC–MS metabolite peaks were manually integrated and identified using El-Maven (v.0.12.0) by matching with a previously established in-house library 33 .

For metabolic analysis of cultured cancer cells and tumours, frozen tumour tissue was homogenized in 80% methanol containing 20 mM NEM using a Precellys cold tissue homogenizer (Bertin) at a ratio of 20 mg tissue to 1 ml solvent. Following homogenization, samples were transferred to –80 °C for 30 min and then placed on regular ice for 30 min with vortexing every 10 min. Next, samples were centrifuged at 17,000g for 10 min and 800 µl supernatant was dried in a vacuum evaporator (Thermo). Samples were reconstituted in 90 µl of 50% acetonitrile (Fisher Scientific, A955) and transferred to glass vials for LC–MS analysis. For mouse serum collection, the animals were anaesthetized with isoflurane, and blood was collected via the retro-orbital venous sinus into BD microtainer tubes (BD, 365967). Serum was obtained by centrifuging blood samples at 10,000g for 5 min and stored at –80 °C. Then 20 µl serum was mixed with 2 µl 200 mM NEM (Thermo Scientific, 040526.06) and extracted with 80% methanol. Next, 900 µl serum extract was dried in a vacuum evaporator (Thermo), reconstituted in 90 µl of 50% acetonitrile (A955, Fisher Scientific) and transferred to glass vials for LC–MS analysis. LC–MS analysis was carried out by the URMC Metabolomics Resource. For LC–MS analysis, metabolite extracts were analysed by high-resolution mass spectrometry with an Orbitrap Exploris 240 (Thermo) coupled to a Vanquish Flex liquid chromatography system (Thermo). A volume of 2 µl samples was injected into a Waters XBridge XP BEH Amide column (150 mm length × 2.1 mm i.d., 2.5 µm particle size) maintained at 25 °C, with a Waters XBridge XP VanGuard BEH Amide (5 mm × 2.1 mm i.d., 2.5 µm particle size) guard column. For positive-mode acquisition, mobile phase A was 100% LC–MS-grade H 2 O with 10 mM ammonium formate and 0.125% formic acid. Mobile phase B was 90% acetonitrile with 10 mM ammonium formate and 0.125% formic acid. For negative-mode acquisition, mobile phase A was 100% LC–MS-grade H 2 O with 10 mM ammonium acetate, 0.1% ammonium hydroxide and 0.1% medronic acid (Agilent). Mobile phase B was 90% acetonitrile with 10 mM ammonium acetate, 0.1% ammonium hydroxide and 0.1% medronic acid. The gradient was 0 min, 100% B; 2 min, 100% B; 3 min, 90% B; 5 min, 90% B; 6 min, 85% B; 7 min, 85% B; 8 min, 75% B; 9 min, 75% B; 10 min, 55% B; 12 min, 55% B; 13 min, 35%, 20 min, 35% B; 20.1 min, 35% B; 20.6 min, 100% B; 22.2 min, 100% B, all at a flow rate of 150 μl min −1 , followed by 22.7 min, 100% B; 27.9 min, 100% B at a flow rate of 300 μl min −1 , and finally 28 min, 100% B at flow rate of 150 μl min −1 , for a total length of 28 min. The H-ESI source was operated in positive mode at spray voltage 3,500 or negative mode at spray voltage 2,500 with the following parameters: sheath gas 35 a.u., aux gas 7 a.u., sweep gas 0 a.u., ion transfer tube temperature 320 °C, vaporizer temperature 275 °C, mass range 70 to 1,000 m/z, full scan MS1 mass resolution of 120,000 FWHM, RF lens at 70% and standard AGC. Data-dependent MS2 fragmentation for compound identification and annotation was performed via the AquireX workflow (Thermo Scientific), which comprised three deep scans with MS1 resolution at 60,000 and MS2 resolution at 15,000. LC–MS data were analysed using Compound Discover (v.3.3, Thermo Scientific) and El-Maven software 69 for peak area determination and compound annotation. Compounds were annotated by matching to LC–MS method-specific retention time values of external standards and MS 2 spectral matching to external standards and the mzCloud database (Thermo Scientific).

Human biospecimens, clinical data and metabolite analysis of human samples

Human biospecimens and deidentified clinical data were provided by the Wilmot Cancer Institute Biobank Shared Resource (BSR) at the University of Rochester. All samples were collected under Institutional Review Board-approved protocols (STUDY61977 and STUDY7108), and all participants provided written informed consent. A total of 16 participants were enrolled, all of whom were women with a median age of 64 years (range of 25–88 years). Neoadjuvant treatment was administered in 4 participants (25%). The cohort included 81% white, 6% Black and 13% with race not reported. Most tumours were invasive ductal carcinoma (94%) with 6% classified as other histology. Oestrogen receptor positivity was observed in 81% of cases and progesterone receptor positivity in 62%. HER2 status was distributed as 0 in 56%, 1+ in 32%, 2+ in 6% and 3+ in 6%. Tumour staging showed 44% T1 and 56% T2 with no T3 or T4 disease. Nodal staging included 38% N0, 38% N1, 6% N2 and 18% NX. All participants were M0 at diagnosis with no metastatic disease. Tumour grade was 1 in 12%, 2 in 50% and 3 in 38%.

Peripheral whole blood was collected in the perioperative setting via a venous catheter immediately before surgery using K 2 -EDTA-coated tubes (BD Biosciences, 367899). Samples were centrifuged at 2,500g for 10 min at 4 °C to separate plasma. The plasma supernatant was aliquoted, snap-frozen in liquid nitrogen and stored at –80 °C before analysis.

Immediately following excision, tumour specimens were delivered to Surgical Pathology and reviewed by certified pathologists at the University of Rochester Medical Center, who allocated remnant malignant tissue for research. Research biospecimens were promptly transferred to the BSR laboratory, rinsed with ice-cold PBS and gently blotted dry. Tumour tissues (100–500 mg) were dissected into smaller fragments to cover the filter surface of 0.22 µm nylon-filtered microcentrifuge tubes (Corning, 8160), taking care to minimize mechanical cell lysis. Tubes were centrifuged at 300g for 5 min at 4 °C to collect TIF. The isolated TIF was snap-frozen in liquid nitrogen and stored at –80 °C before analysis.

Quantitative metabolite profiling of the TIF and serum human patient samples was performed based on a protocol adapted from a previous study 28 . In brief, chemical standard libraries of 149 metabolites in seven pooled libraries were prepared and serially diluted in high-performance liquid chromatography (HPLC)-grade water from in a dilution series from 5 mM to 1 µM to generate external standard pools, which are used for calibration of isotopically labelled internal standards and to quantify concentrations of metabolites for which internal standards were not available. We then measured 0.3 µl of each sample using a 1 µl Hamilton syringe (Hamilton, 80135) and extracted these volumes in 20 µl of a 75:25:0.1 HPLC grade acetonitrile–methanol–formic acid extraction mix with the following labelled stable isotope internal standards: 2 H 9 -choline (DLM-549), 13 C 4 -3-hydroxybutyrate (CLM-3853), 13 C 6 , 15 N 2 -cystine (CNLM-4244), 13 C 3 -lactate (485926, Sigma-Aldrich), 13 C 6 -glucose (CLM-1396), 13 C 3 -serine (CLM-1574), 13 C 2 -glycine (CLM-1017), 13 C 5 -hypoxanthine (CLM-8042), 13 C 2 , 15 N-taurine (CNLM-10253), 13 C 3 -glycerol (CLM-1510), 2 H 3 -creatinine (DLM-3653), 13 C 10 -kynurenic acid (80445, Sigma-Aldrich), 13 C 10 , 15 N 5 -adenosine (CNLM-3806), 13 C 8 -indole-3-carboxaldehyde (CLM-10745), 15 N 4 -inosine (NLM-4264), 13 C 10 -kynurenine (CLM-9884) and 13 C 4 -methylmalonic acid (CLM-9426). All standards were obtained from Cambridge Isotope Laboratory unless stated otherwise. Samples in extraction mix were vortexed for 10 min at 4 °C and centrifuged at 15,000 rpm for 10 min at 4 °C to pellet insoluble material. Next, 18 µl of the soluble polar metabolite supernatant was transferred to sample vials for analysis by LC–MS as previously described 4 , 28 . Once LC–MS analysis was performed, Skyline software was used for metabolite identification. External standard libraries were used to confirm the m/z and retention time for each metabolite. For quantitative analysis, when internal standards were available, external standard libraries were used to quantify concentrations of isotopically labelled internal standards in the extraction mix. Once internal standard concentrations were obtained, the peak areas of the unlabelled metabolites in the samples of culture medium were compared with the peak area of the quantified internal standard to determine the metabolite concentration in the sample.

For metabolites for which an internal standard was not present in the extraction mix, external standard libraries were used to perform analysis of relevant metabolite concentrations. In brief, the peak area of the metabolite was normalized to the peak area of an isotopically labelled internal standard with similar elution time, both in samples from culture medium and external standard library dilutions. Using the external standard library dilutions, we created a standard curve based on the linear relationship of the normalized peak area and the concentration of the metabolite, excluding those metabolites with an r 2 < 0.95. This standard curve was then used to interpolate the concentration of the metabolite in the samples.

Proteomic analysis

Cells were seeded in 10-cm dishes and cultured for 7 days in cystine-free medium supplemented with heavy isotope-labelled 13 C 6 , 15 N 2 -cystine (Cambridge Isotope Laboratories, CNLM-4244-H). At the end of this 7-day labelling period (t = 0 h), samples were collected to verify complete incorporation of the heavy label into cellular proteins. The remaining cells were washed three times with PBS and transferred to 6-well plates containing cystine-free medium supplemented with unlabelled GSH. After an additional 4 days (t = 96 h), cells were collected by trypsinization, pelleted, snap-frozen and stored at –80 °C. Cell pellets were then lysed in 50 µl of 5% SDS, 100 mM TEAB and sonicated (QSonica) for 5 cycles, with a 1 min resting period on ice after each cycle. Samples were then centrifuged at 15,000g for 5 min to pellet cellular debris, and the supernatant was collected. Protein concentration was determined using a BCA kit (Thermo Scientific), after which samples were diluted to 1 mg ml −1 in 5% SDS and 50 mM TEAB. Next, 25 µg protein from each sample was reduced with 2 mM dithiothreitol, followed by incubation at 55 °C for 60 min. Iodoacetamide was added to 10 mM and incubated in the dark at room temperature for 30 min to alkylate the proteins. Phosphoric acid was added to 1.2%, followed by six volumes of 90% methanol, 100 mM TEAB. The resulting solution was added to S-Trap micros (Protifi) and centrifuged at 4,000g for 1 min. The S-Traps containing trapped protein were washed twice by centrifuging through 90% methanol and 100 mM TEAB. Then 1 µg of trypsin was brought up in 20 µl of 100 mM TEAB and added to the S-Trap, followed by an additional 20 µl of TEAB to ensure the sample did not dry out. Samples were incubated at 37 °C overnight. The next morning, the S-Trap was centrifuged at 4,000g for 1 min to collect the digested peptides. Sequential additions of 0.1% trifluoroacetic acid (TFA) in acetonitrile and 0.1% TFA in 50% acetonitrile were added to the S-trap, centrifuged and pooled. Samples were frozen and dried in a Speed Vac (Labconco), then re-suspended in 0.1% TFA before MS analysis.

Peptides were injected onto a 75 µm × 2 cm trap column (Thermo Fisher) before re-focusing on an Aurora Elite 75 µm × 15 cm C18 column (IonOpticks) using a Vanquish Neo UHPLC (Thermo Fisher) connected to an Orbitrap Astral mass spectrometer (Thermo Fisher). Solvent A was 0.1% formic acid in water, whereas solvent B was 0.1% formic acid in 80% acetonitrile. Ions were introduced to the mass spectrometer using an Easy-Spray source operating at 2 kV. The gradient began at 1% B and ramped to 5% B in 0.1 min, increased to 30% B in 12.1 min, increased to 40% in 0.7 min and finally increased to 99% B in 0.1 min and was held for 2 min to wash the column for a total run time of 15 min. After each run was completed, the column was re-equilibrated with 1% B before the next injection. The Orbitrap Astral was operated in data-independent acquisition (DIA) mode, with MS1 scans acquired in the Orbitrap at a resolution of 240,000, with a maximum injection time of 5 ms over a range of 380–980 m/z. DIA MS2 scans were acquired in the Astral mass analyser with a 3 ms maximum injection time using a variable windowing scheme, using 2 Da windows from 380 to 680 m/z, 4 Da windows from 680 to 800 m/z and 8 Da windows from 800 to 980 m/z. The HCD collision energy was set to 25%, and the normalized AGC was set to 500%. Fragment ions were collected over a scan range of 150–2,000 m/z. The cycle time was 0.6 s.

For data analysis, the raw data were processed with DIA-NN (v.1.9.2; ) 70 . For all experiments, data analysis was carried out using library-free analysis mode in DIA-NN. To annotate the library, the Homo sapiens UniProt ‘one protein sequence per gene’ database (UP0000005640_9606, downloaded 7 April 2021) was used, with ‘deep learning-based spectra and RT prediction’ enabled. For precursor ion generation, the maximum number of missed cleavages was set to 1, the maximum number of variable modifications to 1 for Ox(M), the peptide length range to 7–30, the precursor charge range to 2–4, the precursor m/z range to 380–980 and the fragment m/z range to 150–2,000. The quantification was set to ‘Robust LC (high precision)’ mode with normalization disabled, MBR enabled, protein inferences set to ‘Genes’ and ‘Heuristic protein inference’ turned off. MS1 and MS2 mass tolerances, along with the scan window size, were automatically set by the software. To quantify light and heavy cysteine-containing peptides, the following parameters were included in the ‘Additional Options’ pane in the DIA-NN software: fixed-mod Carb, 57.021464, C, label; lib-fixed-mod Carb; channels Carb,L,C,0; Carb,H,C,4.007099; original-mods. Precursors were subsequently filtered at library precursor q value (1%), library protein group q value (1%) and posterior error probability (50%). Protein quantification was carried out using the MaxLFQ algorithm as implemented in the DIA-NN R package ( ), and the number of peptides quantified in each protein group was counted as implemented in the DiannReportGenerator package ( ) 71 .

Statistical analysis

All statistical analysis was completed using either R or GraphPad Prism (v.10.6.1).