A lysis-derived PMSF and orthovanadate oxidant defines a distinct cysteine-engagement profile

Abstract

Reactive cysteines serve important functions in proteins, and characterizing their engagement by different electrophiles facilitates biological discovery and covalent drug development. Here, we show that the common lysis buffer components PMSF and orthovanadate generate a lysis-derived oxidant that engages cysteines during cell lysis. This oxidant sulfonylates N-acetyl-D-glucosamine kinase (NAGK) C217, producing a mobility shift on SDS-PAGE. C217 lies within the ATP-binding pocket, and a C217S mutant exhibits reduced ATP affinity and enzymatic activity. Competitive IAA-ABPP chemoproteomics further showed that the PMSF/orthovanadate oxidant defines a cysteine-engagement profile that partially differs from that of pervanadate. These findings reveal an unrecognized source of chemical reactivity during protein extraction that expands the toolkit for cysteine-engagement profiling and underscores how sample preparation chemistry shapes chemoproteomic measurements.

Graphical Abstract

eTOC Blurb

Cysteine reactivity profiling helps identify functionally important and potentially druggable cysteines. Schaffer et al. show that PMSF and orthovanadate generate a lysis-derived oxidant that reveals a functionally important cysteine in NAGK and provides a complementary strategy for profiling cysteine engagement.

Introduction

The terminal sulfhydryl group of cysteine renders it functionally distinct within the proteome^1,2. Deprotonation of this sulfhydryl group generates a thiolate anion that is far more reactive toward electrophiles and more susceptible to oxidation than the protonated thiol^1,3. Because of this reactivity, cysteines are highly suited to act as active site nucleophiles that catalyze substrate bond cleavage or form covalent intermediates, are sites of redox regulation, serve as ligands for metal binding, and promote structural integrity or enzymatic activity through disulfide bond formation^3,4. Beyond these roles, cysteines can be pharmacologically targeted by covalent inhibitors^5–7, resulting in multiple clinically-approved compounds^8–10. There has thus been sustained effort to map cysteine reactivity, oxidation state, and ligandability across the proteome^4,11–18.

Proteome-wide cysteine reactivity profiling was developed by labeling highly reactive cysteines with low concentrations of iodoacetamide-alkyne (IAA) followed by copper-catalyzed azide-alkyne cycloaddition (CuAAC) to biotin enrichment tags for mass spectrometry (MS)-based detection¹⁶. Numerous studies built on this strategy to assess cysteine reactivity in response to various electrophilic scout fragments across multiple cell lines and conditions^{6,11,14,19,20}. Related strategies coupling oxidation with IAA-based labeling have been used to define context-dependent cysteine oxidation^12,21,22. Further, different electrophilic compounds and oxidants have been found to engage partially distinct cysteine populations, generating insights into the chemical and structural determinants of cysteine engagement^6,19,23–25. These insights can guide covalent inhibitor development by identifying targetable cysteines and assessing warhead selectivity²⁶. However, the chemical diversity of oxidants and electrophiles capable of engaging cysteines remains incompletely explored, limiting both mechanistic understanding and the engagement approaches available for cysteine reactivity profiling.

Here, we discovered that an oxidant generated by the widely used lysis buffer components phenylmethylsulfonyl fluoride (PMSF) and sodium orthovanadate (Na₃VO₄) can engage cysteines. As part of a study on the relevance of the N-acetyl-D-glucosamine (GlcNAc) salvage kinase NAGK (Figure 1A) in pancreatic cancer, we observed that NAGK underwent a 2 kDa mobility shift on SDS-PAGE in response to glutamine deprivation that was dependent on the presence of Na₃VO₄ in the lysis buffer²⁷. We initially aimed to characterize this presumed modification but found that the mobility shift was driven by PMSF and Na₃VO₄-mediated sulfonylation at NAGK C217. The susceptibility of NAGK C217 to sulfonylation suggested it was a reactive, potentially functional, cysteine. C217 is in the ATP binding pocket of NAGK, and we found that its mutation to serine reduced ATP affinity and decreased enzymatic activity in cells. Since artifactual oxidation by PMSF and Na₃VO₄ engaged a functional cysteine, we sought to identify other similarly engaged cysteines. We combined PMSF/Na₃VO₄-mediated oxidation during cell lysis with competitive iodoacetamide-alkyne activity-based protein profiling (IAA-ABPP)¹⁹ to identify reduced cysteines labeled by IAA following oxidant exposure. We compared the resulting cysteine-engagement profile with that of pervanadate, a mixture of highly oxidized vanadium compounds²⁸, and found that the two oxidant labeling systems yielded partially distinct IAA labeling patterns. We thus describe a lysis buffer artifact that can be harnessed as a complementary cysteine-engagement strategy.

Figure 1. The ATP binding pocket of NAGK is altered during cell lysis in response to glutamine deprivation.

(A) Schematic of the hexosamine biosynthetic pathway (HBP), focused on the salvage enzyme N-acetyl-D-glucosamine kinase (NAGK). The HBP synthesizes UDP-GlcNAc and derivative sugars from fructose-6-phosphate, glutamine (Gln), acetyl-CoA, and uridine triphosphate (UTP) for use in glycosylation. Following glycan degradation, salvage enzymes recycle hexosamine sugars into the HBP. Galactokinase 2, GALK2; cytidine monophosphate N-acetylneuraminic acid synthetase, CMAS; N-Acetylglucosamine, GlcNAc; N-acetylgalactosamine, GalNAc; N-Acetylneuraminic acid, Neu5Ac.

(B) PANC-1 cells were grown in the indicated concentrations of glucose and glutamine (Q) for 24 or 48 hours prior to lysis and analysis on SDS-PAGE/Western blot. AKT pS473 serves as a nutrient deprivation control.

(C) PANC-1 cells expressing inducible pInducer20 (pInd20) GFP or NAGK-HA wild type (WT) or mutants were grown in 500 ng/ml doxycycline and 3.6 or 0.05 mM Q for 24 hours prior to lysis. Exposure, exp.

(D) Quantification of the upper and lower HA band intensities from the 3.6 and 0.05 mM Q WT and S271A conditions in Figure 1C. No significant differences were detected by one-way ANOVA with Šídák’s correction. Data are presented as mean ± s.d.

(E) Structure of NAGK generated from PDB 2ch6. The left image is of the NAGK homodimer, with one monomer in yellow/orange and the other in gray/blue. The small N-terminal domain is in yellow/gray, and the large C-terminal domain is in orange/blue. The boxed region indicates the ATP binding pocket, enlarged at right. The right image highlights the location of S271 near the ATP binding site.

(F) PANC-1 cells were grown in 3.6 or 0.05 mM Q for 24 hours prior to lysis without the specified protease or phosphatase inhibitor.

(G) Recombinant His-NAGK was incubated in 0.5% IGEPAL lysis buffer with or without 1 mM PMSF and 20 mM Na₃VO₄.

(H) PANC-1 cells were grown in 3.6 or 0.05 mM Q for 24 hours and lysed in 0.5% IGEPAL lysis buffer containing the indicated concentrations of ATP and/or GlcNAc.

All assays are representative of n=3 independent replicates. Protein lysates were extracted in 0.5% IGEPAL buffer with 1 mM PMSF and 20 mM Na₃VO₄ unless otherwise indicated. See also Figure S1, and Data S1 and S4.

Design

In this study, we found that Na₃VO₄ and PMSF generate an oxidant that engages a subset of cysteines during cell lysis. The combination of Na₃VO₄ and PMSF has been shown to oxidize NADH²⁹ but not cysteines. We inadvertently discovered that these reagents could do so when we observed a 2 kDa mobility shift on SDS-PAGE of NAGK that was driven by PMSF and Na₃VO₄-mediated sulfonylation of C217. In this study, we describe our efforts to understand what was driving the mobility shift and how we distinguished it from canonical drivers of altered mobility, such as posttranslational modification. This has practical implications, as PMSF and Na₃VO₄ are frequent lysis buffer components and unintentional cysteine oxidation by this combination could go unrecognized. Since labeling by the PMSF/Na₃VO₄ oxidant identified a functionally relevant cysteine in NAGK, we extended this approach to identify other cysteines engaged by this oxidant. We utilized competitive IAA-ABPP¹⁹, as it is a well-established approach in which cysteines are exposed to an electrophile of interest followed by IAA labeling of unreacted cysteines. Depletion of a cysteine from the resulting analysis supports engagement by the electrophile of interest. Applying IAA-ABPP, we compared the PMSF/Na₃VO₄ oxidant with pervanadate and found partially distinct cysteine engagement profiles by the two oxidant systems. Through careful characterization, we defined the PMSF/Na₃VO₄ oxidant as an artifact of cell lysis with the goals of describing this as a potential outcome of using these reagents and of defining a complementary cysteine engagement approach.

Results

The ATP-binding pocket of NAGK is altered by the lysis buffer in response to nutrient deprivation

We previously observed that NAGK undergoes a Na₃VO₄-dependent 2 kDa mobility shift on SDS-PAGE in response to glutamine deprivation²⁷. The modification also occurred in response to glucose deprivation in some cells and in almost every human cancer cell line tested following 48 hours of glutamine deprivation (Figures 1B, S1A, and S1B; data not shown). The modification persisted for at least 24 hours following refeeding after glutamine deprivation (Figure S1C). Given the Na₃VO₄ dependence²⁷, we initially hypothesized that NAGK was phosphorylated in response to nutrient stress, but we did not detect differential NAGK-HA phosphorylation by LC-MS/MS following glutamine deprivation (Data S1, tabs 1 and 2). In addition, the mobility shift persisted following lambda phosphatase treatment of immunoprecipitated NAGK-HA (Figure S1D) and was retained in alanine mutants of all annotated NAGK phosphorylation sites³⁰ (Figures 1C and 1D). Notably, mutation of NAGK S271, located near the ATP binding pocket³¹ (Figure 1E), to alanine resulted in stronger basal modification of NAGK in cells grown in nutrient replete media (Figures 1C and 1D), suggesting S271 is proximal to and influences the modification.

The modification appeared inconsistent with a typical posttranslational modification, as it depended strongly on lysis buffer composition and volume. For example, the modification did not occur under strongly denaturing lysis conditions (e.g. 2% SDS) and was more pronounced with larger lysis buffer volumes (Figures S1E and S1F). The modification also required PMSF (Figure 1F), occurring in a PMSF and Na₃VO₄ dose-dependent manner (Figures S1G and S1H), with higher concentrations of either reagent inducing the modification independent of nutrient status. PMSF potency appeared batch and cell line dependent, but 0.5–1 mM was sufficient to observe the modification in response to glutamine deprivation. AEBSF, a structurally similar serine protease inhibitor with enhanced aqueous stability, did not mimic PMSF’s capacity to induce the modification in the presence of Na₃VO₄ (Figure S1I), suggesting PMSF hydrolysis may contribute to the modification.

To test whether the lysis buffer alone was sufficient to induce the modification, we incubated recombinant His-NAGK with and without PMSF and/or Na₃VO₄. The modification occurred on recombinant NAGK in the presence of both reagents, (Figure 1G), confirming the modification is lysis buffer-derived. The S271A mutant suggested that the modification occurred within or near the active site (Figures 1C–1E), so we tested whether ATP and/or GlcNAc in the lysis buffer could inhibit the modification by blocking active site access. Increasing concentrations of ATP blunted the modification, an effect enhanced by GlcNAc, while GlcNAc alone had minimal effect (Figure 1H). Similar results were observed on recombinant NAGK (Figure S1J). These data indicated that NAGK was modified in or near the ATP binding pocket by PMSF and Na₃VO₄ during lysis, but the nature of the modification and its relationship to glutamine deprivation remained unclear.

NAGK is sulfonylated at C217 in the ATP binding pocket by the lysis buffer

The phosphatase inhibitory activity of Na₃VO₄ arises from its ability to bind the phosphotransfer catalytic residue in the enzyme’s active site^32,33. Molybdate and tungstate have a similar mechanism of action^32,33 but did not induce the modification (Figure S2A), making a vanadate adduct less likely. Furthermore, such a mechanism would not readily explain the reliance on PMSF.

Vanadium compounds can act as oxidizing agents, and PMSF has been reported to catalyze the oxidative capacity of Na₃VO₄ toward NADH²⁹. In line with oxidation underlying the modification, adding reducing agents to the lysis buffer prevented it dose-dependently (Figures 2A, S2B, and S2C). H₂O₂ weakly induced the modification on recombinant His-NAGK (Figure S2D), further supporting an oxidative mechanism. Cysteines are frequent targets of oxidation³, so we pretreated cells with the cell-permeable alkylating agent N-ethylmaleimide (NEM) to block cysteines prior to lysis. This prevented the modification and generated a smaller, NEM-dependent shift, potentially representing a NEM adduct (Figure 2B). Cysteines in some proteins, such as PTP1B and Ran, can be sulfonylated (converted to cysteic acid) by pervanadate^28,34–36, producing a ~2 kDa mobility shift on Ran³⁵. We found that addition of pervanadate to the lysis buffer in place of orthovanadate induced a similar mobility shift on NAGK that was independent of nutrient stress (Figure 2C), likely due to the strong oxidizing capacity of pervanadate. Although PMSF was included when initially assessing pervanadate’s oxidative capacity (Figures 2C and S2E), it is not required for pervanadate to induce the modification (Figure S2F).

Figure 2. PMSF and Na₃VO₄ sulfonylate NAGK C217 during cell lysis.

(A) PANC-1 cells were grown in 3.6 or 0.05 mM Q for 24 hours prior to lysis in buffer with the indicated concentrations of Tris (2-carboxyethyl) phosphine (TCEP).

(B) PANC-1 cells were grown in 3.6 or 0.05 mM Q for 24 hours and incubated ± 20 mM N-ethylmaleimide (NEM) for 10 minutes prior to lysis with 1 mM PMSF and ± 20 mM Na₃VO₄.

(C) PANC-1 cells were grown in 3.6 or 0.05 mM Q for 24 hours prior to lysis in buffer containing 1 mM PMSF and the indicated concentrations of Na₃VO₄ or pervanadate, or H₂O₂ + catalase as a control for pervanadate synthesis.

(D) Location of C211 and C217 in the ATP binding pocket of NAGK. Images generated from PDB 2ch6.

(E) PANC-1 cells expressing doxycycline-inducible pInducer20 (pInd20) GFP or the specified NAGK-HA constructs were grown in 3.6 or 0.05 mM Q and 500 ng/ml doxycycline for 24 hours prior to lysis.

(F) The upper, slower mobility band of NAGK is enriched for NAGK C217 sulfonylation (i.e. cysteic acid) following incubation with PMSF and Na₃VO₄. Recombinant His-NAGK was incubated ± 1 mM PMSF and 20 mM Na₃VO₄ in 0.5% IGEPAL lysis buffer and run on SDS-PAGE. The upper and lower bands were excised for mass spectrometry analysis. Normalized sulfonylated peptide intensity is shown, with a line connecting the intensities of the upper and lower bands within each replicate. No significant differences were detected by two-way ANOVA with Šídák’s correction (n=3 independent replicates).

(G) Representative MS/MS spectra of a tryptic peptide from human NAGK containing C217 in its carbamidomethylated (unmodified; left) and sulfonylated (right) forms. Carbamidomethylation reflects alkylation of unmodified cysteine with iodoacetamide. Annotated b- and y-ion series confirm the peptide sequence and site-specific modification.

(H) PANC-1 cells expressing pInd20 GFP or the specified NAGK-HA constructs were grown as in Figure 2E for 24 hours prior to lysis in buffer with 0.6 mM PMSF and 20 mM Na₃VO₄. The SDS lysis buffer does not facilitate sulfonylation (Figure S1E) and is shown as a control for baseline mobility of the various mutants.

All assays are representative of n=3 independent replicates. Protein lysates were extracted in 0.5% IGEPAL buffer with 1 mM PMSF and 20 mM Na₃VO₄ unless otherwise indicated. See also Figures S2, S3, and S4, and Data S1 and S4.

There are two cysteines within the ATP binding pocket of NAGK, C211 and C217³¹ (Figure 2D). An alanine mutant of C217, but not C211, blunted the PMSF/Na₃VO₄ (Figure 2E), pervanadate (Figure S2E), and NEM (Figure S2G) induced mobility shifts. We then incubated recombinant His-NAGK with and without PMSF and Na₃VO₄, resolved the product by SDS-PAGE, and excised the upper and lower bands for LC-MS/MS analysis. In support of our biochemical observations, sulfonylation of C217 was enriched on the upper band of protein (Figures 2F and 2G). This is consistent with reducing agents preventing the modification (Figure 2A), as they buffer the oxidative capacity of the lysis buffer. It also explains the modification’s persistence in SDS-PAGE, as once cysteine is fully oxidized to cysteic acid, it cannot be reversed by reducing agents³.

Interestingly, we did not observe sulfonylation of NAGK in mT8 cells, a murine pancreatic cancer cell line³⁷, but when murine NAGK was knocked out and replaced with human NAGK, the human protein was sulfonylated regardless of nutrient status (Figure S3A). Murine NAGK is one amino acid shorter than human NAGK and differs at 28 amino acids, most of which are distal to C217 (Figure S3B and S3C). However, human NAGK C217 is followed by arginine whereas murine NAGK C217 is followed by glutamine (Figure S3B and S3C). The positive charge of arginine stabilizes the thiolate state of cysteine¹², which is more susceptible to oxidation than the protonated thiol¹. The neutral charge of Q218 in murine NAGK wouldn’t similarly support the thiolate state. Consistently, human R218Q and R218A mutants were less susceptible to PMSF/Na₃VO₄ oxidation when expressed in PANC-1 cells, while mutation to negatively charged aspartic acid rendered human NAGK unresponsive to sulfonylation (Figure 2H). Together, these data suggest that cell type, nutrient status, and proximal amino acids dictate whether NAGK C217 is sulfonylated by the PMSF/Na₃VO₄ oxidant.

Glutamine deprivation increases NAGK C217 engagement by iodoacetamide

To determine whether NAGK C217 was engaged by canonical cysteine-alkylating agents, we lysed cells in the presence of IAA, followed by CuAAC-mediated biotinylation and streptavidin enrichment of the labeled proteins. Glutamine deprivation increased IAA labeling of NAGK in a C217- and C211-dependent manner (Figure S4A). Similar results were observed with the cysteine-reactive electrophile maleimide-biotin (MB), although labeling was entirely dependent on C211 (Figure S4B), likely reflecting steric exclusion of the bulkier MB probe from C217, which, notably, is the target of PMSF/Na₃VO₄ sulfonylation (Figure 2F). This increased labeling occurred throughout the lysate in PANC-1 cells following glutamine deprivation (Figures S4A and S4B), but not in glutamine-deprived DLD-1 cells and not consistently in glucose-deprived cells (Figures S4C and S4D), despite NAGK C217 sulfonylation occurring during cell lysis in response to these conditions (Figures 1B and S1B). Thus, NAGK C217 engagement by PMSF/Na₃VO₄ sulfonylation during lysis can correlate with IAA or MB labeling, but these processes are not uniformly coupled across conditions and cell lines. However, it appears likely that the same underlying process drives both increased oxidation and IAA labeling of NAGK C217 in response to glutamine deprivation.

Glutathione buffers PMSF/Na₃VO₄-mediated NAGK C217 sulfonylation during lysis

We next investigated why glutamine deprivation increased NAGK C217 sulfonylation during lysis. The protonation status of a cysteine depends on intracellular pH⁴ and protein intrinsic factors including nearby amino acids¹², posttranslational modifications^12,38, and protein conformation⁶. NAGK showed no evidence of glutamine deprivation-induced posttranslational modification (e.g. phosphorylation, Figures 1C and S1D), making that a less likely cause. The global increase in IAA alkylation following glutamine deprivation suggested an environmental factor was driving heightened cysteine labeling. We therefore measured intracellular pH in response to nutrient deprivation, as increased pH could promote cysteine deprotonation. However, glutamine deprivation decreased intracellular pH in PANC-1 cells (Figure S5A).

We next considered factors that could decrease cellular reductive buffering capacity. Glutamine deprivation depletes glutathione³⁹, a major contributor to reductive buffering capacity⁴⁰, potentially allowing the PMSF/Na₃VO₄ oxidant to persist in the lysis buffer. We confirmed that glutathione levels decrease in response to nutrient deprivation (Figures 3A and S5B). To deplete glutathione without nutrient deprivation, we treated cells with L-Buthionine-(S,R)-Sulfoximine (BSO), an inhibitor of γ-glutamylcysteine synthetase⁴¹, the first and rate-limiting enzyme in glutathione synthesis (Figures 3A and S5B). BSO increased sulfonylation of NAGK C217 (Figures 3B and S5C) and global cysteine alkylation (Figures 3C and S5D) in PANC-1 and DLD-1 cells, indicating that intracellular glutathione buffers PMSF/Na₃VO₄-mediated labeling of NAGK C217 by sulfonylation during lysis. BSO is reported to have minimal impact on cysteine ligandability in other cell lines⁶, likely due to different labeling protocols. Here, alkylating agents are present in the lysis buffer, whereas the aforementioned study added them post-sonication. Post-sonication labeling focuses on protein-intrinsic cysteine reactivity, while labeling during lysis or in situ captures the native cellular state but may also reflect environmental effects on the probe. To test this, we labeled cells with Iodoacetyl-PEG2-Biotin (IAB) during lysis or following sonication and found that the increase in labeling from glutamine deprivation or BSO treatment was diminished with labeling post-sonication (Figure S5E). IAB also gave a notably smaller increase in labeling following glutamine deprivation than IAA or MB (Figures S4A, S4B, and S5E), highlighting differences in probe behavior.

Figure 3. Glutathione buffers PMSF/Na₃VO₄-mediated NAGK C217 sulfonylation during lysis.

(A) PANC-1 cells were grown in the indicated media for 24 hours prior to measurement of reduced glutathione (GSH) and oxidized glutathione (GSSG). Significance was determined using a two-way ANOVA with Dunnett’s correction. BSO, L-Buthionine-(S,R)-Sulfoximine; GSEE, glutathione ethyl ester.

(B and C) PANC-1 cells were grown in the indicated media for 24 hours prior to lysis in 0.5% IGEPAL lysis buffer with PMSF (0.6 – 1 mM, depending on replicate) and 20 mM Na₃VO₄ (B) or ± 50 μM maleimide-biotin (MB; C).

(D and E) PANC-1 cells were grown in 3.6 or 0.05 mM Q media ± 10 mM GSEE for 24 hours prior to lysis in 0.5% IGEPAL lysis buffer with PMSF (0.5 – 0.75 mM, depending on replicate) and 20 mM Na₃VO₄ (D) or ± 50 μM maleimide-biotin (E).

(F and G) PANC-1 cells were grown in 3.6 or 0.05 mM Q media for 24 hours ± 1 mM Tris(2-methoxycarbonylethyl)phosphine hydrochloride (tmTCEP) for 1 hour or ± 1 μM KI696 for 24 hours prior to lysis in 0.5% IGEPAL lysis buffer with 0.6 mM PMSF and 20 mM Na₃VO₄ (F) or ± 50 μM maleimide-biotin (G).

(H) PANC-1 cells were grown in the indicated media for 24 hours prior to measurement of total glutathione (reduced GSH + GSSG). Significance was determined using a one-way ANOVA with Dunnett’s correction.

(I) Empty vector (EV) or human NAGK-HA were constitutively expressed (PGK) in clones of murine mT8 cells with CRISPR (px459) knockout of NAGK or a control locus. Cells were grown in the indicated conditions for 24 hours prior to lysis in 0.5% IGEPAL lysis buffer with 0.6 mM PMSF and 20 mM Na₃VO₄. Ctrl, control.

All assays are representative of n=3 independent replicates. Data are presented as mean ± s.d. *p < 0.05, ***p ≤ 0.001, ****p < 0.0001. See also Figure S5, and Data S1 and S4.

We then tested whether increasing glutathione attenuated PMSF/Na₃VO₄-mediated labeling by sulfonylation during lysis. Supplementation with 10 mM glutathione ethyl ester (GSEE), a cell-permeable form of glutathione⁴², partially restored glutathione levels during glutamine deprivation (Figure 3A) and reduced both labeling of NAGK C217 by PMSF/Na₃VO₄-mediated sulfonylation (Figure 3D) and global cysteine alkylation (Figure 3E) following glutamine deprivation. We then tested additional ways of augmenting cellular antioxidant capacity with KI696, an activator of the pro-antioxidant transcription factor NRF2^43,44, and Tris(2-methoxycarbonylethyl)phosphine hydrochloride (tmTCEP), a cell-permeable TCEP derivative⁴⁵. KI696 reduced basal PMSF/Na₃VO₄-mediated sulfonylation of NAGK C217 (Figure 3F) and cysteine alkylation (Figure 3G) during lysis of PANC-1 cells grown in full glutamine. KI696 did not induce NRF2 expression in cells grown in 0.05 mM glutamine, and sulfonylation and global alkylation were unchanged (Figures 3F and 3G). NRF2 orchestrates a broad antioxidant response⁴⁴, including glutathione synthesis^46,47 (Figure 3H), likely explaining the observed decrease in C217 labeling by sulfonylation following KI696 treatment. Conversely, tmTCEP had no effect on C217 labeling by sulfonylation or global cysteine alkylation (Figures 3F and 3G), consistent with the low cellular GSSG levels observed (Figure 3A). Given the differential susceptibility of human versus murine NAGK C217 to labeling by the PMSF/Na₃VO₄ oxidant during lysis of mT8 cells (Figure S3A), we tested whether BSO or KI696 could alter PMSF/Na₃VO₄-mediated labeling of either ortholog. Murine NAGK C217 remained unreactive when mT8 cells were treated with BSO (Figure 3I), even though BSO globally increased cysteine alkylation in these cells (Figure S5F), consistent with sequence differences reducing murine NAGK C217 reactivity to the PMSF/Na₃VO₄ oxidant (Figures 2H, S3B, and S3C). Treatment of mT8 cells with KI696 did, however, reduce PMSF/Na₃VO₄ labeling of human NAGK C217 (Figure 3I), suggesting basal glutathione from mT8 cells is insufficient to quench the PMSF/Na₃VO₄ oxidant in the lysis buffer.

Together, these data demonstrate that intracellular glutathione depletion increases cysteine labeling during cell lysis by both the PMSF/Na₃VO₄ oxidant and alkylating agents, most likely by leaving more oxidant or alkylating agent available to react with protein cysteines when glutathione-depleted cells are lysed. We did observe nutrient-deprivation independent changes in C217 labeling by PMSF/Na₃VO₄ when variants of NAGK were expressed bearing mutations with sufficient proximity to C217 to influence its reactivity (Figures 1C, 1D, and 2H). Thus, despite being an artifact of the lysis protocol, labeling by the PMSF/Na₃VO₄ oxidant appears to report on cysteine reactivity in a manner analogous to iodoacetamide and directed our attention to a reactive cysteine in a protein of interest.

NAGK C217S has decreased ATP affinity in vitro

Given NAGK’s emerging relevance in disease and normal physiology^27,48,49, we next sought to characterize how C217, as well as nearby C211, contribute to NAGK enzymatic activity. We assessed substrate affinity (Km) and maximal velocity (Vmax) for purified GST-NAGK wild type (WT) and cystine mutants using the coupled pyruvate kinase/lactate dehydrogenase ATPase continuous assay (PK-LDH assay)⁵⁰ (Figure S6A). Mutation of C217 to serine substantially decreased ATP affinity, mildly reduced GlcNAc affinity, and had no effect on affinity for the muramyl dipeptide, an alternative NAGK substrate⁴⁸ (Figures 4A–4C and S6B), consistent with C217’s location in the ATP binding pocket. However, a C217A mutant did not impact ATP affinity (Figure S6B), likely because the smaller methyl group of alanine does not interfere with catalysis. Mutation of C211 did not impact substrate affinity (Figures 4A–4C and S6B). These findings align with a previous report that C217S reduces NAGK activity in vitro⁵¹. C217 is highly conserved (Figure S6C), while C211 is not, consistent with its functional importance. None of the mutants impacted Vmax (Figures 4A–4C and S6B), although Vmax for the muramyl dipeptide was substantially lower than for GlcNAc (Figure 4C).

Figure 4. NAGK C217 promotes enzymatic activity.

(A and B) Michaelis Menten plots of the initial velocity versus [substrate] for NAGK WT and cysteine to serine mutants. NAGK was incubated with the indicated concentration of ATP (A) or GlcNAc (B) and saturating concentrations of the second substrate. The associated Km and Vmax values are in Figure 4C. n=4 independent assays.

(C) Km and Vmax values from the assays shown in 4A and 4B, extended to include muramyl dipeptide. n=4 independent assays for each substrate; each set of assays is represented by a different background color.

(D) ¹³C₂ GlcNAc tracing assay schematic. PANC-1 cells were grown in 3.6 or 0.05 mM Q for 21 hours followed by the addition of 10 mM ¹³C₂ GlcNAc for three hours prior to harvest; ¹²C GlcNAc was used in parallel controls. For ¹³C₂ GlcNAc-treated cells, unlabeled (¹²C) GlcNAc-6-P is primarily derived from de novo synthesis while ¹³C₂ GlcNAc-6-P is derived from NAGK activity.

(E-G) Area under the curve normalized to protein concentration for isotopes of GlcNAc, GlcNAc-6-P, and UDP-GlcNAc. PANC-1 cells with endogenous knockout of NAGK and inducible expression of empty vector (EV), NAGK-HA WT or C217S were grown in 3.6 or 0.05 mM Q and doxycycline (see Figure S6E for concentrations) for 21 hours, followed by the addition of 10 mM ¹³C₂ GlcNAc for 3 hours. Significance was determined using a two-way ANOVA with Tukey’s correction. The significance of select comparisons of ¹³C₂ GlcNAc-6-P is shown; for significance of all comparisons, see Data S2. Data are presented as mean ± s.d.; n=3 independent biological replicates. *p < 0.05, ***p < 0.001, ****p < 0.0001.

See also Figure S6, and Data S2 and S3.

NAGK C217S has decreased activity in cells

To assess C217’s contribution to NAGK activity in cells, we generated PANC-1 lines with knockout of endogenous NAGK and rescue with inducible empty vector (EV), WT or C217S NAGK-HA (Figures S6D–S6F). The C217S mutant did not express as robustly as WT NAGK-HA using constitutive promoters, so we used an inducible system to equalize NAGK expression across conditions (Figures S6E and S6F). We grew the cells in 3.6 or 0.05 mM glutamine for 21 hours, followed by addition of 10 mM N-[1,2-¹³C₂]acetyl-D-glucosamine (¹³C₂ GlcNAc) for three hours prior to harvest (Figures 4D–4G and S6G–S6I). Cells expressing WT NAGK-HA had higher levels of ¹³C₂ GlcNAc-6-P, the direct NAGK phosphorylation product, than cells expressing the C217S mutant, demonstrating decreased C217S activity in cells (Figure 4F). Meanwhile, levels of ¹³C₂ GlcNAc-6-P were nearly absent in cells with knockout of NAGK, indicating specificity for NAGK activity (Figure 4F), while ¹³C₂ UDP-GlcNAc, being further downstream (Figure 4D), was a less specific readout (Figure 4G), consistent with previous NAGK KO tracing data²⁷. ¹³C₂ GlcNAc-6-P decreased in glutamine deprived cells expressing WT, but not C217S, NAGK-HA, suggesting glutamine deprivation reduces WT activity. We did not, however, consistently observe a similar reduction in ¹³C₂ GlcNAc-6-P following glutamine deprivation across all replicates in a ¹³C₂ GlcNAc tracing experiment conducted with endogenous NAGK in PANC-1 cells (Figures S6J–S6O), indicating the reduction in WT activity may not hold across all systems. Given that ATP levels decrease during glutamine deprivation (Figure S6P) and C217S has reduced ATP affinity (Figures 4A–4C and S6B), we anticipated decreased C217S activity under these conditions. The absence of this effect suggests C217 may influence cellular activity beyond ATP affinity. Together, these data demonstrate that C217 enhances ATP affinity in vitro and promotes cellular NAGK activity relative to the C217S mutant.

The PMSF/Na₃VO₄ oxidant defines a distinct cysteine-engagement profile by competitive IAA-ABPP

Having identified a key active site cysteine in NAGK, we next asked whether PMSF/Na₃VO₄ could more broadly engage cysteines. We lysed cells in increasing concentrations of PMSF/Na₃VO₄ or in pervanadate and analyzed a panel of proteins with previously characterized reactive cysteines^{35,36,52–56}. In addition, we probed with an antibody that recognizes the sulfonylated active site cysteine of protein tyrosine phosphatases (oxPTP). Several proteins displayed one or more slower mobility bands with increasing PMSF/Na₃VO₄, although not all proteins tested were affected (e.g. SRC and Ran) (Figure 5A). Vinculin appeared as multiple faster and slower migrating bands, potentially indicative of more extensive oxidative damage^57,58. In addition, a ~50 kDa band showed increasing signal with the oxPTP antibody as oxidative capacity increased. Some of the proteins tested were modified by PMSF/Na₃VO₄ but not pervanadate, and vice versa, indicating that while both reagents can irreversibly oxidize cysteines during lysis, they do not engage the same targets equivalently. Pretreatment of cells with NEM prior to lysis blunted some of the PMSF/Na₃VO₄-induced bands (Figure S7A), while TCEP in the lysis buffer prevented all of them (Figure S7B), consistent with oxidation underlying the mobility shifts.

Figure 5. The PMSF/Na₃VO₄ oxidant defines a distinct cysteine-engagement profile by competitive IAA-ABPP.

(A) PANC-1 cells were grown in 3.6 or 0.05 mM Q for 24 hours and lysed in 0.5% IGEPAL lysis buffer containing the indicated concentrations of PMSF and Na₃VO₄, or 20 mM pervanadate (without PMSF). Representative of n=3 independent biological replicates.

(B) IAA-ABPP-based labeling strategy to identify cysteines oxidized by PMSF/Na₃VO₄ and pervanadate.

(C,D) PCA (C) and hierarchical clustering (D) of IAA-alkylated peptide abundance.

(E) Volcano plots from proDA-based differential abundance analysis of the indicated comparisons. proDA models missing values probabilistically without imputation, accounting for the missing-not-at-random dropout expected when cysteines are oxidized, blocking IAA labeling. Condition color (Figure 5D) indicates peptide depletion in that condition, e.g. purple = depleted with pervanadate treatment. Circles, peptides with at least one observed value in both conditions; triangles, peptides undetected across all replicates in one condition; squares, peptides undetected across all replicates in both conditions. The five most depleted and five most enriched peptides per comparison are labeled by gene name and cysteine position.

(F) PANC-1 cells were lysed in 0.5% IGEPAL lysis buffer containing the indicated concentrations of PMSF and Na₃VO₄, or 20 mM pervanadate (without PMSF). Representative of n=3 independent biological replicates.

See also Figures S7, S8, and S9; Tables S2, S3, and S4, and Data S4.

To define these effects at scale, we combined oxidant exposure during lysis with competitive IAA-ABPP-based proteomics^16,19 (Figure 5B). This assay reports cysteine engagement indirectly by measuring remaining IAA-labelable cysteines after oxidant exposure and does not constitute direct site-by-site mapping of oxidized cysteines. We lysed PANC-1 cells in no oxidant, 5 mM pervanadate, or two concentrations of PMSF plus Na₃VO₄ followed by labeling in 1 mM IAA to alkylate unreacted cysteines. A subset of IAA-free lysate was left unlabeled or labeled with IAB as western blot controls for protein oxidation and decreased alkylation, respectively (Figures S7C and S7D). IAA-labeled proteins were biotinylated, pulled down, digested, and eluted in acid (Figure S8A), then alkylated peptides were analyzed by mass spectrometry (Table S2). After filtering and collapsing missed tryptic cleavages, we identified 25,865 alkylated peptides. (Figure S8B and S8C, Table S3). Principal component analysis (PCA) revealed that overall oxidative capacity was the dominant source of variance across samples, while oxidant identity separated PMSF/Na₃VO₄ from pervanadate-treated samples along the second principal component (Figures 5C). Similarly, hierarchical clustering grouped samples based on oxidant type and concentration (Figure 5D). Consistent with broad cysteine engagement, both oxidant systems decreased the intensity of many IAA-labeled peptides relative to control lysates (Figure 5D, S8D, and S8E). Conversely, some cysteines gained IAA labeling under oxidizing conditions (Figure 5D), perhaps reflecting oxidation-induced conformational changes^59,60 that expose normally buried cysteines. Differential abundance analysis identified peptides altered across all conditions, as well as those preferentially depleted by either PMSF/Na₃VO₄ or pervanadate (Figures 5E and S8F, Table S4). Together, these data indicate that the PMSF/Na₃VO₄ oxidant accesses a broad but selective subset of cysteines and that its target spectrum is only partially overlapping with that of pervanadate.

To orthogonally assess candidate targets identified by IAA-ABPP, we selected proteins spanning a range of depletion and enrichment behaviors and examined them for oxidation-induced mobility shifts. Many of these proteins underwent some degree of mobility shift (Figure 5F), but IAA alkylation patterns were not necessarily predictive of oxidation detected by western blot (Figure S9). This divergence is expected for several reasons: IAA-ABPP can capture both reversibly and irreversibly oxidized cysteines, whereas only irreversibly oxidized cysteines persist under SDS-PAGE reducing conditions; some relevant peptides, such as the one containing NAGK C217, are not detected in the proteomics data; and the effect of an added negative charge on SDS-PAGE mobility is position dependent. Thus, the two approaches are complementary in the information they provide and can be tailored to address specific questions. For example, introducing a reduction step in IAA-ABPP between oxidation and alkylation would select for irreversibly oxidized cysteines.

Discussion

PMSF and Na₃VO₄ have been used in protein lysis buffers for decades^61–64 as independent components with distinct purposes. Here, we find that these reagents react to form an unknown oxidant that engages cysteines. We first observed this as a 2 kDa mobility shift of NAGK on SDS-PAGE that we mistook for a posttranslational modification. The assays we conducted to identify the underlying cysteic acid provide a framework to disambiguate future instances of unintentional PMSF/Na₃VO₄ oxidation from biologically derived mobility shifts. Of these, addition of a reducing agent to the lysis buffer is the most straightforward to implement. Further, these observations underscore how cellular state can chemically impact cysteine labeling and thus shape the interpretation of cysteine reactivity. Our finding that the cysteine underlying the NAGK mobility shift was functional, coupled with similar shifts on additional proteins with known functional cysteines (Figure 5A), indicates that other cysteines engaged by the PMSF/Na₃VO₄ oxidant could represent cysteines of interest. Thus, the discovery that PMSF and Na₃VO₄ oxidize cysteines can help identify when this occurs as an unintended artifact but can also be used as a complementary approach to engage and study cysteines. As an illustration of the value of complementary approaches to cysteine profiling, NAGK C217 was more reactive to the PMSF/Na₃VO₄ oxidant and pervanadate than to canonical cysteine alkylating agents, which largely preferred C211 (Figures S4A and S4B), and its associated peptide was not detected in our IAA-ABPP proteomics.

We used IAA-ABPP to identify other cysteines engaged by the PMSF/Na₃VO₄ oxidant because it is an established methodology to profile cysteine engagement by an electrophile of interest, such as oxidants or scout fragments, by labeling and enriching cysteines available for alkylation following electrophile exposure. Since IAA-ABPP does not provide a direct atlas of cysteine oxidation, biochemical assays similar to those conducted for NAGK are required to demonstrate that decreased IAA labeling following oxidant exposure represents oxidation and not label depletion for secondary reasons (e.g., conformation change following oxidation elsewhere on the protein). We considered direct analysis of sulfonylated peptides, but there are currently no chemical probes for the enrichment of cysteic acid-containing peptides⁶⁵. There are several reported methodologies that facilitate their enrichment^66,67 or identify them directly from a complex mixture⁶⁸, but these have not been robustly scaled to proteomics. This approach would also miss cysteines engaged by the PMSF/Na₃VO₄ oxidant that are not fully oxidized to cysteic acid.

Using IAA-ABPP, we compared cysteine engagement profiles following cell lysis in the presence of the PMSF/Na₃VO₄ oxidant versus pervanadate. Pervanadate has been used to characterize active-site cysteine oxidation in protein tyrosine phosphatases (PTPs)^34,69, study tyrosine kinase signaling through PTP inhibition in cells^70,71, and oxidize other targets including Ran and SRC^35,36, but its activity at the proteomics level had not been assessed. We found that the PMSF/Na₃VO₄ oxidant engages a partially distinct population of cysteines relative to pervanadate, indicating that different oxidant systems access distinct regions of cysteine reactivity space. This substrate selectivity is analogous to that observed among electrophilic compounds^6,11,20 and other oxidants^24,25, suggesting that oxidant chemistry, like electrophile structure, can be a key determinant of cysteine targeting.

Our interest in NAGK led to the discovery that two common lysis buffer components, PMSF and Na₃VO₄, react to form a cysteine-targeting oxidant. This facilitated the identification of a key active site cysteine in NAGK and of a PMSF/Na₃VO₄-based cysteine engagement approach complementary to existing resources. Overall, our study reveals that cysteine engagement profiles reflect not only intrinsic protein properties, but also the chemical environment in which labeling occurs. These findings underscore that cysteine reactivity is context-dependent and can be shaped by both biological and experimental conditions. This has important implications for the interpretation of chemoproteomic experiments, particularly those performed under conditions of metabolic or oxidative stress.

Limitations of the study

The major limitation of this study is that the identity of the PMSF/Na₃VO₄ oxidant remains unknown, although we may have some insight into the underlying chemistry. The related sulfonyl fluoride-based serine protease inhibitor AEBSF does not replicate PMSF’s oxidant activity (Figure S1I). The sulfur-fluoride bond in PMSF is rapidly hydrolyzed in aqueous solution, releasing hydrofluoric acid and a sulfonic acid group, whereas the equivalent bond in AEBSF is stable. We speculate that the released fluoride or the sulfonyl group could complex with orthovanadate in a manner that generates a reactive vanadium species, potentially involving peroxo bond formation. Similar peroxo bonds are thought to underlie the oxidative capacity of pervanadate^28,72. While both vanadium-based systems may ultimately act through peroxo intermediates, highly oxidized pervanadate species bearing multiple peroxo bonds⁷³ may differ in substrate specificity from a putative fluoride- or sulfonyl-coordinated peroxo vanadium complex with fewer such bonds. The chemistry appears selective for the vanadium oxoanion, as molybdate and tungstate did not react with PMSF to sulfonylate NAGK (Figure S2A). A second limitation is that optimal PMSF and Na₃VO₄ concentrations vary by target cysteine (Figures 5A and 5D–5F). Thus, optimizing reagent concentrations and lysis buffer volumes may be a prerequisite to applying this method. The same holds for cysteine-reactive compounds generally, though the abundance of prior studies has helped establish standard working concentrations.

Resource availability

Lead Contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, John Blenis, job2064@med.cornell.edu

Materials Availability

Constructs and cell lines generated in this study are available upon request.

Data and Code Availability

• IAA-ABPP proteomics data (Figure 5) was deposited to the ProteomeXchange Consortium via PRIDE⁷⁴ partner repository with the dataset identifier PXD076162.

• Mass spectrometry data identifying sulfonylation of NAGK C217 (Figure 2F and 2G) have been deposited to the ProteomeXchange Consortium via the PRIDE⁷⁴ partner repository with the dataset identifier PXD081578.

• Mass spectrometry data assessing potential differential phosphorylation of NAGK in PANC-1 cells grown in 3.6 vs 0.05 mM Q glutamine (Data S1, tabs 1 and 2) have been deposited to the MassIVE repository with the dataset identifier MSV000102508 (Title: PMSF and orthovanadate generate a cysteine-targeting oxidant with distinct selectivity) at the permanent URL: ftp://massive-ftp.ucsd.edu/v13/MSV000102508/ or accessible via the web interface at https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=b99761b013bf44fd9edcad979c4fd202

• All original code has been deposited at Zenodo at https://doi.org/10.5281/zenodo.21287567 and is publicly available.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR Methods

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell lines and culture conditions

Cell lines

Human pancreatic epithelioid carcinoma PANC-1 cells (male, ATCC, CRL-1469), human embryonic kidney 293T cells (female, ATCC, CRL-3216), human pancreatic adenocarcinoma HPAC cells (female, ATCC, CRL-2119), human colorectal adenocarcinoma DLD-1 cells (male, ATCC, CCL-221), human breast acantholytic squamous cell carcinoma HCC1806 cells (female, ATCC, CRL-2335), human breast adenocarcinoma MDA-MB-231 cells (female, ATCC, CRM-HTB-26), and human lung carcinoma A549 cells (male, ATCC, CRM-CCL-185) were obtained from ATCC. Mouse mT8 cells³⁷ (sex unknown) adapted to 2D culture were a kind gift from Rohit Chandwani. 293T, PANC-1, HPAC, MDA-MB-231, and mT8 cells were cultured in DMEM (Gibco, 11965-118) medium with 10% fetal bovine serum (FBS, Sigma, F0926) and 1% Penicillin-Streptomycin (PS, Thermo Fisher Scientific, 15140-163). HCC1806, A549, and DLD-1 cells were cultured in RPMI (Corning, 10-040-CV) medium with 10% FBS and 1% PS. These are referred to below as normal growth medias. Cells were grown at 37°C in 5% CO₂ in a humidified chamber and were routinely checked for mycoplasma.

Plating of cells for experimental analysis

In general, cells were plated such that approximately equal protein concentration was obtained across conditions at the point of lysis. 2e5 and 3.5e5 PANC-1 cells were plated per well in 6 well dishes for growth in 3.6 or 0.05 mM glutamine medium, respectively; 2e5 and 5e5 DLD-1 cells were plated per well in 6 well dishes for growth in 1.85 or 0.05 mM glutamine media, respectively; and 6e4 and 1.8e5 mT8 cells were plated per well in 6 well dishes for growth in 3.6 or 0.05 mM glutamine medium, respectively. For Figure S1A, between 6.4e4 and 1.7e5 HPAC cells were plated per well of a 6 well dish depending on the condition. For Figure S1C, 5.6e5 PANC-1 cells were seeded in 6 cm dishes on day 0. Early assays analyzing the NAGK mobility shift had the same number of cells plated regardless of nutrient condition (e.g. per well of a 6 well dish: 1.8e5 or 2.5e5 PANC-1 for 48 or 24 hour assays; 2.5e5 HCC1806, DLD-1, and A549; and 3e5 MDA-MB-231). This was later optimized to obtain equal protein concentration at lysis, although we found that mild density differences at lysis did not have a strong impact on the susceptibility of NAGK sulfonylation.

Experimental growth conditions

Unless otherwise specified, cells were grown in normal growth medium for one day after seeding prior to growth in assay medium. For assays with cells grown in experimental medium for 24 and 48 hours, cells were seeded on day 0. On day 1, 48-hour timepoints were washed twice in PBS and experimental medium (see below) was added. On day 2, 24-hour timepoints were washed twice in PBS and experimental medium was added, and 48-hour timepoints were fed with fresh experimental medium. Cells were harvested on day 3. For nutrient deprivation assays, dishes were washed 2x in room temperature PBS prior to administration of the following experimental medias:

Full nutrient DMEM: DMEM (Gibco, 11965-092) with 10% dialyzed FBS (dFBS, Sigma, F0392) and 1% PS; this medium contains 3.6 mM glutamine and 22.5 mM glucose after dFBS/PS addition.

0.2 mM glucose DMEM: glucose-free DMEM (Gibco, 11966-025) with 10% dFBS and 1% PS was supplemented to 0.2 mM glucose through the addition of full nutrient DMEM (see above).

0.05 mM glutamine DMEM: glutamine-free DMEM (Gibco, 11960-044) with 10% dFBS and 1% PS was supplemented to 0.05 mM glutamine through the addition of full nutrient DMEM.

Full nutrient RPMI: RPMI (Corning, 10-040) with 10% dFBS and 1% PS; this medium contains 1.85 mM glutamine and 10 mM glucose after dFBS/PS addition.

0.2 mM glucose RPMI: glucose-free RPMI (Corning, 10-043) with 10% dFBS and 1% PS was supplemented to 0.2 mM glucose through the addition of full nutrient RPMI.

0.05 mM glutamine RPMI: glutamine-free RPMI (Corning 15-040) with 10% dFBS and 1% PS was supplemented to 0.05 mM glutamine through the addition of full nutrient RPMI.

For media used in ¹³C₂ N-acetyl-D-glucosamine (GlcNAc) tracing assays, see “¹³C₂ GlcNAc tracing”.

Glutathione ethyl ester (GSEE, Sigma G1404 or Cayman Chemical 14953), KI696 (Cayman Chemical, 42039), and L-Buthionine-(S,R)-Sulfoximine (BSO, Cayman Chemical, 14484) were added to full nutrient or 0.05 mM glutamine media at the time of media administration. GSEE (10 mM working concentration) and BSO (1 mM working concentration) were directly reconstituted into media immediately prior to addition to cells. KI696 was reconstituted to 1 mM in DMSO and stored at −20°C, then diluted to 1 μM in media immediately prior to use (final media [DMSO] = 0.1%). Tris(2-methoxycarbonylethyl)phosphine hydrochloride (tmTCEP, AstaTech 66756) was reconstituted to 1 M in DMSO and stored at −20°C, then added to media at 1 mM for one hour prior to lysis (final media [DMSO] = 0.1%). pInducer20 gene expression was induced with the indicated concentration of doxycycline (EMD Millipore, 324385) at the time cells were treated with experimental media and/or compounds.

Method Details

Cloning

Mutagenesis and cloning of NAGK overexpression constructs

NAGK (Homo sapiens) in pDONR221 was obtained from Harvard PlasmidID (cat# HsCD00044726). All mutants of NAGK were generated in the pDONR construct using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs, E0552; see Table S1 for all cloning primers). An LR clonase (Invitrogen, 11791-100) reaction was performed to transfer the wild type and mutant NAGK cDNA into pInducer20⁷⁵ (Addgene, 44012) and/or pLenti PGK Neo DEST (w531-1)⁷⁶ (Addgene, 19067). NAGK-HA cloned into the PGK destination vector had additional mutation of the Kozak sequence to generate lower levels of overexpression and a synonymous mutation at R85 to mutate the PAM sequence to confer resistance to a sgRNA (the sgRNA was not used in this manuscript). The pInducer20 Empty Vector (EV) and PGK EV controls were generated through an LR clonase reaction with the pENTR1A entry plasmid (pENTR1A no ccDB (w48-1)⁷⁶, Addgene, 17398). pInducer20 GFP was previously reported⁷⁷. NAGK was cloned into the pGEX-2T construct (laboratory stock; backbone from GE Healthcare, GenBank U13850.1) through PCR amplification of the desired pDONR NAGK cDNA, digest of the PCR amplicon and pGEX-2T construct with XmaI and EcoRI followed by gel purification (Invitrogen, K210012), and subsequent ligation with T4 DNA ligase (New England Biolabs, M0202).

Cloning of CRISPR/Cas9 knockout constructs

The transient CRISPR/Cas9 construct pSpCas9(BB)-2A-Puro (PX459) V2.0⁷⁸ (Addgene, 62988) was digested with BbsI and gel purified, and the constitutive CRISPR/Cas9 construct lentiCRISPR v2-Blast (a gift from Mohan Babu, Addgene, 83480) was digested with BsmBI and gel purified. A 10 μl mix containing 100 pmole each forward and reverse sgRNA oligo (see Table S1 for sequences) and 0.5 μl T4 PNK (New England Biolabs, M0201) in 1x (1 μl) T4 ligation buffer (New England Biolabs, B0202) was incubated at 37°C to phosphorylate the oligos, then at 95°C for 5 minutes followed by a 1°/minute temperature decrease to anneal the oligos. The annealed sgRNAs (1 μl) were then ligated to 75 ng digested backbone in a 10 μl mixture containing 1 μl T4 ligase (New England Biolabs, M0202) and 1 μl T4 ligation buffer (New England Biolabs, B0202). The murine sgControl sequence is from Zafra et al.⁷⁹, the murine sgNAGK#4 sgRNA was designed by CRISPick (Doench et al.)⁸⁰, the human control sgNT2 sequence is also from Doench et al.⁸⁰, and the human sgNAGK sequence is from Kim et al.⁴⁹ See Table S1 for cloning primer sequences.

Generation of cell lines

Lentiviral production

To generate lentivirus, 4e6 293T cells were seeded per 10 cm plate (day 0) and transduced the following day (day 1) with a mixture containing 300 μl OPTI MEM (Gibco, 31985-070), 2.25 μg psPAX2 (a gift from Didier Trono, Addgene, 12260), 0.75 μg pMDG.2 (a gift from Didier Trono, Addgene, 12259), 3 μg plasmid of interest, and 18 μl X-tremeGENE HP DNA Transfection Reagent (Roche, 6366546001) that was incubated at room temperature for 20 minutes prior to drop-wise addition to the cells. On day 2, media was changed and 10 ml fresh media was added. On day 4, viral media was collected, passed through 0.22 μM filters, and stored at −20°C for later use.

Generation of inducible overexpression cell lines

To generate PANC-1 cells with pInducer20 expression of NAGK or control constructs (GFP or EV), 1e6 PANC-1 cells were plated in a 10 cm dish. Either on the day of plating or the following day, one tenth the pInducer20 virus generated from a 10 cm dish (see “Lentiviral production) was added per plate. The following day, cells were split and selected with 600 μg/ml G418. Once selection was complete, cells were maintained in 300 μg/ml G418.

Generation of PANC-1 NAGK knockout/rescue lines

To generate PANC-1 cells with knockout of endogenous NAGK and rescue with EV, NAGK-HA WT or C217S, 3.5e5 PANC-1 cells were seeded in 6 cm plates and infected with 350 μl lentivirus (see “Lentiviral production) containing pLentiCRISPR blast sgNT2 or sgNAGK (see “Cloning of CRISPR/Cas9 knockout constructs”). The following day, cells were split and selected with 20 μg/ml Blasticidin S HCl (Gibco, A11139-03). Following selection, cells were seeded as above and infected with 350 μl lentivirus containing pInducer20 EV, NAGK-HA WT or C217S, selected in 600 μg/ml G418, and maintained in 10 μg/ml Blasticidin S HCl and 300 μg/ml G418.

NAGK knockout/rescue mT8 cells

To generate mT8 cells with knockout of endogenous murine NAGK (mNAGK) and rescue with human NAGK (hNAGK), endogenous NAGK was knocked out with transient expression of px459 (see “Cloning of CRISPR/Cas9 knockout constructs”). To this end, 1.5e5 cells were seeded per well of a six well plate on day 0. On day 1, cells were fed with 2 ml DMEM with 10% FBS without antibiotics and transfected with a mix of 200 μl OPTI MEM, 4 μg px459 construct, and 20 μl FuGENE HD Transfection Reagent (Promega, E2311) that was allowed to incubate for 2–3 minutes prior to addition to cells. The following day, cells were split 1:1 into 6 μg/ml puromycin. Puromycin was removed after 24 hours, and cells were allowed to recover prior to the generation of single cell clones via FACS. Clones were verified with western blot and sequencing of genomic DNA amplified with the following primer sequences (see “Cloning of CRISPR/Cas9 knockout constructs” and Table S1 for sgRNA sequences):

Control Forward: TGCCAGGATGTCTTGACTATTCATG

Control Reverse: GAAACCACTGAGGTATGTAAGAGAGG

NAGK#4 Forward: GACCAGAGAACTTGGCCTTCATG

NAGK#4 Reverse: GAAAAGGTCTCTAGGGCCAGAGC

PCR amplicons from clones were sequenced with the Control Forward and NAGK#4 Forward primers. Selected clones were seeded in 6 well plates at 1.5e5 cells/well and infected with 500 μl lentivirus expressing PGK empty vector or human wild type NAGK-HA and maintained in 300 μg/ml G418 following selection.

Preparation of select reagents

Preparation of sodium orthovanadate

Preparation of sodium orthovanadate (Na₃VO₄, sodium vanadate, Fisher Scientific, S454-50) was based on the “Na₃VO₄ Stock Solution” protocol from Cold Spring Harbor Protocols. A 200 mM solution of Na₃VO₄ was prepared in water, and the pH was adjusted to 10 using NaOH or HCl as needed. The solution was boiled until clear and cooled to room temperature. The pH was readjusted to 10. The solution was repeatedly boiled, cooled, and pH adjusted to 10 until the solution remained clear (instead of yellow) at pH 10. Aliquots were stored at −20°C.

Preparation of pervanadate

Preparation of pervanadate was based on the protocol described in Huyer et al.²⁸ A mixture of 40 mM Na₃VO₄ (see “Preparation of sodium orthovanadate”) and 200 mM H₂O₂ (Sigma, H1009, diluted to 1M in 20 mM HEPES pH 7.3 prior to use) was made in a final volume of 1 ml water and incubated at room temperature for 5 minutes. To remove excess H₂O₂, ~130 U (40 μl of a ~3.3 U/μl stock) catalase from bovine liver (Sigma, C1345) was added to the reaction for ~2 minutes. The resulting 40 mM pervanadate stock consists of a number of complexes²⁸ and was made fresh for each experiment and used immediately. To control for the potential oxidizing effects of H₂O₂, H₂O₂ was diluted to 200 mM in water and treated with catalase as above.

Preparation of phenylmethylsulfonyl fluoride

A 200 mM stock of phenylmethylsulfonyl fluoride (PMSF, Sigma-Aldrich, P7626) was made in ethanol (KOPTEC, V1016) and stored at −20°C. Please note that PMSF generates hydrofluoric acid and hydrogen fluoride gas upon addition to aqueous buffers and should be used in the fume hood to prevent inhalation. The higher concentrations of PMSF used in this study are sufficient to produce irritating fumes.

Cell lysis protocol

Cells were washed 1x in 4°C PBS prior to lysis on ice (SDS and Laemmli lysis were at room temperature). Unless otherwise indicated, samples from 6 well plates were lysed in 200 μl lysis buffer, samples in 15 cm dishes for NAGK-HA immunoprecipitation were lysed in 1 ml 0.5% IGEPAL lysis buffer, samples in 15 cm dishes for cysteine alkylation and pulldown (HA or streptavidin) were lysed in 700 μl lysis buffer, and samples in 6 cm dishes were lysed in 400 μl lysis buffer. See below for lysis buffer components; see “Cysteine alkylation” for preincubation with or lysis in cysteine alkylating agents. Cells lysed in 0.5% IGEPAL, RIPA, and 1% Triton X-100 lysis buffer were lysed on ice and moved to 4°C for at least ten minutes, scraped into 1.5 ml sample tubes on ice, and centrifuged for 15 minutes at 4°C and 21k × g to remove insoluble debris. The supernatant was moved to fresh 1.5 ml sample tubes. Cells lysed in 2% SDS or Laemmli buffer were scraped to 1.5 ml sample tubes and did not require clarification. Samples were quantified using the DC Protein Assay (Bio-Rad, 5000112); samples lysed in Laemmli lysis buffer were not quantified and concentration was extrapolated using that of lysates lysed in parallel using other lysis buffers. Following quantification, 3x Laemmli buffer (160 mM Tris pH 6.8, 2.55% SDS, 25.5% glycerol, 15% β-mercaptoethanol, and bromophenol blue) was added to lysates (1 part 3x Laemmli buffer for every 2 parts lysate), excluding those directly lysed in 1x Laemmli buffer. Samples were heated to 95°C for several minutes and stored at −20°C.

Lysis buffer components

Unless otherwise stated, we used the 0.5% IGEPAL lysis buffer throughout the study for detection of sulfonylation.

0.5% IGEPAL lysis buffer

10 mM sodium fluoride (NaF, Sigma-Aldrich, S1504), 20 mM Na₃VO₄ unless otherwise indicated (Fisher Scientific, S454-50), 0.5–1 mM PMSF unless otherwise indicated (Sigma-Aldrich, P7626; PMSF generates toxic hydrogen fluoride gas upon exposure to aqueous solutions- use in the fume hood), 1.5 μM aprotinin (Sigma-Aldrich, A1153), 84 μM leupeptin hemisulfate (Research Products International, L22035), 1 μM pepstatin A (Research Products International, P30100), 50 mM Tris (a 1 M Tris pH 6.0 stock was used for lysis buffer with Na₃VO₄; pH 8.0 or pH 7.6 was used for lysis buffer without Na₃VO₄ because Na₃VO₄ is basic and increases the lysis buffer pH), 150 mM NaCl, and 0.5% v/v IGEPAL CA-630 (Sigma Aldrich, I3021). The exact concentration of PMSF required to differentially induce NAGK sulfonylation in cells grown in 3.6 versus 0.05 mM glutamine could vary in a batch and cell line dependent manner, but unless otherwise indicated, we used between 0.5 and 1 mM. PMSF must be added to the lysis buffer immediately prior to use as it undergoes rapid hydrolysis in aqueous buffers and will no longer induce sulfonylation of NAGK. If indicated, 4-(2-Aminoethyl)benzenesulfonyl fluoride hydrochloride (AEBSF, Sigma-Aldrich, SBR00015-1ML), Tris(2-carboxyethyl)phosphine hydrochloride (TCEP, Millipore Sigma, 580561), reduced glutathione (Thermo Fisher, 78259), dithiothreitol (DTT, Promega, V3155), sodium molybdate dihydrate (Sigma-Aldrich, M1003), sodium tungstate dihydrate (Sigma-Aldrich, 223336), ATP (Adenosine 5′-triphosphate magnesium salt, Sigma-Aldrich, A9187), and/or GlcNAc (Chem-Impex, 01427) were added to the lysis buffer at the indicated concentrations.

RIPA lysis buffer

150 mM NaCl, 1% v/v IGEPAL CA-630, 0.5% sodium deoxycholate (Sigma-Aldrich, D6750), 0.1% v/v sodium dodecyl sulfate (SDS, Amresco, M107), 50 mM Tris pH 7.4, 10 mM NaF, 20 mM Na₃VO₄, 1 mM PMSF, 1.5 μM aprotinin, 84 μM leupeptin, and 1 μM pepstatin A.

Triton X-100 lysis buffer

25 mM HEPES (Sigma-Aldrich, H4034) pH 7.4, 100 mM NaCl, 10% v/v glycerol, 1% v/v Triton X-100 (Sigma-Aldrich, T8787), 10 mM NaF, 20 mM Na₃VO₄, 1 mM PMSF, 1.5 μM aprotinin, 84 μM leupeptin, and 1 μM pepstatin A.

2% SDS lysis buffer

2% v/v SDS, 50 mM Tris pH 6.8, 10 mM NaF, 20 mM Na₃VO₄, 1 mM PMSF, 1.5 μM aprotinin, 84 μM leupeptin, and 1 μM pepstatin A. In Figure S6E, 5 mM EDTA was added to the SDS lysis buffer.

Laemmli buffer

For direct lysis of cells in Laemmli buffer, 3x Laemmli buffer (see “Cell lysis”) was diluted to 1x in 0.5% IGEPAL lysis buffer. The 1x Laemmli stock in IGEPAL lysis buffer was then used to lyse cells. For components of the 3x Laemmli buffer, see “Cell lysis.”

SDS-PAGE and Western Blot

11% or 14% SDS-PAGE gels were cast in large glass plates (C.B.S. Scientific: NGP-190NR, VGC-1520, VGE-1519, VGS-1519, GPC0002). The resolving gel was ~12.5 cm long and consisted of 11% acrylamide (14% for 14% gels, National Diagnostics, EC-890), 0.375 M Tris pH 8.8, 0.1% SDS, 0.1% ammonium persulfate (APS, Sigma-Aldrich, A3678), 0.04% TEMED (Sigma-Aldrich, T7024). The stacking gel consisted of 5% acrylamide, 0.127 M Tris HCl pH 6.8, 0.1% SDS, 0.1% APS, 0.1% SDS, and 0.1% TEMED. Typically, 20 μg whole cell lysate protein was loaded per sample (30 μg was loaded for the HOXB6 blot in Figure 5F). To observe the NAGK mobility shift, the gel was resolved for ~4.5 hours at 250 V until approximately 3 cm of separation was obtained between the 25 and 37 kDa protein standards (Bio-Rad; 1610375). Proteins were transferred (30 V, 4°C, overnight) to 0.45 μM pore size nitrocellulose membrane (0.22 μM for proteins of ~20 kDa or less, Amersham, 10600002 or BioRad 1620115, 1620112) using a discontinued variant of BioRad’s Trans-Blot Cell system. Unless otherwise indicated, primary antibodies were used at 1:1000 in a 1:1 mix of Intercept (TBS) Blocking Buffer (LICORbio, 927-60050) and Tris-buffered saline with Tween 20 (TBST). The primary antibodies used were rabbit polyclonal anti-NAGK (Proteintech, 15051-1-AP), rabbit monoclonal anti-Akt (pan) (C67E7) (Cell Signaling Technology, 4691S; used at 1:1000 in 5% milk in TBST), rabbit monoclonal anti-phospho-Akt (Ser473) (D9E) (Cell Signaling Technology, 4060; used at 1:1000 in 5% milk in TBST), mouse monoclonal anti-Vinculin (Sigma, V9264; used at 1:10,000 in Intercept or 5% milk), rabbit monoclonal anti-HA-Tag (C29F4) (Cell Signaling Technology, 3724), rabbit polyclonal anti-PTPN1/PTP1B (Proteintech, 11334-1-AP), rabbit polyclonal anti-Ran (Proteintech, 10469-1-AP), rabbit monoclonal anti-Nrf2 [EP1808Y] (Abcam, ab62352), rabbit polyclonal anti-p38 MAPK (Cell Signaling Technology, 9212), rabbit polyclonal anti-phospho-p38 MAPK (Thr180/Tyr182) (Cell Signaling Technology, 9211), rabbit monoclonal anti-BiP (C50B12) (Cell Signaling Technology, 3177), rabbit polyclonal anti-SOD1 (AB Clonal, A0274), mouse monoclonal anti-GAPDH-71.1 (Sigma Aldrich, G8795), mouse monoclonal anti-Oxidized PTP Active Site (oxPTP, R&D Systems, MAB2844), rabbit polyclonal anti-Ras (Cell Signaling Technology, 3965), rabbit monoclonal anti-Src (32G6) (Cell Signaling Technology, 2123), rabbit polyclonal anti-KIF3A (Proteintech, 13930-1-AP), mouse monoclonal anti-HoxB6 (Santa Cruz Biotechnology, sc-166950, 1:500), rabbit polyclonal anti-PCCB (Proteintech, 11139-1-AP), CIAPIN1 (Proteintech, 12638-1-AP), rabbit polyclonal anti-SHB (Proteintech, 31967-1-AP), rabbit polyclonal anti-BLVRB (Proteintech, 17727-1-AP), rabbit polyclonal anti-ARHGEF7 (Proteintech, 14092-1-AP), rabbit monoclonal anti-S6 Ribosomal Protein (5G10) (Cell Signaling Technology, 2217), rabbit polyclonal anti-AP1M1 (Proteintech, 12112-1-AP), rabbit polyclonal anti-ALDOA (Proteintech, 11217-1-AP), and rabbit polyclonal anti-CLIP-115/CLIP2 (Proteintech, 25088-1-AP).

The following secondary antibodies were used, all from LICORbio: IRDye^® 800CW Donkey anti-Mouse IgG Secondary Antibody (926-32212), IRDye^® 800CW Donkey anti-Rabbit IgG Secondary Antibody (926-32213), IRDye^® 680RD Donkey Anti-Mouse IgG Secondary Antibody (926-68072), IRDye^® 680RD Donkey anti-Rabbit IgG Secondary Antibody (926-68073). Secondary antibodies were used at 1:10,000 in the same buffer as the respective primary antibody. Biotinylated probes were detected using Streptavidin (LICORbio, 926-32230, 1:2000 in 1:1 Intercept:TBST + 0.02% SDS). Membranes were developed using the LI-COR Odyssey CLx system. Select bands were quantified using LICORbio’s Image Studio software.

Immunoprecipitation followed by λ phosphatase

PANC-1 cells stably expressing pInducer20 GFP or NAGK-HA were seeded in 15 cm plates in normal growth media (see “Cell culture”), 3.6e6 and 5.6e6 cells per plate for 3.6 and 0.05 mM glutamine, respectively. The following day, cells were washed 2x in 15 ml room temperature PBS and fed with 20 ml 3.6 or 0.05 mM glutamine DMEM (see “Cell culture: Experimental growth conditions”) containing 500 ng/ml doxycycline (EMD Millipore, 324385). The next day, cells were washed once in 10–15 ml cold PBS and lysed on ice in 1 ml 0.5% IGEPAL lysis buffer with 40 mM Na₃VO₄ as opposed to the typical 20 mM (see “Lysis buffer components: 0.5% IGEPAL lysis buffer”), as the larger volume of residual PBS in a 15 cm dish dilutes the lysis buffer. Lysates were clarified by centrifugation at 21k × g for 15 minutes at 4°C then quantified using the DC assay (Biorad, 5000112). Some whole cell lysate was saved for western blot analysis (typically about 100 μl). 70 μl α-HA agarose (Sigma Aldrich, A2095) per sample was washed 3x in IP buffer (50 mM Tris pH 8.0, 150 mM NaCl, 0.5% IGEPAL) with 1 minute, 4°C, 200 × g centrifugation between washes. 2 mg of each sample was loaded to the washed α-HA agarose. Samples were incubated for 2–3 hours, rotating at 4°C, then washed 4x in IP buffer and 2x in PBS as above. NAGK-HA was eluted with 75 μl 400 μg/ml Hemagglutinin (HA) Peptide (Sigma-Aldrich, I2149) in PBS for 30 minutes at room temperature with agitation. The supernatant was recovered, the elution step was repeated, and the two eluates were combined. Samples were divided in half and incubated in 9 μl PMP buffer, 9 μl 10 mM MnCl2, ± 2 μl λ protein phosphatase (all components of NEB, P0753S) for 30 minutes at 30°C. 3x Laemmli buffer was added 1:2 to each sample prior to heating to 95°C and storage at −20°C. Samples were analyzed on 11% SDS-PAGE followed by western blot (see “SDS-PAGE and Western Blot”).

Mass spectrometry analysis of immunoprecipitated NAGK for phosphorylation

Cell culture and immunoprecipitation

PANC-1 cells stably expressing pInducer20 GFP (for 3.6 mM glutamine only) or NAGK-HA (for 3.6 and 0.05 mM glutamine) were seeded in 15 cm plates in normal growth media (see “Cell culture”), with 3.6e5 cells per plate for all growth conditions (2–6× 15 cm plates per condition). The following day, cells were washed 2x in 15 ml room temperature PBS and fed with 20 ml 3.6 or 0.05 mM glutamine DMEM (see “Cell culture: Experimental growth conditions”) containing 500 ng/ml doxycycline (EMD Millipore, 324385). Two days later, cells were lysed as described in “Immunoprecipitation followed by λ phosphatase”. Samples from the same condition across multiple dishes were combined, and 5× 2 mg immunoprecipitations were set up as described in “Immunoprecipitation followed by λ phosphatase”. Following washes, one immunoprecipitation per condition was eluted in 90 μl 3x Laemmli buffer without β-mercaptoethanol and bromophenol blue by heating sample to 95°C for 5 minutes. Eluted samples were centrifuged briefly to pellet beads, and the Laemmli supernatant containing the elution fraction was transferred to the next of the five immunoprecipitation samples/condition. Elution was repeated with the same 90 μl across all five immunoprecipitation samples/condition to keep sample volume low. After all samples were eluted, 3x normal Laemmli buffer was added to each elution, and samples were heated to 95°C and stored at −20°C. Immunoprecipitated samples were run on 11% SDS-PAGE, the gel was stained with SimplyBlue SafeStain (Invitrogen, 465034), and the top and bottom bands of each sample were excised for MS analysis.

In-Gel Digestion and Peptide Desalting

18 excised gel bands corresponding to 9x upper (slower migrating) and 9x-lower (faster migrating) immunoprecipitated HA-NAGK, or GFP only controls, each performed in triplicate, were subjected to standard in-gel digestion⁸¹. Briefly, gel pieces were destained, and proteins were reduced with dithiothreitol (DTT) and alkylated with iodoacetamide (with the alkylation reaction subsequently quenched by additional DTT). Proteins were digested in-gel overnight at 37°C using a combination of Lys-C (Wako/Fuji Chemical) and sequencing-grade modified trypsin (Promega, cat. no. V5111). Resulting peptides were extracted from the gel slices using aqueous mixtures of acetonitrile and formic acid, and dried to completeness in a vacuum concentrator. Prior to LC-MS/MS analysis, the extracted peptides were reconstituted and desalted using hand-packed C18 STAGE (stop and go extraction) tips⁸².

Liquid Chromatography and Mass Spectrometry (LC-MS/MS)

Desalted peptides were analyzed on an Orbitrap Fusion mass spectrometer equipped with a FAIMS Pro interface and a nanoelectrospray ion source (Thermo Fisher Scientific), coupled inline to an EASY-nLC 1000 high-performance liquid chromatography system. Peptides were separated on a reverse-phase analytical column (75 μm inner diameter) packed in-house with C18 resin. Mobile phase A consisted of 0.1% formic acid in water, and mobile phase B consisted of 0.1% formic acid in 100% acetonitrile. The chromatographic separation was achieved using a 45-min gradient: 5% to 26% B over 35 min at 350 nL/min, followed by an increase to 50% B over 3 min at 400 nL/min, a ramp to 100% B over 0.5 min at 600 nL/min, and a final column wash at 100% B for 6.5 min.

The mass spectrometer was operated in data-dependent acquisition (DDA) mode using the FAIMS interface. The method continuously cycled through five different compensation voltages (CV: −40, −50, −60, −70, and −80 V). To optimize the detection of phosphopeptides, selected precursor ions were subjected to a dual-fragmentation MS2 strategy. First, precursors were fragmented using higher-energy collisional dissociation (HCD) and detected in the Orbitrap at a resolution of 30,000. Second, a multi-stage activation (MSA) event triggered by the neutral loss of phosphoric acid (97.9769 Da) was enabled, wherein triggered precursors were subjected to collision-induced dissociation (CID) and similarly analyzed in the Orbitrap. Complete, mass spectrometry instrument methods are deposited as text files in the MassIVE repository (see “Data and Code Availability”).

Data Processing and Protein Identification

Raw mass spectrometry data files were searched against the canonical human UniProt reference proteome database (downloaded May 11, 2019), concatenated with a reversed-sequence decoy database⁸³. Searches were performed using the Comet search engine (version 2020.01 rev. 4)⁸⁴. The precursor ion mass tolerance was set to 20 ppm and fragment ion mass tolerance was set to 0.02 Da. Carbamidomethylation of cysteine (+57.02146 Da) was defined as a fixed modification. Oxidation of methionine (+15.9949 Da) and phosphorylation of serine, threonine, and tyrosine (+79.96633 Da) were set as variable modifications, allowing for up to three variable modifications per peptide. To strictly control the false discovery rate (FDR), peptide spectral matches (PSMs) and protein identifications were filtered to a 1% FDR using linear discriminant analysis (LDA)⁸⁵.

In vitro of sulfonylation of recombinant NAGK

For western blot analysis

50 ng His-NAGK (Novus Biologicals, NBP1-99050) was incubated in 50 μl 0.5% IGEPAL lysis buffer (see “0.5% IGEPAL lysis buffer”) with or without the indicated components (e.g. PMSF, Na₃VO₄, ATP, GlcNAc) for ten minutes on ice. 3x Laemmli buffer was added 1:2 to samples, samples were heated to 95°C, stored at −20°C, and run on 11% SDS-PAGE followed by western blot as indicated in “SDS-PAGE and Western Blot”.

In Figure S2D, 50 ng recombinant His-NAGK was incubated in 50 mM Tris (pH 8.0 for conditions without PMSF and Na₃VO₄; pH 6.0 for conditions with PMSF and Na₃VO₄) and 150 mM NaCl with or without 1 mM PMSF, 20 mM Na₃VO₄, the indicated concentration of H₂O₂ (Sigma Aldrich, H1009), 20 U catalase (Sigma-Aldrich, C1345; for some conditions, the catalase was first heat inactivated at 95°C for 5 minutes), and/or the equivalent mg to 20 U catalase of bovine serum albumin (BSA, P212121, CI-00535-250G). H₂O₂ and PMSF/Na₃VO₄ were added to the Tris/NaCl buffer followed by catalase or BSA. After a 3–4 minute incubation to allow the catalase to react with H₂O₂, NAGK was added to the reaction buffer.

MS analysis

To identify the modification underlying the mobility shift, 2 μg His-NAGK (Novus Biologicals, NBP1-99050) was incubated in 600 μl 0.5% IGEPAL lysis buffer ± 1 mM PMSF and 20 mM Na₃VO₄ for ten minutes on ice. The samples were concentrated by centrifugation at 4k × g to <100 μl using Amicon Ultra centrifugal filters (Amicon, UFC200324). 3x Laemmli buffer was added to samples, and samples were heated to 95°C, stored at −20°C, and run on 11% SDS-PAGE. The gel was stained with SimplyBlue SafeStain (Invitrogen, 465034), and the top and bottom bands of each sample (± 1 mM PMSF and 20 mM Na₃VO₄) were excised for MS analysis.

Protein samples were digested in-gel with trypsin overnight at 37 °C following reduction with dithiothreitol (DTT) and alkylation with iodoacetamide. The resulting peptides were concentrated by vacuum centrifugation to near dryness and desalted using C18 stage-tip columns prior to LC-MS/MS analysis.

Liquid chromatography–mass spectrometry (LC-MS/MS) was performed using an EASY-nLC 1000 system coupled online to a Fusion Lumos mass spectrometer (Thermo Fisher Scientific). Peptide separation was carried out using a 75 μm × 20 cm in-house packed column (ReproSil-Pur C18-AQ, 3 μm; Dr. Maisch GmbH, Germany) with Buffer A (0.1% formic acid in water) and Buffer B (0.1% formic acid in acetonitrile) as mobile phases. Peptides were eluted with a linear gradient of 3–30% Buffer B over 50 minutes, followed by 30–90% Buffer B over 10 minutes, at a flow rate of 300 nL/min.

For the analysis of the protein modifications, the mass spectrometer was operated in data-dependent acquisition (DDA) mode. Full MS scans were acquired in the Orbitrap mass analyzer over a range of 300–1500 m/z at a resolution of 60,000 (at m/z 200). The top 15 most intense precursor ions (charge states 2–6) were selected for fragmentation using higher-energy collisional dissociation (HCD) with a normalized collision energy of 35 and an isolation window of 1.6 m/z. MS/MS spectra were acquired in the Orbitrap at a resolution of 15,000. Automatic gain control (AGC) targets were set to 1e6 for MS scans and 5e4 for MS/MS scans, with a maximum injection time of 50 ms for both.

DDA raw files were processed using MaxQuant version 1.6.17.0 (Max Planck Institute, Munich, Germany). Protein identification was performed against the UniProt human NAGK protein sequence (downloaded August 7, 2021). Variable modifications included carbamidomethylation of cysteine, methionine oxidation, protein N-terminal acetylation, sulfinylation and sulfonylation of cysteine, and PMSF modification of serine and threonine. Precursor and fragment ion mass tolerances were set to 7 ppm and 20 ppm, respectively, allowing up to two missed cleavages. Both peptide and protein identifications were filtered at a 1% false discovery rate (FDR) using a decoy database of reversed protein sequences. Identified peptides with modifications were manually inspected.

Cysteine alkylation for western blot

NEM alkylation prior to lysis

2.5e5 PANC-1 cells were seeded per well into normal growth medium (see “Cell culture”) in 6-well plates. The following day, cells were washed 2x with PBS and fed with 2 ml 3.6 or 0.05 mM glutamine DMEM (see “Cell culture: Experimental growth conditions”). If PANC-1 cells contained the pInducer20 construct, 500 ng/ml doxycycline was added at the time of the media change. The next day, a 20 mM N-ethylmaleimide (NEM, Pierce, 23030) stock was made immediately prior to use in serum-free DMEM with or without glutamine for cells grown in 3.6 or 0.05 mM glutamine, respectively. Cells were washed twice with PBS, incubated in 20 mM NEM for 10 minutes, washed 4x with cold PBS, and lysed on ice in 0.5% IGEPAL lysis buffer (see “0.5% IGEPAL lysis buffer”) with or without Na₃VO₄ as indicated. Samples were collected to 1.5 ml sample tubes and clarified by centrifugation at 21k × g for 15 minutes at 4°C. Supernatant was transferred to a fresh sample tube, samples were quantified with the DC Assay, 3x Laemmli buffer was added to samples, and samples were heated to 95°C for several minutes and stored at −20°C.

Iodoacetamide-alkyne, streptavidin pulldown

PANC-1 cells expressing the indicated doxycycline-inducible pInducer20 construct were seeded in 15 cm plates to yield similar protein concentration at lysis (3.6e6 cells for growth in 3.6 mM glutamine or 5.6e6 cells for growth in 0.05 mM glutamine) and grown in normal growth medium overnight. The following day, plates were washed 2x in PBS and fed with 20 ml 3.6 or 0.05 mM glutamine DMEM (see “Cell culture: Experimental growth conditions”) containing 500 ng/ml doxycycline. The next day, cells were lysed at room temperature in 700 μl 0.5% IGEPAL lysis buffer (see “Lysis buffer components: 0.5% IGEPAL lysis buffer”) made with 50 mM Tris pH 8.0, no Na₃VO₄, and ± 50 μM N-Hex-5-ynyl-2-iodo-acetamide (iodoacetamide alkyne or IAA, Oakwood Chemicals, 95226). Lysates were incubated at room temperature for 40 minutes, clarified at 21k × g for 15 minutes, and quantified using the DC Assay (Bio-Rad, 5000112). Protein concentration and volume were standardized across samples. Approximately 3000 μg protein was then used to set up a click chemistry reaction (exact amount varied between biological replicates but was consistent within an experiment). To this end, 200 μM desthiobiotin azide (Click Chemistry Tools/Vector Laboratories, CCT-1107-5), 5.43 mM tris-hydroxypropyltriazolylmethylamine (THPTA, Click Chemisty Tools/Vector Laboratories, CCT-1010), 1 mM CuSO₄ (Sigma-Aldrich, PHR1477), and 16.3 mM (+)-Sodium L-ascorbate (Sigma, A4034) were added to each sample for a final volume of approximately 1 ml. The click reaction proceeded for 30 minutes at room temperature in the dark, and samples were snap frozen. Within a week, samples were thawed, and protein was precipitated with 5 volumes of acetone for 20 minutes at −20°C to facilitate the removal of unreacted click reagents. Protein was pelleted at 10k × g for 10 minutes at 4°C, and the supernatant was discarded. The sample was spun again to remove residual acetone. Samples were air dried for 10 minutes, and protein was reconstituted in 300 μl 2% SDS in PBS by heating to 95°C for 15–20 minutes with vortexing and/or pipetting every ~5 minutes. Insoluble protein was removed by centrifugation for 15 minutes at 21k × g at room temperature. The supernatant was transferred to a new tube and diluted to 0.1% SDS in PBS. Samples were quantified using the DC assay and the volume and concentration were equalized. 200 μl streptavidin agarose (Sigma-Aldrich, S1638) per sample was washed 3x in PBS with 2-minute, 500 × g centrifugations to remove washes, and samples were added to washed beads. Typically, 1.5–2 mg protein was added per sample to streptavidin agarose (the exact amount varied between biological replicates but was consistent within an experiment). Samples were rotated at room temperature for 2–3 hours; washed 3x in 10–12 ml PBS at room temperature with 2-minute, 500 × g centrifugation to remove the washes; transferred to 1.5 ml sample tubes; and eluted by boiling in 80 μl 3x Laemmli buffer. An equal volume of elution per sample was then analyzed on 11% SDS-PAGE followed by western blot (see “SDS-PAGE and Western Blot”).

Maleimide-biotin, αHA immunoprecipitation

PANC-1 pInducer20 cell lines were seeded and treated until lysis as described in “Iodoacetamide-alkyne, streptavidin pulldown”. Plates were washed in cold PBS on ice and lysed at room temperature in 700 μl lysis buffer containing 50 mM Tris pH 10.0 (the higher pH Tris was to accommodate the maleimide-biotin probe that was reconstituted in acetic acid), 150 mM NaCl, 0.5 % IGEPAL CA-630, 10 mM NaF, 1.5 μM aprotinin, 84 μM leupeptin, 1 μM pepstatin A, and either 50 μM maleimide-biotin (Sigma, B1267, from a 40 mM stock solution prepared in acetic acid) or acetic acid alone as a vehicle control. Samples were incubated in the dark at room temperature for 40 minutes. Lysates were clarified (15 minutes at 21k × g, 4°C) and quantified with the DC Assay, and protein concentration was equalized. 70 μl αHA-conjugated agarose beads (Sigma-Aldrich, A2095) per sample were washed 3x in IP buffer (50 mM Tris pH 8.0, 150 mM NaCl, and 0.5% IGEPAL) with 200 × g, 1 minute centrifugation at 4°C. 2 mg each sample was added to the washed beads and incubated at 4°C, rotating, for 2–3 hours. Samples were then washed 5x in IP buffer with 200 × g centrifugation for 1 minute at 4°C. Immunoprecipitated samples were eluted by heating to 95°C for 5 minutes in 50 μl 3x Laemmli buffer and stored at −20°C. An equal volume of elution per sample was analyzed on 11% SDS-PAGE followed by western blot (see “SDS-PAGE and Western Blot”).

Maleimide-biotin, streptavidin pulldown

PANC-1 (3.6e6 for full nutrient media, 5e6 for 0.2 mM glucose, 5.6e6 for 0.05 mM glutamine), HCC1806 (2.8e6 for full nutrient media, 6e6 for each 0.2 mM glucose and 0.05 mM glutamine) and DLD-1 (2.4e6 for full nutrient media, 4.5e6 for 0.2 mM glucose, and 5e6 for 0.05 mM glutamine) were plated in 15 cm plates on day 1. On day 2, cells were washed 2x in PBS and fed with 20 ml media containing full nutrients, 0.2 mM glucose, or 0.05 mM glutamine (see “Cell culture: Experimental growth conditions”). On day 3, lysates were labeled with maleimide-biotin, clarified, quantified, and equalized as described in “Maleimide-biotin, αHA immunoprecipitation”, and then snap frozen in liquid nitrogen for storage at −80°C for up to a week. Samples were then thawed, precipitated, reconstituted, equalized (1.4–4.5 mg protein was used for streptavidin pulldown depending on the cell type and biological experiment but was consistent within each cell type and biological replicate), and processed for streptavidin pulldown as described in “Iodoacetamide-alkyne, streptavidin pulldown”.

Maleimide-biotin, whole cell lysate analysis

Cells were seeded in 6 well dishes (PANC1: 2e5 for growth in full nutrient media, 3.5e5 for growth in 0.05 mM glutamine media; DLD-1: 2.5e5 for growth in full nutrient media, 5e5 for growth in 0.05 mM glutamine media; mT8: 6e4 for growth in full media, 1.8e5 for growth in 0.05 mM glutamine media) and grown in normal growth medium overnight. The following day, plates were washed 2x in PBS and fed with 2 ml full nutrient or 0.05 mM glutamine media containing any additional indicated compounds (see “Cell culture: Experimental growth conditions”). The next day, cells were washed in cold PBS on ice and lysed at room temperature in 200 μl lysis buffer as described in “Maleimide-biotin, αHA immunoprecipitation”. Lysates were clarified (15 minutes at 21k × g, 4°C) and quantified with the DC Assay. 3x Laemmli buffer was added to samples, and samples were heated to 95°C for several minutes prior to storage at −20°C. Samples were analyzed on 11% SDS-PAGE followed by western blot (see “SDS-PAGE and Western Blot”).

Iodoacetyl-PEG2-Biotin labeling

PANC-1 cells were seeded into normal growth medium in 10 cm dishes, 1.4e6/plate for full nutrient media and BSO treatment and 2.2e6 for 0.05 mM glutamine media. The following day, cells were washed twice with PBS and fed with 3.6 mM glutamine media ± 1 mM BSO, or 0.05 mM glutamine media (see “Experimental growth conditions”). To label cells after lysis, plates were washed 2x in 10 ml cold PBS on ice and scraped to 2 ml sample tubes in 1 ml PBS. Cells were pelleted at 200 × g for 3 minutes and resuspended in 450 μl DBPS (Corning, 21-031-CV) containing 10 mM NaF, 1.5 μM aprotinin, 84 μM leupeptin hemisulfate, and 1 μM pepstatin A. Cells were disrupted using a Branson Sonifier 250 (2× 8 pulses at a 30% duty cycle and output of 1.75). Lysates were clarified by centrifugation at 9000 × g for 10 minutes and quantified with the DC assay. Protein concentration was standardized across samples within each biological assay (between 1.17 and 1.5 mg/ml). 450 μl lysate was incubated with 50 μM Iodoacetyl-PEG2-Biotin (IAB, 0.9 μl/sample of a 25 mM stock; ThermoFisher Scientific, 21334) for 40 minutes rotating in the dark at room temperature. 3x Laemmli buffer was added to samples, and samples were heated to 95°C for several minutes prior to storage at −20°C. To label cells at the point of lysis, plates were washed 1x in cold PBS on ice and lysed in 500 μl 50 mM Tris pH 8.0, 150 mM NaCl, 0.5% IGEPAL CA-630, 10 mM NaF, 1.5 μM aprotinin, 84 μM leupeptin hemisulfate, and 1 μM pepstatin A, ± 50 μM IAB. Lysates were labeled for 40 minutes at room temperature, rotating and protected from light. Lysates were clarified by centrifugation for 15 minutes at 21k × g, 4°C. Supernatant was transferred to a fresh 1.5 ml sample tube, and protein was quantified with the DC assay. 3x Laemmli buffer was added to each sample, and samples were heated to 95°C for several minutes and stored at −20°C. Samples were analyzed on 11% SDS-PAGE followed by western blot (see “SDS-PAGE and Western Blot”).

IAA-ABPP Proteomics

Cysteine Labeling

PANC-1 cells were seeded in 12 × 15 cm plates, 3.5e6 cells/plate, on day 0 for lysis on day 2. The four oxidant conditions used were no oxidant, 1.2 mM PMSF and 40 mM Na₃VO₄, 4 mM PMSF and 60 mM Na₃VO₄, and 5 mM pervanadate. Higher PMSF/Na₃VO₄ concentrations were used to lyse 15 cm plates than to lyse 6-well plates so the lysis volume could be kept relatively small (1 ml) while maintaining an approximately equivalent oxidative capacity per mg. A repeated caution from above: PMSF generates hydrofluoric acid and hydrogen fluoride gas upon addition to aqueous buffers and should be used in the fume hood to prevent inhalation during dilution and lysis buffer use, especially at the higher concentrations of 1–4 mM used here On day 2, 2x stocks of IGEPAL lysis buffers were prepared without PMSF or Na₃VO₄ and with Tris pH 8.0 for eventual lysis with no oxidant or 5 mM pervanadate and with Tris pH 6.0 for eventual lysis in 1.2 mM PMSF plus 40 mM Na₃VO₄ or 4 mM PMSF plus 60 mM Na₃VO₄ because Na₃VO₄ is basic. See “Lysis buffer components: 0.5% IGEPAL lysis buffer” for making 1x IGEPAL lysis buffer. 40 mM pervanadate was prepared fresh for each replicate and immediately prior to use as described in “Preparation of select reagents: Preparation of pervanadate”. 1x lysis buffers for each oxidant condition were prepared immediately prior to lysis of each replicate by adding water, PMSF, Na₃VO₄, and/or pervanadate aliquots to the 2x IGEPAL lysis buffer. Within each replicate, cells were washed once on ice in cold PBS, then lysed on ice in the fume hood in 1 ml/plate of the respective lysis buffer. Cells were immediately scraped and approximately 5 minutes after lysis, 990 ul lysate was added to a 1.5 ml sample tube containing 10 μl of 100 mM N-hex-5-ynyl-2-iodoacetamide (iodoacetamide-alkyne, IAA; Oakwood Chemicals, 95226) for a final concentration of 1 mM IAA. Samples were incubated for 15 minutes in the dark at room temperature, with shaking at 750 rpm on an Eppendorf ThermoMixer C, prior to snap freezing in liquid nitrogen (see next paragraph for next steps).

During IAA labeling, two aliquots of residual lysate were 1) labeled with 1 mM IAB (Thermo Scientific, 21334; 2 ul 25 mM stock in 48 μl lysate) for 15 minutes in the dark at room temperature for western blot control of decreased labeling with oxidation or 2) left unlabeled for western blot confirmation that oxidation occurred via expected mobility shifts. Control lysates were then snap frozen in liquid nitrogen. After all triplicate lysates were processed, control samples were thawed on ice and clarified for 30 minutes at 21 k × g, 4°C. Supernatants were transferred to fresh 1.5 ml sample tubes, quantified with the DC Assay, and mixed with 3x Laemmli sample buffer for western blot analysis.

IAA-labeled proteins were thawed, clarified at 15k rpm for 10 minutes at 4°C, and conjugated to dialkoxydiphenylsilane (DADPS) biotin azide (Vector Laboratories, Cat# CCT-1330) to enable streptavidin enrichment and acid-mediated peptide release, thereby improving proteomic depth by eliminating endogenously biotinylated species. Each condition (no oxidation, 1.2 mM PMSF/40 mM Na₃VO₄, 4 mM PMSF/60 mM Na₃VO₄, and 5 mM pervanadate) was prepared in triplicate. After protein quantification, 1.1 mg of total protein from each sample was subjected to biotinylation. Samples were adjusted to a final volume of 700 μL with lysis buffer. To derivatize biotin on labeled cysteine residues, 15 μL of 50 mM DADPS biotin azide (Vector Laboratories, Cat# CCT-1330) and 300 μL of copper-catalyzed click reaction premix were added to initiate the copper click reaction. The premix consisted of 100 μL 100 mM Tris-hydroxypropyltriazolylmethylamine (THPTA), 100 μL 20 mM Copper(II) sulfate (CuSO₄, Sigma-Aldrich, C1297-100), and 100 μL 300 mM sodium ascorbate (Sigma-Aldrich, A7631). Click chemistry reactions were done for 1.5 h at 25 °C in the dark with agitation at 750 rpm. To remove unreacted chemicals, 600 μL ice-cold methanol and 100 μL chloroform were added, and the mixture was vortexed briefly. White and cloudy precipitated proteins were centrifuged at 20,000 × g for 10 min at 4 °C, and the upper aqueous phase was carefully discarded. The protein pellet was washed with 1,000 μL ice-cold methanol, then centrifuged (20,000 × g, 10 min, 4 °C) and the supernatant removed. This methanol wash step was repeated, and the protein pellet was subsequently air-dried for at least 15 min.

Bead enrichment

The dried protein pellet was reconstituted in 600 μL 2.5% SDS/PBS buffer and allowed to dissolve completely (typically 30 min, variable by sample). After quantification, the same amount of biotinylated samples was taken, and the volume was adjusted to the same using 2.5% SDS/PBS buffer. The sample was transferred to a 15 mL conical tube and diluted with PBS to reduce SDS concentration to below 0.125%. Streptavidin Sepharose High Performance affinity resin (200 μL, Cytiva, 17511301) was added, and samples were incubated overnight at 4 °C with rotation. Beads were pelleted by centrifugation at 1,700 ×g for 5 min, washed sequentially (2 times each) with 1% SDS/PBS (1 mL), PBS (1 mL), 6 M urea (1 mL, Fisher Chemical, U15-500), and PBS (1 mL), with centrifugation at 1,700 × g for 3 min between washes. Beads were resuspended in 700 μL 100 mM ammonium bicarbonate (ABC, Sigma-Aldrich, A6141). Proteins were subsequently incubated with 7 μL of 0.5 M DTT at room temperature (RT) for 30 min at 750 rpm and 14 μL of 0.5 M iodoacetamide (Sigma-Aldrich, I6125) in the dark at RT for 30 min. Next, 0.7 μL of 1 M CaCl₂ and 2 μg Trypsin Gold, Mass Spectrometry Grade (Promega, V5280) was added, and digestion proceeded overnight at 37 °C (16 h).

Elution

After digestion, the beads were centrifuged again at 1,700 × g for 3 min, and the supernatant was carefully removed without disturbing the beads. Beads were washed twice with PBS, followed by two washes with LC-grade H2O (Millipore Sigma, 1153331000). The beads were resuspended in 1 mL of 2% formic acid and incubated for 2 hours in a thermomixer (25 °C, 1250 rpm). The formic acid solution containing eluted peptides was transferred to a new 1.5 mL LoBind Eppendorf microtube (Eppendorf, 022431081) after the beads were pelleted (2,000 × g, 5 min). The beads were additionally incubated with 300 μL 1:1 acetonitrile:2% formic acid solution for 30 min, and the solution was collected in the same microtube. The eluted peptide (total 1.4 mL) was lyophilized under speed-vac and desalted using a C18 column before going on injection to the mass spectrometry.

DIA Data Collection

The UltiMate 3000 (Thermo Fisher Scientific, MA, USA) liquid chromatography system was connected to the timsTOF Pro 2, an ion-mobility spectrometry quadrupole time of flight mass spectrometer (Bruker Daltonik, Bremen, Germany). Samples were reconstituted in 0.1% formic acid (FA, ThermoFisher Scientific, 28905) and 200 ng peptide was separated by AUR3-15075C18 column (15 cm length, 75 μm i.d, 1.7 μm particle size, 120 Å pore size) with a 60 min gradient starting at 4% buffer B (80% acetonitrile- Millipore Sigma, 1000291000- with 0.1% FA) followed by a stepwise increase to 28% in 25 min, 44% in 10 min, 90% in 10min and stayed there for 7 min, then equilibrates at 4% for 8 min. The column flow rate was maintained at 400 nL/min with a column temperature of 50 °C. DIA data was acquired in the diaPASEF mode. We defined 22 × 40 Th precursor isolation windows from m/z 349 to 1229. To adapt the MS1 cycle time, we set the repetitions to variable steps (2–5) in the 13-scan diaPASEF scheme in our experiment. During PASEF MSMS scanning, the collision energy was ramped linearly as a function of the mobility from 59 eV at 1/K0 = 1.6 Vs/cm2 to 20 eV at 1/K0 = 0.6 Vs/cm2. All acquired MS data was deposited to the ProteomeXchange Consortium via PRIDE partner repository with the dataset identifier PXD076162.

DIA Database Search & Quantification

Raw DIA data were processed and analyzed using Spectronaut version 18 (Biognosys AG, Switzerland) with default parameters. Searches were performed against the Homo sapiens proteome (UniProtKB/Swiss-Prot Reviewed; UP000005640_9606, 20,654 entries). Retention time prediction was performed using dynamic iRT calibration. Spectronaut dynamically determined optimal extraction windows based on extensive mass calibration, iRT alignment, and gradient stability. False discovery rate (FDR) thresholds were set to 1% at both the precursor and protein levels. Decoy generation was configured as “mutated,” in which a random number of amino acid position swaps (minimum of 2, maximum of half the peptide length) were introduced. Local normalization was applied for quantitative analysis. Peptides passing the 1% Q-value cutoff were used to calculate protein group intensities using the MaxLFQ algorithm. For N-hex-5-ynyl-2-iodoacetamide-labeled cysteine residues, C₁₄H₂₄N₄O₂ (m/z 280.19047) was included as a variable modification. Non-imputable peptides identified under the background imputation setting were regarded as background noise and excluded from the dataset.

Data processing

Data processing code was written with AI assistance by Claude and ChatGPT. Cysteine residue numbers were appended to each peptide by matching the associated protein IDs against the reviewed and unreviewed canonical sequences in UP000005640. If the peptide matched to more than one protein ID, the cysteine location in both IDs is reported. If an ID was not present in the reference proteome, the position was reported as “ambiguous” and the peptide was matched against the entire proteome, with its location in each matched protein reported. The unmodified Spectronaut output with appended cysteine residue numbers is in Table S2.

Competitive IAA-ABPP without isotopic labeling will result in missing not at random (MNAR) peptides because when a cysteine is oxidized, it will not be labeled with IAA, so it will not be detected by MS. If enough instances of a given cysteine are sufficiently oxidized, the associated peptide will fall below the MS detection limit and be reported as missing, or “NaN” in our data.

To filter out noise in this context, peptides were kept for further analysis if all three replicates within at least one condition had detected peak areas (i.e. not “NaN”). 1,229 peptides of the original 33,525 were removed with this filter. Missed cleavage events were then accounted for if peptides shared the same Uniprot ID, cysteine position(s), and modification(s) by summing peak area intensities per sample and compressing the combined peptides into one line of data. A “ | “ between peptides in columns D and E of Table S3 indicates the peak area data are a summation of those peptides from the Spectronaut output (Table S2). This compressed the data by 6,431 rows to 25,865. Detected peak areas were then log2 transformed.

PCA and Hierarchical clustering

PCA and Hierarchical clustering codes were written with AI assistance by Claude. To conduct principal component analysis and hierarchical clustering of the replicates, missing values (“NaNs”) were imputed from log2-transformed data using the QRILC algorithm implemented in the imputeLCMD R package (version 2.1)^86,87 with default parameters, under the assumption that missing values were missing not at random due to left-censoring at the per-sample detection limit. Prior to PCA and hierarchical clustering with heatmap visualization, each peptide’s intensities were z-score scaled across the 12 samples so that high-abundance peptides did not dominate. PCA was performed using the default settings in scikit-learn (Python, version 1.8.0)⁸⁸ with samples as observations, peptides as features, and components capped at 12. Hierarchical clustering was performed using complete linkage with Euclidean distance using the default settings of the pheatmap R package (version 1.0.13)⁸⁹, with z-score scaling implemented with scale = “row”. The row dendrogram is not displayed due to the density of peptide features.

Differential abundance analysis

Differential abundance analysis was performed using the proDA R package (version 1.20.0)^90,91 in R (version 4.4.2) with AI assistance from Claude. We chose this model because proDA models missing values probabilistically without imputation by estimating sample-specific dropout curves and treating missing intensities as censored observations, and we expected missing-not-at-random values due to oxidation-based dropout of peptides. A linear model was fit to the log2-transformed, non-imputed peak area matrix (25,865 peptides × 12 samples) using condition as the experimental factor, with no oxidant specified as the reference level and default parameters applied. Variance estimates were regularized across peptides using empirical Bayes shrinkage. Differential abundance was tested for low PMSF/VO4 vs. no oxidant, high PMSF/VO4 vs. no oxidant, pervanadate vs. no oxidant, and high PMSF/VO4 vs. pervanadate, the latter implemented as a linear contrast within the fitted model. P-values were adjusted using the Benjamini–Hochberg method. Peptides were considered significantly differentially abundant at adjusted p < 0.05 and |log2 fold change| > 1. In the context of IAA-ABPP, negative log2 fold changes indicate reduced IAA labeling in the treatment condition relative to no oxidant.

To assess whether fold change estimates were dependent on proDA posterior mean log2 abundance, MA plots were generated for each pairwise comparison using a custom Python script (Matplotlib, version 3.10.8⁹²). The x-axis represents the proDA posterior mean log2 abundance across all 12 samples (avg_abundance in Table S4), and the y-axis represents the proDA log2 fold change for that comparison. A smoothed trend line was overlaid by dividing peptides into 40 equal-frequency bins along the abundance axis, calculating the median log2 fold change per bin, and fitting a cubic spline through the bin medians using SciPy⁹³ make_interp_spline (version 1.17.0, Python).

Volcano plots were generated in R using ggplot2 (version 4.0.2)⁹⁴ for each pairwise comparison. Points were colored by direction of significant change using condition-specific colors, with non-significant peptides shown in grey. Points were shape-encoded by detection status: circles indicate peptides with at least one observed numeric value in both conditions, triangles indicate peptides where one condition was entirely undetected across all three replicates, and squares indicate peptides where both conditions were entirely undetected. The five most differentially abundant depleted and five most enriched peptides meeting significance thresholds per comparison were labeled with gene name and mapped cysteine position, prioritized by a combined score of −log10(adjusted p-value) × |log2 fold change|.

Chimera X

Images of the NAGK structure³¹ were generated by loading Protein Data Bank (PDB) 2ch6 into UCSF Chimera X (v1.8)⁹⁵. The C/B homodimer was hidden, and the A/D homodimer is shown.

Purification of GST-NAGK

pGEX-2T GST NAGK constructs (see “Cloning”) were transformed into BL21(DE3) Competent E. coli (New England Biolabs, C2527H) and grown on Luria-Bertani (LB) Agar Plates with Carbenicillin (Teknova, L1010) overnight at 37°C. A single colony was then grown in 6 ml LB shaking overnight at 37°C and diluted into 250 ml LB the following day with continued culture at 37°C. When the culture reached an OD600 of 0.6–0.65, 500 μM Isopropyl β-D-1-thiogalactopyranoside (IPTG, Sigma-Aldrich, I6758) was added to induce GST-NAGK expression for 4–5 hours at 37°C, shaking. Samples were pelleted at 4,000 × g for 10 minutes, the supernatant was removed, and pellets were snap frozen and stored at −80°C for up to a week. Pellets were thawed and reconstituted in 15 ml cold PBS containing 1 mM PMSF and 2 mM DTT. Samples were sonicated in 4× 30s pulses on a Branson Sonifier 250 at 50% duty cycle with an output of 4 and placed on ice between pulses. Samples were further solubilized with the addition of 1% Triton X-100 and incubated for 30 minutes at 4°C with gentle shaking. Samples were then clarified by centrifugation for 20 minutes at 15k × g and 4°C. Glutathione agarose (500 μl resin volume per sample, ThermoFisher Scientific, 16100) was washed 3 times in cold PBS with 3 minute, 500 × g, 4°C centrifugations. The clarified lysate was transferred to a conical containing the washed glutathione agarose and incubated for one hour at room temperature with rotation. The resin was washed 3x in 12 ml PBS containing 2 mM DTT (500 × g, 3 minutes, 4°C) and moved to a microcentrifuge tube for elution. GST-NAGK was eluted in 500 μl 10 mM reduced glutathione (Thermo Fisher, 78259) in 50 mM Tris pH 8.0 for 10 minutes, rotating, at room temperature. The sample was then centrifuged at 500 × g for 2 minutes to pellet the resin. The eluted supernatant was moved to a fresh microcentrifuge tube, the elution was repeated two additional times on the resin, and all eluates were pooled. To remove the glutathione and to exchange the sample into storage buffer, samples were diluted in 15 ml 20 mM Tris pH 8.0, 200 mM NaCl and concentrated on Amicon Ultra-15 Centrifugal Filter Units (Amicon, UFC903024) by centrifugation at 4.2k × g at 4°C until the sample volume was 250 μl. The dilution and concentration step was repeated three additional times. Glycerol was added to 20% final volume, and samples were aliquoted, snap frozen, and stored at −80°C. To quantify protein, samples were run on an SDS-PAGE gel in parallel with a BSA standard curve, protein was stained using SimplyBlue Safe Stain (Invitrogen, LC6060), and protein concentration from scanned gels was quantified in Fiji (v2.9.0/1.53t)⁹⁶.

PK-LDH assay

A reaction mix containing 25 mM Hepes pH 7.5, 100 mM KCl, 10 mM MgCl₂, 1 mM DTT, 4 U/ml Pyruvate Kinase (MP Biomedicals, 0215199905), 4 U/ml Lactate Dehydrogenase (Sigma-Aldrich, L1254), 200 μM NADH (Sigma-Aldrich, N7410-15VL; or Santa Cruz Biotechnology, sc-215562), 1 mM phosphoenolpyruvate (Sigma-Aldrich, P7127), and a concentration range of 4, 2, 1, 0.5, 0.25, 0.125, 0.0625, 0.03125, 0.015625, and 0 mM GlcNAc (Chem-Impex, 01427), Adenosine 5′-triphosphate disodium salt hydrate (ATP, Sigma-Aldrich, A3377), or muramyl dipeptide (MedChem Express, HY-127090) was prepared. Assays testing the GlcNAc and muramyl dipeptide Kms used 4 mM ATP; those testing the ATP Km used 4 mM GlcNAc. The assay was initiated with 5 nM of the indicated GST-NAGK for a final assay volume of 100 μl, the reaction was mixed, the plate was briefly spun down, and the absorbance (Abs.) at 340 nM was recorded on an Envision 2104 plate reader (PerkinElmer) every minute for one hour at room temperature to measure the decrease in [NADH] as it was oxidized to NAD⁺. A standard curve of 200, 150, 100, 75, 50, 25, 12.5, and 0 μM NADH was run with each assay, the background absorbance at 0 μM NADH was subtracted from each value, and the slope was calculated (Abs./[NADH]). For experimental conditions, the slope of the initial velocity (linear portion) of each resulting curve (Abs. @ 340 nM over time per substrate concentration for each NAGK wild type or mutant: -Abs./min.5 nM E; E = enzyme) was calculated using GraphPad Prism software, and the slope of the background NADH consumption rate with 0 μM of the variable substrate was subtracted for each condition. The reaction rate (μmol ATP consumed/min.mg E) for each NAGK variant and [substrate] was calculated using the slope of the standard curve (Abs./[NADH]), the background corrected slope of the initial velocity, the concentration of NAGK in mg/nmol, and the reaction volume:

$$\left(-\text{Abs}./\text{min}.5\text{nM E}\right)*\left(\left[\text{NADH}\right]/\text{Abs}.\right)*\left(0.0001\text{L}\right)*\left(1/0.0001\text{L}\right)*\left(\text{nmol}/0.3745571\text{mg}\text{E}\right)=\mu \text{mol}/\text{min}.\text{mg E}$$

The resulting reaction rates for each [substrate] and NAGK variant from four replicate assays were plotted against [substrate] to generate a Michaelis Menten plot from which the Km and Vmax were derived using GraphPad Prism software.

¹³C₂ GlcNAc tracing

Sample preparation

Parental or NAGK KO/rescue PANC-1 cells (see “Generation of PANC-1 NAGK knockout/rescue lines”) were plated into normal growth medium (see “Cell culture”) in 10 cm dishes at a seeding density of 1.4e6 cells for growth in 3.6 mM glutamine media or 2.2e6 cells for growth in 0.05 mM glutamine media. Five plates per growth condition and GlcNAc isotopologue (¹²C or ¹³C₂) were seeded. The following day, plates were washed 2x in PBS and fed with 7 ml phenol red free DMEM without glucose or glutamine (Gibco, A1443001) supplemented with 10% dialyzed FBS (Sigma, F0392), 22.523 mM glucose (Sigma, G7021), 1% PS, and either 3.6 or 0.05 mM glutamine (Sigma, G8540). For cells with inducible expression of NAGK, cells were treated with the indicated concentration of doxycycline (Figure S6E) at the time of the media change. Approximately 21 hours after the media change and three hours prior to extraction, 10 mM ¹²C GlcNAc (Chem-Impex, 01427) or ¹³C₂ GlcNAc (N-[1,2-¹³C₂]acetyl-D-glucosamine, Omicron Biochemicals, Inc, Glc-006) was added to each plate from 1 M stocks, with 5 plates per condition per isotopologue. Three hours later, cells were washed on ice 2x with 5 ml cold 0.9% w/v NaCl in H2O (Sigma, 34877). 1 ml −80°C 80% methanol (Sigma, 34860) in H2O (Sigma, 34877) was added to each plate to extract metabolites, and plates were stored at −80°C until all plates were processed. Samples were scraped to 2 ml microcentrifuge tubes (Thermo Fisher, 88379) on ice; the larger microcentrifuge tube size was used to provide adequate head space for later sonication. Samples from the NAGK knockout versus WT versus C217S tracing assay were stored at −80°C and batch processed on subsequent days. Samples were pulse sonicated for 1× 20 seconds on a Branson Sonifier 450 at 50% duty cycle with an output of ~2.2. Samples were incubated on ice for 10 minutes and spun at 21k × g for 10 mins at 4°C. The supernatant containing the metabolites was moved to a new tube and stored at −80°C. The remaining protein pellet was stored at −80°C until it was lysed in a 2% SDS, 50 mM Tris pH 6.8 lysis buffer with heating to 95°C and quantified using the DC Assay. Within 1–2 weeks, metabolite samples were dried under nitrogen (Organomation Associates Inc., OA-SYS N-EVAP 111) or in a SpeedVac (Eppendorf, Vacufuge plus).

Liquid chromatography-high resolution MS

Metabolite samples were analyzed on a Vanquish Duo UHPLC (Thermo Fisher Scientific) coupled to a Thermo QExactive Plus Orbitrap equipped with a HESI II probe. 2 μL of sample resuspended in 50μL of LC-MS grade 95/5 water/methanol were injected onto a ZIC-pHILIC 150 × 2.1 mm 5 μm particle size column (EMD Millipore) with a ZIC-pHILIC 20 × 2.1 guard column at a temperature of 25°C. Mobile phases were acetonitrile (solvent A) and 20 mM ammonium carbonate with 0.1% (v/v) ammonium hydroxide in water (solvent B). Initial conditions consisted of a mobile phase composition of 20% B and a flow rate of 0.150 mL/min. A linear gradient program of 20% B to 80% B from 0.5–20 min was performed, followed by a return to initial conditions from 20–20.5 min and an isocratic hold at 20% B from 20.5–28 min. The Orbitrap was operated in polarity switching mode with 70–1000 m/z full scans utilizing an insource fragmentation energy of 1. For each analyte, isotopologue, and internal standard, the peak corresponding to the [M+H]+ or other predominant ion at 5 ppm was integrated in Tracefinder 4.1 (Thermo Scientific).

¹³C₂ data analysis

For each isotope of interest per metabolite per sample, the area under the curve was normalized to protein concentration. The normalized area under the curve was averaged for the five technical replicates within each condition, and the averaged values for each biological experiment were graphed.

ATP Detection Assay

On day 1, PANC-1 cells were seeded into normal growth medium (see “Cell culture”) in 96 well plates (BD Falcon, 353072). Between 2000–8000 cells were seeded per well to yield approximately equal confluency across conditions at the assay end point; duplicate wells were set up for each growth condition (24 and 48 hours growth in 3.6 or 0.05 mM glutamine media). On day 2, media was removed from the 48-hour timepoint wells and replaced with 100 μl 3.6 or 0.05 mM glutamine media (see “Experimental growth conditions”). On day 3, media was removed from the 24 and 48-hour timepoint wells and replaced as above. On day 4, the confluence of each well was measured on an Incucyte S3 (Sartorius), and the ATP concentration was then measured using the ATP Detection Assay Kit – Luminescence (Cayman, 700410) per the manufacturer’s protocol. Luminescence was measured on an Envision 2104 plate reader (PerkinElmer). The resulting pmol ATP per condition was normalized to % confluence.

Cysteine conservation

The conservation of NAGK C211 and C217 was determined by querying the human NAGK sequence (Uniprot ID Q9UJ70) by BLAST against the Uniprot database. We selected the top 250 hits and aligned them using Uniprot’s Align tool (2022 version). The amino acids aligning to human C211 and C217 were quantified with a custom R script, and the results were graphed using GraphPad Prism software.

BCECF, AM assay

PANC-1 (1.3e4 cells/well for full nutrient media, 1.6e4 for 0.2 mM glucose, 1.8e4 for 0.05 mM glutamine media) and DLD-1 (8e3 for full nutrient media, 1.1e4 for 0.2 mM glucose, and 1.4e4 for 0.05 mM glutamine media) were seeded into normal growth medium (see “Cell culture”) in 96 well clear bottom, black-sided plates (Corning, 3603). Twelve wells per media condition were seeded. The following day, cells were washed once with PBS and fed with media containing full nutrients, 0.2 mM glucose, or 0.05 mM glutamine (see “Experimental growth conditions”). The next day, cells were washed 2x in HBSS (SAFC, 55037C) then incubated in 100 μl 2 μM BCECF, AM reagent (ThermoFisher Scientific, B1150) in HBSS for 10 minutes in the dark at room temperature. Cells were washed 2x with live cell imaging solution (LCIS, Invitrogen, A14291DJ) pre-warmed to 37°C, and then the pH of each well was read in a third wash of the same on an Envision 2104 plate reader (PerkinElmer) using a D505 Dichroic mirror (PerkinElmer, 2100–4030), a 492/8 nM excitation filter (PerkinElmer, 2100–5220) for pH dependent reads, a 440/8 nM excitation filter (PerkinElmer, 2100–5660) for isobestic reads, and a 535/25 nM emission filter (PerkinElmer, 2100–5120) for both reads. The LCIS was removed, and three wells per condition were given 100 ul 100 mM KH₂PO₄ (Sigma, P5655) buffers adjusted to pH 6.2, 6.8, 7.4, and 8.0 with NaOH (12 wells total) containing 10 μM nigericin (Invitrogen, tlrl-nig) and 10 μM valinomycin (Cayman Chemical Company, 10009152). This served to equilibrate intracellular and extracellular pH for use as a standard curve. The plate was incubated at 37°C for 5–6 minutes then read on the Envision 2104 using the above protocol. For PANC-1 cells, the pH was calculated using the BCECF, AM formula pH = pKa + log[((R-R_A)/(R_B-R))*(F_A(λ2)/F_B(λ2))], where pKa = 6.97, λ1 = 490 (pH dependent), λ2 = 440 (isobestic), F = fluorescent intensity, R = F_(λ1)/F_(λ2) intensity, R_A = ratio at the most acidic point of calibration, R_B = ratio at the most basic point of calibration, F_A = fluorescent intensity at the most acidic point of calibration, F_B = fluorescent intensity at the most basic point of calibration. The pH of DLD-1 cells was too close to F_B(λ1) (pH 8.0), so the formula did not work accurately, and the pH was instead calculating by plotting a standard curve using all pH calibration points (nearly identical pH’s were calculated using the formula and standard curve method in PANC-1 cells).

Quantification of total and oxidized glutathione

PANC1 (6,000 cells/well for full nutrient media ± 1 mM BSO, 10 mM GSEE, 1 μM KI696 or 1:1000 DMSO control; 7,000/well for 0.2 mM glucose; 8,500/well for 0.05 mM glutamine ± 10 mM GSEE, 1 μM KI696 or 1:1000 DMSO) and DLD-1 (5,000 cell/well for full media ± 1 mM BSO, 6,500/well for 0.2 mM glucose; 7,500/well for 0.05 mM glutamine) were plated into normal growth media (see “Cell culture”) in 96 well plates in duplicate per growth condition. One day after plating, media was removed, and cells were washed once with PBS and fed with the indicated media (full nutrient, 0.2 mM glucose, or 0.05 mM glutamine, see “Experimental growth conditions”) ± compounds as noted above. GSEE and BSO were reconstituted directly into media immediately prior to use. KI696 was prepared as a 1 mM stock in DMSO and stored at −20°C prior to addition to the media. Following 24 hours of growth in the experimental media, plates were imaged on an Incucyte S3 (Sartorius) to measure confluence. Total or oxidized glutathione was then measured using the GSH/GSSG-Glo Assay (Promega, V6611) per the manufacturer’s instructions. The average luminescence from duplicate wells was calculated and background corrected. Average luminescence values that were negative following background correction were set to zero. Concentrations of total and oxidized glutathione were calculated using interpolation of values against a nonlinear standard curve in GraphPad Prism software using the “Hyperbola (X is concentration)” function. Values were normalized to the average percent confluent for duplicate wells calculated on the Incucyte, and the concentration of reduced glutathione normalized to % confluence was calculated per the manufacturer’s instructions [total glutathione normalized to percent confluence – (2x GSSG normalized to percent confluence)]. Normalized values were graphed in GraphPad Prism software, and significance was determined as indicated in each figure legend, dependent on experimental setup.

Code availability

The code used to quantify cysteine conservation (Figure S6C) and all code related to the IAA-ABPP data (Figures 5, S8, and S9) has been deposited at Zenodo at https://doi.org/10.5281/zenodo.21287567 and is publicly available.

QUANTIFICATION AND STATISTICAL ANALYSIS

For Figures 1–4 and associated Supplemental Figures, GraphPad Prism software (Version 10.6 or lower) was used to perform statistical analyses. Statistical details for each assay can be found in the associated figure legend. Unless otherwise indicated, data are presented as mean ± standard deviation (s.d.) and n represents the number of independent replicates. Significance was defined as p < 0.05. No data were excluded from the analyses. We did not use methods to predetermine sample size or to determine whether the data met assumptions of the statistical approach, as most had small sample sizes of n=3–4. Samples were not blinded.

For statistical analysis of the proteomics data in Figure 5, we used Python and R packages. The R packages used were the QRILC function from the imputeLCMD R package (version 2.1) with default settings, pheatmap with default settings (version 1.0.13), proDA (version 1.20.0) with default setting. The Python packages used were scikit-learn (version 1.8.0), and SciPy make_interp_spline (version 1.17.0). For proDA analysis of differential abundance, peptides were considered significantly differentially abundant at adjusted p < 0.05 and |log2 fold change| > 1. See STAR Methods subsection “Data processing” in “IAA-ABPP Proteomics” for data filtering prior to statistical analysis. Statistical details for each analysis can be found in STAR Methods under the subsections “PCA and Hierarchical clustering” and “Differential abundance analysis” in “IAA-ABPP Proteomics”.

Supplementary Material

1

Data S1: Source data for mass spectrometry of NAGK-HA immunoprecipitated from PANC-1 cells to look for potential differential phosphorylation following glutamine deprivation; related to Figures 1 and S1, and data underlying graphs in Figures 1–3, S4, and S5

2

Data S2: Source data underlying graphs in Figure 4 and unprocessed GlcNAc tracing data for Figures 4E–G and S6G–I

3

Data S3: Source data underlying graphs in Figure S6 and unprocessed GlcNAc tracing data for Figures S6J–O

4

Data S4: Uncropped western blots and associated replicates from Main Figures 1–3 and 5.

5

Data S5: Uncropped western blots and associated replicates from Supplemental Figures 1–7.

6

Document S1: Figures S1-S9, Table S1 (related to STAR Methods), and Protocol S1

7

Table S2: Spectronaut output of IAA-ABPP data with cysteine residue appended; related to Figures 5, S8, and S9.

8

Table S3: Filtered and processed IAA-ABPP data; related to Figures 5, S8, and S9.

9

Table S4: proDA differential abundance analysis of IAA-ABPP data; related to Figures 5 and S8.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
rabbit polyclonal anti-NAGK	Proteintech	Cat#15051-1-AP; RRID:AB_2152368
rabbit monoclonal anti-Akt (pan) (C67E7)	Cell Signaling Technology	Cat#4691; RRID:AB_915783
rabbit monoclonal anti-Phospho-Akt (Ser473) (D9E)	Cell Signaling Technology	Cat#4060; RRID:AB_2315049
mouse monoclonal anti-Vinculin	Sigma-Aldrich	Cat#V9264; RRID:AB_10603627
rabbit monoclonal anti-HA-Tag (C29F4)	Cell Signaling Technology	Cat#3724; RRID:AB_1549585
rabbit polyclonal anti-PTPN1/PTP1B	Proteintech	Cat#11334-1-AP; RRID:AB_10642566
rabbit polyclonal anti-Ran	Proteintech	Cat#10469-1-AP; RRID:AB_2176484
rabbit monoclonal anti-Nrf2 [EP1808Y]	Abcam	Cat#ab62352; RRID:AB_944418
rabbit polyclonal anti-p38 MAPK	Cell Signaling Technology	Cat#9212; RRID:AB_330713
rabbit polyclonal anti-phospho-p38 MAPK (Thr180/Tyr182)	Cell Signaling Technology	Cat#9211; RRID:AB_331641
rabbit monoclonal anti-BiP (C50B12)	Cell Signaling Technology	Cat#3177; RRID:AB_2119845
rabbit polyclonal anti-SOD1	AB Clonal	Cat#A0274; RRID:AB_2757087
mouse monoclonal anti-GAPDH-71.1	Sigma-Aldrich	Cat#G8795; RRID:AB_1078991
mouse monoclonal anti-Oxidized PTP Active Site	R&D Systems	Cat#MAB2844; RRID:AB_664170
rabbit polyclonal anti-Ras	Cell Signaling Technology	Cat#3965; RRID:AB_2180216
rabbit monoclonal anti-Src (32G6)	Cell Signaling Technology	Cat#2123; RRID:AB_2106047
rabbit polyclonal anti-KIF3A	Proteintech	Cat#13930-1-AP; RRID:AB_2132063
mouse monoclonal anti-HoxB6	Santa Cruz Biotechnology	Cat#sc-166950; RRID:AB_10610644
rabbit polyclonal anti-PCCB	Proteintech	Cat#11139-1-AP; RRID:AB_2877746
rabbit polyclonal anti-CIAPIN1	Proteintech	Cat#12638-1-AP; RRID:AB_2079665
rabbit polyclonal anti-SHB	Proteintech	Cat#31967-1-AP; RRID:AB_3742501
rabbit polyclonal anti-BLVRB	Proteintech	Cat#17727-1-AP; RRID:AB_2065087
rabbit polyclonal anti-ARHGEF7	Proteintech	Cat#14092-1-AP; RRID:AB_2227567
rabbit monoclonal anti-S6 Ribosomal Protein (5G10)	Cell Signaling Technology	Cat#2217; RRID:AB_331355
rabbit polyclonal anti-AP1M1	Proteintech	Cat#12112-1-AP; RRID:AB_2058339
rabbit polyclonal anti-ALDOA	Proteintech	Cat#11217-1-AP; RRID:AB_2224626
rabbit polyclonal anti-CLIP-115/CLIP2	Proteintech	Cat#25088-1-AP; RRID:AB_2879890
IRDye® 800CW Donkey anti-Mouse IgG Secondary Antibody	LICORbio	Cat#926-32212; RRID:AB_621847
IRDye® 800CW Donkey anti-Rabbit IgG Secondary Antibody	LICORbio	Cat#926-32213; RRID:AB_621848
IRDye® 680RD Donkey Anti-Mouse IgG Secondary Antibody	LICORbio	Cat#926-68072; RRID:AB_10953628
IRDye® 680RD Donkey anti-Rabbit IgG Secondary Antibody	LICORbio	Cat#926-68073; RRID:AB_10954442
Monoclonal Anti-HA−Agarose antibody produced in mouse	Sigma-Aldrich	Cat#A2095; RRID:AB_257974
Bacterial and virus strains
BL21(DE3) Competent E. coli for protein purification	New England Biolabs	Cat#C2527H
Biological samples
Chemicals, peptides, and recombinant proteins
IRDye® 800CW Streptavidin	LICORbio	926-32230
Glutathione reduced ethyl ester	Millipore Sigma	Cat#G1404
Glutathione ethyl ester	Cayman Chemical	Cat#14953
KI696	Cayman Chemical	Cat#42039
L-Buthionine-(S,R)-Sulfoximine	Cayman Chemical	Cat#14484
TRI(2-METHOXYCARBONYLETHYL)PHOSPHINE HCL	AstaTech	Cat#66756
X-tremeGENE HP DNA Transfection Reagent	Roche	Cat#6366546001
FuGENE HD Transfection Reagent	Promega	Cat#E2311
Sodium Vanadate (Na3VO4)	Fisher Scientific	Cat#S454-50
Hydrogen peroxide 30% (H2O2)	Sigma-Aldrich	Cat#H1009
Catalase from bovine liver	Sigma-Aldrich	Cat#C1345
Phenylmethanesulfonyl fluoride (PMSF)	Sigma-Aldrich	Cat#P7626
Sodium fluoride	Sigma-Aldrich	Cat#S1504
aprotinin	Sigma-Aldrich	Cat#A1153
leupeptin hemisulfate	Research Products International	Cat#L22035
pepstatin A	Research Products International	Cat#P30100
IGEPAL CA-630	Sigma-Aldrich	Cat#I3021
4-(2-Aminoethyl)benzenesulfonyl fluoride hydrochloride (AEBSF) ready made solution	Sigma-Aldrich	Cat# SBR00015-1ML
Tris(2-carboxyethyl)phosphine hydrochloride (TCEP)	Millipore Sigma	Cat#580561
reduced glutathione	ThermoFisher Scientific	Cat#78259
sodium molybdate dihydrate	Sigma-Aldrich	Cat#M1003
sodium tungstate dihydrate	Sigma-Aldrich	Cat#223336
Dithiothreitol, molecular grade	Promega	Cat#V3155
Adenosine 5-triphosphate magnesium salt	Sigma-Aldrich	Cat# A9187
N-acetyl-D-glucosamine (GlcNAc)	Chem-Impex	Cat#01427
Intercept (TBS) Blocking Buffer	LICORbio	Cat#927-60050
Hemagglutinin (HA) Peptide	Sigma-Aldrich	Cat#I2149
Recombinant Human N-Acetyl-D-Glucosamine Kinase/NAGK His Protein	Novus Biologicals	Cat#NBP1-99050
SimplyBlue SafeStain	Invitrogen	Cat#465034
N-ethylmaleimide (NEM)	Pierce	Cat#23030
N-Hex-5-ynyl-2-iodo-acetamide (IAA)	Oakwood Chemicals	Cat#95226
Desthiobiotin Azide	Click Chemistry Tools (now Vector Laboratories)	Cat#CCT-1107-5
tris-hydroxypropyltriazolylmethylamine (THPTA)	Click Chemistry Tools (now Vector Laboratories)	Cat# CCT-1010
Copper(II) sulfate (CuSO4)	Sigma-Aldrich	Cat#PHR1477
Copper(II) sulfate (CuSO4)	Sigma-Aldrich	Cat#C1297-100
(+)-Sodium L-ascorbate	Sigma-Aldrich	Cat#A4034
(+)-Sodium L-ascorbate	Sigma-Aldrich	Cat#A7631
Streptavidin−Agarose from Streptomyces avidinii	Sigma-Aldrich	Cat#S1638
Biotin-maleimide	Sigma-Aldrich	Cat#B1267
EZ-Link Iodoacetyl-PEG2-Biotin	ThermoFisher Scientific	Cat#21334
dialkoxydiphenylsilane (DADPS) biotin azide	Vector Laboratories	Cat#CCT-1330
Streptavidin Sepharose High Performance affinity resin	Cytiva	Cat#17511301
iodoacetamide	Sigma-Aldrich	Cat#I6125
Trypsin Gold, Mass Spectrometry Grade	Promega	Cat#V5280
Isopropyl β-D-1-thiogalactopyranoside (IPTG)	Sigma-Aldrich	Cat#I6758
Glutathione agarose	ThermoFisher Scientific	Cat#16100
Pyruvate kinase	MP Biomedicals	Cat#0215199905
Lactate dehydrogenase	Sigma-Aldrich	Cat#L1254
β-Nicotinamide adenine dinucleotide (NADH), reduced disodium salt	Sigma-Aldrich	Cat# N7410-15VL
β-Nicotinamide adenine dinucleotide (NADH), reduced disodium salt	Santa Cruz Biotechnology	Cat#sc-215562
Phospho(enol)pyruvic acid monopotassium salt	Sigma-Aldrich	Cat#P7127
Adenosine 5-triphosphate (ATP) disodium salt hydrate	Sigma-Aldrich	Cat#A3377
muramyl dipeptide	MedChem Express	Cat#HY-127090
N-[1,2-13C2]acetyl-D-glucosamine	Omicron Biochemicals, Inc	Cat#Glc-006
nigericin	Invitrogen	Cat#tlrl-nig
valinomycin	Cayman Chemical Company	Cat#10009152
Critical commercial assays
BCECF, AM (2’,7’-Bis-(2-Carboxyethyl)-5-(and-6)-Carboxyfluorescein, Acetoxymethyl Ester)	ThermoFisher Scientific	Cat#B1150
GSH/GSSG-Glo Assay	Promega	Cat# V6611
ATP Detection Assay Kit - Luminescence	Cayman Chemical	Cat#700410
Q5 Site-Directed Mutagenesis Kit	New England Biolabs	Cat#E0552
DC Protein Assay	Bio-Rad	Cat#5000112
Lambda Protein Phosphatase	New England Biolabs	Cat#P0753S
Deposited data
IAA-ABPP proteomics data (Figure 5)	This paper; ProteomeXchange Consortium	PXD076162
Mass spectrometry data identifying sulfonylation of NAGK C217 (Figure 2F and 2G)	This paper; ProteomeXchange Consortium	PXD081578
Mass spectrometry data assessing potential differential phosphorylation of NAGK in PANC-1 cells grown in 3.6 vs 0.05 mM Q glutamine (Data S1, tabs 1 and 2)	This paper; MassIVE	MSV000102508l; ftp://massive-ftp.ucsd.edu/v13/MSV000102508/
Experimental models: Cell lines
Human: PANC-1 cells	ATCC	Cat#CRL-1469
Human: 293T cells	ATCC	Cat#CRL-3216
Human: HPAC cells	ATCC	Cat#CRL-2119
Human: DLD-1 cells	ATCC	Cat#CCL-221
Human: HCC1806 cells	ATCC	Cat#CRL-2335
Human: MDA-MB-231 cells	ATCC	Cat#CRM-HTB-26
Human: A549 cells	ATCC	Cat#CRM-CCL-185
Mouse: mT8 cells	Laboratory of Rohit Chandwani; original source: Boj et al.37	N/A
Mouse: mT8 cells with: clonal knockout of a control locus (sgCtrl sequence: Zafra et al.79) and expression of PGK empty vector (EV); clonal knockout of NAGK (sgNAGK#4 sequence: Doench et al.80) and expression of PGK-EV or PGK-NAGK-HA (human)	This paper, derived from above mT8 cells	N/A
Human: PANC-1 cells with: pLentiCRISPR nontargeting control (sgNT2 sequence: Doench et al.80) and expression of pInducer20 EV; pLentiCRISPR knockout of NAGK (sgNAGK sequence: Kim et al.49) and expression of pInducer20 EV, human NAGK-HA WT, or human NAGK-HA C217S	This paper, derived from above PANC-1 cells	N/A
Experimental models: Organisms/strains
Oligonucleotides
sgRNA targeting sequence: mouse control (sgCtrl/sgControl), GACATTTCTTTCCCCACTGG	Zafra et al.79	N/A
sgRNA targeting sequence: mouse NAGK (sgNAGK#4), GACAGGGCCAAACAGAAGGC	Doench et al.80	N/A
sgRNA targeting sequence: human nontargeting control (sgNT2), GCCCCGCCGCCCTCCCCTCC	Doench et al.80	N/A
sgRNA targeting sequence: human NAGK (sgNAGK), gGCCTAGGGCCTATCTCTGAG	Kim et al.49	N/A
Genomic PCR amplification and sequencing primers for mouse sgCtrl/sgNAGK#4 clones: see STAR Methods, Generation of cell lines, NAGK knockout/rescue mT8 cells	This paper	N/A
Cloning primers: See Table S1	This paper; above references for sgRNAs	N/A
Recombinant DNA
NAGK (Homo sapiens) in pDONR221	Harvard PlasmidID	Cat#HsCD00044726
pInducer20	Meerbrey et al.75	Addgene Plasmid #44012
pLenti PGK Neo DEST (w531-1)	Campeau et al.76	Addgene Plasmid #19067
pENTR1A no ccDB (w48-1)	Campeau et al.76	Addgene Plasmid #17398
pGEX-2T	Laboratory stock; original backbone from GE Healthcare	pGEX-2T, 4948 bp, GenBank U13850.1
pInducer20-GFP	Published plasmid from our laboratory: Schild et al.77	N/A
pSpCas9(BB)-2A-Puro (PX459) V2.0	Ran et al.78	Addgene Plasmid #62988
lentiCRISPR v2-Blast	A gift from Mohan Babu	Addgene Plasmid #83480
psPAX2	a gift from Didier Trono	Addgene Plasmid #12260
pMDG.2	a gift from Didier Trono	Addgene Plasmid #12259
pInducer20 expressing empty vector (EV) or human NAGK-HA wild type, S70A, S74A, S76A, T106A, Y205A, S271A, C211A, C217A, C211A/C217A, C217S, R218Q, R218A, or R218D	This paper	N/A
pLenti PGK Neo DEST (w531-1) expressing empty vector (EV) or human NAGK-HA wild type	This paper	N/A
pSpCas9(BB)-2A-Puro (PX459) V2.0 expressing sgRNAs against a murine control locus (sgCtrl sequence: Zafra et al79) or murine NAGK (sgNAGK#4 sequence: Doench et al.80)	This paper	N/A
lentiCRISPR v2-Blast expressing a nontargeting sgRNA (sgNT2 sequence: Doench et al.80) or a sgRNA against human NAGK (sgNAGK sequence: Kim et al.49)	This paper	N/A
pGEX-2T expressing GST-NAGK (human) wild type, C211A, C217A, C211A/C217A, C211S, C217S, or C211S/C217S	This paper	N/A
Software and algorithms
imputeLCMD, v2.1 (R package)	Lazar et al.86, Lazar et al.87	https://CRAN.R-project.org/package=imputeLCMD
scikit-learn, v1.8.0 (Python package)	Pedregosa et al.88	https://scikit-learn.org/stable/index.html
pheatmap, v1.0.13 (R package)	Kolde, R.89	https://CRAN.R-project.org/package=pheatmap
proDA, v1.20.0 (R package)	Ahlmann-Eltze et al.90, Ahlmann-Eltze et al.91	https://www.bioconductor.org/packages/release/bioc/html/proDA.html
Matplotlib, v3.10. (Python package)	Hunter, J.D.92	https://matplotlib.org/stable/
SciPy, v1.17.0 (Python package)	Virtanen, P. et al.93	https://scipy.org/
ggplot2, v4.0.2 (R package)	Wickham, H.94	https://ggplot2.tidyverse.org/
GraphPad Prism, v10.6 or lower	Dotmatics	https://www.graphpad.com/
ChimeraX, v1.8	Meng et al.95	https://www.cgl.ucsf.edu/chimerax/
Spectronaut, v18	Biognosys AG	https://biognosys.com/software/spectronaut/
Claude, Sonnet 4.6	Anthropic	https://claude.ai
ChatGPT, v5.3	OpenAI	https://chatgpt.com/
Tracefinder 4.1	ThermoFisher Scientific	https://www.thermofisher.com/us/en/home/industrial/mass-spectrometry/liquid-chromatography-mass-spectrometry-lc-ms/lc-ms-software/lc-ms-data-acquisition-software/tracefinder-software.html
MaxQuant, v1.6.17.0	Max Planck Institute	https://maxquant.org/
Fiji, v2.9.0/1.53t	Schindelin, J.et al.96	https://fiji.sc/
Code generated for Figure S6C and IAA-ABPP data (Figures 5, S8, S9)	This paper	https://doi.org/10.5281/zenodo.21287567
Other
Amicon Ultra Centrifugal Filter, 3 kDa MWCO	Amicon	UFC200324
Amicon Ultra Centrifugal Filter, 30 kDa MWCO	Amicon	UFC903024

Significance.

Reactive cysteines regulate protein function and are increasingly targeted in chemoproteomics and covalent drug discovery. Because different electrophiles engage distinct cysteine populations, expanding the range of available cysteine-engagement chemistries can reveal new functional sites while improving understanding of electrophile selectivity. Here, we show that the commonly used lysis buffer components PMSF and orthovanadate unexpectedly generate a lysis-derived oxidant that engages cysteines during sample preparation. This oxidant identified a functionally important cysteine in N-acetyl-D-glucosamine kinase (NAGK) and produced a cysteine-engagement profile that only partially overlapped with that of pervanadate. Our findings reveal an unrecognized source of chemical reactivity during protein extraction that can both influence chemoproteomic measurements and be repurposed as a complementary strategy for identifying and characterizing functional cysteines.

Highlights.

• PMSF and orthovanadate oxidize NAGK C217, causing a mobility shift on SDS-PAGE

• A C217S mutant of NAGK has reduced ATP affinity and enzymatic activity

• The PMSF/orthovanadate oxidant engages a distinct subset of cysteines during cell lysis

• PMSF/orthovanadate and pervanadate define partially distinct cysteine-engagement profiles