Functional interplay among thiol-based redox signaling, metabolism, and ferroptosis unveiled by a genetic variant of TP53

Significance

Thiol-mediated protein cysteine modifications represent important, but often overlooked, physiologic regulatory determinants. We previously reported that the codon 47 variants of TP53 associate with differences in the intracellular abundance of low molecular weight (LMW) thiol metabolites, directly promoting enhanced resistance to the oxidative death process of ferroptosis in S47 cells. We now report that these polymorphic variants also associate with distinct metabolic profiles, including the redox-mediated, reversible reprogramming of glycolytic metabolism. The observed metabolic differences are mechanistically linked to changes in the activities of key physiologic regulatory proteins, including ATF4, GAPDH, and G6PD. The results support LMW thiol abundance as a critical factor in the regulation of adaptive, coordinated, metabolic signaling networks and oxidative stress response pathways.

Abstract

The p53 tumor suppressor protein is a transcription factor and master stress response mediator, and it is subject to reduction-oxidation (redox)-dependent regulation. The P47S variant of TP53, which exists primarily in African-descent populations, associates with an elevated abundance of low molecular weight (LMW) thiols, including glutathione (GSH) and coenzyme A (CoA). Here we show that S47 and P47 cells exhibit distinct metabolic profiles, controlled by their different redox states and expression of Activating Transcription Factor-4 (ATF4). We find that S47 cells exhibit decreased catabolic glycolysis but increased use of the pentose phosphate pathway (PPP), and an enhanced abundance of the antioxidant, NADPH. We identify ATF4 as differentially expressed in P47 and S47 cells and show that ATF4 can reverse the redox status and rescue metabolism of S47 cells, as well as increase sensitivity to ferroptosis. This adaptive metabolic switch is rapid, reversible, and accompanied by thiol-mediated changes in the structures and activities of key glycolytic signaling pathway proteins, including GAPDH and G6PD. The results presented here unveil the important functional interplay among pathways regulating thiol-redox status, metabolic adaptation, and cellular responses to oxidative stress.

Reduction-oxidation (redox) reactions are dynamic events critical to most cellular processes, including bioenergetics, metabolism, and immunity (1, 2). Altered redox regulation contributes to human disease. One major component of redox signaling centers on the rapid, reversible modification of critical protein cysteine residues in response to changing environmental conditions (3–6). The amino acid cysteine plays a central role in protein structure and function, contributing to protein stability and the formation of intra- and intermolecular disulfide bonds. Its highly reactive thiol or sulfhydryl side chain (-SH) switches between oxidized and reduced states in response to redox changes and is susceptible to a variety of posttranslational modifications (PTMs). Among these is S-thiolation, a reversible process by which the -SH groups of protein cysteines form mixed disulfides with low molecular weight (LMW) thiol compounds, such as glutathione (GSH, S-glutathionylation, RS-SG), coenzyme A (CoA, S-CoAlation, RS-SCoA), or other cysteines (Cys, S-cysteinylation, RS-SCys). Protein S-thiolation occurs as part of an adaptive response to oxidative and metabolic stress, but also contributes to normal cellular metabolism. This PTM protects proteins from damaging sulfhydryl overoxidation (RS-O⁻, RS-O₂⁻), aggregation, and subsequent proteolysis. By rapidly and reversibly altering protein structure and function, S-thiolation/dethiolation can serve as a molecular switch to help regulate and integrate redox-dependent signaling networks during normal physiology and in pathophysiologic settings (1–6).

l-cysteine is a rate-limiting precursor for the synthesis of both GSH and CoA. Like cysteine itself, GSH is a major cellular antioxidant, and S-glutathionylation is the most frequently investigated form of S-thiolation in mammalian cells (3–5). Although less well studied, recent studies indicate that CoA also functions in redox signaling and protein cysteine modification, and emerging investigations on eukaryotic and prokaryotic cells have putatively identified a diverse set of proteins that are subject to S-CoAlation (7–11). However, the role of S-thiolation in antioxidant defense, protein modification, and redox signaling remains poorly understood. Moreover, for the vast majority of proteins, the physiologic settings and the functional consequences of differential S-CoAlation and S-glutathionylation have yet to be established.

As a sequence-specific DNA-binding protein and transcription factor, p53 is referred to as a master stress-response mediator. Although best known for its role in tumorigenicity, p53 also functions in many cellular processes, including metabolism and redox regulation (12–14). The human TP53 gene is distinguished by the presence of functionally impactful single-nucleotide polymorphisms (SNPs), including a SNP at codon 47 (P47S, rs1800371, G/A) (15, 16). The P47S variant encodes a serine (S47) instead of proline (P47) in the p53 protein. Interestingly, the S47 variant exists predominantly in individuals of African descent, at an allele frequency of about 1.2% in African Americans, and about 5% in Africans (15–20). We recently discovered that the S47 variant up-regulates the cellular abundance of CoA/GSH and that these metabolites alter the activity of p53 (21). p53 function is subject to redox regulation, and its activities regulate the expression of a number of genes that impact cellular redox status (12–14, 21–23). We find that the different activities of the P47 and S47 variants correlate with elevated levels of LMW thiols in S47 cultured cells and in mouse tissues, which feedback on the p53 protein to alter its structure and function (21). Importantly, the phenotypes of the P47 and S47 cells and livers can be reversed, by altering the cellular redox environment (21).

To define how the S47 variant might impact p53 function and disease risk, we previously generated a humanized knockin mouse model of the P47S polymorphism (human TP53 knockin, or Hupki); we also analyzed human lymphoblastoid cell lines generated from homozygous S47 and P47 individuals, as well as primary and immortalized cells derived from the S47 and P47 Hupki mice (18, 19, 21). While the activities of the P47 and S47 isoforms are comparable in many respects, we find that human and mouse cells with the S47 variant show enhanced resistance to ferroptosis (19, 21, 24). Ferroptosis is a nonapoptotic programmed cell death process characterized by the accumulation of lethal lipid peroxides (25–28). This oxidative process is implicated in several human pathologies and contributes to p53’s role in tumor suppression and normal cell stress responses (19, 21, 29, 30). Cells with the S47 variant exhibit altered regulation of a subset of genes that modulate ferroptosis sensitivity (19, 21, 24). The S47 phenotype manifests as: greater resistance to ferroptotic cell death in cultured mouse and human cells; increased liver damage in a mouse model of ferroptosis; increased age-related cancer risk in mice and humans; and an altered macrophage profile and response to bacterial infections in mice (19–21, 24).

Given its unique characteristics, we used the P47S polymorphism to interrogate the physiologic consequences resulting from a shift in CoA/GSH-mediated redox homeostasis and to identify factors and pathways that might be targeted therapeutically in disorders of redox dysregulation. We now find that P47 and S47 cells exhibit distinct metabolic profiles which are tied to the different redox states and differences in the expression of the cysteine-responsive transcription factor ATF4. S47 cells exhibit reduced catabolic glycolysis, but enhanced use of the pentose phosphate pathway (PPP) and an increased abundance of metabolites that contribute to antioxidant defense. This physiologic outcome can be reversed: we find that modifying CoA/GSH or AT4 levels promotes changes in the structure and function of several key glycolytic pathway proteins that associate with their different metabolic profiles and ferroptotic responses. These data support a model wherein reversible S-thiolation of critical proteins is an important regulatory determinant in adaptive, coordinated cell signaling networks. Our studies also reinforce the impact of the metabolic state on ferroptosis sensitivity. This investigation generated insight into the diverse functions of cysteine and CoA in redox homeostasis and in signal transduction pathways and provides a framework for defining the underlying mechanisms and physiological consequences.

Results

LMW Thiol Abundance and Metabolic Reprogramming.

In addition to transporter-mediated uptake, cellular cysteine can be obtained from the diet and, in some cells, it can be synthesized de novo via the transsulfuration pathway (TSP). TSP activation helps maintain cellular thiol pools, and protects against ferroptosis (21, 31–34), and S47 cells have elevated TSP metabolites (ref. 21 and Fig. 1A). Our steady-state metabolite analysis confirmed that S47 cells have higher cysteine levels (Fig. 1A). We also noted differences in the abundance of several glycolytic pathway metabolites (Fig. 1B). It is well established that, after import, glucose is converted to glucose-6-phosphate (G6P), which is shunted into three main pathways: 1) catabolic glycolysis; 2) the PPP and anabolic pathways; and 3) glycogen storage. Wild type p53 (WT, P47) promotes the use of the catabolic glycolysis pathway that generates pyruvate (12, 35). However, we find that, relative to P47, the S47 cells have reduced levels of pyruvate, but increased NADPH (Fig. 1B). The oxidative PPP is a major source of NADPH. Immunohistochemical staining also revealed enhanced glycogen stores in S47 mouse livers, as well as the spontaneous, age-related development of hepatic fibrosis; these are not evident in P47 mice (Fig. 1C). Liver fibrosis is indicative of underlying tissue damage and inflammation, and its presence is consistent with impaired ferroptosis (21, 36, 37). Together, these data indicate that S47 cells exhibit a switch in glucose metabolism from catabolic glycolysis into branching pathways, including PPP and glycogen storage.

Fig. 1.

P47 and S47 variants exhibit different metabolic profiles. (A and B) Diagram illustrating relative differences in transsulfuration pathway (A) and glycolytic pathway (B) metabolites. Values are presented as the ratio between the mean values obtained in S47, relative to P47, samples (S47/P47). (C) Representative mouse liver tissue sections examined for glycogen by PAS staining, and for fibrosis by Sirius Red staining of collagen. (Scale bars, 50 μm.)

Glycolytic Pathway Proteins Exhibit Different Activities and Conformations in P47 and S47 Cells.

To determine the underlying basis for the differences in metabolic profiles between P47 and S47 cells, we analyzed three key glycolytic pathway enzymes, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), glucose-6-phosphate dehydrogenase (G6PD), and pyruvate kinase M2 (PKM2). The rationale for this selection is that GAPDH and PKM2 play a primary role in promoting glucose flux toward the synthesis of pyruvate. G6PD, on the other hand, is the first, rate-limiting enzyme in the PPP (38, 39). Our analyses revealed that, relative to P47, S47 cells exhibit reduced GAPDH enzymatic activity (Fig. 2A), and an increase in G6PD enzymatic activity (Fig. 2B); these findings are in line with the observed differences in glycolytic pathway metabolites (Fig. 1B). We reasoned that, if CoA/GSH abundance plays a role in the differences in glycolytic metabolites between P47 and S47, then changing their cellular levels should impact the enzymatic activities of GAPDH and/or G6PD. Consistent with this hypothesis, we find that exogenous administration of either CoA or GSH to P47 cells promotes a rapid, dose-dependent increase in intracellular CoA and GSH levels (Fig. 2 C and D). We also find reduced GAPDH activity (Fig. 2A), enhanced G6PD enzymatic activity (Fig. 2B), and an increase in the cellular NADPH/NADP⁺ ratio (Fig. 2E), all of which are comparable to what we observe in control S47 cells. We have confirmed that the protein expression levels of GAPDH, G6PD, PKM2, and p53 are comparable in P47 and S47 cells, and they remain unchanged following altered levels of CoA and GSH (Fig. 2F). Note, in response to exogenous administration of LMW thiols we detect a small shift in the migration of the P47 protein to that of the S47 variant (Fig. 2F), which may result from S-thiolation. Thus, the distinct metabolic profiles in the P47 and S47 cells associate with changes in the activities of these key regulatory proteins, rather than with alterations in their abundance.

Fig. 2.

Effect of P47S polymorphism and redox environment on protein structure and activity. (A and B) P47 and S47 cells were treated as indicated for 5 h and analyzed for relative GAPDH (A) or G6PD (B) enzymatic activity. Means and SD are shown (n = 3, *P < 0.05, **P < 0.01). (C) Cells were treated as indicated and analyzed for relative abundance of CoA or GSH. Means and SD are shown (n = 3, *P < 0.05, **P < 0.01). (D and E) Cells treated as in A and B were analyzed for GSH (D) or NADPH/NADP⁺ ratio (E). Means and SD are shown (n = 3, *P < 0.05, **P < 0.01). (F) P47 and S47 cells were treated as indicated for 24 h. Cell lysates were prepared and examined by Western blotting for the expression of the indicated proteins. (G) The same lysates as in F were treated with BMH, resolved by SDS/PAGE, and probed by Western blotting for the indicated proteins. (H) P47 and S47 Hupki mice were treated as described in SI Appendix, Materials and Methods. Liver lysates were prepared, treated with BMH, and probed by Western blotting for GAPDH.

Like p53, GAPDH and PKM2 are subject to various PTMs. For instance, in vitro and in vivo S-thiolation by CoA or GSH can alter GAPDH protein structure and inhibit its enzymatic activity (10, 11, 40). PKM2 is subject to complex allosteric regulation and redox-mediated PTM, switching between a high activity tetrameric form that promotes glycolysis and a low activity dimeric/monomeric conformation that results in the increased use of the PPP (41). Exposure of cells to exogenous cysteine inhibits PKM2 tetramer formation and activity, leading to a decrease in levels of pyruvate and downstream metabolites, including ATP (42, 43). G6PD activity can be inactivated by dethiolation, and restored by GSH or cysteine (44), although the precise molecular basis for this effect remains to be defined. We find that the S47 cells have increased cysteine and NADPH levels, but lower cellular pyruvate abundance (Figs. 1 A and B and 2E). There also is less ATP in S47 cells and mouse livers (SI Appendix, Fig. S1A). These observations are the expected outcomes of increased G6PD, but decreased GAPDH and PKM2, activities and are consistent with a redirection of metabolism away from catabolic glycolysis.

We next sought to test the hypothesis that the redox state and protein conformation of PKM2, GAPDH, and G6PD are altered in P47 and S47 cells, consistent with their activity differences in response to a changing thiol environment. To do so, we used the thiol reactive reagent, bismaleimidohexane (BMH), which covalently conjugates free (reduced) sulfhydryl groups and cross-links cysteine residues within 13 Å, thereby providing a way to assess changes in protein cysteine status. Because not all cysteines are modified, the BMH cross-linking pattern observed for any protein is a reflection of those residues that are modified, the spacing of the affected cysteines, and the tertiary structure. We find that the BMH cross-linking patterns of GAPDH are distinct in P47 and S47 cultured cells and liver tissue (Fig. 2 G and H). Similarly, the sulfhydryl cross-linking studies also show distinct p53, G6PD, and PKM2 protein gel migration patterns in P47 and S47 cells (Fig. 2G). Notably, the observed altered mobility of these enzymes following BMH cross-linking is reversible with exogenous CoA. For example, we find that treating P47 cultured cells with CoA for 5 or 24 h, or treating P47 mice with CoA once daily for 3 d, changes the protein cross-linking patterns to resemble that in the S47 samples (Fig. 2 G and H and SI Appendix, Fig. S1B). In comparing the effect of different LMW thiols, we find that providing 1 mM CoA to the culture medium under these conditions proved to be more effective in altering the BMH cross-linking patterns of these proteins, when compared to 3 mM GSH or 3 mM N-acetyl-L-cysteine (NAC) (Fig. 2G). We also confirmed that treating S47 cells and livers with diethyl maleate (DEM) to deplete cellular CoA and GSH abundance (SI Appendix, Fig. S1 C and D), has the opposite effect, generating protein structures like those in the P47 cells (Fig. 2H and SI Appendix, Fig. S1B).

Biotin Pull-Down Assays Reveal CoA Binding to p53 and Other Cellular Proteins.

We next sought to test the hypothesis that proteins like p53 and PKM2 might be subject to direct modification by CoA. To do this we devised a biotin-avidin pull-down approach as a way to help identify CoA-bound proteins in vivo. For these initial proof-of-principle assays, we generated a biotinylated-CoA derivative (biotin-CoA) and treated P47 and S47 cells with this compound for 6 h before capturing biotin-CoA and associated proteins using NeutrAvidin resins. Potential CoA-interacting proteins were eluted using dithiothreitol (DTT), which cleaves the disulfide bond in the spacer arm of the biotin attached to CoA. The associated proteins were resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS/PAGE) and detected by Western blotting. Cells treated with biotin alone served as a negative control. Using this assay, we obtained evidence for biotin-CoA binding (S-CoAlation) to p53 as well as to PKM2, GAPDH, and G6PD in both P47 and S47 cells (Fig. 3A and SI Appendix, Fig. S2A). In contrast, we detected no significant interaction between biotin-CoA and the protein BCL-xL (Fig. 3A). Also, biotin-CoA-mediated interaction with p53, PKM2, GAPDH, or G6PD was reduced in cells cultured in media containing excess untagged CoA or GSH (Fig. 3B and SI Appendix, Fig. S2B). In this pull-down assay, the untagged CoA and biotin-tagged CoA comparably modify the cross-linking patterns of all of the proteins we analyzed (p53, GAPDH, PKM2, and G6PD) (Figs. 2G and 3C) in the P47 cells, but did not affect protein abundance (Figs. 2F and 3A).

Fig. 3.

Biotin-CoA pull-down assays detect in vivo protein S-CoAlation. (A) P47 cells were treated with biotin or biotin-CoA for 6 h. Cell lysates were prepared and examined by Western blotting for the expression of the indicated proteins (input). For Avidin IP, the same lysates were captured using NeutrAvidin resins, separated by SDS/PAGE, and probed for the proteins indicated. (B) Protein lysates were prepared from P47 cells pretreated with CoA for 1 h followed by biotin or biotin-CoA for 6 h, and examined by Western blotting for the expression of the indicated proteins (input). The same lysates were captured using NeutrAvidin resins, separated by SDS/PAGE, and probed for the proteins indicated. (C) The same lysates as in A were treated with BMH and examined for the migration/structure of the indicated proteins.

Redox Signaling, Glycolytic Flux, and Ferroptosis.

We next investigated the potential for interplay among redox status, glucose metabolism, and ferroptosis sensitivity. To do so, we employed several compounds that alter the cellular redox state and/or glycolytic metabolism, as follows: propargylglycine (PPG) decreases cysteine biosynthesis by inhibiting the TSP enzyme cystathionine γ-lyase (32); TEPP46 promotes catabolic glycolysis by activating PKM2 (45); pyruvate, or its precursor glutamine, decreases CoA, GSH, and cysteine levels (46–49). Importantly, for all of our studies, we use a defined medium containing physiologically relevant concentrations of pyruvate, cystine, and glutamine as described (refs. 11, 21, 50 and SI Appendix, Materials and Methods).

We confirmed that exposure of cells to the ferroptosis inducer glutamate promotes a substantial depletion of cellular GSH in the P47 cells (Fig. 4A). As expected, this response is blunted in S47 cells (Fig. 4A) likely because of the higher endogenous levels of CoA, GSH, and cysteine. Under these treatment conditions, PPG, TEPP46, pyruvate, and glutamine are more effective than glutamate in decreasing GSH abundance in the S47 cells (Fig. 4A). Consistent with an influence of the redox environment on the metabolic state of the cell, each compound promoted an increase in GAPDH activity (Fig. 4B), an increase in NADH/NAD⁺ ratio (Fig. 4C), a decrease in G6PD activity (Fig. 4D), and a decrease in NADPH/NADP⁺ ratio (Fig. 4E). In each case these compounds restored the levels in S47 cells to those observed in the untreated P47 cells. In contrast, treating P47 cells with CoA or GSH, to increase thiol abundance, modified the metabolic profile to resemble that in the untreated, control S47 cells (Fig. 2 A–E).

Fig. 4.

Cross-talk among cellular thiol levels, glycolysis, and ferroptosis. (A–E) P47 and S47 cells, treated for 5 h as indicated, were analyzed for relative abundance of GSH (A), GAPDH activity (B), NADH/NAD⁺ ratio (C), G6PD activity (D), or NADPH/NADP⁺ ratio (E). Means and SD are shown (n = 3, *P < 0.05, **P < 0.01). (F) Lysates from cells treated as in A–E were treated with BMH and probed by Western blotting for the indicated proteins. (G) Cells were treated as indicated for 24 h, and cell viability was assessed by MTT assays. Means and SD are shown (n = 4, *P < 0.05).

These metabolic compounds also were tested for their ability to promote altered gel migration patterns of the affected proteins, indicative of modified cysteine status. The results show that S47 cells treated with PPG, TEPP46, pyruvate, or glutamine exhibit the predicted change in protein cross-linking patterns to what is observed in P47 cells; this is evident for each of the four proteins examined (p53, GAPDH, G6PD, and PKM2) (Fig. 4F). Likewise, decreasing cellular CoA and GSH abundance by treating S47 cells with DEM (SI Appendix, Fig. S1 B and C), or subjecting the cells to cystine starvation (SI Appendix, Fig. S3 A and B), altered the protein migration patterns of these enzymes to mirror those in P47 cells.

We next sought to test the link between these metabolites and their influence on protein conformation and metabolism, with sensitivity to ferroptosis. We find that cotreating the S47 cells with glutamate and either PPG, TEPP46, pyruvate, or glutamine promoted an increase in glutamate-induced cell death to levels similar to those in the P47 cells (Fig. 4G). Cell viability determinations were comparable when assayed by release of lactate dehydrogenase (LDH) from damaged cells or by MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] analyses (Fig. 4G and SI Appendix, Fig. S4 A and B). In both the P47 and S47 cells, an increase in glutamate-induced cell death associated with elevated reactive oxygen species (ROS) production (SI Appendix, Fig. S4C). The results are consistent with the proposed role of catabolic glycolysis and glutaminolysis in cysteine-deprivation-induced ferroptosis (47, 51). Further, glutamate-induced loss of cell viability in P47 and S47 cells correlated with the enhanced expression of ferroptosis and p53 biomarkers (SI Appendix, Fig. S4 D–F). We confirmed that the resulting cell death could be suppressed by the ferroptosis inhibitor liproxstatin-1 (LIP-1), as well as by administration of CoA, GSH, or NAC (Fig. 4G). Similarly, the exogenous administration of pyruvate or glutamine to S47 cells also enhanced cell death in response to another inducer of ferroptosis, the small molecule erastin (SI Appendix, Fig. S4G).

ATF4 Expression Influences Redox Signaling, Glycolytic Flux, and Ferroptosis Sensitivity.

The integrated stress-response transcription factor ATF4 is a mediator of ferroptosis sensitivity (28, 52–54). This key regulator of cysteine and glutathione metabolism also is dynamically regulated in response to cysteine/glutathione concentrations (6, 52–55). MDM2 binds to, and regulates the activity of, ATF4 (56), and our finding of decreased MDM2 induction in S47 cells (21) prompted us to test ATF4 expression in P47 and S47 cells. This analysis revealed that S47 cultured cells and mouse liver have lower ATF4 protein levels than P47 samples (Fig. 5 A and B). We also find that cystine/cysteine starvation, or treating S47 cells with compounds that reduce cellular thiol levels, such as PPG and pyruvate, promotes an increase in ATF4 protein abundance (Fig. 5C and SI Appendix, Fig. S5A). To test the hypothesis that reduced ATF4 function is an important element in the altered metabolic profile and differential ferroptosis sensitivity of the S47 cells, we generated S47 cells constitutively expressing ATF4 (Fig. 5D). Relative to S47, the S47/ATF4 cells exhibit changes in both G6PD and GAPDH activities (Fig. 5E), as well as the basal expression levels of ATP, GSH, glutamine, and NADPH (Fig. 5F and SI Appendix, Fig. S5B); in each case, the phenotype of the S47/ATF4 cells approximates that of the P47 cells. These results are consistent with the known role of ATF4 in promoting catabolic glycolysis and redox regulation (6, 28, 52–55). This ATF4-mediated change in the metabolic profile of the S47 cells was accompanied by a corresponding modification in the BMH cross-linking patterns of the p53, GAPDH, PKM2, and G6PD proteins, to resemble those in the P47 cells (Fig. 5G and SI Appendix, Fig. S5C). Moreover, enhancing ATF4 protein expression sensitized the S47 cells to glutamate-induced ferroptotic cell death, LDH release, and cellular ROS production (Fig. 5H and SI Appendix, Fig. S5 D and E), which were associated with increased expression of the ATF4 target genes and ferroptosis biomarkers (52) CHAC1 and SLC7A11/xCT (SI Appendix, Fig. S5F). Thus, restoring ATF4 protein function in S47 cells is sufficient to reset critical metabolic signaling pathways, redox status, and ferroptosis sensitivity.

Fig. 5.

Effect of ATF4 restoration on cross-talk among cellular thiol levels, glycolysis, and ferroptosis in S47 cells. (A and B) Cell lysates were prepared from P47 and S47 cells (A) or livers (B) and examined by Western blotting for the expression of the indicated proteins. (C) P47 and S47 cells were cultured with or without added l-cystine. Cell lysates were probed by Western blotting for ATF4 or ribosomal protein S6. (D) Western blot analysis of cell lysates from P47, S47, and S47/ATF4 cells for expression of the indicated proteins. (E and F) P47, S47, and S47/ATF cells were analyzed for G6PD and GAPDH enzymatic activities (E), as well as GSH abundance and NADPH/NADP⁺ ratio (F). Means and SD are shown (n = 4, *P < 0.05, **P < 0.01). (G) The same lysates as in D were treated with or without BMH, resolved by SDS/PAGE, and probed by Western blotting for the indicated proteins. (H) Cells were treated as indicated for 24 h, and cell viability was assessed by MTT assays. Means and SD are shown (n = 4, **P < 0.01).

Discussion

Cystine/cysteine homeostasis and redox-mediated PTMs of protein cysteines represent important, but often overlooked, mechanisms in the control and integration of cell signaling pathways (4–9). Cells containing the S47 variant of p53 are resistant to ferroptosis due to increased levels of the metabolites CoA and GSH and altered expression of ferroptotic genes (21). Here we show that the elevated abundance of LMW thiols also modifies the metabolism of these cells, in a manner that is linked to redox-mediated changes in the activities of key regulatory proteins, including G6PD, ATF4, and p53.

The S47 isoform of p53, like the P47 variant cultured in media containing excess CoA/GSH, exists predominantly in the monomeric/inactive conformation, rather than the tetrameric/active state (Figs. 2G and 4F). This redox reversible structure of p53 may have evolved to allow for rapid cycling between active and inactive states in response to changing cellular conditions (57, 58). We carried out differential scanning fluorimetry (DSF) to generate thermal melting profiles of full-length recombinant P47 and S47 proteins. The results reveal that the P47 protein has a higher melting temperature (T_m) relative to S47; this difference can be recapitulated by incubating the P47 protein with GSH or CoA (SI Appendix, Fig. S6). Together, the available data suggest a model in which the cysteine residues of the S47 isoform may be modified by LMW thiols more readily than those in the P47 isoform, leading to altered expression of a set of p53 target genes that regulate redox homeostasis and metabolism. This would initiate a cycle in which even an initial modest increase in the abundance of LMW thiols could be sufficient to propel further changes in the structure and functions of the S47 variant, ultimately leading to concomitant, but reversible, changes in cellular redox homeostasis and in the activities of redox-sensitive proteins like ATF4, GAPDH, PKM2, and G6PD. Such a model would be consistent with the age-related increase in fibrosis noted in older S47 mice (Fig. 1C). Interestingly, this cascade model also has features in common with one addressing the impact of dysregulated cysteine homeostasis and ATF4 expression in Huntington’s disease (54).

Like p53, the ATF4 transcription factor is redox regulated, plays an important role in the cellular response to oxidative stress, helps balance redox homeostasis, and promotes catabolic glycolysis. Also like p53, the role of ATF4 in promoting a prodeath or prosurvival response to stress, is complex and context dependent. A number genes implicated in cysteine-depletion-mediated ferroptotic cell death are ATF4 targets, including CHAC1 and SLC7A11/xCT (52), and germline deletion of ATF4 can render primary neurons resistant to glutamate-induced ferroptosis (6, 53). Here we find that restoring ATF4 protein expression promotes a metabolic reprogramming in the S47 cells toward catabolic glycolysis, enhances ferroptosis sensitivity, and alters redox homeostasis. Thus, changes in cysteine/glutathione metabolism can have a widespread impact on cellular physiology and response to oxidative stress.

While reduced glycolysis and enhanced use of the PPP and branching pathways are characteristic of many cancer cells, they also are properties of several normal cell types, such as stem-like cells and certain immune cell types. As an example, two major classes of macrophages, referred to as M1 and M2, exhibit glycolytic and functional differences; targeting key metabolic enzymes can change the major immune functions of these and other myeloid cells. Likewise, alterations in CoA or GSH metabolism have been linked to deregulated macrophage plasticity and sensitivity to bacterial infections (59, 60). It is of interest, therefore, that P47 and S47 mice exhibit distinct macrophage phenotypes and differential responses to bacterial infections (24). Our investigations were carried out using nontransformed cells, so it will be important to also address the role of thiol-based redox signaling in tumor cells. As an example, a recent study found that depriving pancreatic ductal adenocarcinoma (PDAC) cells of cysteine increased the level of lipid ROS and promoted ferroptosis. Interestingly, while GSH loss alone did not lead to ferroptosis in these tumor cells, a reduction of CoA together with GSH was effective (34).

Not all protein cysteines are subject to S-thiolation, which likely occurs in a site- and protein-specific manner (8, 9). However, such PTM pathways have remained challenging to study due to the scarcity of available reagents and approachable methods to characterize targets. Herein we have devised a protocol to assess CoAlation of proteins. One limitation of this pull-down assay is the need to add exogenous CoA, which currently narrows its use in examining alterations in protein CoAlation under different conditions. Nonetheless, this tractable approach should be useful for detecting S-CoAlated or S-glutathionylated proteins when a specific subset of proteins are of interest and should allow more rapid identification of potential targets of redox regulation. Differences in the cellular abundance of CoA and GSH have been linked to several human pathologies, including neurodegenerative diseases, cancers, viral infections, and immune dysregulation. A better understanding of the functional cross-talk among pathways regulating redox signaling, metabolic adaptation, and ferroptosis is expected to generate new insight on redox-mediated signaling in diverse aspects of human health and disease.

Materials and Methods

All experimental procedures involving mice were conducted in accordance with protocols approved by The Institutional Animal Care and Use Committee, Office of Animal Welfare of the Perelman School of Medicine at the University of Pennsylvania, and conformed to the guidelines outlined in the National Institutes of Health Guide for Care and Use of Laboratory Animals (61). Materials and experimental procedures for cell culture, animal studies, RNA isolation, and quantitative RT-PCR, metabolite and viability studies, periodic acid Schiff (PAS) and Sirius Red staining, protein isolation, BMH cross-linking, Western blotting, biotin-CoA pull-down assays, and quantification and statistical analysis are described in SI Appendix.

Data Availability.

All study data are included in the article text and SI Appendix.