MESH1 is a cytosolic NADPH phosphatase that regulates ferroptosis

Abstract

Critical to the bacterial stringent response is the rapid relocation of resources from proliferation toward stress survival through the respective accumulation and degradation of (p)ppGpp by RelA and SpoT homologues. While mammalian genomes encode MESH1, a homologue of the bacterial (p)ppGpp hydrolase SpoT, neither (p)ppGpp nor its synthetase has been identified in mammalian cells. Here, we show that human MESH1 is an efficient cytosolic NADPH phosphatase that facilitates ferroptosis. Visualization of the MESH1-NADPH crystal structure revealed a bona fide affinity for the NADPH substrate. Ferroptosis-inducing erastin or cystine deprivation elevates MESH1, whose overexpression depletes NADPH and sensitizes cells to ferroptosis, whereas MESH1 depletion promotes ferroptosis survival by sustaining the levels of NADPH and GSH and by reducing lipid peroxidation. The ferroptotic protection by MESH1 depletion is ablated by suppression of the cytosolic NAD(H) kinase, NADK, but not its mitochondrial counterpart NADK2. Collectively, these data shed light on the importance of cytosolic NADPH levels and their regulation under ferroptosis-inducing conditions in mammalian cells.

Extended Data Figure 1,. MESH1 contributes to the NADPH phosphatase activity in cells.

NADPH phosphatase activity of HEK-293T was estimated by incubating NADPH with cell lysates with different MESH1 status; phosphate release over time was measured by the malachite green assay and normalized by the total lysate protein (μmol/min/mg total lysate protein). a, Significantly decreased cellular NADPH phosphatase activity by silencing MESH1 in HEK-293T cells. b, Increased cellular NADPH phosphatase activity with overexpression of WT MESH1, but not with overexpression of enzymatically deficient MESH1 (MESH1-E65A) in HEK-293T cells. Black dots represent siNT or empty vector samples, blue dots represent siMESH1 samples, red dots represent MESH1-WT samples, and grey dots represent MESH1-E65A samples. Error bars represent s.d. (n=3 biologically independent samples per condition). Statistical analysis: one way ANOVA with Tukey HSD post-hoc, **P<0.01, N.S., not significant.

Extended Data Figure 2.. MESH1 asymmetric unit, omit map, and Zn²⁺ anomalous phasing.

a, Crystallographic asymmetric unit is shown with two protomers of hMESH1 in the ribbon diagram and NADPH molecules in the stick model. b, 2mFo-DFc omit maps for the ADP (left and middle) and nicotinamide riboside moieties (right) of the NADPH molecules contoured to 0.8σ in purple mesh. c, The active site of each protomer is shown along with density from Zn²⁺ anomalous scattering contoured to 6σ in black mesh.

Extended Data Figure 3.. Up-regulation of MESH1 protein and its regulation of NADP(H) and cell survival during ferroptosis.

a-b, The levels of MESH1 protein were determined by Western blots of MESH1 and β-tubulin protein in RCC4 cells treated with (a) DMSO vs. erastin (3 µM; 12h) or (b) 200 or 0 µM for 20h. All the representative western blots were repeated for 3 biologically independent times with similar results. c, Overexpression of WT MESH1, but not the MESH1 E65A mutant, significantly reduced the intracellular NADP(H) content in RCC4 cells. d, Overexpression of MESH1-WT, but not MESH1-E65A, sensitized the cell death to erastin treatment in RCC4 cells. Black dots represent empty vector samples, red dots represent MESH1-WT samples, and grey dots represent MESH1-E65A samples. e, MESH1 silencing restored the intracellular NAPDH level decreased by erastin treatment. HEK-293T cells transfected with the NADPH fluorescent biosensor, iNAP1, and siNT or two independent siMESH1 RNAs were treated with 20 µM of erastin. The ratio of fluorescence excited at 420 nm and 485 nm indicated the cellular NADPH level. The fluorescence ratios of individual treatments were normalized to that of the control condition. (n=6 biologically independent samples). f, The viability of RCC4 cells transfected with siNT (black) or two independent siMESH1 (blue) and treated with indicated concentrations erastin for one day. Each dot represents one sample; there are three biologically independent samples (n=3) in each group unless otherwise specified. Error bars indicate s.d. Statistical analysis: one-way ANOVA with Tukey HSD post-hoc test, *P<0.05, **P<0.01.

Extended Data Figure 4.. MESH1 depletion conferred resistance to ferroptosis in multiple assays.

a-c, Erastin-induced cell death is rescued by MESH1 silencing in RCC4 cells using multiple cell death and viability assays: a, Cytotox-Flour for measuring protease release in the media due to cell death; b, Crystal violet staining for attached viable cells; and c, CellTiter-Glo for intracellular ATP and viability. Error bars indicate s.d. (n=3 biologically independent samples per condition). Statistical analysis: one-way ANOVA with Tukey HSD post-hoc test, *P<0.05, **P<0.01.

Extended Data Figure 5.. MESH1 depletion increased resistance to ferroptosis in multiple cell lines.

a-h. Erastin-induced ferroptotic cell death of indicated cell lines was rescued by MESH1 silencing as accessed by CellTiter-Glo: a, HEK-293T cells exposed to erastin 2.5 μM for one day; b, lung adenocarcinoma cell line H1975 with erastin 10 μM for two days; c, breast cancer cell line MDA-MB-231 with erastin 2.5 μM for one day; d, prostate cancer cell line PC3 with erastin 10 μM for one day; e, fibrosarcoma cell line HT1080 with erastin 2.5 μM for one day; f, Ewing’s sarcoma cell line A673 with erastin 2.5 μM for one day; g, pancreatic carcinoma cell line PANC1 with erastin 2.5 μM for one day, and h, clear cell renal carcinoma cell line 786-O with erastin 5 μM for one day. Black dots represent siNT samples, blue dots represent siMESH1 samples. Error bars indicate s.d. (n=3 biologically independent samples per condition). Statistical analysis: one-way ANOVA with Tukey HSD post-hoc test, *P<0.05, **P<0.01.

Extended Data Figure 6.. MESH1-depletion increased the ratio of reduced glutathione over oxidized glutathione.

a-d, The level of indicated metabolites in RCC4 transfected with non-targeting (NT) siRNA or MESH1-targeting siRNA were quantified by mass spectrometry and normalized by DNA content, a, cysteine; b, oxidized glutathione (GSSG); c, reduced glutathione (GSH) (effect size 1.746, p=0.0491); and d, normalized intensity ratio of GSH/GSSG. n=5 biologically independent samples per condition. Each box plot was defined by a lower line (at the 25th percentile), a central line (at the centre), and a higher line (at the 75th percentile). The whisker boundaries marked the minima and the maxima of the data. Statistical analysis: two-tail student’s t-test, *P<0.05.

Extended Data Figure 7.. MESH1 silenced cells were re-sensitized to erastin-induced cell death upon NADK silencing.

Percentage of cell viability of HEK-293T cells after treatment with erastin 10 μM for one day, showing simultaneously silencing MESH1 with NADK, but not NADK2, restored sensitivity to erastin-induced ferroptosis. Black dots represent siNT samples, blue dots represent siMESH1 samples. Error bars indicate s.d. (for siNT and siMESH1 n=6 biologically independent samples per condition; for other genetic conditions n=3 biologically independent samples). Statistical analysis: ANOVA with Tukey HSD post-hoc test *P<0.05, **P<0.01.

Extended Data Figure 8.. Relative expression levels of genes involved in the NADPH and glutathione metabolisms.

Three plates of RCC4 cells were sequenced and the relative expression levels of the indicated genes were calculated by RPKM (Reads per kilo base per million mapped reads). Error bars indicate s.d. (n=3 biologically independent samples per condition).

Figure 1.. MESH1 is a mammalian NADPH phosphatase.

a, Chemical similarity between NADPH and ppGpp, and the proposed chemical reaction of MESH1. b, Detection of the hMESH1-catalyzed phosphate release from NADPH by color change (yellow to green) in the malachite green assay. c, Linear product accumulation of the NADPH dephosphorylation reaction catalyzed by hMESH1 (n=3 biologically independent samples, error bars indicate s.d.). d, Validation of the formation of NADH by mass spectrometry analysis. This experiment was repeated for three independent times with similar results. e, Enzymatic characterization of hMESH1 toward NADPH and NADP⁺ (mean ± s.d., n=4 independent experiments). f, Effects of Zn²⁺ substitution and active site mutation on the specific activity of purified recombinant hMESH1 (mean ± s.d., n=3 independent experiments). g. hMESH1 is a significant contributor to the NADPH phosphatase activity in RCC4 lysates. Black dots represent siNT samples, and blue dots represent siMESH1 samples. n=4 biologically independent samples for each condition; Samples were analyzed by Western blot to confirm knock-down efficiency. h, Cell fractionation separating cytosol from nucleus and mitochondria, showing that MESH1 is specifically enriched in the cytosolic compartment. This representative western blot was repeated independently for 3 times on RCC4 cells and HEK-293T cells with similar results. Error bars indicate s.d. One-way ANOVA with Tukey HSD post-hoc test, *P<0.05, **P<0.01.

Figure 2.. Structure of the hMESH1 (D66K)-NADPH complex.

a, The architecture of the hMESH1 active site. MESH1 is shown in the ribbon diagram, NADPH in the stick model, and the Zn²⁺ (grey) and Na⁺ (pink) ions are shown as spheres. Secondary structures are labeled, with MESH1-NADPH-interacting motifs annotated in magenta. b, Coordination of the active site Zn²⁺ ion in a distorted octahedral geometry. NADPH and sidechains of the Zn²⁺-binding residues are shown in the stick model. The zinc ion (grey) and its coordinating water molecule (red) are shown as spheres. The signature ‘HD’ motif is annotated in blue. c, The schematic illustration of the NADPH recognition by MESH1. Polar interactions are denoted with dashed lines, and vdW contacts are shown with curved lines. The locations of the metal ions and NADPH-interacting water molecules are denoted as circles. d, Molecular recognition of the 2’-phospho-adenosine diphosphate moiety. e, Molecular recognition of the nicotinamide riboside moiety.

Figure 3.. MESH1 regulates cellular NADP(H) levels and ferroptosis.

a, A schematic model describing the roles of NADPH, MESH1, NADK and NADK2 in the cytosolic and mitochondrial NADPH/glutathione metabolisms and the targets of reported ferroptosis-inducing agents. b, Changes of MESH1 RNA levels in treated RCC4 cells. Left panel, RCC4 cells treated with DMSO or erastin (1μM); right panel, RCC4 cells cultured under regular (200 μM cystine) or cystine-deprived (0 μM cystine) media. RNA levels were quantified by rt-qPCR relative to the house keeping gene β-actin. c, Percentage changes of NADP(H) after one day treatment with 20 μM erastin (left) or cystine deprivation (right) when compared with control treatments (left: DMSO, right: full media) in RCC4 cells transfected with non-targeting (NT) siRNA or 2 distinct MESH1-targeting siRNAs. (left: n=6 biologically independent samples for siNT, n=3 biologically independent samples for siMESH1-CDS and siMESH1–3’UTR; right: n=3 biologically independent samples for each group). The MESH1 knock-down efficiency was validated by Western blot. This representative western blot was repeated for 3 biologically independent times with similar results. d, The lipid peroxidation of RCC4 cells with or without one-day treatment of erastin (1μM) accessed by C11-BODIPY and flow cytometry in cells transfected with non-targeting (NT) siRNA or 2 distinct MESH1-targeting siRNAs. The percentage of cells showing increased fluorescence (indicated by the bar) for each treatment are labeled. e, Time course of cell death in RCC4 cells transfected with non-targeting (NT) siRNA or 2 distinct MESH1-targeting siRNAs under erastin (0.625 μM) treatment for up to one week. f-i, The viability of RCC4 cells transfected with siNT (black) or two independent siMESH1 (blue) and treated with different concentrations of indicated ferroptosis-inducing conditions or agents for one day: f, cystine deprivation; g, sulfasalazine; h, ML-162; and i, RSL3. Each dot represents one sample; (b, e, f-i): n=3 biologically independent samples in each group. If unspecified, siMESH1 indicates siMESH1-CDS. RCC4 cells were transfected with siRNA for 2 days and then incubated with erastin for one day (c, d) or seven days (e). Black dots represent siNT samples, and blue dots represent siMESH1 samples. Error bars indicate s.d. Statistical analysis: (b,) two-tail Student’s t-test. (c) One-way ANOVA with Tukey HSD post-hoc test. (f, g) Two-way ANOVA with Tukey HSD post-hoc test. *P<0.05, **P<0.01, ***P<0.005, N.S., not significant.

Figure 4.. The ferroptosis resistance phenotypes are caused by the loss of the cytosolic NADPH phosphatase activity of MESH1.

a, Percentage of cell viability of RCC4 cells after erastin (5 μM) treatment for one day. Lentivirus with the empty vector, MESH1-WT, or MESH1-E65A, were stably transduced into RCC4 prior to siRNA transfection. The experiments were repeated for 3 biologically independent times with similar results. The knockdown and overexpression efficiencies were validated by Western blot. b, Knockdown of the cytosolic enzyme NADK, but not mitochondrial enzyme NADK2, eliminated the ferroptosis protection in MESH1 silenced cells. c, Co-silencing of NADK, but not NADK2, reduced the NADP(H) levels accomplished with MESH1 silencing. d, Normalized levels of MESH1, NADK and NADK2 in cells treated with indicated siRNAs. The mRNA abundance was quantified by rt-qPCR relative to the house keep gene β-actin, and presented in fold change (2^∆∆CT). This representative rt-qPCR were repeated for 3 independent times with similar results. If unspecified, siMESH1 indicates siMESH1-CDS. RCC4 cells were transfected with siRNA for 2 days and then incubated with erastin for one day (a, b). Black dots represent siNT samples, and blue dots represent siMESH1 samples. Error bars indicate s.d. (for (a,b,c,d) n=3 biologically independent samples per condition). Statistical analysis: (a, b, c) two-way ANOVA with Tukey HSD post-hoc test, *P<0.05, **P<0.01, ***P<0.005, N.S., not significant.