Hibernator-derived cells exhibit increased cold resistance

a, Schematic of cold cell viability counting, consisting of 1 day cold exposure (4°C), followed by a short rewarming at 37°C for 15, 30, or 60 minutes before trypan blue staining. b, Number of viable K562 cells based on trypan blue staining after one day at 4°C and subsequent rewarming for 24 hours at 37°C. Numbers are normalized to initial cell counts. Blue shaded regions indicate 4°C exposure and shaded pink regions indicate 37°C exposure. Red square indicates calculated cell counts after one day at 4°C based on the viable cell number measured after 24 hour rewarming. c, Viability of K562 cells was assessed by trypan blue staining after incubation at 37°C for 0, 15, 30, or 60 minutes following 24 hours at 4°C (n = 4). Cells incubated at 37°C for 15, 30, and 60 minutes show a significant increase in cell counts compared to cells counted without rewarming (n = 4 samples per condition, ***P = 0.0007, ****P < 0.0001). d, Viability of hibernator (BHK-21, HaK)- and non-hibernator (HeLa, RPE1, HT1080, K562)-derived cell lines at 4°C as measured by trypan blue staining. Hibernator-derived lines show significantly increased cold resistance compared to lines derived from non-hibernators at 7 days 4°C (n = 4 samples per data point, ****P < 0.0001). e, f, Fluorescence images of hibernator- and non-hibernator-derived cell lines after 4 days at 4°C. Cultures were stained with 1 µg/mL Hoechst 33342 and 1 µg/mL propidium iodide (PI) to distinguish live and dead cells. g, Viability of hibernator- and non-hibernator-derived cell lines at 4°C as measured by LDH release (n = 4 samples per data point, ****P < 0.0001). All values show mean ± SEM, with significance measured by one-way ANOVA adjusted for multiple comparisons by Dunnett’s test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Unbiased CRISPR screens identify the GPX4 pathway as necessary for cold-induced survival in hibernator-derived BHK-21 cells

a, Schematic of CRISPR screen paradigm, consisting of four passages of control cells kept at 37°C and three cycles of cold exposure (4°C) for 4 days followed by rewarming (37°C) for 2 days. Yellow dots indicate points of sample collection. b, Volcano plot showing median log2 fold-change of genes after three cycles of cold exposure and rewarming (4°C) compared to cells following three passages at 37°C. Red dots indicate selectively required genes with a median log2 fold-change < −1 or > 1 and FDR < 0.1. c, Heatmap of the median log2FC of ferroptosis-related genes after three cycles of cold exposure and rewarming (4°C) compared to three passages at 37°C. d, Partial schematic of the selenocysteine incorporation pathway. Left: Production of the GPX4 selenoprotein requires recoding of a UGA codon to the amino acid selenocysteine (Sec). This process involves a cis-acting SECIS element within the Gpx4 mRNA 3’ UTR, SECIS binding protein 2 (SECISBP2), a specific eukaryotic elongation factor (EEFSEC), and Sec-charged tRNA. Right: The Sec-charged tRNA is generated by the combined action of Phosphoseryl-tRNA kinase (PSTK), Selenophosphate synthetase 2 (SEPHS2), and selenocysteine synthase (SEPSECS). e-h, Median log2 fold-change (log2FC) of 10 guides per targeted gene, showing guide depletion over three cycles of cold exposure and rewarming. Significance between Cycle 1 versus Cycle 3 is measured by two-way ANOVA adjusted for multiple comparisons by Dunnett’s test. Significance between 37°C and 4°C for each cycle is measured by two-way ANOVA adjusted for multiple comparisons by Bonferroni’s test. e, Gpx4, f, Eefsec, g, Secisbp2, h, Pstk. i, Schematic of CRISPR screen paradigm, showing cells exposed to 4°C continuously for 15 days. Yellow dot indicates point of sample collection. j, Volcano plot showing median log2 fold-change in abundance of guides targeting the indicated genes after 15 days of 4°C exposure compared to one passage at 37°C. Red dots indicate selectively required genes with a median log2 fold-change < −0.5 or > 0.5 and FDR < 0.10. k, Heatmap of the median log2FC in abundance of guides targeting ferroptosis-related genes after 15 days of 4°C exposure compared to 37°C control cultures. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Panels a, d and i were created with BioRender (https://BioRender.com/v19u474, https://BioRender.com/b57h207 and https://BioRender.com/i24y807 respectively)

GPX4 activity is required for BHK-21 cell cold tolerance

a, Stable Gpx4 knockout (KO) BHK-21 cell lines exhibit reduced cold tolerance. Left: Western blot of wild-type (WT) and individual Gpx4 KO clone lysates for GPX4 and β-actin loading control. Right: Viability of Gpx4 KO lines is significantly lower than WT BHK-21 cells at 7 days 4°C by trypan blue staining (n = 4, ****P < 0.0001), with complete cell death by four days at 4°C. b, c, Reintroduction of wild-type Syrian hamster GPX4 (GPX4), but not a catalytically dead form of GPX4 (mGPX4) rescues cold-induced cell death in two independent BHK-21 Gpx4 KO clonal cell lines. Left panels: Western blots for HA and GPX4 along with β-actin loading control. Right panels: Expression of WT hamster GPX4 showed significantly higher cell viability at 7 days 4°C compared to the corresponding parental Gpx4 KO, GFP-, and mGPX4-expressing lines by trypan blue staining (n = 4, ****P < 0.0001). d, e, Treatment with the GPX4 inhibitors RSL3 (d) or ML162 (e) results in enhanced cold-induced death in BHK-21 cells by 4 days at 4°C as measured by trypan blue exclusion (n = 4, ****P < 0.0001). f, Cold-induced BHK-21 cell death upon RSL3 treatment occurs via ferroptosis. BHK-21 cells were placed at 4°C and treated with RSL3 (1 µM) and/or the ferroptosis inhibitor ferrostatin-1 (Fer-1, 1 µM) or iron chelator DFO (100 µM) for 4 days (n = 4). Treatment with RSL3 resulted in significantly lower cell viability than no treatment as determined by one-way ANOVA adjusted for multiple comparisons by Tukey’s HSD (****P < 0.0001). All values show mean ± SEM, with significance measured by two-tailed t test, unless otherwise indicated. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Genome-wide CRISPR screens identify GPX4 as a suppressor of cold-induced cell death in human cells

a, Schematic of CRISPR screen paradigm, consisting of three cycles of 5 days of cold exposure (4°C) interrupted by 3 day rewarming (37°C) periods. Yellow dots indicate points of sample collection. b, Volcano plot showing median log2 fold-change in abundance of guides targeting the indicated genes after three cycles of 4°C cold exposure compared to three passages at 37°C (Control). Red dots indicate selectively required genes with a median log2 fold-change < - 0.5 or > 0.5 and FDR < 0.1. c-g, Combined median log2 fold-change (log2FC) of 5 sgRNAs per targeted gene, showing sgRNA depletion over three cycles of cold exposure. Significance between Cycle 1 versus Cycle 3 is measured by two-way ANOVA adjusted for multiple comparisons by Dunnett’s test. Significance between 37°C and 4°C for each cycle is measured by two-way ANOVA adjusted for multiple comparisons by Bonferroni’s test. c, GPX4, d, EEFSEC, e, SECISBP2, f, PSTK, g, SEPSECS. h, Volcano plot showing median log2 fold-change in abundance of guides targeting the indicated genes after three cycles of 4°C cold exposure in the presence or absence of 1 µM of ferrostatin-1. Red dots indicates selectively required genes with a log2 fold-change < −0.5 or > 0.5 and FDR < 0.1. i, Heatmap of the median log2FC in abundance of guides targeting ferroptosis-related genes after three cycles of cold exposure (4°C) compared to 37°C control cultures with and without ferrostatin-1. j-n, Depletion of 5 sgRNAs per gene over three cycles of cold exposure and rewarming with and without ferrostatin-1 (1 uM) treatment. j, GPX4 (**P = 0.0047), k, EEFSEC (**P = 0.0059), l, SECISBP2 (ns, P = 0.5457), m, PSTK (*P = 0.0393), n, SEPSECS (**P = 0.0029) as measured by two-tailed t-test at Cycle 3. All values show mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Panel a created with BioRender.com/q71x739.

Endogenous GPX4 activity is limiting for K562 cell cold tolerance

a, Stable GPX4 knockout (KO) K562 cell lines exhibit reduced cold tolerance. Left: Western blotting of wild-type (WT) and individual GPX4 KO clones for GPX4 and β-actin loading control. Right: Viability of GPX4 KO lines is significantly lower than WT K562 cells as measured by trypan blue staining (n = 4, ****P < 0.0001), with complete cell death by one day at 4°C. Significance measured by two-tail t-test at 7 days 4°C. b, c, Treatment with the GPX4 inhibitors RSL3 (b) or ML162 (c) results in enhanced K562 cold-induced death by 4 days at 4°C (n = 4, ****P < 0.0001) as measured by two-tailed t-test. d, Cold- and RSL3-induced K562 cell death occurs via ferroptosis. K562 cells were placed at 4°C and treated with RSL3 (1 µM) and/or the ferroptosis inhibitor ferrostatin-1 (Fer-1, 1 µM) or iron chelator DFO (5 µM) for 4 days (n = 4). Treatment with RSL3 resulted in significantly lower cell viability than no treatment as determined by one-way ANOVA adjusted for multiple comparisons by Tukey’s HSD (***P = 0.0002). e, Reintroduction of wild-type human (hs) or hamster (ma) GPX4, but not catalytically dead forms of GPX4 (Gpx4 U46S), rescues cold-induced cell death in a K562 GPX4 KO clonal cell line (n = 4). Left: Western blot for GPX4 levels and β-actin loading control. Right: Expression of WT human GPX4 and hamster GPX4 resulted in significantly higher cell viability compared to the corresponding parental GPX4 KO, GFP-, and mutGPX4-expressing lines by trypan blue staining, as measured by one-way ANOVA adjusted for multiple comparisons by Dunnett’s test at 7 days 4°C (****P < 0.0001). f, Overexpression of wild-type human or hamster GPX4, but not catalytically dead forms of GPX4 (mutGPX4), suppresses cold-induced cell death in a K562 cells (n = 4). Left: Western blot for GPX4, with β-actin loading control. Right: Expression of WT human GPX4 and hamster GPX4 resulted in significantly higher cell viability compared to wild-type, GFP-, and mutGPX4-expressing K562 lines by trypan blue staining, as measured by one-way ANOVA adjusted for multiple comparisons by Dunnett’s test at 7 days 4°C (****P < 0.0001). All values show mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

GPX4 is required for cold tolerance across both several hibernator and non-hibernator mammalian species and cell types

a, Viability of hibernator cells (Syrian hamster embryonic fibroblasts, Greater horseshoe bat embryonic fibroblasts, and 13-lined ground squirrel embryonic fibroblasts) and non-hibernator cells (SV40-immortalized mouse embryonic fibroblasts, human adult kidney fibroblasts, human adult dermal fibroblasts, rat adult dermal fibroblasts, and mouse adult dermal fibroblasts) exposed to 4°C as measured by trypan blue staining (n = 4, ****P < 0.0001). b, Human kidney fibroblast GPX4 knockout cells exhibit reduced cold tolerance compared to WT cells. Left: Western blot of wild-type (WT) and GPX4 KO cells for GPX4 and β-actin loading control. Right: Viability of GPX4 KO cells is significantly lower than WT cells at 4°C as measured by trypan blue staining (n = 4, ****P < 0.0001). c, Gpx4 activity is essential for cold survival of primary human kidney fibroblasts. Human kidney fibroblasts were placed at 4°C and left untreated or treated with RSL3 (1 μM) and/or the ferroptosis inhibitor ferrostatin-1 (Fer-1, 1 μM) for 7 days (n = 4, ****P < 0.0001). d, Representative fluorescence images of human kidney fibroblasts after 4 days at 4°C with no treatment, 1 μM Fer-1, 1 μM RSL3, or 1 μM Fer-1 and 1 μM RSL3. Cultures were stained with Hoechst 33342 and propidium iodide (PI) to identify live cells. e, Gpx4 activity is essential for cold survival in hibernator cells (Syrian hamster embryonic fibroblasts, Greater horseshoe bat embryonic fibroblasts, and 13-lined ground squirrel embryonic fibroblasts) and non-hibernator cells (SV40-immortalized mouse embryonic fibroblasts, human adult kidney fibroblasts, human adult dermal fibroblasts, rat adult dermal fibroblasts, and mouse adult dermal fibroblasts). Cells were placed at 4°C and left untreated, treated with RSL3 (1 μM), and/or ferrostatin-1 (Fer-1, 1 μM) for 7 days (n = 4). f, Expanded Day 4 data from e) indicates that Gpx4 activity is essential for fibroblast survival in the cold across several hibernator and non-hibernator species. Cells were placed at 4°C and left untreated or treated with RSL3 (1 μM) and/or ferrostatin-1 (Fer-1, 1 μM) for 7 days (n = 4, ****P < 0.0001). g-h, Gpx4 activity is essential in g) mouse primary cortical neuron cultures and h) hamster primary cortical neuron cultures. Cells were placed at 4°C and left untreated, treated with RSL3 (1 μM) and/or the ferroptosis inhibitor ferrostatin-1 (Fer-1, 1 μM) for 1 or 4 days (n = 4). RSL3 treatment increased cell death, which was rescued by ferrostatin-1. All values show mean ± SEM, with significance measured by one-way ANOVA adjusted for multiple comparisons with Tukey’s HSD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Permeability to trypan blue changes rapidly upon cell rewarming

a, Number of viable K562 cells based on trypan blue staining after one day at 4°C and subsequent rewarming for 24 hours at 37°C. Numbers are normalized to initial cell counts. Dots indicate viable cell number based on trypan blue staining of cells after incubation at room temperature for 15, 30, or 60 minutes. Blue shaded regions indicate 4°C exposure and shaded pink regions indicate 37°C exposure. Red square indicates calculated cell counts after one day at 4°C based on the viable cell number measured after 24 hour rewarming. b, Viability of K562 cells was assessed by trypan blue staining after incubation at room temperature for 0, 15, 30, or 60 minutes following 24 hours at 4°C (n = 4). Cells incubated at room temperature for 15 minutes show a significant increase in cell counts compared to cells counted immediately (**P = 0.0016). Cells incubated at room temperature for 30 minutes show a significant increase in cell counts compared to a 15-minute incubation (**P = 0.0023), while no significant difference in viability was observed between cells incubated for 30 or 60 minutes (ns, P = 0.4059). c-f, Viability of cells was assessed by trypan blue staining after incubation at 37°C for 0, 15, 30, or 60 minutes following 24 hours at 4°C (n = 4). c, HEK293T, d, RPE1, e, HeLa, f, BHK-21. No significant difference in cell viability was observed between cells incubated for 30 or 60 minutes for HEK293T (ns, P = 0.3122), RPE1 (ns, P = 0.5137), HeLa (ns, P = 0.9735), and BHK-21 (ns, P = 0.9998) cells. All values show mean ± SEM, with significance determined by one-way ANOVA adjusted for multiple comparisons by Tukey’s HSD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Depletion of Core Essential Genes in Genome-Wide BHK-21 Screens

a-d, Genes ranked by median fold-change (log2) in genome-wide BHK-21 screens. a, after three cycles of cold exposure and rewarming (Cycle 3 4°C), b, matched constant 37°C control condition (Cycle 3 37°C), c, after 15 days of 4°C exposure (15 Days 4°C), d, matched constant 37°C control condition (15 Days 37°C). Core essential genes24 indicated in red are positioned below based on gene rank to demonstrate their depletion in each screen condition. e-f, Boxplots showing log2 fold change in representation for the population of control sgRNAs (gray; n = 250) or sgRNAs targeting core essential genes24 (blue; n = 4635) over e) three cycles of cold exposure and rewarming (4°C) or f) constant 37°C control conditions. The line within each box represents the median, the bounds of each box represent the first and third quartiles, and the whiskers extend to the furthest data point within 1.5 times the interquartile range. A two-sided Kolmogorov-Smirnov test was used to test the difference between each pair of control/essential-gene-targeting sgRNA distributions (estimated p-value < 2.2e-16 for all six pairs in e and f).

RSL3 treatment has no effect on the viability of cold-exposed Gpx4 KO BHK-21 cells

a-d, Wild-type BHK-21 cells (a) and three independent Gpx4 KO BHK-21 clonal lines (b-d) were treated with RSL3 (1 µM) and ferrostatin-1 (Fer-1, 1 µM) as indicated for 24 hours at 4°C (n = 4) prior to trypan blue staining. Wild-type BHK-21 cells show a significant decrease in cell viability when treated with RSL3 compared to no treatment (**P = 0.0013), whereas Gpx4 KO lines show no significant changes in viability (ns, P > 0.05). All values show mean ± SEM, with significance measured by one-way ANOVA adjusted for multiple comparisons by Dunnett’s test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Growth of BHK21 and K562 GPX4 knockout cells and effects of RSL3 treatment on human kidney fibroblast GPX4 knockout cells

a, Viable K562 cells recorded as percentage relative to start based on trypan blue staining over the course of 3 days at 37°C. Cell growth is significantly decreased in GPX4 K562 KO clones compared to WT K562 cells. Supplementation of liproxstatin-1 (2.5 µM) increases cell viability in GPX4 KO cells and not WT cells at 37°C. b, Viable BHK-21 cells recorded as percentage relative to start based on trypan blue staining over the course of 2 days. Cell growth is significantly decreased in Gpx4 BHK-21 KO cells compared to WT BHK-21 cells. Supplementation of liproxstatin-1 (2.5 µM) increases cell viability in Gpx4 KO cell and not WT cells at 37°C. c, Human kidney GPX4 KO cells were placed at 4°C and left untreated or treated with RSL3 (1 μM) for 4, and 7 days (n = 4). Treatment with RSL3 does not confer significant additional death (n=4 per timepoint and condition) as measure by two-tailed t-test. All values show mean ± SEM, with significance measured by one-way ANOVA adjusted for multiple comparisons with Tukey’s HSD unless otherwise specified. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Ferroptosis inhibitors and iron chelators increase cold cell viability in a dose-dependent manner

a-h, K562 (a-d) and HT1080 (e-h) cells were treated with varying concentrations of the ferroptosis inhibitors, ferrostatin-1 and liproxstatin-1, and iron chelators, deferoxamine and 2’2’-pyridine, for four days at 4°C prior to assaying cell viability by trypan blue staining (n = 3). All values show mean ± SEM.

Cells derived from non-hibernators undergo cold-induced ferroptotic cell death

a, K562 cells were treated with ferrostatin-1 (1 µM) (****P < 0.0001), liproxstatin-1 (1 µM) (****P < 0.0001), Z-VAD-FMK (1 µM) (ns, P = 0.9760), or necrostatin-1 (1 µM) (ns, P = 0.9835) for 4 days at 4°C prior to trypan blue staining (n = 4). b, HT1080 cells treated with ferrostatin-1 (1 µM) (****P < 0.0001), liproxstatin-1 (1 µM) (****P < 0.0001), Z-VAD-FMK (1 µM), or necrostatin-1 (1 µM) for 4 days at 4°C prior to trypan blue staining (n = 4). c, HeLa cells treated with ferrostatin-1 (1 µM) (****P < 0.0001), liproxstatin-1 (1 µM) (****P < 0.0001), Z-VAD-FMK (1 µM) (ns, P > 0.9999), or necrostatin-1 (1 µM) (ns, P = 0.9987) for 10 days at 4°C prior to trypan blue staining (n = 4). d, RPE1 cells treated with ferrostatin-1 (1 µM) (****P < 0.0001), liproxstatin-1 (1 µM) (****P < 0.0001), Z-VAD-FMK (1 µM) (ns, P = 0.9291), or necrostatin-1 (1 µM) (ns, P > 0.9999) for 10 days at 4°C prior to trypan blue staining (n = 4). e, K562 cells treated with deferoxamine mesylate (5 µM) (***P = 0.0002) or 2’2’-pyridine (10 µM) (****P < 0.0001) for 4 days at 4°C prior to trypan blue staining (n = 3). f, HT1080 cells treated with deferoxamine mesylate (100 µM) (****P < 0.0001) or 2’2’-pyridine (5 µM) (***P = 0.0002) for 4 days at 4°C prior to trypan blue staining (n = 3). g, HeLa cells treated with deferoxamine mesylate (5 µM) (****P < 0.0001) or 2’2’-pyridine (5 µM) (****P < 0.0001) for 10 days at 4°C prior to trypan blue staining (n = 3). h, RPE1 cells treated with deferoxamine mesylate (100 µM) (****P < 0.0001) or 2’2’-pyridine (5 µM) (***P < 0.0001) for 10 days at 4°C prior to trypan blue staining (n = 3). All values show mean ± SEM, with significance measured one-way ANOVA adjusted for multiple comparisons by Dunnett’s test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Depletion of Core Essential Genes in Genome-Wide K562 Screens

a-d, Genes ranked by median fold-change (log2) in genome-wide K562 screens. a, after three cycles of cold exposure (Cycle 3 4°C), b, matched constant 37°C control condition (Cycle 3 37°C), c, three cycles of cold exposure with 1 µM ferrostatin-1 (Cycle 3 4°C + Fer-1), d, matched constant 37°C control condition with 1 µM ferrostatin-1 (Cycle 3 37°C + Fer-1). Core essential genes24 (red bars) are positioned below based on gene rank to demonstrate their depletion in each screen condition. e-f, Boxplots showing the log2 fold change in representation for the population of control sgRNAs (gray; n = 500) or sgRNAs targeting core essential genes24 (blue; n = 3219) over e) three cycles of cold exposure (4°C) or f) constant 37°C control conditions. The line within each box represents the median, the bounds of each box represent the first and third quartiles, and the whiskers extend to the furthest data point within 1.5 times the interquartile range. A two-sided Kolmogorov-Smirnov test was used to test the difference between each pair of control/essential-gene-targeting sgRNA distributions (estimated p-value < 2.2e-16 for all six pairs in e and f).

Top 20 enriched pathways at Cycle 3 (4°C vs. 37°C) in K562 cells

a, Graphical representation of the top 20 enriched, differentially expressed gene sets (204 genes; FDR < 0.1) from the genome-scale CRISPR-Cas9 screen in K562 cells (Cycle 3 4°C vs. 3 passages at 37°C). GO_Biological Processes Panther overrepresentation test was used to determine enriched gene sets. Functional annotation analysis of the selectively required genes identified pathways related to translational readthrough, selenocysteine incorporation, protein insertion into the ER, glycosylation, fatty acid metabolism, and mitochondrial respiration. A full list of pathways is provided in Table S6.

RSL3 treatment has no effect on the viability of cold-exposed GPX4 KO K562 cells

a-d, Wild-type K562 cells (a) and three independent GPX4 KO K562 clonal lines (b-d) were treated with RSL3 (1 µM) and ferrostatin-1 (Fer-1, 1 µM) as indicated for 8 hours at 4°C (n = 4) prior to trypan blue staining. Wild-type K562 cells show a significant decrease in cell viability when treated with RSL3 compared to no treatment (*P = 0.0213), whereas GPX4 KO lines show no significant changes in viability (ns, P > 0.05). All values show mean ± SEM, with significance measured by one-way ANOVA adjusted for multiple comparisons by Dunnett’s test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Cell permeable glutathione and glutathione precursors do not increase cold cell viability a,

K562 cells were placed at 37°C and treated with cell permeable glutathione GSH-MEE (1 µM) and/or the glutathione synthesis inhibitor buthionine sulfoximine (BSO, 1 µM) for 1 day (n = 3). Treatment with BSO resulted in increased cell death (**P = 0.0018) that was rescued by GSH-MEE (ns, P = 0.9961). b, Treatment with GSH-MEE has no effect on K562 cold-induced death after 3 days at 4°C as measured by trypan blue staining (n = 3, 5 µM; ns, P = 0.1558). c, K562 cells were placed at 37°C and treated with ferroptosis inducer Erastin (10 µM) and N-acetylcysteine (NAC, 10 µM) for 1 day (n = 3). Treatment with Erastin resulted in increased cell death (***P = 0.0006) that was rescued by NAC (ns, P = 0.1091). d, Treatment with N-acetylcysteine does not increase cold cell viability in K562 cells after 2 days at 4°C as measured by trypan blue staining (n = 3, 10 µM; *P = 0.0352). e, Treatment with L-Cystine does not increase cold cell viability in K562 cells after 2 days at 4°C as measured by trypan blue staining (n = 3, 100 µM; ns, P = 0.9843). All values show mean ± SEM, with significance measured by one-way ANOVA adjusted for multiple comparisons by Dunnett’s test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Glutathione biosynthesis genes show increased depletion in cold-exposed K562 cells

a,c, Log2 fold-change (log2FC) of 10 guides per targeted gene in BHK-21 genome-wide screen, showing guide depletion over three cycles of cold exposure and rewarming. a, Gclc, c, Gss. b, d, Log2 fold-change (log2FC) of 5 guides per targeted gene in K562 genome-wide screen, showing significant guide depletion over three cycles of cold exposure. b, GCLC, d, GSS. Significance between Cycle 1 versus Cycle 3 is measured by two-way ANOVA adjusted for multiple comparisons by Dunnett’s test. Significance between 37°C and 4°C for each cycle is measured by two-way ANOVA adjusted for multiple comparisons by Bonferroni’s test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns P > 0.05.

Significant genes from Cycle 3 of genome-wide BHK-21 screen (median log2 fold-change < −1 or > 1 and FDR < 0.1)

Significant genes from 15 days 4°C of genome-wide BHK-21 screen (median log2 fold-change < −0.5 or > 0.5 and FDR < 0.1)

Significant genes from Cycle 3, 4°C + ferrostatin-1 vs 4°C of genome-wide K562 screen (median log2 fold-change < −0.5 or > 0.5 and FDR < 0.1)