Drug-Tolerant Persister Cancer Cells Are Vulnerable to GPX4 Inhibition Matthew J

letter doi:10.1038/nature24297

Drug-tolerant persister cancer cells are vulnerable to GPX4 inhibition Matthew J. Hangauer1,2,3, Vasanthi S. Viswanathan4, Matthew J. Ryan4, Dhruv Bole3, John K. Eaton4, Alexandre Matov5, Jacqueline Galeas3, Harshil D. Dhruv6, Michael E. Berens6, Stuart L. Schreiber4,7,8, Frank McCormick3 & Michael T. McManus1,2

Acquired drug resistance prevents cancer therapies from achieving hydroperoxidase GPX46. We included GPX4 inhibitors within a stable and complete responses1. Emerging evidence implicates a key diverse collection of compounds screened against BT474 parental and role for non-mutational drug resistance mechanisms underlying persister cells, and found the GPX4 inhibitors RSL3 and ML210 to be the survival of residual cancer ‘persister’ cells2–4. The persister cell among the compounds most selectively lethal to persister cells, with pool constitutes a reservoir from which drug-resistant tumours may minimal effect on parental cells or non-transformed MCF10A cells emerge. Targeting persister cells therefore presents a therapeutic (Fig. 1b, c and Extended Data Fig. 1c). Given the observed phenotypic opportunity to impede tumour relapse5. We previously found that and gene-expression similarities between persister cells from different cancer cells in a high mesenchymal therapy-resistant cell state are cancer types2–4,7, we wondered whether sensitivity to GPX4 inhibition dependent on the lipid hydroperoxidase GPX4 for survival6. Here might be a general persister cell vulnerability. Notably, we found that we show that a similar therapy-resistant cell state underlies the sensitivity to GPX4 inhibition extended beyond BT474 persister cells behaviour of persister cells derived from a wide range of cancers and drug treatments. Consequently, we demonstrate that persister cells acquire a dependency on GPX4. Loss of GPX4 function results in a Parental cells Persister cells selective persister cell ferroptotic death in vitro and prevents tumour 2 μM lapatinib relapse in mice. These findings suggest that targeting of GPX4 may represent a therapeutic strategy to prevent acquired drug resistance. We sought to identify therapeutically exploitable vulnerabilities in an 9 days experimental model of minimal residual disease, drug-tolerant persister cells. Persister cells, which are present across a wide range of tumour No types, survive cytotoxic exposure to targeted therapy or chemotherapy 37 days through poorly understood, reversible, non-mutational mechanisms2. lapatinib We focused our initial work on the HER2-amplified breast cancer line 2 μM BT474, which, upon treatment for nine or more days with cytotoxic lapatinib concentrations of lapatinib, reveals a small population of quiescent surviving persister cells (Fig. 1a). Removal of lapatinib allows the persister cells to regrow and re-acquire sensitivity to lapatinib. Subsequent 9 days lapatinib treatment re-derives persister cells. The reversibility of drug resistance in BT474 persister cells, which is also observed in other Rederived persister cells Regrown cells persister cell models2 including A375 melanoma (Extended Data b c Fig. 1a), is indicative of a non-mutational resistance mechanism. 1.2 GPX4 inhibitor (RSL3) To identify cellular processes that may represent selective depend- 1.2 1.0 GPX4 inhibitor (ML210) encies in persister cells, we performed RNA sequencing (RNA-seq) 1.0 0.8 on BT474 persister cells and drug naive parental cells (Extended 0.8 0.6 Data Tables 1, 2 and Supplementary Tables 1–4). In persister cells, we 0.6 observed upregulation of stemness markers CD133 and CD44 consistent 0.4 GPX4 inhibitor (ML210) 0.4 with earlier observations in other persister cell models2 (Extended 0.2 0.2 Fraction of surviving cells

Data Table 1). We also observed upregulation of mesenchymal mark- GPX4 inhibitor (RSL3) Fraction of surviving cells 0 0 ers and downregulation of epithelial markers (Extended Data Table 1). Chemical inhibitors Chemical inhibitors This encouraged us to explore further the role that mesenchymal state biology may play in persister cell therapy resistance. In addition, a Figure 1 | Small-molecule screen in drug-tolerant persister cells. broad downregulation of antioxidant gene-expression in persister cells, a, A small fraction of BT474 cells enter into a reversible, quiescent, including NRF2 target genes, directed our attention towards possible drug-tolerant persister state in response to 9 or more days of treatment with 2 μM lapatinib. Scale bars, 100 μm. b, Small-molecule inhibitor persister cell vulnerabilities in oxidative stress defence mechanisms screen in which pre-derived BT474 persister cells were treated with 1 μM (Extended Data Fig. 1b and Extended Data Table 2). inhibitors while maintained in 2 μM lapatinib. c, Small-molecule inhibitor We recently reported that cancer cells in a high mesenchymal therapy- counter-screen in parental BT474 cells. Images in a are representative of resistant state are selectively sensitive to ferroptosis, an oxidative two independent experiments. Data in b and c are from one biologically form of cell death that can be induced by inhibition of the lipid independent sample from a single screening experiment.

1Department of Microbiology and Immunology, University of California San Francisco, 513 Parnassus Avenue, San Francisco, California 94143, USA. 2UCSF Diabetes Center, University of California San Francisco, 513 Parnassus Avenue, San Francisco, California 94143, USA. 3UCSF Helen Diller Family Comprehensive Cancer Center, University of California San Francisco, 1450 3rd Street, San Francisco, California 94143, USA. 4Broad Institute, 415 Main Street, Cambridge, Massachusetts 02142, USA. 5DataSet Analysis LLC, 155 Jackson Street, San Francisco, California 94111, USA. 6Cancer and Cell Biology Division, The Translational Genomics Research Institute, 445 N 5th Street, Phoenix, Arizona 85004, USA. 7Howard Hughes Medical Institute, Chevy Chase, Maryland 20815, USA. 8Department of Chemistry and Chemical Biology, Harvard University, 12 Oxford Street, Cambridge, Massachusetts 02138, USA.

a Breast b Melanoma cdLung Ovarian

1.0 1.0 1.0 1.0

Parental 0.5 Persister 0.5 0.5 0.5 Fractional viability

(normalized to DMSO) 0 0 0 0 10 100 1,000 10 100 1,000 10 100 1,000 10 1001,000 [RSL3] (nM) [RSL3] (nM) [RSL3] (nM) [RSL3] (nM)

efBreast Melanoma ghLung Ovarian

1.0 1.0 1.0 1.0

0.5 0.5 0.5 0.5 Fractional viability

(normalized to DMSO) 0 0 0 0 10 100 1,000 10 100 1,000 10 100 1,000 10 1001,000 [ML210] (nM) [ML210] (nM) [ML210] (nM) [ML210] (nM)

ijWT GPX4 KO 1GPX4 KO 2 WT GPX4 KO 1GPX4 KO 2 NS **** **** NS ** *** 1.0 1.0 1.0 1.0 1.0 1.0 ostatin)

0.5 0.5 0.5 ostatin) 0.5 0.5 0.5 (normalize d (normalized to ferr to ferr Fractional viability 0 0 0 Fractional viability 0 0 0 2 μM ferrostatin-1 + ––+ + – 2 μM ferrostatin-1 + – + – + – Figure 2 | Persister cells are vulnerable to GPX4 inhibition. parental cells (j) after 3 days ferrostatin-1 withdrawal. Data are means and a–h, Fractional viability of breast (BT474) (a, e), melanoma (A375) s.d., and from three (a–h) or three (i, j) biologically independent samples. (b, f), lung (PC9) (c, g) and ovarian (Kuramochi) (d, h) cancer parental or *P < 0.05; ** P < 0.01; ** * P < 0.001; ** * * P < 0.0001; NS, not significant persister cells treated with GPX4 inhibitor RSL3 (a–d) or ML210 (e–h) for (P > 0.05); two-tailed t-tests. All data are representative of two separate 3 days. DMSO, dimethylsulfoxide. i, j, Fractional viability of A375 GPX4 experiments. wild-type (WT) or knockout (KO; clones 1 and 2) persister cells (i), and to a variety of other persister cell models (Fig. 2a–h and Extended Data Lipid hydroperoxides must be transported by the lipid transporter Fig. 1d). In each case, we found the parental cells to be far less sensitive SCP2 for cell death to occur13,14. Accordingly, we found that persister to GPX4 inhibitors (Fig. 2a–h). We also observed a less pronounced cells were protected from GPX4 inhibition by chemical inhibition of differential sensitivity of persister cells to the ferroptosis inducer SCP2 (Fig. 3d and Extended Data Fig. 2a, b). Also, in agreement with erastin, which acts upstream of GPX4 by depleting the GPX4 cofactor the non-apoptotic, caspase-independent nature of ferroptosis, we glutathione8 (Extended Data Fig. 1e, f). found that persister cell death from GPX4 inhibition is not rescued by We further confirmed the GPX4 dependency of persister cells the pan-caspase inhibitor Z-VAD-FMK (Fig. 3e and Extended Data using GPX4 knockout A375 melanoma cells (Fig. 2i, j, Extended Data Fig. 2a–c). These ferroptosis characteristics were also observed in A375 Fig. 1g and Supplementary Data). A hallmark of ferroptosis is the GPX4-knockout persister cells upon removal of ferrostatin-1 (Fig. 3f). accumulation of lipid hydroperoxides, and lipophilic, small-molecule Taken together, these observations indicate that GPX4 inhibition antioxidants have been shown to rescue cells from ferroptosis9–11. induces canonical ferroptosis in persister cells. Upon withdrawal of the ferroptosis-rescuing lipophilic antioxidant On the basis of our observation that antioxidant genes are down- ferrostatin-1, we observed that A375 GPX4-knockout, but not GPX4 regulated in persister cells (Extended Data Fig. 1b and Extended Data wild-type, persister cells undergo ferroptotic death6 (Fig. 2i). By Table 2), we hypothesized that there may be a disabled antioxidant contrast, the loss of GPX4 has minimal effect on parental A375 cell program that contributes to the sensitivity to ferroptosis. Glutathione viability (Fig. 2j). Furthermore, we found that GPX4 inhibitors are not and NADPH are two reducing cofactors that have been demonstrated synergistic with lapatinib treatment of parental BT474 cells, demon- to protect cells from the toxic effects of lipid peroxidation and prevent strating that GPX4 dependence is specific to the persister cell state ferroptosis10,15,16. We found that persister cells have markedly decreased (Extended Data Fig. 1h). In addition, although p53 can regulate ferrop- levels of both reducing cofactors (Fig. 4a, b and Extended Data Fig. 3a). tosis sensitivity12, we found that persister cells are dependent on GPX4 This is consistent with our observation that persister cells have down- regardless of p53 status (Fig. 2). These findings support the compelling regulated expression of glutathione and NADPH biosynthesis genes hypothesis that GPX4 is a selective dependency of a drug-resistant (Extended Data Table 2). In addition, replacement of glutathione via persister cell state that is shared by cancer cells from many lineages treatment with the glutathione metabolic precursor N-acetyl cysteine subjected to diverse therapeutic regimens. partially rescues persister cells from ferroptosis (Extended Data To test whether multiple persister cell types undergo canonical Fig. 3b). Persister cells have a specific sensitivity to lipid peroxidation, ferroptosis10 upon GPX4 inhibition, we treated persister cells with a rather than general sensitivity to oxidative stress, because persister cells series of ferroptosis rescue compounds. We found that co-treatment are not rescued from GPX4 inhibition by the antioxidants ebselen and with the lipophilic antioxidants ferrostatin-1 or liproxstatin-1 rescued EUK134 (Extended Data Fig. 3b); exhibit only modestly increased various persister cell types from GPX4 inhibition (Fig. 3a and Extended sensitivity to endogenous reactive oxygen species (ROS) induced by Data Fig. 2a–c). Ferroptosis is an iron-dependent process, and although DMNQ treatment (Extended Data Fig. 3c); and are relatively insensitive persister cells have less labile iron than parental cells (Extended Data to inhibition of an unrelated antioxidant enzyme, SOD1 (Extended Fig. 2d), we observed that iron is nonetheless essential for GPX4 inhibitor- Data Fig. 3d). These findings indicate that global downregulation of induced persister cell death, because iron chelation with deferoxamine NRF2 targets and other antioxidant genes, and decreased glutathione partially rescues persister cells (Fig. 3b). Consistent with the established and NADPH act to impair persister cell lipid peroxidation defence and role for lipoxygenase-catalysed lipid hydroperoxidation for ferroptosis, contribute to a dependence on GPX4. we observed that treatment with lipoxygenase inhibitors rescued per- Consistent with a disabled lipid peroxidation defence, we observed sister cells from GPX4 inhibition (Fig. 3c and Extended Data Fig. 2a, b). that after one hour of GPX4 inhibition and before any cell death, there

ab c ab** c 1.0 1.0 1.0 NS

1.0 1.0 1.0

0.5 0.5 0.5 0.5 0.5 *** 0.5 escence Fractional viability uor Relative DCF Fractional viability Fractional viability Relative [GSH] (normalized to DMSO) (normalized to DMSO) (normalized to DMSO)

0 0 0 Relative [NADPH] 0 0 0 – 1 μM RSL3 – +++ 1 μM RSL3 – + + 1 μM RSL3 +++ 1 μM RSL3 – + – + 1 μM RSL3 – + – + 2 μM ferrostatin-1 – ––100 μM DFO ––+ 1 μM PD146176 – – + – + Parental Persister ental Parental Persister 0.5 μM liproxstatin-1 – – – + 5 μM NDGA – – – + cells cells Par Persister cells cells

de f 1.5 d e 1.0 1.0 Pretreatment 2 weeks 2 months regrowth regrowth GPX4 WT ostatin) 1.0 1.5 GPX4 KO **** ** 1.0 1.0 0.5 0.5 *** 1.0 0.5 1.0 0.5 0.5 Fractional viability Fractional viability

Fractional viability 0.5 (normalized to DMSO ) (normalized to DMSO ) (normalized to ferr

0 0 0 Relative tumour volum e 0.7 1 M RSL3 – 1 M RSL3 – ++ 246810 12 14 Fractional viability μ + ++ μ 2 μM ferrostatin-1+– – – – (normalized to DMSO) 0 0 0 Days after ferrostatin 1 μM SCPI-2 – ––+ 50 μM ZVAD ––+ 1 μM PD146176 ––+ –– 1 μM RSL3 – + – + – + withdrawal 1 μM SCPI-4 – ––+ 1 μM SCPI-2 ––––+ – – – – 50 μM ZVAD + Figure 4 | Persister cells have a disabled antioxidant program and Figure 3 | GPX4 inhibition causes ferroptosis in persister cells. depend on GPX4 in vivo. a, b, Glutathione (GSH) (a) and NADPH (b) a–e, Fractional viability of BT474 persister cells treated with RSL3 and levels in BT474 cells treated with RSL3 for 1 h. c, ROS levels, as measured co-treated with lipophilic antioxidants ferrostatin-1 and liproxstatin-1 (a), by DCF staining, in BT474 cells treated with RSL3 for 1 h. d, Relapse iron chelator deferoxamine (DFO) (b), lipoxygenase inhibitors PD146176 of A375 GPX4 wild-type or knockout (clone 1) tumours. Mice bearing and NDGA (c), lipid transporter SCP2 inhibitors SCPI-2 and SCPI-4 GPX4 wild-type and knockout tumours on opposing flanks dosed with (d), or pan-caspase inhibitor Z-VAD-FMK (ZVAD) (e). f, Fractional ferrostatin-1 were treated with dabrafenib and trametinib to shrink viability of A375 GPX4 knockout (clone 1) persister cells with ferrostatin-1 tumours to their minimal size. Ferrostatin-1 was then withdrawn and replaced by ferroptosis rescue compounds or ZVAD. Data are means, and tumour relapse was monitored. See Source Data for individual data points. from two biologically independent samples. All data are representative of e, Fractional viability after 24-h RSL3 pretreatment of BT474 cells before two separate experiments. derivation of persister cells (left), RSL3 treatment of cells regrown from persister cells upon lapatinib removal for 15 days (middle), or 2 months is a marked increase in 2′7 ′-dichlorofluorescein (DCF) staining of ROS, (right). Data are means and s.d., from three (a, e left and middle) or two which indicates lipid peroxidation in persister cells but not parental cells (b, c, e right) biologically independent samples, and four animals (d). < < < < (Fig. 4c and Extended Data Fig. 3e). Upon lipid hydroperoxide accumu- *P 0.05; ** P 0.01; ** * P 0.001; ** * * P 0.0001; NS, not significant (P > 0.05); two-tailed t-tests. All data are representative of two separate lation, free reduced glutathione levels are expected to decrease, because experiments. reduced glutathione covalently ligates, via glutathione-S-transferase catalysed or non-enzymatic glutathionylation, to oxidized products of lipid peroxidation, such as aldehydes17. Consistent with Our data reveal that co-treatment of cancer cells with a targeted ther- this, we observed a decrease in levels of reduced glutathione specifically apy or chemotherapy agent together with a GPX4 inhibitor effectively in persister cells immediately after GPX4 inhibition that mirrored the reduces the residual persister cell pool. However, co-treatment with increase in DCF staining (Fig. 4a and Extended Data Fig. 3a). Notably, multiple drugs often results in increased toxic side effects that can com- a recent study found that persister cells are sensitive to the aldehyde promise patient safety and treatment efficacy. Interestingly, we found dehydrogenase inhibitor disulfiram3. Although disulfiram was shown that pretreatment of BT474 parental cells with RSL3 alone for 24 hours to cause apoptosis rather than ferroptosis, and is not rescued by decreases the number of persister cells remaining after subsequent ferrostatin-1 (Extended Data Fig. 3f), the persister cell sensitivity to lapatinib treatment (Fig. 4e). This finding is consistent with recent disulfiram may be more broadly reflective of the disabled antioxidant reports that demonstrate that persister cells may pre-exist before drug program. treatment3,18. We also observed that BT474 persister cells allowed to We reasoned that an effective GPX4 inhibitor could potentially be regrow after washout of lapatinib retain their full sensitivity to RSL3 for used in a clinical setting to prevent tumour relapse. To test this possibility at least two weeks but for less than two months (Fig. 4e and Extended in a preclinical model, we developed an A375 melanoma xenograft Data Fig. 3i), indicating that the impaired antioxidant program in model in which tumours respond and relapse upon treatment with persister cells is not immediately repaired upon drug removal and dabrafenib and trametinib. Because neither RSL3 nor ML210 are regrowth. These findings demonstrate that pre- or post-treatment with systemically bioavailable, we instead adopted a recently developed GPX4 inhibitors, rather than co-treatment, may be adequate to deplete genetic strategy6 to test GPX4 dependence of residual tumours in vivo. the pool of persister cells that survive targeted therapy or chemotherapy. We used GPX4-knockout or wild-type A375 cells (Extended Data This staggered treatment strategy may be clinically beneficial in maxi- Fig. 1g and Supplementary Data) to produce xenograft tumours mizing efficacy while minimizing the potential increase in toxicity that in NCr nude mice. We then treated the mice with dabrafenib and might be anticipated from simultaneous co-treatment with two agents. trametinib to shrink the tumours while concurrently dosing the mice with Non-mutationally drug-tolerant persister cells have recently gained ferrostatin-1 to mask any effects of GPX4 deletion before or during the attention for their potential role in acquired drug resistance in which initial response to drug treatment. Once tumours had been reduced to they may constitute a reservoir of surviving cells from which fully their minimal volume, ferrostatin-1 was withdrawn, unmasking the drug-resistant cells eventually emerge to cause tumour relapse2,4,5. GPX4-knockout effect in these residual tumours. Upon further dosing Here we have shown that GPX4 inhibition is a promising strategy to of mice with dabrafenib and trametinib, without ferrostatin-1, the selectively eradicate this residual cell reservoir (Extended Data Fig. 3j). GPX4 wild-type tumours relapsed and the GPX4-knockout tumours Although existing GPX4 inhibitors, such as RSL3 and ML210, are did not (Fig. 4d and Extended Data Fig. 3g). By contrast, parental A375 valuable tool compounds in cell culture settings, their poor pharma- GPX4-knockout and wild-type cells both formed tumours without cokinetic properties preclude their systemic use in vivo. Given the ferrostatin-1 equally well (Extended Data Fig. 3h). Taken together, these therapeutic promise for inducing ferroptosis in persister cells and A375 melanoma xenograft data are consistent with an essential role for drug-resistant cancers6,19, the development of a potent bioavailable GPX4 in residual A375 tumours in vivo. GPX4 inhibitor is an urgent priority. Because GPX4 genetic deletion

00 Month 2017 | VOL 000 | NATURE | 3 © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved. RESEARCH letter is lethal in adult mice13,20, further studies will be needed to determine 15. Shimada, K., Hayano, M., Pagano, N. C. & Stockwell, B. R. Cell-line selectivity improves the predictive power of pharmacogenomic analyses and helps whether a suitable therapeutic window exists for treatment with GPX4 identify NADPH as biomarker for ferroptosis sensitivity. Cell Chem. Biol. 23, inhibitors. In summary, we have found that GPX4 inhibition results in 225–235 (2016). ferroptotic death in cancer persister cells across many lineages, present- 16. Yang, W. S. et al. Regulation of ferroptotic cancer cell death by GPX4. Cell 156, 317–331 (2014). ing a novel strategy to prevent tumour relapse. 17. Guéraud, F. et al. Chemistry and biochemistry of lipid peroxidation products. Online Content Methods, along with any additional Extended Data display items and Free Radic. Res. 44, 1098–1124 (2010). Source Data, are available in the online version of the paper; references unique to 18. Shaffer, S. M.et al. Rare cell variability and drug-induced reprogramming as a these sections appear only in the online paper. mode of cancer drug resistance. Nature 546, 431–435 (2017). 19. Roh, J. L., Kim, E. H., Jang, H. J., Park, J. Y. & Shin, D. Induction of ferroptotic cell received 20 July 2016; accepted 19 September 2017. death for overcoming cisplatin resistance of head and neck cancer. Cancer Lett. 381, 96–103 (2016). Published online 1 November 2017. 20. Yoo, S. E. et al. Gpx4 ablation in adult mice results in a lethal phenotype accompanied by neuronal loss in brain. Free Radic. Biol. Med. 52, 1820–1827 1. Groenendijk, F. H. & Bernards, R. Drug resistance to targeted therapies: déjà vu (2012). all over again. Mol. Oncol. 8, 1067–1083 (2014). 2. Sharma, S. V. et al. A chromatin-mediated reversible drug-tolerant state in Supplementary Information is available in the online version of the paper. cancer cell subpopulations. Cell 141, 69–80 (2010). 3. Raha, D. et al. The cancer stem cell marker aldehyde dehydrogenase is Acknowledgements We acknowledge technical support from the UCSF required to maintain a drug-tolerant tumor cell subpopulation. Cancer Res. 74, ES Cell Targeting Core and UCSF Preclinical Therapeutics Core. This work 3579–3590 (2014). was supported by grants from the National Cancer Institute (NCI) of the 4. Hata, A. N. et al. Tumor cells can follow distinct evolutionary paths to become National Institutes of Health (NIH) (Cancer Target Discovery and Development resistant to epidermal growth factor receptor inhibition. Nat. Med. 22, 262–269 Network grant U01CA168370 to M.T.M. and F.M., U01CA217882 to M.T.M., (2016). U01CA176152 to S.L.S., U01CA168397 to M.E.B., and R01CA212767 to 5. Oxnard, G. R. The cellular origins of drug resistance in cancer. Nat. Med. 22, M.T.M.), Susan G. Komen for the Cure Postdoctoral Fellowship KG1101214 to 232–234 (2016). M.J.H., and the Howard Hughes Medical Institute (S.L.S.). 6. Viswanathan, V. S. et al. Dependency of a therapy-resistant state of cancer cells on a lipid peroxidase pathway. Nature 547, 453–457 (2017). Author Contributions M.T.M. and F.M. directed the project; M.J.H., V.S.V., S.L.S., 7. Fan, W. et al. MET-independent lung cancer cells evading EGFR kinase F.M. and M.T.M. wrote the manuscript; M.J.H. performed chemical screens, RNA- inhibitors are therapeutically susceptible to BH3 mimetic agents. Cancer Res. seq analysis, and all cell culture experiments except V.S.V. and M.J.R. performed 71, 4494–4505 (2011). Kuramochi cell GPX4 inhibitor assays, D.B. assisted with ROS, glutathione and 8. Yagoda, N. et al. RAS–RAF–MEK-dependent oxidative cell death involving NADPH assays, and J.G. performed western blots; M.J.H. directed CRISPR voltage-dependent anion channels. Nature 447, 865–869 (2007). editing of A375 cells and in vivo experiments that were performed at UCSF core 9. Yang, W. S. & Stockwell, B. R. Ferroptosis: death by lipid peroxidation. Trends facilities; V.S.V. and J.K.E. contributed reagents; A.M., H.D.D. and M.E.B. provided Cell Biol. 26, 165–176 (2016). advice and project support. 10. Dixon, S. J. et al. Ferroptosis: an iron-dependent form of nonapoptotic cell death. Cell 149, 1060–1072 (2012). Author Information Reprints and permissions information is available at 11. Skouta, R. et al. Ferrostatins inhibit oxidative lipid damage and cell death www.nature.com/reprints. The authors declare no competing financial in diverse disease models. J. Am. Chem. Soc. 136, 4551–4556 (2014). interests. Readers are welcome to comment on the online version of the 12. Jiang, L. et al. Ferroptosis as a p53-mediated activity during tumour paper. Publisher’s note: Springer Nature remains neutral with regard suppression. Nature 520, 57–62 (2015). to jurisdictional claims in published maps and institutional affiliations. 13. Friedmann Angeli, J. P. et al. Inactivation of the ferroptosis regulator Correspondence and requests for materials should be addressed to Gpx4 triggers acute renal failure in mice. Nat. Cell Biol. 16, 1180–1191 M.T.M. ([email protected]) or F.M. ([email protected]). (2014). 14. Kriska, T., Pilat, A., Schmitt, J. C. & Girotti, A. W. Sterol carrier protein-2 (SCP-2) Reviewer Information Nature thanks N. Chandel and P. Vandenabeele and involvement in cholesterol hydroperoxide cytotoxicity as revealed by SCP-2 the other anonymous reviewer(s) for their contribution to the peer review inhibitor effects.J. Lipid Res. 51, 3174–3184 (2010). of this work.

Methods HiSeq 2000 instrument. 145 million to 160 million single-end, 50-base-pair reads Cell culture. HER2-amplified breast cancer BT474 (ATCC), EGFR-mutant non- were obtained per replicate. small-cell lung cancer PC9 (Sigma-Aldrich) and Kuramochi ovarian cancer (JCRB) RNA-seq data analysis. Fastq sequence files were aligned to hg18 using TopHat cells were cultured in RPMI-1640 medium, and BRAF-mutant A375 melanoma v.2.0.6, allowing only uniquely mapped reads with the ‘-g 1’ option. Differential (ATCC) cells were cultured in DMEM, each supplemented with penicillin, strep- expression testing and assignment of FPKM and fold change values to genes was tomycin and 10% FBS. A375 GPX4-knockout cells were cultured in the presence performed with CuffDiff v.2.0.2, comparing the two BT474 parental cell replicates of 2 μM ferrostatin-1 unless otherwise noted. MCF10A (ATCC) cells were cultured to the three BT474 persister cell replicates. For CuffDiff, quartile normalization in Mammary Epithelial Cell Growth Medium supplemented with bovine pituitary and ‘-compatible-hits-norm’ options were used to assign reads to a GTF file con- extract, gentamycin, amphotericin B (Lonza MEGM BulletKit), and 100 ng ml−1 taining all human RefSeq NM and NR genes which was downloaded from the cholera toxin (Sigma). Each cell line was maintained in a 5% CO2 atmosphere at UCSC Genome Browser and converted into a CuffDiff readable format using 37 °C. Cell line identities were confirmed by STR fingerprinting and all were found Cuffcompare as described in the Cufflinks manual. to be negative for mycoplasma using the MycoAlert kit (Lonza). The set of genes with minimum fragments per kilobase of transcript per million Persister cell derivation and treatments. Persister cells were derived from treat- (FPKM) mapped reads greater than 1 in either parental cells or persister cells, ment of HER2-amplified breast cancer BT474, EGFR-mutant non-small-cell lung greater than twofold upregulation or downregulation in persister cells relative to cancer PC9 or Kuramochi ovarian cancer cells with 2 μM lapatinib, 2 μM erlotinib, parental cells, and significant false discovery rate corrected P value (q < 0.05) as or co-treatment with 10 μM carboplatin and 100 nM paclitaxel, respectively, for at calculated by CuffDiff were used for core analysis with Ingenuity Pathway Analysis least 9 days with fresh drug added every 3 days. A375 persister cells were derived (IPA) software to generate IPA results for canonical pathways, upstream regula- from 9 or more days of treatment with 2 μM vemurafenib (Fig. 2i), 25 μM vemu- tors, and functions (Supplementary Tables 2–4). Analysis of differential expression rafenib (Extended Data Fig. 1a), 1 μM dabrafenib and 100 nM trametinib (Fig. 3f and among genes within previously established panels of mesenchymal and epithelial Extended Data Fig. 2a, b), or 2 μM dabrafenib and 200 nM trametinib (Fig. 2b, f). marker genes21 (Extended Data Table 1) and human antioxidant genes22 (Extended A375 GPX4-knockout cells were maintained in medium with 2 μM ferrostatin-1 Data Fig. 1b and Extended Data Table 2) was also performed using FPKMs and during persister cell derivation with drug. Chemotherapy-derived BT474 persister fold changes calculated by CuffDiff. cells were similarly derived from co-treatment with 25 μM carboplatin and 1.25 nM Chemical inhibitor screens. BT474 cells were seeded in duplicate 12-well tissue cul- paclitaxel (Extended Data Fig. 1d). Unless otherwise noted, small-molecule inhibi ture plates and were allowed to reach 75% confluency. To derive persister cells, each tors were added to pre-derived persister cells in 12-well tissue culture plates that well was treated with 2 μM lapatinib for 9 days, with fresh media containing 2 μM were maintained under constant drug exposure throughout subsequent small- lapatinib added every 3 days. Small-molecule inhibitors at 1 μM concentration molecule inhibitor treatment. Cell viability was assessed using CellTiter Glo were then added in lapatinib-containing media and, after five days of incubation, (Promega) after 3 days of small-molecule treatment, or 3 days after ferrostatin-1 each well was measured using CellTiter Glo reagent. CellTiter Glo measurements removal in the case of A375 GPX4 knockout cell experiments, unless otherwise were made by transferring 200 μl of the CellTiter Glo solution per well into Costar noted. 96-well clear flat bottom plates for readout using a Promega GloMax Explorer For testing RSL3 pretreatment before persister cell derivation and RSL3 treat- luminometer. The CellTiter Glo luminescent signal was normalized to negative ment on cells regrown from persister cells (Fig. 4e), the following experimental control wells by calculating the ratio of luminescence signal in the test versus setup was used. In Fig. 4e (left), BT474 parental cells were pretreated for 24 h with negative control wells. Negative control wells contained media with 2 μM lapatinib 1 μM RSL3 alone, then RSL3 was removed and persister cells were subsequently but no additional chemical inhibitors. Each small-molecule inhibitor was also derived from two weeks of treatment with 2 μM lapatinib alone followed by a cell tested at 1 μM concentration in parental BT474 cells in 12-well plates, in which cells viability assay. For Fig. 4e (middle and right), pre-derived BT474 persister cells were seeded in duplicate wells at 25% confluency and treated with chemical inhi from 2 μM lapatinib were allowed to regrow upon removal of lapatinib for 15 days bitors. After incubation for 5 days, cell viability was measured with CellTiter Glo. (middle) or 63 days (right) and were then treated for 3 days with 1 μM RSL3 alone CRISPR-mediated GPX4 deletion. Guide RNAs targeting the first exon of human followed by a cell viability assay. GPX4 were designed using the CRISPR Design tool (F. Zhang laboratory, MIT). Chemicals. (Z)-4-hydroxytamoxifen, altretamine, baicalein, BWA4C, deferox- pX330 (F. Zhang laboratory, MIT) expression plasmids encoding the single-guide amine mesylate salt, erlotinib hydrochloride, ferrostatin-1, gossypol, HA14-1, RNAs 5′-CACGCCCGATACGCTGAGTG-3′ (for GPX4 knockout clone 1) hydroxychloroquine, liproxstatin-1, ML210, PBIT, salinomycin, trigonelline or 5′-CTTGGCGGAAAACTCGTGCA-3′ (for GPX4 knockout clone 2) and hydrochloride, tunicamycin, zileuton and Z-VAD-FMK were purchased from Streptococcus pyogenes Cas9 were generated and transfected into A375 cells in Sigma-Aldrich. ABT-263, ABT-737, dabrafenib, paclitaxel, canertinib, fluvastatin, the presence of 2 μM ferrostatin-1 using Lipofectamine 3000 (Life Technologies). GSK2110183, GX15-070, LBH589, (1S,3R)-RSL3 (RSL3), romidepsin, sara- The cells were subjected to two rounds of transfection. In the first round, the catinib, trametinib, trichostatin A, TW-37 and vorinostat were purchased from CRISPR plasmids were co-transfected with a vector encoding mCherry and puro- Selleck Chemical. Mifepristone, pladienolide B, SB-431542, vemurafenib and mycin resistance. Then, 24 h after the transfection, the cells were selected with XAV939 were purchased from Santa Cruz Biotechnology. Compound A (Mirk 1 μg ml−1 puromycin for 3 days. The selected cells were expanded and transfected inhibitor), erastin, MIM1 and SP600125 were purchased from EMD Millipore. again with the pX330 CRISPR constructs and a pMax-GFP reporter. 24 h after SCPI-2 and SCPI-4 were purchased from ChemBridge. Metformin and PD146176 the transfection, the cells were sorted and the GFP-positive, mCherry-negative were purchased from Enzo Life Sciences. AEW541 and JQ1 were purchased cells were retained. Cells were then transferred into 96-well plates at ~1 cell per from Cayman Chemical. WEHI-539 was purchased from MedChem Express. well and expanded as clones. Genomic DNA was extracted from all clones with Everolimus was purchased from Cell Signaling Technology. 17-DMAG was pur- the NucleoSpin Blood kit (Macherey-Nagel). The GPX4 target sites were ampli- chased from Biotang USA. Herboxidiene was purchased from Apollo Scientific. fied from 100 ng of genomic DNA with Phusion DNA Polymerase (NEB) with Sorafenib tosylate was purchased from AK Scientific. Spliceostatin A was pur- primers forward 5′-GTAAAACCGGACCAGAAGTACAAG-3′ and reverse chased from AdooQ Bioscience. Lapatinib ditosylate was purchased from LC 5′-CCCACCTGCTTCCCGAACTG-3 ′. Purified PCR products were TOPO cloned Laboratories. Nordihydroguaiaretic acid (NDGA) was purchased from Acros (Life Technologies) and sequenced with primer M13F (Quintara). 5 TOPO colonies Organics. Carboplatin was purchased from Hospira or Sigma. All chemicals were from each clone were sequenced to confirm frameshifts in GPX4. stored as stock solutions in DMSO (Sigma-Aldrich) or nuclease-free water (Life A375 GPX4 knockout clones 1 and 2 were confirmed to lack GPX4 protein by Technologies). western blot. For this, 3 million cells were seeded in 10-cm tissue culture plates and RNA-seq library preparation. To generate BT474 persister cells for RNA-seq, grown to 75% confluence, and the next day, cells were trypsinized, washed with BT474 cells were plated in five 15-cm tissue culture plates per replicate, for three PBS, and lysed using RIPA buffer (Thermo Scientific) supplemented with phos- replicates, and were allowed to reach 75% confluency before being treated with 2 μM phatase inhibitor (Sigma) and protease inhibitor (Sigma). Lysates were centrifuged lapatinib for nine days. Simultaneously, parental BT474 cells were cultured for at 10,000g at 4 °C for 12 min, and the protein concentration of the supernatant nine days in lapatinib-free medium in two replicate 15-cm tissue culture plates. was determined using the Bio-Rad Protein Assay. 40 μg of protein per sample was Approximately 2 million persister or parental cells were used per replicate for mixed with sample buffer (Bio-Rad) and denatured at 95 °C for 5 min. Samples RNA isolation. RNA was isolated using Trizol (Life Technologies), following the were separated by SDS–PAGE (NuPage 4–12% Bis-Tris Gel, Life Technologies), run manufacturer’s instructions. RNA samples were quantified using a NanoDrop with BioRad Precision Plus Protein Kaleidoscope ladder, and transferred to a nitro- (ThermoFisher), and RNA integrity was determined using a RNA 6000 Nano Kit cellulose membrane using an iBlot system (Life Technologies). Membranes were (Agilent) on an Agilent 2100 Bioanalyzer. All RNA samples had a measured RNA blocked with 5% BSA for 1 h at room temperature, and then incubated with either integrity number (RIN) greater than 9.3. RNA samples were sent to the UC Davis a 1:1,000 dilution of anti-GPX4 (rabbit) polyclonal antibody (Abcam, ab41787) Genome Center for non-strand specific polyA enrichment, Illumina library prepa- or 1:5,000 dilution of anti-β-actin (mouse) monoclonal antibody (Sigma, A5441) ration including multiplex barcodes and RNA-seq across six lanes of an Illumina at 4 °C overnight. LICOR secondary antibodies were then incubated with the

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved. RESEARCH letter membrane for 1 h at room temperature, and the membrane was imaged using the Protocol #AN142193. Tumour size criteria permitted a maximum volume of LICOR Odyssey Imaging System. 2,000 mm3, however, careful observation of the animals and body condition score Glutathione and NADPH measurements. NADP/NADPH-Glo and GSH-Glo of two or less superseded any laboratory measurement. These limits were not assay kits (Promega) were used to measure NADPH levels in 20,000 cells and exceeded in any experiments. For the formation of xenograft tumours without glutathione levels in 50,000 cells per replicate following manufacturer’s instructions. ferrostatin-1 (Extended Data Fig. 3h), 10 million A375 GPX4 wild-type or GPX4 To measure total glutathione (GSH + GSSG), oxidized glutathione (GSSG) was knockout (clone 1) cells were mixed in a 1:1 PBS:Matrigel mixture and injected into reduced by incubating test samples with 500 μM tris(2-carboxyethyl)phosphine the left or right flanks, respectively, of female athymic mice aged 10–12 weeks with (TCEP) for 10 min before luminescence measurement, following the manufac- a mean body weight of 24 g (NCr Nude, Taconic). Tumour volume measurements turer’s instructions. Each assay was read out using a Promega GloMax Explorer were made twice a week. All mice were included in the analysis. For the tumour luminometer. relapse experiments (Fig. 4d and Extended Data Fig. 3g), 10 million A375 GPX4 Intracellular reactive oxygen species measurements. BT474 cells plated in wild-type or knockout (clone 1) cells were mixed in a 1:1 PBS:Matrigel mixture replicate 10-cm tissue culture plates were treated with 2 μM lapatinib for at least containing 2.5 μM ferrostatin-1 and injected into left and right flanks, respectively, 9 days, with fresh media added every 3 days, to derive persister cells. For cells of male athymic NCr nude mice aged 10–12 weeks with a mean body weight re-grown from persister cells (Extended Data Fig. 3i), lapatinib was subsequently of 25 g. Mice were dosed daily intraperitoneally with 0.2 mg kg−1 ferrostatin-1 removed from persister cells and fresh lapatinib-free media was replaced every while tumours were initially established and during subsequent treatment with 3 days for 15 days. Then, for both the persister cell experiments (Fig. 4c and 100 mg kg−1 dabrafenib and 1 mg kg−1 trametinib twice daily by oral gavage. Once Extended Data Fig. 3e) and regrown cell experiments (Extended Data Fig. 3i), tumours had shrunk and tumour volumes had stabilized, ferrostatin-1 treatment cells were treated with 1 μM RSL3 or ML210 for 1 h. RSL3 or ML210 was removed was ceased while dabrafenib and trametinib treatment was continued and tumours and cells were trypsinized. For each replicate, 100,000 cells were pelleted and cell relapsed. All mice for which any tumour relapsed were included in the analysis. pellets were resuspended in full media containing 25 μM 6-carboxy-2′,7 ′-dichlo - Statistical analyses. No statistical methods were used to predetermine sample size. rodihydrofluorescein diacetate, di(acetoxymethyl ester) (carboxy-DCFDA, For mouse experiments, the mice were not randomized. The investigators perform- acetoxymethyl ester) (Thermo Fisher Scientific) in HBSS with 1% FBS. After ing tumour volume measurements were blinded. Statistical tests were performed 30 min of incubation with DCF dye, cells were pelleted again, resuspended with GraphPad Prism v.7.0a. Unless otherwise noted, P values were calculated in HBSS with 1% FBS and green fluorescence was analysed with a SONY using unpaired, two-tailed t-tests assuming unequal variance. Graphing of data SH800S Cell Sorter. Because HER2-bound lapatinib is green fluorescent23, was performed with GraphPad Prism v.7.0a and R v.3.2.1. data for persister cells were normalized to unstained persister cells, and parental Data availability. All data are available from the corresponding authors upon cell data were normalized to unstained parental cells. A minimum of 3,000 cells reasonable request. RNA-seq data have been deposited in the NCBI Genome were analysed for each persister and parental replicate. Data were processed Expression Omnibus (GEO) under the accession code GSE84896. with FloJo software. Iron measurements. Total labile iron (Fe(ii) and Fe(iii) combined) was meas- 21. Salt, M. B., Bandyopadhyay, S. & McCormick, F. Epithelial-to-mesenchymal ured using the colourimetric Iron Assay Kit (Abcam) following the manufacturer’s transition rewires the molecular path to PI3K-dependent proliferation. Cancer instructions. In total 1.4 million BT474 persister or parental cells were tested per Discov. 4, 186–199 (2014). 22. Gelain, D. P. et al. A systematic review of human antioxidant genes. Front. Biosci. replicate for two biological replicate samples. Colourimetric signal was read out 14, 4457–4463 (2009). using a BioTek Synergy 2 Microplate Reader. 23. Wilson, J. N., Liu, W., Brown, A. S. & Landgraf, R. Binding-induced, turn-on Xenograft experiments. The University of California San Francisco Institutional fluorescence of the EGFR/ERBB kinase inhibitor, lapatinib.Org. Biomol. Chem. Animal Care and Use Committee (IACUC) approved the xenograft studies in 13, 5006–5011 (2015).

Extended Data Figure 1 | Additional data demonstrating persister cell parental cells (f) were treated with erastin for 5 days. g, Western blot ferroptosis sensitivity. a, A375 melanoma persister cells are reversibly demonstrating GPX4 knockout in two distinct A375 clones (clone 1, drug-resistant. Scale bars, 400 μm. b, Global antioxidant gene expression KO1; clone 2, KO2). For gel source data, see Supplementary Fig. 1. is downregulated in BT474 persister cells. P value calculated using a two- h, BT474 parental cells co-treated with 2 μM lapatinib and RSL3 or ML210 tailed Wilcoxon signed-rank test. c, d, MCF10A parental cells (c) and for 3 days. Data are means and s.d. from 3 (c) or 2 (d–h) biologically BT474 persister cells derived from carboplatin and paclitaxel (d) were independent samples. c, NS, not significant (P > 0.05); two-tailed t-test. treated with RSL3 or ML210 for 3 days. e, f, BT474 persister cells (e) or All data are representative of two separate experiments.

Extended Data Figure 2 | Additional data demonstrating that GPX4 ML210 and ferroptosis rescue compounds for 3 days. d, Relative inhibition causes ferroptosis in persister cells. a, b, A375 persister concentration of total labile iron in BT474 parental and persister cells. cells were treated with RSL3 (a) or ML210 (b) and ferroptosis rescue Data are means and s.d. from 2 biologically independent samples. All data compounds for 3 days. c, PC9 persister cells were treated with RSL3 or are representative of two separate experiments.

Extended Data Figure 3 | Additional data demonstrating that Fig. 4d. Ferrostatin-1 was withdrawn on day 10. See Source Data for persister cells have a disabled antioxidant program and depend on individual data points. h, Untreated A375 GPX4 wild-type or GPX4 GPX4 in vivo. a, Reduced glutathione (GSH) and reduced plus oxidized knockout clone 1 tumour formation without ferrostatin-1 dosing. See glutathione (GSH + GSSG) levels in BT474 cells. b–d, Fractional viability Source Data for individual data points. i, ROS levels (DCF staining) in of BT474 persister cells treated with RSL3 and antioxidant compounds cells regrown without lapatinib for 15 days from BT474 persister cells and (b), endogenous ROS-generating compound DMNQ (c), or the SOD1 then treated with 1 μM RSL3 for 1 h. j, Persister cell GPX4 dependence inhibitor LCS-1 (d) for 3 days. e, ROS levels (DCF staining) in BT474 cells model. Data are means and s.d. from three (a, e, i) or two (b–d, f) treated with ML210 for 1 h. f, BT474 persister cells were co-treated with biologically independent samples or four (g) or five (h) animals. *P < 0.05; ALDH inhibitor disulfiram and ferrostatin-1 for 3 days. g, Tumour volume ** P < 0.01; ** * P < 0.001; ** * * P < 0.0001; NS, not significant (P > 0.05); measurements for the full time course of the experiment presented in two-tailed t-tests. All data are representative of two separate experiments.

Extended Data Table 1 | Expression of epithelial–mesenchymal transition and stem markers

RNA-seq data for mesenchymal, epithelial and stemness marker genes in BT474 parental and persister cells.

Extended Data Table 2 | Expression of NRF2 target genes

RNA-seq data for NRF2 target genes in BT474 parental and persister cells.

Corresponding author(s): Michael T. McManus Initial submission Revised version Final submission Life Sciences Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form is intended for publication with all accepted life science papers and provides structure for consistency and transparency in reporting. Every life science submission will use this form; some list items might not apply to an individual manuscript, but all fields must be completed for clarity. For further information on the points included in this form, see Reporting Life Sciences Research. For further information on Nature Research policies, including our data availability policy, see Authors & Referees and the Editorial Policy Checklist.

` Experimental design 1. Sample size Describe how sample size was determined. For mouse experiments, no statistical methods were used to predetermine sample size. 2. Data exclusions Describe any data exclusions. No data were excluded. 3. Replication Describe whether the experimental findings were All experiments were performed in two or three biological replicates, and reliably reproduced. independently reproduced as indicated in figure legends. 4. Randomization Describe how samples/organisms/participants were For mouse experiments, the mice were not randomized. allocated into experimental groups. 5. Blinding Describe whether the investigators were blinded to For mouse experiments, the investigators performing tumour volume group allocation during data collection and/or analysis. measurements were blinded. Note: all studies involving animals and/or human research participants must disclose whether blinding and randomization were used.

6. Statistical parameters For all figures and tables that use statistical methods, confirm that the following items are present in relevant figure legends (or in the Methods section if additional space is needed). n/a Confirmed

The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement (animals, litters, cultures, etc.) A description of how samples were collected, noting whether measurements were taken from distinct samples or whether the same sample was measured repeatedly A statement indicating how many times each experiment was replicated The statistical test(s) used and whether they are one- or two-sided (note: only common tests should be described solely by name; more complex techniques should be described in the Methods section) A description of any assumptions or corrections, such as an adjustment for multiple comparisons The test results (e.g. P values) given as exact values whenever possible and with confidence intervals noted A clear description of statistics including central tendency (e.g. median, mean) and variation (e.g. standard deviation, interquartile range)

Clearly defined error bars June 2017

See the web collection on statistics for biologists for further resources and guidance.

1 ` Software nature research | life sciences reporting summary Policy information about availability of computer code 7. Software Describe the software used to analyze the data in this Statistical analyses for all experiments were performed using GraphPad Prism 7.0a, study. except RNA-seq analysis which was performed using TopHat v2.0.6, CuffDiff v2.0.2, and pathway analysis with Ingenuity Pathway Analysis software as described in the Methods section.

For manuscripts utilizing custom algorithms or software that are central to the paper but not yet described in the published literature, software must be made available to editors and reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). Nature Methods guidance for providing algorithms and software for publication provides further information on this topic.

` Materials and reagents Policy information about availability of materials 8. Materials availability Indicate whether there are restrictions on availability of All materials used are commercially available except CRISPR-modified cell lines unique materials or if these materials are only available which are all available upon request. for distribution by a for-profit company. 9. Antibodies Describe the antibodies used and how they were validated anti-GPX4 (rabbit) polyclonal Ab (Abcam, ab41787) and anti-β-Actin (mouse) mAb for use in the system under study (i.e. assay and species). (Sigma, A5441) were validated in human cells by the manufacturers. 10. Eukaryotic cell lines a. State the source of each eukaryotic cell line used. BT474 (ATCC), PC9 (Sigma-Aldrich), Kuramochi (JCRB), A375 (ATCC) and MCF10A (ATCC).

b. Describe the method of cell line authentication used. STR fingerprint analysis

c. Report whether the cell lines were tested for All lines were tested and found to be negative for mycoplasma with MycoAlert mycoplasma contamination. (Lonza).

d. If any of the cell lines used are listed in the database None are listed in this database. of commonly misidentified cell lines maintained by ICLAC, provide a scientific rationale for their use.

` Animals and human research participants Policy information about studies involving animals; when reporting animal research, follow the ARRIVE guidelines 11. Description of research animals Provide details on animals and/or animal-derived Athymic female mice aged 10-12 weeks of average body weight 24 grams were materials used in the study. used for the experiment in ED Fig 3h, and male mice aged of average body weight 25 grams were used in the experiment in Fig 4d/ED Fig 3g.

Policy information about studies involving human research participants 12. Description of human research participants Describe the covariate-relevant population Not applicable. characteristics of the human research participants. June 2017