Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Letter
  • Published:

RETRACTED ARTICLE

Fatty acids and cancer-amplified ZDHHC19 promote STAT3 activation through S-palmitoylation

A Retraction to this article was published on 17 June 2020

This article has been updated

Abstract

Signal transducer and activator of transcription 3 (STAT3) has a critical role in regulating cell fate, inflammation and immunity1,2. Cytokines and growth factors activate STAT3 through kinase-mediated tyrosine phosphorylation and dimerization3,4. It remains unknown whether other factors promote STAT3 activation through different mechanisms. Here we show that STAT3 is post-translationally S-palmitoylated at the SRC homology 2 (SH2) domain, which promotes the dimerization and transcriptional activation of STAT3. Fatty acids can directly activate STAT3 by enhancing its palmitoylation, in synergy with cytokine stimulation. We further identified ZDHHC19 as a palmitoyl acyltransferase that regulates STAT3. Cytokine stimulation increases STAT3 palmitoylation by promoting the association between ZDHHC19 and STAT3, which is mediated by the SH3 domain of GRB2. Silencing ZDHHC19 blocks STAT3 palmitoylation and dimerization, and impairs the cytokine- and fatty-acid-induced activation of STAT3. ZDHHC19 is frequently amplified in multiple human cancers, including in 39% of lung squamous cell carcinomas. High levels of ZDHHC19 correlate with high levels of nuclear STAT3 in patient samples. In addition, knockout of ZDHHC19 in lung squamous cell carcinoma cells significantly blocks STAT3 activity, and inhibits the fatty-acid-induced formation of tumour spheres as well as tumorigenesis induced by high-fat diets in an in vivo mouse model. Our studies reveal that fatty-acid- and ZDHHC19-mediated palmitoylation are signals that regulate STAT3, which provides evidence linking the deregulation of palmitoylation to inflammation and cancer.

This is a preview of subscription content, access via your institution

Access options

Buy this article

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: STAT3 is S-palmitoylated at evolutionarily conserved cysteine residues.
Fig. 2: A signalling relay involving STAT3 phosphorylation and palmitoylation promotes STAT3 dimerization in response to cytokine and fatty acids.
Fig. 3: ZDHHC19 mediates STAT3 palmitoylation.
Fig. 4: ZDHHC19 is amplified in LSCC, and promotes tumorigenesis in vitro and in vivo.

Similar content being viewed by others

Data availability

The data supporting the findings of this study are available within the paper. Uncropped raw images from western blots are shown in Supplementary Fig. 1. Source Data for all graphs are available online. Tumour sample information is shown in Supplementary Tables 1, 2. All other data are available from the corresponding author upon reasonable request.

Change history

  • 17 June 2020

    An amendment to this paper has been published and can be accessed via a link at the top of the paper.

References

  1. Yu, H., Lee, H., Herrmann, A., Buettner, R. & Jove, R. Revisiting STAT3 signalling in cancer: new and unexpected biological functions. Nat. Rev. Cancer 14, 736–746 (2014).

    Article  CAS  Google Scholar 

  2. Mertens, C. & Darnell, J. E. Jr. SnapShot: JAK–STAT signaling. Cell 131, 612 (2007).

    Article  CAS  Google Scholar 

  3. Yu, C. L. et al. Enhanced DNA-binding activity of a Stat3-related protein in cells transformed by the Src oncoprotein. Science 269, 81–83 (1995).

    Article  ADS  CAS  Google Scholar 

  4. Aaronson, D. S. & Horvath, C. M. A road map for those who don’t know JAK–STAT. Science 296, 1653–1655 (2002).

    Article  ADS  CAS  Google Scholar 

  5. Chen, B., Sun, Y., Niu, J., Jarugumilli, G. K. & Wu, X. Protein lipidation in cell signaling and diseases: function, regulation, and therapeutic opportunities. Cell Chem. Biol. 25, 817–831 (2018).

    Article  CAS  Google Scholar 

  6. Resh, M. D. Palmitoylation of proteins in cancer. Biochem. Soc. Trans. 45, 409–416 (2017).

    Article  CAS  Google Scholar 

  7. Zheng, B. et al. 2-Bromopalmitate analogues as activity-based probes to explore palmitoyl acyltransferases. J. Am. Chem. Soc. 135, 7082–7085 (2013).

    Article  CAS  Google Scholar 

  8. Martin, B. R. & Cravatt, B. F. Large-scale profiling of protein palmitoylation in mammalian cells. Nat. Methods 6, 135–138 (2009).

    Article  CAS  Google Scholar 

  9. Ren, W., Jhala, U. S. & Du, K. Proteomic analysis of protein palmitoylation in adipocytes. Adipocyte 2, 17–27 (2013).

    Article  CAS  Google Scholar 

  10. Resh, M. D. Trafficking and signaling by fatty-acylated and prenylated proteins. Nat. Chem. Biol. 2, 584–590 (2006).

    Article  CAS  Google Scholar 

  11. Chamberlain, L. H. & Shipston, M. J. The physiology of protein S-acylation. Physiol. Rev. 95, 341–376 (2015).

    Article  CAS  Google Scholar 

  12. Yuan, Z. L., Guan, Y. J., Chatterjee, D. & Chin, Y. E. Stat3 dimerization regulated by reversible acetylation of a single lysine residue. Science 307, 269–273 (2005).

    Article  ADS  CAS  Google Scholar 

  13. Park, M. J. et al. SH2 domains serve as lipid-binding modules for pTyr-signaling proteins. Mol. Cell 62, 7–20 (2016).

    Article  CAS  Google Scholar 

  14. Herbert, D. et al. High-fat diet exacerbates early psoriatic skin inflammation independent of obesity: saturated fatty acids as key players. J. Invest. Dermatol. 138, 1999–2009 (2018).

    Article  CAS  Google Scholar 

  15. Hayashi, T. et al. High-fat diet-induced inflammation accelerates prostate cancer growth via IL6 signaling. Clin. Can. Res. 24, 4309–4318 (2018).

    Article  CAS  Google Scholar 

  16. Park, E. J. et al. Dietary and genetic obesity promote liver inflammation and tumorigenesis by enhancing IL-6 and TNF expression. Cell 140, 197–208 (2010).

    Article  CAS  Google Scholar 

  17. Fukata, Y., Iwanaga, T. & Fukata, M. Systematic screening for palmitoyl transferase activity of the DHHC protein family in mammalian cells. Methods 40, 177–182 (2006).

    Article  CAS  Google Scholar 

  18. Rana, M. S. et al. Fatty acyl recognition and transfer by an integral membrane S-acyltransferase. Science 359, eaao6326 (2018).

    Article  Google Scholar 

  19. Giordano, V. et al. Shc mediates IL-6 signaling by interacting with gp130 and Jak2 kinase. J. Immunol. 158, 4097–4103 (1997).

    CAS  PubMed  Google Scholar 

  20. Lowenstein, E. J. et al. The SH2 and SH3 domain-containing protein GRB2 links receptor tyrosine kinases to ras signaling. Cell 70, 431–442 (1992).

    Article  CAS  Google Scholar 

  21. Swarthout, J. T. et al. DHHC9 and GCP16 constitute a human protein fatty acyltransferase with specificity for H- and N-Ras. J. Biol. Chem. 280, 31141–31148 (2005).

    Article  CAS  Google Scholar 

  22. Wang, J. et al. Integrative genomics analysis identifies candidate drivers at 3q26-29 amplicon in squamous cell carcinoma of the lung. Clin. Cancer Res. 19, 5580–5590 (2013).

    Article  CAS  Google Scholar 

  23. Bromberg, J. F. et al. Stat3 as an oncogene. Cell 98, 295–303 (1999).

    Article  CAS  Google Scholar 

  24. Justilien, V. et al. The PRKCI and SOX2 oncogenes are coamplified and cooperate to activate Hedgehog signaling in lung squamous cell carcinoma. Cancer Cell 25, 139–151 (2014).

    Article  CAS  Google Scholar 

  25. Louie, S. M., Roberts, L. S. & Nomura, D. K. Mechanisms linking obesity and cancer. Biochim. Biophys. Acta 1831, 1499–1508 (2013).

    Article  CAS  Google Scholar 

  26. Yang, J. J. et al. Dietary fat intake and lung cancer risk: a pooled analysis. J. Clin. Oncol. 35, 3055–3064 (2017).

    Article  CAS  Google Scholar 

  27. Beyaz, S. et al. High-fat diet enhances stemness and tumorigenicity of intestinal progenitors. Nature 531, 53–58 (2016).

    Article  ADS  CAS  Google Scholar 

  28. Pascual, G. et al. Targeting metastasis-initiating cells through the fatty acid receptor CD36. Nature 541, 41–45 (2017).

    Article  ADS  CAS  Google Scholar 

  29. Tian, H. et al. Systematic siRNA screen unmasks NSCLC growth dependence by palmitoyltransferase DHHC5. Mol. Cancer Res. 13, 784–794 (2015).

    Article  CAS  Google Scholar 

  30. Yang, C. H., Yue, J., Fan, M. & Pfeffer, L. M. IFN induces miR-21 through a signal transducer and activator of transcription 3-dependent pathway as a suppressive negative feedback on IFN-induced apoptosis. Cancer Res. 70, 8108–8116 (2010).

    Article  CAS  Google Scholar 

  31. Niu, J. et al. USP10 inhibits genotoxic NF-κB activation by MCPIP1-facilitated deubiquitination of NEMO. EMBO J. 32, 3206–3219 (2013).

    Article  CAS  Google Scholar 

  32. Nelson, E. A. et al. Nifuroxazide inhibits survival of multiple myeloma cells by directly inhibiting STAT3. Blood 112, 5095–5102 (2008).

    Article  CAS  Google Scholar 

  33. Percher, A. et al. Mass-tag labeling reveals site-specific and endogenous levels of protein S-fatty acylation. Proc. Natl Acad. Sci. USA 113, 4302–4307 (2016).

    Article  ADS  CAS  Google Scholar 

  34. Braakman, I. & Hebert, D. N. Analysis of disulfide bond formation. Curr. Protoc. Protein Sci. 90, 14.1.1–14.1.21 (2017).

    Article  Google Scholar 

  35. Thomas, M. et al. Full deacylation of polyethylenimine dramatically boosts its gene delivery efficiency and specificity to mouse lung. Proc. Natl Acad. Sci. USA 102, 5679–5684 (2005).

    Article  ADS  CAS  Google Scholar 

  36. Niu, J. et al. Induction of miRNA-181a by genotoxic treatments promotes chemotherapeutic resistance and metastasis in breast cancer. Oncogene 35, 1302–1313 (2016).

    Article  CAS  Google Scholar 

  37. Hu, Y. & Smyth, G. K. ELDA: extreme limiting dilution analysis for comparing depleted and enriched populations in stem cell and other assays. J. Immunol. Methods 347, 70–78 (2009).

    Article  CAS  Google Scholar 

  38. Jafari, R. et al. The cellular thermal shift assay for evaluating drug target interactions in cells. Nat. Protoc. 9, 2100–2122 (2014).

    Article  CAS  Google Scholar 

Download references

Acknowledgements

This work was supported by Samuel M. Fisher Memorial–Melanoma Research Alliance (MRA) Established Investigator Award, the Idea Award from Prostate Cancer Research Program, US Department of Defense (W81XWH-17-1-0361) and grants from National Institutes of Health (R01CA181537, R01DK107651-01 and R01CA238270-01 to X.W., and R01CA160979 to D.A.F.). We thank T. Maniatis for the expression vector of STAT2, M. Fukata for the expression vectors of the ZDHHC proteins, J. Hersch for commenting and editing of the manuscript, the Confocal Imaging Core at Cutaneous Biology Research Center of Massachusetts General Hospital with the Shared Instrumentation Grant (1S10RR027673-01), and the Taplin Mass Spectrometry Core at Harvard Medical School for proteomic studies.

Reviewer information

Nature thanks Trever G. Bivona, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Author information

Authors and Affiliations

Authors

Contributions

X.W. conceived the concepts, designed the experiments and supervised the studies. J.N. designed and performed STAT3 palmitoylation, dimerization, ZDHHC-related biochemistry and cell biology experiments with the help of B.C., Y.S. and S.R.W. J.N. and Y.S. performed cancer cell biology and tumour-sphere experiments. B.C. performed the palmitoylation assays, confocal imaging and RAS palmitoylation experiments. Y.S. and J.N. designed and performed in vivo mouse experiments. Y.S. performed STAT3 disulfide assays, cancer stem-cell analysis, bioinformatics and immunohistochemical analysis of the LSCC tissue microarray. B.Z. and G.K.J. synthesized the chemical probes. B.Z. identified STAT proteins from mass spectrometry studies. A.N.H. contributed to the LSCC patient-derived xenograft model. M.M.-K. contributed to pathology studies of samples from patients with LSCC. D.A.F. and S.R.W. contributed to experimental design and studies of STAT3 signalling. STAT3 palmitoylation has been independently reproduced by J.N., Y.S. and B.C. multiple times. J.N., Y.S., B.C. and X.W. analysed the data. J.N., S.Y, B.C. and X.W. wrote the manuscript with input from all co-authors.

Corresponding author

Correspondence to Xu Wu.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 STAT3 is S-palmitoylated at evolutionarily conserved cysteine residues and mutation of STAT3 palmitoylation sites does not affect disulfide formation and protein stability.

a, HEK293A cells were incubated for 4 h with DMSO or 50 μM clickable probes. Cell lysates were reacted with biotin azide for enrichment of labelled proteins with streptavidin beads and identified by mass spectrometry. The peptide spectral counts are shown in the table. b, HEK293A cells were labelled with 50 μM chemical probe (alk-C16) for 4 h. Cell lysates were reacted with biotin azide and precipitated with streptavidin beads, and subjected to western blot using anti-STAT3 antibody. Endogenous STAT3 palmitoylation was analysed by western blot. c, HEK293A cells were transfected with MYC–STAT3 wild type and labelled with 50 μM chemical probe (alk-C16) for 4 h. Cell lysates were subjected to streptavidin blot, after anti-MYC immunoprecipitation and subsequent click reaction. d, HEK293A cells were transfected with empty vector or GFP–STAT1 and labelled with 50 μM chemical probe (alk-C16) for 4 h. After click reaction, cell lysates were subjected to streptavidin blot, showing detection of STAT1. e, Analysis similar to that in c was performed in HEK293A cells transfected with Flag–STAT1α and Flag–STAT1β. f, HEK293A cells were transfected with Flag–STAT3 or Flag–STAT5 and labelled with 50 μM chemical probe (alk-C16) for 4 h. Cell lysates were reacted with biotin azide and precipitated with streptavidin beads, and subjected to western blot using anti-Flag antibody. g, h, Analysis similar to that in e was performed in HEK293A cells transfected with Flag–STAT3, Flag–STAT2, Flag–STAT4 or Flag–STAT6. i, HEK293A cells were pulse-labelled with 50 μM chemical probe (alk-C16) for 4 h, and subsequently chased by the addition of an excess of 50 μM BSA-conjugated palmitic acid. Cells were collected at the indicated time point, and subjected to analysis of STAT3 palmitoylation. j, Schematic of Flag–STAT3 and Flag–STAT3 truncation-mutant constructs. k, HEK293A cells were transfected with vector control, Flag-tagged wild-type STAT3, STAT3(ΔSH2), STAT3(1–585) or STAT3(586–770) mutant, and labelled with 50 μM chemical probe (alk-C16) for 4 h. STAT3 palmitoylation levels were analysed by click reaction and streptavidin bead pulldown, followed by western blotting using anti-Flag antibody. l, Alignment of STAT3 protein sequence among different species. m, HEK293A cells were transfected with Flag-tagged wild-type STAT3, or the STAT3(C687S), STAT3(C712S) or STAT3(C687S/C712S) (here labelled 2CS) mutants, and total cell lysates were subjected to APE assay. Samples were analysed by western blot using anti-Flag antibody. The top band indicates the palmitoylated STAT3. n, The palmitoylated STAT3 in m was quantified using ImageJ. The data are presented as mean ± s.e.m., n = 3 biologically independent experiments. P value was determined by two-tailed t-test. o, HEK293A cells were transfected with Flag-tagged wild-type STAT3, or the STAT3(C687S), STAT3(C712S) or STAT3(C687S/C712S) mutants, and total cell lysates were reduced with DTT. Samples were analysed by western blot using anti-Flag antibody. (p) HEK293A cells were transfected with Flag-tagged wild-type STAT3 or STAT3(C687S/C712S) mutant, and free thiols of protein were blocked by NEM. Disulfide bonds were reduced by DTT. Maleimide–PEG was applied to attach at the reduced thiols. Samples were analysed by western blot using anti-Flag antibody. q, HEK293A cells were transfected with Flag-tagged wild-type STAT3, or the STAT3(C687S), STAT3(C712S) or STAT3(C687S/C712S) mutants, and total cell lysates were heated to the indicated temperatures. Samples were analysed by western blot using anti-Flag antibody. Soluble fractions were plotted versus temperatures. r, HEK293A cells were transfected with Flag-tagged wild-type STAT3, or the STAT3(C687S), STAT3(C712S) or STAT3(C687S/C712S) mutants, and cells were heated to the indicated temperatures. Total cell lysates were analysed by western blot using anti-Flag antibody. Soluble fractions were plotted versus temperatures. In bi, k, or, the experiments were independently repeated at least three times with similar results. For gel source data, see Supplementary Fig. 1.

Source data

Extended Data Fig. 2 STAT3 palmitoylation promotes STAT3 nuclear translocation and synergically enhances STAT3 signalling activity with phosphorylation.

a, HEK293A cells were pretreated with ruxolitinib (1 μM) for 0.5 h and then labelled with 50 μM chemical probe (alk-C16) for 2 h, with the incubation with IFNγ (1 ng ml−1). Endogenous STAT3 palmitoylation was analysed by click reaction and streptavidin bead pulldown, and subjected to western blotting. b, HEK293A cells were transfected with Flag-tagged wild-type STAT3 or Flag–STAT3(C687S) and then labelled with 50 μM chemical probe (alk-C16) for 2 h, with the incubation with IL-6 (20 ng ml−1). STAT3 palmitoylation was analysed by western blot, after click reaction and streptavidin bead pulldown. c, HEK293A cells were transfected with MYC-tagged wild-type STAT3, MYC–STAT3(C687S) or MYC–STAT3(Y705F), and treated with IL-6 (20 ng ml−1, 1 h). Whole-cell lysates were immunoprecipitated with anti-MYC antibody, followed by immunoblotting using indicated antibodies. d, Confocal microscopy showing changes in subcellular localization of wild-type STAT3 and STAT3(C687S/C712S) mutant in HEK293A stable cell lines. Cells were stained with anti-Flag antibody (green), anti-lamin B1 antibody (red) and DAPI (blue). Lamin B1 displays the nuclear membrane. The red line indicates the position of the Z-stack. Western blot showing that the wild-type STAT3 and STAT3(C687S/C712S) mutant were expressed at comparable levels in HEK293A stable cell lines. The levels of nuclear-localized STAT3 were quantified by measuring the Flag (STAT3) fluorescence intensity in the nucleus, using the confocal software to define the selected region of interest on the basis of nuclear DAPI signal. The data are presented as mean ± s.e.m. n = 225 cells with wild-type STAT3 and n = 300 cells with STAT3(C687S/C712S) mutant. P value was determined by two-tailed Student’s t-test. Scale bars, 20 μm. e, HEK293A cells were transfected with Flag-tagged wild-type STAT3, or Flag–STAT3(C687S/C712S) or Flag–STAT3(K685S), with or without CBP. Whole-cell lysates were analysed by western blot using indicated antibodies. f, HEK293A cells were co-transfected with Flag-tagged wild-type STAT3 or Flag–STAT3(C687S/C712S) along with GFP–STAT1. Whole-cell lysates were analysed by anti-Flag immunoprecipitation, followed by immunoblotting using the indicated antibodies. g, Analysis similar to that in f was performed in HEK293A cells transfected with Flag–STAT3 and HA–JAK1, followed by anti-HA immunoprecipitation. h, HEK293A cells were co-transfected with Flag–STAT3 and MYC–STAT3, pretreated with 5,15-DPP (10 μM) and then treated with IL-6 (20 ng ml−1, 1 h). Whole-cell lysates were analysed by anti-Flag immunoprecipitation, followed by immunoblotting using the indicated antibodies. i, HEK293A cells were transfected with Flag–STAT3 treated with 5,15-DPP alone or in combination with IL-6 (20 ng ml−1, 1 h), and labelled with 50 μM chemical probe (alk-C16). STAT3 palmitoylation was analysed by click reaction and streptavidin bead pulldown, and subjected to western blotting. j, The acyl chain-binding pocket according to the STAT3 crystal structure (RCSB Protein Data Bank code 4E68). Hydrophobic amino acids are shown in yellow. k, Mutational analysis of selected residues involved in STAT3 palmitoylaiton. l, HEK293A cells were transfected with MYC- and Flag-tagged STAT3 as indicated. Whole-cell lysates were immunoprecipitated with anti-MYC antibody, followed with immunoblotting using indicated antibodies. m, STAT3-null (Stat3−/−) mouse embryonic fibroblast (MEF) cells were transfected with vector control, Flag-tagged wild-type STAT3 or STAT3(C687S/C712S) mutant. Cells were labelled with 50 μM alk-C16 probe for 2 h, with the incubation with IL-6 (20 ng ml−1). Western blotting using anti-Flag antibody in the streptavidin bead pulldown samples indicates the palmitoylation levels of STAT3. n, Stat3−/− mouse embryonic fibroblast cells were co-transfected with STAT3 reporter construct (m67-luciferase reporter) and Renilla luciferase control construct, and a vector control, a Flag-tagged wild-type STAT3 or STAT3(C687S/C712S) mutant. Cells were then treated with IL-6 (20 ng ml−1) for 8 h. Luciferase activity was obtained from triplicate experiments and normalized to the Renilla luciferase. Relative fold induction was plotted as shown. o, Experiment similar to that in n was performed in HEK293A cells. p, Stat3−/− mouse embryonic fibroblast cells were transfected with vector control, Flag-tagged wild-type STAT3 or STAT3(C687S/C712S) mutant. Cells were treated with IL-6 (20 ng ml−1) for 8 h. The mRNA levels of STAT3 target genes (BCL2, BCL2L1 and MMP9) were analysed by quantitative PCR with reverse transcription (qRT–PCR). q, Experiment similar to that in p performed in HEK293A cells. In nq, the data are presented as mean ± s.e.m. n = 3 biologically independent samples. P value were determined by two-tailed Student’s t-test. In ac, ei, km, the experiments were independently repeated at least three times with similar results. For gel source data, see Supplementary Fig. 1.

Source data

Extended Data Fig. 3 Palmitic acid promotes STAT3 nuclear localization and activity through induction of dimerization in synergy with cytokine stimulation.

a, C57BL/6 mice were fed with an NFD (10% kcal from fat) or HFD (60% kcal from fat) for 2 weeks. Levels of STAT3 palmitoylation and phosphorylation at Y705 were analysed by APE and western blot in mouse lung tissues. b, Experiment similar to that in a was performed in liver tissue. c, U3A cells were treated with various fatty acids at 100 μM or IL-6 (20 ng ml−1) for 6 h. Luciferase activity was measured, and relative fold induction from triplicate experiments were calculated. d, Confocal microscopy showing changes in subcellular localization of STAT3 in HEK293A cells treated with palmitic acid (50 μM, overnight) and/or IL-6 (20 ng ml−1, 1 h). Cells were stained with anti-STAT3 antibody (red) and DAPI (blue). The yellow line indicates the position of the Z-stack. The levels of nuclear-localized STAT3 were quantified by measuring the STAT3 fluorescence intensity in the nucleus, using the confocal software to define the selected region of interest on the basis of nuclear DAPI signal. Scale bars, 20 μm. The data are presented as mean ± s.e.m., n = 217 cells (control), n = 246 cells (palmitic-acid-treated), n = 261 (IL-6-treated), and n = 288 cells (palmitic-acid- and IL-6-treated). P values were determined by two-tailed Student’s t-test. e, HEK293A cells were transfected with STAT3–luciferase reporter with Renilla control constructs, and treated with 20 ng ml−1 IL-6 or 100 μM palmitic acid at the indicated time points. Luciferase activity was measured, and relative fold induction from triplicate experiments was plotted with mean ± s.e.m. n = 3 biologically independent samples. f, HEK293A cells were treated with 20 ng ml−1 IL-6 or 100 μM palmitic acid, and the expression of indicated genes was quantified with qRT–PCR. g, HEK293A cells were transfected with Flag–STAT3 with or without MYC-tagged wild-type STAT3, and treated with 100 μM palmitic acid for 2 h. Whole-cell lysates were analysed by anti-Flag immunoprecipitation, followed by immunoblotting using the indicated antibodies. h, Experiment similar to that in g, performed in HEK293A cells with or without treatment with NH2OH. i, HEK293A cells were co-transfected with m67–luciferase reporter and Renilla reporter constructs along with vector control, Flag-tagged wild-type STAT3 or STAT3(C687S/C712S) mutant. After 48 h, cells were treated with palmitic acid (100 μM, 8 h) and/or IL-6 (20 ng ml−1, 8 h). Normalized luciferase activity from triplicate experiments was plotted as shown. j, HEK293A cells were transfected with vector control, Flag-tagged wild-type STAT3 or STAT3(C687S/C712S) mutant. Cells were treated with IL-6 (20 ng ml−1) for 8 h. The mRNA levels of STAT3 target-gene BCL2 were analysed by qRT–PCR. Relative fold change (normalized to GAPDH) is shown, from triplicate experiments. In c, f, i, j, the data are presented as mean ± s.e.m. n = 3 biologically independent samples. P values were determined by two-tailed Sudent’s t-test. In a, b, g, h, the experiments were independently repeated at least three times with similar results. For gel source data, see Supplementary Fig. 1.

Source data

Extended Data Fig. 4 ZDHHC19 mediates STAT3 palmitoylaiton and Ras proteins are not involved in ZDHHC19-mediated STAT3 signalling.

a, b, HEK293A cells were co-transfected with construct encoding Flag–STAT3 and HA-tagged ZDHHC proteins. After 48 h, cells were labelled for 4 h with 50 μM palmitoylation probe (alk-C16). Cell lysates were reacted with biotin azide and subjected to streptavidin bead pulldown. STAT3 palmitoylation was shown by western blot. c, HEK293A cells were transfected with Flag–STAT3 alone or in combination with HA–ZDHHC19, HA–ZDHHC5 or HA–ZDHHC 18, and labelled with 50 μM probe (alk-C16) for 4 h, followed by analysis of STAT3 palmitoylation. d, HEK293A cells were transfected with Flag–STAT3 alone or in combination with HA–ZDHHC19, HA–ZDHHC5 or HA–ZDHHC18. Whole-cell lysates were analysed by anti-Flag immunoprecipitation, followed by immunoblotting using the indicated antibodies. e, HEK293A cells were transfected as in d. Total cell extracts were analysed by western blot after immunoprecipitation with anti-HA antibody. f, HEK293A cells were transfected with Flag-tagged wild-type STAT3 or Flag–STAT3(K685S), with or without ZDHHC19. Palmitoylation of STAT3 was analysed as in c. g, h, HEK293A cells were transfected with HA-tagged wild-type ZDHHC19 or the inactive mutant ZDHHC19(C142S), and labelled with 50 μM alk-C16 or C16–BYA. Cell lysates were subjected to streptavidin bead pulldown. STAT3 palmitoylation was showed by western blot. i, HEK293A cells were transfected with STAT3–luciferase reporter and Renilla control, co-transfected with wild-type ZDHHC19 or ZDHHC19(C142S) as shown, and treated with IL-6 (20 ng ml−1) and/or palmitic acid (100 μM) for 8 h. Luciferase activity was measured, and relative fold induction was plotted from triplicate experiments. j, HEK293A cells were co-transfected with ZDHHC19-shRNA and wild-type ZDHHC19 or ZDHHC19(C142S), and the expression of the indicated genes was quantified with qRT–PCR. km, KNS62 control cells and ZDHHC19-knockout stable cells were treated with IL-6 (20 ng ml−1) and/or palmitic acid (100 μM) for 8 h. The mRNA levels of STAT3 target genes (BCL2, BCL2L1 and MMP9) were analysed by qRT–PCR. n, Schematic of wild-type GRB2 and SH3-domain truncation mutants, and alignment of the SH3 binding motif sequences of ZDHHC19. o, HEK293A cells were co-transfected with MYC-tagged wild-type ZDHHC19 and GFP-tagged wild-type GRB2, or GRB2 with an N-terminal SH3 deletion (ΔN-SH3) or C-terminal SH3 deletion (ΔC-SH3). Whole-cell lysates were subjected to immunoprecipitation using anti-MYC antibody, followed by western blotting using the indicated antibodies. p, HEK293A cells were co-transfected with MYC-tagged wild-type ZDHHC19 or ZDHHC19(P18A) mutant, Flag-tagged STAT3 and GFP-tagged GRB2, followed by co-immunoprecipitation assay. q, HEK293A cells were treated with IL-6 (20 ng ml−1) and analysed by anti-STAT3 co-immunoprecipitation assay, followed by western blotting using the indicated antibodies. r, s, HA-tagged ZDHHC19 was co-transfected with HA-tagged HRAS (r) or NRAS (s) into HEK293A cells. After 24 h, cells were incubated with fresh medium containing 10% dialysed FBS for 2 h, and subsequently labelled with 50 μM probe (alk-C16) for 4 h. Cell lysates were reacted with biotin azide and subjected to SDS–PAGE. Streptavidin blot was used to detect HRAS (r) or NRAS (s) palmitoylation, as described for the detection of STAT3 palmitoylation in Methods section. Comparable protein loading was confirmed by anti-HA and β-actin western blotting. t, u, Treatment with the depalmitoylase inhibitor palmostatin B (10 μM) or overexpression of the depalmitoylase ABHD17A has no effect on the expression of STAT3 target genes (BCL2, BCL2L1 and MMP9) in HEK293A cells that express oncogenic HRAS(G12V) (t) or NRAS(G12V) (u). v, Treatment with different concentrations of palmostatin B (1 μM, 5 μM and 10 μM) has no effect on the expression of STAT3 target genes (BCL2, BCL2L1 and MMP9) in melanoma SK-MEL-2 cells that contain the NRASQ61R mutation. In im, tv, the data are presented as mean ± s.e.m. n = 3 biologically independent samples. P values were determined by two-tailed Student’s t-test. In ah, os, the experiments were independently repeated at least three times with similar results. For gel source data, see Supplementary Fig. 1.

Source data

Extended Data Fig. 5 The ZDHHC19 gene is amplified and highly expressed in LSCC and correlated with poor clinical outcomes.

a, Alteration frequency and oncoprint diagram of ZDHHC19 gene-alteration summary in patients with cancer (11,413) obtained from cBioPortal (The Cancer Genome Atlas, TGCA). b, The regions of the genome that are substantially amplified across a set of LSCC samples from TCGA were identified using GISTIC2 from Gene Pattern (Broad Institute). ZDHHC19 was confirmed as one of the genes in the amplified 3q29 region. c, Genes in the amplified 3q29 region are correlated with the STAT3 target-genes BCL2L1 and BCL2 across two datasets from the Gene Expression Omnibus (GEO) (GSE28571 (n = 100 samples) and GSE73403 (n = 69 samples)). Twenty-eight genes were analysed. P values were calculated by two-tailed Pearson correlation. d, Venn diagram of genes in the amplified 3q29 region that are significantly correlated with the STAT3 target-genes BCL2L1 and BCL2 across the two GEO datasets (GSE28571 (n = 100 samples) and GSE73403(n = 69 samples)). Thirty-six genes from the GSE28571 dataset and 30 genes from the GSE73403 datset were analysed. P values were determined by two-tailed Pearson correlation. e, The plots of P values versus q values of oncogenes that show co-occurrence or mutual exclusivity with ZDHHC19 in LSCC from TCGA (n = 501 samples). Two hundred and thirty-one genes (162 examples of co-occurrence and 69 examples of mutual exclusivity) were analysed. The P and q values were determined by Fisher’s exact test and false-discovery-rate (FDR) test, respectively. f, The plots of P values versus q values of tumour suppressors that show co-occurrence or mutual exclusivity with ZDHHC19 in LSCC from TCGA (n = 501 samples). One hundred and eighty genes (115 examples of co-occurrence and 65 examples of mutual exclusivity) were analysed. The P and q values were determined by Fisher’s exact test and FDR test, respectively. g, Gene list of oncogenes or tumour suppressors that show significant co-occurrence or mutual exclusivity with ZDHHC19 in LSCC from TCGA (n = 501 samples). The P and q values were determined by Fisher’s exact test and FDR test, respectively. h, Comparison of ZDHHC19 expression in samples from patients with LSCC or lung adenocarcinoma, obtained from the cBioPortal TCGA dataset (n = 1,097 samples). The P values were determined by two-tailed Student’s t-test. i, The expression level of ZDHHC19 in human lung-cancer cell lines was grouped into four subtypes: LSCC cell lines (lung squ) (n = 29 samples); lung large-cell carcinoma cell lines (large lung) (n = 14 samples); lung adenocarcinoma cell lines (lung adeno) (n = 53 samples); and lung small-cell lung carcinoma cell lines (small lung) (n = 53 samples). Data were obtained from Cancer Cell Line Encyclopedia database (CCLE, www.broadinstitute.org/ccle) and plotted using GraphPad Prism. P values were determined by two-tailed Student’s t-test. j, Kaplan–Meier curves of 68 patients with LSCC, after stratification by the median level of ZDHHC19, were used for depicting survival time. Patient data were derived from GSE73403 and analysed using GraphPad Prism. P values were determined by log-rank (Mantel–Cox) test. k, Kaplan–Meier curves of patients with LSCC, after stratification by the median level of ZDHHC19, were used for depicting survival time. Patient data were analysed using http://kmplot.com. P values were determined by log-rank (Mantel–Cox) test. l, Pearson analysis of gene-expression data from patients with LSCC (GSE73403 dataset (n = 69 samples)) was used for depicting the correlation between BCL2 and ZDHHC19. Data were plotted and analysed using GraphPad Prism software. m, Analysis similar to that in l; gene-expression data from patients with LSCC (GSE73403 dataset (n = 69 samples)) was used for depicting the correlation between BCL2L1 and ZDHHC19. n, Pearson analysis of gene-expression data from patients with LSCC (GSE28571 dataset (n = 100 samples)) was used for depicting the correlation between BCL2 and ZDHHC19. o, Analysis of gene-expression data from patients with LSCC (GSE28571 dataset (n = 100 samples)) was used for depicting the correlation between BCL2L1 and ZDHHC19. One hundred samples were plotted and analysed using GraphPad Prism software. In lo, the centre line indicates a line of best fit through the data of two variables in Pearson correlation coefficient model, and the dotted lines indicate the 95%-confidence band.

Source data

Extended Data Fig. 6 ZDHHC19 expression correlates with STAT3 nuclear localization in LSCC.

a, Immunohistochemistry staining showing correlation of ZDHHC19 expression with STAT3 nuclear localization. One hundred and thirty-one biologically independent samples were analysed. b, Statistical analysis of Pearson correlation between immunohistochemistry staining scores of ZDHHC19 expression and STAT3 nuclear localization. c, Summary of tissue samples from patients with LSCC, and the scores of ZDHHC19 expression and STAT3 nuclear localization.

Source data

Extended Data Fig. 7 ZDHHC19-mediated STAT3 palmitoylaiton facilitates LSCC tumour-cell growth, colony formation and migration in vitro through induction of STAT3 activity.

a, HCC95 LSCC cells were pretreated with ruxolitinib (1 μM) for 30 min and then labelled with 50 μM probe alk-C16 for 2 h with the incubation of IL-6 (20 ng ml−1). Endogenous STAT3 palmitoylation was analysed by click reaction and streptavidin bead pulldown, followed by western blotting. b, HCC95 cells were treated as in a. Whole-cell lysates were analysed by anti-STAT3 immunoprecipitation, followed by western blotting using the indicated antibodies. c, The LSCC cell lines HCC95 and KNS62 were transfected with control shRNA or ZDHHC19 shRNA (two shRNA clones, shRNA3 and shRNA5, were used and are denote here as 3 and 5, respectively), and labelled with 50 μM alk-C16 probe for 4 h. Cell lysates were reacted with biotin azide, and precipitated with streptavidin beads. Endogenous STAT3 palmitoylation was determined by western blot. d, SK-MES-1 cells were transfected with ZDHHC19 shRNA and labelled with 50 μM palmitoylation probe (alk-C16) for 4 h. Cell lysates were reacted with biotin azide and subjected to streptavidin bead pulldown. STAT3 palmitoylation was shown by western blot. e, HCC95 control-shRNA cells and ZDHHC19-shRNA (shRNA3, denoted here as 3) stable cells were pretreated with ruxolitinib (1 μM) for 30 min, and then labelled with 50 μM probe (alk-C16) for 2 h with the incubation of IL-6 (20 ng ml−1). Endogenous STAT3 palmitoylation was determined by western blot. f, Confocal immunofluorescence imaging showed STAT3 localization in HCC95 ZDHHC19-shRNA (shRNA3) stable cells and vector-control cells. Cells were stained with anti-STAT3 antibody (green) and DAPI (blue). The levels of nuclear-localized STAT3 were quantified by measuring the STAT3 fluorescence intensity in the nucleus, using the confocal software to define the selected region of interest on the basis of the nuclear DAPI signal. Scale bar, 20 μm. The data are presented as mean ± s.e.m., n = 275 cells (control) and n = 320 cells (ZDHHC19 shRNA). P value was determined by two-tailed Student’s t-test. g, Cell proliferation was determined in HCC95 ZDHHC19-shRNA (shRNA3) stable cells or control-shRNA cells. The data are presented as mean ± s.e.m. n = 3 biologically independent samples. P values were determined by two-way ANOVA, followed by Bonferroni’s test. h, Cell proliferation was determined in HCC95 control cells or HCC95 cells with CRISPR–Cas9 knockout of ZDHHC19. The data are presented as mean ± s.e.m. n = 6 biologically independent samples. P value was determined by two-way ANOVA, followed by Bonferroni’s test. i, Colony formation of HCC95 control cells or HCC95 cells with CRISPR–Cas9 knockout of ZDHHC19. The colony number was quantified. The data are presented as mean ± s.e.m. n = 5 biologically independent samples. P value was determined by two-tailed Student’s t-test. j, Cell proliferation showing KNS62 cells with ZDHHC19-shRNA knockdown or control shRNA. The data are presented as mean ± s.e.m. n = 12 biologically independent samples. P value was determined by two-way ANOVA, followed by Bonferroni’s test. k, Cell proliferation was determined in KNS62 control cells or KNS62 cells with CRISPR-Cas9 knockout of ZDHHC19. The data are presented as mean ± s.e.m. n = 6 biologically independent samples. P value was determined by two-way ANOVA, followed by Bonferroni’s test. l, Colony formation of HCC95 control cells or HCC95 cells with CRISPR–Cas9 knockout of ZDHHC19. The colony number was quantified. The data are presented as mean ± s.e.m., n = 6 biologically independent samples. P value was determined by two-tailed Student’s t-test. m, Cell proliferation was determined in SK-MES-1 control cells or SK-MES-1 cells with CRISPR–Cas9 knockout of ZDHHC19. The data are presented as mean ± s.e.m. n = 12 biologically independent samples. P values were determined by two-way ANOVA, followed by Bonferroni’s test. n, Colony formation of SK-MES-1 control cells or SK-MES-1 cells with CRISPR–Cas9 knockout of ZDHHC19. The colony number was quantified. The data are presented as mean ± s.e.m. n = 6 biologically independent samples. P value was determined by two-tailed Student’s t-test. o, The migration ability of HCC95 ZDHHC19-shRNA (shRNA3) stable cells or control-shRNA cells was measured using transwell migration assay. The numbers of invading cells were quantified. The data are presented as mean ± s.e.m. n = 5 biologically independent samples. P values were determined by two-tailed Student’s t-test. p, Crystal staining showing cell growth of KNS62 ZDHHC19-knockdown or control stable cells transfected with STAT3C or vehicle control. In ae, p, the experiments were independently repeated at least three times with similar results. For gel source data, see Supplementary Fig. 1.

Source data

Extended Data Fig. 8 STAT3 palmitoylaiton through ZDHHC19 is involved in maintaining the cancer stem-cell niche.

a, Photomicrographs of HCC95, KNS62 and HCC827 parental adherent cells (left) and tumour spheres (right) in low-adherence culture. The experiment was independently repeated at least three times with similar results. b, qRT–PCR analysis of the expression level of stem-cell markers in tumour spheres of HCC95 cells, compared to parental HCC95 cells. Fold change was normalized to 18S rRNA. The data are presented as mean ± s.e.m. n = 3 biologically independent samples. P values were determined by two-tailed Student’s t-test. c, d, Experiments similar to those in b were performed in KNS62 and HCC827 cells. The data are presented as mean ± s.e.m. n = 3 biologically independent samples. P values were determined by two-tailed Student’s t-test. e, HCC95 empty-vector control cells and STAT3- or ZDHHC19-knockout stable cells were cultured in low-attachment plates with 25 μM palmitic acid for 7 days. Phase-contrast photomicrographs showing tumour-sphere formation. Numbers of spheres were counted from five randomly selected fields. The data are presented as mean ± s.e.m. n = 5 biologically independent samples. P values were determined by two-tailed sStudent’s t-test. f, KNS62 empty-vector control cells and ZDHHC19-shRNA knockdown stable cells were cultured in low-attachment plates with 25 μM palmitic acid for 5 days. Fluorescent (calcein AM staining) and phase-contrast photomicrographs showing tumour-sphere formation. The data are presented as mean ± s.e.m. n = 12 biologically independent samples. P values were determined by two-tailed Student’s t-test. g, SK-MES-1 empty-vector control cells and ZDHHC19-shRNA knockdown stable cells were cultured in low-attachment plates with 25 μM palmitic acid for 5 days. Fluorescent (calcein AM staining) and phase-contrast photomicrographs showing tumour-sphere formation. The data are presented as mean ± s.e.m. n = 12 biologically independent samples. P values were determined by two-tailed Student’s t-test. h, KNS62 empty-vector control cells and ZDHHC19-shRNA knockdown stable cells were cultured in low-attachment plates with 25 μM palmitic acid for 5 days. The cells were trypsinized, and seeded in plates again. Fluorescent (calcein AM staining) photomicrographs showing secondary tumour-sphere formation after five days. The data are presented as mean ± s.e.m. n = 4 biologically independent samples. P values were determined by two-tailed Student’s t-test. i, KNS62 empty-vector control cells and ZDHHC19-shRNA knockdown stable cells transfected with vehicle control or constitutive active STAT3C were cultured in low-attachment plates for five days. Fluorescent (calcein AM staining) photomicrographs showing tumour-sphere formation. Tumour-sphere numbers were quantified. The data are presented as mean ± s.e.m. n = 4 biologically independent samples. P values were determined by two-tailed Student’s t-test.

Source data

Extended Data Fig. 9 HFD-induced tumour growth of LSCC in vivo is dependent on ZDHHC19-mediated STAT3 palmitoylaiton.

a, Representative image of the xenograft tumours isolated from HCC95 empty-vector control and ZDHHC19-knockout xenografts, as indicated. b, Weights of the tumours were measured and plotted. The data are presented as mean ± s.e.m. n = 10 biologically independent samples. P values were determined by two-tailed Student’s t-test. c, Body-weight change of mouse fed with an NFD or HFD. Data are normalized to original weight. The data are presented as mean ± s.e.m. n = 5 mice. d, STAT3 palmitoylation was analysed by APE and western blotting in human-derived cell tumour tissues. The palmitoylated STAT3 is quantified using ImageJ. The data are presented as mean ± s.e.m. n = 3 biologically independent samples. P values were determined by two-tailed Student’s t-test. e, Representative images of immunohistochemical staining of Ki-67 in paraffin-embedded xenograft tumour tissues collected from the indicated groups. Scale bar, 200 μm. Numbers of Ki-67-positive cells were counted from five randomly selected fields. The data are presented as mean ± s.e.m. n = 10 biologically independent samples. P values were determined by two-tailed Student’s t-test. f, Representative images of immunohistochemical staining of STAT3. g, qRT–PCR analysis of the expression level of STAT3 target genes (BCL2, BCL2L1 and MMP9) was performed in tumour tissues collected at the end of the experiment. Fold change was normalized to 18S rRNA. The data are presented as mean ± s.e.m. n = 10 biologically independent samples. P values were determined by two-tailed Student’s t-test. h, HCC95 shRNA-control and ZDHHC19-shRNA (shRNA3) stable cell lines were injected into mice. Representative image of the xenograft tumours isolated from the indicated groups. i, Tumour growth was monitored and read-out by the tumour volume. The data are presented as mean ± s.e.m. n = 10 biologically independent samples (control); n = 9 biologically independent samples (shZDHHC19). P value was determined by two-way ANOVA, followed by Bonferroni’s test. j, Weight of the tumours was measured, and the data are presented as mean ± s.e.m. n = 10 biologically independent samples (control); n = 9 biologically independent samples (shZDHHC19). P value was determined by two-tailed Student’s t-test. k, qRT–PCR analysis of the expression level of STAT3 target genes in HCC95 shRNA-control or ZDHHC19-shRNA (shRNA3) stable-cell xenografts. The data are presented as mean ± s.e.m. n = 10 biologically independent samples (control); n = 9 biologically independent samples (shZDHHC19). P values were determined by two-tailed Student’s t-test. l, Representative images of immunohistochemical staining of STAT3 and Ki-67 in paraffin-embedded xenograft tumour tissues collected from the indicated groups. Ki-67-positive cells were quantified as in e. The data are presented as mean ± s.e.m. n = 5 biologically independent samples. P value was determined by two-tailed Student’s t-test.

Source data

Extended Data Fig. 10 HFD facilitates tumour growth in PDX model of LSCC.

a, qRT–PCR analysis of the expression level of ZDHHC19 was performed in a series of PDX tumour tissues (PDX1–PDX7). Fold change was normalized to 18S rRNA. The data are presented as mean ± s.e.m. n = 3 technical replicates. b, Representative image of the xenograft tumours isolated from a PDX mouse with an NFD or HFD. c, Weight of the tumours was measured. Data are presented as mean ± s.e.m. n = 8 biologically independent samples (NFD); n = 10 biologically independent samples (HFD). P values were determined by two-tailed Student’s t-test. d, STAT3 palmitoylation was analysed by APE and western blotting in PDX tumour tissues. The palmitoylated STAT3 was quantified using ImageJ. The data are presented as mean ± s.e.m. n = 5 biologically independent samples. P values were determined by two-tailed Student’s t-test. e, qRT–PCR analysis of the expression level of STAT3 target genes (BCL2, BCL2L1 and MMP9) was performed in tumour tissues collected at the end of the experiment. Fold change was normalized to 18S rRNA. The data are presented as mean ± s.e.m. n = 8 biologically independent samples (NFD); n = 10 biologically independent samples (HFD). P values were determined by two-tailed Student’s t-test. f, Representative images of immunohistochemical staining of STAT3 and ZDHHC19 in paraffin-embedded PDX tumour tissues collected from the indicated groups. g, Transcriptional level of ZDHHC19 shows the knockdown efficiency in primary PDX cells by qRT–PCR. The data are presented as mean ± s.e.m. n = 3 biologically independent samples. P values were determined by two-tailed Student’s t-test. h, Representative images of tumour-sphere formation using primary tumour cells isolated from LSCC PDX tumour tissues. i, Schematic showing that fatty acids and amplified ZDHHC19 promote STAT3 activation through S-palmitoylation. In f, h, the experiments were independently repeated at least three times with similar results.

Source data

Supplementary information

Supplementary Figure 1

Uncropped gel data scans of Figs. 1–4 and Extended Data Figs 1–10. The loading controls were run in the same gel. Blots from the same gel were indicated in the figure.

Reporting Summary

Supplementary Table 1

Clinical characteristics of de-identified patient samples for lung squamous cell carcinoma tissue microarray cohort shown in Fig. 4. Age, sex, tumor site, clinical and pathology diagnosis, and histology grade of the tumor are listed.

Supplementary Table 2

Clinical characteristics of de-identified patient samples used for lung squamous cell carcinoma patient-derived xenograft models (PDX) shown in Fig. 4. Age, sex, histology grade, clinical and pathology diagnosis, biospecimen type and the sources are listed.

Source data

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Niu, J., Sun, Y., Chen, B. et al. Fatty acids and cancer-amplified ZDHHC19 promote STAT3 activation through S-palmitoylation. Nature 573, 139–143 (2019). https://doi.org/10.1038/s41586-019-1511-x

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41586-019-1511-x

This article is cited by

Search

Quick links

Nature Briefing: Cancer

Sign up for the Nature Briefing: Cancer newsletter — what matters in cancer research, free to your inbox weekly.

Get what matters in cancer research, free to your inbox weekly. Sign up for Nature Briefing: Cancer