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.

  • Article
  • Published:

Tumour hypoxia causes DNA hypermethylation by reducing TET activity

This article has been updated

Abstract

Hypermethylation of the promoters of tumour suppressor genes represses transcription of these genes, conferring growth advantages to cancer cells. How these changes arise is poorly understood. Here we show that the activity of oxygen-dependent ten-eleven translocation (TET) enzymes is reduced by tumour hypoxia in human and mouse cells. TET enzymes catalyse DNA demethylation through 5-methylcytosine oxidation. This reduction in activity occurs independently of hypoxia-associated alterations in TET expression, proliferation, metabolism, hypoxia-inducible factor activity or reactive oxygen species, and depends directly on oxygen shortage. Hypoxia-induced loss of TET activity increases hypermethylation at gene promoters in vitro. In patients, tumour suppressor gene promoters are markedly more methylated in hypoxic tumour tissue, independent of proliferation, stromal cell infiltration and tumour characteristics. Our data suggest that up to half of hypermethylation events are due to hypoxia, with these events conferring a selective advantage. Accordingly, increased hypoxia in mouse breast tumours increases hypermethylation, while restoration of tumour oxygenation abrogates this effect. Tumour hypoxia therefore acts as a novel regulator of DNA methylation.

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

Access options

Rent or buy this article

Prices vary by article type

from$1.95

to$39.95

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

Figure 1: Effect of hypoxia on 5hmC in vitro.
Figure 2: Genomic profiles of 5(h)mC in MCF7 following hypoxia.
Figure 3: Effect of hypoxia on hypermethylation in TCGA.
Figure 4: Effect of hypoxia on TET activity in human tumours.
Figure 5: Effect of vessel pruning and normalization on 5hmC and TSG hypermethylation.

Similar content being viewed by others

Accession codes

Data deposits

Microarray and sequencing data are available at the Gene Expression Omnibus under accession GSE71403.

Change history

  • 16 September 2016

    The affiliations of authors V.N.K., K.P.K. and M.M. were corrected in the HTML version of the paper.

References

  1. Esteller, M. Epigenetics in cancer. N. Engl. J. Med. 358, 1148–1159 (2008)

    CAS  PubMed  Google Scholar 

  2. Struhl, K. Is DNA methylation of tumour suppressor genes epigenetic? eLife 3, e02475 (2014)

    PubMed  PubMed Central  Google Scholar 

  3. Weisenberger, D. J. et al. CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nat. Genet. 38, 787–793 (2006)

    CAS  PubMed  Google Scholar 

  4. Mack, S. C. et al. Epigenomic alterations define lethal CIMP-positive ependymomas of infancy. Nature 506, 445–450 (2014)

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  5. Tahiliani, M. et al. Conversion of 5-methylcytosine to 5-hydroxymethylcytosine in mammalian DNA by MLL partner TET1. Science 324, 930–935 (2009)

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  6. Shen, L. et al. Genome-wide analysis reveals TET- and TDG-dependent 5-methylcytosine oxidation dynamics. Cell 153, 692–706 (2013)

    CAS  PubMed  PubMed Central  Google Scholar 

  7. Xiao, M. et al. Inhibition of α-KG-dependent histone and DNA demethylases by fumarate and succinate that are accumulated in mutations of FH and SDH tumor suppressors. Genes Dev . 26, 1326–1338 (2012)

    CAS  PubMed  PubMed Central  Google Scholar 

  8. Figueroa, M. E. et al. Leukemic IDH1 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair hematopoietic differentiation. Cancer Cell 18, 553–567 (2010)

    CAS  PubMed  PubMed Central  Google Scholar 

  9. Xu, W. et al. Oncometabolite 2-hydroxyglutarate is a competitive inhibitor of α-ketoglutarate-dependent dioxygenases. Cancer Cell 19, 17–30 (2011)

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Yang, H. et al. Tumor development is associated with decrease of TET gene expression and 5-methylcytosine hydroxylation. Oncogene 32, 663–669 (2013)

    CAS  PubMed  Google Scholar 

  11. Ploumakis, A. & Coleman, M. L. OH, the places you’ll go! Hydroxylation, gene expression, and cancer. Mol. Cell 58, 729–741 (2015)

    CAS  PubMed  Google Scholar 

  12. Schofield, C. J. & Ratcliffe, P. J. Oxygen sensing by HIF hydroxylases. Nat. Rev. Mol. Cell Biol. 5, 343–354 (2004)

    CAS  PubMed  Google Scholar 

  13. Hanahan, D. & Folkman, J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 86, 353–364 (1996)

    CAS  PubMed  Google Scholar 

  14. Vaupel, P., Höckel, M. & Mayer, A. Detection and characterization of tumor hypoxia using pO2 histography. Antioxid. Redox Signal. 9, 1221–1235 (2007)

    CAS  PubMed  Google Scholar 

  15. Williams, K. et al. TET1 and hydroxymethylcytosine in transcription and DNA methylation fidelity. Nature 473, 343–348 (2011)

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  16. Ito, S. et al. Tet proteins can convert 5-methylcytosine to 5-formylcytosine and 5-carboxylcytosine. Science 333, 1300–1303 (2011)

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  17. Mariani, C. J. et al. TET1-mediated hydroxymethylation facilitates hypoxic gene induction in neuroblastoma. Cell Reports 7, 1343–1352 (2014)

    CAS  PubMed  Google Scholar 

  18. Zhao, B. et al. Redox-active quinones induces genome-wide DNA methylation changes by an iron-mediated and Tet-dependent mechanism. Nucleic Acids Res . 42, 1593–1605 (2014)

    CAS  PubMed  Google Scholar 

  19. Blaschke, K. et al. Vitamin C induces Tet-dependent DNA demethylation and a blastocyst-like state in ES cells. Nature 500, 222–226 (2013)

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  20. Koivunen, P. et al. Transformation by the (R)-enantiomer of 2-hydroxyglutarate linked to EGLN activation. Nature 483, 484–488 (2012)

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  21. Bachman, M. et al. 5-Hydroxymethylcytosine is a predominantly stable DNA modification. Nat. Chem. 6, 1049–1055 (2014)

    CAS  PubMed  PubMed Central  Google Scholar 

  22. Chowdhury, R. et al. Selective small molecule probes for the hypoxia inducible factor (HIF) prolyl hydroxylases. ACS Chem. Biol. 8, 1488–1496 (2013)

    CAS  PubMed  Google Scholar 

  23. Taberlay, P. C., Statham, A. L., Kelly, T. K., Clark, S. J. & Jones, P. A. Reconfiguration of nucleosome-depleted regions at distal regulatory elements accompanies DNA methylation of enhancers and insulators in cancer. Genome Res. 24, 1421–1432(2014)

    Google Scholar 

  24. Weinstein, J. N. et al. The Cancer Genome Atlas Pan-Cancer analysis project. Nat. Genet. 45, 1113–1120 (2013)

    PubMed  PubMed Central  Google Scholar 

  25. Buffa, F. M., Harris, A. L., West, C. M. & Miller, C. J. Large meta-analysis of multiple cancers reveals a common, compact and highly prognostic hypoxia metagene. Br. J. Cancer 102, 428–435 (2010)

    CAS  PubMed  PubMed Central  Google Scholar 

  26. Feinberg, A. P. & Irizarry, R. A. Evolution in health and medicine Sackler colloquium: Stochastic epigenetic variation as a driving force of development, evolutionary adaptation, and disease. Proc. Natl Acad. Sci. USA 107 (Suppl. 1), 1757–1764 (2010)

    ADS  CAS  PubMed  Google Scholar 

  27. Kuchnio, A. et al. The cancer cell oxygen sensor PHD2 promotes metastasis via activation of cancer-associated fibroblasts. Cell Reports 12, 992–1005 (2015)

    CAS  PubMed  Google Scholar 

  28. Vogelstein, B. et al. Cancer genome landscapes. Science 339, 1546–1558 (2013)

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  29. Landan, G. et al. Epigenetic polymorphism and the stochastic formation of differentially methylated regions in normal and cancerous tissues. Nat. Genet. 44, 1207–1214 (2012)

    CAS  PubMed  Google Scholar 

  30. Laukka, T. et al. Fumarate and succinate regulate expression of hypoxia-inducible genes via TET enzymes. J. Biol. Chem. 291, 4256–4265 (2016)

    CAS  PubMed  Google Scholar 

  31. Paez-Ribes, M. et al. Antiangiogenic therapy elicits malignant progression of tumors to increased local invasion and distant metastasis. Cancer Cell 15, 220–231, (2009)

    CAS  PubMed  PubMed Central  Google Scholar 

  32. Heist, R. S. et al. Improved tumor vascularization after anti-VEGF therapy with carboplatin and nab-paclitaxel associates with survival in lung cancer. Proc. Natl Acad. Sci. USA 112, 1547–1552 (2015)

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  33. Sermeus, A. et al. Hypoxia induces protection against etoposide-induced apoptosis: molecular profiling of changes in gene expression and transcription factor activity. Mol. Cancer 7, 27 (2008)

    PubMed  PubMed Central  Google Scholar 

  34. Schmittgen, T. D. & Livak, K. J. Analyzing real-time PCR data by the comparative C(T) method. Nat. Protocols 3, 1101–1108 (2008)

    CAS  PubMed  Google Scholar 

  35. Majmundar, A. J., Wong, W. J. & Simon, M. C. Hypoxia-inducible factors and the response to hypoxic stress. Mol. Cell 40, 294–309 (2010)

    CAS  PubMed  PubMed Central  Google Scholar 

  36. Feng, J., Liu, T. & Zhang, Y. Using MACS to identify peaks from ChIP-Seq data. Curr. Protoc. Bioinformatics Chapter 2, Unit 2.14, (2011)

  37. Durand, R. E. & Raleigh, J. A. Identification of nonproliferating but viable hypoxic tumor cells in vivo. Cancer Res. 58, 3547–3550 (1998)

    CAS  PubMed  Google Scholar 

  38. Lippert, A. R., Van de Bittner, G. C. & Chang, C. J. Boronate oxidation as a bioorthogonal reaction approach for studying the chemistry of hydrogen peroxide in living systems. Acc. Chem. Res. 44, 793–804 (2011)

    CAS  PubMed  PubMed Central  Google Scholar 

  39. Dickinson, B. C., Tang, Y., Chang, Z. & Chang, C. J. A nuclear-localized fluorescent hydrogen peroxide probe for monitoring sirtuin-mediated oxidative stress responses in vivo. Chem. Biol. 18, 943–948 (2011)

    CAS  PubMed  PubMed Central  Google Scholar 

  40. Vichai, V. & Kirtikara, K. Sulforhodamine B colorimetric assay for cytotoxicity screening. Nat. Protocols 1, 1112–1116 (2006)

    CAS  PubMed  Google Scholar 

  41. He, Y. F. et al. Tet-mediated formation of 5-carboxylcytosine and its excision by TDG in mammalian DNA. Science 333, 1303–1307 (2011)

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  42. Taiwo, O. et al. Methylome analysis using MeDIP-seq with low DNA concentrations. Nat. Protocols 7, 617–636 (2012)

    CAS  PubMed  Google Scholar 

  43. McCarthy, D. J., Chen, Y. & Smyth, G. K. Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation. Nucleic Acids Res . 40, 4288–4297 (2012)

    CAS  PubMed  PubMed Central  Google Scholar 

  44. Parker, J. S. et al. Supervised risk predictor of breast cancer based on intrinsic subtypes. J. Clin. Oncol. 27, 1160–1167 (2009)

    PubMed  PubMed Central  Google Scholar 

  45. Murtagh, F. & Legendre, P. Ward’s hierarchical agglomerative clustering method: which algorithms implement Ward’s criterion? J. Classif. 31, 274–295 (2014)

    MathSciNet  MATH  Google Scholar 

  46. Nielsen, T. O. et al. A comparison of PAM50 intrinsic subtyping with immunohistochemistry and clinical prognostic factors in tamoxifen-treated estrogen receptor-positive breast cancer. Clin. Cancer Res. 16, 5222–5232 (2010)

    CAS  PubMed  PubMed Central  Google Scholar 

  47. Du, P. et al. Comparison of Beta-value and M-value methods for quantifying methylation levels by microarray analysis. BMC Bioinformatics 11, 587 (2010)

    CAS  PubMed  PubMed Central  Google Scholar 

  48. Wauters, E. et al. DNA methylation profiling of non-small cell lung cancer reveals a COPD-driven immune-related signature. Thorax 70, 1113–1122 (2015)

    PubMed  Google Scholar 

  49. Yu, M. et al. Tet-assisted bisulfite sequencing of 5-hydroxymethylcytosine. Nat. Protocols 7, 2159–2170 (2012)

    CAS  PubMed  Google Scholar 

  50. Nazor, K. L. et al. Application of a low cost array-based technique - TAB-Array - for quantifying and mapping both 5mC and 5hmC at single base resolution in human pluripotent stem cells. Genomics 104, 358–367 (2014)

    CAS  PubMed  Google Scholar 

  51. Morris, T. J. et al. ChAMP: 450k Chip Analysis Methylation Pipeline. Bioinformatics 30, 428–430 (2014)

    CAS  PubMed  Google Scholar 

  52. Liu, F., Song, Y. & Liu, D. Hydrodynamics-based transfection in animals by systemic administration of plasmid DNA. Gene Ther . 6, 1258–1266 (1999)

    CAS  PubMed  Google Scholar 

  53. Pelicano, H., Carney, D. & Huang, P. ROS stress in cancer cells and therapeutic implications. Drug Resist. Updat. 7, 97–110 (2004)

    CAS  PubMed  Google Scholar 

  54. Booth, M. J. et al. Quantitative sequencing of 5-methylcytosine and 5-hydroxymethylcytosine at single-base resolution. Science 336, 934–937 (2012)

    ADS  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

We thank G. Peuteman, T. Van Brussel, J. Serneels and K. Kurz for assistance, C. Chang for NucPE1, G.-L. Xu for Tet-triple knockout ES cells. H.Z. and B.T. hold FWO-F postdoctoral fellowships. This work was supported by funding from the ERC (CHAMELEON 617595 to D.L.; EU-ERC269073 to P.C.; CHAMELEO 334420 to B.T.), from the FWO-F (G065615N, G070615N) to D.L., from the IUAP (P7/03) and the Flemish Government (Methusalem) to P.C., and from the DFG (EXC114 (CiPSM), grants CA275/8-5, GRK2062/1 and SPP1784) to T.C.

Author information

Authors and Affiliations

Authors

Contributions

B.T. and D.L. conceived and supervised the project, designed experiments and wrote the manuscript. B.T. and F.D.A. performed in vitro experiments and analysed data, helped by L.V.D.; M.L.C. and A.P. analysed Tet Michaelis–Menten kinetics; animal models were provided by E.H., F.A. (xenografts), M.M. (sFlk1), A.K. and P.C. (Phd2+/−); V.N.K. contributed ideas, L.S. and K.P.K. provided reagents; J.S. quantified nucleotides by LC–MS, supervised by T.C.; B.G. quantified metabolites. H.Z. analysed TCGA tumours; B.T., H.Z. and B.B. performed bioinformatics and statistics.

Corresponding authors

Correspondence to Bernard Thienpont or Diether Lambrechts.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Additional information

Reviewer Information Nature thanks R. S. Johnson, M. Rehli and Y. Xiong for their contribution to the peer review of this work.

Extended data figures and tables

Extended Data Figure 1 Hypoxia-induced changes in 5hmC, 5mC and TET expression.

Global 5hmC/C and 5mC/C content of DNA, TET1, TET2 and TET3 mRNA expression and hypoxia marker gene expression in 15 cell lines grown for 24 h under normoxic (21% O2, white) or hypoxic (0.5% O2, red) conditions. TET mRNA copy number is expressed relative to B2M for human cell lines (HepG2, HT-1080, MCF10A, H358, MCF7, Hep3B, A549, H1299, SK-N-Be2c and SHSY5Y), and to Hprt for mouse cell lines (LLC, N2A, 4T1, mES and Tet1-KO ES cells). Shown are cell lines derived from liver cancer (HepG2 and Hep3B), lung cancer (H358, A549, H1299 and LLC), breast cancer (MCF7 and 4T1), fibrosarcoma (HT1080), neuroblastoma (SK-N-Be2c and SHSY5Y), normal breast epithelium (MCF10A) and the inner cell mass of blastocyst-stage mouse embryos (mES and Tet1-KO ES cells). ALDOA and BNIP3 are expected to be increased, and HIF1A to be decreased upon hypoxia. The global 5fC content of ES cells is depicted, but was undetectable in cancer cell lines. Bars represent the mean ± s.e.m. of five different replicate samples. DNA and RNA from these replicates were extracted from cells derived from the same stock vial but grown on different days. *P < 0.05, **P < 0.01, ***P < 0.001 by paired t-tests.

Extended Data Figure 2 Impact of hypoxia on TET expression.

a, Changes in Tet1, Tet2 and Tet3 expression in mouse cell lines, at the protein level (top row, n = 6) and the mRNA level (bottom row, n = 5). Middle row: representative immunoblot images of Hif1a, Tet1, Tet2 and Tet3. α-Tubulin serves as loading control, and expression of the corresponding coding gene (Tuba1a) was used to normalize mRNA expression, enabling a direct comparison of relative protein and relative mRNA expression changes. For the same reason, mRNA expression was depicted relative to control conditions, in contrast to the absolute levels shown in Extended Data Fig. 1. Changes in Tet mRNA and protein expression correlate strongly (Pearson’s R = 0.855, P = 4 × 10−4). For example, both 4T1 and N2A cells displayed increased Tet2 expression at the protein and mRNA level. Likewise, ES cells showed no pronounced changes at the protein or mRNA level. The overall expression of Tet enzymes was not altered in any of these cell lines. For gel source data, see Supplementary Fig. 1. b, Hif1β ChIP-seq at the promoters of TET1, TET2 and TET3 and at hypoxia markers genes (Bnip3 and Aldoa), with peaks or promoter regions highlighted using coloured boxes. Green and red boxes correspond to overexpression and no overexpression (specified in the figure panel) of the corresponding gene, respectively, as determined using TaqMan in Extended Data Fig. 1. Scale: reads per million reads and per base pair. c, Left, TET2 expression in MCF7 cells transfected with control (white) or TET2-targeting (purple) siRNAs. Right, corresponding 5hmC levels as determined using LC–MS. d, 5hmC levels as determined by LC–MS, in wild-type (white) and Tet1-knockout (purple) ES cells grown under 21% (left) and 0.5% (right) O2 tensions. Bars in c and d represent the mean ± s.e.m. of five replicate samples from cells derived from the same stock vial but grown on different days. *P < 0.05, **P < 0.01, ***P < 0.001 by paired t-tests (a, c, d).

Extended Data Figure 3 Effects secondary to hypoxia.

ae, ROS production and redox state of MCF7 cells cultured for 24 h under control (21% O2, white) or hypoxic (0.5% O2, red) conditions. Shown are capillary gas chromatography mass spectrometry (GC–MS) quantifications of changes in the cellular energy state as represented by the adenylate energy charge (AEC) (calculated as [ATP + 0.5 × ADP]/[ATP + ADP + AMP]) (a); the reducing equivalents of the cell as represented by the relative NADH and NADPH levels (calculated as NADH/[NAD+ + NADH] and NADPH/[NADP+ + NADPH]); and the reductive capacity of the cell as represented by the levels of glutathione (calculated as GSH/[GSH + GSSG × 2]). b, c, Quantification (b) and representative FACS intensity traces (c) of total ROS levels in MCF7 cells exposed to hypoxia or H2O2, as assessed using 2`,7`-dichlorodihydrofluorescein diacetate (DCF-DA). d, Nuclear ROS in MCF7 cells as assessed using the nuclear peroxy emerald 1 probe (NucPE1)39. MCF7 cells were exposed to 21% (control) or 0.5% (hypoxia) O2 for 24 h, after which live cells were loaded with NucPE1 (5 μM) and Hoechst 33342 (10 μg ml−1) in O2 pre-equilibrated PBS for 15 min. After washing, control cells were incubated with H2O2 (0.5 mM in PBS) as a positive control, or with water (control and hypoxia cells) in PBS for 20 min. Cells were washed again and immediately imaged by confocal microscopy. Representative images are shown. Scale bar, 50 μm. e, The nuclear NucPE1 signal, averaged across >100 nuclei and expressed relative to control conditions. f, LC–MS quantification of 8-oxoG concentrations in DNA of cells lines cultured for 24 h under control (21% O2, white) and hypoxic (0.5% O2, red) conditions. 8-oxoG serves as a marker of nuclear ROS53. gi, GC–MS quantification of changes in the indicated metabolite levels in mouse ES cells (g), MCF10A cells (h) and MCF7 cells (i) grown for 24 h under control (21% O2, white), hypoxic (0.5% O2, red) or glutamine-free (21% O2, black) conditions. j, Quantities of α-ketoglutarate and 2-hydroxyglutarate in MCF7 cells, expressed relative to α-ketoglutarate levels in MCF7 cells grown under control conditions (21% O2). k, LC–MS quantification of 5hmC levels in response to hypoxia (0.5% O2) and glutamine-free culture conditions. l, Growth of cell lines cultured for 24 h under control (21% O2, white) and hypoxic (0.5% O2, red) conditions, as assessed using a sulforhodamine B colorimetric assay. Changes in cell density after 24 h are depicted relative to control conditions (21% O2). m, IOX2-induced changes in the global 5hmC content of DNA, in TET mRNA expression and in hypoxia marker gene expression of five cell lines treated for 24 h with DMSO (carrier, white) or IOX2 (50 μM, blue). n, 5mC hydroxylation activity of nuclear lysates from MCF7 cells grown for 24 h under 21% or 0.5% O2 (white or red). Bars represent the mean ± s.e.m. of 5 (b, k, m), 6 (a, ce), 16 (gj) or 24 (l) samples prepared on different days. *P < 0.05, **P < 0.01, ***P < 0.001 by t-test (b, e, h–m).

Extended Data Figure 4 Genomic profiles of 5mC and 5hmC.

Shown are results from DIP-seq of DNA from MCF7 cells cultured for 24 h under 21% or 0.5% O2 (control and hypoxia), with examples of 5hmC-DIP-seq (top) and 5mC-DIP-seq (bottom) read depths (FPM, fragments per base pair per million fragments) at regions surrounding the transcription start site of NSD1, FOXA1 and CDKN2A. These show 5hmC loss (FDR < 5%) and a 5mC gain that is more subtle, perhaps because the resolution of 5mC-DIP-seq is limiting: regions rich in 5hmC tend to be poorer in 5mC54, and thus have less substrate available for pull-down. 5mC-DIP-seq moreover captures all methylated sites, so most of the 5mC-DIP-seq signal does not derive from sites that are actively turning over 5hmC.

Extended Data Figure 5 Effect of hypoxia on hypermethylation frequency in tumours.

a, Immunofluorescence analysis of patient-derived tumour xenografts, stained for pimonidazole (white), 5hmC (red), DNA (propidium iodide, blue) and pan-cytokeratin (green). Shown are representative images of a breast and two endometrial tumour xenografts. The inset on the right shows box plots illustrating the signal in normoxic pimonidazole-negative nuclei (blue), and in hypoxic pimonidazole-positive nuclei (red). b, Hypoxia marker gene expression clusters, with the first three clusters used to define normoxic, intermediate and hypoxic tumours. c, Unsupervised clustering of 1,000 CpGs showing the highest average methylation increase in tumour versus corresponding normal tissues. The first three clusters were used to define tumours of low, intermediate and high hypermethylation. The colour bar above the clusters annotates each tumour as normoxic, intermediate or hypoxic, as determined in b. d, Box plots showing the relative expression (z-score) of genes in tumours in which they have either 0 or ≥1 hypermethylation event in their promoter, stratified into normoxic, intermediate and hypoxic tumours (blue, grey and red, respectively). Diamonds indicate mean, box plot wedges indicate 2× the standard error of the median. Genes with ≥1 hypermethylation events in their promoters have a lower average expression level (P < 0.01 for each tumour type). e, Fraction of genes having a promoter that is rich, intermediate or poor in CpGs, out of all gene promoters that are assessed on the 450k array, and out of all gene promoters that are frequently hypermethylated in the indicated tumour types. f, Fraction of 1,742 TET wild-type tumours and 39 TET mutant that are normoxic, intermediate and hypoxic. P > 0.2 for all comparisons. g, Cell proliferation marker gene46 expression clusters, with the first two clusters used to define high-proliferative and low-proliferative tumours. h, hypermethylation frequencies in low- and high-proliferative tumours, with asterisks representing P values from linear models correcting for variables specified in Supplementary Table 8. i, Partial correlation coefficient (partial R2) estimates of the relative contribution of tumour characteristics (annotated in TCGA) to the variance in hypermethylation observed in these tumours. Partial R2 values were obtained from linear model estimation using ordinary least squares, and expressed as a fraction of the total variance (that is, total R2) explained by the model when taking into account all indicated variables, as indicated between brackets under each tumour type. *P < 0.05, **P < 0.01, ***P < 0.001 by t-test (a) or by generalized linear model (h).

Extended Data Figure 6 Functional annotation of genes more frequently hypermethylated in hypoxic tumours.

a, Ontology terms enrichment analysis of genes that are more frequently hypermethylated at their gene promoters in hypoxic than normoxic tumours, for eight tumour types characterized in the TCGA pan-cancer effort. A representative set of terms is displayed, selected from terms enriched in most tumour types. P values as defined by the grey-scale insert. Enrichment calculated using topGO. b, Selected examples of hypermethylation frequencies in the promoter of key TSGs (PTEN, CDKN1C, ATM) more frequently hypermethylated in normoxic than hypoxic tumours. c, Hypermethylation frequency in the promoter of selected genes involved in the processes indicated. P < 0.05 for all genes (asterisks are not displayed). Bars in b and c represent the hypermethylation frequency ± s.e.m. P values in a by Fisher’s exact test.

Extended Data Figure 7 Effect of hypoxia on TET activity in human tumours.

a, The t-value of correlation between hypermethylation and expression of TET or DNMT genes across 3,141 tumours of 8 tumour types (bladder, breast, colorectal, head and neck, kidney, lung adenocarcinoma, lung squamous, and uterine carcinoma) profiled in TCGA for gene expression and DNA methylation, while correcting for tumour type, hypoxia and proliferation. The dotted line represents P < 0.05, negative t-values represent inverse correlations. b, Hypoxia metagene signature applied to 63 glioblastoma multiforme tumours from TCGA. c, Boxplots showing the relative expression (z-score) of genes in tumours in which they have either 0 or ≥1 hypermethylation event in their promoter, stratified into wild-type IHD1 tumours that are normoxic (n = 19), intermediate (n = 21) and hypoxic (n = 17) (blue, grey and red, respectively), and IDH1R138-mutated tumours (n = 4, yellow). Diamonds indicate mean, box plot wedges indicate 2× the standard error of the median. Genes with ≥1 hypermethylation events in their promoters have a lower average expression level. No hypermethylation events were detected in wild-type IHD1 normoxic tumours. d, Hypoxia metagene signature applied to 12 normoxic and 12 hypoxic non-small-cell lung tumours. *P < 0.05, ***P < 0.001 by t-test (c).

Extended Data Figure 8 5hmC, hypoxia and TSG hypermethylation in mouse breast tumours.

a, Frequency of hypermethylation events in the promoters of all genes, all oncogenes and all TSGs as annotated28, in 695 human breast tumours available through TCGA and stratified into normoxic, intermediate and hypoxic subsets. b, c, DNA was extracted from 53 tumours developing in MMTV-PyMT mice of the indicated ages (c) or weights (d) and sequenced to a depth of ~500×. Plotted are z-scores of hypermethylation (y axis, exponential) for 15 TSGs, relative to the tumours from 11-week-old mice. The dotted line represents the threshold for a Bonferroni-adjusted P < 0.05, and bold darker dots are used for tumours displaying significantly increased hypermethylation events. d, DNA extracted from 20 normal mammary glands from 14-week-old mice, PCR-amplified for 15 TSGs and sequenced to a depth of ~500x. Plotted are z-scores of hypermethylation relative to 11-week-old tumours. e, Staining of PyMT tumours for 5hmC (red), DNA (propidium iodide, blue), pimonidazole (white) and PyMT (green), and fraction of PyMT-positive cells in normoxic and hypoxic areas. The area outlined corresponds to the hypoxic, pimonidazole-positive section, arrowheads point to PyMT-negative cells. Scale bar, 25μm. The bar chart inset illustrates the relative number of PyMT-positive cells in normoxic and hypoxic areas (grey and red, respectively; n = 19). f, Ki67-positive cells in PyMT tumours: representative image of staining for DNA (propidium iodide, blue), Ki67 (red) and pimonidazole (green). Scalebar, 50 μm. The bar chart inset illustrates the quantification of Ki67-positive cells in normoxic and hypoxic areas (grey and red, respectively) across 6 tumours, analysing 3 fields of view with over 150 cells per field of view. g, CD45-positive cells in PyMT tumours: representative image of staining for DNA (propidium iodide, blue), 5hmC (red), pimonidazole (green) and CD45 (white). Scale bar, 100 μm. The bar chart inset illustrates the quantification of CD45-positive cells in normoxic and hypoxic areas (white and red, respectively) of 11 tumours, capturing on average ~2,500 nuclei per analysis. ***P < 0.001 in (a) by Fisher’s exact test, significance relative to all genes.

Extended Data Figure 9 Manipulation of tumour oxygenation in mouse breast tumours, and effects on 5hmC, TSG hypermethylation and confounders.

a, Plasma sFlk1 concentrations at the indicated times after hydrodynamic injection with an empty (n = 7) or sFlk1-overexpression plasmid (n = 5) (grey and red, respectively). b, c Quantification of tumour vessel number (b) and hypoxic areas (c) of tumours from transgenic MMTV-PyMT mice, hydrodynamically injected with an empty or sFlk1-overexpression plasmid, with representative images of blood vessels stained for CD31 (b) and hypoxic areas stained for pimonidazole adducts (c). Scale bar, 100 μm. d, Changes in RNA expression of hypoxia marker genes that are known to be downregulated (Mrc1) or upregulated (Bnip3, Car9, Ddit4) in hypoxic conditions. e, 5hmC levels (y axis) across mouse chromosome 18 (x axis) in 400 kb bins, with the location of RefSeq genes (middle), and differences in 5hmC levels (lower). 5hmC levels were determined using shallow TAB-seq, and chromosome 18 was selected because it has large stretches of gene deserts that illustrate the 5hmC depletion in these areas (n = 3). 5hmC levels decrease by 12.4% ± 3.5 after sFlk1 overexpression, although technical limitations of TAB-seq (incomplete 5hmC protection or bisulfite-conversion) may partially obscure the magnitude of effects. f, Hypermethylation in tumours developing in 12-week-old mice receiving hydrodynamic injection with an empty (n = 19) or sFlk1-overexpressing plasmid (n = 24) 3 weeks earlier. DNA was bisulfite-converted, PCR-amplified for the indicated oncogenes, and sequenced to a depth of ~500×. Plotted are z-scores of hypermethylation (y-axis, exponential), relative to the more normoxic tumours (that is, empty). The dotted line represents the threshold at 5% FDR, and bold darker dots the tumours displaying significantly increased hypermethylation events. gj, Relative weights of tumours from tg(MMTV-PyMT) mice, hydrodynamically injected with an empty (grey, n = 19) or sFlk1-overexpression plasmid (red, n = 24) (g), and corresponding RNA expression of Ptprc (the gene encoding CD45, n = 5) (h), of Tet enzymes (i, n = 15 for empty plasmid, n = 12 for sFlk1-overexpressing plasmid) and of cell proliferation markers (j, n = 5 for each). k–m, As in df but for 16-week old transgenic MMTV-PyMT mice of the indicated genotype. n = 5 (k), n = 3 for Phd2+/+; n = 4 for Phd2+/− (l) and n = 9 (m). n, As in d, but for 16-week-old Tie2-Cre;Tg(MMTV-PyMT) mice of the indicated genotypes (n = 5). o, DNA was extracted from 17 breast tumours developing in Tie2-Cre;Phd2fl/WT;Tg(MMTV-PyMT) mice (blue) and 13 breast tumours developing in Tie2-Cre;Phd2WT/WT;Tg(MMTV-PyMT) mice (grey), all 16 weeks old. DNA was bisulfite-converted, PCR-amplified for the indicated TSGs (left) and oncogenes (middle) and sequenced to a depth of >500×. Plotted are z-scores of hypermethylation (y axis, exponential), relative to the more normoxic, Phd2WT/fl, tumours. The dotted line represents the threshold for a Bonferroni-adjusted P < 0.05, and bold darker dots the tumours displaying significantly increased hypermethylation events. Right, 5hmC levels ± s.e.m. across a metagene in tumours of 16-week-old mice with the indicated genotype (n = 3 for Phd2fl/fl; n = 4 for Phd2WT/fl). pu, Relative weights of tumours from Phd2+/−;tg(MMTV-PyMT) mice and Phd2+/+;Tg(MMTV-PyMT) mice (n = 10 and 9 resp.) (pr) and from Tie2-Cre;Phd2fl/WT;Tg(MMTV-PyMT) and Tie2-Cre;PhdWT/WT;Tg(MMTV-PyMT) mice (n = 17 and 13, respectively) (s u), and the corresponding RNA expression of cell proliferation markers (n = 5, p, s), of Tet enzymes (n = 5, q, t) and of Ptprc (n = 5) (r, u). #P < 0.10, *P < 0.05, **P < 0.01, ***P < 0.001 by t-test.

Supplementary information

Supplementary Figure

This file contains the uncropped scans with size marker indication for Extended Data Figure 2b. (PDF 741 kb)

Supplementary Tables

This file contains Supplementary Tables 1-12. (XLSX 21073 kb)

PowerPoint slides

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Thienpont, B., Steinbacher, J., Zhao, H. et al. Tumour hypoxia causes DNA hypermethylation by reducing TET activity. Nature 537, 63–68 (2016). https://doi.org/10.1038/nature19081

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nature19081

This article is cited by

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

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