Skip to main content

Thank you for visiting 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.

Super-enhancers delineate disease-associated regulatory nodes in T cells


Enhancers regulate spatiotemporal gene expression and impart cell-specific transcriptional outputs that drive cell identity1. Super-enhancers (SEs), also known as stretch-enhancers, are a subset of enhancers especially important for genes associated with cell identity and genetic risk of disease2,3,4,5,6. CD4+ T cells are critical for host defence and autoimmunity. Here we analysed maps of mouse T-cell SEs as a non-biased means of identifying key regulatory nodes involved in cell specification. We found that cytokines and cytokine receptors were the dominant class of genes exhibiting SE architecture in T cells. Nonetheless, the locus encoding Bach2, a key negative regulator of effector differentiation, emerged as the most prominent T-cell SE, revealing a network in which SE-associated genes critical for T-cell biology are repressed by BACH2. Disease-associated single-nucleotide polymorphisms for immune-mediated disorders, including rheumatoid arthritis, were highly enriched for T-cell SEs versus typical enhancers or SEs in other cell lineages7. Intriguingly, treatment of T cells with the Janus kinase (JAK) inhibitor tofacitinib disproportionately altered the expression of rheumatoid arthritis risk genes with SE structures. Together, these results indicate that genes with SE architecture in T cells encompass a variety of cytokines and cytokine receptors but are controlled by a ‘guardian’ transcription factor, itself endowed with an SE. Thus, enumeration of SEs allows the unbiased determination of key regulatory nodes in T cells, which are preferentially modulated by pharmacological intervention.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Figure 1: SE structure predicts lineage- and stage-specific transcription.
Figure 2: Transcription factors with major roles in TH-cell differentiation occupy SEs.
Figure 3: Bach2 is endowed with the highest p300-enriched SE in T cells.
Figure 4: Rheumatoid arthritis risk genes with SE structure are selectively targeted by the JAK inhibitor tofacitinib.

Accession codes

Primary accessions

Gene Expression Omnibus

Data deposits

All ChIP- and RNA-sequencing data sets have been deposited in the Gene Expression Omnibus under accession number GSE60482.


  1. 1

    Andersson, R. et al. An atlas of active enhancers across human cell types and tissues. Nature 507, 455–461 (2014)

    ADS  CAS  Article  Google Scholar 

  2. 2

    Hnisz, D. et al. Super-enhancers in the control of cell identity and disease. Cell 155, 934–947 (2013)

    CAS  Article  Google Scholar 

  3. 3

    Lovén, J. et al. Selective inhibition of tumor oncogenes by disruption of super-enhancers. Cell 153, 320–334 (2013)

    Article  Google Scholar 

  4. 4

    Whyte, W. A. et al. Master transcription factors and mediator establish super-enhancers at key cell identity genes. Cell 153, 307–319 (2013)

    CAS  Article  Google Scholar 

  5. 5

    Parker, S. C. et al. Chromatin stretch enhancer states drive cell-specific gene regulation and harbor human disease risk variants. Proc. Natl Acad. Sci. USA 110, 17921–17926 (2013)

    ADS  CAS  Article  Google Scholar 

  6. 6

    Dowen, J. M. et al. Control of cell identity genes occurs in insulated neighborhoods in Mammalian chromosomes. Cell 159, 374–387 (2014)

    CAS  Article  Google Scholar 

  7. 7

    Okada, Y. et al. Genetics of rheumatoid arthritis contributes to biology and drug discovery. Nature 506, 376–381 (2014)

    ADS  CAS  Article  Google Scholar 

  8. 8

    Rada-Iglesias, A. et al. A unique chromatin signature uncovers early developmental enhancers in humans. Nature 470, 279–283 (2011)

    ADS  CAS  Article  Google Scholar 

  9. 9

    Kieffer-Kwon, K. R. et al. Interactome maps of mouse gene regulatory domains reveal basic principles of transcriptional regulation. Cell 155, 1507–1520 (2013)

    CAS  Article  Google Scholar 

  10. 10

    Mousavi, K. et al. eRNAs promote transcription by establishing chromatin accessibility at defined genomic loci. Mol. Cell 51, 606–617 (2013)

    CAS  Article  Google Scholar 

  11. 11

    Hu, G. et al. Expression and regulation of intergenic long noncoding RNAs during T cell development and differentiation. Nature Immunol. 14, 1190–1198 (2013)

    CAS  Article  Google Scholar 

  12. 12

    Ciofani, M. et al. A validated regulatory network for Th17 cell specification. Cell 151, 289–303 (2012)

    CAS  Article  Google Scholar 

  13. 13

    Roychoudhuri, R. et al. BACH2 represses effector programs to stabilize Treg-mediated immune homeostasis. Nature 498, 506–510 (2013)

    ADS  CAS  Article  Google Scholar 

  14. 14

    Wei, L. et al. Discrete roles of STAT4 and STAT6 transcription factors in tuning epigenetic modifications and transcription during T helper cell differentiation. Immunity 32, 840–851 (2010)

    CAS  Article  Google Scholar 

  15. 15

    Vahedi, G. et al. STATs shape the active enhancer landscape of T cell populations. Cell 151, 981–993 (2012)

    CAS  Article  Google Scholar 

  16. 16

    McAllister, K. et al. Identification of BACH2 and RAD51B as rheumatoid arthritis susceptibility loci in a meta-analysis of genome-wide data. Arthritis Rheum. 65, 3058–3062 (2013)

    CAS  Article  Google Scholar 

  17. 17

    Franke, A. et al. Genome-wide meta-analysis increases to 71 the number of confirmed Crohn's disease susceptibility loci. Nature Genet. 42, 1118–1125 (2010)

    CAS  Article  Google Scholar 

  18. 18

    Sawcer, S. et al. Genetic risk and a primary role for cell-mediated immune mechanisms in multiple sclerosis. Nature 476, 214–219 (2011)

    ADS  CAS  Article  Google Scholar 

  19. 19

    Ferreira, M. A. et al. Identification of IL6R and chromosome 11q13.5 as risk loci for asthma. Lancet 378, 1006–1014 (2011)

    CAS  Article  Google Scholar 

  20. 20

    Cooper, J. D. et al. Meta-analysis of genome-wide association study data identifies additional type 1 diabetes risk loci. Nature Genet. 40, 1399–1401 (2008)

    CAS  Article  Google Scholar 

  21. 21

    Lovén, J. et al. Revisiting global gene expression analysis. Cell 151, 476–482 (2012)

    Article  Google Scholar 

  22. 22

    Jostins, L. et al. Host-microbe interactions have shaped the genetic architecture of inflammatory bowel disease. Nature 491, 119–124 (2012)

    CAS  Article  Google Scholar 

  23. 23

    Beecham, A. H. et al. Analysis of immune-related loci identifies 48 new susceptibility variants for multiple sclerosis. Nature Genet. 45, 1353–1360 (2013)

    CAS  Article  Google Scholar 

  24. 24

    McLean, C. Y. et al. GREAT improves functional interpretation of cis-regulatory regions. Nature Biotechnol. 28, 495–501 (2010)

    CAS  Article  Google Scholar 

  25. 25

    Ghisletti, S. et al. Identification and characterization of enhancers controlling the inflammatory gene expression program in macrophages. Immunity 32, 317–328 (2010)

    CAS  Article  Google Scholar 

  26. 26

    Nakayamada, S. et al. Early Th1 cell differentiation is marked by a Tfh cell-like transition. Immunity 35, 919–931 (2011)

    CAS  Article  Google Scholar 

  27. 27

    Wei, G. et al. Genome-wide analyses of transcription factor GATA3-mediated gene regulation in distinct T cell types. Immunity 35, 299–311 (2011)

    CAS  Article  Google Scholar 

  28. 28

    Hawkins, R. D. et al. Global chromatin state analysis reveals lineage-specific enhancers during the initiation of human T helper 1 and T helper 2 cell polarization. Immunity 38, 1271–1284 (2013)

    CAS  Article  Google Scholar 

  29. 29

    Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008)

    Article  Google Scholar 

  30. 30

    Zang, C. et al. A clustering approach for identification of enriched domains from histone modification ChIP-Seq data. Bioinformatics 25, 1952–1958 (2009)

    CAS  Article  Google Scholar 

  31. 31

    Shen, L., Shao, N., Liu, X. & Nestler, E. ngs.plot: Quick mining and visualization of next-generation sequencing data by integrating genomic databases. BMC Genomics 15, 284 (2014)

    Article  Google Scholar 

  32. 32

    Morris, A. P. et al. Large-scale association analysis provides insights into the genetic architecture and pathophysiology of type 2 diabetes. Nature Genet. 44, 981–990 (2012)

    CAS  Article  Google Scholar 

Download references


The authors thank B. Afzali, A. Nussenzweig, A. Poholek, S. Canna, A. Richard and E. Mathe for critically reading this manuscript. We are grateful to R. Faryabi, H.-Y. Shih, W. Resch, M. Ombrello, Z. Deng and E. Remmers for their contributions to experimental and analytical components of this study. We also thank H. Sun, G. Gutierrez-Cruz, J. Simone, J. Lay and K. Tinsley for their excellent technical support. This study used the high-performance computational capabilities of the Biowulf Linux cluster at the National Institutes of Health. R.R. is supported by a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (grant number 105663/Z/14/Z). This work was supported by the Intramural Research Program of NIAMS and by NCI grant R01 CA186714.

Author information




G.V., V.S., F.S.C. and J.J.O’S. participated in the study design. Y.K., Y.F. and K.J. performed sequencing experiments. Z.T. and Y.R. supervised and performed sequencing experiments. Y.F. and M.G. performed tofacitinib-related experiments. G.V. performed computational analysis. S.C.J.P., M.R.E. and S.R.D. participated in statistical analysis relevant to human genetics. R.R. and N.P.R. supervised and performed experiments involving Bach2-deficient cells. Y.K., Y.F. and S.R.D. participated in writing of the methodology. G.V., V.S. and J.J.O’S. wrote the manuscript and all authors reviewed it. J.J.O’S. supervised the project.

Corresponding authors

Correspondence to Golnaz Vahedi or John J. O’Shea.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Extended data figures and tables

Extended Data Figure 1 SE structures are lineage-specific.

a, Histone acetyltransferase p300 is distributed asymmetrically across the genome in CD4+ T cells with a subset of enhancers (SEs) containing exceptionally high amounts of p300 binding. Graph demonstrates the ranked distribution of p300 binding measured by ChIP-seq in TH2 and TH17 cells. b, Closely related CD4+ T-cell populations have distinct SE landscapes. Common and cell-type-specific SE domains in T-cell subsets are illustrated for various fractions of overlapping genomic regions (f = 0.1, 0.3, 0.5 and 0.7). The overlapping pattern of SEs across CD4+ T cells was statistically significant when these annotations were shuffled across the genome (P value < 10−5). c, Lineage-specific presence of SEs for master transcription factor genes in T cells. Genomic loci of genes encoding T-BET, GATA3 and ROR-γt exhibit SE structures in TH1, TH2 and TH17 cells, respectively. Black bar represents the genomic location of SEs. d, The genomic locus of the gene encoding gp130, Il6st, accommodates an SE with a high level of transcription. Black bar represents the genomic location of SEs.

Extended Data Figure 2 Transcription factor enrichment at SEs.

a, Lineage-specific transcription factors are enriched at cell-type-specific SEs. Binding patterns of STAT4, STAT6 and STAT3 revealed preferential binding at TH1-, TH2- and TH17-specific SE regions, respectively. Furthermore, transcription factors T-BET, GATA3, and HIF-1α and ROR-γt were enriched at lineage-specific SEs. Strong binding of BATF, BACH2 and IRF4 was present in SEs of all three cell types. Maps of cell-type-specific SEs were constructed as described in Fig. 1b. Normalization of y-axis takes into account the variable sizes of genomic regions and also the corresponding library size (that is, the total read count) (Methods). b, CTCF binding demarcates the boundaries of SEs. Normalized binding profile of CTCF protein revealed the enrichment of CTCF at boundaries of SE regions. c, Comparing the enrichment of transcription factors at constituent enhancers of SEs and TEs reveals the preferential binding of STAT3 at SEs while other transcription factors demonstrated comparable binding at SEs and TEs.

Extended Data Figure 3 Identity of SE-associated genes.

a, SEs delineate genes that have a central role in the biology of specific cell lineages. Gene ontology (GO) functional categories relevant to cytokine binding are enriched at SE-associated genes in T cells. In ES cells, SE structures primarily encompass DNA-binding proteins and transcriptional repressor functions. In macrophages, chemokine and cytokine activity were the most prominent categories. Using a complementary approach to that described in Fig. 2a, we characterized genes in proximity to SEs. The top GO molecular functions using GSEA were chosen. To calculate the statistical significance of these gene categories, we shuffled the SE regions around the genome 105 times, delineating the gene sets in proximity to the random genomic domains. We then assessed the relative proportion of a gene set captured in the actual data versus the shifted SEs. −Log10 P values for this permutation test are reported in the bar graph. b, GO functional category relevant to cytokines binding is enriched at SE-associated genes in T cells and, to a lesser extent, in macrophages but not in mouse ES cells (mESCs) or myotubes. To explore whether “cytokine binding” is specific to the SE structure in CD4+ T cells, we explored its association within the SE structures of other cell types. The GO molecular function associated with cytokine binding (GO:0019955) was chosen. To calculate the statistical significance of this gene category, we shuffled the SE regions of mouse ES cells, macrophage, myotubes and CD4+ T cells around the genome 105 times, delineating the gene sets in proximity to the random genomic domains associated with each cell type. We then assessed the relative proportion of the gene set captured in the actual data versus the shifted SEs. P values for this permutation test are reported in the bar graph. c, d, BACH2 preferentially represses SE genes. Wild-type and Bach2-deficient CD4+ T cells were polarized to induced regulatory T cells (iTreg cells) and were subjected to total RNA extraction. RNA standards ‘spiked-in’ were added in proportion to the number of cells present in the sample. The resulting transcriptome data measured by RNA-seq were processed by using standard normalization methods and then renormalized based on spiked-in reads (RPKM) (see Methods). Transcript abundance measured by RNA-seq was evaluated in wild-type and Bach2-deficient cells at SE- and TE-associated genes compared to remaining genes (RPKM). Cumulative distribution (c) and violin plots (d) show the (log2) fold change in gene expression for wild-type versus Bach2-deficient cells for these three groups of genes. SE genes are preferentially affected by loss of BACH2 compared with TE genes (P value < 2.2 × 10−16, Kolmogorov–Smirnov test) or remaining genes (P value < 2.2 × 10−16, Kolmogorov–Smirnov test). P values for the violin plots (d) were calculated using the Wilcoxon rank-sum test. e, BACH2 selectively affects SE genes and such selectivity remains statistically significant when controlling for the higher levels of gene expression for the SE genes. Genes were ranked based on their transcriptional activity in Treg cells. We focused on the top 500 highly expressed genes and explored the effect of BACH2 on three categories among them: genes with SEs (77), with TEs (125), and without either SEs or TEs (298). Expression levels among these three categories of genes were comparable (Wilcoxon rank-sum test, P value = 0.644). However, BACH2 selectively affected highly expressed SE genes in contrast to those with TEs or no enhancers (Kolmogorov–Smirnov test, P value = 9.813 × 10−7 and 4.669 × 10−8).

Extended Data Figure 4 BACH2 acts as a guardian transcription factor.

a, b, Loss of BACH2, STAT4 and STAT6 has the most selective impact on the expression of SE genes. a, The fold change in expression (in RPKM) between wild-type and knockout samples was calculated for SE genes and an equal number of randomly selected TE genes. b, For each transcription factor, the difference between SEs and TEs was quantitated using Kullback–Leibler distance. The larger difference between SEs and TEs for BACH2, STAT4 and STAT6 suggests the more selective impact of these transcription factors on SEs. STAT4 and T-BET transcriptome data were under TH1, STAT6 under TH2, STAT3, BATF and IRF4 under TH17 and BACH2 under iTreg conditions. c, SE-associated genes in CD4+ T cells are repressed by BACH2. To ensure accurate inference of the effect of BACH2 on the transcriptome, spiked-in RNA standards were added. The gene set enrichment analysis (GSEA) of SE-associated genes revealed that SE genes were enriched in genes repressed by BACH2 when transcript levels were renormalized using spiked-in RNA standards. d, BACH2 acts as a repressor of SE-associated genes. Comparison of the transcriptome data measured by RNA-seq in wild-type and Bach2-deficient cells (DE-seq analysis for three wild-type and knockout samples, FDR < 0.05 and fold change > 1.5) revealed that 348 SE genes were repressed while 176 were induced by this protein. Integration of BACH2 binding data measured by ChIP-seq characterized the direct targets of BACH2. e, The GSEA of TE-associated genes revealed that TE genes are not enriched in genes repressed by BACH2. f, g, BACH2-associated transcriptional repression at some SE domains correlates with the downregulation of nearby genes such as Rbpj (f) and Socs1 (g). h, Genes and noncoding transcripts endowed with SE architecture in CD4+ T cells are tightly and negatively controlled by the ‘guardian’ transcription factor BACH2, which itself has a rich cassette of regulatory elements. Examples were selected based on direct binding of BACH2 at the gene body or SE regions measured by ChIP-seq.

Extended Data Figure 5 RA risk genes with SE structure are selectively targeted by a JAK inhibitor, tofacitinib.

a, Genetic variants in high linkage disequilibrium (LD) with SNPs associated with autoimmune disorders such as RA, IBD, MS and T1D exhibit preferential enrichment in SEs versus TEs in human CD4 T cells. Variants in LD with SNPs in each disease were determined from the 1000 Genomes Project using r2 = 0.9 and distance limit = 500 by SNAP toolbox. The heatmap depicts the percentages of SNPs and total number of SNPs per 10 Mb within SEs and TEs. b, Tofacitinib treatment has a selective impact on SE versus TE genes in human T cells. Violin plots depict the fold change (log2) in transcript levels due to tofacitinib treatment at SE versus TE genes in CD4+ T cells. The P values were calculated based on the Wilcoxon signed-rank test. c, Highly expressed genes in T cells with SEs are selectively affected by tofacitinib. For each donor, the top 100 highly expressed genes in non-treated cells were selected and categorized as having SEs or not. The P values were calculated based on the Wilcoxon signed-rank test. d, RA risk genes with SEs are selectively targeted by a JAK inhibitor, tofacitinib. Violin plots depict the fold change in expression (log2) after tofacitinib treatment of human CD4+ T cells at RA risk genes with or without SEs (a donor with no spiked-in standard in RNA-seq). P values were calculated using the F-test.

Extended Data Figure 6 Tofacitinib selectively affects autoimmune disease risk genes with SE structure in T cells.

a, RA, IBD, and MS risk genes are linked to SEs in CD4+ T cells. The candidate genes associated with RA7, IBD22, MS23 and T2D32 were chosen based on recent meta-analyses of GWAS data. More than half of RA risk genes (53/98) contained SEs in CD4+ T cells. In line with the enrichment of SNPs associated with IBD and MS in T-cell SEs (Fig. 4a), around half of IBD (91/216) and MS risk genes (36/87) were associated with SEs in T cells. In contrast, T2D risk genes showed little association with SEs (4/65) (Fisher’s exact test, P value = 0.4). bd, RA and IBD risk genes with SEs are selectively targeted by a JAK inhibitor, tofacitinib. Cumulative plots depict the fold change in expression (log2) at RA (b), IBD (c) and MS (d) risk genes with or without SEs after 0.3 µM tofacitinib treatment of human CD4+ T cells (P values, Kolmogorov–Smirnov test).

PowerPoint slides

Source data

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Vahedi, G., Kanno, Y., Furumoto, Y. et al. Super-enhancers delineate disease-associated regulatory nodes in T cells. Nature 520, 558–562 (2015).

Download citation

Further reading


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.


Quick links