Pancreatic cancer foiled by a switch of tumour subtype

Pancreatic ductal adenocarcinoma (PDAC) is one of the most deadly cancers in Western society1 because it tends to be diagnosed late and responds poorly to therapy. PDAC tumours fall into two main groups24: a ‘classical’ subtype, and a more aggressive ‘squamous’ subtype in which pancreatic cells fail to undergo proper differentiation. The squamous subtype often involves mutations in members of the COMPASS-like complex — a group of methyltransferase and demethylase enzymes that, respectively, add or remove methyl groups from lysine amino-acid residues on histone proteins, around which DNA is packaged. Such histone modification can lead to changes in the expression of histone-associated genes involved in pancreatic-cell differentiation. Writing in Cancer Cell, Andricovich et al.5 demonstrate the role of mutations in one member of this complex, KDM6A, in driving the squamous PDAC subtype.

The KDM6A gene is found on the X chromosome, and, in males, the presence of a KDM6A mutation can co-occur with mutation of a related gene on the Y chromosome, UTY. Andricovich et al. found that KDM6A and UTY mutations were associated with the squamous subtype of PDAC, and with shortened length of patient survival. They then used mouse models to confirm that functional KDM6A acts to suppress the development of PDAC. Mice harbouring Kdm6a gene mutations developed aggressive, poorly differentiated squamous tumours that showed protein- and gene-expression patterns characteristic of human tumours of the squamous subtype2. These defects were more pronounced in females than in males, consistent with the fact that females carry two copies of Kdm6a in their cells and males have one copy of Kdm6a on the X chromosome and Uty on the Y.

KDM6A is a demethylase that removes a methyl modification dubbed H3K27me3 from lysine residue 27 on histone H3 proteins. But Andricovich and colleagues found that only a small percentage of H3K27me3 marks were altered in cells from Kdm6a-mutant mice, compared with controls. This observation led the authors to posit that altered KDM6A demethylase activity was not the driver of Kdm6a-mutant PDAC. Instead, they found that loss of Kdm6a resulted in changes in other histone modifications, specifically at groups of gene-regulatory DNA sequences called super-enhancers, whose activation promotes the expression of certain genes that are highly expressed in PDAC.

In particular, the researchers observed changes in the distribution across super-enhancers of a different methyl modification (dubbed H3K4me1) and a modification involving acetyl groups at lysine 27 of histone H3 (H3K27ac). Such changes were associated with a repositioning of the COMPASS-like complex to these regions. The authors found that the altered histone modifications led to activation of some super-enhancers, and, in some cases, to an increased reach — an ability to regulate distant genes that the super-enhancer cannot normally influence. These findings indicate that KDM6A exerts its tumour-suppressive role not only through its demethylase activity, but also by altering the position of the COMPASS-like complex, enabling other enzymes to modify histones. The authors also showed that the increase in the reach and activation of super-enhancers led to the activation of genes involved in squamous-subtype-like differentiation.

Because Kdm6a-mutant PDAC in mice was not associated with significant H3K27me3 demethylation, the authors hypothesized that the alternative functions of the aberrant COMPASS-like complex promoted PDAC, and might therefore be vulnerable to drug treatment. This hypothesis is supported by the fact that mutant UTY, which helps to drive PDAC in males, lacks demethylase activity. Andricovich et al. therefore analysed the ability of various drugs that target other histone modifications to prevent the growth of KDM6A-deficient human cancer cells in vitro.

The authors found that cells harbouring mutations in KDM6A or other genes of the COMPASS-like complex were highly sensitive to inhibitors of BET-family proteins. These proteins bind to histone lysine residues that have been modified by acetyl groups, and recruit the cell’s transcriptional machinery to promote gene expression. Various studies6 have shown that BET inhibitors can displace the BET protein BRD4 from acetylated lysines at super-enhancer regions, thereby reducing the expression of cancer-causing genes (oncogenes) such as MYC. Because KDM6A mutations lead to altered lysine acetylation at super-enhancers, it makes sense that these drugs could be effective in this setting. Indeed, Andricovich et al. showed that the BET inhibitor JQ1 decreased BRD4 binding to the super-enhancers that regulate MYC and other oncogenes, and so decreased the expression of these genes.

Finally, the authors demonstrated that this drug treatment was also effective in vivo. The tumours of Kdm6a-deficient mice treated with JQ1 were smaller than those of mice that did not receive the drug, and had well-differentiated features typical of the classical PDAC subtype. This indicates that BET inhibitors have the potential to reverse the effects of the histone-modification remodelling that occurs in the squamous subtype (Fig. 1a). Targeting histone modifications and altered gene-regulatory networks to cause a ‘class switch’ to a more differentiated, less aggressive subtype of cancer might provide a promising therapeutic strategy. In support of the idea that modulating these factors can alter cancer progression, other studies have shown that enhancer reprogramming and large-scale losses of DNA methylation play a part in the spread of cancer7,8.

Figure 1 | KDM6A protein in pancreatic cancer. a, Pancreatic ductal adenocarcinomas (PDACs) can be catagorized into two main subtypes: classical tumours and more-aggressive, squamous tumours. Andricovich et al.5 have provided evidence from mice and humans that mutations in the gene KDM6A cause changes in the patterns of molecular modifications to histone proteins, around which DNA is packaged. This histone remodelling leads to the expression of genes associated with squamous PDAC. However, the authors show that treatment with a small molecule called JQ1 prevents this subtype switch. b, This finding adds to the list of PDAC subgroups that can be targeted with drug treatments. Most PDAC tumours involve mutations in the gene KRAS, which cannot be targeted by drugs, but some KRAS wild-type tumours lack this mutation and carry other mutations that can be targeted. In addition, two subgroups of KRAS-mutant tumours carry defects in DNA-repair pathways, which can be targeted by different drugs.

Our increased understanding of the molecular underpinnings of cancer has hugely improved treatments for many tumours, although in PDAC the relative lack of obvious drug targets has presented a challenge. There are some cases of PDAC that involve oncogenes for which inhibitors do exist9. However, most PDAC tumours harbour oncogenic mutations in the gene KRAS, for which inhibitors are not available. But there are two clear groups of people with KRAS-mutant PDAC tumours characterized by deficiencies in specific DNA-repair pathways that can be targeted by drugs10,11 (Fig. 1b). Patients harbouring KDM6A mutations (and possibly other mutations in genes of the COMPASS-like complex) might represent another subgroup, who would benefit from therapies targeting BET function. Moreover, BET inhibitors could have broader activity if combined with other inhibitors of histone remodelling, as previously reported12.

It is to be hoped that more molecular biomarkers will soon be discovered that, like KDM6A mutations, can predict tumour responsiveness to a particular therapy. This research avenue provides cause for optimism that improved outcomes for people with pancreatic cancer will be the norm — and not the exception — in the near future.

Nature 557, 500-501 (2018)

doi: https://doi.org/10.1038/d41586-018-05129-6


  1. 1.

    Siegel, R. L., Miller, K. D. & Jemal, A. CA Cancer J. Clinicians 67, 7–30 (2017).

    Article  Google Scholar 

  2. 2.

    Bailey, P. et al. Nature 531, 47–52 (2016).

    PubMed  Article  CAS  Google Scholar 

  3. 3.

    Collisson, E. A. et al. Nature Med. 17, 500–503 (2011).

    PubMed  Article  CAS  Google Scholar 

  4. 4.

    Moffitt, R. A. et al. Nature Genet. 47, 1168–1178 (2015).

    PubMed  Article  CAS  Google Scholar 

  5. 5.

    Andricovich, J. et al. Cancer Cell 33, 512–526 (2018).

    PubMed  Article  CAS  Google Scholar 

  6. 6.

    Shi, J. & Vakoc, C. R. Mol. Cell 54, 728–736 (2014).

    PubMed  Article  CAS  Google Scholar 

  7. 7.

    McDonald, O. G. et al. Nature Genet. 49, 367–376 (2017).

    PubMed  Article  Google Scholar 

  8. 8.

    Roe, J.-S. et al. Cell 170, 875–888 (2017).

    PubMed  Article  CAS  Google Scholar 

  9. 9.

    The Cancer Genome Atlas Research Network. Cancer Cell 32, 185–203 (2017).

    PubMed  Article  CAS  Google Scholar 

  10. 10.

    Waddell, N. et al. Nature 518, 495-501 (2015).

    PubMed  Article  CAS  Google Scholar 

  11. 11.

    Le, D. T. et al. N. Engl. J. Med. 372, 2509–2520 (2015).

    PubMed  Article  CAS  Google Scholar 

  12. 12.

    Mazur, P. K. et al. Nature Med. 21, 1163–1171 (2015).

    PubMed  Article  CAS  Google Scholar 

Download references


Nature Briefing

An essential round-up of science news, opinion and analysis, delivered to your inbox every weekday.

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing