Epigenetic dysregulation is now widely recognized as a prominent feature of cancer cells. However, mutations in epigenetic modifier genes such as DNA and histone methyltransferases or 5-methylcytosine (5mC) oxidases can explain only a limited fraction of misregulated epigenomes in cancer. A pervasive feature of cancer epigenomes is CpG island DNA hypermethylation. Despite the fact that these methylation events have been cataloged in thousands of publications and in some instances, have been used to classify tumors, we still do not understand the mechanisms how these DNA methylation changes arise. In several types of tumors, a large percentage of the cancer-associated DNA hypermethylation events at CpG islands occur at Polycomb-marked genes. Therefore, one can view the occupancy of CpG islands by Polycomb complexes as one mechanism that protects them from de novo DNA methylation. However, this is likely not the only methylation-protective mechanism that operates at CpG islands. Mammalian genomes encode a relatively small set of 12 proteins that contain a CXXC-type zinc finger domain known to tightly bind to unmethylated CpG-rich DNA sequences. All of these proteins are known or suspected chromatin modifiers. The family includes the 5mC oxidases (the TET proteins), as well as histone lysine methyltransferases and demethylases. We propose that the common function of the majority of the CXXC proteins is to protect CpG islands from DNA methylation. Furthermore, metabolic disturbances of the TCA cycle can affect the function of multiple a-ketoglutarate-dependent dioxygenase enzymes. Remarkably, 5 of the 12 CXXC proteins belong to this class of enzymes. We hypothesize that there is functional redundancy in multiple mechanisms that protect CpG islands from DNA methylation. Overall, our main hypothesis is that a breakdown of multiple CpG island protection mechanisms, including downregulation of Polycomb components and dysfunction of CXXC proteins will lead to CpG island hypermethylation in cancer. We propose to identify these protective factors by a systematic approach using gene inactivation of Polycomb components and of CXXC proteins and comprehensive chromatin mapping studies combined with genome-wide DNA methylation assays after manipulating the different CpG island binding factors. The study will use human bronchial cells and will focus on lung cancer for which we have previously established comprehensive lists of tumor-methylated CpG islands. We consider that a combined deficiency of Polycomb components and of CXXC protein function will explain the common hypermethylation of Polycomb target genes observed in tumors. The role of accumulation of TCA cycle metabolites that inhibit dioxygenases and loss of Polycomb proteins will be tested to determine if DNA methylation can be targeted in this way to resemble a prevalent cancer-like pattern. These studies will provide insights into the long- sought mechanisms of DNA hypermethylation in cancer.
Aberrations in epigenetic control are key features of cancer cells. Among these, DNA hypermethylation of CpG islands is one of the most prevalent and pervasive occurrences in tumors, but the mechanistic basis of these events has remained unknown. Focusing on lung cancer, we will use genetic functional assays in combination with epigenetic profiling to investigate the mechanisms how CpG islands are protected from DNA methylation in cancer and how this protection breaks down in tumors.