Keywords

FormalPara What You Will Learn in This Chapter

Alterations in chromatin function and epigenetic mechanisms are a hallmark of cancer. The disruption of epigenetic processes has been linked to altered gene expression and to cancer initiation and progression. Recent cancer genome sequencing projects revealed that numerous epigenetic regulators are frequently mutated in various cancers. This information has not only started to be utilized as prognostic and predictive markers to guide treatment decisions but also provided important information for the understanding of the molecular mechanisms of epigenetic regulation in both physiological and pathological conditions. Furthermore, the reversible nature of epigenetic aberrations has led to the emergence of the promising field of epigenetic therapy that has already provided new therapeutic options for patients with malignancies characterized by epigenetic alterations, laying the basis for new and personalized medicine.

8.1 Epigenetics and Cancer

Cancer is a group of more than 100 different and distinct diseases in which abnormal cells divide without control and can invade nearby tissues. Cancer is the second leading cause of death globally. In the recent years, there have been tremendous efforts and remarkable advances to understand and treat this disease. In particular, the newest sequencing technologies have made possible to obtain a complete DNA sequence of large numbers of cancer genomes. These analyses identified genetic alterations not only between normal and cancer genomes but also between genomes of tumors from patients affected by the same type of cancer. This tremendous effort had provided relevant information for the history of tumor development at molecular level and led the basis for a personalized treatment approach that allows doctors to select treatments based on a molecular understanding of their disease.

The earliest indications of an epigenetic link to cancer originated from gene expression and DNA methylation studies. The past decade has seen a remarkable acceleration in the validation of the concept that cancer is not only a genetic disease but also one of epigenetic abnormalities. This view has been significantly strengthened by whole-genome sequencing results showing that numerous epigenetic regulators are frequently the target of mutations and epimutations in cancer cells, with an intriguing interplay between the two. The abundance of cancer mutations involving these genes practically affects all levels of epigenetic regulation, including key players in DNA methylation, histone modifications, and chromatin organization but also substrates for these modifications, such as histones. The identification of mutations in writers, readers, and erasers, as well as alterations in the epigenetic landscape in cancers, do not only imply a causative role for these factors in cancer initiation and progression but also provide potential targets for therapeutic intervention.

In this chapter, we describe the best studied epigenetic and chromatin alterations found in cancers (DNA methylation, histone H3K27 trimethylation, histone acetylation, and chromatin remodeling factors), how these alterations occur, and how they contribute to cancer initiation and progression. We also provide examples of the development of epigenetic inhibitors and strategies for their therapeutic application.

8.2 DNA Methylation and Cancer

Aberrant DNA methylation was the first epigenetic abnormality to be identified in human cancers (Feinberg and Vogelstein 1983). DNA methylation provides a stable gene silencing mechanism that plays an important role in regulating gene expression and chromatin architecture in association with histone modifications and other chromatin associated proteins (see book ► Chap. 1 of Wutz). Alterations detected in cancers can be promoter hypermethylation and the consequent silencing of tumor suppressor genes (► Sect. 8.2.1), global hypomethylation that has been associated with genomic instability (► Sect. 8.2.2), and alterations of DNA methylation at imprinting control regions (see book ► Chap. 5 of Grossniklaus) with a consequent loss of imprinting (► Sect. 8.2.3). Regions of low-density methylation near CpG islands, known as ‘shores’, exhibit great variation in DNA methylation, including hypomethylation and hypermethylation, across different types of cancers.

8.2.1 DNA Hypermethylation in Cancer

Many genes associated with CpG islands undergo de novo methylation in cancer. The “CpG island methylator phenotype (CIMP)” is defined as frequent methylation of multiple CpG islands and has been found in many types of cancers (◘ Fig. 8.1). The CIMP status is uniquely associated with specific clinicopathological characteristics in individual cancer types and has the potential to provide information for cancer diagnosis and the stratification of patients for specific therapeutic treatments.

Fig. 8.1
figure 1

Normal and cancer genomes exhibit distinct DNA methylation profiles. Schematic showing the difference in DNA methylation at defined gene regulatory elements and repeats between normal and cancer cells. These alterations include hypermethylation of CpG islands and hypomethylation at transposable elements and pericentromeric heterochromatin. These changes cause transcriptional silencing of tumor suppressor genes, the increase in the expression of oncogenes, and genome and chromosome instability. White circle, unmethylated CpG; black circle, methylated CpG

Full size image

Since DNA methylation is associated with gene repression, it has been suggested that new/aberrant acquisition of DNA methylation contributes to the repression of genes that are active in normal tissue and probably involved in tumor suppression activitiesFootnote 1. Accordingly, hypermethylation in cancers has been associated with the suppression of genes involved in cancer-related pathways, such as cell cycle control, DNA repair, apoptosis, and angiogenesis. The profiles of hypermethylation of the CpG islands in tumor suppressor genes are specific to the cancer type. Examples of this are the methylation at the promoter of the tumor suppressor gene VHL (associated with von Hippel-Lindau disease) in renal cancer, the cell cycle control gene p16 in many types of cancer, the DNA mismatch repair MutL Homolog 1 (MLH1) gene in colorectal cancer, and breast-cancer susceptibility 1 (BRCA1) gene in breast and ovarian cancers. Although these are important examples of DNA methylation-mediated gene repression, it was also shown that a large fraction of genes acquiring DNA methylation in cancer were already transcriptionally repressed in the normal tissue. It was proposed that the repression of these genes was mainly acting through PcG proteins that establish H3K27me3 (see book ► Chap. 3 of Paro). It was also shown that genes targeted by DNA methylation in cancer have a low, poised transcription state in embryonic stem cells and are implicated in the maintenance of stemness and self-renewal in normal stem cells. Their reactivation occurs at defined stages of development through the removal of the repressive PcG-complexes. Thus, the acquisition of DNA methylation at genes that are already in a repressed state might serve to either further downregulate and/or establish a stable repression that prevents activation, thereby contributing to a stem cell-like state of cancer. However, mechanisms that target de novo methylation at genes critical to cancer remain yet elusive.

8.2.2 DNA Hypomethylation in Cancer

Loss of 5-methylcytosine (5mC) was the first epigenetic abnormality to be identified in human cancer. The loss of DNA methylation is mainly occurring at repetitive DNA sequences, coding regions, and introns (◘ Fig. 8.1). The extensive demethylation of DNA observed in tumor progression was suggested to be a source of the continually generated cellular diversity associated with cancers. During the development of a neoplasmFootnote 2, the degree of hypomethylation of genomic DNA increases as the lesion progresses from a benign proliferation of cells (hyperplasia)Footnote 3 to an invasive and potentially metastatic cancerFootnote 4 (Fraga et al. 2004).

Hypomethylation of DNA in cancer has several implications, including gene activation, loss of imprinting, reactivation of transposable elements, and genome instability. DNA methylation is particularly concentrated at repetitive elements and may limit their genomic activity. Loss of DNA methylation can favor mitotic recombination, leading to deletions, translocations, and chromosomal rearrangements. Furthermore, hypomethylation of DNA in malignant cells can reactivate transposable elements, such as the long interspersed nuclear element LINE 1, which accounts for 17% of the human genome (◘ Fig. 8.1). The large number of these repeats can, when reactivated, promote translocations to other genomic regions, thereby affecting genome integrity. Hypomethylation can also occur at pericentromeric regions of the chromosome (near the site of attachment of the mitotic spindle) and rearrangements in the pericentromeric heterochromatin of chromosomes 1 or 16 are found in many types of cancers.

8.2.3 Loss of Imprinting Through Alterations of DNA Methylation

The Beckwith-Wiedemann syndrome (BWS) is an important example of disruption of genome imprinting through alteration in DNA methylation (see book ► Chap. 5 of Grossniklaus). BWS is characterized by macrosomia, macroglossia, and abdominal wall defects, and exhibits a predisposition to tumorigenesis (Soejima and Higashimoto 2013). In particular, the development of embryonal tumors (i.e. Wilms’ tumor, hepatoblastoma, and rhabdomyosarcoma) is a recurring feature of BWS. The relevant imprinted chromosomal region in BWS that is epigenetically altered is 11p15.5, which consists of two imprinting domains, CDKN1C/KCNQ1OT1 and IGF2/H19 (◘ Fig. 8.2) (see also book ► Chap. 5 of Grossniklaus).

Fig. 8.2
figure 2

Loss of imprinting through alteration of DNA methylation in cancer. Schema representing aberrant imprinting patterns at the chromosome 11p15 imprinting cluster found in Beckwith-Wiedemann Syndrome (BWS) patients. Loss of methylation at the differentially methylated regions (DMRs) KvDMR1 or H19-DMR and the resulting changes in the expression of imprinted genes are shown. Pink boxes represent maternally genes. Blue boxes are paternally expressed genes. White circle, unmethylated CpG; black circle, methylated CpG

Full size image

In the imprinting domain CDKN1C/KCNQ1OT1, CDKN1C (Cyclin Dependent Kinase Inhibitor 1C) is maternally expressed, whereas the long non-coding RNA KCNQ1OT1 is paternally expressed. The imprinting control region (ICR) of CDKN1C/KCNQ1OT1 domain is the differentially methylated region (DMR) KvDMR1, which overlaps with the promoter of KCNQ1OT1. KvDMR1 is methylated on the maternal allele and unmethylated on the paternal allele. Fifty percent of BWS patients display a loss of DNA methylation on KvDMR1 at the maternal allele, which leads to the expression of KCNQ1OT1 and repression of the cell cycle inhibitor CDKN1C. Representative phenotypes of this epigenetic alteration include omphalocele and hemihyperplasiaFootnote 5. Of note is that downregulation of CDKN1C has been reported in several other cancers. It has been proposed that the repression of CDKN1C on the paternal chromosome is mediated by the long non-coding RNA (lncRNA) KCNQ1OT1, which coats the surrounding locus. KCNQ1OT1, through the interaction with H3K9- and H3K27-specific histone methyltransferases G9a and EZH1/2, the latter part of Polycomb Repressive Complex 2 (PRC2) (see book ► Chap. 3 of Paro), leads to the repression of CDKN1C in cis.

In about 5% of BWS patients, gain of DNA methylation occurs on the normally unmethylated maternal H19-DMR, which is the ICR of IGF2/H19 domain (◘ Fig. 8.2). H19-DMR is located 2 kb upstream of H19 . In normal cells, differential parental methylation of H19-DMR leads to paternal expression of IGF2 (the insulin-like growth factor gene), and maternal expression of the non-coding RNA gene H19. On the maternal allele, IGF2 is repressed due to CTCF binding the unmethylated H19DMR. The insulator formed by CTCF blocks enhancers downstream of H19 from accessing IGF2 promoters (see also ► Chap. 5 of Grossniklaus). On the paternal allele, IGF2 is expressed since H19-DMR is methylated, which impairs CTCF binding and results in the contact of enhancers with IGF2 promoters (Bell and Felsenfeld 2000). Gain of methylation of H19-DMR on the maternal allele induces loss of imprinting in BWS, leading to biallelic expression of IGF2 and reduced expression of H19. Loss of imprinting of IGF2 is also a risk factor for colorectal cancer and disrupted genomic imprinting contributes to the development of Wilms’ tumor. The mechanism by which gain of methylation at H19-DMR occurs is still unknown.

8.2.4 Mutations in the DNA Methylation Machinery in Cancers

High-resolution cancer genome sequencing efforts have discovered mutations in genes encoding epigenetic regulators that have roles as writers, readers, or erasers of DNA methylation and/or chromatin states (see also book ► Chap.1 of Wutz). In this section, we describe the mutations at the de novo DNA methyltransferase 3a (DNMT3A ) and the methylcytosine dioxygenase Ten-eleven translocation 2 (TET2) and their roles in cancer.

8.2.4.1 Mutations of de novo DNA Methyltransferase 3a

The de novo DNA methyltransferase 3a (DNMT3A) is one of the most frequently mutated genes across a range of haematological malignanciesFootnote 6, especially in acute myeloid leukaemia (AML) and T cell lymphoma. DNMT3A mutations in AML were found with frequencies of up to 22%, indicating that DNMT3A acts as critical tumor suppressor (Yang et al. 2015). Patients with DNMT3A mutations had significantly worse survival than patient with DNMT3A wild-type. Arginine 882 (R882) is located within the catalytic domain of the methyltransferase and represents a mutational hotspot that accounts for around 60% of DNMT3A mutations in AML. The R882 mutation results in a hypomorphic protein (i.e. partial loss of enzymatic activity) that inhibits wild-type DNMT3A by blocking its ability to form active tetramers, which represent the most active form of the DNA methyltransferase.

DNMT3A was found to be essential for the differentiation of hematopoietic stem cells (HSCs). Conditional deletion of Dnmt3a in mouse HSCs promoted self-renewal over differentiation and HSCs with a complete knock-out of the Dnmt3a gene (Dnmt3a-KO) dramatically outcompete their wild-type counterparts and accumulate in the bone marrow (◘ Fig. 8.3). Dnmt3a-KO HSCs display significant genome-wide hypomethylation. In particular, HSC-associated enhancers display dominant hypomethylation in Dnmt3a-KO HSCs. Accordingly, Dnmt3a-KO HSCs show upregulation of HSC multipotency genes and downregulation of differentiation factors, and their progeny exhibit global DNA hypomethylation and incomplete repression of HSC-specific genes.

Fig. 8.3
figure 3

DNMT3A controls hematopoietic stem cell fate. DNMT3A is a critical factor for the epigenetic silencing of hematopoietic stem cell (HSC) regulatory genes. Dnmt3a-KO HSCs show substantial CpG island hypermethylation, upregulate HSC multipotency genes, and downregulate differentiation factors. DNMT3A loss progressively impairs HSC differentiation and expands HSC numbers in the bone marrow

Full size image

It has been proposed that DNMT3A mutations serve as a pre-leukaemic lesion since somatic DNMT3A mutations arise in HSCs many years before malignancies develop. It has also been suggested that the enhanced self-renewal capacity of HSCs with DNMT3A mutations would be advantageous for second pro-oncogenic hits that drive tumor development (Brunetti et al. 2017). DNM3A mutations were found to negatively correlate with other known AML alterations, such as chromosomal translocations t(15;17), t(8;21) and inversion inv(16), rearrangements involving the histone lysine N-methyltransferase KMT2A (MLL), and mutations affecting TET2, the enzyme responsible for DNA hydroxymethylation (see book ► Chap. 1 of Wutz and ► Chap. 3 of Paro). The mutual exclusion of DNMT3A mutations and these genomic alterations suggests a role in similar epigenetic pathways. In contrast, mutations of DNMT3A in AML co-occur with mutations of nucleophosmin (NPM1) and the internal tandem duplication in the receptor tyrosine kinase FLT3 gene (FLT3ITD). Sixty percent of patients with DNMT3A mutations carry an NPM1 mutation, whereas only 13% of patients with wild-type DNMT3A harbor an NPM1 mutation (Yang et al. 2015). Similarly, FLT3ITD mutations are specifically enriched in patients with DNMT3A mutations. When patients with FLT3ITD mutations were classified according to DNMT3A status, those carrying a DNMT3A mutation had a significantly worse outcome, and a high relapse rate. Recent extensive genomic and epigenomic sequencing data suggested that the occurrence of all three mutations is non-random, and that NPM1mut/FLT3ITD/DNMT3Amut AML is a distinct entity (Cancer Genome Atlas Research et al. 2013). However, how these mutations act together to cause leukaemia remains elusive. HSCs with DNMT3A mutations persist even after chemotherapy and during relapse, indicating that mutant HSCs might represent a reservoir for the re-evolution of the disease in a relapse.

8.2.4.2 Mutations of Ten-Eleven Translocation 2 (TET2)

Somatic mutations at the Ten-eleven translocation 2 (TET2 ) gene have frequently been identified in a wide variety of hematologic malignancies, including AML, myelodysplastic syndrome (MDS), chronic myelomonocytic leukemia (CMML), and myeloproliferative neoplasms (MPN). In particular, somatic deletions and inactivating mutations in TET2 were identified in 10–20% of the MPN and MDS cases and in 7–23% of the AML cases. TET2 belongs to a family of three proteins (TET1, TET2, and TET3) that catalyze the successive oxidation of 5-methylcytosine (5meC) to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC), finally resulting in the replacement of 5meC by an unmodified cytosine (see book ► Chap. 1 of Wutz). TET2 mutations associated with the disease are largely loss-of-function mutations that impair the enzymatic activity, resulting in the failure of the 5mC-to-5hmC conversion, and eventually impairment of DNA demethylation. Studies in mice have shown that Tet2 loss leads to increased hematopoietic stem cell self-renewal and progressive myeloproliferation (the expansion of a multipotent hematopoietic progenitor stem cell), with features characteristic of human CMML. Thus, it appears that impaired 5meC-to-5hmC conversion confers clonal dominance to HSCs and exerts differentiation pressure toward the myeloid lineage. As in the case of DNMT3A, mutations in TET2 are considered pre-leukaemic as the mutation per se does not induce haematologic malignancies but rather does so in conjunction with other driver mutations. The pool of pre-leukaemic HSCs containing mutations at epigenetic regulators, such as TET2 or DNMT3A, has been considered a cellular reservoir that should be targeted therapeutically for more durable remissions.

Genome-wide methylation analyses revealed CpG hypermethylation of many active enhancer elements in human AML patients with TET2 mutations and similar results were obtained using a Tet2-dependent leukemia mouse model. The loss of 5hmC and gain of methylation at these enhancer elements lead to decreased H3K27ac and downregulation of neighboring genes, including tumor suppressor genes.

TET enzymes are Fe(II)/α-ketoglutarate (αKG)-dependent dioxygenases. Isocitrate dehydrogenases IDH1, IDH2, IDH3 are the enzymes that generate αKG. Genome-wide sequencing studies identified mutations in IDH1 and IDH2 in several cancers, including AML. IDH1/2 mutations gain the function to generate the novel oncometabolite D-2-hydroxyglutarate (D2HG) from αKG. D2HG inhibits αKG-dependent enzymes such as TET2. In AML, TET2 and IDH1/2 mutations are mutually exclusive, suggesting that the IDH1/2-TET2 pathway acts to suppress AML. The role and function of αKG and IDH in physiological and pathological conditions is described in book ► Chap. 9 of Santoro.

8.2.5 Epigenetic Inhibitors of DNA Methyltransferases in Cancer Therapy

Two inhibitors of DNA methyltransferases, 5-azacytidine (azacitidine) and 5-aza-2′-deoxycytidine (decitabine), have been approved for clinical use in haematological malignancies such as MDS, AML, and CMML (◘ Fig. 8.4). Azacitidine and decitabine are nucleotide analogues. Azacitidine can be incorporated into RNA and DNA whereas decitabine only into DNA. In particular, these nucleotide analogues can be incorporated during DNA replication instead of cytosine.

Fig. 8.4
figure 4

DNA methyltransferase inhibitors. Chemical structure of cytidine nucleoside and DNA methyltransferases inhibitors azanucleosides

Full size image

Maintenance of DNA methylation occurs after the passage of the replication fork where hemi-methylated DNA is generated that is then methylated at the opposite strand by DNMT1 (see book ► Chap. 1 of Wutz). The reaction of DNMTs comprises the attack at the C6 position of the pyrimidine ring of cytosine, leading to the intermediate formation of a covalent DNMT-DNA complex, followed by formation of 5mC (◘ Fig. 8.5). In the case of azacitidine or decitabine, however, the replacement of the carbon with an aza-group impairs the transfer of the methyl-group to the C-5 position (◘ Fig. 8.5). Consequently, the reaction arrests, the DNMT remains covalently trapped on the DNA, and the restoration and maintenance of methylation after the passage of the replication fork is impaired. Thus, treatment with DNMT inhibitors leads to a passive DNA demethylation process (see book ► Chap. 1 of Wutz).

Fig. 8.5
figure 5

Scheme showing the action of DNMT inhibitors. a The reaction of DNMTs comprises the attack at the C6 position of the pyrimidine ring of cytosine, leading to the intermediate formation of a covalent DNMT-DNA complex followed by formation of 5mC. Methylation of the C5 position induces a shift of electrons and releases the enzyme. b In the case of azacitidine or decitabine, the replacement of the carbon with an aza-group impairs the transfer of the methyl-group to the C-5 position. Consequently, the reaction arrests and the DNMT remains covalently trapped on the DNA. c Because of the semiconservative nature of DNA replication, a DNA sequence carrying symmetrical methylation marks on both strands gives rise to two hemi-methylated double strands, which can be restored to fully a methylated status by the maintenance DNA methyltransferase DNMT1. Azacitidine and decitabine can be incorporated into DNA during replication. Treatment with these compounds leads to loss of DNMT activity because the enzyme becomes irreversibly bound to the aza-residues in DNA and is no longer available for further catalysis. Loss of DNA methylation is the result of the dilution of DNA methylation during several rounds of DNA replication (passive DNA demethylation), resulting in transcriptional activation of genes previously silenced by DNA methylation

Full size image

Azacitidine is the first therapy to have demonstrated a survival benefit for patients with MDS (Fenaux et al. 2009). However, the molecular mechanisms governing the impressive responses seen in MDS are largely unknown. It is important to note that, in addition to the demethylating effect of these compounds, also the trapping of DNMT proteins on the DNA might indirectly contribute to an anti-cancer effect. Recently, it has also been shown that treatment with DNMT inhibitors induces immune signaling in cancer cells through the upregulation of a wide range of genes implicated in immune signaling and increased numbers of immune cells in the tumor microenvironment. Transcriptional silencing mediated by DNA methylation is also related to the recruitment of co-repressors and deacetylation or methylation of histone marks at promoters and enhancers. Accordingly, the combinatorial use of DNMT and histone deacetylases (HDAC) inhibitors (described in the next paragraph) was shown to enhance reactivation of aberrantly silenced genes in tumor cells and cause reductions in tumor burden.

8.3 Polycomb Group Proteins and Cancer

In ► Chap. 3 “cellular memory”, you have learnt that Polycomb group (PcG) proteins mainly function as members of two large multisubunit complexes, Polycomb Repressive Complex 1 and 2 (PRC1 and PRC2). PRC1 catalyzes the monoubiquitination of histone H2A at lysine 119 (H2AK119ub1) whereas PRC2 catalyzes the trimethylation of histone H3 at lysine 27 (H3K27me3) (see book ► Chap. 3 of Paro). Both of these post-translational modifications of histones are associated with transcriptional silencing. PcG proteins have been shown to regulate diverse biological processes during embryonic development, such as cell fate and lineage decisions, cellular memory, stem cell function, tissue homeostasis and regeneration. Because of the involvement of PcG proteins in so many key cellular processes, it should not be now surprising that alterations in the PcG machinery have frequently been found in several cancers. These changes include mutations and differential expression of the writers and erasers of H3K27me3 (8.3.1) and mutations in genes encoding histone H3 (8.3.2).

8.3.1 Alterations of PcG Activity in Cancer

Several PcG proteins are differentially expressed in tumors compared to the corresponding normal tissue (◘ Fig. 8.6). PRC2 components EZH2 and SUZ12 and PRC2 component BMI1 (see book ► Chap.3 of Paro) are often found overexpressed in several malignancies with an aggressive phenotype. In particular, more than half of the hormone-refractory prostate cancers, representing the late-stage cancer, exhibit amplification of the EZH2 gene, which encodes the catalytic subunit of PRC2 responsible for H3K27me3. This amplification results in increased EZH2 protein expression that consequently alters the H3K27me3 genomic landscape. High levels of BMI1 and EZH2 predict advanced disease and a poor prognosis in several human cancer types. In particular, EZH2 level is an important factor for assessing the progression and prognosis of prostate cancers.

Fig. 8.6
figure 6

Alterations of PcG activity in cancer. EZH2 is the catalytic subunit of PRC2 that also contains SUZ12 and EED. Loss-of-function mutations in EZH2 gene decrease H3K27me3 levels and activate the expression of genes implicated in cancer growth and invasion, conferring a poor prognosis. Amplification of EZH2 or mutations leading to its hyperactivation increase H3K27me3, thereby silencing tumor suppressor genes and promoting a cancer stem cell-like state. Similarly, inactivating mutations in the histone demethylase gene UTX cause an increase in H3K27me3

Full size image

Cancers also exhibit several PcG missense mutations and chromosomal translocations. For example, heterozygous mutations replacing a single tyrosine in the catalytic SET domain of the EZH2 protein (Tyr641) occur in 21.7% of patients with diffuse large B-cell lymphoma and 7.2% with follicular lymphoma. This mutation alters the catalytic activity of EZH2 by inhibiting the monomethylation of H3K27 and increasing the catalytic efficiency for subsequent di- and di- to trimethylation reactions (hyperactivated EZH2 mutants). Furthermore, PcG proteins can physically interact with a number of fusion transcription factors originating from translocations, such as the promyelocytic leukemia zinc finger-retinoic acid receptor α (PLZF-RARα) that, together with PRC2, induces the transcriptional silencing of target genes, thereby mediating leukemic transformation. Consistent with the critical role of H3K27me3 regulation in cancers, inactivating somatic mutations in the histone lysine demethylase gene UTX (KDM6A) have also been found in several cancers that consequently showed high H3K27me3 levels.

Although the role of PRC2 in cancer is mainly linked to H3K27me3 and transcriptional repression, there are cases of additional, non-classical functions. For example, in castration-resistant prostate cancers (CRPCs, i.e. cancers that do not respond to androgen deprivation therapy and hence representing the final and most aggressive stage of the disease), EZH2 acts as a coactivator for critical transcription factors. This functional switch is dependent on the phosphorylation of EZH2 at Ser21, which promotes the interaction with the Androgen Receptor (AR) and the activation of AR-target genes that are critical for disease progression (Xu et al. 2012).

Since PcG proteins are critical for maintaining stem cell-like characteristics of adult as well as embryonic stem cells (ESCs), it has been proposed that abnormally elevated levels of PcG proteins may lead to the generation and maintenance of cancer stem cells. Tumors are composed of heterogeneous populations of cells, which differ in their phenotypic and genetic features. It has been suggested that cellular heterogeneity within a tumor is organized in a hierarchical manner. In particular, evidence indicates the presence of a small subpopulation of cancer stem cells with stem cell-like characteristics that give rise to heterogeneous cancer cell lineages and undergo self-renewal to maintain their reservoirFootnote 7. Cancer stem cells were shown to have an enhanced capacity for therapeutic resistance, immune evasion, invasion, and metastasis (Prager et al. 2019). In pluripotent ESCs, PRC1 and PRC2 bind to promoters of genes encoding regulators of developmental processes that must remain silenced to maintain ESC self-renewal. During ESC differentiation, PcG factors are removed from these genes that consequently become de-repressed, thereby setting new transcriptional programs to ensure proper differentiation. Furthermore, PcG proteins are required for cell fate determination since they restrict alternative fates once cells differentiate towards a specific cell lineage. Since PRC2 is important for balancing proliferation versus differentiation, PRC2 has been shown to promote a de-differentiated phenotype of several cancers (see also book ► Chap. 3 of Paro). Increased PRC2 activity in stem or progenitor cells might promote self-renewal over differentiation by repressing differentiation genes or genes controlling cell proliferation, such as CDKN2A (Laugesen et al. 2016). For example, elevated EZH2 levels led to growth of Ewing tumors and a consequential inhibition of endothelial and neuroectodermal differentiation. The important role of PcG proteins in cancer makes them attractive targets for cancer therapy.

8.3.2 Mutations of Affecting Lysine 27 of Histone H3 Occur in Multiple Cancers

Consistent with the critical role of altered PRC2 activity, mutations in the substrate, lysine 27 of histone H3, have also been reported in cancers. Whole-genome sequencing of paediatric high-grade gliomas have identified gain-of-function mutations in histone H3 genes, specifically histone 3A (H3F3A) and histone H3B (HIST1H3B), encoding histone H3 variants H3.3 and H3.1, respectively (Schwartzentruber et al. 2012; Wu et al. 2012). In 78% of paediatric diffuse intrinsic pontine gliomas (DIPGs) and in 22% of non-brainstem paediatric glioblastomas (non-BS-PGs), the H3F3A or HIST1H3B genes contained a mutation leading to the substitution of lysine 27 to methionine (H3K27M). Somatic mutations at H3F3A that replace glycine 34 with arginine (H3G34R) were also found in 14% of non-BS-PGs.

H3K27M mutant tumors are frequently found in the thalamus, pons, and spinal cord and show poor prognosis whereas H3G34R mutations are seen in tumors of the cerebral hemispheres, suggesting a different cellular origin for these tumors. At molecular level, H3K27M inhibits the enzymatic activity of PRC2 through interaction with the EZH2 subunit and causes a profound reduction of wild-type H3K27me3 levels (Lewis et al. 2013) (◘ Fig. 8.7).

Fig. 8.7
figure 7

H3K27M mutations in genes encoding the histone variants H3.3 and H3.1. H3K27M mutations are recurrently found in pediatric high-grade glioma (HGG) and diffuse intrinsic pontine glioma (DIPG). K27M mutations in genes coding for histone variants H3.3 and H3.1 result in a global reduction of H3K27me3 levels, leading to derepression of PRC2 target genes

Full size image

The alteration in H3K27me3 levels has been thought to drive a transcriptional program that promotes tumor initiation and progression. The strong reduction of H3K27me3 in H3K27M tumors is quite remarkable since the human genome contains 16 distinct histone H3 encoding genes and only one of the two H3F3A genes contains the H3K27M mutation. Thus, the K27M mutation of only one histone H3 gene causes a drastic reduction of H3K27me3 at all the other wild-type histone H3s that, in diploid cells, are encoded by 31 genes. Importantly, concomitant with H3K27me3 reduction, the levels of the active H3K27ac mark increase, suggesting a switch in the epigenetic landscape of H3K27M tumors and the activation of genes linked to proliferation.

There have been several hypotheses to explain how H3K27M impairs PRC2 activity. It has been suggested that H3K27M might sequester PRC2 to chromatin and inhibit its activity. Crystal structure and in vitro biochemical assays showed that PRC2 has high binding affinity for mutant H3K27M histones, leading to the stalling of PRC2 and preventing propagation of the H3K27me3 mark. Furthermore, in vitro biochemical analyses revealed that PRC2 activity is inhibited by di-nucleosomes composed of H3K27M-H3K27me3. On the other hand, it has also been proposed that H3K27M might exclude PRC2 from chromatin. Indeed, H3K27M associates with chromatin regions enriched in H3K27ac whereas PRC2 is excluded from these regions. These results support a model in which H3K27M excludes PRC2 binding, which thus induces aberrant accumulation of H3K27ac at nucleosomes containing the wild-type H3 and the mutated H3K27M.

The majority of the heterotypic H3K27M-H3K27ac nucleosomes were found at actively transcribed genes and colocalized with BRD2 and BRD4, which are part of the Bromodomain (BRD) and Extra-Terminal motif (BET) protein family. BRD-containing proteins regulate gene expression primarily through recognition of histone acetyl residues, leading to the recruitment of protein complexes that modulate gene expression (see also 8.4.2). The development of inhibitors of epigenetic readers as anticancer agents is now an intense area of research. In particular, the selective small-molecule inhibitor JQ1 binds competitively to bromodomains and impairs their binding to acetylated histones (Qi 2014). The association of BRD2 and BRD4 proteins with chromatin regions enriched in H3K27M-H3K27ac heterotypic nucleosomes suggested a potential role of BRD proteins in DIPG pathogenesis. Accordingly, treatment with JQ1 strongly inhibits the proliferation of H3K27M-DIPG cells in vivo and DIPG xenograft miceFootnote 8 show substantially improved survival, identifying BET proteins as potential therapeutic targets of H3K27M-DIPG. It has also been suggested that increasing H3K27me3 in H3K27M-DIPG might be an effective therapeutic strategy. This can be achieved either by enhancing PRC2 methyltransferase activity or inhibiting H3K27 demethylase activity. H3K27me3 can be demethylated by the KDM6 subfamily K27 demethylases JMJD3 and/or UTX. Accordingly, treatment with the JMJD3 K27 histone demethylase inhibitor GSKJ4 increased cellular H3K27 methylation in patient-derived H3K27M-DIPG tumor cells and showed potent antitumor activity both in vitro against H3K27M cells and in vivo against H3K27M-DIPG xenografts (◘ Fig. 8.8).

Fig. 8.8
figure 8

Methylation and demethylation of H3K27. Left panel. EZH2, the catalytic component of PRC2, methylates H3K27 and promotes a transcriptionally repressed chromatin state. In contrast, JMDJD3 or UTX are demethylases that remove methyl groups from H327K, inducing an open and transcriptionally active chromatin state. Right panel. H3K27M inactivates EZH2 activity with a consequent decrease of H3K27me3 levels and transcriptional activation. Pharmacological inhibition of JMJD3 demethylase by GSKJ4 increases methylation at wild-type H3K27 and inhibits gene expression. Treatment with GSKJ4 showed potent antitumor activity in H3K27M tumor cells

Full size image

8.3.3 EZH2 Inhibitors in Cancer Therapy

Since PcG proteins have been proven to be bona fide and cancer stem cell markers, PcG proteins became attractive targets for both cancer prevention and therapy. Inhibitors targeting EZH2 activity have been developed. Three of these inhibitors (GSK126, Tazemetostat, and CPI-1205) are now undergoing phase I or II clinical trials. GSK126 is an S-adenosylmethionine (SAM) competitor and a highly selective EZH2 inhibitor. However, these inhibitors are not selective for hyperactive EZH2 mutants that contain mutations in the substrate-binding pocket (Laugesen et al. 2016). Thus, targeting EZH2 might be a relevant therapeutic strategy in cancers not expressing hyperactivated EZH2 mutant variants.

8.4 Histone Acetylation and Deacetylation in Cancers

Histone acetyltransferases (HATs) are a class of enzymes that transfer acetyl groups from acetyl-CoA cofactors to lysine residues at histones (see book ► Chap. 1 by Wutz and ► Chap. 9 by Santoro). Histone acetylation is a highly reversible process since histone deacetylases (HDACs) act as erasers by removing the acetyl group. Histone acetylation is associated with active transcription, especially at promoters and enhancers, and facilitates the recruitment of co-regulators and RNA polymerase II complexes. In contrast, HDACs usually establish gene repression. As discussed in the previous section, the acetyl group can be recognized by proteins containing bromodomains (BRD) that, in turn, often recruit factors linked to transcriptional activation.

The impairment of the balance between acetylation and deacetylation can affect gene expression. This situation is often found in cancers with altered acetylation patterns. Alterations in the acetylation/deacetylation balance can originate because of the abnormal recruitment of HDACs or the reduced activity of HATs, leading to repression of tumor suppressor genes. On the other hand, it can also occur an increased HAT activity with consequent activation of oncogenes (◘ Fig. 8.9). These mechanisms can alter the normal cell cycle, block or revert differentiation, impair apoptosis, and facilitate proliferation.

Fig. 8.9
figure 9

Alterations of HATs and HDACs in cancers. Alterations in the acetylation/deacetylation balance can influence the expression of tumor suppressor genes and oncogenes, favoring the tumorigenic process. High expression or the abnormal recruitment of HDACs or the reduced expression or activity of HATs can lead to the repression of tumor suppressor genes. In contrast, mutation or low expression of HDACs or high expression, oncogenic fusion, or aberrant recruitment of HATs can lead to the activation of oncogenes

Full size image

8.4.1 Alterations of Histone Acetyltransferases in Cancer

HATs are divided into three families, depending on their structural homology. Gcn5-related N-acetyltransferases (GNAT) include GCN5 and PCAF. MYST acetyltransferases include MOZ, MOF, TIP60, and HBO1. The last class consists of p300/cAMP-responsive element-binding proteins that include CBP and p300.

Appropriate acetylation within cells is important since upregulation or downregulation of HATs is associated with tumorigenesis and poor prognosis (Di Cerbo and Schneider 2013). For example, aberrant lysine acetylation mediated by CBP/p300 has been implicated in the genesis of multiple haematologic cancers. A high frequency of alterations in HAT genes has been reported in small-cell lung cancers, an aggressive lung tumor subtype with poor prognosis, and non-Hodgkin B-cell lymphomas. These mutations are most often point mutations in proximity to the HAT catalytic domain, resulting in loss of enzymatic activity. In acute lymphoid leukemia (ALL), a significant percentage of patients (18.3%) have mutations in the HAT domain of CBP. These mutations were shown to impair histone acetylation and transcriptional regulation of CBP targets. Since several of these mutations acquired at relapse were detected in subclones at diagnosis, it was suggested that they may confer resistance to therapy.

Genes encoding HATs can also act as oncogenes. Although less frequent, it has been reported that chromosomal translocations involving HAT genes can generate chimeric proteins that retain HAT catalytic activity and bromodomains. Beside these cases, HATs can also be involved in the regulation of oncogenic fusions. For example, AML1-ETO, the most frequent fusion protein in AMLs, is acetylated by p300. This acetylation is essential to promote self-renewal and induce leukemogenesis. These examples indicate that because of the various cellular functions linked to HATs and their specific substrates (histones but also non-histone proteins), they can act as either tumor suppressors or oncogenes, depending on the cellular or molecular context and cancer type.

8.4.2 Acetyl-Lysine Recognition Proteins and Cancer

The bromodomain (BRD) is a conserved protein module (reader) that recognizes acetyl-lysine. Depending on the structure, each BRD has preference for different acetylated histones. BRD-containing proteins are often implicated in the regulation of transcription by the recruitment of different molecular partners. HATs, such as PCAF, GCN5, p300, and CBP also contain a BRD.

As discussed earlier, an important class of BRD-containing proteins is the BET protein family, including BRD2, BRD3, BRD4, and BRDt. BET proteins are critical mediators of transcriptional activity through the recognition of acetylated histones and the recruitment of cofactors for gene activation (◘ Fig. 8.10).

Fig. 8.10
figure 10

Targeting readers of histone acetylation to modify gene expression. The bromodomain (BRD) is a conserved protein module that recognizes acetyl-lysine (reader). BRD-containing proteins are often implicated in the activation of transcription by the recruitment of different molecular partners. BRD inhibitors prevent the interaction between the BRD and the acetyl group, causing the downregulation of genes including some that may play critical roles in cancer

Full size image

BRD4 and BRD2 play an important role in transcription elongation of genes controlling cell proliferation by recruiting the positive transcription elongation factor complex (P-TEFb) to acetylated chromatin through their BRDs. BET proteins were shown to play important roles in tumorigenesis, for instance BRD4, which is required for the maintenance of AML with sustained expression of Myc. In breast cancer, BRD3/4 interacts with the histone H3K36 methyltransferase WHSC1 and promotes the expression of estrogen receptor alpha (ERα), thereby contributing to tamoxifenFootnote 9 resistance in ER-positive breast cancers. The fact that BRDs are a potentially druggable target has encouraged the discovery and development of several small-molecule inhibitors in recent years (◘ Fig. 8.10).

JQ1 and I-BET762 are two representative inhibitors of BET family proteins that competitively bind to BRD4 (Filippakopoulos et al. 2010). The efficacy of JQ1 was initially tested in NUT midline carcinoma (NMC) cells that express BRD4 fused to the nuclear protein in testis (NUT). The BRD4-NUT oncoprotein contributes to carcinogenesis by interfering with epithelial differentiation. Treatment of NMC cell lines and xenograft models with JQ1 was shown to displace BRD4-NUT from chromatin, inducing differentiation and specific anti-proliferative effects. To date, the efficacy of JQ1 was demonstrated in many other malignancies, including hematological malignancies and a variety of solid tumors, such as glioblastoma, medulloblastoma, hepatocellular carcinoma, colon cancer, pancreatic cancer, prostate cancer, lung cancer, and breast cancer (Perez-Salvia and Esteller 2017). Currently, there is a lot of effort in generating inhibitors targeting BRDs of non-BET proteins, including the BRD of CBP and p300. To date, many inhibitors targeting BRD proteins have been investigated in clinical trials.

8.4.3 Alterations of Histone Deacetylases in Cancer

In humans, the genome encodes 18 histone deacetylases (HDACs). HDACs can be divided into four classes, based on their primary homology to yeast HDACs. Class I HDACs include HDAC1, HDAC2, HDAC3, and HDAC8. Class II HDACs are represented by HDAC4, HDAC5, HDAC6, HDAC7, HDAC9, and HDAC10. Class III HDACs belong to the Sirtuin 2 (Sir2) family, which is composed of SIRT1, SIRT2, SIRT3, SIRT4, SIRT5, SIRT6, and SIRT7. Class IV contains only HDAC11. Classes I, II, and IV are Zn2+-dependent HDACs whereas Sir2-like proteins (sirtuins) are nicotinamide adenine dinucleotide (NAD+)-dependent HDACs (see also book ► Chap. 9 by Santoro). HDACs can also deacetylate non-histone proteins. Some HDACs are subunits of complexes, such as Sin3 and NuRD. The association with these complexes was shown to increase they catalytic activity.

Alterations of HDAC activity in cancers is generally associated with aberrant deacetylation and the inactivation of tumor suppressor genes. Altered expression and mutations in genes encoding HDACs have been linked to tumor development due to inactivation of tumor suppressor genes (Ropero and Esteller 2007). Overexpression of HDAC1 has been found in gastric, breast, pancreatic, hepatocellular, lung, and prostate carcinomas and, in most of the cases, HDAC1 up-regulation is associated with poor prognosis. HDAC1, HDAC2, and HDAC3 were shown to be highly expressed in renal cell, colorectal, and gastric cancers as well as in classical Hodgkin’s lymphoma. Mutations affecting HDACs were also reported in several cancers. Loss of HDAC2 protein expression in sporadic carcinoma was associated microsatellite instability. The best-known mechanism showing the contribution of HDACs in cancer is their interaction with oncogenic fusion proteins. In acute promyelocytic leukemia (APL), fusion proteins containing RAR-PML and RAR-PLZF bind to retinoic acid-responsive elements (RAREs) and recruit HDAC-containing complexes. This action prevents the binding of the retinoic acid receptor and represses the expression of genes implicated in the differentiation of myeloid cells. Furthermore, the fusion of ETO or TEL to AML1 converts a transcriptional activator to a constitutive transcriptional repressor through recruitment of a HDAC complex, thereby repressing the expression of genes critical to differentiation and promoting acute myeloid leukemia.

8.4.4 HAT and HDAC Inhibitors in Cancer Therapy

Aberrant HAT activity in cancers can be targeted by using HAT inhibitors. An example is provided by bi-substrate inhibitors for PCAF, p300, and TIP60 that mimic two substrates of HATs: the cofactor acetyl coenzyme A (Ac-CoA) and a peptide resembling the lysine substrate. However, these molecules are not membrane-permeable. Several synthetic compounds have been designed to selectively target HATs, such as A-485, a potent, selective, and drug-like catalytic inhibitor of p300 and CBP (Lasko et al. 2017). As described in 8.4.3, targeting of the histone acetylation readout through inhibition of the readers of histone acetylation (i.e. inhibitors targeting BRD proteins) is an attractive strategy for the pharmacological treatment of malignancies. Accordingly, BRD inhibitors have been investigated in clinical trials, whereas there currently are no clinical trials with HAT inhibitors.

HDAC family members are attractive targets for drug design and a variety of HDAC-based combination strategies have been developed for the treatment of cancers (Ropero and Esteller 2007). The inhibition of HDACs has shown promising antitumor effects. Several classes of natural and synthetic HDAC inhibitors have been identified, such as butyrates, hydroxamic acid, benzamides, and cyclic tetrapeptides. Trichostatin A (TSA) is a natural compound that inhibits the activity of HDACs and induces cancer cell differentiation and apoptosis. SAHA (vorinostat) is a pan-HDAC competitive inhibitor that structurally belongs to the group of hydroxamic acids. SAHA was the first FDA-approved HDAC inhibitor and is clinically effective in the treatment of refractory, primary cutaneous T-cell lymphomas. Specific HDAC inhibitors have also successfully been generated. SHI-1:2 is a benzamide inhibitor that shows HDAC1/HDAC2-specific inhibitory activity. Several HDAC inhibitors are currently in clinical trials and some of them have already been approved for disease treatment. As described above, mutations affecting HDACs were reported in several cancers. Mutations causing loss of HDAC2 protein expression and/or enzymatic activity rendered cells more resistant to the usual antiproliferative and proapoptotic effects of histone deacetylase inhibitors. Thus, the mutational status of HDAC encoding genes in individual cancers should be taken into consideration for pharmacogenetic treatments.

8.5 Chromatin Remodeling Factors and Cancer

Chromatin remodeling factors are multi-subunit complexes that use the energy of ATP hydrolysis to reposition, eject, slide, or alter the composition of nucleosomes. These processes play key roles in transcription by enabling access of DNA-binding proteins and the transcriptional machinery to DNA in order to facilitate expression. In eukaryotes, four families of chromatin remodeling complexes have been characterized: the switching defective/sucrose non-fermenting (SWI/SNF) family, the imitation-switch (ISWI) family, the nucleosome remodeling and histone deacetylase complex (NuRD), and the inositol 80 (INO80) family. These complexes contain distinct types of catalytic ATPases and associate with several factors that specify targets and contribute to diverse biological processes, including the regulation of gene expression. An extensive description of the activities of these chromatin remodeling factors can be found in book ► Chap. 3 of Paro. Here, the alterations of these complexes in cancer and their functional consequences are described.

8.5.1 SWI/SNF Complexes and Cancer

In mammalian cells, the SWI/SNF complexes are generally grouped into two types of complexes: BAF complexes containing the ATPase BRG1 or BRM and several alternate core subunits, such as ARID1A or ARID1B, and PBAF (polybromo-associated BAF) complexes containing the BRG1 ATPase associated with factors like PBRM1 and ARID2. Cancer genome sequencing studies have revealed that >20% of all cancers harbor mutations in SWI/SNF-encoding genes (Valencia and Kadoch 2019). Evidence for a driving role of BAF complex alterations in cancer first came from the identification of mutations at the SMARCB1 locus encoding a subunit shared by both SWI/SNF BAF and PBAF complexes. The SMARCB1 gene undergoes biallelic inactivation in ∼98% of malignant rhabdoid tumors (MRTs), a highly aggressive paediatric cancer. Loss of ARID1A represents a commonly mutated gene in ovarian, endometrial, gastric, pancreas, breast, brain, prostate, lung, and liver cancers. The large majority of ARID1A mutations are frameshift indels or nonsense point mutations resulting in loss of the protein.

SWI/SNF complexes were suggested to oppose epigenetic silencing by PcG proteins. For example, in Drosophila, PcG proteins were shown to maintain repression of Hox genes during embryogenesis, while the SWI/SNF complex promotes Hox gene activation. Studies in MRT showed that loss of SMARCB1 destabilizes SWI/SNF complexes on chromatin, so they are unable to oppose PcG-mediated repression at bivalent promoters required for differentiation (Nakayama et al. 2017). Intriguingly, while core components of the SWI/SNF complex are frequently inactivated, PcG proteins are frequently overexpressed in cancers. For example, EZH2 is often overexpressed in ovarian clear cell carcinomas (OCCCs) whereas ARID1A is mutated in ∼57% of OCCCs. The link between PcG and SWI/SNF complexes is further supported by studies showing that the inhibition of EZH2 acts in a synthetic lethal manner in ARID1A-mutated ovarian cancer cells and the ARID1A mutational status correlates with the response to EZH2 inhibitors (Bitler et al. 2015).

Gain-of-function perturbations of SWI/SNF subunits have also recently been discovered. Human synovial sarcoma show a recurrent chromosomal translocation, t(X;18)(p11.2;q11.2), that fuses the SS18 gene on chromosome 18 to one of three closely related genes on the X chromosome, SSX1, SSX2, and, rarely, SSX4, resulting in in-frame fusion proteins. SS18 is a subunit of BAF complexes. When SS18 is fused to SSX, a protein is produced that alters the SWI/SNF complex by evicting the SWI/SNF subunit SMARCB1. This altered SWI/SNF complex is redirected to other genomic loci, relieving H3K27me3 repression at genes critical to cancer. One of these genes encodes the transcription factor Sox2 that is essential for proliferation in synovial sarcoma. Because of the important role of SWI/SNF complexes in cancer, several compounds targeting SWI/SNF subunits have been developed. For example, the bromodomain of the SWI/SNF subunit BRD7 and BRD9 has been targeted by specific chemicals and ligand base degrader compounds, which were shown to attenuate the proliferation of tumor cells with SWI/SNF mutations (Remillard et al. 2017).

8.5.2 ISWI Complexes and Cancer

Mammals have two homologs of ISWI, SNF2L (encoded by the SMARCA1 gene) and SNF2H (encoded by the SMARCA5 gene). In mammals, five ISWI complexes have been identified: ACF, CHRAC, WICH, RSF, and NoRC. ISWI complexes are typically heterodimers and composed of an ISWI protein (SNF2H or SNF2L) and another subunit that provides additional specificity through targeting and recruitment of additional chromatin regulators.

RSF1 is an ISWI complex composed of SNF2H and the remodeling and spacing factor 1 (RSF1). This complex can reposition the nucleosome for transcriptional regulation. RSF1 is overexpressed in multiple types of tumors, including breast cancer, and correlates with their aggressiveness in terms of tumor size and stage. Overexpression of RSF1 was associated with the amplification of the 11q13.5 chromosomal region. Elevated expression of RSF1 was shown to promote the expression of genes regulated by NF-kB, including some involved in the evasion of apoptosis and inflammation, which is necessary for the development of chemoresistance in ovarian cancer cells (Yang et al. 2014).

BAZ2A (also known as TIP5) interacts with SNF2H to form the nucleolar remodeling complex NoRC (Bersaglieri and Santoro 2019). BAZ2A was initially identified as the repressor of ribosomal rRNA genes in healthy and differentiated cells. In prostate cancer, BAZ2A is highly expressed. Alterations in BAZ2A levels are not caused by somatic structural or sequence variations but more likely by post-transcriptional misregulation involving loss of microRNA miR-133a. In prostate cancer, BAZ2A silences the expression of numerous genes that are frequently repressed in metastatic prostate cancers. Data have also shown that BAZ2A is required for the initiation of prostate cancer driven by the loss of PTEN, the most commonly lost tumor suppressor gene in primary prostate cancer (Pietrzak et al. 2020). BAZ2A overexpression was shown to be tightly associated with a prostate cancer subtype displaying CIMP and prostate cancer recurrence. Thus, BAZ2A might serve as a useful marker for metastatic potential in prostate cancer.

8.5.3 The NuRD Complex and Cancer

NuRD (Mi-2) is macromolecular protein complex that combines chromatin remodeling with protein deacetylase activity (Hoffmann and Spengler 2019). The remodeling subcomplex consists of an ATPase (chromodomain helicase DNA-binding protein 3/4/5; CHD3/4/5) whereas the histone deacetylase activity is attributed to HDAC1/2. NuRD also encompasses several non-enzymatic components including methyl-CpG-binding domain 2/3 (MBD2/3), retinoblastoma-binding proteins 4/7 (RBBP4/7), metastasis-associated proteins 1/2/3 (MTA1/2/3) and GATA zinc finger domain containing proteins 2A/B (GATAD2A/B) (Allen et al. 2013).

Multiple subunits belonging to NuRD complex protein families are overexpressed in a variety of human cancers. These include MTA1/2/3, HDAC1, and HDAC2. The elevated expression of these factors was associated with gene repression and implicated in processes such as DNA damage repair and the maintenance of genomic integrity. MTA1 is overexpressed in various cancers, correlating with cancer progression and poor outcome. MTA3 has been shown to directly interact with BCL-6, a master regulator of B cell differentiation that plays a crucial role in diffuse large B cell lymphoma. The MBD3 subunit was shown to directly interact with JUN, an oncoprotein important in many malignancies. Recently, CHD4, one of the catalytic subunits of NuRD, was shown to be essential in fusion-positive rhabdomyosarcoma (FP-RMS), a rare paediatric sarcoma with a low mutational burden that exhibits features of skeletal myogenesis. The most common chromosomal translocation observed in FP-RMS is PAX3-FOXO1, both encoding transcription factors. This fusion generates a novel transcription factor with altered transcriptional power and target genes. Surprisingly, although CHD4 is classically defined as a repressor, in FP-RMS it acts as co-activator by interacting with super-enhancers where it generates a chromatin architecture permissive for binding of PAX3-FOXO1 and activating its downstream oncogenic program (Marques et al. 2020). These examples further indicate that because of the various cellular functions mediated by chromatin remodeling activities, their readout depend on the cellular or molecular context and cancer type.

8.5.4 The INO80 Complex and Cancer

The INO80 family of remodeling enzymes is currently composed of two classes of enzymes, Ino80 and Swr1 (Watanabe and Peterson 2010). In mammals, the chromatin-remodeling complexes of the INO80 subfamily are INO80, Snf2-related CBP activator protein (SRCAP), and p400. The mammalian INO80 complex is composed of at least 13 subunits. INO80 has been implicated in many crucial cellular functions, including transcriptional regulation, DNA replication and repair, telomere maintenance, and chromosome segregation. Subunits of the INO80 complex are frequently amplified in cancers, including lung squamous carcinoma, 50% of pancreatic cancers, and 45% of bladder cancers. Increased expression of INO80 has been functionally associated with tumor progression. INO80 is overexpressed in BRAF- and NRAS-mutated melanoma cancer cells. Mechanistically, INO80 interacts with super-enhancers through transcription factors such as MITF and Sox9. INO80 binding reduces nucleosome occupancy and facilitates Mediator recruitment (see also book ► Chap. 3 of Paro), thus promoting oncogenic transcription. Expression of INO80 was also found upregulated in anaplastic thyroid carcinoma, an aggressive and lethal cancer with extrathyroidal invasion, distant metastasis, and resistance to conventional therapies. Interestingly, downregulation of INO80 was shown to decrease the expression of stem cell marker genes as well as to attenuate stem cell-specific properties including the ability to form tumors. These results suggest that the role of INO80 in cancer cells is linked to its stem cell-promoting function. Accordingly, INO80 has been found to selectively activate pluripotency genes in ESCs (Wang et al. 2014), supporting the notion that genes and pathways important for ESC maintenance are often reactivated in cancer.

Take-Home Message

  • Alterations of epigenetic processes affect gene expression and promote cancer initiation and progression

  • Epigenetic regulators are frequently mutated in cancer

  • The reversible nature of epigenetic aberrations has led to the emergence of the promising field of epigenetic therapy

  • Alterations of DNA methylation are frequently observed in cancers. Tumor suppressor genes can be silenced through promoter hypermethylation. Global hypomethylation has been associated with genomic instability. Alterations of DNA methylation at imprinting control regions, with a consequent loss of imprinting, have also been linked to cancer.

  • Component of the DNA methylation/demethylation machinery, DNMT3A and TET2, are frequently mutated in haematological malignancies. Alterations in the DNA methylation profile have been linked to enhanced self-renewal capacity of HSCs.

  • Azacitidine and decitabine are nucleotide analogues that are used as inhibitors of DNA methyltransferases. These nucleotide analogues can be incorporated during DNA replication instead of cytosine and covalently bind DNMT1, impairing the inheritance of DNA methylation during DNA replication (passive DNA demethylation).

  • Component of the PcG machinery are frequently mutated in cancers. These changes include mutations (loss of function or hyperactivation) and differential expression of the writers and erasers of H3K27me3 and mutations in genes encoding histone H3 (H3K27M). Increased PRC2 activity has been shown to promote a de-differentiated phenotype of several cancers by promote self-renewal over differentiation through the repression of differentiation genes or genes controlling cell proliferation.

  • The mutation one histone H3 gene (H3K27M), identified in paediatric high-grade gliomas, exerts a dominant negative effect on H3K27me3.

  • Expression of HATs and HDACs are frequently altered in cancers, leading to an acetylation/deacetylation imbalance that can influence the expression of tumor suppressor genes and oncogenes, favoring the tumorigenic process.

  • BRD is a conserved protein module (reader) that recognizes acetyl-lysine. BRD-containing proteins are often implicated in the regulation of transcription by the recruitment of different molecular partners. JQ1 is a small molecule inhibitors that competitively binds to the BRD of BRD4 transcription factor and displaces it by its target genes, causing differentiation and specific anti-proliferative effects. Their efficacy was demonstrated in many malignancies.

  • Chromatin remodeling factors are also frequently mutated in cancers. In particular, ARID1A, the gene encoding a subunit of SWI/SNF BAF complex, is mutated in 57% in ovarian clear cell carcinomas. These mutations typically cause the loss of ARID1A protein expression and correlate with the response to EZH2 inhibitors (synthetic lethality).