Histone H3 G34 Tail Mutations in Cancer: Actions in Cis and Trans to Alter Chromatin and Gene Expression

The first histone mutations identified in childhood glioblastoma affected transcription-associated histone H3.3. Histone H3.3 is deposited by specific chaperone systems at sites upstream and downstream from the start site of transcription. H3.3 mutations upset the cellular balance between gene activation–associated histone H3 lysine 36 methylation (H3K36me2/3) and repressive H3K27me3, mutually exclusive chromatin modifications. H3K36me2, which can be focused near the start site of transcription, is commonly created by the NSD family of histone methyltransferases (HMT), whereas H3K36me3 is produced solely by SETD2, accumulating in gene bodies as transcription proceeds. H3K36me2/3 prevents the action of EZH2, the enzymatic component of the polycomb repressive complex 2 (PRC2), which creates H3K27me3. H3.3K27M mutations are most commonly found in childhood diffuse pontine glioma, whereas mutations of H3.3K36M are found mostly in chondroblastoma. These mutations act both in cis on the histone tail in which they occur and in trans, across the genome. An H3K27M-mutant histone molecule cannot be methylated on residue 27 and is subject to increased action by the NSD1/2/3 and SETD2 HMTs. Moreover, H3.3K27M sequesters EZH2, the enzymatic component of the PRC2, and prevents its ability to methylate H3.3, H3.1, or H3.2. The result is a global decrease in H3K27me3 and increase in H3K36me2/3. Similarly, H3K36M abrogates residue 36 methylation in cis, stimulates the activity of H3K27me3 on the mutant molecule, and sequesters H3K36 HMTs, leading to a global decrease in H3K36 and increase in H3K27 methylation. In addition, these mutations lead to altered distribution of chromatin modifications, further driving changes in expression. For example, deposition of H3.3K36M at specific promoters increases H3K27me3, leading to gene repression, but at the same time attracts PRC2 away from other genomic sites, leading to aberrant gene activation at sites abandoned by PRC2.

The histone H3G34 mutations also upend H3K27/H3K36 methylation patterns. H3.3G34W was identified in >90% of giant cell tumor of bone (GCT), accounting for 5% of bone tumors (1). GCT is a locally aggressive neoplasm characterized by proliferation of a primitive mesenchymal lineage and attraction of myeloid and multinucleated giant cells resembling osteoclasts, which damage bone. H3.3G34R/V/D mutations are associated with high-grade gliomas. H3.1G34 mutations are occasionally found in a variety of carcinomas (2). Mutations changing glycine 34 to more bulky amino acids occlude the entry of the histone tail into a narrow channel in SETD2 (3). Accordingly, exogenous H3G34-mutant histone cannot be di- or trimethylated at H3K36 in vitro, but still appears to accumulate some H3K36 mono- and dimethyl marks in vivo (4). Furthermore, unlike H3K36M, expression of this histone mutant did not block H3K36 methylation of endogenous histone (5). This might suggest that the G34 mutation only affects chromatin locally through decreased H3K36me3. However, G34R brain tumor cell lines showed increased expression of a set of genes associated with neural development, increased H3K36me3, and RNA polymerase II recruitment (6), indicating that H3G34 mutation could cause redistribution of chromatin marks across the genome. Another group showed that the H3.3G34R led to increases in H3K36me3, particularly at regions enriched for the KDM4A/B/C H3K36 demethylases (7), suggesting that the mutation might enhance H3K36 methylation by sequestration of the demethylases. H3.3G34R gains the ability to bind the PHD domain of the RACK7 transcriptional cofactor (8), which can deactivate genes through the recruitment of the KDM5C H3K4 histone demethylase. Hence, accumulating evidence suggests pleiotropic activity of H3.3G34 mutations.

A prior study of 86 primary GCTs, and cell lines derived from patients with and without H3.3G34W mutations, showed that mutant cells had increased proliferation, invasion, and migration (9). G34W specimens displayed decreased expression of genes associated with the E2F transcription factor, including osteoprotegerin, an inhibitor of osteoclast function, potentially explaining the bone resorption characteristic of GCT. Khazaei and colleagues (10) reverted or disrupted the G34W mutation in tumor cell lines derived from primary GCT, demonstrating that this decreased growth of cells in vitro and in vivo, while reexpression of mutant H3.3 stimulated neoplastic growth and the recruitment of multinucleated giant cells to tumors. RNA-sequencing analysis of isogenic cells showed that G34W increased expression of genes involved in cell adhesion and extracellular matrix and decreased expression of genes involved in muscle contraction. Single-cell RNA sequencing of primary tumors identified subpopulations of cells within primary tumors and found that stromal cells harboring the G34W mutation mostly resembled myofibroblasts blocked at an intermediate state of differentiation. Mass spectrometry analysis to determine the complement of secreted proteins from malignant cells showed that the aberrant stromal cells secreted a number of extracellular matrix components that could attract monocytes and osteoclasts, helping explain the bone destruction of GCT.

Although this work reinforces that H3.3G34W is the basis of neoplastic growth and attraction of inflammatory cells to the tumor, how and why this pattern of gene expression comes about remains to be fully explained. To understand the epigenetic basis of altered gene expression in GCT, the authors analyzed histone changes associated with H3.3G34W. Mass spectrometry showed, in accordance with prior findings, that on G34W-mutant H3.3, H3K36me3 was decreased and H3K37me3 was increased. H3G34W peptides displayed increased levels of H3K36me2, which suggests that the NSD enzymes can still act on the mutant histone in vivo. While in GCT cell lines H3.3G34W made up only 3% of histone H3, H3K27me3 levels on nonmutant histones H3.3 and H3.1 decreased by 30% to 40%, suggesting an important effect in trans. This change in H3K27me3 may play a role in the observed chromatin and gene expression changes and could be due to sequestration of the KDM4 demethylases for H3K36 and H3K9. Global H3K36me3 levels were unchanged in the presence of H3.3G34W, but H3K36me2 levels were not measured and could well be elevated.

Chromatin immunoprecipitation sequencing (ChIP-seq) analysis by Khazaei and colleagues showed that H3.3G34W incorporation generally paralleled H3.3 localization in the genome. In G34W cells there were areas with both gain and loss of H3.3. Furthermore, despite its inhibitory effect on SETD2, H3.3G34W was not exclusively associated with regions that lost H3K36me3, and genome-wide studies showed more areas of H3K36me3 gain than loss. How the G34W mutant alters the accumulation of H3.3 remains to be explained, but might involve aberrant interactions of H3.3G34W with histone chaperones or other proteins. Mass spectrometry by Lim and colleagues identified a unique set of binding partners specific for H3.3G34W, including SWI/SNF components ARID2 and SMARCA1, as well as HDAC2 (9). Hence, G34W might attract novel sets of proteins from other sites in the genome, changing local chromatin configuration. This group also found that chromatin in H3.3G34W-mutant cells was less accessible to micrococcal nuclease; this might be potentially due to the fact that H3.3 interaction with histone H1 interferes with chromatin compaction and this interaction is disrupted by G34W.

Further indicating an imbalance in PRC2 function, ChIP-seq for H3K27me3 by Khazaei and colleagues showed that in the presence of G34W there were 25% more sites that lost H3K27me3 and PRC2, predominantly from intergenic regions, than gained the mark. Gains of H3K27me3/PRC2 were more focused on promoter and intergenic regions. This, combined with the global decline in H3K27me3, suggests that there is a deficiency of PRC2 activity in these cells, which may lead to a competition among genes for the remaining active machinery. Conversely, regions lacking K3K27me3 showed enhanced H3K36me2 or H3K9me3, which could arise from the release of PRC2 from these regions, followed by occupancy and action of alternative histone methyl transferases and/or by the sequestration of the KDM4 demethylases. In addition to the authors' idea that H3.3G34W leads to H3.3 eviction by in cis inactivation of SETD, decreased H3K36me3, and progressive gene repression by PRC2 recruitment, there may be additional mechanisms by which G34W activates and represses genes (Fig. 1).

Many of the genes downregulated in the presence of G34W, demonstrating gene body H3K27me3 gain and H3K36me3 loss, encoded muscle contractile proteins, suggesting a differentiation block of the malignant stroma cells. Identification of the sequence-specific transcription factors and chromatin regulators present on these genes before the appearance of H3.3G34W may offer insights into their selective silencing. Similarly, the features of genes encoding extracellular matrix proteins that increased H3K36me3 remain to be elucidated. Even at this increasingly detailed but incomplete level of understanding, there are translational implications of the alteration of H3K36 and H3K27 imbalance. In situations including the overexpression of NSD2, loss of the H3K27me3 demethylase KDM6A, and mutations of SWI/SNF proteins, a shift to PRC2-mediated repression can be targeted with EZH2 inhibitors. The results of the studies by Khazaei and colleagues (10) and Lim and colleagues (9) predict that PRC2 inhibitors might increase muscle differentiation and cell-cycle regulatory genes and inhibit GCT growth. In contrast, SETD2 or NSD inhibitors might block expression of aberrantly activated extracellular matrix genes. Therefore, epigenetic drugs and targeting the gene products stimulating bone resorption might augment current surgical treatment of GCT.

No disclosures were reported.