Epigenetic determinants of metastasis

Abstract

Genetic analyses of cancer progression in patient samples and model systems have thus far failed to identify specific mutational drivers of metastasis. Yet, at least in experimental systems, metastatic cancer clones display stable traits that can facilitate progression through the many steps of metastasis. How cancer cells establish and maintain the transcriptional programmes required for metastasis remains mostly unknown. Emerging evidence suggests that metastatic traits may arise from epigenetically altered transcriptional output of the oncogenic signals that drive tumour initiation and early progression. Molecular dissection of such mechanisms remains a central challenge for a comprehensive understanding of the origins of metastasis.

Keywords: Cancer; Epigenetic; Metastasis.

Figure 1

Alterations in DNA methylation patterns as a source of cancer progression traits. (A) Loss of DNA methylation expands the HIF2A tumour‐initiating pathway output in renal cell carcinoma (RCC) to promote metastasis. In the presence of oxygen, HIF2A is normally targeted for proteosomal degradation, but in VHL mutant RCC cells HIF2A is constitutively expressed and it drives tumorigenesis. DNA demethylation in metastatic cells increases the HIF2A pathway target gene spectrum to include pro‐metastatic CYTIP expression. (B) Melanocyte lineage factor MITF signalling output is expanded through DNA demethylation in support of metastatic progression in melanoma. MITF drives differentiation/pigmentation and proliferation/survival in melanocyte and melanoma cells. Loss of DNA methylation allows MITF to bind additional gene promoters and induce expression of metastasis‐promoting TBC1D16. (C) Metabolic alterations can induce global alterations in DNA methylation. In glioblastoma, IDH1/2 mutants produce the oncometabolite 2‐hydroxyglutarate (2‐HG) which inhibits the activity of TET enzymes. TETs normally mediate demethylation of DNA by converting 5‐methylcytocine (5mC) to 5‐hydroxymethylcytosine (5hmC). Thus, the accumulation of 2‐HG leads to increased DNA methylation resulting in the global CpG island hypermethylator phenotype (CIMP). This can lead to tumour suppressor silencing or loss of the binding of methylation sensitive DNA‐binding factors such as CTCF. CTCF functions as an insulator protein that demarcates chromatin domains. In glioblastoma, CIMP‐induced loss of CTCF binding can allow aberrant PDFGRA activation. Thus, unspecific large‐scale alterations in DNA methylation can result in specific cancer phenotypes. Similar mechanism could activate metastasis genes as well.

Figure 2

Aberrant polycomb repressive complex 2 activity as a source of metastatic cancer phenotypes. (A) In normal cells, PRC2 accumulates at inactive promoters and it functions as a stabilizer of gene silencing through the deposition of the H3K27me3 repressive mark. (B) In tumour cells, depending on the context, PRC2 can both promote and inhibit cancer progression. In some cancers, PRC2 is important for the stable suppression of genes that inhibit metastasis. On the other hand, reduced PRC2 activity can in some contexts promote metastasis by inducing epigenetic instability, which leads to increased transcriptional plasticity. This can facilitate the activation of pro‐metastatic genes such as CXCR4.

Figure 3

Long non‐coding RNAs in the epigenetic regulation of metastasis. (A) LncRNA HOTAIR found in breast cancer metastasis binds and activates PRC2 and LSD1 to remodel the histone methylation landscape. This can induce a pro‐metastatic gene expression profile. (B) SChLAP1 lncRNA expression is found in metastatic prostate cancer cells. Mechanistically, SChLAP1 binds and antagonizes SNF5, a component of the SWI/SNF complex. (C) CCAT2 is transcribed from the enhancer region 335 kb upstream of MYC. Accumulation of CCAT2 in metastatic colon cancer cells stimulates expression of WNT target genes including MYC by promoting TCF4 binding. CCAT2 also mediates enhancer‐promoter interaction upstream of MYC to increase MYC expression in metastatic cells.

Figure 4

Optimization of oncogenic signal output facilitates metastatic cancer progression. Thus far, no metastasis‐specific genetic pathways have been identified. This suggests a model whereby metastatic progression is supported by the same oncogenic pathways that drive tumour initiation and early progression. However, the output of these pathways is not constant but evolves during cancer progression in support of the most aggressive tumour phenotypes. In a mutually non‐exclusive manner, this modulation of pathway output can be both quantitative and qualitative. For example, during metastatic progression a given pathway may become more active generally through increased expression of its core effectors, but there may also be more specific alterations in the expression of individual target genes. At the molecular level, this fine‐tuning of oncogenic signalling can occur through multiple mechanism including genetic and epigenetic alterations, changes in the abundance of transcriptional cofactors and post‐transcriptional modulation of gene products. Non‐cell autonomous factors, whereby signalling is regulated through interaction with stromal components and extracellular matrix, can also modulate the strength of oncogenic signalling in cancer cells and consequently increase metastatic fitness.