In this issue, Deblois and colleagues show how taxane-resistant triple-negative breast cancer cells evade viral mimicry response as a result of metabolic alteration, DNA hypomethylation, and relocation of histone H3K27 trimethylation (H3K27me3). This adaptation confers a therapeutic vulnerability to the inhibition of the H3K27me3 methyltransferase EZH2 in resistant cells, leading to tumor growth inhibition by viral mimicry reactivation.
See related article by Deblois et al., p. 1312.
Tumor progression by resistance to pharmacologic therapy is a major issue for managing patients with cancer, with few treatment alternatives, and is associated with poor prognosis. This is particularly the case for triple-negative breast cancer (TNBC), which initially tends to respond to treatment but often develops resistance, leading to the advancement of the disease. Thus, deciphering by which mechanism the cancer cells adapt to the cytotoxic pressure of the drugs is of utmost importance. During the last decades, efforts by the research community have led to the expansion of tumor biomarker discovery, allowing clinicians to choose adequate therapeutic strategies for both original tumors and relapses, which has led to increased patient lifespan. These biomarkers are present at all levels from DNA to proteins. In this field of opportunities, epigenetics is gaining importance for diagnosis and prognosis but also in the prediction of response to treatment. In this last application, both DNA methylation and histone marks can be reversed by specific inhibitors (1) changing chemosensitivity profiles. For example, the use of the EZH2 inhibitor tazemetostat (Tazverik) has recently been approved by the FDA for patients carrying an activating EZH2 mutation in advanced or metastatic epithelioid sarcoma not eligible for complete resection, and adult patients with relapsed or refractory (R/R) follicular lymphoma (2). The ongoing study of tazemetostat and similar drugs in various clinical trials is also encouraging for other neoplasia types, and it is an illustrative example of how epigenetics can initially influence prognosis and guides therapy development. In this regard, the recent interest in gene expression regulation by the H3K27me3 repressive mark in TNBC and the concomitant significant upregulation of EZH2 in TNBC compared with the other breast cancer subtypes (3) makes the study by Deblois and colleagues in this issue (4) a useful reference for better drug resistance understanding and therapy development. The mechanism of resistance to therapy pinpointed herein (4) is the escape from viral mimicry response. This response normally occurs through the activation of human endogenous retroviral elements (HERV) that leads to double-stranded RNA (dsRNA) production and IFN-mediated T-cell recognition. In resistant cells, the DNA methylation pattern impedes the retroviral element activation, and so the use of hypomethylating agents has been shown to reactivate HERVs, leading to apoptosis and tumor growth inhibition (5, 6).
In breast cancer, patients with BRCA1 and BRCA2 germline mutations are sensitive to PARP inhibitors, but cell lines with CpG island methylation of BRCA1 have also been shown to be sensitive to these inhibitors (7). Comparative studies revealed that germline mutations or CpG island methylation had the same pattern of gene expression and, overall, a similar phenotype. The authors found that close to 37% of the TNBC tumors exhibited methylation of BRCA1, offering a potential therapeutic solution for these patients (7). More recently, it was shown in TNBC patient-derived xenograft (PDX) models that the tumors, initially sensitive to docetaxel, a type of taxane, contained a resistant subpopulation of CD49f-positive cells, also called integrin α6, a marker of stem cells (8). This subpopulation is enriched during treatment due to apoptosis of the sensitive cells, which can lead to metastasis and fatal progression. This context underlines the evolving dimension of tumors, the interplay between the different intratumor subpopulations, and the selective character of tumor treatments. To avoid the emergence of treatment resistance of nonresponsive tumor cells due to intratumor heterogeneity, single-cell DNA methylation and histone modification profiling (9) has emerged as a promising tool to identify and study these clones that can provide clues for possible multitherapy approaches. Clinically, the need for patient stratification with TNBC is important, and gene methylation signatures can help to predict how the patients will respond to treatment (10).
In this context, Deblois and colleagues (4) go a step further with the finding that metabolic reprogramming, DNA methylation, and chromatin marks are interdependent to escape viral mimicry response. Indeed, the decrease of the universal methyl-donor S-Adenosyl-Methionine (SAM) metabolism in taxane-resistant TNBC cells leads to DNA hypomethylation, which could have a pro-response effect, but the H3K27me3 reallocation at specific chromatin domains prevents the response activation.
Using paclitaxel-resistant cell line models generated from the TNBC cell lines, the authors observed that methionine metabolism was altered in resistant models compared with parental cells. This alteration leads to exogenous l-methionine dependency of taxane-resistant cells for proliferation and an improved antioxidant capacity. As SAM is the methyl donor of the DNA methyltransferases (DNMT), the authors investigated whether SAM reduction leads to DNA hypomethylation. They observed a significant decrease of DNA methylation in the resistant models relative to the parental cells. Global DNA hypomethylation was also assayed by identification of differentially methylated regions (DMR) between taxane-resistant and parental cells. These data were also confirmed in taxane-resistant PDX models. Importantly, hypomethylation was enriched in intergenic regions in both resistant cell lines and resistant PDXs, and hypermethylation marks were mostly found at gene promoter regions associated with IFNα/β and cytokine signaling gene signature.
As SAM is also the methyl donor of histone methyltransferases, the authors examined whether the levels and the distribution of the histone marks were affected by methionine metabolism alteration. The global level was not affected, but the distribution was significantly different between resistant and parental cells, meaning a reallocation of these repressive marks. This reallocation leads to close cluster formations grouped in Large Organized Chromatin Lysine (K) domains (LOCK). To understand the role of the H3K27me3 marks in LOCKs, they classified them as taxane-resistant or parental cells. They identified close to 50% more taxane-resistant LOCKs compared with the parental ones, but importantly, their average size was more than 20 times wider (630 Kb versus 29 Kb), meaning a gain of widespread heterochromatin domains over gene-poor regions in taxane-resistant TNBC cells driven by H3K27me3 reallocation. By comparing DNA methylation and H3K27me3 location, the authors found that the differentially methylated CpGs inside H3K27me3 LOCKs are hypomethylated in taxane-resistant cells, supporting a compensatory repressive role of H3K27me3 reallocation in LOCKs over DNA-hypomethylated regions. To go further, they showed that taxane resistant–specific H3K27me3 LOCKs overlapped with families of transposable elements (TE) such as LINE1, LTR, and HERV. After having shown that only parental cells accumulate dsRNA under paclitaxel treatment, they validated that H3K27me3 reprogramming provides a transcriptional repression of the hypomethylated TEs in the resistant cells. Treatment of resistant models with two EZH2 inhibitors (UNC1999 and GSK343) led to dsRNA accumulation and induced the expression of a subset of HERVs. Moreover, the gene expression profile of taxane-treated relative to untreated parental cells positively associated with the viral mimicry response gene signature, whereas an inverse correlation was observed with taxane-resistant relative to parental cells. Using gene expression data from a cohort of patients with TNBC treated in part with paclitaxel, they observed that genes from the viral mimicry signature were significantly downregulated in patients with TNBC with progressive disease upon treatment, compared with patients responding to treatment. In taxane-resistant cells treated with the EZH2 inhibitor, they observed an induction of gene expression related to IFN response, and an additional TE expression upon depletion of EZH2 by siRNAs. Thus, using paclitaxel-resistant organoids derived from PDXs, they confirmed the awakening of viral mimicry response through EZH2 inhibitor treatment, validating that reallocation of H3K27me3 marks in taxane-resistant TNBC represses TEs and viral mimicry response. Importantly, the inhibition of EZH2 by both inhibitor or siRNAs led to growth and survival decrease of the taxane-resistant cells only, and IFN response inhibition by depletion of IFIH1 rescued cell growth inhibition, confirming the interplay between TEs, dsRNAs, and IFN expression for viral mimicry response. Finally, the authors validated the specific effect of UNC1999 in mice with paclitaxel-sensitive versus paclitaxel-resistant PDXs, where the resistant PDXs had significant growth delay. All these results validate the viral mimicry escape of the taxane-resistant TNBC cells, driven by epigenetic rewiring, TEs, and IFN response repression, and also confirm the epigenetic vulnerability of these cells by the use of EZH2 inhibitor.
It would be interesting to further explore the drugs that awake the viral mimicry response in cancer, and those that inhibit the pathway (Fig. 1). The authors focused on paclitaxel, but we can wonder whether other taxane treatments and other classes of anticancer drugs in breast cancer and beyond might also block the viral mimicry response to escape drug-related tumor inhibition. Future research with the approved tazemetostat would be valuable to tackle this and other issues. These studies, together with improved mouse models closer to the real clinical situation, would permit a better evaluation of the efficacy of EZH2 inhibitors in the context of breast tumors. It is noteworthy that EZH2 inhibitor efficacy was independent of its mutational status in this model, but the EZH2 inhibitor has only been approved in the context of EZH2-mutant tumors. In TNBC, the lack of expression of hormone receptors and HER2 overexpression/amplification reduce chemotherapy options, and the heterogeneity of the tumors is an important point to consider for tumor eradication. Overall, this study highlights the importance of epigenetics in the establishment of chemotherapy resistance mechanisms in TNBC, and how targeting DNA methylation and histone modifications could be a useful strategy to restore sensitivity to the pharmacologic treatment.
Disclosure of Potential Conflicts of Interest
M. Esteller is a consultant for Ferrer International and Quimatryx. No potential conflicts of interest were disclosed by the other author.