A Synthetic Lethality-Based Strategy to Treat Cancers Harboring a Genetic Deficiency in the Chromatin Remodeling Factor BRG1—Letter

This is the authors' abstract. We don't add key points for this paper.

We would like to comment on a recently published research article in your journal, entitled “A Synthetic Lethality-Based Strategy to Treat Cancers Harboring a Genetic Deficiency in the Chromatin Remodeling Factor BRG1,” by Oike and colleagues (1). This article attempts to demonstrate that BRG1-deficient cancer cells can be inhibited by the suppression of the homolog SWI/SNF catalytic subunit, Brahma (BRM). This is an intriguing concept, which if correct would present potential new therapeutic avenues. The thrust of this article, however, appears seemingly contradictory to a subset of published data from a number of different laboratories about the functional roles of BRG1 and BRM in cancer. In this letter, we would like to promote a balanced discussion by covering those data that both oppose and support this publication as well as explain how these diverging observations might be reconciled.It is worth noting that the SWI/SNF complex functions as a catalyst for gene expression by opening and closing the chromatin, thereby allowing access to the DNA. It has been convincingly demonstrated by several laboratories that SWI/SNF is a cofactor for RB and that loss of BRG1 and/or BRM impairs the function of RB family members (p107 and p130), and in particular RB (2, 3). Consistent with this observation is the fact that reexpression of BRG1, BRM, or BAF47 into deficient cell lines has been reported to inhibit growth (4). Moreover, the inactivation of BRM, BRG1, and BAF47 in mice is tumorigenic to varying degrees (5). These data would seemingly orient one's thinking to accept that SWI/SNF subunit expression blocks cancer, whereas loss of SWI/SNF subunits promotes cancer. As such, BRG1, BRM, and BAF47 are not infrequently referred to in the literature as tumor suppressors or tumor susceptibility genes (4). Oike and colleagues (1) showed that knockdown of BRM in BRG1-deficient cell lines resulted in growth inhibition, which would seem contrary to BRM's accepted role as a tumor susceptibility/suppressor protein. Although data appear to contradict the observations made by Oike and colleagues (1), several other published articles do support their experimental observation. First, Becker and colleagues (6) found that knockdown and expression of BRG1 in the WMM1175 cancer cell line has no effect on proliferation or cell-cycle regulation, whereas p16 could induce cell-cycle arrest in this cell line. Moreover, knockdown of BRM in BRG1-deficient melanoma cells results in activation of p21 and compromises cell line growth in soft agar, a measure of tumorigenicity (7). Finally, BRG1 overexpressed in breast cancer inversely correlates with 5-year survival, and in vitro knockdown of BRG1 compromises cell proliferation in both breast and glioma cancer cell lines (8, 9).Depending on the cell type and environmental conditions, the observed biologic effects that occur when SWI/SNF subunits are either inactivated or restored will depend on which transcription factors and other key cellular proteins are using this complex. As such, perturbation of the SWI/SNF complex could have very different and seemingly conflicting effects. Any observed effect may depend on the levels of BRG1 and BRM present in a given cancer cell—for example, low levels of BRM expression could promote growth within a BRG1-mutated tumor cell, whereas even lower levels might completely inactivate SWI/SNF and cause growth inhibition to occur in the same BRG1-mutated tumor cells. Furthermore, BRG1 is mutated in the majority of studied cell lines (including the ones used in this article); however, BRG1 is not frequently mutated in primary lung tumors, as illustrated by the data from this article (1). As such, an important question to answer is whether this observed effect will occur even when BRG1 is not mutated—for example, if it were epigenetically silenced, which would likely yield low levels of expression that are undetectable by Western blot. Unfortunately, this question may remain unanswered, as there are currently no known cell lines with the correct phenotype (epigenetically silenced) to test this issue. Finally, there are other genes that may be linked to BRM expression and that could also account for the observation by Oike and colleagues, such as Egr1, miR199a-5p, and miR199a-3p, which are expressed at high levels in BRM-negative cancer cell lines and not expressed in BRM-positive cells (10). The induction of these microRNAs could cause the observed growth inhibition to occur via the regulation of other genes.Although this article clearly describes a novel mechanism, it is somewhat problematic in that certain data are only shown in one cell line, such as the induction of p21 in H1299 cells. In addition, these authors conducted colony-forming assays over periods of 10 to 12 days; and yet, in this assay, they also used transient transfection of shRNA interference, in which BRM knockdown effects would be expected to last for only 5 to 6 cell replications, or about 5 to 6 days, depending on the cell line and growth rate. However, more importantly, the main thrust of this article is to show that the silencing of BRM in BRG1-deficient or BRM/BRG1-deficient cell lines leads to inhibition of BRG1-deficient cancer cell lines. However, we and others have derived stable cell lines by knocking down BRM in BRG1-deficient cell lines, such as A549, H661, and H157, in addition to knocking down BRG1, BRM, or both in BRG1/BRM-deficient cell lines such as SW13, H522, and C33A (11, 12). Whether the observations made by Oike and colleagues are insights that could be translated into novel therapeutic avenues remains unanswered and in need of further investigation.Nevertheless, the authors may have made a significant and major contribution to the literature about the mutational status of BRG1 in cancer. Hence, this article has a great deal of merit worth noting. BRM has been found to be epigenetically silenced in probably most, if not all, BRM-deficient cancer cell lines, and loss of BRM in tumors occurs without mutations. To this end, the mechanism of BRM silencing, including HDAC9 and GATA 3 overexpression, can be observed in primary tumors, suggesting that BRM epigenetic silencing occurs in primary lung tumors (12). In contrast, current scientific dogma about the mechanism of BRG1 silencing has been shaped primarily by findings in cell lines, in which most BRG1-deficient cell lines (∼70%) harbor abrogating mutations (13). Mutations as the driving force behind BRG1 silencing has gained further support by a number of recent NextGen sequencing publications that have identified the presence of primarily missense mutations in BRG1-deficient tumors. While mutations have been detected, however, far fewer missense, indel, or other abrogating mutations (<5%) have been identified than the frequency of BRG1-deficient tumors would suggest. For example, in the NextGen sequencing studies, BRG1 truncating mutations (14) were identified in approximately 5% of lung adenocarcinomas. However, this frequency of mutation is well below the 15% to 30% loss of BRG1 expression observed in lung cancer (4). Thus, abrogating BRG1 mutations appears to contribute to, but cannot fully account for, the loss of BRG1 expression in the majority of cases. Some current research articles and reviews, remarkably, report that BRG1 is silenced through mutations and omit that BRG1 is silenced more frequently than mutations occur; such statements leave the reader to infer that mutations are the major mechanism of inactivation. Contradicting this view is the authors' observation that BRG1-deficient tumors lack mutations: This is an important and potentially underemphasized finding of this article. Reversing epigenetically silenced genes may well be a novel form of gene therapy. To this end, a number of compounds can readily reactivate BRM expression, resulting in growth inhibition in cancer cells.The authors then propose the idea of synthetic lethality, based on their experimental data from cell lines, which seemingly contradicts a wealth of experimental data as noted above. However, as there are data in the literature to support their findings, this would seem to make the functionality of BRM apparently dual faceted, depending on the circumstances. To the authors' credit, a novel and more persuasive corollary of their data—and one that is aligned with published findings—is that BRG1 could be epigenetically silenced akin to BRM silencing and thus targetable by therapy.See the Response, p. 4948D. Reisman is CEO of Zenagene and has ownership interest in a patent on BRM polymorphisms. No potential conflicts of interest were disclosed by the other authors.