Targeting SWI/SNF mutant cancers with tyrosine kinase inhibitor therapy

KEYWORDS:

• SWI/SNF
• SMARCB1
• receptor tyrosine kinase
• tyrosine kinase inhibitor
• cancer
• signal transduction

The SWI/SNF chromatin remodeling complex orchestrates the repositioning of nucleosomes and is necessary for the regulation of multiple fundamental processes including DNA replication and repair, gene expression, and cell division [1]. Genes encoding subunits of the SWI/SNF complex are potent tumor suppressors and inactivating mutations in these genes are found in ~20% of cancers [2,3]. This class of cancers is almost universally fatal and there currently are no effective treatments, in part because the SWI/SNF complex is deemed undruggable. For an overview of the fundamentals of SWI/SNF structure and function, readers are directed to recent excellent reviews [1,4]. Emerging evidence indicates that exploiting tyrosine kinase dependencies arising from SWI/SNF subunit loss may be an effective strategy to overcome cancers driven by these tumor suppressor genes. Specifically, recent preclinical data suggest that tyrosine kinase inhibitors (TKIs) have utility in the treatment of this class of cancers. This editorial summarizes the current knowledge of tyrosine kinase dependencies in SWI/SNF mutant cancers and discusses outstanding questions and barriers that need to be overcome in order to translate these preclinical results into patient benefit.

SMARCB1 is a core subunit of the SWI/SNF complex located at the 22q11 chromosomal region. It is capable of forming two distinct and mutually exclusive complexes containing either the SMARCA2 or the SMARCA4 ATP-dependent helicases [4]. Loss of SMARCB1 is a driver in a number of cancer types, including rhabdoid tumors (RTs), epithelioid sarcomas, renal medullary carcinoma, epithelioid malignant peripheral nerve sheath tumors, meningiomas and extraskeletal myxoid chondrosarcomas [5]. Tyrosine kinase dependencies associated with SMARCB1 deficiency have been most well studied in RTs and include oncogene addiction to the epidermal growth factor receptor (EGFR) and Human EGFR 2 (HER2) [6–8]. Screening three RT cell lines across a panel of 129 small molecular inhibitors, Singh et al. showed that targeting EGFR-HER2 signaling using the EGFR-HER2 TKI lapatinib was effective in blocking tumor cell migration and inducing apoptosis in vitro [8]. Similarly, Kuwahara et al. found that RT cell lines displayed high levels of EGFR expression and inhibiting EGFR phosphorylation with the EGFR TKI gefitinib was capable of reducing cell growth both in vitro and in vivo [7]. These findings are consistent with a recent phosphoproteomic analysis of Smarcb1-deficient RT murine cells derived from a genetically engineered mouse model which showed that Smarcb1 loss led to the upregulation of EGFR phosphorylation and activation of the AKT pro-survival pathway [6]. Interestingly, in this study, lapatinib suppressed EGFR phosphorylation while gefitinib had the opposite effect of elevating receptor activation, although the mechanistic basis of this counter-intuitive finding was not pursued.

In addition to EGFR and HER2, other receptor tyrosine kinases (RTKs) such as fibroblast growth factor receptors (FGFRs), insulin-like growth factor 1 receptor (IGF-1R), and the c-MET receptor are also active in SMARCB1-deficient cancers. Wohrle et al. found that RT cell lines express high levels of FGFRs, including FGFR1 and FGFR2 [9]. They showed that ectopic expression of SMARCB1 in these cell lines led to the downregulation of FGFR protein expression and that the mechanism involved direct binding of SMARCB1 to the FGFR2 promoter locus. Importantly, the use of the FGFR inhibitor NVP-BGJ398 was capable of blocking FGFR signaling through the ERK pathway and inhibiting RT cell growth in vitro and in vivo. Wong et al. recently found that dual inhibition of platelet-derived growth factor receptor alpha (PDGFRα) and FGFR1 induces synergistic cytotoxicity in RT cells [10]. D’Cunja et al. demonstrated that IGF-1R was expressed in RT patient specimens while its cognate ligands IGF-1 and IGF-II were highly expressed in RT cell lines [11]. This finding suggests that an autocrine/paracrine loop may exist in RT cells where high levels of ligand lead to the activation of IGF-1R ultimately promoting tumor growth. Given that the authors did not perform SMARCB1 reexpression experiments, it remains unclear if SMARCB1 directly regulates IGF-1R and/or ligand expression. However, silencing of IGF-1R levels using antisense oligonucleotides was capable of suppressing RT cell proliferation and inducing apoptosis. In addition, IGF-1R downregulation sensitized RT cells to chemotherapy in the form of doxorubicin and cisplatin. Brenca et al. published a short report describing the expression and coactivation of EGFR and c-Met in the VAESBJ epithelioid cell line which lacks SMARCB1 [12]. These receptors activated the AKT and ERK pathways and combined inhibition of both RTKs using two different TKIs had a synergistic effect on cell proliferation versus single-agent treatment alone. Collectively, these studies indicate that SMARCB1 loss results in the upregulation of multiple RTKs and employing TKI combinations may be necessary to block RTK coactivation and suppress compensating growth-promoting signals [13,14].

Increased levels of tyrosine kinases are not only associated with a loss in the SMARCB1 subunit. In a recent study by Papadakis et al., the authors showed that the loss of SWI/SNF subunits SMARCE1, ARID1A, and SMARCA4 results in EGFR upregulation in non-small-cell lung cancer (NSCLC) cell lines [15]. Similar to the regulation of FGFR2 by SMARCB1, SMARCE1 was found to control EGFR transcriptional levels by direct interaction with the regulatory elements of the EGFR locus. The SMARCE1-induced upregulation of EGFR conferred resistance of NSCLC cell lines to c-MET and ALK TKIs. Correspondingly, blocking EGFR activation using EGFR-specific TKIs sensitized SMARCE1-deficient NSCLC cells to MET and ALK kinase inhibitors. Along these lines, Fillmore et al. demonstrated a genetic interaction between SMARCA4 and EGFR in NSCLC [16]. They found that SMARCA4 loss is correlated with high EGFR expression in NSCLC patient specimens and that SMARCA4-containing chromatin remodeling complexes bind to the EGFR regulatory element to disrupt gene transcription in a NSCLC cell line model. Employing a chemical inhibitor screen, Miller et al. showed that mutations in the SWI/SNF subunit ARID1A in ovarian clear cell carcinoma (OCCC) cell lines conferred sensitivity to the multi-target TKI dasatinib [17]. They resolved the cellular target of dasatinib in OCCC cell lines as the YES1 tyrosine kinase and demonstrated that dasatinib could inhibit tumor growth in xenograft models of OCCC. These data provide proof-of-principle evidence that exploiting tyrosine kinase dependencies has utility in a broad range of SWI/SNF subunit deficiencies and supports the use of TKI therapy; in particular the repurposing of FDA approved inhibitors, as a general strategy for this class of cancers.

There remain several conceptual gaps in our knowledge of the regulation of RTK expression by the SWI/SNF complex. It is not clear from the existing data if the loss of different SWI/SNF subunits leads to the upregulation of a similar set of RTK species. Comprehensive analysis of tyrosine kinase signaling across extensive panels of SWI/SNF mutant versus wild-type cancer cell lines using tools such as phosphoproteomics, in combination with computational strategies to elucidate kinase dependencies [18,19] may shed light on the general utility of specific TKIs such as gefitinib, lapatinib, or dasatinib in the treatment of a broad range of SWI/SNF-deficient cancers. In addition, we do not fully understand how cellular context contributes to kinase dependencies associated with SWI/SNF subunit loss. This context specificity is most apparent in the case of SMARCB1-deficiency in RTs. RTs are genomically simple tumors with low rates of mutations and no additional gene amplifications or deletions [20]. Despite SMARCB1 loss being the sole driver of disease, several studies have now shown that different RT cell lines exhibit distinct kinase dependencies [8,9]. Furthermore, Wohrle et al. found that genetic knockdown of SMARCB1 in a panel of non-RT cell lines was incapable of inducing FGFR2 expression [9]. This suggests that molecular heterogeneity and cell-type-specific signaling is present in genomically homogenous RTs and may dictate specific RTK dependencies. With the recent identification of distinct epigenetic subgroups in RTs of the central nervous system [21], it remains to be determined if the observed differences in kinase dependencies is linked to specific epigenetic subgroups in SMARCB1-deficient tumors. Finally, we are only just starting to appreciate the role of interactions between SWI/SNF subunit mutations with other genetic aberrations in promoting tumorigenesis. A prominent example is in OCCC where ARID1A mutations coexist with PIK3CA mutations. Genetically engineered mouse models harboring both mutations promote tumor growth by inducing IL-6 overproduction [22]. How these genetic interactions alter the landscape of kinase dependencies and sensitivity to TKIs such as dasatinib is unknown.

Given the increasing amounts of cancer genome sequencing data that is emerging from the International Cancer Genome Consortium (ICGC) sequencing efforts, it is likely that additional SWI/SNF mutations will be identified in new cancer types in the near future. It should be noted that the role of SWI/SNF complex function in cancer is not completely understood and likely to be context- and tumor specific. For instance, wild-type SMARCA4 is upregulated and has been shown to be an oncogene in breast cancer and direct targeting of SMARCA4 may be a useful approach in this context [23]. It remains to be seen if the success of TKI therapy in cancers such as gastrointestinal stromal tumors (GIST) and mutant EGFR-driven NSCLC will be replicated in SWI/SNF mutant cancers. However, the discovery that a subset of SWI/SNF mutant cancer cell lines is susceptible to TKI therapy provides some confidence that exploiting RTK dependencies has utility in the treatment of this class of hard-to-treat cancers. It also highlights the fact that kinase dependencies are not limited to cancers harboring kinase amplifications and mutations but may also be induced by alterations in non-kinases that are widely considered to be undruggable. Addressing the outstanding questions described above will be key to overcoming some of the barriers to translating this therapeutic approach into the clinical setting and ultimately delivering benefit to patients.

Declaration of interest

P.H. Huang is part of a patent application in relation to methods and materials for treating cancer. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Additional information

Funding

This work was supported by grants from the Institute of Cancer Research, Cancer Research UK, Sarcoma UK, Royal Marsden Cancer Charity, and BBSRC.