Synthetic Lethal Screens Identify Vulnerabilities in GPCR Signaling and Cytoskeletal Organization in E-Cadherin–Deficient Cells

Abstract

The CDH1 gene, which encodes the cell-to-cell adhesion protein E-cadherin, is frequently mutated in lobular breast cancer (LBC) and diffuse gastric cancer (DGC). However, because E-cadherin is a tumor suppressor protein and lost from the cancer cell, it is not a conventional drug target. To overcome this, we have taken a synthetic lethal approach to determine whether the loss of E-cadherin creates druggable vulnerabilities. We first conducted a genome-wide siRNA screen of isogenic MCF10A cells with and without CDH1 expression. Gene ontology analysis demonstrated that G-protein-coupled receptor (GPCR) signaling proteins were highly enriched among the synthetic lethal candidates. Diverse families of cytoskeletal proteins were also frequently represented. These broad classes of E-cadherin synthetic lethal hits were validated using both lentiviral-mediated shRNA knockdown and specific antagonists, including the JAK inhibitor LY2784544, Pertussis toxin, and the aurora kinase inhibitors alisertib and danusertib. Next, we conducted a 4,057 known drug screen and time course studies on the CDH1 isogenic MCF10A cell lines and identified additional drug classes with linkages to GPCR signaling and cytoskeletal function that showed evidence of E-cadherin synthetic lethality. These included multiple histone deacetylase inhibitors, including vorinostat and entinostat, PI3K inhibitors, and the tyrosine kinase inhibitors crizotinib and saracatinib.Together, these results demonstrate that E-cadherin loss creates druggable vulnerabilities that have the potential to improve the management of both sporadic and familial LBC and DGC. Mol Cancer Ther; 14(5); 1213-23. Ó2015 AACR.

Transcriptomic profiling and quantitative high-throughput (qHTS) drug screening of CDH1 deficient hereditary diffuse gastric cancer (HDGC) cells identify treatment leads for familial gastric cancer

“…We found c.1380delA CDH1 SB.mhdgc-1 cells sensitive to taxanes targeting TUBB1 as well as agents targeting mitosis like the aurora kinase inhibitors, a finding also made by Telford and colleagues in the MCF10A CDH1 (−/−) system [12, 29]. Elevated phosphoinositide signaling, both by direct detection of PI(4,5)P2 and PI(3,4,5)P3s messenger intermediates as well as by network analysis of differential gene expression profiling may be a consequence of increased GPCR signaling, which was the most enriched functional cluster (enrichment score = 10.01) in the synthetic lethality screen and confirmed by increased sensitivity to JAK2 inhibition, an immediate downstream effector kinase of GPCR signaling complexes, in the isogenic MCF10A CDH1 (−/−) system studied by Telford et al [12]. Overall, drug sensitivities with overlap to Telford and colleagues’ findings in the CDH1(−/−) isogenic mutant MCF10A model include inhibitors of the PI3K/mTOR axis, including PI-103 followed up in our validation studies, mTOR, aurora kinase inhibitors as well as inhibitors of c-Src kinase.…”

“…These findings of select drug response profiles of hereditary SB.mhdgc-1 gastric cancer cells to MEK and mTOR inhibition, in combination with gene expression dysregulation of ERK and phosphoinositide-mediated signaling networks, also appear to be in line with elevated phospho-ERK: total ERK and phospho-Akt (T308): total Akt ratios observed as a measure of dysregulated signal transduction activity in SB.mhdgc-1 cells. Additionally, these results, in part, overlap with data obtained from comparative gene expression profiling, synthetic RNA lethality and drug screening in the isogenic MCF10A CDH1(−/−) breast fibroblast model as well as prior large-scale gene profiling studies of sporadic clinical gastric cancer specimens [12, 23, 28]. …”

“…Similarly, the recently released data from The Cancer Genome Atlas (TCGA) initiative suggests genomically defined leads, for example, of increased sensitivity to receptor tyrosine kinase or cell cycle modulating agents in the chromosomal unstable (CIN) identified gastric cancers, or increased sensitivity to PI3K kinase, JAK2, and immune checkpoint blockade in the EBV positive subtype [25]. In contrast, leads for drug sensitivities unique to familial gastric cancer due to germline CDH1 mutations are to date largely derived from a synthetic lethality and drug screen in an isogenic CDH1 knockout breast fibroblast model (MCF10A CDH1(−/−)) [12], CDH1-negative systems overexpressing wild type versus mutant forms of CDH1 [26, 27], or correlative tissue studies of the early unique T1a lesions of prophylactic gastrectomy specimens [8, 10]. That hereditary diffuse gastric cancer (HDGC) due to germline CDH1 variants, and sporadic gastric cancers with somatic CDH1 perturbations are unique subtypes of gastric cancer, likely to harbor different drug sensitivity profiles and hence offer opportunities for genotype-directed personalized treatment approaches, is supported by a number of observations; for example, in a detailed CDH1 profiling effort of 174 sporadic and 72 familial gastric cancer specimens Corso et al reported significant differences in clinical outcomes depending on presence and type of CDH1 alternation with the worst survival rates observed across all examined gastric cancers in cases with structural CDH1 defects and in particular in familial tumors [11].…”

“…On the other, there were also significant differences between the pharmacological profile of MCF10A CDH1 (−/−) mutant cells reported by Telford and coworkers and the drug phenotype of c.1380delA CDH1 SB.mhdgc-1 in our study. Among these, for example, sensitivity to MEK or EGFR inhibition was not seen in E-cadherin deficient MCF10A cells and sensitivity to the c-Src kinase inhibitor saracatinib was only shown at one of the three concentrations tested to be significant between the CDH1 MCF10A isogenic cell line pair with overall generally a more modest effect on cell viability [12]. Also, there was discordancy between the sensitivities to HDAC inhibitors and anti-apoptosis inhibitors to BCL2 and XIAP between the two systems.…”

“…In contrast, recent differential gene expression, synthetic lethality, and high throughput drug screening studies in isogenic E-cadherin deficient (−/−) breast MCF10A cells identified several select genetic and pharmacological vulnerabilities in CDH1(−/−) mutant cells [12]. While some of the observed activity profiles in CDH1-deficient cells, like select sensitivities to Src kinase in the CDH1(−/−) mutant MCF10A isoform, were in line with early signaling perturbations observed in T1a cancers of gastrectomy specimens of HDGC patients, other signaling aberrations observed in clinical specimens later in the transition of invasion beyond the gastric mucosa, like FAK and STAT3 kinase activation, were not found to be associated with select drug sensitivities in the MCF10A CDH1(−/−) in vitro system [10, 12]. As the MCF10A CDH1(−/−) isogenic system captures predominantly early vulnerabilities with predominantly chemopreventative translational value, it is thus not known if the synthetic lethalities and drug sensitivities discovered to selectively occur in the CDH1(−/−) mutants capture vulnerabilities of E-cadherin deficient gastric cancers in a more evolved stage which might harbor greater therapeutic value.…”