Inter and intra-tumoral heterogeneity are major stumbling blocks in the treatment of cancer and are responsible for imparting differential drug responses in cancer patients. Recently, the availability of high-throughput screening datasets has paved the way for machine learning based personalized therapy recommendations using the molecular profiles of cancer specimens. In this study, we introduce Precily, a predictive modeling approach to infer treatment response in cancers using gene expression data. In this context, we demonstrate the benefits of considering pathway activity estimates in tandem with drug descriptors as features. We apply Precily on single-cell and bulk RNA sequencing data associated with hundreds of cancer cell lines. We then assess the predictability of treatment outcomes using our in-house prostate cancer cell line and xenografts datasets exposed to differential treatment conditions. Further, we demonstrate the applicability of our approach on patient drug response data from The Cancer Genome Atlas and an independent clinical study describing the treatment journey of three melanoma patients. Our findings highlight the importance of chemo-transcriptomics approaches in cancer treatment selection.
The propensity of cancer cells to transition between epithelial and mesenchymal phenotypic states via the epithelial–mesenchymal transition (EMT) program can regulate metastatic processes, cancer progression, and treatment resistance. Transcriptional investigations using reversible models of EMT, revealed the mesenchymal-to-epithelial reverting transition (MErT) to be enriched in clinical samples of metastatic castrate resistant prostate cancer (mCRPC). From this enrichment, a metastasis-derived gene signature was identified that predicted more rapid cancer relapse and reduced survival across multiple human carcinoma types. Additionally, the transcriptional profile of MErT is not a simple mirror image of EMT as tumour cells retain a transcriptional “memory” following a reversible EMT. This memory was also enriched in mCRPC samples. Cumulatively, our studies reveal the transcriptional profile of epithelial–mesenchymal plasticity and highlight the unique transcriptional properties of MErT. Furthermore, our findings provide evidence to support the association of epithelial plasticity with poor clinical outcomes in multiple human carcinoma types.
Recent evidence has implicated the transmembrane co-receptor neuropilin-1 (NRP1) in cancer progression. Primarily known as a regulator of neuronal guidance and angiogenesis, NRP1 is also expressed in multiple human malignancies, where it promotes tumor angiogenesis. However, non-angiogenic roles of NRP1 in tumor progression remain poorly characterized. In this study, we define NRP1 as an androgen-repressed gene whose expression is elevated during the adaptation of prostate tumors to androgen-targeted therapies (ATTs), and subsequent progression to metastatic castration-resistant prostate cancer (mCRPC). Using short hairpin RNA (shRNA)-mediated suppression of NRP1, we demonstrate that NRP1 regulates the mesenchymal phenotype of mCRPC cell models and the invasive and metastatic dissemination of tumor cells in vivo. In patients, immunohistochemical staining of tissue microarrays and mRNA expression analyses revealed a positive association between NRP1 expression and increasing Gleason grade, pathological T score, positive lymph node status and primary therapy failure. Furthermore, multivariate analysis of several large clinical prostate cancer (PCa) cohorts identified NRP1 expression at radical prostatectomy as an independent prognostic biomarker of biochemical recurrence after radiation therapy, metastasis and cancer-specific mortality. This study identifies NRP1 for the first time as a novel androgen-suppressed gene upregulated during the adaptive response of prostate tumors to ATTs and a prognostic biomarker of clinical metastasis and lethal PCa.
CMTM6 is a critical regulator of cell surface expression of PD-L1 in tumor cells, but little is known about the transcriptional regulation of CMTM6. Here we report that the expression of CMTM6 positively correlates with the epithelial to mesenchymal transition (EMT) score in breast cancer cell lines and with the major EMT marker Vimentin in triple-negative breast cancers (TNBC). We showed that CMTM6 is concomitantly overexpressed with PD-L1 in breast mesenchymal compared with the epithelial cells. Driving a mesenchymal phenotype in SNAI1-inducible MCF-7 cells (MCF-7Mes cells) increased both PD-L1 and CMTM6. CMTM6 silencing in MCF-7Mes cells partially reduced cell surface expression of PD-L1, indicating that a proportion of the PD-L1 on the surface of MCF-7Mes cells depends on CMTM6. We also found a positive correlation between CMTM3 and CMTM7 expression with EMT score in breast cancer cells, and with Vimentin in TNBC patients. Dual knockdown of CMTM6 and CMTM7 significantly decreased PD-L1 surface expression in MCF-7Mes cells, indicating that both CMTM6 and CMTM7 regulate the expression of PD-L1. This study highlights the importance of CMTM6 and CMTM7 in EMT-induced PD-L1 and suggests that EMT, CMTM6 or CMTM7 modulators can be combined with anti-PD-L1 in patients with highly aggressive breast cancer.
Background: Triple-negative breast cancers (TNBC) have a relatively poor prognosis and responses to targeted therapies. Between 25 and 39% of TNBCs are claudin-low, a poorly differentiated subtype enriched for mesenchymal, stem cell and mitogen-activated signaling pathways. We investigated the role of the cell-surface co-receptor NRP1 in the biology of claudin-low TNBC. Methods: The clinical prognostic value of NRP1 was determined by Kaplan–Meier analysis. GSVA analysis of METABRIC and Oslo2 transcriptomics datasets was used to correlate NRP1 expression with claudin-low gene signature scores. NRP1 siRNA knockdown was performed in MDA-MB-231, BT-549, SUM159 and Hs578T claudin-low cells and proliferation and viability measured by live cell imaging and DNA quantification. In SUM159 orthotopic xenograft models using NSG mice, NRP1 was suppressed by shRNA knockdown or systemic treatment with the NRP1-targeted monoclonal antibody Vesencumab. NRP1-mediated signaling pathways were interrogated by protein array and Western blotting. Results: High NRP1 expression was associated with shorter relapse- and metastasis-free survival specifically in ER-negative BrCa cohorts. NRP1 was over-expressed specifically in claudin-low clinical samples and cell lines, and NRP1 knockdown reduced proliferation of claudin-low cells and prolonged survival in a claudin-low orthotopic xenograft model. NRP1 inhibition suppressed expression of the mesenchymal and stem cell markers ZEB1 and ITGA6, respectively, compromised spheroid-initiating capacity and exerted potent anti-tumor effects on claudin-low orthotopic xenografts (12.8-fold reduction in endpoint tumor volume). NRP1 was required to maintain maximal RAS/MAPK signaling via EGFR and PDGFR, a hallmark of claudin-low tumors. Conclusions: These data implicate NRP1 in the aggressive phenotype of claudin-low breast cancer and offer a novel targeted therapeutic approach to this poor prognosis subtype.
IGF-I can bind to the extracellular matrix protein vitronectin (VN) through the involvement of IGF-binding proteins-2, -3, -4, and -5. Because IGF-I and VN have established roles in tumor cell dissemination, we were keen to investigate the functional consequences of the interaction of IGF-I, IGF binding proteins (IGFBPs), and VN in tumor cell biology. Hence, functional responses of MCF-7 breast carcinoma cells and normal nontumorgenic MCF-10A mammary epithelial cells were investigated to allow side-by-side comparisons of these complexes in both cancerous and normal breast cells. We demonstrate that substrate-bound IGF-I-IGFBP-VN complexes stimulate synergistic increases in cellular migration in both cell types. Studies using IGF-I analogs determined this stimulation to be dependent on both heterotrimeric IGF-I-IGFBP-VN complex formation and the involvement of the IGF-I receptor (IGF-IR). Furthermore, the enhanced cellular migration was abolished on incubation of MCF-7 and MCF-10A cells with function blocking antibodies directed at VN-binding integrins and the IGF-IR. Analysis of the signal transduction pathways underlying the enhanced cell migration revealed that the complexes stimulate a transient activation of the ERK/MAPK signaling pathway while simultaneously producing a sustained activation of the phosphatidylinositide 3-kinase/AKT pathway. Experiments using pharmacological inhibitors of these pathways determined a requirement for phosphatidylinositide 3-kinase/AKT activation in the observed response. Overexpression of wild type and activated AKT further increases substrate-bound IGF-I-IGFBP-VN-stimulated migration. This study provides the first mechanistic insights into the action of IGF-I-IGFBP-VN complexes and adds further evidence to support the involvement of VN-binding integrins and their cooperativity with the IGF-IR in the promotion of tumor cell migration.