By stratifying EOC cell lines according to their EMT
By stratifying EOC cell lines according to their EMT stages, we observed a significant higher expression of DDR1 in the epithelial-like cell lines compared to low or undetectable DDR1 in mesenchymal-like cell lines. Similar trend was observed in the tumour samples, with the lowest expression of DDR1 in the Mes subtype, which is associated with poor prognosis. This suggests a loss of DDR1 as EOC Spectinomycin hydrochloride hydrate undergo EMT. To date, the research in EOC has been limited by the lack of an animal model that mimics the disease progression in human. A representative disease model would be useful to elucidate the changes in EMT regulators (possible upregulation of DDR1) during early ovarian tumourigenesis, followed by subsequent cancer metastasis (possible downregulation of DDR1). To assess whether the loss of DDR1 could drive EMT in EOC, we performed RNAi knockdown of DDR1 in OVCAR3 and OVCA429 cell lines and found that the protein levels of E-cadherin remained unchanged. This is consistent with a previously reported study in breast carcinoma, which demonstrated unchanged levels of EMT markers after modulation of DDR1 expression, but a downregulation of DDR1 by ZEB1 was observed upon H-Ras-induced EMT (Koh et al., 2015). Therefore, the loss of DDR1 is an effect of EMT but downregulation of DDR1 may not be sufficient to induce a strong EMT phenotype. However, other studies have shown that downregulation of DDR1 could significantly disrupt the stability of E-cadherin at cell-cell junctions (Yeh et al., 2011), as well as the collective migratory behaviour of epithelial cells (Hidalgo-Carcedo et al., 2011). In our DDR1-knockdown EOC cells, we did observe a more spread-out morphology in siDDR1 #10 transfected OVCAR3, which could indicate a partial change to a more mesenchymal-like phenotype, but this would require future validation. Importantly, many studies have shown that DDR1 promotes cancer cell migration and invasion (Rammal et al., 2016) through its regulation of actomyosin contractility, a function that is independent of its activation by collagen (Hidalgo-Carcedo et al., 2011). Thus, we speculate that epithelial-like EOC cells, which keep stable cell-cell adhesion, would require DDR1 to relocate collectively; whereas the more solitary mesenchymal-like EOC cells would adopt a different migratory behaviour without the involvement of DDR1. Therefore, the EMT-based stratification in EOC and/or other types of cancers, may be useful to predict patient-response for anti-DDR1 drugs. Nevertheless, the expression of DDR1 will remain to be an important predictor, as we observed differential DDR1 expression (high and low) among the epithelial-like cell lines. Although most EMT-related studies focus on inhibiting EMT-inducing pathways, drugs that target epithelial-specific genes such as DDR1 may still benefit patients by impairing the DDR1-dependent pathways involved in cell proliferation and collective migration, specifically in epithelial-like cancer cells. Despite increasing studies on the functions of DDR1, relatively little is known about how DDR1 is regulated. In response to DNA damage, the expression of DDR1 is directly activated by the tumour suppressor p53, which would lead to the accumulation of p53 through MAPK signaling, thus forming a positive feedback loop (Ongusaha et al., 2003). Another transcription factor ZEB1, a well-documented EMT driver, has been reported to repress the transcription of DDR1 through binding at its promoter (Koh et al., 2015). Besides transcription factors, microRNA-199a-5p has been shown to regulate DDR1 post-transcriptionally through binding to its 3′ UTR (Shen et al., 2010). Three recent findings have suggested that DDR1 could also be epigenetically controlled. In patients with idiopathic nonobstructive azoospermia, DDR1 is one of the top ten hypermethylated genes (Ramasamy et al., 2014). In non-small cell lung cancer, a decrease in DNA methylation at DDR1 was found compared to non-tumour lung tissues (Nelson et al., 2012), which is consistent with the DDR1 overexpression described in lung cancer (Ford et al., 2007). In endometrial tissues, a lower CpG methylation of DDR1 was detected during the midsecretory phases, corresponding to its increased expression (Houshdaran et al., 2014). Here, we showed that the CpG methylation levels of DDR1 promoter correlated with the EMT stages of EOC cells, which is consistent with the decreasing expression of DDR1. Our data suggested that the loss of DDR1 gene expression along the EMT gradient is mediated by CpG hypermethylation at its promoter. In addition to gene expression profiling and EMT stratification, the detection of aberrant DNA methylation signals in cancer may help to develop better biomarkers for future drug development. As reviewed by Rammal et al., a number of inhibitors targeting DDR1 have been generated (Rammal et al., 2016), and a combined inhibition of DDR1 and Notch signaling has been suggested as a promising therapeutic strategy for KRAS-driven lung adenocarcinoma (Ambrogio et al., 2016). Understanding the role of DDR1 in EOC progression may open up new avenues to novel therapeutic strategies.