br Material and methods br Results br Discussion

Material and methods
Results
Discussion The diagnosis of pancreatic ductal adenocarcinoma still means a dismal prognosis for patients. The highly homologous T-box transcription factor TBX3 is overexpressed in various solid tumours mediating proliferation, EMT, increased invasion/migration, and the ability for purchase Phos-tag to bypass senescence (Shan et al., 2015; Burgucu et al., 2012; Beukers et al., 2015; Fan et al., 2004; Yarosh et al., 2008). TBX3 is overexpressed in human PDAC, but its precise role in the development and perpetuation of this tumour remains unclear. However, there is evidence for TBX3 as a promoter of metastasis (Fillmore et al., 2010) and its expression has been linked to EMT, as suggested by TBX3-mediated repression of E-cadherin (Wang et al., 2015; Peres and Prince, 2013). In line with previous data from Wang et al. (Wang et al., 2015) we confirm TBX3 overexpression in PDAC patient samples when compared to normal pancreatic tissue. For further evaluation of its oncogenic potential we used well-established and well-characterized human PDAC cell lines and overexpressed TBX3. We could not detect relevant changes in proliferation, which is well in line with observations in other solid tumours like melanoma. Interestingly, TBX3 expression even inhibits cell proliferation in some tumours (Peres et al., 2010). In contrast to TBX3, TBX2 is a well-known inducer of proliferation, and is known to be a downstream target of TBX3 (Muller et al., 2012). One could hypothesize that the lack of proliferation may result from a direct inhibition of TBX2 as a result of TBX3 overexpression. Furthermore, others have already shown TBX3 to be a key mediator for TGFβ1 signalling, thus negatively regulating proliferation and inducing migration and invasion (Li et al., 2013). On the other hand, we detected a significantly increased migratory and invasive capacity in different PDAC cell lines that depended on TBX3 expression. In CAM assays as an in vivo platform to study tumorigenic potential, TBX3 overexpression resulted in a significantly increased tumour formation capacity and consecutively larger tumours, which almost penetrated the CAM. This is well in line with our in vitro data showing increased migration and invasion, and is supported by the role of TBX3 in malignant melanoma, where it drives invasiveness (Peres and Prince, 2013). Of note, previous reports describe TBX3-dependent down-regulation of E-cadherin (Boyd et al., 2013), which is recognized as a crucial step during EMT and subsequent metastatic spread of tumours (Vanharanta and Massague, 2013). While we made similar observations in CAM tumours as suggested by immunohistochemistry and qPCR, we were unable to detect TBX3-mediated induction of EMT in vitro. Neither the epithelial BxPC3 nor the more mesenchymal Panc1 cells showed a TBX3-dependent regulation of EMT marker genes. This might be in line with recent reports showing that EMT is dispensable for metastasis but induces chemoresistance in pancreatic cancer (Zheng et al., 2015), and further supports our dataset where challenging cell culture with low serum, hypoxia or chemotherapy showed no TBX3-dependent effects. Another explanation for the increased tumour formation might by the higher intrinsic tumour-initiating capacity of TBX3 overexpressing cells and a TBX3-related induction of angiogenesis, which we found to be elevated after TBX3 overexpression. Since we have shown previously that TBX3 is critically involved in the self-renewal and the pluripotency of induced pluripotent stem cells (iPSCs) (Weidgang et al., 2013; Russell et al., 2015b), the regulation of a “stemness” phenotype together with the observation of increased aggressiveness in tumour formation led us to investigate whether TBX3 serves as a mediator of a cancer stem cell phenotype in pancreatic cancer. Using sphere formation as an established surrogate assay for self-renewal in vitro, the overexpression of TBX3 resulted in significantly enhanced sphere formation. Moreover, TBX3 is highly expressed within spheres. Pancreatic cancer stem cells are usually enriched by sorting for surface markers such as CD133 (Hermann et al., 2007) or CD44/CD24/EpCAM (Li et al., 2007), or in sphere cultures that promote the survival and expansion of undifferentiated cells in anchorage-independent conditions. Neither approach identifies a pure population, and surface markers have been under intense debate since they may vary considerably on inter-individual differences between patients, (chemo-)therapies, culture techniques, and utilized antibodies (Wicha, 2006). TBX3 is indeed overexpressed in CD133+ cells and sphere cultures. We therefore proceeded to investigate the functional role of TBX3 in CSCs. We have shown previously that TBX3 reinforces an autocrine TBX3–NODAL–SMAD2 signalling loop to mediate mesendodermal differentiation in embryonic stem cells (Weidgang et al., 2013; Russell et al., 2015b; Weidgang et al., 2016). The signalling pathways shared between embryonic stem cells and cancer cells prompted us to use a recent TBX3 ChIP-sequencing dataset to identify putative self-renewal pathways in CSCs. Not only did we identify key signalling networks for the self-renewal of pancreatic CSCs such as TGFB, NODAL and ACTIVIN signalling, but we also validated these findings in a set of core genes previously described to be highly relevant in these processes. Interestingly, the endogenous overexpression of TBX3 induces NODAL and ACTIVIN, whereas exogenous treatment with recombinant ACTIVIN induces TBX3 expression. This autocrine feedback loop is most likely responsible for the self-renewal of CSCs. Pharmacological inhibition of key ACTIVIN receptors (ALKs 4/5) repressed key effects of the TBX3–NODAL–SMAD2 signalling loop such as gene expression and sphere formation. We have shown previously that NODAL/ACTIVIN signalling is active in CSCs and that it is critical for their perpetuation and tumorigenic potential (Lonardo et al., 2011). Previous studies have already provided evidence for NODAL/ACTIVIN-induced angiogenesis: (i) inhibition of ALK-4 reduces tumour-angiogenesis in breast cancer cell lines (Siragam et al., 2012) and (ii) Smad2/Smad3 pathway activation mediated by ALK4 and ALK5 activates Growth Differentiation Factor 11 (GDF11)–induced migration and angiogenesis in endothelial progenitor cells (Williams et al., 2013). This newly postulated mechanism deserves further investigation in PDAC, since we could demonstrate a relevant induction of both VEGF-expression in PDAC cells after TBX3 overexpression and an increase in VEGF and FGF-2 positive vascular structures in the CAM-assay.