These data raised from human tissues suggested a potential
These data raised from human tissues suggested a potential link between TGF-β1 signaling and linear invadosome machinery that may explain TGF-β involvement in HCC progression and invasion. Using HCC cultured cells, we confirmed that TGF-β1 stimulation increases the production and secretion of collagen I and LOXL2 and we demonstrated that DDR1 is also a target of TGF-β1 signaling. As previously shown for LOXL2 (Wong et al., 2014), we established that TGF-β1-dependent regulation of DDR1 is mediated by the canonical signaling pathway involving Smad4. Thus, the regulation of DDR1 expression depends on various factors, as IGF-I was described to also modulate its expression in breast cancer hmg-coa reductase through the PI3K/AKT pathway and the miR-199a-5p miRNA (Mata et al., 2016). Interestingly, without the presence of type I collagen, TGF-β1 is unable to promote invadosome formation and associated matrix degradation. But in presence of type I collagen, TGF-β1 enhances the percentage of HCC cells exhibiting linear invadosomes and this increase is correlated with an improvement of the degradation of the ECM. We demonstrated that this effect is dependent on the collagen receptor DDR1 as its depletion decreases TGF-β-induced invadosome formation. Thus, our study is in favor of a role of DDR1 in the invasive capabilities of liver cancer cells. Accordingly, several studies described the association of DDR1 overexpression with the HCC tumor stage (Shen et al., 2010). DDR1 expression was also correlated with the recurrence of HCC (Jian et al., 2012). Interestingly, miR-199a-5p was described to regulate DDR1 expression (Mata et al., 2016); and a decrease of miR-199a-5p contributes to increase cell invasion by functional deregulation of DDR1 activity in HCC. However, this study revealed that the effect of miR-199a-5p on DDR1 varies among individuals and HCC cell lines (Shen et al., 2010). Thus, it will be interesting to further test the impact of TGF-β1 stimulation on the expression of this miRNA in HCC cell lines. Surprisingly, DDR2 was also found overexpressed in HCC and implicated in HCC invasion and metastasis formation. In this study, the authors demonstrated that DDR2 play this role through ERK signaling pathway and via a stabilization of the EMT factor, SNAIL1 (Xie et al., 2015). In addition, another study described a role of DDR2 in HCC cell proliferation (Park et al., 2015). So it will be necessary to determine the specific role or the redundancy of these two receptors in the invasion process of liver tumor cells. In addition to the receptor, TGF-β1 also regulates the DDR1 ligand meaning the type I collagen and the enzyme involved in its maturation (LOXL2). We previously demonstrated that an increase of type I collagen concentration is associated with the presence of more linear invadosomes in cells (Juin et al., 2012). In vivo, type I collagen is cross-linked, this modification increases the diameter and the stiffness of the fibrils (Parry et al., 1978). Matrix stiffness is now a well-known parameter implicated in tumor invasion (Levental et al., 2009). Indeed, the matrix stiffness modulates cell adhesion, migration and invasion (Kai et al., 2016). Several studies described the importance of matrix rigidity in invadosome formation and/or activity (Collin et al., 2008, Collin et al., 2006, Linder and Wiesner, 2015, Parekh and Weaver, 2015). Modulating the type I collagen fibril cross-linking using chemical compounds or LOXL2 as a physiological cross-linker, we demonstrated that an increase of collagen cross-linking is associated with an increase of the cell capacity to form linear invadosomes. Thus, DDR1 could be considered as a mechanosensor for the type I collagen fibril stiffness. However, the signaling of this mechanosensing is not elucidated. As mentioned before, DDR1 belongs to the RTK family; however, we demonstrated that the kinase activity of DDR1 is not necessary for linear invadosome formation and activity (Juin et al., 2014). Now, it will be necessary to determine how DDR1 signal as a mechansensor for type I collagen fibrils.