In this study we stably transfected the highly invasive
In this study, we stably transfected the highly invasive, triple-negative (ER-, PR-, HER2-), human breast cancer cell line MDA-MB-231 with wild-type (WT) NHERF1, with NHERF1 mutated from serine to alanine such that they could no longer be phosphorylated in serine 279 (S279A), in serine 301 (S301A) or in either (S279A/S301A double mutant, Dmuts). We investigated the contribution of these residues of NHERF1 to cancer cell phenotype and invasion in vitro and in tumor growth and metastasis in vivo. Our results indicate that the NHERF1 phosphorylation state differentially controls (i) NHE1 activity and invadopodial-dependent ECM degradation, (ii) vasculogenic mimicry (VM) and the angiogenic secretome, (iii) soft-agar and 3D growth and (iv) the metastatic organotropism of cancer cells. We conclude that the possibility to phosphorylate the two serines separately or together permits the cell to very finely tune its phenotypic behavior and eventual metastatic organotropism.
Materials and methods
Discussion The complex set of processes required to produce metastatic lesions is based on tumor cell phenotypic plasticity, which enables the cell to acquire pro-metastatic phenotypes and organotropism [3,63,64]. This plasticity relies on specific signal transduction systems  which are orchestrated by scaffolding proteins. NHERF1 has been shown to finely regulate the various metastatic behaviors of cancer Tunicamycin [6,7,30] and this capacity was demonstrated to occur through its PDZ binding domains, permitting the shift from one to another phenotypic outcome, and thus conferring a specific metastatic competence to the cell [24,30]. However, the in situ mechanism(s) underlying this regulation of PDZ function is still only partially known [32,66]. Here, we have identified the phosphorylable S279 and S301 of NHERF1 as key determinants in the regulation of metastatic phenotype expression and organotropic outcome. S279A and S301A phosphorylation-dead NHERF1 mutant-expressing cancer cells have (i) a greatly reduced invasive capacity (Fig. 1, Fig. 2), suggesting that the cooperative action of phosphorylation at S279 and S301 of NHERF1 are necessary for these properties; (ii) a release of the inhibition of anchorage-independent tumor growth observed in WT NHERF1-overexpressing cells, especially by the phosphorylation of S301 (Fig. 3); (iii) an increased ability to form tumor vascular channel-like structures (Fig. 4A–B); together with (iv) changes in angiogenic factor secretion (Fig. 4C), and finally (v) a switch in metastatic organotropic choice. Indeed, cells over-expressing WT-NHERF1 displayed a greatly enhanced lung metastatic load consisting of high numbers of micro-metastases, while cells over-expressing the NHERF1 double S279A/S301A mutant were almost completely redirected towards bone metastasis (i.e. high osteotropism) (Fig. 5). The phosphorylation of NHERF1 serines 279 and 301 has been shown to be mediated by the cdc2/cyclin B kinase [, , ] and is involved in NHERF1 ability to oligomerize during mitosis , in NHERF1 increased proteosomal degradation (by human papillomavirus oncoproteins) and subsequent PI3K/AKT signaling pathway activation  and in actin cytoskeleton remodelling and cell-ECM adherence . We indeed found that cdc2/cyclin B kinase is involved in phosphoNHERF1-driven regulation of both anchorage-independent 3D growth and invadopodia-mediated ECM digestion since incubation of WT-NHERF1 cells with the cdc2 kinase inhibitor, roscovitine, dose-dependently decreases NHERF1 phosphorylation and replicates the increase in anchorage-independent growth and decrease in invadopodia proteolytic activity observed in the Dmut clone (Supplemental Fig. 3). Human breast cancer cells are able to form vasculogenic mimicry (VM)-like channels, and overexpression of WT-NHERF1 impairs this ability through its PDZ2 binding domain (HRF2 clone) . Importantly, VM and angiogenic secretome are increasingly regulated by S279 and S301, such that the double S279A/S301A mutant, like the PDZ2 mutant, fully restored these abilities. These results suggest that NHERF1 can orchestrate, via reversible phosphorylation, a bidirectional transition program called “mesenchymal-vasculogenic transition” [30,67,68]. This transition could, in part, be regulated by the release of pro-angiogenic molecules. Since VM was reported to be resistant to angiogenesis inhibitors in tumor therapy [68,69], elucidation of molecular mechanisms by which deregulated NHERF1 signaling promotes both angiogenesis and VM could open new therapeutic strategies against tumor microcirculation.