In the present study we investigated the migratory activity of the two hybrid cell lines M13MDA435-1 and M13MDA435-3 in dependence of EGF and SDF-1α stimulation in comparison to M13SV1-EGFP-Neo breast epithelial cells exhibiting stem cell characteristics and HS578T-Hyg breast cancer cells. Our data show that each hybrid clone exhibited a unique EGF and SDF-1α mediated migratory activity, which was most likely attributed to a unique signal transduction cascade kinetics pattern in each hybrid cell clone.
Cell migration is a complex process and is directed by the interplay of several signal transduction pathways initiated by various ligands such as growth factors, chemokines and extracellular matrix components that activate growth factor receptors, chemokine receptors and integrins [41, 42]. The involvement of different pro-migratory pathways initiated by different inducers is nicely seen in M13SV1-EGFP-Neo breast epithelial cells exhibiting stem cell properties. EGF treatment resulted in a markedly induced migratory activity of the cells. In accordance to M13MDA435-3 hybrid cells the PI3K inhibitor Ly294002 impaired both the spontaneous and EGF induced migration of M13SV1-EGFP-Neo cells. However, the inhibitory effect of Ly294002 on the spontaneous migration of the cells was rather moderate (about 25%), whereas the inhibitory effect of Ly294002 on the EGF induced migration of M13SV1-EGFP-Neo cells was much higher (about 90%). This indicates that inhibition of PI3K/AKT signalling impairs both the spontaneous and the EGF induced migration.
Since only M13SV1-EGFP-Neo breast epithelial cells showed an EGF-dependent calcium influx and EGFR/HER2/PLC-γ1 signalling has been suggested as a key regulatory step in cell migration [34, 43] we conclude that the markedly increased migratory activity of the cells in response to EGF was attributed to the induction of this signal transduction cascade mediated by EGFR/HER2 heterodimer signalling. Moreover, data of Falasca and colleagues provided evidence that PLC-γ1 signalling does also depend on PI3K activity . PI3K generates phosphatidylinositol-3,4,5-triphosphate to which PLC-γ1 binds with its pleckstrin homology domain . By doing so PLC-γ1 translocates to the plasma membrane and is subsequently activated by tyrosine phosphorylation . PI3K is highly activated by HER2/HER3 signalling [45, 46]. In the context of breast cancer HER2/HER3 heterodimer signalling has been referred to function as an oncogenic unit . Thus inhibition of PI3K activity by Ly294002 also impairs PLC-γ1 signalling, which might be a suitable explanation for the effective inhibition of Ly294002 on the EGF induced migration of M13SV1-EGFP-Neo cells initiated by the interplay of EGFR/HER2/PLC-γ1 and HER2/HER3/PI3K signalling. Similar findings were recently demonstrated by Balz et al. demonstrating that the EGF induced calcium influx concomitant with the migratory activity was markedly decreased in EGFR/HER2/HER3 positive MDA-HER2 breast cancer cells treated with the PI3K inhibitor wortmannin .
This assumption is further substantiated by PD98059 data. MAPK signalling does not interfere with PLC-γ1 signalling and because of that inhibition of MAPK signalling did not result in a decreased EGF-induced migration of M13SV1-EGFP-Neo cells. Moreover, these data further show the differential roles of signal transduction cascades in regulating cell migration. Even though MAPK signalling is involved in the spontaneous migration of M13SV1-EGFP-Neo cells, the inhibition of this pathway by PD98059 is superimposed by the PLC-γ1 signalling pathway.
Of interest was the finding of the diametrically opposed effect of the PI3K inhibitor Ly294002 on M13MDA435-1 and M13MDA435-3 hybrid cells. While Ly294002 effectively inhibited the migratory activity of M13MDA435-3 hybrid cells, the migration of M13MDA435-1 hybrids was potently stimulated by this compound. We assume that this effect might be attributed to a differential RAF-AKT crosstalk  in both hybrid cell lines. The RAF-AKT crosstalk describes the interplay between AKT and RAF-1 signalling, whereby a strong PI3K/AKT activation leads to a AKT dependent inhibition of RAF-1 concomitant with abrogation of MAPKp42/44 signalling . Inhibition of AKT signalling, e.g., by Ly294002 treatment, relieves this block and restores RAF-1-MAPKp42/44 signalling . Moelling et al. observed increased phosphorylated MAPKp42/44 levels in IGF and Ly294002 co-treated cells , which is similar to our work.
The putative RAF-AKT crosstalk in M13MDA435-1 hybrid cells was further validated by Western Blot analysing RAF-1 phosphorylation at position S259, which is consistent with AKT mediated RAF-1 inactivation . In fact, Western Blot analysis of LY294002 treated M13MDA435-1 hybrid cells showed decreased pRAF-1 S259 levels, which were neither detected in M13MDA435-3 hybrid cells nor in the parental cells. This finding would be in view with the assumption of an active RAF-AKT crosstalk in M13MDA435-1 hybrid cells. Treatment of M13MDA435-1 cells with Ly294002 blocks PI3K/AKT signalling, thereby impairing AKT mediated RAF-1 S259 phosphorylation, which in turn restores RAF-1-MAPKp42/44 signalling including MAPKp42/44 phosphorylation.
Whether the differential activity of the RAF-AKT crosstalk was attributed to the cell fusion process or to another mechanism is unclear. Cell fusion is a random process associated with chromosomal instability, loss of single chromosomes and aneuploidy in emerging hybrid cells [6, 10, 11]. As a consequence each hybrid clone originate individually and because of that each hybrid clone exhibit a unique gene expression profile concomitant with a unique signal transduction cascade pattern. On the other hand, Rommel and colleagues showed that the regulation of RAF-AKT crosstalk also depended on the differentiation state of the cell . AKT activation inhibited the RAF-MAPK pathways in differentiated myotubes, but not in their myoblast precursors, which might be attributed to a stage-specific ability of AKT to form a complex with RAF . Thus, the deactivated RAF-AKT crosstalk in M13MDA435-3 hybrid cells might be a relict of stem cell-like phenotype of M13SV1-EGFP-Neo cells. In this cell line no RAF-AKT crosstalk was observed. Activation of the RAF-AKT crosstalk in M13MDA435-1 hybrid cells might thus be attributed to some kind of a maturation/differentiation process that may have occurred in this hybrid cell line during its evolution.
The finding that both hybrid cells responded to EGF with an increased migratory activity, whereas the parental tumour cell line MDA-MB-435-Hyg did not, indicate the potency of cell fusion in changing the fate of cancer cells. Moreover, cell fusion could not only revert the migratory phenotype of a parental cancer cell, but also could give rise to hybrid cells each exhibiting a unique signal transduction cascade pattern. This knowledge would of interest in case of anti-cancer therapies specifically targeting single molecules and pathways. As mentioned above, the HER2/HER3 heterodimer has been referred to act as an oncogenic unit in breast cancer [45, 46] due to strong activation of the pro-survival PI3K/AKT pathway. Because of that HER3 has been recommended as a suitable target for novel anti-cancer therapies. For instance, inhibition of HER2/HER3 heterodimer signalling either by shHER3 mediated knock-down or pertuzumab/trastuzumab treatment was correlated with markedly reduced growth of BT474M1 and MDA-MB-175 induced tumours in vivo . Similar results were obtained by a Cre-mediated HER3 deletion or by a chemically stabilized HER3 antisense oligonucleotide, in a murine model of mammary carcinoma . While such strategies would effectively block HER3/PI3K/AKT signalling it would be of interest to investigate the effect of this blockage in the hybrid cells used in this study. Inhibition of AKT signalling, e.g., by Ly294002 treatment, can restore RAF-1-MAPKp42/44 signalling . Thus, a similar effect should occur in case of HER3 inhibition, e.g., by using recombinant monoclonal antibodies targeting HER2 and HER3 molecules.
In addition to EGF we also investigated the migratory activity of the cells within the presence of SDF-1α since the parental M13SV1-EGFP-Neo cell line and both hybrid cell lines were slightly positive for CXCR4. The SDF-1α receptor CXCR4 has been linked to the organ-specific metastatic spreading of breast cancer [51, 52]. The finding that both hybrid cells were positive for CXCR4 may suggest that cell fusion might be a mechanism how (breast) cancer cells could acquire the ability to metastasise in an organ-specific manner. Several lines of evidence indicated that fusion of tumour cells with other cells, such as macrophages, gave rise to hybrid cells exhibiting an increased metastatogenic capacity [5, 6, 31, 53]. However, cell migration data revealed that the migratory activity of both hybrid cell lines (and the parental breast cancer cell line) was markedly impaired by this chemokine, which is contrary to parental M13SV1-EGFP-Neo breast epithelial cells. Here CXCR4 expression was correlated to a slightly, but not significantly increased migratory activity of the cells upon SDF-1α stimulation.
The SDF-1α receptor CXCR4 belongs to the family of G-protein coupled receptors (GPCRs) [54, 55]. Recent data indicated that CXCR4 signalling may not be limited to Gαi as first thought, but that CXCR4 can couple to other Gα proteins such as Gαq, Gαo and Gαs [54, 56]. Thus, the differential SDF-1α migratory activities of the analysed cells might be attributed to differential expression levels of Gα subunits. For instance, adenyl cyclase is activated by Gαs, but blocked by Gαi . In neutrophil granulocytes the interleukin-8 (IL-8) dependent activation of the adenyl cyclase/PKA pathway has been associated with the induction of a stop-signal, thereby impairing the cells overall and fMLP induced migration in a dose-dependent manner . Activation of the adenyl cyclase/PKA pathway promotes the sequestration of cytosolic calcium in cells , thereby impairing cell migration.
Holland and colleagues demonstrated that association of G-protein αβγ-heterotrimers with CXCR4 receptor and induction of a SDF-1α specific signalling did only occur in highly invasive breast cancer cell lines, but not in non-invasive cell lines . Here, the blockade of non-metastatic cell lines seemed to be due to the inability of G protein α and β subunits to form a heterotrimeric complex with CXCR4, which, on the other hand, was observable in highly invasive cell lines .
However, both hybrid cell lines as well as the parental MDA-MB-435-Hyg breast cancer cell line responded to SDF-1α treatment with a decreased migratory activity. Thus we assume that these cells harbour a functional CXCR4 receptor coupled with a functional G-protein αβγ-heterotrimer. In addition to heterotrimeric G-protein dependent signals CXCR4, like all GPCRs, is a substrate for G protein receptor kinases (GRKs) . GRK-mediated phosphorylation of CXCR4 creates a binding site for β-arrestins, thereby enabling a heterotrimeric G-protein independent signalling . The CXCR4-β-arrestin complex is a potent inducer of the MAPKp42/44 pathway via RAF-1 . MAPKp42/44 phosphorylation was observed in SDF-1α treated M13MDA435-1 hybrid cells. However, both Gαi and Gαq coupled GPCRs do also stimulate MAPK activation  and thus it remains unclear whether MAPKp42/44 was activated in M13MDA435-1 hybrid cells in a heterotrimeric G-protein dependent or independent manner.