Lysophosphatidic acid (LPA), a natural phospholipid with an 18 carbon fatty acid chain, isolated from crude soybean lecithin extracts, is generated during thrombin-induced platelet aggregation.1, 2 It is involved in several important biological functions, such as cell proliferation, wound healing and neurite retraction via the cell surface G-protein-coupled receptors (GPCRs) LPA1, LPA2 and LPA3 (formerly termed Edg-2, Edg-4 and Edg-7, respectively).3 Besides its physiological role, LPA also promotes cancer progression, in particular, that of epithelial ovarian cancer. Once termed “ovarian cancer activating factor,” LPA was shown to be necessary for the intraperitoneal growth of human ovarian carcinoma cells in nude mice and to be present at high concentrations in the plasma and ascites of ovarian cancer patients and thus is regarded as a potential biomarker for epithelial ovarian cancer.4–7 The fact that ovarian cancer cell express the LPA receptors LPA2 and LPA3, whereas normal ovarian surface epithelium expresses LPA1 also links LPA to the progression of ovarian cancer.8–10
Several mechanisms have been shown to be involved in the LPA-mediated promotion of invasion and migration of ovarian cancer cells. Biochemically, LPA induces urokinase secretion, promotes metalloproteinase MMP2 activation and induces the production of cytokines and growth factors, such as vascular endothelial growth factor, interleukin-8 and interleukin-6.9, 11–14 Morphologically, extensive studies have shown that LPA facilitates stress fiber formation and focal adhesion assembly in ovarian cancer cells.15–18 However, it has not known whether LPA affects cell–cell adhesions in ovarian cancer cells. One recent study suggested that it may act as a scattering factor to induce colony dispersal and cell migration by disrupting adherens junctions in various epithelial cancer cell lines.19 This new evidence has raised further interests in whether cell–cell contacts are altered by LPA, using ovarian cancer cells as a model. LPA signaling involves extensive tyrosine phosphorylation events, especially in the activation of focal adhesion proteins, such as focal adhesion kinase (FAK) or paxillin.15, 16 Src family kinases have been suggested to play a major role in controlling GPCR effects by their non-receptor tyrosine kinase activity.20 LPA is known to induce shape changes in neuronal cells, accompanied by a significant increase in the activity of the prototypic member of the Src family kinases, p60src kinase.21 Src family kinase activity is also heavily involved in regulating cell–cell adhesion22, 23 by altering tyrosine phosphorylation of the adherens junction complex components β-catenin and p120-catenin.24, 25 It is therefore reasonable to suspect that LPA might regulate cell–cell adhesion by modulating Src family kinase activity. Among the components of the adherens junction complex, p120-catenin was originally identified as a Src kinase substrate26, 27 before it was found to act as a docking protein at adherens junctions by binding to cadherins.28, 29 p120-catetnin has been shown to be not only an important regulator of the stability of adherens junctions and turnover of cadherins,30 but also a modulator of RhoGTPase activity.31, 32 However, how external stimuli modulate its function remains poorly understood. In our study, we examined the morphological changes at the intercellular adherens junction caused by LPA stimulation, and showed that LPA induces cell–cell junction dispersal in ovarian cancer cells by activating Src family kinase and facilitating their association with the adherens junction scaffold protein p120-catenin. Our data provide the first evidence that p120-catenin is involved in one of the GPCR ligand signaling pathways in ovarian cancer cells.
Material and methods
Reagents and antibodies
L-α-LPA (oleoyl sodium salt, LPA 18:1), purchased from Sigma-Aldrich (St. Louis, MO) was dissolved in phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4, 1.4 mM KH2PO4, pH 7.4; PBS) containing 2% fatty acid-free bovine serum albumin (Sigma-Aldrich). 4-Amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP2) was purchased from Biomol International L.P. (Plymouth Meeting, Philadelphia, PA). The primary antibodies used for immunoblotting and/or immunostaining were anti-phospho-Src (Tyr416, affinity-purified rabbit polyclonal, Cell Signaling Technology, Beverly, MA), anti-p120-catenin, anti-β-catenin, anti-α-catenin, anti-E-cadherin (all mouse monoclonal from BD/Transduction Laboratories, Lexington, KY), anti-cytokeratin, anti-vimentin (both mouse monoclonal from Dako, Carpinteria, CA), anti-N-cadherin (rabbit polyclonal, Takara, Japan), anti-phosphotyrosine (4G10, Upstate Biotechnology, Lake Placid, NY), anti-c-Src (mouse monoclonal, Upstate Biotechnology), anti-β-actin (mouse monoclonal, Sigma-Aldrich) and anti-Fyn (Santa Cruz Biotechnology, Santa Cruz, CA). FITC-conjugated goat anti-rabbit IgG antibodies were purchased from Chemicon (Temecula, CA), Texas red-conjugated horse anti-mouse IgG antibodies from Vector (Burlingame, CA), alkaline phosphatase-conjugated goat anti-rabbit IgG and anti-mouse IgG antibodies from Promega (Madison, WI) and horseradish peroxidase-conjugated goat anti-mouse IgG antibodies and Luminol reagent from Santa Cruz Biotechnology (Santa Cruz, CA).
Cell culture and drug treatment
Ovarian cancer cell lines SKOV3 and OVCAR3 were obtained from the American Type Culture Collection (Manassas, VA). SKOV3 cultures were maintained in Dulbecco's modified Eagle medium (DMEM, GIBCO-Invitrogen, Grand Island, NY), while OVCAR3 cells were maintained in RPMI 1640 medium (GIBCO-Invitrogen), supplemented with 10% fetal bovine serum (FBS, GIBCO-Invitrogen) and 1% penicillin–streptomycin (100 IU/ml–100 μg/ml; GIBCO-Invitrogen) in a humidified 5% CO2 incubator at 37°C. Prior to LPA treatment, cell cultures were switched to serum-free medium for 16–18 hr for starvation. In experiments using the selective Src family kinase inhibitor PP2, 10 μM PP2 was added to the culture 30 min before LPA treatment and was present during LPA treatment. The vehicle control of LPA was 2% BSA in PBS. For calcium depletion experiments, SKOV3 cells were washed twice with PBS and placed in low calcium medium (5 μM Ca2+) for 30 min.
Live cell imaging
Images of live cells were examined, using an Olympus IMT-2 microscope (Olympus, Tokyo, Japan). Pictures were taken every 5 min for a total period of 3 hr and the images were digitized, using a Nikon 995 digital camera (Nikon, Tokyo, Japan).
Ovarian cancer cells SKOV3, plated at a density of 1 × 105 cells/cm2 on uncoated glass coverslips in 35-mm culture dishes, were fixed in acetone for 10 min at −20°C, followed by brief rehydration with PBS before being incubated for 90 min at 37°C with primary antibodies against β-catenin (rabbit polyclonal, Sigma-Aldrich; 1:100 dilution) and double-labeled against p120-catenin (mouse monoclonal, BD/Transduction Laboratories; 1:50 dilution). After extensive rinsing for 3 × 5 min with PBS, they were incubated for 90 min at room temperature with a mixture of FITC-conjugated goat anti-rabbit IgG antibodies (Chemicon, Temecula, CA; 1:50 dilution) and Texas red-conjugated goat anti-mouse IgG antibodies (Vector, Burlingame, CA; 1:50 dilution). After rinsing for 3 × 5 min with PBS, they were mounted in GEL/MOUNT™ (Biomeda, Foster City, CA) and examined, using a Zeiss Axiophot microscope (Carl Zeiss, Oberkocheu, Germany) equipped with epifluorescence. The images acquired were digitized, using a Nikon D1X digital camera (Nikon, Tokyo, Japan).
Immunoprecipitation and immunoblotting
For immunoprecipitation and immunoblotting, SKOV3 cells were rinsed with ice-cold PBS, then incubated for 30 min at 0°C with 100 μl of RIPA buffer (50 mM Tris-HCl, pH 7.4, 1% NP-40, 150 mM NaCl, 1 mM EDTA) containing a mixture of protease inhibitors (1 mM PMSF and 1 μg/ml each of aprotinin, leupeptin and pepstatin) and phosphatase inhibitors (1 mM NaF and 1 mM Na3VO4). Lysates were prepared by scraping off the cells with a rubber policeman and sonication for 3 × 10 sec. The protein concentration was determined using a Bio-Rad DC protein assay kit (Bio-Rad Laboratories, Hercules, CA). For immunoprecipitation, 100 μl of cell lysates (400 μg of protein) was precleared by addition of 10 μl of protein G-Sepharose beads (Pharmacia, Uppsala, Sweden), incubation on a rocker at 4°C for 30 min and removal of the protein G-Sepharose beads by centrifugation at 13,000g for 10 min at 4°C. About 4–6 μg of immunoprecipitating antibody was added to the supernatant, followed by incubation overnight at 4°C with end-over–end rotation, then protein G-Sepharose beads were added to the lysates for an additional 1 hr, spun down by centrifugation at 13,000g for 1 min at 4°C, washed 4 times with 500 μl of lysis buffer, resuspended in Laemmli sample buffer and boiled for 5 min at 95°C. Denatured proteins were separated by 7.5% SDS-polyacrylamide gel electrophoresis, then transferred to PROTRAN® BA85 nitrocellulose membranes (PerkinElmer Life Science, Boston, MA). Strips of the membranes were blocked for 1 hr at room temperature in Tris-buffered saline (150 mM NaCl and 50 mM Tris-HCl, pH 7.4; TBS) containing 5% skimmed milk and 0.1% Tween-20, then incubated overnight at 4°C with primary antibodies. After 3 × 10-min washing with TBS the strips were incubated for 2 hr at room temperature with either alkaline phosphatase-conjugated goat anti-rabbit or goat anti-mouse IgG antibodies (Promega, Madison, WI; 1:8,000 dilution in TBS/0.1% Tween-20 containing 5% skimmed milk, pH 8.2) and bound antibody detected by substrate development, using nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate in 100 mM NaCl, 100 mM Tris-base, 5 mM MgCl2, pH 9.5.
In membrane stripping experiments, the blots were stripped using 25 mM glycine-HCl, pH 2.0, 1% (w/v) SDS, then were washed and reprobed with primary antibodies followed by horseradish peroxidase-conjugated secondary antibodies. Bound antibodies were detected using Western Blotting Luminol Reagent (Santa Cruz Biotechnology) and Hyperfilm ECL (Amershampharmacia Biotech, Buckinghamshire, England, UK) before X-ray film exposure and development.
Bands on nitrocellulose membranes and films were scanned and quantified, using Gel-Pro Analyzer 3.1 software (Media Cybernetics, MD).
Knockdown of Fyn kinase
SMARTPool small interfering RNA (siRNA) duplexes targeting human Fyn kinase and controls (non-targeting or targeting cyclophilin B) were purchased from Dharmacon (Lafayette, CO). The siRNAs were preincubated with Dharmafect transfection reagent (Dharmacon) according to the manufacturer's protocol. SKOV3 cells at 70% confluency were incubated in siRNA (100 nM)-containing antibiotic-free media supplemented with 10% FBS for 48 hr, then were lysed and processed for immunoblotting with anti-Fyn antibodies. For cell dispersal experiments, SKOV3 cells incubated with non-targeting control or Fyn-specific siRNAs were switched to serum-free media overnight at 48 hr after transfection prior to LPA stimulation.
Differences between groups were evaluated using Student's t-test. A p-value < 0.05 was considered statistically significant.
Lysophosphatidic acid induces cell dispersal on ovarian cancer cell line SKOV3
LPA had different effects on cell dispersal on the ovarian cancer cell lines SKOV3 and OVCAR3. Live cell imaging showed that SKOV3 cells (Fig. 1a, upper panels), but not OVCAR3 cells (Fig. 1a, lower panels), responded to LPA by cell dispersal. In SKOV3 cells, the intercellular junction started to dissociate after about 30 min of LPA treatment (asterisk at 30′ in Fig. 1a) and cell dispersal was seen after 75 min of LPA stimulation (asterisk at 75′ in Fig. 1a). We compared the expression of the epithelial markers (E-cadherin and cytokeratin) and mesenchymal markers (N-cadherin and vimentin) between these 2 cell lines by immunostaining. OVCAR3 cells, which formed compact colonies in in vitro culture conditions, were stained positive only for the epithelial markers E-cadherin and cytokeratin but not the mesenchymal markers N-cadherin and vimentin (Fig. 1b, upper panel). SKOV3 cells, which appeared more spindle-shaped, were stained positive for both epithelial and mesenchymal markers (Fig. 1b, lower panel, N-cad and vimentin). These results showed that LPA induces cell dispersal only on ovarian cancer cell line SKOV3, which harbors both epithelial and mesenchymal characteristics.
LPA induces the formation of “half-junctions” in SKOV3 cells
We next examined the dose responses of the effect of LPA on intercellular junctions in SKOV3 cells. Double immunofluorescence staining of β-catenin and p120-catenin showed that, under control culture conditions, SKOV3 cells formed intercellular junctions with a continuous linear staining pattern (Figs. 2a and 2b). At low LPA concentrations of 50 nM (data not shown) and 500 nM (Figs. 2c and 2d), no significant changes were detected in SKOV3 cells. However, at higher concentrations of 1 μM (Figs. 2e and 2f), 2 μM (Figs. 2g and 2h) and 2–20 μM (data not shown), LPA induced the formation of “half-junctions” in SKOV3 cells. The changes in junction staining patterns were similar at all concentrations higher than 1 μM, so that the concentration of 2 μM LPA was chosen for use throughout our study. We then compared the staining of β-catenin and N-cadherin in SKOV3 cells. After incubation with LPA for 30 min, SKOV3 cells dissociated from each other, which resulted in the transition of intact intercellular junctions (Figs. 3a and 3b) into “half-junctions,” seen as a discontinuous staining pattern (Figs. 3c and 3d). To confirm the observed changes in staining pattern was a specific phenomenon, we performed calcium depletion, which was known to cause junction breakdown. As shown in Figure 3, SKOV3 cells were also dispersed by calcium depletion, resulting in the same junction staining pattern as that induced by LPA (Figs. 3e and 3f). Thus, the morphological change of intact adherens junctions to “half-junctions” can be regarded as the intermediate phenotype during the process of junction dissolution.
LPA induces sequential events of intercellular junction dispersal in SKOV3 cells
In a time-frame study, cell dispersal and the transition of intact intercellular junctions (Figs. 4a–4c, arrows) to half-junctions in SKOV3 cells began after 15 min of LPA stimulation, concomitant with the disruption of the continuous linear pattern (Figs. 4d–4f, arrows). At this time, some of the β-catenin and some of the p120-catenin remains colocalized at the cell junctions (Figs. 4d–4f, arrows), but there was also some cytoplasmic accumulation of β-catenin (Fig. 4d, large arrowheads), which initially was colocalized with p120-catenin (Figs. 4e and 4f, large arrowheads). After prolonged stimulation (30 and 60 min), free β-catenin unbound to p120-catenin (Figs. 4g–4l, small arrowheads) started to accumulate in the cytoplasm. These results show that, initially, intact complexes separated as “half-junctions,” which remains located at the cell–cell contact areas, then disassembly occurred into individual junction proteins.
LPA activates Src family kinases and promotes their association with the junctional protein p120-catenin
LPA had been shown to induce cell rounding by transiently activating Src family kinases in neuronal cells.21 We suspected that LPA-induced intercellular junction dispersal might be mediated by Src kinase activation. SKOV3 cells were therefore incubated with LPA and Src family kinase activation was studied by immunoblotting with antibodies specific for tyrosine 416-phosphorylated Src family kinase (pSrc), which cross-reacts with other Src-family members. In a time-course study using Western blots, incubation of SKOV3 cells with LPA induced a transient activation increase in pSrc levels by 2 min, followed by signal attenuation by 15 min (Figs. 5a and 5b). Double-immunofluorescence staining was then used to clarify the involvement of Src in LPA-induced junction dispersal. After LPA treatment, an increase of pSrc immunoreactivity was detected at the cell–cell contact areas, which was colocalized with p120-catenin (Figs. 6e–6g, arrows). To determine whether the LPA-induced activation of Src promoted its association with the p120-catenin, the cells were incubated with LPA for 0, 2, 15 or 30 min and then p120-catenin was immunoprecipitated followed by immunoblotting, using anti-pSrc antibodies. As shown in Figure 6a, the association of pSrc with p120-catenin was significantly increased after 15 and 30 min of LPA treatment.
A selective tyrosine kinase inhibitor abolishes LPA-induced junction dispersal and the “half-junction” phenotype in SKOV3 cells
To verify the involvement of activated Src family kinases in LPA-induced junction dispersal, SKOV3 cells were pre-incubated with or without the selective tyrosine kinase inhibitor PP2, then incubated with LPA for 30 min in the continued presence or absence of PP2. After LPA stimulation, non-PP2-treated SKOV3 cells dispersed and exhibited a discontinuous staining pattern at the intercellular junctions (Figs. 7b and 7f), while addition of PP2 alone to non-LPA-treated cells did not alter the intact junctional staining pattern (Figs. 7c and 7g) compared to that seen in control cells (Figs. 7a and 7e). However, PP2 treatment of LPA-treated cells prevented cell dispersal, abolished the “half-junctions” phenotype and maintained the intercellular junction morphology (Figs. 7d and 7h).
Src family kinase Fyn associated with p120-catenin is activated after LPA stimulation
To further investigate, which Src family kinase might mediate the early signaling events at the adherens junction, we used the anti-pSrc antibody for immunoprecipitations followed by immunoblotting with specific antibodies against c-Src or Fyn. As shown in Figure 8a, c-Src was barely detectable in the immunoprecipitates either in control or LPA-treated SKOV3 cells. However, after LPA stimulation, a significant increase in Fyn was seen in the immunoprecipitates (Fig. 8b, lower panel), the association of which to p120-catenin was also increased (Fig. 8b, middle panel). In addition, the p120-catenin associated with the pSrc pool showed higher levels of tyrosine phosphorylation (Fig. 8b, upper panel). Reciprocal immunoprecipitation of Fyn confirmed that LPA did not alter the association between Fyn and p120-catenin but induced tyrosine phosphorylation of Fyn kinases (Fig. 8c).
To examine the role of Fyn in the LPA-induced cell dispersal of ovarian cancer SKOV3 cells, we used siRNA targeting of human Fyn kinase followed by live cell observation. The specific silencing effect of Fyn kinase siRNA compared to non-targeting siRNA in SKOV3 cells was verified at 48-hr post-transfection by immunoblotting (Fig. 9a, upper panel). In the non-targeting controls, LPA still induced SKOV3 cell and intercellular junction dispersal (Fig. 9b, upper panels, asterisk and arrows), but knockdown of Fyn kinase blocked both effects (Fig. 9b, middle panels). To confirm the LPA-induced cell dispersal was mediated by Fyn kinase, we next silenced the prototypic Src family kinase c-Src in SKOV3 cells. Interestingly, there was a delay on SKOV3 cells responding to LPA on cell dispersal (Fig. 9b, lower panels, asterisk and arrows). The intercellular junction in the control group started to dissociate after 30 min of LPA treatment. However, the dissociation did not occur until 90 min of LPA treatment in the Src knockdown group.
Effect of LPA on the association of β-catenin with active Src family kinases and α-catenin
Increased tyrosine phosphorylation of β-catenin has been shown to lead to a decrease in cell–cell adhesion via dissociation of α-catenin and actin cytoskeleton from the cadherin–β-catenin complex.33 To clarify the possible downstream events regulating junction proteins during LPA-induced cell dispersal, we performed immunoprecipitation experiments on SKOV3 cells. As shown in Figure 10a, β-catenin co-immunoprecipitation with pSrc increased after LPA treatment, resulting in increased levels of tyrosine phosphorylation of β-catenin. In parallel to the increased tyrosine phosphorylation of β-catenin, the co-immunoprecipitation between β-catenin with α-catenin decreased after LPA treatment (Figs. 10b and 10c). These results suggested that LPA altered the association between α-catenin and β-catenin possibly by tyrosine phosphorylation of β-catenin by Src family kinase, which further explains how LPA induces cell and junction dispersal in SKOV3 cells.
LPA has been shown to cause a colony dispersal of cancer cell lines A431 and HT-29.19 However, it remains unknown how LPA regulates the intercellular junction in promoting ovarian cancer cell motility. In the present study, we demonstrated, for the first time, that LPA induced cell–cell junction dispersal and caused the transition of intact intercellular junctions to a “half-junction” phenotype in ovarian cancer SKOV3 cells. The intermediate morphology of the cell–cell junction induced by LPA was compatible with that observed in our Ca2+ depletion experiments and with that documented in the literature.34 This suggests that occurrence of “half-junctions” might represent the transition phase during cell–cell dissociation before the disassembly of the entire junction complex. By time-lapse tracing experiments, Jourquin et al. (2006) demonstrated that, after LPA treatment for 30 min–1 hr, A431 cells begin to loss cell–cell contact, which is preceded by sequential morphological events involving membrane ruffling and lamellipodia formation.19 These temporal sequences were in agreement with the results of our time-frame study, which showed that “half-junctions” were seen after 15 min of LPA treatment in SKOV3 cells. Thus, “half-junctions” could serve as the intermediate phenotypic indicator of junction breakdown.
In our study, we demonstrated that LPA induced rapid and transient activation of Src kinase in ovarian cancer SKOV3 cells, which agrees with previous observations in other cell systems, such as neuroblastoma N1E-115, COS-7 and PC-12 cells.21, 35, 36 The involvement of Src kinase in LPA signaling in fibroblasts has been shown mainly by indirect evidence including inhibition of Src kinase activity by Src kinase inhibitors or Src mutants.37 Several mechanisms have been suggested to mediate LPA-induced Src kinase activation and are context-specific.20 Indirect cross-talk between GPCRs and receptor tyrosine kinases and focal adhesion complexes has been shown to play a role in Src kinase activation. In COS-7 cells, the selective Src kinase inhibitor PP1 inhibits LPA-induced EGFR-dependent tyrosine phosphorylation of Shc and Gab1 and Gab1–Grb2 complex formation,38 suggesting that Src kinase activity downstream of LPA-induced EGFR transactivation is crucial. In PC-12 cells, LPA was found to induce tyrosine phosphorylation of the FAK family kinase, Pyk2, leading to Src kinase activation, suggesting the contribution of focal adhesion complexes to Src kinase activation.36 In ovarian cancer cells, there is no direct evidence for the mechanism involved in LPA-induced Src kinase activation. However, it has been shown that Src kinase activity is involved in endothelin-1-induced EGFR transactivation in ovarian cancer cells.39 Preliminary data from our laboratory also suggest cross-talk between LPA and EGFR signaling in Src kinase activation (unpublished data). LPA-induced EGFR transactivation and Src kinase activation might therefore play important roles in signal convergence in ovarian cancer.
It has been proposed that LPA-induced junction dissociation is a passive consequence of cells pulling apart from each other.19 However, in our study, we provide evidence that LPA induces active signaling events at cell–cell junctions, at least in the model of SKOV3 cells. We demonstrated that, after rapid activation by LPA, Src family kinases show increased association with p120-catenin at cell–cell junctions. We also showed that blocking of Src kinase activity with a selective Src kinase inhibitor, PP2, prevented the transition of intact cell–cell junctions to “half-junctions.” These results indicate that signals mediated by Src family kinases and p120-catenin are involved in LPA-induced ovarian cancer cell dispersal. It remains unclear how p120-catenin acts as a relay station for signals coming from external stimuli, such as growth factors or inflammatory cytokines, to regulate cell–cell adhesion. One possibility could be a change in the phosphorylation status (especially tyrosine residues) of p120-catenin. However, we only detected marginal changes in p120-catenin tyrosine phosphorylation after LPA treatment of SKOV3 cells (unpublished data). A recent study also suggested that phosphorylation of p120-catenin at Y228 might not directly regulate junction stability.40 Thus, the role of p120-catenin phosphorylation in cell dispersal remains inconclusive. On the other hand, p120-catenin has been shown to act as a scaffold protein allowing tyrosine kinases Fyn and Fer to associate with the adherens junction complex.23, 24, 41 Fyn and Fer further modulate the β-catenin–α-catenin interaction by phosphorylation of β-catenin at tyrosine residue 142.24 We demonstrated that LPA increased the association of tyrosine-phosphorylated β-catenin with activated Src family kinase and altered the β-catenin–α-catenin interaction in SKOV3 cells. Moreover, we showed that LPA activated the Src family kinase member Fyn but not c-Src to associate with p120-catenin and silencing Fyn kinase abolished the LPA-induced cell dispersal. However, we did observe a delayed effect of c-Src knockdown on LPA-induced cell dispersal, suggesting that c-Src might participate in the early stages of cell dispersal, such as the alteration of focal adhesions and other Src family kinase (ex. Fyn) might be able to compensate. Our data further support the hypothesis that LPA induces cell junction dispersal by activating Fyn kinase docking to p120-catenin, allowing signaling via the β-catenin–α-catenin complex.
Only cancer cell lines with the characteristics similar to the epidermal carcinoma cell line A431, such as the colon cancer cell lines DLD-1, HT-29 and HCA-7, are known to respond to LPA by cell dispersal.19 Our findings add the ovarian cancer cell line SKOV3 to this list. In contrast, LPA did not induce cell dispersal or “half-junction” formation in the ovarian cancer cell line OVCAR3. From our results and others, OVCAR3 cells harbor the characteristics of epithelial cells, which are positive only for E-cadherin42 and cytokeratin, while SKOV3 cells have the mixed epithelial and mesenchymal phenotypes. We speculate that the differential cell–cell junction dissociation response to LPA in SKOV3 cells might be related to the expression of mesenchymal N-cadherin. Recent evidence has shown that p120-catenin activates Rac1 and promotes cell motility by cooperating with the mesenchymal cadherins, cadherin-11 and N-cadherin.43 E-cadherin might compete with these for binding to p120-catenin since E-cadherin-mediated cell adhesion has been shown to suppress growth factor signaling.43, 44 Forced expression of a dominant-negative E-cadherin in OVCAR3 cells has been shown to disrupt adherens junctions and to increase cell migration in wounding assays when the leading front of the cells dispersed as single mesenchymal-like cells.45 This might explain the lack of a cell-dispersal response to LPA in more adhesive cell lines, such as OVCAR3. Further modification of the expression levels of E-cadherin and N-cadherin in these cell lines would be crucial to clarify this issue.
In summary, we show, for the first time, that LPA induces cell–cell junction dispersal in the ovarian cancer cell line SKOV3 by activating one of the Src family kinase, Fyn and increasing the Fyn kinase-p120-catenin association that results in altered tyrosine phosphorylation of β-catenin and its interaction with α-catenin. p120-catenin serves as an important scaffold that collects signals from the cell microenvironment to modulate cell–cell adhesion. Further information on p120-catenin phosphorylation will be crucial in understanding the detailed mechanism, by which LPA regulates cadherin-based cell–cell adhesions.
Dr. Ruby Yun-Ju Huang was supported by a FIGO/ESRF Postdoctoral Research Fellowship. We thank Dr. Jean-Paul Thiery for his critical review and suggestion of this manuscript.