Cell migration plays an essential role in various physiological and pathological phenomena such as embryogenesis, wound repair, inflammation and invasion and metastasis of cancer cells.1, 2 Cell migration is primarily mediated by integrin binding to extracellular matrices (ECM) and regulated by organization of the actin cytoskeleton and the formation of focal adhesion complexes, which are induced through activation of the Rho-family small GTP-binding proteins.3–5 Integrins not only act as a scaffold but also generate important signals leading to cell migration. Integrin-mediated signaling events that regulate cytoskeletal organization require Rho-family GTPases and conversely integrin-mediated adhesion itself regulates Rho-family GTPases.6 Cell motility is closely related to morphology, growth,7 differentiation and survival, in which different signals from other cell surface receptors are also implicated.8 It remains to be elucidated, however, how the integrin and growth-factor signaling pathways are integrated into cell migration.
Paxillin (p68-kD) is one of the integrin assembly proteins and can interact directly with several integrin assembly proteins, including vinculin, talin, β1 integrin, focal adhesion kinase (FAK), c-Src and Csk.9, 10 Three human isoforms of paxillin (α, β and γ) and 2 homologous murine isoforms (α and β) have been identified,11, 12 and the α form (original paxillin) appears to play a more dominant role.13 Integrin-mediated tyrosine phosphorylation also enables paxillin to interact with various signaling molecules: tyrosine residues (Y) 31 and 118 being especially predominant targets of phosphorylation by kinases and creating binding sites for the SH2 domain of adaptor protein Crk.14–16 Thus, paxillin plays a pivotal role in cell adhesion, migration and further oncogenic transformation.9, 10 Because several cytokines and growth factors also induce tyrosine phosphorylation of paxillin,9, 10 it is suggested that the signals from both cytokine- and growth-factor receptors and those from integrins converge on paxillin.
In efforts to evaluate the invasive capacity of cancer cells and to understand the mechanisms of cancer-host interactions including transcellular (transmonolayer) migration of cancer cells, we have developed an in vitro system: rat ascites hepatoma MM1 cells are allowed to invade a mesothelial cell monolayer (MCL) and the numbers of penetrating tumor cells and tumor colonies are determined as the invasive capacity of MM1 cells.17 Using this system, we have shown that MM1 cells require fetal calf serum or lysophosphatidic acid (LPA) for transcellular migration and that the RhoA-ROCK (Rho-associated kinase p160) pathway plays an essential part in LPA-induced transcellular migration of MM1 cells.18–22 Recently, the Rho-ROCK pathway was also shown by other investigators, to be involved in intrahepatic metastasis of human hepatocellular carcinomas by means of orthotopic implantation into SCID mice.23
Our previous studies on cell signaling generated by LPA showed that LPA evokes tyrosine phosphorylation of FAK and paxillin, concomitant with actin-polymerization in MM1 cells. These events have been demonstrated to occur downstream of RhoA and to be closely involved in transcellular migration.20, 21, 24 Moreover, because anti-fibronectin (FN) antibodies (Abs), anti-β1 integrin Abs and cyclo-GRGDSPA remarkably suppressed LPA-induced migration, the interaction between FN and β1 integrin is also necessary for LPA-induced migration of MM1 cells.25 These findings show that both LPA-induced activation of the RhoA-ROCK pathway and integrin signaling via the FN-β1 integrin interaction are essential and that paxillin may play an important role in integrating both signals into the migration of MM1 cells.
In the present study, we examined the role of and characterized the tyrosine phosphorylation of paxillin, especially at Y31 and Y118, in LPA-FN-cooperated migration of MM1 cancer cells.
MATERIAL AND METHODS
Drugs and chemicals
Bovine serum albumin (BSA) fraction V, FN from bovine plasma and 1-oleoyl-sn-lysophosphatidic acid (LPA) were purchased from Sigma (St. Louis, MO). LPA was dissolved in phosphate-buffered saline (PBS) supplemented with 0.1% BSA. 3-Amino-N-(aminoiminomethyl)-5-(dimethylamino)-6-chloropyrazine-2-carboxamide hydrochloride (amiloride) was from Research Biochemicals International (Natick, MA) and dissolved in PBS. This is an amiloride derivative that exhibits enhanced potency and selectivity as an inhibitor of the Na+/H+ antiporter (IC50 = 6.9 μM).26Clostridium botulinum C3 exoenzyme was a kind gift from Dr. B. Syuto (Department of Veterinary Medicine, Iwate University). Monoclonal anti-phosphotyrosine Abs (clone 4G10) were purchased from Upstate Biotechnology (Lake Placid, NY). Horseradish peroxidase (HRP)-conjugated secondary Abs were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal anti-paxillin Abs were from Transduction Laboratories Inc. (Lexington, KY). Rabbit anti-paxillin Abs (Ab199–217) and the paxillin β isoform-specific rabbit Abs were described previously.11 Phosphorylation site-specific Abs that recognize paxillin when phosphorylated at tyrosine 31 (anti-pY31 paxillin), tyrosine 118 (anti-pY118 paxillin), or tyrosine 181 (anti-pY181 paxillin) were obtained from BioSource International (Camarillo, CA). These Abs are affinity-purified rabbit polyclonal Abs highly selective for the targeted phosphorylation site (anti-pY181 paxillin: Yano et al., unpublished results).27
Cells and culture conditions
Cells of MM1, a highly invasive clone of rat ascites hepatoma AH130, were grown in a dispersed and floating manner and maintained by serial passages in modified Eagle's minimum essential medium (MEM; Nissui, Tokyo) containing 2-fold concentrations of amino acids and vitamins (modified MEM), supplemented with 10% (v/v) fetal bovine serum (FBS; JRH Biosciences, Lenexa, KS).17 To examine the effect of RhoA inactivation, MM1 cells were pretreated with 10 μg/ml of C3 exoenzyme for 24 hr, under which conditions no significant effects were observed on cell proliferation. Cell numbers and cell viability were determined with a hemocytometer and the Trypan-blue-exclusion test, respectively.
Morphological observation of MM1 cells
MM1 cells (3 × 105 cells) were incubated in 2 ml of FCS-free modified MEM with or without amiloride and 10-min later, 25 μM LPA was added or not added to the culture. Immediately the culture was seeded onto a 35-mm dish coated with 10 μg of FN. After 1-hr incubation, at least 200 MM1 cells were observed under a phase-contrast microscope (Olympus IX70, Tokyo), the proportion of cells exhibiting pseudopodia formation being estimated.25
Phagokinetic motility assay
Cell motility was determined by means of a slight modification of the phagokinetic motility assay previously described.7 Briefly, uniform carpets of colloidal gold particles were prepared on glass slides. Because MM1 cells grow in suspension, poly-L-lysine (PLL)-coated glass slides (Matsunami Glass, Osaka) were used for the adherence of MM1 cells to the slides. The glass slides were further coated with 10 μg/ml of FN, at which concentration the motility-inducing activity as to MM1 cells was highest in the presence of LPA.25 The slides were then placed in 35-mm tissue culture dishes containing 2 ml of serum-free modified MEM and then 1.2 × 105 MM1 cells that had been washed once with modified MEM were added to each dish, with or without 25 μM LPA, followed by incubation at 37°C for 7 hr in a CO2 incubator. The area of gold particles scraped by MM1 cells was measured with an image analyzer (Olympus XL-20, Tokyo). Cell motility was expressed as the total area scraped by 100 cells in 7 hr.25
Transcellular migration (in vitro invasion) assay
The assay procedure used for the in vitro invasive capacity of tumor cells was essentially the same as previously described.17, 24 MM1 cells (2 × 105 cells) pretreated with or without an inhibitor (10 μg/ml of C3 exoenzyme for 24 hr, or 50 μM or 200 μM amiloride from 10-min beforehand) were seeded on a confluent MCL and then cultured in serum-free modified MEM in the presence or absence of 25 μM LPA. Twenty hours later, the medium was removed and the resultant monolayer was fixed in situ with 10% formalin in PBS. The numbers of penetrating single tumor cells and tumor cell colonies (collectively called invasion foci) were determined under a phase-contrast microscope. The invasive capacity was expressed as the number of invasion foci/cm.2.
Western blotting and immunoprecipitation
To examine the protein level and tyrosine phosphorylation of paxillin, MM1 cells were washed once with serum-free modified MEM and then incubated in serum-free medium at 37°C for 1 hr in a CO2 incubator. The MM1 cells were then seeded on BSA (10 μg/ml)- or FN (10 μg/ml)-precoated culture dishes in the absence of serum and immediately after seeding, they were stimulated with 25 μM LPA if necessary. After incubation for the indicated period, the cells were collected by brief centrifugation and lysed with lysis buffer [1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 10 mM Tris-HCl (pH 7.2), 1 mM EGTA, 1 mM PMSF, 1 mM sodium orthovanadate, 50 mM sodium fluoride, 10 μg/ml leupeptin and 10 μg/ml aprotinin]. For immunoblotting, the cellular lysate was boiled in reducing SDS sample buffer for 5 min. Fifteen or 20 μg of proteins from each extract were separated by 7.5% SDS-polyacrylamide gel electrophoresis (PAGE) and then transferred to a PVDF membrane. The membrane was blocked with a 1% BSA solution in T-TBS (Tris-HCl-buffered saline supplemented with 0.05% Tween 20) and then incubated with a primary Ab in T-TBS at 4°C overnight. The concentrations of the primary Abs used were as follows: anti-phosphotyrosine, 1 μg/ml; anti-paxillin (Ab199–217), 1:1,000 dilution; anti-paxillin β, 1:1,000 dilution; anti-pY31 paxillin, 1 μg/ml; anti-pY118 paxillin, 0.2 μg/ml; and anti-pY181 paxillin, 0.2 μg/ml. After washing with T-TBS, the membrane was allowed to react with an HRP-conjugated secondary Ab. Signals were visualized using a chemiluminescence detection kit (ECL; Amersham Pharmacia Biotech, Sweden) after extensive washing.
For the immunoprecipitation of paxillin, 100 μg of each cell lysate and monoclonal anti-paxillin Ab were used, coupled with protein G-Sepharose. The immunoprecipitated proteins were solubilized by boiling in reducing SDS sample buffer and separated by 8% SDS-PAGE.
The proportion of cells exhibiting pseudopodia formation was determined by means of the chi-square test. The data on phagokinetic motility and invasive capacity were analyzed for significance by means of the 2-tailed Student's or Welch's t-test. The criterion for significance was p < 0.05.
Induction of cell migration and persistent protein tyrosine-phosphorylation after stimulation with FN + LPA
LPA induces the transcellular migration of MM1 cells through an MCL in FBS-free medium, whereas anti-FN Abs inhibit it. LPA alone can induce pseudopodia formation and phagokinetic motility of MM1 cells on an FN-coated dish, but not on a non-coated dish. These findings show that MM1 cells require the combination of LPA and FN for pseudopodia formation, phagokinetic motility and transcellular migration.20, 25 As shown in Table I, 25 μM LPA induced pseudopodia formation and phagokinetic motility on an FN-coated dish and slide glass, respectively and the transcellular migration of MM1 cells through an MCL. Moreover, it was shown that the phenotype of MM1 cells exhibiting pseudopodia formation was almost equivalent to that of ones capable of phagokinetic motility and transcellular migration.
Table I. Inhibitory Effects of C3 and Amiloride on Pseudopodia Formation, Phagokinetic Motility and Transcellular Migration of MM1 Cells
Transcellular migration (number of invasion foci/cm2)37
Two hundred cells were observed on an FN-coated dish for each measurement. Representative results of two experiments.
p < 0.0001, when compared with the value for cells treated with 25 μM LPA alone, with the chi-square test.
>Mean ± SD of at least three determinations.
The phagokinetic motility assay was performed on an FN-coated slide glass as described in the text.
p < 0.0001, when compared with the value for cells treated with 25 μM LPA alone, with Welch's t-test.
MM1 cells pretreated with or without an inhibitor were seeded on an MCL, and 20 hr later the numbers of penetrating tumor cells and colonies were determined. Significantly;t1> lower than the value for cells treated with 25 μM LPA alone.
To examine tyrosine-phosphorylated proteins in association with the motile phenotype of MM1 cells upon stimulation with FN + LPA, total lysates of MM1 cells stimulated with FN or LPA were subjected to immunoblot analysis with the anti-phosphotyrosine Ab (Fig. 1). Judging from the results of time-course immunoblot analysis (0–90 min after stimulation), the tyrosine-phosphorylation of proteins was most obviously 15–30 min after stimulation (data not shown). After any form of stimulus tested, anti-phosphotyrosine immunoblot analyses revealed the significantly enhanced tyrosine-phosphorylation of 4 prominent proteins with approximately 40-kD, 65–70-kD, 110–120-kD and approximately 140-kD molecular masses. The major components in the 65–70-kD and 110–120-kD protein bands were paxillin and FAK, respectively and maximum tyrosine-phosphorylation of them was observed 10–15 min after the addition of LPA, as previously reported.20 The tyrosine-phosphorylation of paxillin and FAK persisted more intensely at 30-min after stimulation with FN + LPA as compared to that with BSA plus LPA (LPA alone; BSA induces no significant protein tyrosine-phosphorylation in MM1 cells), or FN alone.
The tyrosine-phosphorylation of the 40-kD protein decreased on treatment with PD98059, an inhibitor of mitogen-activated protein kinase (MAPK) kinase (MEK) and immunoblotting with an Ab specific to active MAPK revealed that MAPK was phosphorylated in response to LPA in MM1 cells (Mukai, unpublished results). As shown in Figure 1, the extent of tyrosine-phosphorylation of the 40-kD protein after stimulation with FN alone was lower than that with LPA alone or FN + LPA. Thus, the 40-kD protein appeared to be MAPK. When treated with PD98059, MM1 cells exhibited no significant reduction in migratory activity (Mukai, unpublished results). Contrary to previous reports such as by Klemke et al.,28 the MAPK pathway probably contributes little to the transcellular migration of MM1 cells. The major component(s) in the 140-kD band appeared to be p130Cas (data not shown) and we very recently reported the involvement of its tyrosine-phosphorylation in MM1-cell migration.29 FAK was also recently shown to promote cell migration,30 and therefore we here focused on paxillin.
Overall, there will be a good correlation between the persistent tyrosine-phosphorylation of paxillin and cell migration, both of which were induced by FN + LPA in MM1 cells.
Characterization of tyrosine-phosphorylated paxillin after stimulation with FN + LPA
To confirm and characterize the tyrosine-phosphorylation of paxillin, total cellular lysates of MM1 cells 30-min after stimulation with FN alone, FN + LPA or neither were immunoprecipitated with monoclonal anti-paxillin Ab and then subjected to immunoblot analysis with an anti-phosphotyrosine Ab (Fig. 2, top). Paxillin was faintly phosphorylated after stimulation with FN alone. On the contrary, tyrosine-phosphorylation of paxillin was remarkably detected after stimulation with FN + LPA. Moreover, immunoblots with the anti-phosphotyrosine Ab revealed 2 components of tyrosine-phosphorylated paxillin; a slowly migrating component and a fast migrating one. Stimulation with FN + LPA enhanced tyrosine-phosphorylation of paxillin preferentially on a slowly migrating component. These findings lead us to consider that the persistence of a slowly migrating component of tyrosine-phosphorylated paxillin is involved in cell migration after stimulation of MM1 cells with FN + LPA.
We previously found that the monoclonal anti-paxillin Ab recognizes both the α and β isoforms of paxillin and that MM1 cells contain both the α and, but to a far lesser extent, β isoforms of paxillin. To examine the contribution of phosphorylation of the β isoform, the immunoprecipitates with the anti-phosphotyrosine Ab were immunoblotted with the anti-β isoform of paxillin-specific Ab. Tyrosine-phosphorylation of the β isoform was not detected (data not shown).
Next, to further characterize the phosphorylation status of tyrosine-phosphorylated paxillin, we performed immunoblotting of anti-paxillin immunoprecipitates with phosphorylation site-specific Abs against pY31, pY118 and pY181 (Fig. 2).14, 15 Stimulation with FN + LPA induced tyrosine-phosphorylation of paxillin more intensely at both Y31 and Y118 than that with FN alone and these phosphorylation was predominantly detected on a slowly migrating component. The slowly and fast migrating components of paxillin were shown to contain components tyrosine-phosphorylated at both Y31 and Y118, either simultaneously on 1 molecule or separately. To the contrary, phosphorylation of paxillin at Y181 was constitutive and was not augmented by stimulation with FN or LPA. Under non-stimulated conditions, immunoblots with anti-pY181 Ab showed 1 intense band, whereas no signals were detected by anti-phosphotyrosine Ab (Fig. 2, None lane). This may be because phosphorylation site-specific pY181 Ab is more sensitive to tyrosine-phosphorylated paxillin than generic anti-phosphotyrosine Ab.
As shown in Figures 1 and 2, treatment of MM1 cells with FN alone gave a faint slowly migrating component of tyrosine-phosphorylated paxillin. It is likely that MM1 cells may have residual RhoA activity that cannot be inhibited after serum-starvation, as previously suggested.31, 32
Inhibitory effects of amiloride on pseudopodia formation, transcellular migration and tyrosine-phosphorylation of paxillin
We demonstrated previously that the cell motility (pseudopodia formation, phagokinetic motility and transcellular migration) of MM1 cells and tyrosine-phosphorylation of FAK and paxillin are correspondingly inhibited on pretreatment of MM1 cells with C3 exoenzyme, a specific inhibitor of Rho, or inhibitors of thyrosine kinases such as genistein and herbimycin A.20, 25 We here confirmed that pretreatment with C3 exoenzyme significantly suppressed pseudopodia formation and phagokinetic motility after stimulation with FN + LPA (p < 0.0001) and LPA-induced transcellular migration of MM1 cells (p < 0.001; Table I). To further evaluate the location of tyrosine-phosphorylation of paxillin in the LPA-RhoA-ROCK pathway, we examined the effects of amiloride on cell motility and tyrosine-phosphorylation of paxillin (Table I, Fig. 3). This is because amiloride is a specific inhibitor of the Na+/H+ antiporter acting downstream of the LPA-RhoA-ROCK pathway.33
Treatment with amiloride significantly and dose-dependently inhibited pseudopodia formation after stimulation with FN + LPA and LPA-induced transcellular migration of MM1 cells (Table I). Likewise, amiloride dose-dependently suppressed tyrosine-phosphorylation of paxillin after stimulation with FN + LPA: the total relative densitometric intensity of both bands was 1, 0.58, 0.53, 0.31 and 0.28 with the concentrations of 0 μM, 10 μM, 50 μM, 100 μM and 200 μM, respectively (left halves of Fig. 3). On the contrary, tyrosine-phosphorylation of paxillin after stimulation with FN alone was not significantly inhibited (right halves of Fig. 3). Amiloride treatment of MM1 cells for up to 1 hr had no effect on paxillin expression (data not shown). Therefore, tyrosine phosphorylation of paxillin at both Y31 and Y118 induced by FN + LPA, is essential for MM1-cell migration and is regulated by the Na+/H+ antiporter downstream of ROCK. Tyrosine-phosphorylation of paxillin at Y31 and Y118 weakly induced by FN alone is not located downstream of the Na+/H+ antiporter and is insufficient for cell migration.
Taken together, these results demonstrate that LPA collaborates with FN in persistent tyrosine-phosphorylation of paxillin, especially phosphorylation at both Y31 and Y118 and that this phosphorylated paxillin, regulated by the Na+/H+ antiporter downstream of the LPA-RhoA-ROCK pathway, is essential for MM1 cancer cell migration.
Paxillin comprises an N-terminus containing 5 LD motifs and a C-terminus made up of 4 LIM domains. Paxillin binds to a complex of proteins including PAK (p21 GTPase-activated kinase), Nck and PIX (guanine nucleotide exchange factor). The association of this complex with paxillin is mediated by p95PKL (paxillin-kinase linker), which binds directly to paxillin LD4 and PIX. Paxillin recruits an active PAK/PIX complex to nascent focal adhesion structures potentially via interactions with p95PKL.34 Besides, Kondo et al.35 reported that paxillin transport to sites of integrin macroaggregates at the plasma membrane is regulated by PAG3 (paxillin-associated protein with ARF GTPase-activating protein activity, number 3). Thus, paxillin acts as a mediator of Rho-family GTPase-regulated actin cytoskeletal reorganization through the recruitment of associated proteins together with itself to focal adhesions leading to cell migratory activity. The detailed mechanisms remain to be fully elucidated.
To determine how paxillin tyrosine-phosphorylation is involved in RhoA-mediated cell migration, we utilized amiloride, an inhibitor of the Na+/H+ antiporter. The Na+/H+ antiporter is activated through integrin ligation of ECM and acts cooperatively with myosin-based contractility downstream of the LPA-RhoA-ROCK pathway to mediate ROCK-induced actin stress fiber assembly.33, 36, 37 Inhibition of the Na+/H+ antiporter hampers the recruitment of focal adhesion proteins including paxillin and vinculin and tyrosine-phosphorylation of FAK in fibroblasts.32 We here demonstrate that amiloride inhibits LPA-FN-induced tyrosine-phosphorylation of paxillin at both Y31 and Y118 and LPA-induced transcellular migration of MM1 cells. These findings suggest that tyrosine-phosphorylated paxillin, especially that phosphorylated at both Y31 and Y118, is necessary for MM1-cell migration and that such tyrosine phosphorylation is regulated by the Na+/H+ antiporter downstream of ROCK.
Petit et al.38 demonstrated that overexpression of paxillin mutants in which Y31 or Y118 were replaced by phenylalanine effectively impaired collagen-induced cell migration of NBT-II cells by preventing the formation of the paxillin-Crk complex. Double (both Y31 and Y118)-replaced (31/118F) mutants reduced cell migration most effectively. We also reported that the 31/118F mutant altered the peripheral localization of paxillin and paxillin-containing focal adhesion formation during cell migration of NMuMG cells.27 Moreover, we reported that overexpression of paxillin (α) reduced the in vitro transcellular migration of MM1 cells and that such an effect, however, was not seen with overexpression of the phosphorylation-null mutation of paxillin.29 These findings suggest that tyrosine-phosphorylation of paxillin at Y31 and Y118 is essential for cell migration, probably through the interaction with the SH2 domain of Crk.
In the present study, we further demonstrated that tyrosine-phosphorylation of paxillin at both Y31 and Y118 is induced by the collaboration of LPA and FN, but that phosphorylation of paxillin at Y181 is not augmented by FN or LPA. Paxillin may also be phosphorylated at Y40 in MM1 cells; however, Abs that selectively detect phosphorylation at Y40 were not yet available. FN was reported to phosphorylate paxillin at Y118 in chicken embryonic fibroblasts,15 but paxillin was weakly phosphorylated at both Y31 and Y118 by FN alone in MM1 cells. This may be partially because MM1 cells were grown in a dispersed and floating manner differently from adhesive fibroblasts. Further study, however, is required to determine how paxillin is tyrosine-phosphorylated at both Y31 and Y118 during the migration of MM1 cells. Identification of the kinases that phosphorylate paxillin, especially at Y31 and Y118, during cell migration will contribute to the understanding of the regulation of cytoskeletal organization and cell motility.
Paxillin is phosphorylated at both tyrosine and serine residues.39 It was very recently reported that constitutive paxillin phosphorylation at serine residues promoted NIH3T3 cell migration and melanoma cell metastasis.40 We should study serine/threonine-phosphorylation as well as tyrosine-phosphorylation of paxillin in association with MM1 cancer cell migration. These studies would be helpful in further characterizing the phosphorylation status of 2 components of tyrosine-phosphorylated paxillin. Moreover, FAK was also persistently tyrosine-phosphorylated after stimulation of MM1 cells with FN + LPA. FAK acts as a receptor-proximal bridging protein that links the growth factor-receptor at the amino terminus and the integrin signaling pathway through C-terminal domain-mediated interactions with paxillin and talin.30 The interaction between paxillin and FAK in MM1 cancer cell migration also requires further investigation.
These findings demonstrate that LPA collaborates with FN in persistent tyrosine-phosphorylation of paxillin, especially at both Y31 and Y118 and that such phosphorylated paxillin, regulated by the Na+/H+ antiporter downstream of ROCK, is essential for MM1-cancer cell migration.
This work was supported in part by a Grant-in-Aid for the Second Term Comprehensive 10-Year Strategy for Cancer Control from the Ministry of Health and Welfare of Japan and a Grant-in-Aid for Graduate Students from Hyogo College of Medicine.