Protein kinase C (PKC) is a family of serine/threonine kinases involved in the transduction of a variety of signals. There is increasing evidence to indicate that specific PKC isoforms are involved in the regulation of distinct cellular processes. In glioma cells, PKC α was found to be a critical regulator of proliferation and cell cycle progression, while PKC ε was found to regulate adhesion and migration. Herein, we report that specific PKC isoforms are able to differentially activate extracellular-signal regulated kinase (ERK) in distinct cellular locations: while PKC α induces the activation of nuclear ERK, PKC ε induces the activation of ERK at focal adhesions. Inhibition of the ERK pathway completely abolished the PKC-induced integrin-mediated adhesion and migration. Thus, we present the first evidence that PKC ε is able to activate ERK at focal adhesions to mediate glioma cell adhesion and motility, providing a molecular mechanism to explain the different biological functions of PKC α and ε in glioma cells.
A prominent factor in the pathology of malignant gliomas is their ability to migrate from the primary tumor mass and to diffusely infiltrate the brain parenchyma, prohibiting complete resection of the tumor by surgery (Berens and Giese, 1999; Holland, 2000). A better understanding of the processes controlling glioma migration is thus required in order to improve the therapy of the disease. Several cell surface receptors are involved in promoting glioma cell migration, including the epidermal growth factor receptor (Lund-Johansen et al., 1990; El-Obeid et al., 1997), and several integrins (Friedlander et al., 1996). Downstream of these cell surface receptors, activation of the PKC pathway (reviewed in Newton, 1997) has been implicated in the regulation of glioma cell migration: inhibition of phospholipase C γ, an enzyme involved in PKC activation, blocked glioma cell migration and invasion in vitro (Khoshyomm et al., 1999). Furthermore, inhibition of PKC could inhibit glioma cell motility and invasion (Zhang et al., 1997), while PKC activation with phorbol esters promoted migration in vivo (Tysnes and Laerum, 1993).
We have previously shown that several PKC isoforms regulate different phenotypes of human glioma cells. PKC α was found to regulate cell cycle progression and proliferation (Besson and Yong, 2000), and to negatively regulate adhesion and motility (Besson and Yong, manuscript submitted). On the other hand, PKC ε activation resulted in increased focal adhesion formation, clustering of several integrins, and enhanced integrin-mediated adhesion and motility of glioma cells (Besson and Yong, manuscript submitted). However the molecular mechanism by which these two PKC isoforms regulate distinct pathways remains elusive.
The involvement of PKC in the activation of the ERK pathway (reviewed in Schaeffer and Weber, 1999) has been well documented in other cell systems (El-Shemerly et al., 1997; Schonwasser et al., 1998; Howe and Juliano, 1998; Miranti et al., 1999; Short et al., 2000; Rigot et al., 1998; Traub et al., 1997). In several cell types, PKC was required for integrin-mediated ERK activation (Howe and Juliano, 1998; Miranti et al., 1999; Short et al., 2000; Rigot et al., 1998; Traub et al., 1997). In addition, activation of the ERK pathway following different stimuli at the cell surface was required for cell migration in various cell systems (Rigot et al., 1998; Klemke et al., 1997; Nguyen et al., 1999). For example, ERK activation was required for PKC-induced migration of colon carcinoma cells (Rigot et al., 1998). Also, the ε isoform of PKC was found to mediate ERK1/2 activation following adhesion on fibronectin in bovine aortic endothelial cells (Traub et al., 1997).
In the present report, we provide novel evidence indicating that in human glioma cells, ERKs are activated at focal adhesions following PKC stimulation. Also, we demonstrate that the activation of ERKs at focal adhesions is dependent on PKC ε activity. Moreover, ERK activation is required for PKC-induced integrin-mediated adhesion and motility of human glioma cells.
ERK1/2 are activated downstream of PKC and localize to focal adhesions in human glioma cells
To investigate whether the ERK pathway was activated downstream of PKC in human glioma cells, phospho-specific antibodies that recognize the activated form of ERK1 and ERK2 were used. PMA stimulation rapidly induced a robust and sustained activation of both p44/ERK1 and p42/ERK2, as indicated by their phosphorylation levels on Thr202/Tyr204 (Figure 1). ERK activation was blunted by pre-treatment with PKC-specific inhibitors (Calphostin C, Bisindolylmaleimide I, and Gö6983) 1 h prior to PMA stimulation, indicating that PMA-induced ERK activation was a consequence of PKC activation (Figure 1).
Immunofluorescence experiments performed on glioma cells revealed that phospho-ERK was localized to cellular structures resembling focal adhesions, and this distribution was more pronounced following PMA treatment (Figure 2a). To confirm that activated ERKs were localized to focal adhesions in glioma cells, co-localization studies with vinculin and paxillin, two proteins known to be present at focal adhesions, were performed. Phospho-ERK extensively co-localized with both vinculin (Figure 2b) and paxillin (Figure 2c), and as the number of focal adhesions increased following PMA treatment, so did the phospho-ERK distribution at focal adhesions (Figure 2b,c). The number of focal adhesions following PMA (100 nM) treatment was monitored on vinculin-stained cells using the Image-Pro image analysis software (Figure 3a). A clear increase in the number of focal adhesions was observed at 2 and 6 h following PMA stimulation when compared to vehicle treated cells (control) (Figure 3a). Together, the results indicate that the ERK pathway is activated downstream of PKC and that activated ERKs localize to focal adhesions.
We also monitored the localization of total ERK (independent of its activation state) by immunofluorescence (Figure 3b). ERK could be seen both at focal adhesions and in the nucleus (Figure 3b), and the distribution of ERK between these locations did not appear to be affected by PMA stimulation. Altogether, these results suggest that distinct pools of ERK exist in the cell, and that PMA stimulation of glioma cells induces predominantly the activation of the pool that is present at focal adhesions (Figure 2).
PKC activity is required for activation of ERKs at focal adhesions
We then investigated whether the localization of activated ERK to focal adhesions was dependent on PKC activity. In the presence of either Calphostin C (200 nM) or Bisindolylmaleimide I (5 μM), phospho-ERK failed to localize to focal adhesions and was found predominantly in the cytoplasm and the perinuclear region (Figure 4d,f). The absence of phospho-ERK staining at focal adhesions was not due to the disruption of focal adhesions by the PKC inhibitors since these structures could still be visualized by vinculin staining (Figure 4h,I). Addition of PMA in the presence of PKC inhibitors only weakly rescued the localization of phospho-ERK to focal adhesions (Figure 4e,g). These results indicate that PKC activity is required for ERK activation at focal adhesions.
Different PKC isoforms activate different pools of ERKs in distinct subcellular locations
We previously found that PKC α and ε were controlling different phenotypes of glioma cells, namely, proliferation and adhesion/motility (Besson and Yong, 2000, and manuscript submitted). An attractive hypothesis is that since PKC ε regulates glioma cell motility, it could be the isoform responsible for ERK activation at focal adhesions. To determine whether a specific PKC isoform had an effect on the subcellular localization of ERK activation, we analysed various clones overexpressing either PKC α or ε, or transfected with an antisense PKC ε construct, by immunofluorescence with a phospho-ERK specific antibody (Figure 5). In wild type U251N cells and empty vector transfected cells (pBK) (Figure 5a,b), phospho-ERK was localized mostly to focal adhesions after PMA stimulation, and some phospho-ERK was detected in the nucleus. However, in two clones overexpressing PKC α (Figure 5c,d), a significantly higher proportion of phospho-ERK was localized in the nuclei at all times. In contrast, in clones overexpressing PKC ε (Figure 5e,f), phospho-ERK was extensively localized at focal adhesions. In all these cells (Figure 5a,g), there was an increase in the amount of phospho-ERK localized at focal adhesions following PMA treatment for 2 h. On the other hand, the level of phospho-ERK present in the nuclei of PKC α overexpressing cells was not affected by the PMA stimulation (Figure 5c,d). No difference was observed between PKC α overexpressing cells and their control vector transfected counterparts in their level of ERK activation following PMA stimulation by Western blotting (data not shown). Cells transfected with an antisense PKC ε cDNA (εAS30), in which PKC ε levels are markedly decreased (Besson and Yong, manuscript submitted), provided further evidence that PKC ε is responsible for targeting phospho-ERK to focal adhesions since the activation of ERK at focal adhesions was greatly diminished (Figure 5h) when compared to its empty vector transfected counterpart (pREP) (Figure 5g). Together, the data indicates that PKC α induces the nuclear activation of ERK, while PKC ε is responsible for the activation of ERK at focal adhesions.
In addition to the differential targeting of ERKs by distinct PKC isoforms, we found that PKC α and ε were localized to distinct subcellular locations following activation. It is well known that activated PKCs associate with membranes (Newton, 1997), however, the subcellular localization of pKC isoforms when activated has never been investigated in glioma cells. Thus, we investigated the distribution between nuclear (including the nuclear envelope) and cytoplasmic fractions (including plasma, ER, and golgi membranes) of PKC α and ε following activation. Interestingly, PMA stimulation rapidly translocated PKC α to the nucleus (Figure 6a), while PKC ε was mostly retained in the cytoplasmic fraction (which contains other particulate matter, including the plasma membrane) (Figure 6b). The differential targeting of PKC α and ε could be the basis for the activation of pools of ERKs in distinct subcellular locations.
PKC-induced ERK activation occurs through a MEK1/2-dependent mechanism in human glioma cells
We next attempted to determine by what mechanism PKC activates ERK, as several distinct pathways have been described (El-Shemerly et al., 1997; Schonwasser et al., 1998; Howe and Juliano, 1998; Miranti et al., 1999; Short et al., 2000; Rigot et al., 1998; Traub et al., 1997). Pharmacological studies showed that ERK activation by PKC occurs in a MEK1/2-dependent mechanism, since the MEK1/2 inhibitors PD98005 (25 μM) and U0126 (10 μM) efficiently blocked PMA-induced ERK activation (Figure 7a). Treatment with U0126 also dramatically decreased the phospho-ERK stain by immunofluorescence and completely prevented the PKC-induced activation of ERK at focal adhesions (Figure 7c, right column). Because Fincham et al. (2000) recently reported that targeting of activated ERK to focal adhesions was dependent on v-Src activity; we tested whether the Src-family kinase inhibitor PP2 could prevent ERK activation in our cell system. We found that neither the PKC-induced ERK activation, nor the localization of phospho-ERK to focal adhesion were affected by PP2 treatment (10 μM) (Figure 7b,c), indicating that Src is not involved downstream of PKC in this signaling cascade.
ERK activation downstream of PKC is required for PMA-induced integrin-mediated adhesion and migration
We next tested the effect of ERK inhibition on PKC-induced integrin-mediated adhesion and motility. The presence of the MEK1/2 inhibitors PD98005 (25 μM) or U0126 (10 μM), that prevented PKC-induced ERK activation (Figure 7a,c), completely abolished the PMA-induced increase in adhesion on laminin (Figure 8a), vitronectin (Figure 8b), or fibronectin (Figure 8c) in Es10 cells that overexpress PKC ε. The Src family inhibitor PP2 (10 μM) could also abolish the PMA-induced increase in adhesion on these integrin substrates (Figure 8) (see Discussion).
Similarly, both PD98005 (25 μM) and U0126 (10 μM) reduced the basal and the PMA-induced migration of pBK and Es10 cells (Figure 9). The Src family kinase inhibitor PP2 (10 μM) completely blocked both basal and PKC-induced migration (Figure 9), while the PI 3-kinase inhibitor wortmannin had no effect on PKC-induced migration (Figure 9). Together, these results indicate that ERK activation downstream of PKC is required for PMA-induced integrin-mediated adhesion and migration.
In this report, we present evidence indicating that PKC activation by the phorbol ester PMA induces the activation of ERK1 and ERK2, and that activated ERKs localize to focal adhesions. Moreover, our results suggest that distinct pools of ERK are activated by specific PKC isoforms: PKC α overexpression induced ERK activation in the nucleus, while PKC ε overexpression is responsible for the activation of ERKs at focal adhesions. Also, the activation of ERK downstream of PKC appears to be required for PMA-induced integrin mediated adhesion and migration of human glioma cells.
This is, to our knowledge, the first time that two different PKC isoforms have been shown to spatially regulate ERK activation in distinct subcellular localizations. The ability of several PKC isoforms to activate ERKs within the same cells has been reported previously (Schonwasser et al., 1998), however, the subcellular localization of ERKs was not studied. It remains unclear at this point how PKC mediates this differential activation of ERKs. However, one may speculate that since each PKC isoform seems to translocate to a specific location upon activation, the nuclear envelope in the case of PKC α and the plasma membrane in the case of PKC ε, they could activate different pools of ERKs present in these locations. Consistent with this hypothesis, staining with a total ERK antibody revealed a population of ERK in the nucleus, and another one at focal adhesions in human glioma cells (Figure 3b). It remains to be determined whether PKC α-mediated ERK activation in the nucleus plays a role in the regulation of proliferation by this isoform in human glioma cells (Besson and Yong, 2000). This seems likely since the nuclear translocation of ERKs is generally associated with a proliferative response of cells (Garrington and Johnson, 1999).
Our finding that activated ERKs localize to focal adhesions in glioma cells contrasts with many other cell types, in which ERKs usually translocate to the nucleus and mediates growth factor-induced proliferative responses (Garrington and Johnson, 1999). However, Fincham et al. (2000) recently reported that activated ERK was targeted to focal adhesions following integrin engagement in rat and chick embryo fibroblasts, in a v-Src-dependent manner, and also that activation of the myosin light chain kinase (MLCK) downstream of ERKs was required for proper targeting of activated ERK to focal adhesions. We tested the effect of the Src family kinase inhibitor PP2 and the MLCK inhibitors ML-7 and ML-9 on the PKC-induced ERK activation, and on the localization of phospho-ERK to focal adhesions following PKC activation. Neither of these pharmacological inhibitors prevented PKC-induced ERK activation, nor the localization of activated ERKs to focal adhesions in glioma cells (Figure 7c,b and data not shown), suggesting that Src is not downstream of PKC in this signaling cascade. It is, however, very likely that Src kinases play a critical role in ERK activation downstream of integrins or receptor tyrosine kinases, and possibly, PKC could be involved downstream of Src in this process. Although Src inhibition had no effect on PKC-induced ERK activation and localization to focal adhesions, we found that Src inhibition could block both basal and PKC-induced integrin-mediated adhesion and migration. These results are not very surprising considering the critical role played by Src in focal adhesion assembly/disassembly, and in integrin-mediated signal transduction (Giancotti and Ruoslahti, 1999; Klinghoffer et al., 1999; Sieg et al., 1999). Thus, Src family kinases inhibition likely results in a general inhibition of most processes initiated by integrins (including adhesion and migration), and activation of PKC is not sufficient to rescue this defect.
MLCK has been shown to be a major effector in ERK-mediated cell migration (Klemke et al., 1997; Nguyen et al., 1999). We therefore tested whether MLCK was playing a role in the regulation of adhesion and migration downstream of PKC and ERK in glioma cells. The MLCK inhibitor ML7 had no effect on PKC-induced adhesion; however, it effectively blocked PKC-induced migration of Es10 cells (data not shown). This observation is in agreement with the report by Gillespie et al. (1999) in which they demonstrated the efficacy of MLCK inhibitors to inhibit glioma cell migration.
The exact function of activated ERKs at focal adhesions remains unclear at this point. Although, as our data indicate, ERKs are clearly involved in the regulation of adhesion and migration, their direct targets are still unknown. Likely candidates for ERK phosphorylation at focal adhesions are MLCK, known to be directly activated by ERKs (Klemke et al., 1997; Nguyen et al., 1999; Fincham et al., 2000), and other cytoskeletal or cytoskeleton-associated proteins. In conclusion, we have found that activation of ERK1/2 downstream of PKC and their localization to focal adhesions is required for PKC ε-induced adhesion and migration. Several PKC isoforms appear to mediate different phenotypes of glioma cells: PKC ε regulates adhesion and migration, while PKC α regulates cell cycle and proliferation (Besson and Yong, 2000), possibly through their ability to activate distinct pools of ERK in various subcellular locations. Future therapeutic strategies targeting the PKC pathway in gliomas should aim at inhibiting the activity of both these isoforms.
Materials and methods
Antibodies and reagents
The cDNA for human PKC α was kindly provided by Dr G Finkenzeller (Institut für Molekulare Medizin, Freiburg, Germany). the cDNA for human PKC ε was from ATCC (#80050). Phorbol-12-myristate-13-acetate (PMA), PD98005, U0126, Bisindolylmaleimide I, Calphostin C, Gö6983, poly-L-lysine, laminin, vitronectin, and fibronectin, were from Calbiochem. The monoclonal antibody to vinculin (hVIN-1) and the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide thiazolyl blue (MTT) were from Sigma-Aldrich. Monoclonal antibodies to PKC α (clone 3) and PKC ε (clone 21) were from BD Transduction Laboratories. Rabbit antibodies to p42/44 MAPK (ERK)(#9102) and phospho-p42/44 MAPK (Thr202/Tyr204)(#9101) were from New England Biolabs. Note that the specificity of the P-ERK (Thr202/Tyr204) antibody was extensively investigated by Fincham et al. (2000). The total ERK antibody used for immunofluorescence was from Zymed.
Tissue culture and transfections
Human glioma cells were grown as described previously (Besson and Yong, 2000). Stable transfections were performed using the calcium-phosphate method, and following selection, clones were isolated with cloning rings. Stably transfected cells were kept at all times in the presence of 400 μg/ml G418 (Calbiochem). The PKC cDNAs in sense orientation were inserted in the pBKRSV vector (Invitrogen). αs20 and αs27, and εs1 and εs10 are clones stably overexpressing PKC α or PKC ε, respectively. For antisense studies, the PKC ε cDNA was cloned in antisense orientation in a pREP9 episomal vector. Lines transfected with the empty pBKRSV or pREP9 vector were used as control.
Fractionation into cytosolic and nuclear fractions
Cells were trypsinized, pelleted, and rinsed once in hypotonic buffer (10 mM HEPES (pH 7.9), 1.5 mM MgCl2, 10 mM KCl, 0.5 mM dithiothreitol). Cells were resuspended in hypotonic buffer containing 0.1% Nonidet P-40 and incubated for 10 min at 4°C, with occasional mixing and pipetting. Lysis of the plasma membrane and integrity of the nuclei was checked under a phase-contrast microscope. Nuclei were pelleted by centrifugation at 2000 r.p.m. for 5 min, rinsed once in hypotonic buffer, and lysed in lysis buffer (50 mM HEPES (pH 7.5), 150 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 0.1% Tween-20, 10% glycerol, 1% Nonidet P-40, 0.2% SDS; supplemented with 1 mM dithiothreitol, 10 mM β-glycerophosphate, 10 mM NaF, 1 mM sodium orthovanadate, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 10 μg/ml pepstatin-A, and 1 mM phenylmethylsulfonylfluoride). The nuclear fraction was sonicated for 10 s and debris were cleared by centrifugation for 5 min at 10 000 r.p.m.
Cells were scraped from the plates, lysed in lysis buffer, incubated on ice for 30 min, and homogenized. Western blots were performed as described previously (Besson and Yong, 2000) with the following differences: primary antibodies were incubated overnight at 4°C, and the ERK antibodies were diluted in PBS containing 0.5% Tween-20 and 5% bovine serum albumin.
Cells were grown on glass coverslips for 24 h before treatment. Following the appropriate treatment, cells were rinsed in PBS and fixed in 1% paraformaldehyde at 37°C for 20 min. Coverslips were stored in PBS at 4°C until stained. Cells were permeabilized for 3 min in 0.2% Triton- X-100, rinsed three times in PBS, and incubated overnight with primary antibodies diluted in antibody dilution buffer (PBS supplemented with 3% bovine serum-albumin, 0.05% Tween-20, and 0.08% sodium azide) at 4°C. The coverslips were rinsed three times in PBS. Incubation with secondary antibodies at a 1/500 dilution was for 30 min at 37°C. The coverslips were rinsed three times in PBS and mounted on glass slides. Images were obtained using an Olympus Fluoview 300 confocal laser-scanning microscope using the 60× objective. For the counts of focal adhesions, images of vinculin-stained (at 1/1000) cells were taken using a 40× objective, and focal adhesions were counted using the Image Pro image analysis software; a minimum of 185 cells was counted.
Plates were coated for 1 h at 37°C with poly-L-lysine at 10 μg/ml in PBS. Where needed, further coating with ECM proteins (laminin, vitronectin, or fibronectin) at 10 μg/ml in PBS was carried overnight at 37°C. Cells were trypsinized, counted, and diluted to a concentration of 5×105 cells per ml. When required, cells were incubated with the appropriate inhibitor for 15 min at 4°C (with the exception of PMA which was also added 12 h prior to the experiment). In each well of a 96-well plate, 3×104 cells were seeded and incubated for 10 min at 37°C. Plates were rinsed twice with PBS, and the remaining adherent cells were incubated for 90 min in medium containing 0.1 mg/ml of MTT. After three rinses with PBS, the MTT stain was solubilized in 100 μl dimethylsulfoxyde, and OD was measured at 550 nm.
Cells were seeded at 80% confluency in 60 mm dishes and grown for an additional 24 h. A scratch was done with a rubber policeman in the center of the plate, the plate rinsed with PBS. Cells were incubated for 48 h in growth medium supplemented with the appropriate activator or inhibitor, rinsed with PBS, fixed 10 min in 95% ethanol-5% acetic acid, and stained with hematoxylin. For each plate, pictures were taken on an inverted microscope (Olympus) at 40× magnification. The distance migrated by the cells at each time point was then measured (in mm) on the prints.
We gratefully thank Tammy L Wilson for technical assistance in the adhesion assays. This work is supported by a grant from the Canadian Institutes for Health Research. A Besson is a Research Student of the National Cancer Institute of Canada supported with funds provided by the Terry Fox Run. A Davy is the recipient of an Alberta Cancer Board postdoctoral fellowship. SM Robbins is a scholar of the Alberta Heritage Foundation for Medical Research and holds a Canada Research Chair in Cancer Biology. VW Yong is a Medical Research Council of Canada Scientist and a senior scholar of the Alberta Heritage Foundation for Medical Research.
About this article