Phosphatase-Mediated Crosstalk Control of ERK and p38 MAPK Signaling in Corneal Epithelial Cells

Abstract

purpose. To test the hypothesis that the protein phosphatases PP2A and MKP-1 are involved in controlling epidermal growth factor (EGF)-induced increases in rabbit corneal epithelial cell (RCEC) migration by mediating crosstalk between signaling pathways eliciting EGF receptor control of migration and proliferation.

methods. Western blot analysis was used to determine the phosphorylation status of Erk1/2, p38, and the mitogen-activated protein kinase (MAPK) kinase (MEK1/2) using inhibitors of Erk1/2 or p38 and dominant-negative (d/n) Erk1 or d/n p38 cell lines. Coimmunoprecipitation was used to evaluate protein phosphatase (PP)2A and Erk1/2 interaction. Short-interfering RNA (siRNA) transfection was performed to analyze the involvement of MAPK phosphatase (MKP)-1 in crosstalk. Scratch-wound assay was used to determine EGF-dependent effects on cell migration.

results. EGF (10 ng/mL) induced changes in activation of Erk1/2 and p38, which were enhanced by inhibition with 10 μM SB203580 and 10 μM PD98059, respectively. PP inhibition with sodium orthovanadate (100 μM), okadaic acid (10 nM), or Ro 31–8220 (10 μM) resulted in larger and more prolonged increases in the phosphorylation status of Erk1/2 and p38. After 1 hour, EGF induced 14-fold increases in MKP-1 protein expression. After MKP-1 siRNA transfection, EGF had induced a similar pattern of changes in the phosphorylation status in Erk1/2 and p38 following PP inhibition. EGF-induced cell migration was enhanced by Erk1/2 pathway inhibition and was accentuated after PP inhibition. Conversely, p38 pathway inhibition eliminated this response.

conclusions. EGF-induced changes in Erk1/2 and p38 phosphorylation status are dependent on PP-mediated crosstalk. This control modulates the magnitude of growth factor–induced increases in corneal epithelial cell migration.

Renewal of the corneal epithelium is essential to this tissue layer’s role as a barrier against exogenous noxious environmental agents and bacterial infection. For this function to be preserved, close cell-to-cell apposition is essential for tight junctional integrity. ¹ The continuous sloughing off of the topmost differentiated superficial cell layer results in loss of integrity if these losses are not compensated for by the continuous proliferation of cells in the epithelial basal layer. Autocrine- and paracrine-mediated release of a host of growth factors is essential for this constant renewal to occur. These growth factors affect increases in epithelial cell renewal through the stimulation of cell cognate receptors that are linked to myriad interacting cell-signaling pathways. Such interactions elicit integrated control of cell proliferation and migration, both of which are required for corneal epithelial cell renewal and maintenance of corneal transparency. 

In corneal epithelial cells, various cytokine receptors, or growth factors, mediate the control of cell proliferation and migration through the stimulation of extracellular signal-regulated kinase (ERK) and p38 pathways of the mitogen-activated protein kinase (MAPK) cascade. ^{2 3} The magnitude of each pathway’s responses is affected by their signal strength and by their kinetics of activation. Furthermore, such signaling appears to be modulated through crosstalk because either hepatocyte growth factor (HGF)- or keratinocyte growth factor (KGF)-induced stimulation of migration is accentuated by chemical inhibition of ERK. ³ This latter interaction suggests that the overall effect of each of these growth factors on cell migration comprises an integrated response that is dependent on the extent of crosstalk between the two MAPK signaling pathways. Nevertheless, the precise mechanism(s) affecting such feedback control after epidermal growth factor (EGF) stimulation has not yet been described in this cell layer. 

Numerous studies show that crosstalk between parallel pathways of the MAPK cascade is dependent on increases in activity and expression of protein phosphatase (PP). ^{4 5 6 7 8 9 10 11} Such increases occur in response to activation (i.e., phosphorylation) of a specific branch of the MAPK cascade. Through MAPK-induced increases in phosphatase expression, crosstalk results in the dephosphorylation of a kinase in a parallel pathway. Therefore, the balance between phosphorylation and dephosphorylation of kinase reactions determines the duration of MAPK signaling activation. Evidence is clear in other cell types (e.g., cardiomyocytes, fibroblasts, and macrophages) that kinase activation of ERK and p38 pathways has two simultaneous effects, each of which leads to downstream activation, or phosphorylation, of substrates that affect early gene expression and to inactivation, or dephosphorylation, of a parallel pathway through increases in activity and induction of expression of PP. Such dual effects indicate that the extent of feedback control between the two parallel signaling pathways is determined by increases in PP expression and activity. ^{12 13}  

Members of a large family of tissue-specific PPs that affect the crosstalk between the parallel signaling pathways include PP1, PP2a, and MAPK phosphatase (MKP)-1. It has been shown that rapid increases in the expression and activity of Erk1/2 and p38 occur after cytokine-induced activation. ¹⁴ However, though PPs have critical roles in dictating the magnitudes of different responses to cytokine receptor–mediated stimulation in other tissues, there is no information regarding their roles in mediating crosstalk between the ERK and p38 pathways of the corneal epithelium. 

EGF is one of numerous cytokines proven to be effective in mediating in vitro stimulation of corneal epithelial wound healing. ¹ The kinetics of this response is dependent on the rates of epithelial cell proliferation and migration, which are induced by EGF. The EGF receptor–linked upstream cell signaling pathways that affect increases in proliferation have been partially identified. ¹⁵ Signaling involves the preferential activation of specific protein kinase C isoforms, phospholipase Cγ, ion transport mechanisms, and induction of calcium signaling through the stimulation of capacitative calcium entry and the activation of the ERK branch of the MAPK cascade. ^{15 16 17 18 19 20 21} However, it is unknown whether EGF induces increases in cell migration through stimulation of the p38 limb or whether crosstalk modulates EGF-induced activation of ERK and p38. 

In this study, we show that EGF-induced stimulation of rabbit corneal epithelial cell (RCEC) migration occurs through activation of the p38 pathway, the activation pattern of which is controlled by crosstalk between ERK and p38 signaling pathways. PP2A and MKP-1, whose expression is induced by MAPK activation, are involved. Our results pose the interesting idea that manipulation of PP activity/expression could provide a therapeutic modality for optimizing cytokine-induced responses that affect the rates of wound healing after injury. 

Materials and Methods

Materials

The following antibodies and assay kits were purchased (all from Cell Signaling Technology, Danvers, MA): monoclonal anti–mouse phospho p38 (Thr188/Tyr182) antibody, monoclonal anti–mouse phospho Erk1/2 antibody, p38 MAP kinase assay kit, Erk1/2 MAP kinase assay kit, rabbit anti–Erk1 antibody, rabbit anti–p38 antibody, mouse monoclonal anti–phospho-ERK, MKP-1 rabbit polyclonal IgG, MAP kinase kinase (MEK)-1 rabbit polyclonal IgG (12B), phospho-MEK-1/2 (Ser 218/Ser 222), Actin (H-196) rabbit polyclonal IgG antibody and rabbit polyclonal IgG, PP2A goat polyclonal IgG, phospho PP2A (Tyr307)-R rabbit polyclonal IgG, goat anti–mouse IgG-HRP, goat anti–rabbit IgG-HRP antibody, and MKP-1 short-interfering RNA (siRNA) transfection kit. The following were also purchased: control siRNA (Santa Cruz Biotechnology, Santa Cruz, CA); inhibitors PD98059, U0126, SB203580, and okadaic acid (Biomol, Plymouth Meeting, PA); Ro 31–8220 (Calbiochem, La Jolla, CA); sodium orthovanadate (Na₃VO₄), EGF, and bovine insulin (Sigma-Aldrich, St. Louis, MO); DMEM/F12 medium, fetal bovine serum (FBS), and tetracycline-inducible (tet-on) transfection kit (TREX; Invitrogen Corp., Carlsbad, CA). 

Cell Culture

SV40 adenovirus–immortalized RCECs were a generous gift from Kaoru Araki-Sasaki (Kagoshima Miyata Eye Clinic, Kagoshima, Japan). This cell line exhibits phenotypic and functional properties similar to those of its primary culture counterpart while allowing continuous growth for more than 100 passages. ^{22 23} Passages 12 to 23 were used in the present study. RCECs were cultured in DMEM/F12 medium supplemented with 6% FBS, 5 ng/mL EGF, 5 μg/mL insulin, and 40 μg/mL gentamicin in a humidified incubator with 5% CO₂ and 95% ambient air at 37°C. The medium was replaced every 2 days. Cultures were detached using 0.05% trypsin-EDTA, and cells were seeded in a 35-mm dish at a density of 2 × 10⁴ cells/dish. Cell cycle arrest was achieved by culturing cells in basic medium (serum and EGF-free DMEM/F12) for 24 hours before experimentation. 

Transfection

Tetracycline-inducible RCEC lines were generated with dominant-negative (d/n) cDNA Erk1 and p38 mutant constructs provided by M. H. Cobb (University of Texas Southwestern Medical Center at Dallas) and the TREX system. RCECs coexpressing the tetracycline repressor protein (tetR) were cotransfected with pcDNA6/TR, and cDNA Erk1 and p38 mutant constructs were cotransfected with reagent (Lipofectamine; Invitrogen, Carlsbad, CA) with pcDNA6/TR and pcDNA4/To-Erk1 or with pcDNA4/To-p38 plasmids according to the protocol provided by the manufacturer and as previously described. ²⁴ Transfection efficiency was estimated to be 85%, based on the result of experiments in which cells were transfected with a green fluorescent protein (GFP)-containing construct. 

MKP-1 siRNA Transfection

Functional gene silencing was performed as described with siRNA, which targets MKP-1. ¹⁴ Briefly, RCECs were washed twice with 1× phosphate-buffered saline (PBS) before trypsinization. Cells were centrifuged at 1000 rpm for 5 minutes and resuspended in appropriate growth medium containing 10% FBS, but without antibiotics. Cells were then cultured overnight and allowed to reach 30% to 50% confluence. Transfection with siRNA was accomplished according to the manufacturer’s instructions. In brief, siRNA transfection medium was added to siRNA, gently mixed, and incubated at room temperature for 5 minutes. The siRNA transfection reagent was then added to the siRNA transfection medium, which was mixed gently and incubated for 5 minutes at room temperature. During this 5-minute incubation period, siRNA transfection reagent and siRNA transfection medium were combined to form the siRNA–siRNA transfection reagent complex, which was gently mixed and incubated at room temperature for 20 minutes. This complex was replaced with 600 μL fresh growth medium devoid of antibiotics in which transfected cells were incubated for 72 hours at 37°C before experimentation. 

Western Blot Analysis

Western blot experiments were performed as described. ²⁵ In brief, after reaching 80% confluence, cells were gently washed twice in cold PBS and harvested in 0.5 mL of cell lysis buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM glycerol-phosphate, and 1 mM Na₃VO₄) with a protease inhibitor mixture (1 mM PMSF, 1 mM benzamidine, 10 μg/mL leupeptin, and 10 μg/mL aprotinin). Cell lysates were sonicated and precleared by centrifugation at 13,000 rpm for 15 minutes at 4°C. Protein concentration was measured with a protein assay kit (BCA; Pierce Biotechnology, Rockford, IL). Briefly, 200 μg protein from each sample was mixed with equal volumes of 2× Laemmli buffer and then boiled for 5 minutes to denature the cell lysates. After this procedure, 20 μg denatured protein samples were electrophoresed on 10% polyacrylamide SDS minigels. After protein electrotransfer to polyvinylidene difluoride (PVDF) membranes and blocking with 5% nonfat dry milk, blots were incubated overnight with the appropriate specific primary antibody at 4°C. After primary antibody incubation, blots were washed three times and incubated with the appropriate horseradish peroxidase (HRP) secondary antibody–labeled IgG (1:2000 dilution) for 1 hour at room temperature. Immunoreactive bands were detected with an ECL Plus kit (GE Healthcare, Little Chalfont, UK). Films were scanned, and band density was quantified (SigmaScan Pro 5.0 software; Sigma, St. Louis, MO). Monoclonal anti–Erk1/2, anti–p38, anti–IgG, and β-actin antibodies were used as internal controls to test for equal protein loading. 

Coimmunoprecipitation

Cells were lysed with lysis buffer and spun at 13,000 rpm for 10 minutes. An aliquot of supernatant containing 500 μg protein was exposed to either anti–PP2A antibody or anti–Erk1/2 antibody and gently agitated overnight at 4°C. Subsequently, protein A beads were added and cell lysates were exposed for another 2 hours at 4°C with gentle agitation. Beads were washed twice with cell lysis buffer. The pellets obtained were resuspended in 50 μL cell lysis buffer, to which was added the same volume of 2× Laemmli buffer. The mixture was boiled for 5 minutes and then subjected to Western blot analysis. 

Scratch-Wound Assay

Tetracycline-induced d/n Erk1 and d/n p38 cell lines and RCECs were grown to confluence on 35-mm culture plates supplemented with growth medium. RCECs were starved for 24 hours in basic medium, and d/n cell lines were starved in basic medium in the presence of tetracycline. After starvation, cells were scratch-wounded using a sterile 200-μL pipette tip, washed twice with basic medium to remove suspended cells, and re-fed with medium in the absence or presence of EGF (10 ng/mL). To determine the role of EGF-induced migration, PD98059 or SB203580 was added to the medium with RCECs. Hydroxyurea (5 mM) was also added to the medium to reduce proliferation during the experiment. The progress of cell migration into the wound was photographed at 4, 8, 12, and 24 hours using an inverted microscope (Diaphot 200; Nikon, Tokyo, Japan) coupled to a CCD camera (IC-200 CoolSNAP _fx ; Photometrics, Tucson, AZ). Each experimental condition was repeated three times. Statistical analyses were performed using unpaired Student t-test. Data are shown as mean ± SEM. 

Proliferation

[³H]-Thymidine incorporation was performed as described. ¹⁵ After 24 hours of serum starvation in medium supplemented with 0.5% bovine serum albumin, cells were incubated at 37°C for 1 hour with 1 μCi/mL [³H]-thymidine (3.3–4.8 TBq/mmol) and then washed three times with ice-cold 5% trichloroacetic acid. Cell lysates were solubilized with 0.2 N NaOH and 2% SDS. Radioactivity was monitored using a liquid scintillation counter, and the data were normalized to cellular protein content using a modified Lowry assay. 

Results

EGF-Induced Erk1/2 and p38 MAPK Phosphorylation

To examine the functions of Erk1/2 and p38 in RCECs, cells were exposed to 10 ng/mL EGF, incubated for various time periods, and assayed for EGF-induced changes in MAP kinase activity (phosphorylation). As seen in Figures 1A and 1B , cells exposed to EGF showed time-dependent biphasic changes in the phosphorylation status of Erk1/2 and p38. After a 5-minute exposure to EGF, Erk1 reached a maximum level of phosphorylation that was 20-fold greater than that of its control (Figs. 1A 1C) . During the subsequent 2 hours, Erk1 stimulation waned progressively until it reached baseline. There were also concomitant changes in Erk2 phosphorylation during that time period. Conversely, after 15 minutes, p-p38 increased nearly threefold; this was followed by a slight decline to a plateau level for the next 15 minutes (Figs. 1B 1C) . Subsequently, p38 activation reverted to control level after 60 minutes, somewhat faster than the time required for Erk1 recovery to its baseline level. Neither the p-Erk1/2 nor the p-p38 pattern of change was caused by any differences in protein loading because their total Erk1/2 and p38 protein levels did not vary. 

PP-Mediated Crosstalk between Erk1/2 and p38 Pathways

In other mammalian tissues, such as rat, crosstalk interaction exists between the different branches of the MAPK cascade. ²⁶ In those cases, the extent of stimulation of one branch of the MAPK cascade is sensitive to the levels of activation of a parallel branch of this signaling cascade. We probed for similar interactions by determining whether changes in the activity of either the Erk1/2 or the p38 pathways had a feedback effect on the other parallel pathway. EGF induced increases in Erk1/2 and p38 phosphorylation that were almost completely inhibited by 10 μM PD98059 and 10 μM SB203580, respectively (Fig. 2A) . After 15-minute exposure to PD98059, EGF induced an increase in p38 phosphorylation that was 40% greater than that obtained with EGF alone (Fig. 2B) . Furthermore, after 10-minute exposure to SB203580, EGF (10 ng/mL) induced an increase in Erk1/2 phosphorylation that was nearly twofold greater than that caused by EGF alone (Fig. 2C) . Such an interaction is consistent with evidence provided for the existence of crosstalk in human corneal epithelial cells. ²⁷  

The effects of the broad-spectrum phosphatase inhibitor sodium orthovanadate (100 μM), the MKP-1 inhibitor Ro 31–8220 (10 μM), and the PP2A inhibitor okadaic acid (10 nM) on maximum EGF-induced increases in Erk1/2 and p38 phosphorylation status were also evaluated (Fig. 2) . The addition of sodium orthovanadate increased p38 and Erk1/2 phosphorylation levels (threefold and twofold, respectively) more than did EGF alone. Similarly, Ro 31–8220 and okadaic acid had comparable effects. Thus, the extent of crosstalk between the p38 and the ERK pathways appeared to be associated with changes in activity, expression, or both of PP. 

Activation of d/n Erk1 and d/n p38 Gene Expression

To confirm the effects of MAPK inhibitors on the phosphorylation status of the ERK and p38 pathways, two stably transfected RCEC lines expressing d/n Erk1 and p38 and driven by tetracycline-inducible promoter were used. The expression of d/n p38 was increased maximally to 50-fold by 1 μg/mL tetracycline. Such expression suppressed the formation of EGF-induced p-Erk1 by 36%. Similarly, after 4-hour exposure in the absence of tetracycline, d/n Erk1 expression increased by 50-fold over the basal level. Loss-of-function studies were performed and showed that when these cells were exposed to EGF, MAPK activities were not elicited (data not shown). 

With each of the d/n cell lines, crosstalk between the ERK and the p38 pathways was reevaluated. After the upregulation of d/n Erk1 or p38 expression, we determined whether EGF-induced increases in p-p38 and p-Erk1 formation were enhanced, respectively, over those measured in these two cell lines in the absence of tetracycline. Figure 3Ashows that EGF induced greater increases in p38 phosphorylation in the d/n Erk1 cell line with tetracycline than in the absence of tetracycline. After 15 minutes without tetracycline, EGF-induced a threefold increase in p38 phosphorylation that remained essentially unchanged for up to 90 minutes. In contrast, the initial increase at 5 minutes was 2.5-fold greater in the presence of tetracycline than in its absence. At longer intervals, up to 90 minutes, these increases waned slightly. Figure 3Bshows a trend similar to that described in Figure 3A . In d/n p38 cells, the EGF-induced increases in Erk1 phosphorylation were greater than those measured in the absence of tetracycline. After 5-minute exposure to EGF, the increase was twofold greater in the presence of tetracycline and was sustained for up to 90 minutes. The correspondence between the effects obtained with each of the chemical inhibitors on kinase activation and those measured in the d/n cell lines validated the view that crosstalk occurs between the ERK and p38 pathways. It is noteworthy that the kinetics of dephosphorylation of phospho-p38 in d/n Erk cells were prolonged compared with those for phosphor-Erk1/2 in d/n p38 cells. One possibility is that d/n p38 expression was leaky in the absence of tetracycline. However, no evidence was detected for such expression (data not shown). An untested alternative is that each of these MAPK pathways elicits control of a different cohort of protein phosphatases, in addition to MKP-1 and PP2A. Therefore, loss of Erk1 and p38 function may have differential dephosphorylating effects on each of these kinases. 

MKP-1 Knockdown Effect on EGF-Induced Erk1/2 and p38 Activation

To assess the effect of EGF-induced activation of MKP-1 (Fig. 4) , siRNA MKP-1 was used. Nontransfected RCECs showed a sevenfold increase in MKP-1 protein expression after 60-minute exposure to EGF (10 ng/mL), whereas transfected cells displayed reduced expression of MKP-1, similar to the level of expression displayed by the nontransfected counterpart in the absence of EGF stimulation. The efficiency of transfection was confirmed using the negative control and transfection reagent provided in the transfection kit. 

The role of MKP-1 in mediating crosstalk was then assessed by determining whether EGF induces changes in MKP-1 protein expression and ascertaining the effects of knockdown of MKP-1 protein expression on EGF-induced, time-dependent changes in Erk1/2 and p38 phosphorylation. Figures 5A and 5Bshow that EGF-induced stepwise increases in MKP-1 expression, which reached a maximum level at 60 minutes, were 14-fold greater than those measured in the absence of EGF. After 2 hours, a 50% decline was observed in the amount of MKP-1. Interestingly, such changes in MKP-1 expression are consistent with those shown in the corresponding Western blot analysis of phosphorylated Erk1/2 and p38 (Figs. 1 5C 5D) . In each case, a decline in the amount of phosphorylated Erk1/2 and p38 at a particular time interval was preceded by an increase in MKP-1 expression. Conversely, after MKP-1 knockdown, time-dependent increases in EGF-induced MKP-1 expression were smaller than those in nontransfected cells. EGF-induced increases in Erk1/2 and p38 phosphorylation associated with MKP-1 knockdown were more long-lasting than those in nontransfected cells (Figs. 5C 5D) . In addition, a physiologically irrelevant siRNA counterpart did not suppress MKP-1 protein expression, and the patterns of change in p-Erk1/2 and p-p38 phosphorylation were identical with those in nontransfected cells (data not shown). Therefore, EGF-induced changes in MKP-1 expression contribute to mediating crosstalk between the ERK and p38 pathways. 

Dependence of EGF-Induced MEK1/2 Activation on p38 and ERK Limb-Mediated Crosstalk

The level of ERK phosphorylation affected by feedback control was characterized by determining whether the maximum EGF-induced increases in Erk1/2 phosphorylation after inhibition of p38 activity could also be observed upstream of Erk1/2 at the level of MEK1/2. Figure 6Acompares MEK1/2 phosphorylation status after a 5-minute exposure to EGF with those obtained after exposure to SB203580, U0126, or okadaic acid. Phosphorylation status increase was maximal at 5 minutes and had the same time requirement as did Erk1/2 for maximal activation. In the presence of SB203580, MEK1/2 activation increased another 40% to a level that was higher than that in the absence of SB203580. U0126 caused MEK1/2 activation to decrease by 53%. In contrast, inhibition of PP2A activity with okadaic acid caused MEK1/2 activation to increase by 15%. This effect of okadaic acid was selective for prolonging MEK1/2 phosphorylation because, without EGF, MEK1/2 phosphorylation was unchanging regardless of the presence or absence of okadaic acid. These results are suggestive of a feedback control between the two pathways at a level upstream of Erk1/2. 

To validate such a domain for crosstalk control, EGF-induced MEK1/2 activation was compared in the d/n Erk1 and d/n p38 lines in the presence and absence of tetracycline. The results shown in Figure 6Bindicate that in the d/n Erk1 cell line, MEK1/2 phosphorylation was biphasic in the absence of tetracycline and reached a maximum level after 5-minute exposure to EGF that was followed by a 50% decline to a stable value. In contrast, at 15 minutes, its activation in the presence of tetracycline increased to a maximum value that was threefold higher than in the absence of tetracycline. This higher level of stimulation was essentially sustained for up to 90 minutes. Such a difference indicates that inhibition of Erk1/2 activation has an upstream feedback effect that allows MEK1/2 to maintain its active status. A similar assessment was performed with the d/n p38 cell line (Fig. 6C) . Figure 6Cshows that in the absence of tetracycline, a pattern of change in MEK1/2 similar to that in the d/n Erk1 cell line was observed. With tetracycline, MEK1/2 activation at 5 minutes was 60% greater than in the absence of tetracycline. However, this elevated level of activation waned after 90 minutes and fell to a level indistinguishable from the value measured after 5 minutes in the absence of tetracycline. These results suggest that phosphatase-mediated crosstalk between the ERK pathway and the p38 pathway was initially dependent on EGF-induced stimulation of p38 because, in d/n p38 cells, MEK1/2 activation with tetracycline was greater after d/n p38 induction. 

EGF-Induced Interaction between PP2A and Erk1/2

To determine whether interaction occurs between PP2A and Erk1/2, PP2A and Erk1/2 were immunoprecipitated from cell lysates using corresponding antibodies, and the associated proteins were cross-probed. PP2A was evident in the ERK immunoprecipitates (Fig. 7A) . Evidence for ERK interaction with PP2A was obtained because ERK was pulled-down with the PP2A antibody (Fig. 7B) . 

The effect of changes in crosstalk between the ERK and p38 pathways for the mitogenic response to EGF was determined in three different cell types: nontransfected cells, tetracycline-induced cells with d/n Erk1 expression, and tetracycline-induced cells with d/n p38 expression. Inhibition of the p38 pathway with SB203580 or MKP-1 activity with Ro 31–8220 did not increase the mitogenic response to EGF. Such negative effects suggest that an increase in ERK limb activity after inhibition of either p38 or MKP-1 activity does not affect the mitogenic response to EGF. Interestingly, the declines induced by PD98059 or d/n Erk1 were identical, suggesting that Erk1 activation and translocation are sufficient to stimulate proliferation. 

Crosstalk Effect on EGF-Induced Cell Migration

Our results indicate that EGF-induced crosstalk between the ERK and p38 pathways is PP-mediated. Because it is known that the p38 pathway elicits control of cell migration, we determined the effects of changes in crosstalk on the increases in cell migration induced by EGF. Such interaction can be changed by selective inhibition of each of the two pathways or by chemical inhibition of PP activity. Pathway inhibition was achieved by exposing the cells to PD98059 or SB203580 or using tetracycline-induced d/n Erk1 and d/n p38 cells. In these different groups of cells, we determined whether EGF-induced cell migration was augmented by this EGF. The role of EGF-induced increases in PP activity in mediating crosstalk was evaluated by determining, with the use of nontransfected RCECs, whether EGF-induced stimulation of migration was augmented during exposure to PP inhibitors. The inhibitors used were sodium orthovanadate (pan phosphatase), okadaic acid (PP2A), Ro 31–8220 (MKP-1), and okadaic acid in combination with Ro 31–8220. EGF induced a twofold increase in wound closure rate after 24-hour incubation compared with control cells (Figs. 8A 8B 8C) . Figure 8Ashows that the ERK pathway, inhibited with 10 μM PD98059, enhanced EGF-induced migration by 45%, whereas inhibition of p38 signaling with SB203580 completely suppressed wound closure. Similar results were observed in EGF-stimulated d/n Erk1 and d/n p38 cells, respectively (Fig. 8B) . These declines in migration are in accordance with the known role of p38 MAPK as a mediator of cell migration. Furthermore, wound closure was accelerated through inhibition of EGF-induced increases in PP activity obtained through exposure to 100 μM sodium orthovanadate. This is evident because this inhibitor enhanced the EGF-induced increase in migration by 85%. Two of the PPs that participate in mediating crosstalk are PP2A and MKP-1 because okadaic acid (10 nM) and Ro 31–8220 (10 μM) also enhanced this effect of EGF by 15% and 13%, respectively. Further evidence that each of these phosphatases is involved is indicated by the finding that, in combination, these inhibitors enhanced migration by larger amounts than those obtained with each of the inhibitors. Therefore, phosphatase-mediated crosstalk between the ERK and p38 pathways modulates EGF-induced control of cell migration. 

Discussion

The results of this study show that the EGF-induced patterns of MAPK activation and subsequent deactivation are modulated through ERK and p38 pathway-mediated increases in PP expression. Phosphatase-mediated crosstalk was demonstrated based on changes in the phosphorylation status of components of each of these pathways, changes caused by relatively selective phosphatase inhibitors and occurring after siRNA knockdown of MKP-1 expression. Prolonged and enhanced MAPK activation was observed, suggesting that the time dependence and strength of signaling along the ERK and p38 pathways are modulated through changes in PP expression. This interaction entails a negative feedback effect wherein changes in PP activity modulate EGF receptor control of cell migration. In other tissues for which MAPK signaling has been described, such feedback is elicited through increases in PP expression at either the transcriptional or the posttranscriptional level. ^{4 5 6 7 8 9 10 11 14} Given the rapidity of EGF-induced increases in MKP-1 protein expression, it appears that crosstalk between the two pathways in RCECs resulted from changes at the transcriptional or even the posttranscriptional level. 

That crosstalk is a direct result of MAPK pathway activation is supported by the observation that selective inhibition of either the p38 or the ERK pathway increased, with prolonged stimulation of the parallel pathway. Therefore, the patterns of EGF-induced stimulation of each of these pathways are dependent on the extent of crosstalk between them. Through chemical suppression of ERK pathway signaling and inhibition of phosphatase activity, EGF-induced stimulation of p38-mediated cell migration was augmented. Conversely, inhibition of p38 signaling did not have a reciprocal effect on cell proliferation. As previously suggested, the inability of HGF or KGF to augment stimulation of proliferation through the suppression of p38 signaling could be attributed to the fact that this response had already been maximally stimulated by these growth factors. ³ Such an inability to enhance the mitogenic response to EGF was not caused by drug selectivity because such effects on migration and cell proliferation were repeated in cell lines after tetracycline induction of either d/n Erk1 or d/n p38 expression. 

Figures 1A and 1Bprovide evidence suggesting that ERK pathway stimulation modulates p38 signaling: the maximum increase in Erk1/2 phosphorylation occurred after 5 minutes, whereas the maximum activation of the p38 branch occurred after 15 minutes. This time difference is consistent with the idea that activation precedes induction of a phosphatase-mediated negative feedback response. More direct evidence that ERK limb stimulation has a negative feedback effect on p38 stimulation is that p38 signaling increased after inhibition of ERK pathway signaling (Fig. 2B) . Furthermore, in cells induced to express d/n Erk1, EGF-induced p38 activation was 2.5-fold greater than without tetracycline treatment (Fig. 3A) . Conversely, evidence for p38-induced suppression of Erk1/2 is provided by the observation that during exposure to SB203580 or with tetracycline induction of d/n p38 expression, EGF-induced Erk1/2 phosphorylation increased similarly under both conditions (tetracycline treated and tetracycline untreated) (Fig. 3B) . The fact that there is a correspondence between these two increases validates the selectivity of SB203580 as a p38 inhibitor. 

One of the phosphatases previously identified as affecting crosstalk is MKP-1. ^{11 28} Initially, we found this phosphatase to be a candidate for eliciting feedback because EGF-induced a 14-fold increase in MKP-1 expression after 60 minutes (Figs. 5A 5B) . A role for MKP-1 in crosstalk is now identified based on the effects of either Ro 31–8220 or knockdown of its expression on EGF-induced stimulation of the p38 and ERK pathways. During exposure to either the broad-spectrum phosphatase inhibitor sodium orthovanadate or the relatively selective MKP-1 inhibitor Ro 31–8220, the maximum EGF-induced increases in p38 and Erk1/2 phosphorylation rose to levels that were threefold and twofold greater than those in the absence of either inhibitor (Fig. 2) . The involvement of MKP-1 in mediating crosstalk is further indicated by comparing time-dependent changes in the phosphorylation status of Erk1/2 and p38 after MKP-1 knockdown with those in nontransfected cells. The phosphorylation statuses of Erk1/2 and p38 were enhanced after MKP-1 siRNA knockdown, and dephosphorylation was more gradual than in nontransfected cells (Figs. 5C 5D) . PPs other than MKP-1 are participants in mediating crosstalk because the relatively selective PP2A inhibitor, okadaic acid (10 nM), had effects similar to those of MKP-1 on Erk1/2 and p38 phosphorylation status (Fig. 2) . The involvement of PP2A in affecting control of ERK pathway phosphorylation status is further indicated by the finding of a direct protein–protein interaction between PP2A and Erk1/2 (Fig. 7) . The contributions by MKP-1 and PP2A to crosstalk control have been reported in numerous other studies describing cytokine-induced MAPK signaling. ^{4 5 7 8 29}  

p38 signaling strength, through crosstalk control, modulates ERK pathway signaling. The existence of this link is substantiated by the finding that inhibition of p38 signaling increases at a given time during Erk1/2 phosphorylation (Fig. 2C) , suggesting that the site of crosstalk control occurs upstream of Erk1/2 in the ERK cascade. Accordingly, we determined whether the changes in MEK1/2 phosphorylation status mirrored those seen in Erk1/2 during exposure to EGF in the presence of an MAPK inhibitor or okadaic acid. EGF-induced MEK1/2 phosphorylation status was maximally enhanced at 5 minutes with SB203580, U0126, or okadaic acid (Fig. 6) . The findings obtained with d/n Erk1 and d/n p38 cell lines validate the selectivity of these inhibitory effects on MEK1/2 dephosphorylation (Figs. 6B 6C) . In each of these cell lines, MEK1/2 phosphorylation was greater and less transient than in the nontransfected counterparts. These findings have two implications. First, ERK limb phosphorylation induces expression of a phosphatase that is limiting at the level of MEK1/2 stimulation. Second, p38 pathway stimulation induces expression of a phosphatase with a similarly negative feedback effect on ERK pathway stimulation. Such negative feedback control at the level of MEK1/2 has also been described in human embryonal and parental fibroblasts. ¹⁰  

We considered the impact of MAPK crosstalk control on EGF receptor stimulation of cell proliferation and migration. Regarding the mitogenic response to EGF, neither inhibition of p38 signaling nor phosphatase activity affected EGF-induced increases in proliferation. This negative finding was reported for the same tissue during exposure to HGF or KGF. ³ Crosstalk control of migration is shown by the fact that in cells preincubated with PD98059, the rate of wound closure was 45% greater after EGF stimulation than in the control (Fig. 8A) . The role of p38 activation in mediating migration is further indicated by the fact that in the presence of the p38 antagonist SB203580, wound closure was completely inhibited. Further evidence for crosstalk between the ERK and p38 pathways is indicated by the findings that in d/n Erk1 cells, EGF-induced migration was enhanced by 50%, a value similar to the increase obtained with PD98059. The selectivity of SB203580 as an inhibitor of p38-induced increases in migration is shown by the finding that in d/n p38 cells, EGF failed to stimulate cell migration. This negative response is identical with that obtained in nontransfected cells during exposure to EGF. EGF-induced increases in PP activity do mediate crosstalk control of the p38 pathway because the inhibition of PP activity enhanced migratory responses to EGF. The augmentation obtained with sodium orthovanadate was greater than that obtained with okadaic acid, Ro 31–8220, or a combination of the two, suggesting that the PP assemblage includes PPs other than PP2A and MKP-1. 

In summary, crosstalk control between the ERK and p38 pathways of the MAPK cascade in RCECs is mediated by increases in PP expression. Based on the swiftness of the transient changes in the phosphorylation status of these two pathways, this negative feedback effect probably occurs at the transcriptional or posttranscriptional level. PPs contributing to this feedback control include PP2A and MKP-1. Such feedback limits the magnitude of EGF-induced increases in p38 activity and migration, whereas EGF-induced ERK pathway stimulation of proliferation is unaffected by crosstalk inhibition (Fig. 9) . 

 Supported by National Institutes of Health Grant EY047951.

 Submitted for publication June 13, 2006; revised July 15 and 31, 2006; accepted October 10, 2006.

 Disclosure: Z. Wang, None; H. Yang, None; S.D. Tachado, None; J.E. Capó-Aponte, None; V.N. Bildin, None; H. Koziel, None; P.S. Reinach, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Time-dependent changes in Erk1/2 and p38 MAPK phosphorylation induced by EGF. RCECs were serum starved for 24 hours at 80% to 90% confluence and stimulated with 10 ng/mL EGF for the time indicated. (A) Representative Western blot analysis of anti–phospho-Erk1/2 and (B) phospho-p38 antibodies. Equal loading of proteins was confirmed by reprobing the same blot with total Erk1/2 and p38 antibodies. (C) Summary of time-dependent changes in Erk1/2 and p38 phosphorylation status. Data represent the mean ± SEM of three independent experiments.

Figure 2.

 

Protein phosphatase–mediated crosstalk between Erk1/2 and p38 pathways. RCECs were serum starved for 24 hours and exposed with 10 μM SB203580 (SB), 10 μM PD98059 (PD), 100 μM sodium orthovanadate (SO), 10 μM Ro 31–8220 (Ro), or 10 nM okadaic acid (OA) for 30 minutes. Subsequently, the cells were treated for 30 minutes with 10 ng/mL EGF in the continuous presence of the same inhibitors. (A) Representative blot shows the status of MAP kinase (p-p38 and p-Erk1/2) using specific monoclonal antibodies against active forms and total Erk1/2. Summaries of (B) p38 and (C) Erk1/2 phosphorylation status are shown in (A). Data represent the mean ± SEM of three independent experiments.

Figure 3.

 

Tetracycline activates mutant gene expression in d/n p38 and d/n Erk1. RCECs were transfected with tetracycline (tet)-inducedpcDNA4/To/A-Erk1DN and tetracycline-induced pcDNA4/To/A-p38DN containing full-length cDNA encoding a silent Erk1 or p38 mutants cells. Serum-starved d/n Erk1 or d/n p38 cells were treated with or without 1 μg/mL tetracycline for 24 hours, then stimulated with 10 ng/mL EGF for the times indicated. Levels of activation were evaluated by Western blot analysis. (A) Time course of p-p38 activation induced by EGF in tetracycline-induced d/n Erk1 and its noninduced counterpart. (B) EGF-induced activation of p-Erk1 in tetracycline-induced d/n p38 and its noninduced counterpart. Actin levels were measured as loading controls. Data represent the mean ± SEM of three independent experiments.

Figure 4.

 

Selectivity of siRNA MKP-1 to suppress MKP-1 protein expression. Nontransfected RCECs were exposed to 10 ng/mL EGF for 60 minutes. RCECs were transfected with physiologically irrelevant commercially provided negative control MKP-1 siRNA (neg con siRNA), its transfection reagent (Rea), or siRNA MKP-1 for 72 hours, followed by 24-hour serum starvation. Subsequently, cells were stimulated with 10 ng/mL EGF for 60 minutes. Western blot analysis was then performed using an anti–MKP-1 antibody. Equal loading of proteins was confirmed by reprobing the same blot with anti–IgG antibody. Data represent the mean ± SEM of results in triplicate experiments.

Figure 5.

 

MKP-1 knockdown diminishes time-dependent changes in Erk1/2 and p38 activation induced by EGF. RCECs were transfected with MKP-1 siRNA for 72 hours, followed by 24-hour serum starvation. Subsequently, cells were stimulated with 10 ng/mL EGF for the times indicated. (A) Representative results of nontransfected and MKP-1 siRNA–transfected RCECs probed with anti–MKP-1, anti–phospho-ErK1/2, or anti–phospho-p38 antibody. Equal loading of proteins was confirmed by reprobing the same blot with anti–actin antibody. Summary of EGF-induced activation of (B) MKP-1, (C) p-Erk1/2, and (D) p-p38 displayed in (A). Data represent the mean ± SEM of three independent experiments.

Figure 6.

 

Dependence of EGF-induced MEK1/2 activation on Erk1/2 and p38 pathway stimulation. Nontransfected RCECs, d/n p38, and d/n Erk1 cells were serum starved for 24 hours. (A) Representative Western blot and summary of EGF-induced changes in p-MEK1/2 under indicated conditions. RCECs were exposed for 30 minutes to 10 μM SB203580 (SB), 10 nM okadaic acid (OA), or the ERK antagonist U0126 (10 μM). These treatments were followed by exposure to 10 ng/mL EGF in the presence of the inhibitor for 5 or 15 minutes. Equal loading of proteins was confirmed by reprobing the same blot with MEK1/2 antibody. Representative Western blot indicating p-MEK1/2 formation in (B) d/n Erk1 and (C) d/n p38 in the presence and absence of tetracycline (tet) (1 μg/mL). Both cell lines were exposed to EGF for 5, 15, 45, or 90 minutes. Data represent the mean ± SEM of three independent experiments.

Figure 7.

 

EGF-induced interaction between PP2A and Erk1/2. Whole cell lysates from RCECs were immunoprecipitated (IP) with mouse anti–IgG (negative control), PP2A, and Erk1/2. The obtained immunopellets were then probed with (A) anti–Erk1/2 antibody or (B) anti–PP2A antibody. Data shown are representative of three independent experiments.

Figure 8.

 

EGF induced cell migration. Scratch-wound assay was performed on nontransfected RCECs, tetracycline-induced d/n Erk, or d/n p38. Increases in wound closure induced by EGF alone and in the presence of other modulators were normalized to the level obtained in their absence (i.e., control). (A) Nontransfected cells were exposed to 10 μM PD98059 (PD) or 10 μM SB203580 (SB) for 30 minutes. They were then exposed to 10 ng/mL EGF for another 24 hours. (B) Rates of EGF-induced increases in cell migration measured in d/n Erk1 and their d/n p38 counterparts are shown relative to the control in the absence of EGF in nontransfected cells. (C) In nontransfected cells, EGF-induced migration is compared with those measured in the presence of EGF and with 10 μM SB203580, 10 μM PD98059, 100 μM sodium orthovanadate (SO), 10 μM Ro 31–8220 (Ro), or 10 nM okadaic acid (OA). PP inhibitors were added to the culture medium 30 minutes before EGF was added. Data represent the mean ± SEM of results in triplicate experiments.

Figure 9.

 

Diagram depicting phosphatase-mediated crosstalk control of ERK and p38 MAPK signaling in corneal epithelial cells. EGF-induced changes in Erk1/2 and p38 phosphorylation status are dependent on PP-mediated crosstalk (e.g., MKP-1, PP2A). This control modulates the magnitude of growth factor–induced increases in corneal epithelial cell migration and therefore wound healing. Broken arrows: undefined feedback site at or above MEK level. Okadaic acid (OA), Ro 31–8220 (Ro), and sodium orthovanadate (SO) act as PP2A, MKP-1, and pan phosphatase inhibitors, respectively.

The authors thank Kathryn Pokorny for editorial suggestions. They also thank Luo Lu and his staff for guidance in establishing the tetracycline-inducible cell lines.