Epidermal Growth Factor Induces Fibroblast Contractility and Motility via a Protein Kinase C δ-dependent Pathway

Myosin-based cell contractile force is considered to be a critical process in cell motility. However, for epidermal growth factor (EGF)-induced fibroblast migration, molecular links between EGF receptor (EGFR) activation and force generation have not been clarified. Herein, we demonstrate that EGF stimulation increases myosin light chain (MLC) phosphorylation, a marker for contractile force, concomitant with protein kinase C (PKC) activity in mouse fibroblasts expressing human EGFR constructs. Interestingly, PKCδ is the most strongly phosphorylated isoform, and the preferential PKCδ inhibitor rottlerin largely prevented EGF-induced phosphorylation of PKC substrates and MARCKS. The pathway through which EGFR activates PKCδ is suggested by the fact that the MEK-1 inhibitor U0126 and the phosphatidylinositol 3-kinase inhibitor LY294002 had no effect on PKCδ activation, whereas lack of PLCγ signaling resulted in delayed PKCδ activation. EGF-enhanced MLC phosphorylation was prevented by a specific MLC kinase inhibitor ML-7 and the PKC inhibitors chelerythrine chloride and rottlerin. Further indicating that PKCδ is required, a dominant-negative PKCδ construct or RNAi-mediated PKCδ depletion also prevented MLC phosphorylation. In the absence of PLC signaling, MLC phosphorylation and cell force generation were delayed similarly to PKCδ activation. All of the interventions that blocked PKCδ activation or MLC phosphorylation abrogated EGF-induced cell contractile force generation and motility. Our results suggest that PKCδ activation is responsible for a major part of EGF-induced fibroblast contractile force generation. Hence, we identify here a new pathway helping to govern cell motility, with PLC signaling playing a role in activation of PKCδ to promote the acute phase of EGF-induced MLC activation.

can be deconstructed into a series of orchestrated events: lamellipodial extension, formation of forward adhesions, exertion of contractile forces to pull the cell body forward, and detachment of the rear (1). While each process is required for net cell locomotion, it is not necessarily the case that signals downstream of receptor activation must concomitantly be involved in triggering all of the processes. Despite longstanding anecdotal indications, only recently have formal demonstrations emerged that signaling via EGFR actually elicits cell contractile force generation (2,3) along with the other biophysical processes (4 -6). Because of the central importance of growth factor-induced cell motility in physiological and pathological applications, such as organogenesis, wound repair, and tumor invasion, determination of key pathways involved in connecting EGFR activity to contractile force generation, as well as the other processes underlying motility, is a crucial undertaking.
Myosin motors operating on cytoskeletal actin filaments are presumed to be involved in growth factor-induced cell motility in manner similar to the roles they play in integrin-mediated migration (7). Myosin II is a prominent actor in this context. Myosin II is localized along with actin fibers in the protrusive anterior region and at posterior regions of motile cells, where it is thought to generate contractility, in organizing and breaking cell-substratum adhesion, and/or in reorganizing the actin cytoskeleton (8,9). A recent report finds EGF to induce myosin II heavy chain phosphorylation (at least indirectly), with implications for subcellular localization of the active motor and consequent chemotactic cell movement (10). Phosphorylation at a regulatory serine 19 of the 20-kDa myosin light chain (MLC) subunit of myosin II promotes cell contractility in a variety of cell types responding to diverse stimuli. In addition, serine 19-phosphorylated MLC is enriched near membrane protrusion and retraction areas in motile fibroblast (11). The extent of MLC phosphorylation is regulated not only by protein kinases, such as Ca 2ϩ /calmodulin-dependent MLC kinase (MLCK) and Ca 2ϩ /calmodulin-independent Rho-kinase, but also by myosin phosphatase (MLCP) (12)(13)(14). During haptotactic migration, mitogen-activated protein (MAP) kinase ERK has been shown to mediate MLC phosphorylation through MLCK (15). Thus, the final myosin II contractile force generation machinery is engaged during both chemotactic and haptotactic motility. However, key signaling pathways through which EGFR actuates myosin-based contractility have yet to be identified.
Possible links between EGFR and myosin-based contractility are suggested by a few prior reports. Although ERK activity is required for MLC phosphorylation in haptotactic cell migration (15), ERK activity promotes the extension of membrane protrusions rather than retraction (16). A fair inference from these findings is that transcellular contractility is not solely driven by an ERK "master switch." EGF-induced myosin heavy chain phosphorylation and localization requires a protein kinase C (PKC) intermediary (17). The PKC family of molecules is an attractive candidate for connecting EGFR-elicited signals to myosin-mediated force generation, which is implicated in the contraction of muscle and non-muscle cells (18). PKC consists of a family of at least 11 isoforms. Specific isoforms of PKC are activated by phospholipids, diacylglycerol (DAG) generated by phospholipase C (PLC) or phospholipase D (PLD) from phosphatidylinositol 4,5-bisphosphate (PIP 2 ), fatty acids generated by PLA 2 , and calcium, depending on isoforms. Based on their structural and biochemical properties these PKC isoforms can be divided into three major groups: (i) the classical PKC (cPKC; ␣, ␤I, ␤II, and ␥), which are activated by DAG and are Ca 2ϩdependent; (ii) the novel PKC (nPKC; ␦, ⑀, , , and ), which are activated by DAG but Ca 2ϩ -independent, and (iii) the atypical PKC (aPKC; and ), which do not respond to either DAG or calcium. Importantly, EGFR triggers PKC activity (19) at least in part downstream of phospholipase signaling (20). These data support the hypothesis that one or more PKC isoform, in the classical or novel groups, contributes to EGFRmediated cell contractility during motility.
A picture of how PKC isoforms are regulated, in addition to final activation by lipid-containing molecules, has recently emerged in which direct phosphorylations play a major role (21,22). cPKC and nPKC isoforms contain three conserved serine/ threonine phosphorylation motifs of serine or threonine residues in the catalytic domain; a threonine in the activation loop (Thr-505 in PKC␦) and serines in the hydrophobic (Ser-643 in PKC␦) and C-terminal (Ser-662 in PKC␦) regions (21). An upstream kinase, 3-phosphoinositide-dependent protein kinase-1 (PDK1), phosphorylates the activation loop (23,24), which is necessary for the catalytic activity of cPKC isoforms (25,26). The turn motif and hydrophobic sites then undergo autophosphorylation. The regulation of these sites in nPKC isoforms may mirror those of cPKC isoforms (27), though the existence of a heterologous upstream kinase has been inferred (28). These sequential phosphorylations render cPKC isoforms stable and ready for activation by DAG (29). However, PKC␦ differs from cPKC in its regulation by phosphorylation (22). A study suggests that phosphorylation of the activation loop Thr-505 is not essential for subsequent activation; the PKC␦-specific acidic Glu-500 may assume at least some of the role of threonine phosphorylation (30). On the other hand, phosphorylation at Thr-505 by PDK1 is thought to be required for the stability of the enzyme (24). The functional consequence of phosphorylation of other sites is also not settled, Li et al. (31) demonstrated that the mutation of Ser-643 markedly decreases PKC␦ activity but Stempka et al. (30) showed that the same mutation has no effect on PKC␦ catalytic function. Interestingly, the upstream PDK1 appears itself dependent on phospholipids, PIP3 in particular, for activation (32). In the present model, PI 3-kinase generates PIP3 after activation by receptors with tyrosine kinase, such as EGFR. These investigations, even with the uncertainties, point to many PKC isoforms as potential downstream effectors of growth factor receptor signaling.
In this present contribution, we show that EGFR signaling leads to MLC phosphorylation at serine 19 and consequently to cell contractile force generation and motility in fibroblasts. The pathway to myosin activation required activity of the novel PKC␦ isoform. Moreover, we find that there appear to be two distinct phases of PKC␦ activation, with PLC␥ signaling required only for the acute phase. We have thus identified a novel pathway connection downstream of EGFR activation to one of the crucial biophysical processes underlying cell migration.
Cell Culture-The establishment and maintenance of the NR6 WT, expressing full-length wild-type EGFR, or NR6 cЈ973, expressing signaling-restricted EGFR lacking all autophosphorylation motifs and fails to activate PLC␥, cell lines have been described (19,34,35). Briefly, these constructs were retrovirally transduced in NR6 cells, Swiss 3T3-derived fibroblasts, which lack endogenous EGFRs (35). Cells were cultured in minimum essential medium (MEM)-␣ with 7.5% fetal calf serum plus 2 mM L-glutamine, 1 mM sodium pyruvate, 0.1 mM MEM, nonessential amino acids, and the antibiotics penicillin (100 units/ml), streptomycin (100 g/ml), and G418 (350 g/ml) as the growth medium. Subconfluent cells were passaged with a 1:8 split ratio at 3-day intervals using 0.25% trypsin with 0.25 mM EDTA. Cells were quiesced using restricted or no serum conditions without G418 prior to experiments.
Plasmid Construction and Transfection-The dominant-negative (DN) PKC␦ construct (36) was a generous gift from Dr. Michael Simons (Dartmouth Medical School). The DN construct was generated by replacing the conserved lysine 376 in the ATP binding domain with tryptophan. For stable expression, DN PKC␦ cDNA was subcloned into pCEP4 (Invitrogen) hygromycin resistant vector downstream from a cytomegalovirus promoter (CMVp). CMVp was replaced with a mouse mammary tumor virus promoter (MMTVp) for inducible expression (20). The empty pCEP4 vector, CMVp of which was replaced with MMTVp, was used as control. The construct was stably transfected into NR6 WT cells by electroporation. Cells were trypsinized, pelleted, and resuspended in Opti-MEM medium (Invitrogen) in electroporation cuvette, and the plasmid was added to a total of 30 g. The cells were electroporated at 0.300 kV and 950 F (Gene Pulser, Bio-Rad). 48 h after electroporation, cells were selected in the growth medium containing hygromycin B (Calbiochem) (100 g/ml). Polyclonal cell lines consisting of more than 20 colonies were established. Two independent electroporations and stably transfected cell lines were established and tested. Dexamethasone (2 M for 24 h) was used to induce MMTVdriven DN PKC␦ expression at the same time with quiescence.
siRNA Transfections-siRNA duplexes (siRNAs) were synthesized and purified by IDT (Coralville, IA). The siRNA sequence for targeting PKC␦ (GenBank TM accession number NM_011103) was PKC␦ siRNA (5Ј-AGUACUUGGCAAAGGCAGCTT-3Ј). The siRNA sequence for targeting PKC␣ (GenBank TM accession number NM_011101) was PKC␣ siRNA (5Ј-ACAACCUGGACAGAGUGAATT-3Ј). GFPsiRNA (5Ј-GAC-CCGCGCCGAGGUGAAGTT-3Ј) was used as a negative control (37). Transfection of siRNAs was performed using the manufacturer's protocol for LipofectAMINE TM 2000 (Invitrogen). Briefly, 4 l of 20 M siRNA was mixed with 200 l of Opti-MEM. 4 l of LipofectAMINE TM 2000 was diluted in 200 l of Opti-MEM and incubated at room temperature for 5 min. After the incubation, the diluted LipofectAMINE TM 2000 was combined with the diluted siRNA and then incubated for an additional 20 min at room temperature. Total 400 l of siRNA-LipofectAMINE TM 2000 complexes was applied to each well of cultured NR6 WT fibroblasts at ϳ70% confluence in a 6-well plate.
Immunoblotting and Immunoprecipitation-Cells were grown to confluence in 6-well tissue culture plastic plates. After 24 h of quiescence in MEM␣ with 0.5% dialyzed fetal calf serum, cells were treated. Cells were lysed with sodium dodecyl sulfate (SDS)-sample buffer containing 0.1 M Tris-HCl, 4% SDS, 0.2% bromphenol blue, and 5% ␤-mercaptoethanol. Cell lysates were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Immobilon-P; Millipore, Bedford, MA). Blots were probed by primary antibodies before visualizing with alkaline phosphatase-conjugated or horseradish peroxidaseconjugated secondary antibodies (Promega, Madison, WI) followed by development with a colorimetric method (Promega) or an enhanced chemiluminescence kit (ECL TM ; Amersham Biosciences).
For immunoprecipitation, cells were prepared and treated as described above. Cells were lysed with radioimmune precipitation lysis buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 0.1% SDS, 1% Nonidet P-40, and 0.5% deoxycholate) containing 1 mM EDTA, 2 g/ml aprotinin, 2 g/ml leupeptin and 1 mM phenylmethylsulfonyl fluoride. Cell lysates were cleared by centrifugation at 13,000 ϫ g for 15 min. Soluble proteins were incubated with anti-PKC␦ antibody for 4 -6 h at 4°C. Immunocomplexes were incubated with protein G-agarose and centrifuged. The pellets were washed three times with radioimmune precipitation lysis buffer containing protease inhibitors. Precipitated proteins were subjected to analyze by immunoblotting as described above.
Motility Assay-EGF-induced migration was assessed by the ability of the cells to move into an acellular area (38). Cells were plated on 6-well plastic dishes and grown to confluence in MEM-␣ with 7.5% fetal bovine serum. After 24 h quiescence, an area was denuded by a rubber policeman. Cells were then treated with or without EGF (10 nM) and incubated at 37°C; inhibitors or diluent alone were added at the same time as EGF. Photographs were taken at 0 and 24 h, and the relative distance traveled by the cells was determined.
Isometric Force Measurement-The preparation of three-dimensional fibroblast populated collagen lattices (FPCL) has been described previously (2). Briefly, FPCL containing 7.5 ϫ 10 5 cells were prepared with modifications of Kolodney and Elson (39). A mixture of cells and 0.75 mg/ml type I collagen was poured into the annular space of cylindrical Teflon molds (1 ml/mold) and gelled at 37°C in 5% CO 2 for 1 h. Ring-shaped FPCL gels were removed from the molds and placed in tissue culture medium for ϳ40 h, maintaining the original diameter (15 mm) of the FPCL with sterile parallel spacing rods. This configuration established a tensed FPCL at the time of force measurement.
Contractile force measurements of the WT NR6 and c'973 fibroblasts were performed with an isometric force transduction system. The system uses force transducers (model 52-9545, Harvard Apparatus, South Natick, MA) and mounts one FPCL per transducer vertically. The mounted tissue was placed in an organ bath containing 50 ml HEPES- buffered, serum-free Dulbecco's MEM (37°C). Each FPCL was permitted to reach an equilibrium force over a period of 30 to 60 min prior to addition of 10 nM EGF. The contraction force response of the FPCL was monitored by computer data acquisition. The strain level of the FPCL was maintained throughout so as to measure both the initial contractile force and the persistence of contractile force as a function of time.
Gel Compaction Assay-NR6 WT fibroblasts were grown in floating collagen type I matrices using a modification of described methods (40). Cells were harvested from monolayer culture using 0.25% trypsin/ EDTA and resuspended in quiescent media (serum-free medium containing 1 mg/ml bovine serum albumin). Cells were incubated with inhibitors for 30 min prior to matrix preparation. Neutralized collagen solutions (1 mg/ml) containing 10 6 cells/ml, Ϯ EGF (1 g/ml) and inhibitors (or diluent alone) were dispensed into 24-well culture plates (0.5 ml solution/well). Collagen solutions were left to polymerize for 60 min at 37°C in a humidified incubator with 5% CO 2 and then each matrix was overlaid with 1 ml of quiescent medium Ϯ EGF (1 g/ml) and inhibitors. The matrices were gently released from the surface and sides of each well using a scalpel and incubated for 24 h. Compaction was determined by weighing the matrices after the incubation period. Data are shown as the percentage of the control matrix weight.
Cell viability was assessed following compaction by removal using collagenase digestion. Matrices were washed in phosphate-buffered saline and then incubated with 0.05% trypsin for 10 min followed by collagenase (0.25 mg/ml) for a further 10 min. Fetal bovine serum (20%) was added to quench the collagenase activity and cell viability was determined by trypan blue staining. Cell viability was not significantly affected (Ͻ10%) by ML-7 at 20 mM.
Statistical Analyses-All data are expressed as means Ϯ S.E. of separate experiments. Differences between means were determined by Student's t test for unpaired samples, and those at p Ͻ 0.05 were considered significant.

EGF Induces Phosphorylation of Myosin Light Chain-Ex-
posure of fibroblasts to EGF leads to cell contractility and motility (2,38); both of which are proposed to require actomyosin-based contraction (1). To verify this, we found that EGF leads to phosphorylation of MLC at the activation-specific site serine 19 (Fig. 1). This occurred rapidly in cells expressing the wild type EGFR, but dissipated by more than half over the ensuing 2 h see below). As expected, EGF-induced phosphorylation was blocked by the EGFR-selective inhibitor PD153035 and the MLCK-selective inhibitor ML-7. Similarly, EGFR-mediated compaction and motility was blunted by ML-7 (Fig. 2). While these data were anticipated, they establish directly for the first time that EGF leads to MLC phosphorylation and that motility depends on this contractile mechanism.
EGF Induces Activation of PKC with PKC␦ as a Major Isoform Activated-To determine the key pathway(s) from EGFR to MLC, we focused on PKC. EGF has been shown to induce PKC activities (19). We confirmed that EGF exposure of mouse fibroblasts leads to phosphorylation of PKC targets MARCKS and other target sites (Fig. 3). The EGFR kinase inhibitor PD153035 eliminated these phosphorylations (Fig. 3). Interestingly, PKC phosphorylation, a marker of auto-activation, is noted even at fractional K d levels of EGF (Fig. 3). As little as 0.01 nM EGF results in detectable levels of phospho-PKC bands at 80, 82, and 85 kDa in addition to the unstimulated band at 78 kDa. This pattern of higher molecular weight bands was similar to that noted upon PMA activation (data not shown). That these were due to PKC activation was verified by inhibition by the pan-PKC inhibitor chelerythrine chloride (Fig. 3). The phosphospecific antibody used detects the phosphorylated forms of PKC ␣, ␤I, ␤II, ␦, ⑀, , and .
To ascertain which of the PKC isoforms is activated, we employed phosphospecific antibodies. PKC␣/␤II was present but was not appreciably affected in the presence of EGF (Fig.  4). However, PKC␦ was strongly phosphorylated (Fig. 4); this was not entirely unexpected as such an activation has been suggested previously for fibroblasts (41). PKC␦ was found to be the major contributor to phosphorylation of PKC targets as the isoform-preferential inhibitor rottlerin blocked the majority of phosphorylation of MARCKS and other protein targets (Fig. 4), whereas the PKC␣/␤II inhibitor Gö 6976 had little effect (data not shown).
PLC␥ but Neither ERK nor PI3K Contributes to PKC␦ Activation-Multiple PKC isoforms are activated downstream of the generation of diacylglycerol (DAG) and calcium fluxes {these secondary to IP3-mediated release}. In particular, EGFR triggers PLC␥ to generate both DAG and IP3 resulting in PKC activities (19,20). Interestingly, EGFR activation in NR6 cells leads to only weak calcium fluxes (42,43); this is consistent with the mode of activation of PKC␦ regulation by allosteric lipids but independent of calcium. To probe this link between EGFR and PKC␦, we used both the PLC-specific inhibitor U73122 and NR6 cЈ973 cells that express the truncated cЈ973 EGFR, which fails to activate PLC␥, generate IP3, or mobilize calcium (42). Of interest, phosphorylation of PKC␦ was delayed in the cells expressing cЈ973 EGFR (Fig. 5). This is most probably due to the inability to activate PLC␥ as a similar delayed phosphorylation was noted when the WT EGFR-expressing cells were challenged in the presence of the PLC inhibitor U73122 (Fig. 5). The inactive congener U73343 had no effect on phosphorylation of PKC␦ or MLC (data not shown). This suggests that while PLC␥ is not strictly required for PKC␦ phosphorylation, it contributes to the early phase of activation, a situation not unprecedented (44).
The pathway to the later phase of PKC␦ activation is unknown at present. However, it is unlikely to involve either of the well-described ERK or PI 3-kinase pathways (Fig. 6). ERK activation was completely abrogated by the MEK inhibitor U0126 without any affect on PKC␦ phosphorylation. Similarly, even though EGF does not activate PI 3-kinase robustly in these fibroblasts, inhibitory concentrations of LY294002 had no impact on EGF-induced PKC␦ phosphorylation.
PKC␦ Is Required for EGF-induced Phosphorylation of MLC-Significance of the findings above is buttressed if PKC signaling leads to actomyosin effects. EGF-induced phosphorylation of myosin light chain was completely abrogated not only by the pan-PKC inhibitor chelerythrine, but the PKC␦ isoformpreferential inhibitor rottlerin (Fig. 6). That this loss of MLC phosphorylation did not arise from pancellular inhibitory effects was shown by ERK phosphorylation being maintained even at the highest rottlerin concentrations. Of course, no single inhibitor is truly specific, and rottlerin also uncouples mitochondrial ATP production (45). Thus, an alternative mode of signal abrogation of the PKC␦ isoform was required. Expression of a dominant-negative PKC␦ construct (36) in the NR6 WT cells completely prevented phosphorylation of both PKC␦ and MLC (Fig. 7). Similarly, PKC␦ gene silencing with PKC␦siRNA, a more specific mode of inhibition, abolished EGF-induced MLC phosphorylation, while control transfections with GFPsiRNA or PKC␣siRNA had no effect on this signaling pathway (Fig. 7).
If PKC␦ is a critical signaling intermediary to MLC activation, then interventions should show similar effects for phosphorylation of MLC as they did for PKC␦ (Fig. 6). Again, U0126 and LY294002 did not diminish EGF-induced phosphorylation of MLC (Fig. 6). Interestingly, use of the PLC inhibitor U73122 and stimulation of cells expressing the truncated cЈ973 EGFR also displayed delayed phosphorylation of MLC (Fig. 8). These profiles are strikingly similar to the effects on PKC␦ phosphorylation, further strengthening the upstream position of PKC␦ in MLC activation. PKC␦ Is Required for EGF-induced Cell Motility and Contractility-Contractility is considered necessary for motility to pull the cell body forward (1); this was confirmed in our system as ML-7 prevented this EGF-induced motility (Fig. 2). As such if PKC␦ is the key intermediary in this cascade, then its inhibition should also block motility. All approaches for inhibiting PKC␦ signaling pathway, the pharmacologic agent rottlerin, the molecular dominant-negative PKC␦, and PKC␦siRNA, abrogated EGF-induced motility (Fig. 9). Interestingly, inhibition of PKC␦ signaling, decreased even basal motility. This is not unexpected, similarly to the decrement in basal motility noted in the face of ML-7 (Fig. 2), as haptokinesis also requires MLC kinase activity (15). Still, that the PKC␦siRNA abrogation did not reduce basal motility, may suggest either a lesser decrement in signal reduction using this approach (as noted by some EGF-induced motility retained) or non-PKC␦-specific effects of rottlerin and the DN PKC␦. Compaction of floating collagen gels was similarly blocked by rottlerin (Fig. 10). Both modes of disrupting this putative signaling cascade resulted in significantly reduced gel compaction.
We found a two-stage MLC phosphorylation effect above, with PLC␥ contributing to acute signaling, but later signaling was PLC-independent. While the time-scale integration of motility would not allow us to discern this 30 -60 min delay (46), we could parse this by isometric gel contraction (2). We found cells expressing the cЈ973 EGFR that fails to activate PLC␥ did induce robust contraction but that this response was delayed compared with cells expressing WT EGFR (Fig. 10). EGF induced peak contractility at 0.38 Ϯ 0.10 h in cells expressing the WT receptor, compared with 1.98 Ϯ 0.10 h in those expressing cЈ973 EGFR (n ϭ 5; p Ͻ 0.001); however, the peak force was similar between the cell types (25.1 Ϯ 1.3 dynes in WT versus 24.1 Ϯ 3.2 in cЈ973 NR6 cells). Thus, the cell responses of motility and contraction correlate directly to at least two independent pathways leading from EGFR to PKC␦ and then to myosin-based contraction. DISCUSSION Cell motility results from a highly orchestrated set of events triggered by external signals (1, 7). While the crucial cell processes underlying migration identified by deconstructive analyses, protrusion, de novo adhesion, contraction, and rear release, are all involved in productive motility, only experimental evidence can determine which operate as active governors and which as permissive functions for the various modes of motility induction. This distinction is critical to the modulation of motility, especially for therapeutic purposes, because active governors should proffer for a more targeted intervention. In this report, we tested the hypothesis that active PKC signaling is a key regulatory element in EGF-induced contractility during fibroblast motility. We provide evidence that EGF induces serine 19 phosphorylation of MLC and that this lies downstream of the novel PKC isoform, PKC␦. Molecular or pharmacologic inhibition of PKC␦ blocks subsequent MLC phosphorylation, contractility and motility. Furthermore, we also demonstrate that PLC␥ is responsible for a major portion, the acute phase, of the EGF-induced fibroblast contractility. These data provide for a novel target for intervention to regulate growth factorinduced motility.
Orchestration of cell migration requires that multiple biochemical pathways be coordinated. Certain end effectors for the crucial biophysical processes have been identified recently. The small GTPases of the Rho family and actin binding proteins control the actin cytoskeleton reorganization required for protrusion at the front (47)(48)(49)(50)(51). Some of these same molecular switches regulate the formation of new adhesions (52,53). At the other end of the cell, the intracellular limited protease calpain dictates tail retraction (4, 54 -57), particularly on moderately to highly adhesive substrata (58). These end effectors appear common between motility signaled by adhesion receptors and that triggered by growth factor receptors. However, the signaling pathways that get to the end effectors appear distinct. EGFR and other growth factor receptors require PLC␥ signaling to promote cytoskeletal reorganization at the front (34,48,59,60), while haptotaxis occurs in its absence (34,61). Rear de-adhesion triggered by EGF, IGF-1, or PDGF is limited when ERK or m-calpain is inhibited (3,55), while basal haptokinesis is limited by interference with the -calpain isoform (55,62). Further, during haptokinesis, ERK signaling has been linked to cell contractility, leading to MLCK activation and phosphorylation of serine19 on MLC (15). Herein, we report that inhibition of ERK blocks neither activation of PKC␦ nor phosphorylation of MLC. This is consistent with our earlier finding that ERK inhibition prevented fibroblast detachment and channeled the cell contractility from motility to compaction of the matrix (2). Rather, the same end effect of myosin II contraction is achieved via a PKC␦-dependent pathway.
Association of particular signaling pathways with individual biophysical processes that are distinct from those for haptokinesis and diverge at the immediate postreceptor step carries implications for the regulation of motility and for targeted interventions. Divergence of signaling to separate biophysical processes of motility clearly provides for the possibility that these signals and processes can generate active locomotion under some circumstances but alternative behaviors under others. For instance contractility in the absence of rear de- infection. Separating the key switches between haptokinesis and chemokinesis in response to growth factors suggests that these two modes of motility accomplish fundamentally distinct functions. While these concepts are speculative at present, the existence of process, and induction mode, specific molecular switches enables us to intervene specifically both in discovery and, potentially, in therapies aimed at angiogenesis and tumor invasion.
The definition of this new pathway from EGFR to cell contractility implicating PKC␦ appears at odds with an earlier report. In vitro, using purified enzymes and substrates, PKC was shown to phosphorylate MLC at serine 1, serine 2, and threonine 9, leading to inhibition of MLC function (63). Possible reasons for this discrepancy lie in the different approaches; we are examining in vivo, and likely indirect effects of PKC␦ on MLC whereas the earlier work was in vitro and direct. The recent discovery that PKC phosphorylates the CPI-17 phosphoprotein, a potent inhibitor of MLC phosphatase (64), provides both an explanation for this discrepancy and a potential pathway from PKC␦ to MLC phosphorylation. However, whether this connection is operative in our cells and in wound fibroblasts and is the only connection between PKC␦ and myosinbased contractility, or whether there is a parallel analogous pathway as CPI-17 may not be present in fibroblasts (65), is actively under study but lies beyond the scope of the present communication.

FIG. 9. PKC␦ inhibition abrogates EGF-induced cell motility.
A, cell motility assay was performed with NR6 WT cells in the absence or presence of EGF (10 nM) in the absence or presence of rottlerin (3 M) and Gö 6976 (100 nM), respectively, as described. Cell motility was calculated as -fold increase over basal traveling distance of nontreated cells. B, dexamethasone-inducible DN PKC␦-transfected NR6 WT cells (NR6 WT-DN PKC␦) and the control vector-transfected NR6 WT cells (NR6 WT-Mock) were examined for EGF-induced motility assay. Cells were quiesced, with or without 2 M dexamethasone (Induction) for 24 h. Then, cell motility assay was performed in the presence or absence of EGF (10 nM) as described. Cell motility was calculated as -fold increase over basal traveling distance of nontreated and not DN PKC␦ induced NR6 WT-DN PKC␦ cells. C, RNA interference with siRNA transfection was performed as described. 24 h after transfection, cells were quiesced for 24 h prior to performing a cell motility assay. Cell motility was calculated as -fold increase over basal traveling distance of nontreated cells. The data in the graph are the mean Ϯ S.E. of at least two independent experiments, with each experiment was performed in triplicate. Statistical analysis was performed by Student's t test: **, p Ͻ 0.01.

FIG. 8. Lack of PLC signaling correlates with delayed phosphorylation of MLC.
A, NR6 WT cells and NR6 cЈ973 cells, which fails to activate PLC, were grown to confluence and quiesced for 24 h. Cells were treated with EGF (10 nM) in the absence or presence of 1 M U73122, PLC inhibitor, for 0, 10, 30, 60, 120, and 240 min. Cells were then lysed, and equal volumes of proteins were separated by 15% SDS-PAGE and immunoblotted with anti-phospho-MLC followed by stripping and then reprobed with anti-MLC (total). Shown are representative blots of at least three repeats at all data points. B, phosphorylated MLC levels were enumerated by densitometry (NIH Image) and calculated as a ratio of total MLC. The data in the graph are the mean Ϯ S.E. of three independent experiments. Statistical analysis was performed by Student's t test: *, p Ͻ 0.05; **, p Ͻ 0.01.
A second superficially conflicting aspect might be from earlier reports on PKC activation limiting motility. Strong activation of PKC by PMA has long been known to variously affect cell adhesion and limit growth factor-induced cell motility (66 -68). Conversely, inhibition of PKC has been shown to inhibit cell motility and invasiveness (69 -71). In contrast, we link activation of PKC␦ to increased contractility and subsequent motility; of specific interest is that PKC␦ activity comprises the majority of detectable EGFR-triggered PKC activity in these cells; gene expression analyses of these cells detected appreciable levels of only PKC␣ and PKC␦ (data not shown). However, again we propose that the seeming discrepancy is related to the nature and mode of signaling. In these earlier studies the PMA activation was supraphysiologic resulting in high level activation and subsequent down-regulation of classical and novel PKC isoforms. As PKC is able to phosphorylate numerous cytoskeletal and focal adhesion proteins, it is not surprising that pharmacologic and tonic activation of PKCs have pleiotro-pic effects on cell adhesion and thus motility. In fact, we find that pharmacologic levels of PMA and other phorbol esters abrogate all cell motility in NR6 fibroblasts (data not shown). Herein, however, we activated PKC␦ via a physiological mechanism subject to feedback attenuation (19,20). Thus, the PKC activity is likely more regulated in a physiologically relevant manner.
It may not necessarily surprise that the operative PKC isoform is the novel PKC␦. While at least the acute phase of PKC activity is downstream of PLC␥ signaling, PLC signaling is required for the total activation of PKC as demonstrated in the presence of the pan-PLC inhibitor U73122 or downstream of cЈ973 EGFR, which fails to activate PLC (34). Furthermore, even though EGFR strongly activates PLC␥, calcium mobilization is weak in the NR6 cells (42). More recently, we have found that an EGFinduced calcium flux can be detected in only 20% of the cells exposed (43), while phospho-PLC␥ is found and IP3 is generated in practically all cells (47,48). Thus, a calcium-independent isoform is more likely to be activated in these cells downstream of PLC␥. Intriguing, the EGF-induced PKC␦ phosphorylation at Ser-643 was not inhibited by the PI3K inhibitor LY294002, although PI3K regulates PDK1. Le Good et al. (23) reported that LY294002 blocked serum-stimulated PKC␦ phosphorylation at Thr-505, with Ser-643 phosphorylation as subsequent autophosphorylation (22). However recently, Sonnenburg et al. (72) have shown that PDK1 may phosphorylate cPKC by a PI3K-independent mechanism (72). This other pathway might also control EGFinduced PKC␦ phosphorylation in our cells; as the multiple stimulatory molecules in serum are distinct from EGFR ligands, discrepancies in upstream signaling pathways between serum and EGF are not unexpected (6). Regardless, the mechanism by which the late phase of PKC␦ activation occurs remains to be determined in subsequent studies.
In summary, our results altogether indicate that PKC␦-mediated MLC phosphorylation, presumably through MLCK, plays a pivotal role in growth factor-induced motility by promoting cell contractility. Our findings now open further questions such as what is the pathway between PKC␦ and MLCK activation, and how EGFR activates PKC␦ in the late phase. Therefore, further investigations are needed to clarify this connection.
FIG. 10. PKC␦ regulates EGF-induced cell contractility. A, cell compaction assay was performed with NR6 WT cells in the absence or presence of EGF (1 g/ml) in the absence or presence of rottlerin (500 nM) as described. EGF-induced compaction of fibroblast populated collagen lattices was blocked by abrogation of PKC␦ signaling. Rottlerin limited the ability of EGF to induce the resident fibroblasts to compact the gel and extrude water, thus resulting in a less heavy gel. The data in the graph are the mean Ϯ S.E. of two experiments each with nine repeats. Statistical analysis was performed by Student's t test: **, p Ͻ 0.01. B, isometric contraction of fibroblast populated collagen lattices was assessed in response to saturating concentrations of EGF (100 nM). The lattices were populated by NR6 cells expressing either WT or cЈ973 EGFR. The lattices populated by the NR6 cЈ973, which does not activate PLC␥, present similar peak contractile force but this is delayed in comparison to the lattices with NR6 WT cells. Shown is a representative of five experiments.