ERK and RhoA differentially regulate pseudopodia growth and retraction during chemotaxis.

Nonmotile cells extend and retract pseudopodia-like structures in a random manner, whereas motile cells establish a single dominant pseudopodium in the direction of movement. This is a critical step necessary for cell migration and occurs prior to cell body translocation, yet little is known about how this process is regulated. Here we show that myosin II light chain (MLC) phosphorylation at its regulatory serine 19 is elevated in growing and retracting pseudopodia. MLC phosphorylation in the extending pseudopodium was associated with strong and persistent amplification of extracellular-regulated signal kinase (ERK) and MLC kinase activity, which specifically localized to the leading pseudopodium. Interestingly, inhibition of ERK or MLC kinase activity prevented MLC phosphorylation and pseudopodia extension but not retraction. In contrast, inhibition of RhoA activity specifically decreased pseudopodia retraction but not extension. Importantly, inhibition of RhoA activity specifically blocked MLC phosphorylation associated with retracting pseudopodia. Inhibition of either ERK or RhoA signals prevents chemotaxis, indicating that both pathways contribute to the establishment of cell polarity and migration. Together, these findings demonstrate that ERK and RhoA are distinct pathways that control pseudopodia extension and retraction, respectively, through differential modulation of MLC phosphorylation and contractile processes.

Nonmotile cells extend and retract pseudopodia-like structures in a random manner, whereas motile cells establish a single dominant pseudopodium in the direction of movement. This is a critical step necessary for cell migration and occurs prior to cell body translocation, yet little is known about how this process is regulated. Here we show that myosin II light chain (MLC) phosphorylation at its regulatory serine 19 is elevated in growing and retracting pseudopodia. MLC phosphorylation in the extending pseudopodium was associated with strong and persistent amplification of extracellular-regulated signal kinase (ERK) and MLC kinase activity, which specifically localized to the leading pseudopodium. Interestingly, inhibition of ERK or MLC kinase activity prevented MLC phosphorylation and pseudopodia extension but not retraction. In contrast, inhibition of RhoA activity specifically decreased pseudopodia retraction but not extension. Importantly, inhibition of RhoA activity specifically blocked MLC phosphorylation associated with retracting pseudopodia. Inhibition of either ERK or RhoA signals prevents chemotaxis, indicating that both pathways contribute to the establishment of cell polarity and migration. Together, these findings demonstrate that ERK and RhoA are distinct pathways that control pseudopodia extension and retraction, respectively, through differential modulation of MLC phosphorylation and contractile processes.
Directional cell migration, or chemotaxis, is initiated when cells sense the direction and proximity of a chemoattractant gradient (1)(2)(3). Although significant progress has been made in linking various signals to the general process of chemotaxis, little is known about how these signals facilitate morphological polarization, which is characterized by establishment of a leading pseudopodium (lamellipodium), cell body, and trailing tail region (1,4). Recent evidence indicates that when cells encounter a directional cue such as a chemoattractant gradient, they respond by local activation and amplification of signals on the side facing the stimulus (1)(2)(3)5). For example, pleckstrin homology domain-containing proteins strongly and persistently localize to the plasma membrane facing the gradient (2,3), indicating that intracellular signals are spatio-temporally organized within chemotactic cells. This suggests that cells sense gradients and establish morphological polarity through the formation of an intracellular gradient of signals.
Although it is not well defined, signals organized on the gradient side of cells presumably facilitate localized actin polymerization leading to membrane protrusion in the direction of the chemoattractant (6). The extension of a single dominant leading pseudopodium marks the first sign of morphological polarity prior to cell movement (1,4). Despite its inherent importance to chemotaxis, little is known about gradient-sensing mechanisms and how these responses are transmitted to the actin-myosin cytoskeleton to achieve cell polarity and pseudopodia extension.
Extracellular signal-regulated kinase (ERK) 1 and RhoA are key signals that have been linked to cell migration and chemotaxis. However, their precise roles in mediating cell polarity and their spatio-temporal organization within the cell are still undefined. Recent evidence has shown that ERK mediates cell migration via regulation of myosin light chain kinase (MLCK) and myosin light chain (MLC) phosphorylation (7). Interestingly, MLCK activity and MLC phosphorylation have been shown to localize to the front of migrating cells, suggesting a potential role for ERK and MLCK in mediating contractile processes in the pseudopodium (8,9). MLCK controls contractility through phosphorylation of the regulatory serine 19 on MLC. This phosphorylation event facilitates assembly of a fully functional actin-myosin motor unit capable of generating tension and contractile forces (10 -13). However, RhoA also regulates MLC phosphorylation and contractility through inhibition of myosin phosphatase activity (14,15). Although Rho, ERK, and MLCK are necessary for proper migration, it is not known whether these signals separately or coordinately regulate this process. Moreover, the spatio-temporal organization of these signals in chemotactic cells is not known.
We recently reported a novel method of purifying differentially the pseudopodium and cell body of cells morphologically polarized toward a chemoattractant gradient (16). Using this model, we show here that ERK and RhoA are two separate pathways that differentially regulate myosin function to control pseudopodial dynamics leading to cell polarity and chemotaxis.

EXPERIMENTAL PROCEDURES
Antibodies and Reagents-Phosphospecific antibody to ERK1 and ERK2 (pTpY 185/187 ) was from BIOSOURCE International. Antibodies to ERK2, RhoA, and MEK1 were from Santa Cruz Biotechnology, Inc. Antibody to MLC was kindly provided by Dr. Primal de Lanerolle (University of Illinois at Chicago) or was purchased commercially from Sigma. Phosphospecific antibody to MLC-19 was kindly provided by Dr. Matsumura (9) (Rutgers University). Goat anti-rabbit and goat antimouse horseradish peroxidase antibodies were from Bio-Rad. Rat tail collagen I was from Upstate Biotechnology, Inc., and human fibronectin was from Collaborative Scientific. The MEK1/2 inhibitor PD98059 was obtained from New England BioLabs. LPA was purchased from Sigma. Insulin was purchased from Roche Molecular Biochemicals. Pertussis toxin (PTX), the EGF receptor inhibitor AG1478, the myosin light chain kinase inhibitor ML-7, and the Rho kinase inhibitor Y-27632 were purchased from Calbiochem. Botulinum C3 exoenzyme (C3 toxin) was purchased from Upstate Biotechnology, Inc.
Cell Culture and Cell Transfection-NIH 3T3 fibroblasts were kindly provided by Dr. Tony Hunter (Salk Institute, La Jolla, CA). COS-7 cells were from American Tissue Type Culture Collection. In most instances COS-7 cells were utilized; however, NIH 3T3 fibroblasts were also tested in some cases, and similar results were obtained. Cells were maintained in Dulbecco's modified Eagle's medium (Irvine Scientific) containing 10% fetal bovine serum (Gemini BioProducts), 200 mM Lglutamine, and 50 g ml Ϫ1 gentamicin (Sigma), and 100 mM sodium pyruvate (Sigma). Cells were incubated at 37°C with 5% CO 2 . For transfection experiments, 100-mm dishes of 60 -80% confluent cells were transfected using LipofectAMINE (Invitrogen) according to the manufacturer's protocol. Briefly, 20 l of LipofectAMINE was preincubated with 3.5 g of total DNA in 1 ml of transfection medium for 45 min. The volume was brought to 6 ml and layered over cells for 6 -8 h at 37°C. Cells were used for the appropriate assays within 48 h subsequent to serum starvation overnight. The HA-tagged kinase-dead Raf (HA-Raf-KD) and wild type Raf (HA-Raf-WT) constructs were provided by Dr. Michael Karin (University of California, San Diego). The constitutively active MEK1 mutant (MEKϩ) was provided by Dr. Christopher J. Marshall (Chester Beatty Laboratories, Institute of Cancer Research). The HA-tagged kinase dead mutant of MLCK (pCMV5-dnMLCK) was provided by Dr. Patricia Gallagher (Indiana University School of Medicine). The dominant negative Rho mutant (HA-tagged RhoN19) and wild type Rho construct were kindly provided by Dr. Martin Schwartz (The Scripps Research Institute). pcDNA3.1 (Invitrogen) was used as the empty vector to keep the total amount of DNA to 3.5 g. Cells used in chemotaxis assays were cotransfected with 0.5 g of a reporter construct encoding ␤-galactosidase (pCMV-SPORT-␤-galactosidase; Invitrogen). Cells analyzed by fluorescence staining were cotransfected with 0.5 g of green fluorescent protein reporter construct (pEGFP-C1, Clontech) as a cell transfection marker.
Chemotaxis Assays-Migration assays were performed using Boyden chambers containing polycarbonate membranes (tissue culture-treated 6.5-mm diameter, 10-m thickness, 8.0-m pores, Transwell; Costar) or QCM kit (Chemicon International) as described previously (16). The number of migratory cells/microscopic field (ϫ200) on the underside of the membrane was counted as described under "Experimental Procedures." Cells on a dish were treated with PTX as described above and stimulated with 100 ng/ml LPA for 10 min. Lysates were then immunoblotted using either phosphospecific ERK1/2 or anti-ERK2 antibodies as indicated. Activated Rho (Rho GTP) was determined using the Rhotekin binding domain, which selectively binds Rho GTP. Total Rho protein was determined by immunoblotting using an anti-RhoA antibody. B, an aliquot of cells treated as described in A was allowed to attach to collagen-coated wells for 3 h and then fixed and stained with crystal violet. The number of adherent cells was counted as described above in the migration experiment. C, cells expressing a dominant mutant of Rho (RhoN19) (ϩ) or the mock empty vector (Ϫ) were allowed to migrate for 3 h as described in A in the absence (Ϫ) or presence (ϩ) of 100 ng/ml LPA in the bottom compartment. Transfected cells on a dish were treated with 100 ng/ml LPA for 10 min. Lysates were harvested and immunoblotted for phospho-ERK (ERK1/2) and total ERK (ERK2 loading control) as described above. D, an aliquot of cells treated as described in C was allowed to attach to collagen-coated wells for 3 h and then fixed and stained with crystal violet as described in B. Results shown reflect the mean Ϯ S.D. (error bars) of three replicate experiments.
Membranes were coated on both sides with either 5 g/ml human fibronectin or 5 g/ml collagen I for 2 h at 37°C. 150,000 cells were allowed to attach to the upper membrane surface for 2 h and then stimulated to migrate toward 100 ng/ml LPA in the lower chamber for 3 h. After migration, cells were fixed and stained with crystal violet (Sigma), or transfected cells were fixed in ␤-galactosidase fixative and stained using 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside as a substrate according to the manufacturer's recommendations (Promega Corp.).
Cell Adhesion/Attachment Assays-Collagen-coated cell culture wells were prepared by incubating 5 g/ml collagen I in 24-well cell culture plates (Costar) for 2 h at 37°C. The wells were then blocked with bovine serum albumin for 30 min. Cells were allowed to adhere for the indicated times at 37°C. To assess cellular adhesion, the wells were washed with phosphate-buffered saline and then fixed with crystal violet in 2% ethanol, and cells on the underside were counted directly or were eluted with 10% acetic acid and measured in an enzyme-linked immunosorbent assay plate reader (A 600 ).
Quantitative Pseudopodia Assay-Pseudopodia extension and retraction were monitored using a pseudopodia assay kit (ECM 650; Chemicon International) as described previously (16). Briefly, 150,000 serumstarved cells were placed in the upper chamber of a 3.0-m porous polycarbonate membrane and coated on both sides with the appropriate ECM protein (5 g/ml fibronectin for NIH 3T3 cells and 5 g/ml collagen type I for COS-7 cells). Cells were allowed to attach and spread on the upper surface of the membrane for 2 h, and then stimulated with LPA, insulin, or buffer only, which was placed in the lower chamber to establish a gradient or placed in the upper and lower chamber to form a uniform concentration. Cells were allowed to extend pseudopodia through the pores toward the direction of the gradient for various times. To initiate pseudopodia retraction, the chemoattractant was removed, or an equivalent amount of chemoattractant was placed in the upper chamber to create a uniform concentration. The cell body from the upper surface was removed manually with a cotton swab, and the total pseudopodia protein on the underside was determined using BCA and a microprotein assay system and measured in an enzyme-linked immunosorbent assay plate reader at A 562 (Pierce Chemical Co.). In some instances, pseudopodia were fixed and stained with crystal violet in 2% ethanol and were eluted with 10% acetic acid and measured in an enzyme-linked immunosorbent assay plate reader (A 600 ).
Purification of Pseudopodia-To isolate proteins from growing pseudopodia specifically, 1.5 ϫ 10 6 cells were induced to form pseudopodia toward LPA for 60 min as described above or not treated and then harvested using a pseudopodia isolation kit (ECM 660; Chemicon International) as described previously. Briefly, cells were rinsed in excess cold phosphate-buffered saline and either rapidly fixed in 100% ice-cold methanol (for immunoblotting of whole cell lysates) or not fixed (for Rho GTPase activity only). Cell bodies were removed manually with a cotton swab from the upper membrane surface and pseudopodia from the undersurface scraped into lysis buffer (100 mM Tris, pH 7.4, 5 mM EDTA, 150 mM NaCl, 1 mM sodium orthovanadate, protease inhibitors (mixture tablet; Roche Molecular Biochemicals Corp.)) containing the appropriate detergent for immunoblotting of whole cell lysates (1% SDS) or Rho GTPase activity assay (Triton X-100 according to the manufacturer's recommendation; Upstate Biotechnology, Inc.). Cell bodies were purified in a similar manner except that pseudopodia on the undersurface were removed, and the cell body on the upper surface was scraped into lysis buffer and detergent. Retracting pseudopodia were induced for various times by removing or placing a chemoattractant in the upper chamber to create uniform concentration and then harvested as described above.
Haptotaxis Pseudopodia Purification-For haptotaxis pseudopodia, Transwell Boyden chambers (3.0-m pores) containing polycarbonate membranes were coated with 10 g/ml collagen I on the underside of the membrane only for 2 h at 37°C. 150,000 cells were added to the top of each chamber and allowed to extend pseudopodia to the underside for the indicated times, and lysates were harvested as described above.
Immunoblotting and Rho GTPase Activity Assay-Cell bodies, growing or retracting pseudopodia were purified as described above in the appropriate lysis buffer and then boiled for 10 min (for immunoblotting) or placed on ice for 30 min (for Rho GTPase activity assay). Lysates on ice were clarified by centrifugation for 10 min at 4°C and protein concentration of both lysates determined using BCA and a microprotein assay system (Pierce Chemical Co.). Briefly, 10 -20 g of whole cell lysate was separated by one-dimensional SDS-PAGE and immunoblotted according to standard protocols. Horseradish peroxidase-conjugated goat anti-mouse or goat anti-rabbit secondary antibodies and the enhanced chemiluminescence detection system (Pierce Chemical Co.) were used to visualize the blotted proteins of interest. Rho activity GTPase assay was performed as described in the manufacturer's protocol (Upstate Biotechnology, Inc.) using 100 g of total protein.

ERK and Rho Are Separate Signaling Pathways Necessary for LPA-induced Cell
Chemotaxis-It is known that LPA induces ERK activation through a PTX-sensitive pathway (17), whereas Rho activation is through a PTX-insensitive mechanism (18). This prompted the hypothesis that ERK and Rho may be distinct signaling pathways responsible for cell proliferation and migration, respectively. However, we found that exposure of cells to PTX prevented LPA-induced cell chemotaxis independent of changes in Rho activation (Fig. 1A). Furthermore, inhibition of Rho activity by expression of a dominant negative Rho (19) prevented chemotaxis independent of ERK activation (Fig. 1C). These findings implicate ERK and a PTX-sensitive G␣ i signaling pathway in the regulation of cell chemotaxis in response to LPA (20,21). PTX and dominant negative Rho did not affect cell attachment to the extracellular matrix (ECM), demonstrating that these signals regulate chemotaxis and do not prevent ECM recognition and attachment (Fig. 1, B and D). These findings indicate that ERK and Rho represent two distinct signaling pathways necessary for LPA-mediated chemotaxis.
ERK Activation Is Amplified in Response to a Gradient but Not a Uniform Concentration of Chemoattractant-Chemotactic cells move toward shallow chemoattractant gradients, indicating that they can compare and process extremely small differences in concentrations of extracellular stimuli (2). This is attributable, in part, to the fact that chemotactic cells localize and amplify G protein-coupled signaling pathways in the leading pseudopodium (22). We showed previously that cells exposed to a gradient of LPA readily chemotax, whereas cells exposed to a uniform concentration do not (16). These findings suggest that cells may differentially regulate ERK activity during chemoattractant sensing and cell translocation compared with cells that encounter a uniform concentration of chemokine. Indeed, ERK activation was more rapid and signif-FIG. 2. ERK is amplified in response to a gradient but not a uniform concentration of chemoattractant. Serum-starved COS-7 cells were allowed attach to collagen-coated 3.0-m porous Boyden chambers for 2 h, and 100 ng/ml LPA (A) or 25 g/ml insulin (B) was added either to the upper and lower chamber (Uniform) or to the lower chamber only (Gradient) for the indicated times. Total cellular protein was isolated, and immunoblotting was performed for phospho-ERK and total ERK as described above. In our studies, ERK1 and ERK2 phosphorylations are similar, and thus a representative blot is shown with ERK2 and is representative of at least three replicate experiments.
icantly elevated in cells exposed to a chemoattractant gradient compared with cells exposed to a uniform concentration of chemokine ( Fig. 2A). On the other hand, cells exposed to a gradient of insulin, which does not induce chemotaxis (16), showed significantly delayed and reduced ERK activity compared with cells exposed to a uniform concentration of growth FIG. 3. ERK is necessary for pseudopodial extension but not retraction. A, serum-starved COS-7 cells were allowed to extend pseudopodia on 5 g/ml collagen-coated 3.0-m porous Boyden chambers toward a 100 ng/ml LPA gradient for the indicated times. The cell body on the upper membrane surface or pseudopodia (Pseudo) on the lower membrane surface were isolated and analyzed by one-dimensional SDS-PAGE as described under "Experimental Procedures." Total cellular proteins were also isolated from whole cells attached to the porous filters which were not treated with LPA (NT). Immunoblotting was performed for phospho-ERK and total ERK as described above. B, cells were allowed to extend pseudopodia for 60 min (time 0) as described in A, and then the LPA gradient was removed from the bottom compartment and pseudopodia allowed to retract for the indicated times. The cell body and pseudopodium were purified and immunoblotted for phospho-ERK and total ERK as described above. In our studies we have seen that ERK1 and ERK2 activities correlate well with one another and that in B, ERK2 phosphorylation is representative of changes in ERK activity in general. C, cells were pretreated overnight in the absence (Ϫ) or presence (ϩ) of 50 g/ml PTX. Cells were then examined for pseudopodia extension in response to a 100 ng/ml LPA gradient (ϩ) or not treated (Ϫ) for 60 min. Pseudopodia protein on the underside of the membrane was determined as described under "Experimental Procedures." D, COS-7 cells were transfected with the empty vector (Mock), the vector encoding HA-tagged kinase dead Raf-1 mutant (HA-Raf-KD), or HA-tagged wild type Raf-1 (HA-Raf-WT). Cells were examined for pseudopodia formation in response to LPA as described above. Expression of HA-Raf-KD and HA-Raf-WT was verified using total protein from cells on a dish and immunoblotted for Raf using anti-HA antibody. E, cells were allowed to attach to collagen-coated membranes for 2 h. Cells were then pretreated for 60 min in the absence (Ϫ) or presence (ϩ) of 50 M PD98059, to inhibit MEK activity, or 1 M AG1478, to inhibit EGF receptor activity. Pseudopodia formation was initiated using LPA as a gradient and pseudopodium protein determined as described above. F, cells were allowed to extend pseudopodia toward an LPA gradient for 60 min as described above. The gradient of LPA was then removed, and either buffer (NT) or PD98059 was added to the upper and lower chambers, and the pseudopodia were allowed to retract for the indicated times. The pseudopodia retraction was quantitated as described above. G, COS-7 cells were transfected with either the empty vector or the vector encoding a constitutively active MEK1 mutant (MEKϩ). Cells were examined for pseudopodia formation in response to LPA as described above. Overexpression of MEK1 was verified using total protein from cells on a dish and immunoblotted for MEK1 using anti-MEK1 antibody. H, COS-7 cells were transfected with either the empty vector or the vector encoding a constitutively active MEK1 mutant. Cells were examined for pseudopodia retraction as described above. Results shown are representative of three replicate experiments for A and B and reflect the mean Ϯ S.D. (error bars) of three replicate experiments for C-H. factor (Fig. 2B). In our studies, we have seen that ERK1 and ERK2 activities (i.e. ERK-P) are the same and are representative of changes in ERK activity overall. These findings support the idea that ERK activation is a downstream process of chemosensing mechanisms necessary for directional cell migration.
ERK Is Necessary for Pseudopodial Extension but Not Retraction-We next wanted to determine how ERK activity controls the migration machinery of cells. Stationary cells are often observed to extend and retract pseudopodial-like processes from their cell body as they interact with the surrounding extracellular environment (23)(24)(25). However, directional cell migration or chemotaxis is characterized by protrusion of a dominant leading pseudopodium in the direction of a chemoattractant gradient followed by cell body translocation (1, 4). Therefore, it is likely that establishment of a leading pseudopodium involves the engagement of localized membrane protrusive activity and the uncoupling of retraction mechanisms. Although little is known about the molecular processes that coordinate this event, actin-myosin contractility is likely to be important. ERK has been reported to regulate the actin-myosin contractile machinery of cells through its ability to phosphorylate MLCK leading to increased MLC phosphorylation and contractile activity (26). Furthermore, recent reports indicate that MLCK activity and MLC phosphorylation are elevated in the leading front of migrating cells (8,11,24). However, it is not known how these events are regulated or whether they contribute to pseudopodial growth or retraction, per se. To investigate a possible role for ERK and MLCK in pseudopodial dynamics, we utilized an in vitro model that allows for quantification as well as biochemical analysis of pseudopodia undergoing growth or retraction in response to a chemoattractant gradient (16). Cells were plated on 3.0-m porous membranes attached to a Boyden chamber and then polarized toward an LPA gradient. Cells then extend leading pseudopodia through the pores to the underside of the chamber in the direction of the gradient. Importantly, removal of the LPA gradient from the bottom chamber causes pseudopodia to retract back to the cell body on the upper surface (16). The cell body and extending or retracting pseudopodia can be harvested differentially from the different sides of the filter for analysis. Our previous work had shown that pseudopodia extension through the micropore filters in response to an LPA gradient is linear for 30 -120 min, whereas retraction is linear for up to 120 min (16). Interestingly, we found that ERK activity was increased significantly in extending pseudopodia and decreased during retraction (Fig.  3, A and B). ERK activity in extending pseudopodia was persistent for 15-120 min, whereas initiation of pseudopodia retraction caused a rapid decrease in ERK activity that reached basal levels by 10 min (Fig. 3B). Therefore, ERK activation is associated with extension and growth of pseudopodial processes and not retraction. Inhibition of ERK activity in cells with PTX (27), expression of a dominant blocking form of Raf-1, or treatment with the MEK inhibitor PD98059 (28) prevented pseudopodia extension (Fig. 3, C, D, and E, respectively). However, treatment with an inhibitor of the EGF receptor, AG1478 (29), did not alter pseudopodia growth (Fig. 3E), suggesting that LPA stimulation of pseudopodial extension does not require EGF receptor-mediated ERK activity in trans as suggested previously (30 -32). In addition, expression of a constitutively active form of MEK (33) did not alter basal chemokineinduced pseudopodial extension, suggesting that the activation of the ERK pathway alone is not sufficient to facilitate directional membrane protrusion (Fig. 3G). Importantly, treatment of retracting pseudopodia with PD98059 or the expression of constitutively active MEK1 did not alter membrane retraction, per se. These findings demonstrate that ERK activity was not necessary for this response and that an independent pathway controls this process (Fig. 3, F and H, respectively). It is noteworthy that total ERK protein was the same in the cell body and pseudopodium under growth and retraction conditions, indicating that ERK protein levels remain constant under these conditions (Fig. 3, A and B). Similarly, MEK, the upstream activator of ERK, showed increased activity in growing pseudopodia and decreased during retraction (data not shown). The level of MEK protein was also the same in both conditions (data not shown). Together these findings demonstrate that ERK activity is necessary for pseudopodium extension but not retraction. These findings also indicate that ERK activity is spatially localized to the leading pseudopodium of cells polarized toward a chemoattractant gradient.
Pseudopodial Extension toward a Gradient of the ECM Requires ERK Activation-ECM gradients also provide spatial cues that cause directional pseudopodial extension and haptotactic cell migration (34,35). ERK activity was elevated in haptotactic pseudopodia extending toward a collagen gradient, and PD98059 inhibited this response (Fig. 4, A and B). Again, ERK protein levels were the same in the pseudopodium and cell body. In addition, PTX did not block pseudopodium extension under these conditions, indicating that PTX-sensitive heterotrimeric G proteins are not involved in this response (data not shown). Haptotactic pseudopodia do not readily retract and therefore were not investigated further. These findings provide additional evidence that ERK is necessary for directional pseudopodial extension and establishment of cell polarity.
MLCK Is Necessary for Pseudopodial Extension but Not Retraction-ERK phosphorylates MLCK leading to increased kinase activity and MLC phosphorylation. MLC phosphorylation at serine 19 (MLC-19) leads to increased actin-myosin association and contractility. Expression of a dominant blocking form FIG. 4. Pseudopodial extension toward a gradient of ECM requires ERK activation. A, COS-7 cells were allowed to extend pseudopodia through 3.0-m porous membranes coated only on the underside with 10 g/ml collagen I for 60 and 120 min. Cell body and pseudopodia (Pseudo) lysates were harvested at the indicated times as described above. Lysates from cells in suspension for 15 min (Suspension) or attached to 10 g/ml collagen I for 15 min (Attached) were also harvested for comparison. Immunoblotting was performed for phospho-ERK and total ERK as described above. B, cells pretreated in the absence (NT) or presence of 50 M PD98059 for 1 h were allowed to extend pseudopodia toward collagen I for the indicated times, and pseudopodia protein was quantitated as described above. Results shown are representative of three replicate experiments for A and reflect the mean Ϯ S.D. (error bars) of three replicate experiments for B. of MLCK in cells or treatment with the MLCK kinase inhibitor ML-7 (36) prevented pseudopodia formation and MLC-19 phosphorylation in cells (Fig. 5, A and B). Importantly, as with ERK, inhibition of MLCK activity in retracting pseudopodia with ML-7 did not alter membrane retraction or MLC-19 phosphorylation (Fig. 5, C and D). MLCK (data not shown) and MLC protein levels were the same in the cell body and pseudopodium (Fig. 5D). Thus, ERK and MLCK activity are specifically required for pseudopodial extension but not retraction.
Rho Activity Is Necessary for Pseudopodia Retraction but Not Extension-It is noteworthy that MLC-19 phosphorylation was increased in the pseudopodium compared with the cell body proper, and this does not change during extension or retraction (Fig. 5D). This suggests that MLC phosphorylation and contractility are important for both the growth and retraction processes. If ERK and MLCK activity are not necessary for pseudopodial retraction, what signal regulates this process? Rho may control pseudopodial dynamics through inactivation of myosin phosphatase, leading to increased MLC-19 phosphorylation and contraction (37). Indeed, inhibition of Rho activity in cells with a dominant blocking form of Rho (Fig. 6B), treatment with the Rho inhibitor C3 transferase (38) (Fig. 6A), or treatment with the Rho kinase inhibitor Y-27632 (data not shown) specifically prevented pseudopodial retraction but not extension. In fact, pseudopodial extension was increased compared with control cells (Fig. 6A). Associated with inhibition of Rho activity was decreased MLC-19 phosphorylation in retracting but not extending pseudopodia (Fig. 6, D and C). The increased expression of Rho by itself did not perturb pseudopodial dynamics because expression of wild type Rho in cells did not alter this process (Fig. 6A). Importantly, our previous findings indicated that Rho activity is highest in the extending pseudopodia and decreased during retraction (16). However, Rho activity does not completely return to basal levels in retracting pseudopodia (16) and therefore may be sufficient to facilitate the myosin-mediated contractile process under these conditions. The increased Rho activity seen in the extending pseudopodium may be important as a negative feedback mechanism to regulate Rac activity and protrusive processes as suggested previously (16,39). In support of this, suppression of Rho activity in cells did increase pseudopodial extension (Fig.  6A). Alternatively, Rho activity in the extending pseudopodium may serve a function(s) independent of pseudopodial dynamics. Together these results indicate that ERK and MLCK are important for pseudopodial extension, whereas Rho is necessary for pseudopodial retraction during cell polarization and chemotaxis. These distinct signaling pathways converge on myosin, a key player in pseudopodia dynamics, to regulate chemotaxis differentially. A model depicting the role of ERK and Rho is shown in Fig. 7 and discussed below. DISCUSSION It is intriguing that cells extend and retract pseudopodiallike structures from their membrane surface in an apparent random fashion. These exploratory pseudopodia appear to probe the extracellular environment and thus may serve as Transfected cells on a dish were treated uniformly on top in the absence (Ϫ) or presence (ϩ) of 100 ng/ml LPA for 10 min, and MLC phosphorylation on serine 19 (MLC-19-P) and total MLC phosphorylation was determined by immunoblotting using either phosphospecific MLC-19-P or anti-MLC antibodies as indicated. Densitometry and quantification were performed as in A, and the -fold increase is relative to MLC-19-P in control mock-transfected cells. B, cells were pretreated in the absence (Ϫ) or presence (ϩ) of 5 M MLCK inhibitor, ML-7, or dimethyl sulfoxide (DMSO; control) for 30 min and allowed to extend pseudopodia toward 100 ng/ml LPA for 60 min as described above. Total cellular protein from treated cells was harvested and immunoblotted for MLC-19-P and MLC as described above. Densitometry and MLC-19-P quantification were performed as in A, and the -fold increase is relative to MLC-19-P in control dimethyl sulfoxide-treated cells. C, cells were allowed to extend pseudopodia toward an LPA gradient for 60 min as described above. The gradient of LPA was then removed, and either buffer (NT) or 5 M ML-7 was added to the upper and lower chambers and the pseudopodia allowed to retract for the indicated times. Retracting pseudopodia were quantitated as described above. D, cells were allowed to extend pseudopodia for 60 min (time 0) toward an LPA gradient and then allowed to retract as described above for 60 min in the presence of dimethyl sulfoxide (Ϫ) or ML-7 (ϩ). Cell body and retracting pseudopodia were purified and immunoblotted for MLC-19-P, and total MLC and MLC-19-P were quantified as described above. The -fold increase represents the change in MLC-19-P (ratio of MLC-19-P to total MLC) relative to basal MLC-19-P in the corresponding cell body sample for each condition. Results shown reflect the mean Ϯ S.D. (error bars) of three replicate experiments for A-C and are representative of three replicate experiments for D.
antennae sensing spatial adhesive cues, soluble gradients of chemoattractant, and the proximity of neighboring cells. This process may be important for transmitting information to the interior cell body allowing for proper decisions to be made, such as whether to migrate or remain stationary. However, it is clear that once a directional signal is detected, random pseudopodia extension/retraction ceases, and a single pseudopodium is protruded from the cell body only in the direction of the stimulus. This suggests that molecular components that mediate membrane protrusion are rapidly localized to the side facing the directional cue, whereas retraction mechanisms are suppressed or turned off in this region. In contrast, protrusive mechanisms would be expected to be suppressed, and retractive mechanisms would remain operational on the side opposed to the directional cue. Together these signals are likely to regulate directional migration including pseudopodial extension/retraction as well as the physical turning of the pseudopodium in response to gradient changes. Therefore, it is important to identify specific signals and their spatio-temporal organization within the chemotactic cell to understand how morphological polarity and cell movement are achieved.
LPA is an abundant lipid component present in the serum and is a potent physiological mediator of cell migration, angiogenesis, and of pathological conditions associated with cancer (40 -43). Indeed, high levels of LPA have been detected in the plasma and ascitic fluid of ovarian carcinomas (44,45) where it has been shown to stimulate ovarian tumor cell growth by increasing angiogenesis (46). In addition, increased levels of LPA have been shown to stimulate prostate and breast cancer proliferation (47)(48)(49)(50). Thus, it is important to understand how LPA mediates signals and controls cellular functions including cell migration. Our findings demonstrate that ERK and RhoA are two necessary signaling pathways downstream from the LPA receptor which control the migration machinery of chemotactic cells through regulation of pseudopodia growth and retraction, respectively. Previous findings support the ability for cross-talk between ERK and Rho in motility (38,51). However, in our system, ERK and Rho regulate migration distinctly. Cell migration was inhibited by PTX, suggesting the involvement of the ERK pathway but not Rho. The use of an in vitro model system that allows for biochemical isolation of growing or retracting pseudopodia revealed that ERK and RhoA differentially modulate pseudopodial dynamics in response to an LPA gradient. An interesting finding in this study was that these FIG. 6. Rho activity is necessary for pseudopodia retraction but not extension. A, COS-7 cells were transfected with the empty vector (Mock), the vector encoding HA-tagged dominant negative RhoA mutant (Rho N19), or wild type Rho (wt Rho). Serum-starved cells were examined for pseudopodia extension for the indicated times and quantitated as described above. B, cells pretreated overnight in the absence (NT) or presence of 50 g/ml C3 toxin (C3) were allowed to extend pseudopodia toward an LPA gradient and quantified as described above. C, cells were transfected with either the empty vector or a HA-tagged dominant negative RhoA mutant. Cells were then examined for pseudopodia retraction for the indicated times as described above. D, cells were pretreated with C3 toxin and were also examined for pseudopodia retraction for the indicated times and quantified as described above. E, purified cell bodies and pseudopodia proteins from cells treated as in A were examined for MLC-19-P and MLC as described above. Quantification for densitometric analysis in E was made relative to control mock-transfected cells. F, cells were transfected and treated as described in C, and cell body and retracting pseudopodia (Pseudo) lysates were harvested and examined for MLC-19-P and MLC as described above. Quantification for densitometric analysis in F, consistent with other legends, was made relative to control mock-transfected cells. The -fold increase represents the change in MLC-19-P (ratio of MLC-19-P to total MLC) relative to MLC-19-P in the corresponding cell body control for each condition. Results shown reflect the mean Ϯ S.D. (error bars) of three replicate experiments for A-D and are representative of three replicate experiments for E and F. signals differentially regulate MLC phosphorylation at serine 19, which is necessary for actin-myosin-mediated tension and force. It is likely that ERK phosphorylation of MLC modulates adhesive processes and traction forces as activated ERK localizes to focal adhesions of migratory cells (52). On the other hand, activated ERK is not necessary for formation of actinmediated membrane ruffles and thus may not be directly involved in regulation of actin polymerization and membrane protrusion during pseudopodium extension (7).
Based on our findings and the work of others, a model can be proposed as to the role of ERK and Rho in mediating MLC function in polarized migrating cells (Fig. 7). As a cell encounters a gradient of LPA, Rac and Cdc42 are activated on the side facing the gradient which facilitates actin-mediated pseudopodia extension, which occurs independently of the ECM as shown previously (16, 53). ERK is also spatially activated on the side facing the gradient through a PTX-sensitive G␣ i pathway, leading to increased MLCK activity and MLC phosphorylation. Interaction of the protruding membrane with the ECM stabilizes the pseudopodium and promotes sustained ERK/MLC activation necessary for generation of traction force during pseudopodial expansion (11,54). Activated ERK has been shown to localize to newly forming focal adhesions at the edge of spreading cells, and this localization event requires myosin-mediated contractility (19,52). Thus, it is possible that persistent ERK activity observed in the pseudopodium spatially regulates focal adhesion dynamics and tension at the cell-substratum interface. The continued formation of new focal adhesions at the advancing front may then stimulate additional ERK activity and MLC-mediated contractility in a positive feedback loop. Small adhesions near the leading edge of motile cells have been shown to transmit strong propulsive tractions, which allow the cell to migrate while maintaining its spread morphology (55). ERK may also directly target other cytoskeletal-associated components such as caldesmon, microtubules, Scar/WAVE, or modulate integrin affinity, which are likely to be involved in this process (56 -60).
It is also intriguing that LPA stimulates Rho activity in cells and that Rho is strongly activated in the pseudopodium, yet this signaling event is not necessary for pseudopodia extension per se. In fact, blocking Rho activity increased pseudopodia extension. Localized Rho activity in the pseudopodium may serve as a negative feedback mechanism to control protrusive signals such as Rac and Cdc42, which mediate pseudopodia extension as proposed previously (14,61). RhoA has been shown to down-regulate Cdc42 and Rac1 activity (61). Importantly, in our previous work we showed that Rho activity was elevated significantly in extending pseudopodia, but we also noted that Rho activity did not return to basal levels during pseudopodial retraction (16,19). Thus, a pool of Rho activity is maintained in the retracting pseudopodium and available to facilitate MLC-mediated retraction.
We have shown previously that pseudopodial growth requires the assembly of a CAS/CrkII scaffold, which facilitates translocation and activation of Rac1 at the leading edge (16). We have found that Rac1 mediates pseudopodial extension via CAS/CrkII coupling and that this occurs independently of the ERK pathway (62). It is also likely that there are alternative means by which MLC becomes phosphorylated in migrating cells. The serine/threonine p21-activated kinase, PAK (63,64), is an effector for both Rac and Cdc42 (63)(64)(65) and has been shown to increase MLC phosphorylation, leading to increased contraction and focal adhesions (66). In addition, PAK can directly phosphorylate and activate Raf-1 leading to activation of ERK (67,68). Previous reports in other systems showed that an active PAK mutant decreased the phosphorylation of MLC in baby hamster kidney and in HeLa cells (69) but increased phosphorylation in 3T3 cells (70) (Fig. 7). However, recent evidence indicates that PAK is not required for extension of lamellipodia per se but may be involved in regulation of focal contact turnover (66).
Although early work suggested that MLC contractility was important for tail retraction in migrating cells, a growing body of evidence now points to a prominent role for myosin-mediated Based on our findings and the work of others, a model can be proposed as to the role of ERK and Rho in mediating MLC function in polarized migrating cells. As a cell encounters a gradient of LPA, the LPA receptor (LPAR) becomes ligated and activated leading to ERK activation, which is sustained and amplified on the side of the cell facing the gradient through a PTX-sensitive G␣ i pathway leading to increased MLCK activity and MLC phosphorylation. This modulates the generation of traction force necessary for proper pseudopodial formation. When the LPA gradient is removed or when the cell is exposed to a uniform chemokine concentration, there is a loss of directional ERK signaling and decreased traction force. Pseudopodium retraction is facilitated by Rho-mediated contraction through down-regulation of myosin phosphatase, producing actin-myosin force necessary to retract the pseudopodium. Although it is evident that ERK and Rho are important players in the regulation of MLC phosphorylation, most likely there are additional pathways from the initial signal which help to mediate the extension or retraction of pseudopodia. These signaling molecules include PAK1, protein kinase A, p38, protein kinase C (PKC), and protein kinase G (PKG). Positive regulators of MLC phosphorylation include PAK, p38, and protein kinase C. Negative regulators of MLC phosphorylation include PAK1, protein kinase A, and protein kinase G. contraction in pseudopodia formation. MLCK activity and MLC-19 phosphorylation have recently been shown to be elevated in the leading front of migrating cells (8,9,71). These findings support a role for Rho-mediated myosin regulation of pseudopodial protrusions. More directly, myosin phosphorylation has been linked to cell migration and protrusive events. For example, myosin IIB knockout studies show disordered cell migration of neuroepithelial and differentiated cells in the ventricular walls (71). In addition, neurons cultured from the superior cervical ganglia of B Ϫ /B Ϫ embryonic mice showed decreased rates of neurite outgrowth and lamellipodia formation in growth cones (71,72). However, inhibition of myosin function by microinjection of antibody led to increased migration and membrane protrusion along the entire periphery of cells (73), indicating that myosin activity plays a fundamental role in maintaining the organization of the cytoskeleton of the cell.
Although it is evident that ERK and Rho are important players in the regulation of MLC phosphorylation, most likely there are additional pathways that mediate the extension or retraction of pseudopodia (Fig. 7). For example, the mitogenactivated protein kinase p38 has been correlated with increased MLC phosphorylation (74), and activation of protein kinase C induces the phosphorylation of MLC by inhibiting myosin phosphatase (75). Protein kinase A phosphorylates MLCK on an inhibitory site in vitro (76,77) and may thus help to decrease MLC phosphorylation. However, agents that increase cAMP in vivo do not affect phosphorylation of this site or MLCK activity (78,79), suggesting an alternative means of regulation by protein kinase A. Myosin phosphatase can be activated by protein kinase G either directly (80) or indirectly through the inhibition of Rho (81). These signaling molecules most likely work in concert with ERK and Rho to control and fine tune pseudopodial dynamics.
Our findings show that ERK and Rho are distinct signals that regulate pseudopodium formation and retraction, respectively, through modulation of MLC phosphorylation. Understanding the temporal and spatial regulation of pseudopodia dynamics is important as the ability of cells to form a dominant leading pseudopodium is necessary for cell migration. Our findings help provide an understanding at the molecular level of how pseudopodia are regulated in chemotaxing cells.