Dual Regulation of Actin Rearrangement through Lysophosphatidic Acid Receptor in Neuroblast Cell Lines: Actin Depolymerization by Ca²⁺-α-Actinin and Polymerization by Rho

INTRODUCTION

Lysophosphatidic acid (LPA), a simple yet potent bioactive phospholipid, has been shown to elicit diverse cellular responses in many types of cells (Moolenaar, 1995; Moolenaar et al., 1997; Contos and Chun, 1998; Moolenaar, 1999). These responses are mediated by specific cell-surface G protein-coupled receptors, which are encoded by three cognate genes:lpa₁/edg-2,lpa₂/edg-4, andlpa₃/edg-7 (Hecht et al., 1996; An et al., 1998; Contos and Chun, 1998;Bandoh et al., 1999; Chun, 1999; Chun et al., 1999; Fukushima et al., 2001). LPA₁shares many downstream intracellular signaling pathways with LPA₂, including inhibition of adenylyl cyclase (AC) and activation of phospholipase C (PLC) and the small GTPase Rho (Fukushima et al., 1998, 2001; Ishii et al., 2000). In contrast, LPA₃ links to the former two pathways: AC inhibition and PLC activation, but not Rho stimulation (Ishii et al., 2000).

One prominent cellular response evoked by LPA is rearrangement of the actin cytoskeleton. In fibroblasts, LPA induces actin polymerization, resulting in the formation of cytoplasmic stress fibers that consist of filamentous actin (f-actin) and are associated with cell contraction (Ridley and Hall, 1992). In neuroblastoma cells or primary neuroblasts, LPA induces actin polymerization that produces neurite retraction/cell rounding (Jalink et al., 1993; Fukushima et al., 1998, 2000). In both cell types, LPA induces actin polymerization through the activation of Rho and its downstream Rho-associated kinase (Amano et al., 1997; Hirose et al., 1998). One difference between these cell types is the amount of f-actin that is present under resting conditions. Unlike resting fibroblasts that have little f-actin, resting neuronal cells have abundant f-actin throughout their cell bodies, in both cytoplasm and processes (Jalink et al., 1993; Fukushima et al., 1998, 2000). This raises the question of how LPA affects preexisting abundant f-actin during the remodeling of neuronal cell morphology. In addition, the intracellular actions of LPA through direct interactions with actin-binding proteins have been demonstrated (Meerschaert et al., 1998). Herein, we provide the first evidence that LPA induces both actin depolymerization and polymerization within a single neuronal cell through distinct, receptor-mediated signaling pathways. Actin depolymerization further seems to be involved in regulation of neuronal growth cone morphology.

MATERIALS AND METHODS

Cell Cultures

TR mouse cerebral cortical immortalized neuroblast cells and B103 rat neuroblastoma cells were grown as described previously (Chun and Jaenisch, 1996; Hecht et al., 1996; Fukushima et al., 1998; Ishii et al., 2000). Cells were maintained in DMEM (Invitrogen, Carlsbad, CA) containing 10% fetal calf serum and penicillin/streptomycin. For experiments, cells were grown on (poly)lysine-coated glass coverslips (12 or 40 mm in diameter, 500 cells/coverslip, unless stated otherwise; Fisher Scientific, Tustin, CA). The next day, cells were washed with serum-free Opti-MEM I (Invitrogen) supplemented with 55 μM β-mercaptoethanol, 20 mM glucose, and penicillin/streptomycin, and were further incubated in the serum-free medium for 1 d before analyses. Primary cortical neurons were prepared using embryonic day 12 mice as described previously (Fukushima et al., 2000), seeded on Cell-TaK (BD Biosciences, Franklin Lakes, NJ)-coated glass coverslips (12 mm in diameter, 500-2000 cells/coverslip), and cultured in Opti-MEM containing 5% fetal calf serum.

Time-Lapse Video Microscopy

Time-lapse, video-enhanced differential interference contrast (DIC) microscopy was carried out as described previously (Fukushimaet al., 2000). The 40-mm coverslips were mounted onto a heat-controlled perfusion apparatus (FCS-2; Bioptechs, Butler, PA) set at 37°C, and cells were observed with an inverted microscope (Axiovert 135; Carl Zeiss, Thornwood, NY) by using a 63× oil immersion objective (Plan-Apochromat; Carl Zeiss). Replacement of culture medium with a buffer containing 0.1% fatty acid-free bovine serum albumin (Sigma-Aldrich, St. Louis, MO) was manually performed with a syringe (∼2 ml/min). DIC images were collected every 15 s with a cooled charge-coupled device camera (DEI-47; Carl Zeiss) by using the Scion Image software (Scion, Frederick, MA).

Immunocytochemistry

Cells were fixed for 15 min with 3.7% formaldehyde in the presence of 0.5% Triton X-100 and were then washed with phosphate-buffered saline. To visualize f-actin, cells were incubated with Alexa Fluor 546-labeled phalloidin (1 U/ml; Molecular Probes, Eugene, OR). For double-labeling studies, cells were blocked with 10% normal goat serum and 0.5% bovine serum albumin and incubated for 2 h with a primary antibody. The antibody used was an anti-α-actinin monoclonal antibody (mAb) (1:300; Sigma-Aldrich), anti-gelsolin mAb (2 μg/ml; BD Biosciences), anti-GFP polyclonal antibody for labeling enhanced green fluorescent protein (EGFP) (1:1000; CLONTECH, Palo Alto, CA), or anti-FLAG M2 mAb (1 μg/ml; Sigma-Aldrich). Cells were further incubated with biotinylated anti-mouse IgM, anti-rabbit IgG, or anti-mouse IgG antibodies (5 μg/ml; all from Vector Laboratories, Burlingame, CA), followed by incubation with Alexa 488 streptavidin (1 μg/ml; Molecular Probes) and Alexa 546 phalloidin. For double labeling of transfected TR cells for FLAG and α-actinin, cells were first labeled for FLAG. Cells were then incubated with anti-α-actinin antibody, followed by incubation with Cy3-labeled anti-mouse IgM antibody (1.5 μg/ml; Jackson Immunoresearch Laboratories, West Grove, PA). For double labeling of B103 cells for EGFP and α-actinin, cells were first labeled for EGFP and then stained for α-actinin. Approximately 200 cells/coverslip were observed with an inverted microscope (Axiovert 135) and a 40× objective (Plan NEOFLUAR; Carl Zeiss), and counted in at least five fields, which were selected randomly, but substantially in the middle, top, and bottom of the middle and right and left of the middle of coverslips. Cells were also photographed using a 40× (Plan NEOFLUAR) or 63× oil immersion objective (Plan-Apochromat) and Cy3 or fluorescein filters, and jpg format images created. In some cases, fluorescent images were captured using a charge-coupled device camera (DXM1200; Nikon, Tokyo, Japan) and software ACT-1 (version 2.0; Nikon), and converted to jpg files. All figures of stained cells were generated by using Adobe Photoshop 6.0 (Adobe Systems, Mountain View, CA).

Ca²⁺ Imaging

Cells were cultured on 40-mm coverslips, loaded with 10–20 μM fura 2-acetoxymethyl ester, and subjected to Ca²⁺image analysis (Habara and Kanno, 1991). Briefly, the coverslip was mounted onto a heat-controlled perfusion apparatus (FCS-2; 35°C), and fluorescence images were analyzed by the digital image processor (Argus-100/Ca; Hamamatsu Photonics, Hamamatsu, Japan). The buffer was Opti-MEM I (without phenol red) containing 0.1% fatty acid-free bovine serum albumin. LPA stimulation was performed by the replacement with LPA (1 μM)-containing buffer at 2 ml/min. The images were collected at a 30-s interval before (2 images) and after LPA exposure (12 images) and displayed using pseudocolor.

PLC Assay

TR cells on 12-well plates were prelabeled with [³H]inositol (2 μCi/well), stimulated with LPA for 20 min, and radioactivity in the inositol phosphate fractions measured, as described previously (Ishii et al., 2000).

Rho Assay

Cells on 40-mm (poly)lysine-coated coverslips (20,000 cells seeded/coverslip) were stimulated with LPA and lysed in 500 μl of Rho-binding buffer (50 mM Tris-HCl, 500 mM NaCl, 10 mM MgCl₂, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, and 1× protease inhibitor cocktail, pH 7.2). Cell lysates were centrifuged and the resultant supernatants (450 μl) incubated with agarose beads-conjugated rhotekin Rho binding domain (Upstate Biotechnology, Lake Placid, NY) for 45 min on ice. The beads were washed four times and bead-bound, GTP-bound active forms of Rho specifically detected by Western blot analysis by using anti-RhoA mAb (50 ng/ml; Santa Cruz Biotechnology, Santa Cruz, CA), peroxidase-labeled anti-mouse IgG antibody (100 ng/ml; Vector Laboratories), and ECL Plus (Amersham Biosciences, Piscataway, NJ), according the manufacturers' protocols. An aliquot (15 μl) of the cell lysates was used for detection of total amounts of Rho.

Construction of FLAG-tagged α-Actinin Expression Plasmids

Chick nonmuscle α-actinin cDNA (Waites et al., 1992) was a kind gift from Dr. David R. Critchley (University of Leicester, Leicester, United Kingdom). A full-length α-actinin DNA that encodes 893 amino acids and an α-actinin mutant DNA that lacks Ca²⁺-binding domain sequences and encodes 711 amino acids [designated herein as α-actinin^(-EF)] were amplified with polymerase chain reaction by using Platinum Taq DNA polymerase high-fidelity (Invitrogen). They were subcloned in frame into theNotI/XbaI and NotI/ClaI sites of a pFLAG-CMV-2 mammalian expression vector (Sigma-Aldrich), respectively. The nucleotide sequences were confirmed by BigDye terminator cycle sequencing (Applied Biosystems, Foster City, CA).

Transfection of TR Cells or Primary Cortical Neurons

TR cells were transfected with expression plasmids by using LipofectAMINE PLUS (Invitrogen). Cells were seeded on (poly)lysine-coated glass coverslips (500 cells/12 mm in diameter for immunostaining or 20,000 cells/40 mm in diameter for Western blotting) and incubated for 15 h with DNA–lipid complex (0.1 μg of DNA-1 μl of PLUS reagent-0.5 μl of LipofectAMINE/cm²). Cells were washed and further cultured in serum-free medium for 1 d before use. Transfection of cortical neurons was performed 3 d after seeding using LipofectAMINE2000 (0.15 μg of DNA-0.4 μl of lipid/cm²; Invitrogen). Cells were washed after 24 h and further cultured in serum-free Opti-MEM for 1 d before use.

f-Actin Binding Assay

TR cells on a 9-cm dish were transfected with expression plasmids by using LipofectAMINE2000. Cells were sonicated in 100 μl of lysis buffer (50 mM Tris-Cl, 0.6 M KCl, 1 mM EDTA, 1% Triton X-100, 0.5% deoxycholate-Na, and 1× protease inhibitor cocktail, pH 7.4), and cell lysates diluted with 500 μl of lysis buffer without KCl and deoxycholate-Na. FLAG-tagged proteins were immunoprecipitated using anti-FLAG M2 antibody (3 μg) and protein G-agarose (Santa Cruz Biotechnology) and eluted in 0.1 M glycine-HCl containing 0.1% Triton X-100, pH 3.5. The elutes (100 μl) were neutralized with 10 μl of 0.6 M Tris-Cl, pH 8.2, and subjected to f-actin binding assay. Eluted proteins (40 μl) were incubated with f-actin (10 μg) in 200 μl of binding buffer (20 mM Tris-Cl, 2 mM MgCl₂, 1 mM EDTA, and 0.5 mM dithiothreitol, pH 7.4) with or without 1.1 mM CaCl₂ for 1 h on ice. f-Actin was prepared by incubating rabbit muscle actin (Sigma-Aldrich) in polymerization buffer (5 mM Tris-Cl, 1 mM MgCl₂, 50 mM KCl, and 1 mM ATP, pH 8) at 4°C for 4 h. Reaction mixtures were centrifuged at 65,000 rpm (TLA 100,3; Beckman Coulter, Fullerton, CA) for 1 h and the supernatants (unbound proteins) precipitated by adding trichloroacetic acid. The trichloroacetic acid precipitates and ultracentrifuge precipitates (f-actin–bound proteins) were analyzed by Western blot analysis by using anti-FLAG M2 antibody (50 ng/ml). Bound antibodies were visualized by subsequent incubation with peroxidase-labeled anti-mouse IgG antibody and ECL plus. Images were captured by using ATTO Light Capture (AE-6960; ATTO, Tokyo, Japan).

Western Blotting

Cells were harvested in phosphate-buffered saline and mixed with 2× SDS-sample buffer (Laemmli, 1970). Cell extracts (50 μg of protein) were separated on 10% polyacrylamide gels, and proteins were transferred to polyvinylidene difluoride membranes (Millipore, San Jose, CA), which was followed by probing with anti-FLAG M2 antibody. Membranes were further incubated with peroxidase-labeled anti-mouse IgG antibody and ECL plus and exposed to x-ray films.

Infection of B103 Cells with Retroviruses Expressing LPA Receptors

Production of retroviruses expressing LPA₁or LPA₂ and infection of B103 cells with these viruses were performed as described previously (Ishii et al., 2000).

Reagents

LPA was purchased from Avanti Polar Lipids (Alabaster, AL). Cytochalasin D, pertussis toxin, bisindolylmaleimide I (Go6850), 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid-acetoxymethyl ester (BAPTA-AM), R-(+)-trans-N-(4-pyridyl)-4-(l-aminoethyl)-cyclohexane carboxamide (Y27632), and genistein were purchased from Calbiochem (La Jolla, CA).

RESULTS

Effects of LPA on Cell Shape and Actin Cytoskeleton

To examine the effects of LPA on neuronal cell morphology, we used TR cells, which are immortalized neuroblasts derived from embryonic mouse cerebral cortex (Chun and Jaenisch, 1996; Hecht et al., 1996; Ishii et al., 2000). TR cells extend their bipolar or multipolar processes on (poly)lysine-coated glass coverslips under serum-free culture condition (Chun and Jaenisch, 1996; Ishiiet al., 2000). These cells expresslpa₁ and lpa₂but not lpa₃, and respond to LPA with rapid retraction of their processes, resulting in cell rounding (Hechtet al., 1996; Ishii et al., 2000). Closer observation of TR cells revealed that most cells possess membrane ruffling at the tips of processes (Figure1). After LPA exposure, these structures began to disappear within 1 min and completely disappeared by 4 min (Figure 1). In contrast, the well-documented phenomenon of process retraction (Hecht et al., 1996; Ishii et al., 2000) was also detectable at 1 min after LPA stimulation and was completed by 10–15 min, resulting in cell rounding (see below). LPA-induced loss of membrane ruffling was also observed in TSM cells (Chun and Jaenisch, 1996), another immortalized neuroblast cell type derived from cerebral cortex (our unpublished data). Because membrane ruffling has been shown to contain enriched actin microfilaments (Bray, 1992; Matsudaira, 1994), we examined actin rearrangement during the loss of membrane ruffling.

TR cells were labeled with Alexa Fluor 546 phalloidin, allowing the visualization of f-actin (Figure 2). In most of the resting (control) cells, f-actin was distributed throughout cell bodies and was particularly enriched within membrane ruffling (Figure 2, a and b). This type of labeling was observed in ∼75% of total cells, with the remaining cells showing no prominent staining at the tips (∼15%) or rounded morphologies without processes (∼10%). In this study, cells with these three typical morphologies are referred to as E(+) (extended process with f-actin–enriched membrane ruffling), E(−) (extended processes without f-actin–enriched membrane ruffling), and R (rounded cell body) (Figure 2e).

One to four minutes after 0.5 μM LPA exposure, E(+) cell population decreased, whereas E(−) cell population increased (Figure2, c and f). The f-actin staining observed in E(−) cells was indistinguishable from that observed in cells treated with 200 nM cytochalasin D (CD) (our unpublished data), a drug that promotes actin depolymerization and also inhibits actin polymerization. This suggested that LPA rapidly induced actin depolymerization within ruffling membranes, as CD did (Figure3a). The E(+) cell population decreased between 4 and 15 min and then increased, whereas the E(−) cell population decreased to a basal level between 4 and 60 min (Figure 2f). In contrast, the R cell population reached a peak at 15–30 min and then gradually decreased (Figure 2, d and f). The increase in the R cell population at 15 min was significantly inhibited by pretreatment with CD (Figure 3c), consistent with previous results showing that process retraction and cell rounding require actin polymerization (Jalink et al., 1994). These results indicated that LPA induced rapid actin depolymerization that was associated with loss of membrane ruffling, which was followed by actin polymerization that induced process retraction and cell rounding.

Effects of BAPTA-AM and Y27632 on LPA-induced Actin Cytoskeletal Changes

LPA receptors drive multiple signaling pathways through several types of G proteins, including G_i/o, G_q, and G_12/13 (Contoset al., 2000; Ishii et al., 2000; Fukushimaet al., 2001). The G_i/o pathway is predominantly linked to mitogen-activated protein kinase activation and AC inhibition, but not to actin rearrangement (Weiner and Chun, 1999;Ishii et al., 2000). The G_q pathway is linked to PLC activation (Ishii et al., 2000; Kimuraet al., 2001), which leads to intracellular Ca²⁺ mobilization and protein kinase C (PKC) activation. These molecules can modulate actin rearrangement by means of Ca²⁺-sensitive, actin-associated proteins such as α-actinin or gelsolin, and by phosphorylation of these proteins (Janmey, 1994; Keenan and Kelleher, 1998). The G_12/13 pathway activates Rho and Rho-associated kinases (ROCKs), resulting in actin polymerization (Gohla et al., 1998; Kranenburg et al., 1999). Specific pharmacological inhibitors, which are well established and widely used, were used to determine which pathways were involved in LPA-induced actin depolymerization.

Pretreatment of cells with pertussis toxin (PTX), which inhibits G_i/o activation (Katada et al., 1986), did not alter LPA-induced changes in the E(−) or R cell populations (Figure 3, a and c). Pretreatment of cells with a PKC inhibitor Go6850 (Toullec et al., 1991) also failed to alter the changes in those populations (Figure 3, a and c). At the used concentration (100 nM), this compound shows high selectivity for PKC but does not affect other kinases (Toullec et al., 1991). These results indicated that both G_i/o and PKC pathways were not involved in LPA-induced actin rearrangement. In contrast, when cells were pretreated with BAPTA-AM, which is known to prevent LPA-induced increase in intracellular Ca²⁺ concentrations in many types of cells (Manning and Sontheimer, 1997; Shahrestanifar et al., 1999), LPA-induced transient increases in the numbers of E(−) cells were completely blocked (Figure 3, a and e). The E(+) cell population remained at relatively higher percentages in BAPTA-AM–treated cells compared with the percentages found in control cells (Figure 3d). The R cell population increased with time in BAPTA-AM–treated cells, similar to that in control cells (Figure 3f). Treatment with BAPTA-AM alone did not significantly affect the populations of E(+), E(−), and R, suggesting that this compound showed no effect on actin rearrangement under the resting conditions. These results indicated that Ca²⁺ signaling was involved in LPA-induced actin depolymerization but not actin polymerization. Consistent with this, thapsigargin, a reagent that depletes Ca²⁺ stores and increases intracellular Ca²⁺ concentrations (Thastrup et al., 1990), induced the loss of membrane ruffling in TR cells, accompanied by actin depolymerization [E(−) population; 78% at 15 min after thapsigargin alone treatment].

A highly specific and well-characterized ROCK inhibitor, Y27632 (Uehataet al., 1997; Hirose et al., 1998), was used to examine the involvement of the Rho pathway. Pretreatment of cells with Y27632 did not affect LPA-induced decreases in the E(+) cell population percentages during the first several minutes after LPA exposure (Figure3d). However, in Y27632-treated cell populations, there was no increase in the R cell population (Figure 3, c and f). Instead, a greater increase in the E(−) cell percentages was observed in Y27632-treated cells than observed in nontreated cells (Figure 3, a, b, and e). These results suggested that cells remained in the E(−) cell stage without progressing to the R cell stage. Similar results were obtained in cells pretreated with genistein, a general tyrosine kinase inhibitor (Akiyama et al., 1987) (Figure 3, a and c), consistent with the previous report (Kranenburg et al., 1999). Because genistein has little effect on PKC and other protein kinases at the concentration used (100 nM), our results indicated the involvement of tyrosine kinase in LPA-induced actin polymerization. However, how genistein-sensitive tyrosine kinase(s) interacts with the Rho pathway remains to be determined.

LPA-induced Ca²⁺ Mobilization and Rho Activation

Our pharmacological experiments have suggested two intracellular signals involved in actin rearrangement in TR cells: Ca²⁺ mobilization and Rho activation. To further confirm the data from these experiments, we directly measured both signals and PLC activity. Ca²⁺ mobilization was monitored using fura 2, a fluorescent Ca²⁺indicator. No Ca²⁺ mobilization was observed before LPA exposure (Figure 4, a and b). However, a transient increase in Ca²⁺concentration was detected throughout a cell body, including membrane ruffling, process shaft, and the soma after LPA exposure (Figure 4, a and b). Maximal increase was observed at 2 min after LPA stimulation, consistent with an increase of E(−) cells (Figure 2f). Such Ca²⁺ mobilization was likely to be induced by inositol trisphosphate, because LPA stimulated PLC activity in TR cells (Figure 4c). LPA-induced Rho activation was tested in pull down assay by using agarose beads-conjugated rhotekin Rho binding domain, which specifically binds active GTP-bound forms of Rho (Ren and Schwartz, 2000). Within 4 min after LPA exposure, Rho was activated, and this activation sustained for at least 1 h (Figure 4d). Together with the results using Y27632 (Figure 3), these data indicated that the Rho pathway was involved in LPA-induced process retraction by actin polymerization, as reported previously (Jalink et al., 1993;Hirose et al., 1998; Kranenburg et al., 1999), but not in actin depolymerization.

Subcellular Localization of α-Actinin and f-Actin

The state of actin polymerization is controlled with actin-binding proteins, such as actin-capping or actin cross-linking proteins. These proteins regulate polymerization of globular actin, depolymerization of f-actin, and cross-linking of f-actin in response to intracellular signaling molecules (e.g., Ca²⁺) (Bamburg and Bernstein, 1991; Bray, 1992; Janmey, 1994; Matsudaira, 1994). In view of our data suggesting the role of Ca²⁺ in LPA-induced actin depolymerization (Figures 3 and 4), the involvement of α-actinin or gelsolin, both of which are Ca²⁺-regulated actin-binding proteins, was examined. α-Actinin is an actin cross-linking protein, which has been shown to play a role in neurite outgrowth in neuronal cells and to exist within membranes ruffling of motile cells (Jockusch et al., 1983; Bamburg and Bernstein, 1991; Sobue, 1993). Binding of Ca²⁺ to a nonmuscle type of α-actinin stimulates its dissociation from f-actin, resulting in the accumulation of free f-actin (Condeelis and Vahey, 1982). On the other hand, gelsolin is a Ca²⁺-regulated actin-severing protein, and binding of Ca²⁺ to gelsolin leads to actin depolymerization (Bamburg and Bernstein, 1991; Janmey, 1994).

When TR cells were double labeled for α-actinin and f-actin, expression of α-actinin was observed throughout the soma, particularly within membrane ruffling where prominent f-actin labeling was observed (Figure 5, a and b). We also treated TR cells with LPA, followed by labeling for α-actinin. Within 2 min after LPA exposure, α-actinin labeling either weakened or vanished at the tips of processes with a concomitant reduction in f-actin labeling (Figure 5, c and d). Cell populations with prominent α-actinin labeling at the tips of processes decreased between 2 and 8 min, and then gradually recovered with time (Figure 5e). This temporal profile was consistent with that of E(+) cells (Figure 5e). In contrast, gelsolin distribution showed no obvious overlap with that of f-actin both in control and LPA-treated cells (Figure 5, f and g). These results indicated that LPA-induced actin depolymerization/polymerization within ruffling membranes was correlated with α-actinin accumulation, but not gelsolin, and allowed us to focus on a role of α-actinin.

The signaling pathways regulating α-actinin distribution were explored using BAPTA-AM or Y27632. Cells were pretreated with BAPTA-AM or Y27632, exposed to LPA, and then double labeled for α-actinin and f-actin. In BAPTA-AM–treated cells, α-actinin labeling was prominent and correlated with an accumulation of f-actin within membrane ruffling (Figure 6, a and b). In contrast, cells treated with Y27632 showed loss of both f-actin and α-actinin labeling at the tips of processes after LPA exposure (Figure 6, c and d), as was found in control cells (Figure 5, c and d). These data indicated that Ca²⁺ signaling, but not Rho pathways, was involved in LPA-induced loss of α-actinin and f-actin labeling within membranes ruffling.

Effects of an α-Actinin Mutant on LPA-induced Cytoskeletal Changes

α-Actinin consists of an actin-binding domain, a domain containing spectrin-like repeats, and two EF hand Ca²⁺-binding motifs (Baron et al., 1987a,b; Arimura et al., 1988; Youssoufian et al., 1990; Waites et al., 1992) (Figure7a). To examine the requirement of α-actinin for f-actin depolymerization, a mutant form of α-actinin that lacks two EF hand motifs [designated herein as α-actinin^(−EF)] and thereby could act as Ca²⁺-inseisitive α-actinin was constructed (Figure 7a). The expression of FLAG-tagged proteins in TR cells was confirmed by Western blot analysis by using anti-FLAG antibody (Figure7b). Both FLAG-tagged α-actinin and α-actinin^(−EF) showed f-actin–binding activities in the absence of Ca²⁺ (Figure 7c), as expected from the reports that truncated α-actinin (without spectrin-like domains and EF motifs, or without only EF motifs) can bind f-actin (Tokuue et al., 1991; Hemmings et al., 1992). In addition, f-actin–binding activity of α-actinin^(−EF)was expectedly Ca²⁺ insensitive under conditions where FLAG-tagged full-length α-actinin lost the binding activity (Figure 7c). In TR cells transfected with FLAG-tagged α-actinin or α-actinin^(−EF) plasmid, FLAG immunolabeling always colocalized with f-actin, particularly within membrane ruffling (Figure 7d). However, when these cells were double labeled for FLAG and α-actinin, no or weak labeling of α-actinin was observed within membrane ruffling of FLAG-tagged α-actinin^(−EF)–expressing cells, whereas clear labeling in FLAG-tagged α-actinin–expressing cells. Because anti-α-actinin antibody recognized α-actinin but not α-actinin^(−EF) (our unpublished data), these results indicated that the level of endogenous α-actinin decreased within membrane ruffling of α-actinin^(−EF)–expressing cells. Therefore, exogenously expressed α-actinin^(−EF) could replace endogenous α-actinin to cross-link f-actin within membrane ruffling, and perhaps fail to respond to intracellular Ca²⁺ mobilization by dissociating f-actin and thereby acts as a dominant-negative form.

TR cells were transfected with a control vector (FLAG-bacterial alkaline phosphatase; BAP), or experimental vectors [FLAG-α-actinin and FLAG-α-actinin^(−EF)], exposed to LPA for 4 min and then double labeled for f-actin and FLAG. The E(−) and E(+) cells were counted in the FLAG-positive population. In cells transfected with a control vector, LPA increased the E(−) cell population (28% increase in Figure 7e) with a concomitant decrease in the E(+) cell population (46% decrease), indicating that actin depolymerization occurred as in nontransfected cells (Figure 2f). Similar results were obtained in cells overexpressing the wild-type α-actinin (Figure 7e). In contrast, even although the expression of FLAG-tagged α-actinin^(−EF) was much lower than that of FLAG-tagged wild-type α-actinin in transfected TR cells (Figure 7b), the LPA-induced increase in the E(−) cell population was significantly attenuated by α-actinin^(−EF)overexpression (16% increase in Figure 7e). There was also a significant inhibition of the LPA-induced decrease in the E(+) cell population (29% decrease). These results indicated that overexpression of a mutant α-actinin^(−EF) inhibited LPA-induced actin depolymerization, suggesting that Ca²⁺-regulation through EF hands in α-actinin was, at least in part, responsible for LPA-induced actin depolymerization within membrane ruffling.

Effects of α-Actinin^(−EF) on LPA-induced Morphological Changes in Growth Cone of Primary Neurons

α-Actinin is known to be enriched in neuronal growth cone and suggested to be involved in its morphology (Sobue, 1993). Our finding that α-actinin also accumulated in membrane ruffling at the tips of growing processes of TR cells led us to examine a role of LPA-induced actin depolymerization in growth cone morphology of primary neurons. We used primary cultures consisting of young cortical neurons in which neurite outgrowth was underway and lpa₂ was expressed (Fukushima et al., 2002). These neurons expressed endogenous α-actinin at their growth cone and cell body, colocalized with f-actin (Figure 8, a and b). This overlap was similar to that observed in TR cells, although the fluorescence intensities of these components were not as high as those in TR cells. In control, 38.9% of total neurons possessed intact growth cones, defined as f-actin–enriched, flat and wider structures than neurites. When LPA was exposed to these neurons for 4 min, the population with growth cones was decreased to 16.7%, indicating that LPA induced growth cone collapse. However, no marked process retraction and cell rounding were not observed, probably because these cellular responses were reduced with the progress of morphological maturation of neurons (Fukushima et al., 2002). Similar growth cone collapse was observed in neurons transfected with a control FLAG-BAP vector (Figure8, c–f, the population with growth cone in transfected neurons; 46.2% in control vs. 23.8% in LPA treatment). In contrast, LPA treatment failed to induce marked changes in growth cone morphology of α-actinin^(−EF)–expressing neurons (Figure 8, g–j, the population with growth cone in transfected neurons; 46.5% in control vs. 43.8% in LPA treatment). These results indicated that LPA-induced actin depolymerization through α-actinin was involved in regulation of growth cone morphology of primary neurons.

Heterologous Expression of LPA Receptorslpa₁or lpa₂

Two LPA receptors expressed in TR cells, LPA₁ and LPA₂, have similar signaling properties, including PLC activation and Rho stimulation (Fukushima et al., 1998, 2001; Ishii et al., 2000; Kimura et al., 2001). On the other hand, the LPA molecule itself can directly interact with actin-binding proteins (Meerschaert et al., 1998), and these interactions might be involved in the LPA-induced actin depolymerization. To determine whether LPA receptors or nonreceptor mechanisms are involved in actin depolymerization, either lpa₁ orlpa₂ was heterologously expressed using retrovirus expression systems in a B103 rat neuroblastoma cell line. This cell line expresses none of the three known LPA receptors and shows no cytoskeletal responses to LPA, but responds to LPA with process retraction when either lpa₁ orlpa₂ is introduced (Fukushima et al., 1998; Ishii et al., 2000; Kimura et al., 2001).

Cells were infected withlpa₁-expressing retroviruses that coexpress EGFP, and then double labeled for EGFP and either f-actin or α-actinin. Infected (EGFP-positive) cells showed marked f-actin labeling as well as α-actinin labeling within membrane ruffling as was observed in TR cells (Figure 9, a, b, e, and f). LPA treatment for 2 min resulted in reduced labeling within the tips of processes (Figure 9, c, d, g, h, and i). Such changes in f-actin and α-actinin labeling were also observed inlpa₂-expressing cells (Figure 9i). In the absence of exogenous lpa₁ orlpa₂ expression, actin depolymerization was not observed (Figure 9i). Combined with our previous data (Fukushima et al., 1998; Ishii et al., 2000), both LPA₁ and LPA₂ appeared capable of explaining LPA-induced actin depolymerization associated with membrane ruffling as well as actin polymerization in neurite shafts.

DISCUSSION

We have shown herein that LPA induces two distinct forms of actin rearrangement within single neuroblasts: depolymerization associated with loss of membrane ruffling, and polymerization associated with process retraction. A single type of LPA receptor regulates each actin rearrangement by activating distinct cellular signaling pathways: depolymerization via Ca²⁺-α-actinin and actin polymerization via Rho. These results suggest a novel role for receptor-mediated LPA signaling in regulation of the actin cytoskeleton in neuronal cells.

LPA Induces Both Actin Depolymerization and Polymerization within a Single Cell

Actin microfilaments are primarily regulated through their interactions with actin-binding proteins and/or intracellular signaling messengers produced by extracellular stimuli such as growth factors and extracellular matrix (ECM) (Bamburg and Bernstein, 1991; Bray, 1992;Janmey, 1994; Matsudaira, 1994). Depending on the types of interactions, cells can undergo either polymerization or depolymerization to change their morphologies (e.g., leading edge of migrating cells).

LPA has been shown to be a potent inducer of actin polymerization (Moolenaar, 1995, 1997). LPA-induced actin polymerization results in process retraction and cell rounding in neuronal cells such as N1E-115, or stress fiber formation in fibroblast cells (Ridley and Hall, 1992;Jalink et al., 1993). These cellular responses are mediated by actomyosin interactions through activation of the Rho pathway, producing contractile forces that induce cell shape changes (Jalinket al., 1994; Tigyi et al., 1996; Kozma et al., 1997; Hirose et al., 1998; Kranenburg et al., 1999; Fukushima et al., 2000; Weiner et al., 2001). Herein, we have also shown that Rho is activated by LPA in TR cells and LPA-induced process retraction is blocked by a ROCK inhibitor or CD, which indicates the possible involvement of Rho and actomyosin for process retraction in TR cells (Figure10).

In addition to actin-polymerizing effects, LPA induced more rapid, transient actin depolymerization within the same cell. It is possible that depolymerized actin could be the source for subsequent actin polymerization during process retraction. However, this is unlikely because BAPTA-AM did not block LPA-induced process retraction, whereas Y27632 failed to inhibit LPA-induced loss of f-actin within membrane ruffling. Therefore, LPA appears to activate two independent signaling pathways, with opposing effects on the actin rearrangement, and apparently using distinct pools of actin. Our results also suggest that actin rearrangement induced by LPA may be compartmentalized; actin depolymerization associated with membrane ruffling occurs at the tips of processes, whereas actin polymerization occurs in the shaft of processes (Figure 10).

LPA-induced actin depolymerization has not been documented previously, perhaps because of the use of different cell types or different techniques that focused on longer time periods after LPA exposure. For example, fibroblasts have little cytoplasmic f-actin in resting conditions (i.e., G₀ phase of cell cycle) (Ridley and Hall, 1992), and actin depolymerization might not be detectable after stimulation of the cells. In neuronal cells, actin rearrangement has been examined by fluorometric methods that monitor changes in f-actin in the entire cell body (Jalink et al., 1993), and these methods might not detect local fine changes in f-actin levels.

LPA-induced Actin Depolymerization Involves Ca²⁺-α-Actinin but Not Rho

Several lines of evidence in the present study support an idea that LPA-induced actin depolymerization is mediated by Ca²⁺–α-actinin interactions: 1) α-actinin was colocalized with enriched f-actin within membrane ruffling, and both dissipated after LPA exposure; 2) pretreatment with a Ca²⁺ chelator, BAPTA-AM, inhibited LPA-induced loss of membrane ruffling; 3) intracellular Ca²⁺mobilization was induced in membrane ruffling after LPA stimulation; and 4) overexpression of Ca²⁺-insensitive α-actinin^(−EF), which binds f-actin and replaces endogenous α-actinin, attenuated LPA-induced actin depolymerization. LPA-induced increases in Ca²⁺concentration within membrane ruffling perhaps inhibit the actin cross-linking activity of α-actinin through its EF hands, which in turn leads to the dissociation of α-actinin from f-actin (Bamburg and Bernstein, 1991; Janmey, 1994) (Figure 10). However, actin depolymerization observed as the disappearance of f-actin labeling should normally require actin-depolymerizing or actin-severing activity, which has not been documented for α-actinin. This suggests involvement of another mechanism for actin depolymerization after dissociation of actin filaments resulting from the inhibition of α-actinin activity.

α-Actinin is also known to link actin filaments to receptors for ECM molecules (Pavalko et al., 1991). It has been shown that complexes containing α-actinin, ECM, and ECM receptors are involved in the formation of membrane ruffling (Jockusch et al., 1983; O'Connor et al., 2000). We detected the distribution of β1-integrin, one of the ECM receptors, in membrane ruffling of TR cells by immunocytochemical labeling (our unpublished data). LPA-induced loss of membrane ruffling may involve dissociation of actin filaments from those complexes as well as between filaments.

Our data show a role for α-actinin in actin rearrangement that is different from previous results observed in several cell types. In fibroblasts, α-actinin is recruited along f-actin stress fibers and focal adhesions in response to serum or LPA stimulation (Barry and Critchley, 1994). These cells are likely to use α-actinin in actin rearrangement primarily directed toward polymerization in a Rho-dependent manner, which is different from our data on neuronal cells presented herein. Whether α-actinin is also involved in actin polymerization in TR cells remains to be determined. However, this seems unlikely because overexpression of α-actinin or α-actinin^(−EF) did not alter LPA-induced process retraction (our unpublished data). Thus, in LPA-stimulated TR cells, α-actinin appears to play a role in actin depolymerization but not in actin polymerization.

Both LPA₁ and LPA₂ Could Mediate LPA-induced Actin Depolymerization and Polymerization

Both LPA₁ and LPA₂ can stimulate PLC through PTX-insensitive G proteins (e.g., G_q) in neuronal cells (Ishiiet al., 2000; Kimura et al., 2001), producing inositol trisphosphate that induces Ca²⁺release from intracellular Ca²⁺ stores (Berridge, 1993). This Ca²⁺ mobilization by activation of the G_q-PLC pathway has been also confirmed in the present study (Figure 4) and is probably involved in the inhibition of α-actinin activity (Bamburg and Bernstein, 1991; Janmey, 1994). Both LPA receptors can also activate the Rho pathway that is linked to myosin stimulation, which leads to actin polymerization (Amano et al., 1998; Fukushima et al., 1998; Ishii et al., 2000). Thus, a single LPA receptor is likely to use two independent signaling pathways for actin regulation. This idea is supported by the data from our heterologous experiments using B103 cells, which demonstrated that expression of either LPA receptor was enough for dual regulation of actin rearrangement.

Possible Roles for LPA-induced Actin Rearrangements in Neuronal Cell Morphology

TR cells show many features of cortical neuroblasts (e.g., expression of neuroblast markers, such as brain factor-1,lpa₁, and nestin [an intermediate filament protein]), but also express early neuronal markers, such as neurofilament proteins (Chun and Jaenisch, 1996). Therefore, these cells provide a good, homogenous model system to study intracellular events that occur in primary neuroblasts or migrating or differentiating neurons. Cortical neuroblasts expresslpa₁ (Hecht et al., 1996) and extend radially oriented processes and undergo a “to-and-fro movement,” called interkinetic nuclear migration, in which process retraction and extension are repeated with a radial nuclear movement (Seymour and Berry, 1975). Our previous study has shown that LPA signaling affects cortical neuroblast morphology through actin polymerization (Fukushima et al., 2000). However, how actin depolymerization is involved in the neuroblast morphology remains uncertain. α-Actinin may form a complex with actin filaments and focal adhesion proteins, which functions as scaffolds for signal transduction and/or the attachment of processes (Barry and Critchley, 1994), although its precise localization in neuroblasts remains to be determined. Actin depolymerization through Ca²⁺–α-actinin interactions may produce loss of these scaffolds and/or the detachment of processes, an essential step of interkinetic nuclear migration (Fukushima et al., 2000).

Migrating or differentiating neurons expresslpa₂ (Fukushima et al., 2002) and extend leading processes during migration or axons and dendrites at their final destination, respectively. Ca²⁺ is an important factor for the regulation of growth cone morphology in these processes (Letourneau, 1996), and both α-actinin and f-actin are concentrated in the filopodia of growth cone (Sobue and Kanda, 1989). Our data from the experiments using primary neurons suggest the involvement of Ca²⁺–α-actinin interactions in LPA-induced changes in growth cone morphology, which would implicate LPA as a guidance molecule for migrating or differentiating neurons. Clearly, further investigation should be necessary for clarifying the biologically relevant roles for LPA-induced dual actin rearrangement in the nervous system.

FOOTNOTES

FOOTNOTES

Abbreviations used:

BAP	bacterial alkaline phosphatase
CD	cytochalasin D
DIC	differential interference contrast
ECM	extracellular matrix
EGFP	enhanced green fluorescent protein
f-actin	filamentous actin
LPA	lysophosphatidic acid
PKC	protein kinase C
PLC	phospholipase C
PTX	pertussis toxin

REFERENCES