Small GTP-binding protein Rho stimulates the actomyosin system, leading to invasion of tumor cells.

We have shown previously that Rho plays a pivotal role in 1-oleoyl-lysophosphatidic acid (LPA)-dependent invasion of rat hepatoma cells (MM1). Herein we made stable transfectants of MM1 expressing active and Botulinum exoenzyme C3 (C3)-sensitive (Val14), or active and C3-insensitive (Val14/Ile41) forms of human RhoA. Both transfectants showed greatly promoted invasive ability in vitro in the absence of LPA as well as in vivo, adherence to the dish with scattered shape, and enhanced phosphorylation level of 20-kDa myosin light chain (MLC20). A specific MLC kinase inhibitor (KT5926) could inhibit their invasion and the phosphorylation level of MLC20. Stable active RhoA transfectants of W1 cells (low invasive counterpart of MM1) also demonstrated promoted invasive ability in vitro and in vivo, and enhanced phosphorylation level of MLC20. C3 treatment inhibited the invasiveness of the Val14 RhoA transfectant but not that of the Val14/Ile41 RhoA transfectant. LPA enhanced the invasiveness of both transfectants, and this enhancement was abolished by the C3 treatment. These results suggested that 1) the Rho signaling pathway and actomyosin system were linked in the transmigration of tumor cells, and 2) expressed active RhoA enhanced LPA-induced tumor cell invasion via the activation of endogenous RhoA pathway, indicating a positive feedback mechanism in the activation of RhoA.

Transcellular migration of tumor cells through host cell layer is one of the most crucial steps in cancer invasion and metastasis (1)(2)(3)(4). We have previously developed a cell monolayer invasion assay, in which rat hepatoma cells (AH130) invade through a cultured mesothelial cell monolayer (MCL) 1 of the syngenic rat (5). Transcellular migration of tumor cells could be quantified by counting the number of tumor cells that penetrated through the MCL. Moreover, the invasive ability in vitro of these cells was well correlated with the in vivo ability to make tumor nodules after implantation in the peritoneal cavity of the syngenic rat (6). Using this in vitro system, MM1 cells (a highly invasive cell line isolated from parental AH130 cells) could transmigrate through the MCL in the presence of serum (7), and this effect of serum was completely substituted by 1-oleoyl-lysophosphatidic acid (LPA) (8). The effect of LPA on the induction of transmigration was abolished by the pretreatment of MM1 cells with C3 ADP-ribosyltransferase (C3) (Clostridium botulinum) (9), which is known to specifically inhibit the function of small GTP-binding protein Rho (10,11), suggesting that Rho plays a pivotal role in this transmigration.
Rho protein is a well known member of the p21 Ras superfamily of small GTPases, which exhibits both GDP/GTP binding and GTPase activities. Rho regulates signal transduction from receptors in the membrane to a variety of cellular events related to cell morphology (12), motility (13), cytoskeletal dynamics (14,15), and tumor progression (16,17), as a molecular switch in the cells (for reviews, see Refs. 18 -20). Recently, a series of reports suggested that Rho plays an important regulatory role in induced myosin light chain (MLC) 20 phosphorylation, actomyosin stimulation, and the consequent contraction of smooth muscles (21)(22)(23)(24). Rho-dependent contraction was also found in the neurites of mouse neuroblastoma (NIE-115) and rat pheochromocytoma (PC12) cells. Treatment with C3 inhibited LPA-induced actomyosin-based contractile forces and thereby prevents this neurite retraction and cell rounding (25). Rho also might involve the myosin-based contractility in the formation of focal adhesion and stress fibers in Balb/c 3T3 fibroblast (26).
Here we have focused on how Rho regulates the actomyosin system, leading to transmigration of tumor cells, and how Rho controls signal transduction pathways during the LPA-induced transmigration of tumor cells. Preliminary reports of this work have appeared (27,28).

Cell Culture
Mesothelial cells were isolated from Donryu rat mesentery and cultured in minimum essential medium containing 2-fold concentrations of amino acid and vitamins (M-MEM) supplemented with 10% fetal calf serum (FCS), as reported previously (5). MM1 cells (a highly invasive clone) (5) and W1 cells (a very low invasive clone) (6), isolated from the same parental AH130 cells, were maintained as a suspension in the M-MEM supplemented with 10% FCS and split at a 1:20 ratio every 3 days.

Construction of Mutant RhoA Expression Vectors
The expression plasmids were designed to generate from pEXVactive (Val 14 ) RhoA vector, provided by Dr. A. Hall (University college, London). A 0.63-kilobase pair EcoRI-NotI fragment, containing fulllength human RhoA cDNA tagged at the N terminus with FLAG sequence (8 amino acids; DYKDDDDK) (29), was generated by the polymerase chain reaction (PCR) technique using the forward primer 5Ј-CTTGAATTCATGGACTACAAGGACGACGATGACAAGGCTGCCA-TCCGGAAGAAACTGGTG-3Ј (60-mer) and the reverse primer 5Ј-TTT-GCGGCCGCTCATCACAAGACAAGGCAACC-3Ј (32-mer) with pEXVactive RhoA as the template. PCR was carried out as follows: denaturation for 1 min at 94°C, annealing for 1 min at 56°C, extension for 1 min at 72°C with 30 thermal cycles, following 7-min extension using Pwo DNA polymerase. This fragment contained the FLAG tag sequence at the N terminus of full-length human RhoA cDNA with an EcoRI site engineered before the start ATG codon and a NotI site after the 3Ј-end of the open reading frame. The generated mutant cDNA was introduced into the mammalian expression vector pcDNA3 (Invitrogen, San Diego, CA) at the site of EcoRI and NotI. Mutation of active RhoA (C3-sensitive) to constitutively active RhoA (Val 14 , codon 41 Asn (AAC) to Ile (ATC)) was generated as follows. First PCR used two sets of primers: first set forward primer, 5Ј-CGGAAGAAACTGGTG-3Ј (15-mer); first set reverse primer, 5Ј-ACATAGATCTCAAACACTGT-3Ј (20-mer); second set forward primer, 5Ј-TTGAGATCTATGTGGCAGAT-3Ј (20-mer); second set reverse primer, 5Ј-TCACAAGACAAGGCAACC-3Ј (18-mer) with pcDNA3-FLAG-Val 14 RhoA as the template. 30 cycles of PCR were carried out as follows: denaturation for 1 min at 94°C, annealing for 1 min at 60°C, extension for 1 min at 72°C following a 7-min extension. The PCR products were separated by agarose gel electrophoresis and were extended to generate double-stranded DNA in the presence of Pwo polymerase and dNTPs for 5 min at 72°C. The products were gel-purified and then served as the template for the second series of PCR using the forward primer 5Ј-CGGAAGAAACTGGTG-3Ј (15-mer) and the reverse primer 5Ј-TCACAAGACAAGGCAACC-3Ј (18-mer). The mutant cDNAs were introduced into the expression vector pcDNA3 at the sites of EcoRI and NotI. All plasmids were sequenced with the dideoxy termination method (30) using [␣-35 S]ATP and the Sequi-Gen sequence apparatus (Bio-Rad) to verify the correct substitutions.

Transfection
Cells (MM1 or W1) in log phase growth were centrifuged and resuspended in Ca 2ϩ -and Mg 2ϩ -free PBS at a concentration of 1 ϫ 10 7 cells/ml. 5 g of plasmid was added to 80 l of cell suspension, and electroporation was performed using a Bio-Rad Gene Pulser (capacitance, 25 microfarads; field strength, 2.5 kV/cm) as described previously (27). After the transfection, the cells were grown in the M-MEM with 10% FCS for 48 h, following the selection with G418 (200 -300 g/ml). Clonal transfectants were isolated with limiting dilution method.

Cell Monolayer Invasion Assay
The assay procedure of in vitro invasive ability of tumor cells was described previously (5). Briefly, after mesothelial cells from rat mesentery had reached confluency in a 35-mm dish, the culture medium was removed, and 2 ϫ 10 5 tumor cells were seeded onto the mesothelial cell monolayer in the M-MEM containing LPA or FCS. The number of the penetrated single tumor cells and tumor cell colonies (invasion foci) was counted under a phase contrast microscope (Olympus, Japan) in 16 different visual fields (0.59 mm 2 each). The in vitro invasion ability was quantitatively calculated as the percentage of infiltrated cells out of total seeded cells. In some experiments, tumor cells were pretreated with 50 g/ml C3 for 24 h or with 15 M KT5926 for 30 min, followed by washing twice with M-MEM. Statistical comparisons were made using Student's t test.

Purification of C3 Exoenzyme
Escherichia coli expression of C3 was achieved by use of the pGEX2T vectors, kindly provided by Dr. Larry Feig (Tufts University, Boston, MA), encoding the glutathione S-transferase (GST)-C3 fusion protein.
The purification procedure of recombinant C3 was described previously (31). Briefly, pGEX-transformed E. coli (strain RR1) was grown to midexponential phase, induced for 3 h with 0.1 mM isopropyl-thio-␤galactoside, and centrifuged at 5000 ϫ g for 10 min. The precipitated material was lysed in the lysis buffer (PBS containing 1 mg/ml lysozyme, 1% Triton X-100, 1 mM EDTA, 5 mM dithiothreitol (DTT), 10 g/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride). The bacterial lysate was incubated with glutathione-coupled Sepharose 4B beads (Pharmacia, Sweden) and eluted with glutathione. Separation of C3 from the GST was carried out by the cleavage with 250 units/ml thrombin (Sigma) for 20 h at 22°C.

ADP-ribosylation
MM1 and W1 cells and their active RhoA transfectants were plated at 5 ϫ 10 5 cells/35-mm dish and cultured in M-MEM containing 10% FCS with or without 50 g/ml of C3 for 24 h. The cells were then washed with PBS and spun down by brief centrifugation. Cell pellets were suspended in 50 l of homogenization buffer (20 mM Tris-HCl, pH 7.5, 0.25 M sucrose, 5 mM MgCl 2 , 1 mM EDTA, 1 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 5 M leupeptin) and were homogenized on ice by 10 strokes in a Potter-Elvehjem homogenizer. The homogenate was centrifuged at 700 ϫ g for 10 min, and the supernatant (20 g of protein) or GST-RhoA (60 ng) was incubated at 30°C for 1 h with 1 g of C3 and 50 M [ 32 P]NAD (final concentration of 250 Ci/ml) in 100 mM Tris-HCl (pH 8.0), containing 10 mM nicotinamide, 10 mM thymidine, 10 mM DTT, 5 mM MgCl 2 in a total volume of 100 l. After the reaction, the mixture was added to 400 l of 0.02% sodium deoxycholate and 200 l of 24% trichloroacetic acid. The mixture was chilled on ice for 20 min and then centrifuged at 10,000 ϫ g for 15 min at 4°C. The pellet was dissolved in Laemmli sample buffer (32), heated at 100°C for 3 min and subjected to SDS-12% PAGE. After staining with Coomassie Brilliant Blue R-250, the gel was dried and was exposed for 16 h to Hyperfilm-MP (Amersham Corp.). ADP-ribosylation was estimated with a previously reported method (33,34).

RNA Blot Analysis
Total RNA was prepared from culture cells using Trizol Reagent (Life Technologies). RNA was denatured by the treatment with 1.1 M glyoxal, run in a 1% agarose gel, and then transferred for 16 h by capillary action to a Hybond-Nϩ membrane (Amersham) and UV cross-linked with a cross-linker (1200 watts) (Stratagene). A [␣-32 P]dCTP-labeled probe was made with Multiprime DNA labeling systems (Amersham). Membrane was hybridized for 2 h at 65°C with the Rapid-hyb buffer (Amersham) containing the radioactive probe. After hybridization, the filters were washed twice for 10 min at room temperature in 0.1% SDS, 2 ϫ SSC and twice for 30 min at 65°C in 0.1% SDS, 0.2 ϫ SSC. Processed filter was exposed for 16 h to Hyperfilm-MP.

DNA Probes
The human RhoA cDNA of the plasmid pcDNA3-FLAG-RhoA was cut with NheI and EcoRI, generating a N-terminal FLAG-RhoA fragment of 424 base pairs used for specific detection of the human RhoA. The rat RhoA cDNA was generated with RNA-PCR using 1 g of total RNA from MM1 cells and a set of primers: RhoA forward primer 5Ј-ATG-GCTGCCATCCGGAAGAA-3Ј (20-mer) and RhoA reverse primer 5Ј-TCACAAGACAAGGCAACCA-3Ј (18-mer). The reverse transcriptase reaction was carried out using random hexamers and Moloney murine leukemia virus reverse transcriptase (Takara, Otsu, Japan). The PCR reaction profile included 35 cycles of denaturation for 1 min at 95°C, annealing for 1 min at 65°C, and extension for 1 min at 72°C using Pwo DNA polymerase. The PCR product was separated by agarose gel electrophoresis, subcloned into pCR-ScriptSK(ϩ) vector (Stratagene), and sequenced. 2 The 579 base pairs of rat full-length RhoA cDNA was used for a specific probe for rat RhoA. The RhoA probe for RNA blotting was an equal mixture of the human RhoA probe and the rat RhoA probe.

Expression of Rat MLC20 Protein Using Baculovirus System
Full-length cDNA of MLC20 from MM1 cells was generated with RNA-PCR using 1 g of total RNA prepared from MM1 cells and a set 2 K. Yoshioka and K. Itoh, unpublished results. of primers (forward, 5Ј-ATGTCGAGCAAAAGAGCGAAG-3Ј (21-mer), and reverse, 5Ј-TCAGTCATCTTTGTCTTTCGC-3Ј (21-mer)). PCR conditions were the same as for rat RhoA cDNA. The gel-purified PCR products were cloned into pCR-Script vector and sequenced. The sequence was almost identical to the reported sequence of cultured rat aortic smooth muscle MLC20 (35), besides the 31st codon from CAG (Gln) to GAG (Glu). 3 This cDNA was inserted into pVL1392 transfer vector (Invitrogen) at NotI and EcoRI sites, and recombinant baculoviruses were generated with the plasmid and linear wild type Autographa californica multiple nuclear polyhydrosis virus (AcMNPV) DNA (Invitrogen) using a cationic liposome-mediated technique. The recombinant virus was isolated by a plaque assay according to the instruction manual (Invitrogen). Sf9 cells (1 ϫ 10 7 cells) (Pharmingen, San Diego, CA) were infected with the recombinant baculoviruses and cultured for 3 days in Grace's insect medium supplemented with 10% FCS. The cells were harvested, washed with PBS twice, and homogenized in 1 ml of cell extraction buffer (7 M urea, 10 mM 2-mercaptoethanol, and 50 mM Tris-HCl (pH 8.0)) in ice for 30 min. After centrifugation at 12,000 ϫ g for 30 min at 4°C, an equal volume of ethanol was added to the supernatant. The mixture was then centrifuged at 12,000 ϫ g for 30 min at 4°C again, and an equal volume of ethanol was added to the supernatant and centrifuged at 12,000 ϫ g for 30 min at 4°C again. Final pellet was dissolved with 100 l of the extraction buffer and used for the control of MLC20 of MM1 cells.

Estimation of MLC20 Phosphorylation in Situ
To analyze MLC20 phosphorylation, cells were plated in the M-MEM with 10% FCS and grown to confluency. Confluent cells were then washed twice with the M-MEM without FCS and rendered quiescent by incubating in the medium for 16 h. In some experiments, 15 M of KT5926 was added to the M-MEM and incubated for 30 min and washed with the same medium. 25 M of LPA were added to the same medium and incubated for various times. Medium was removed, cells were washed with PBS twice, and 10% ice-cold trichloroacetic acid was added for 30 min in ice. Cells were centrifuged at 20,000 ϫ g for 10 min, and the precipitated material was washed twice with acetone containing 10 mM DTT to remove trichloroacetic acid, dried at room temperature, and dissolved in Laemmli sample buffer (32). Samples were then heated at 100°C for 3 min and subjected to SDS-12% PAGE. Following electrophoresis, proteins were transferred to a nitrocellulose membrane (0.2 m, Bio-Rad) with the semidry method and blocked with 3% BSA in TPBS (PBS plus 0.05% Tween) for 1 h at room temperature. Primary antibodies were anti-PMLC20 polyclonal Abs (specific for phosphorylated Ser 19 ) at a 1:100 dilution, and secondary antibody was anti-rabbit IgG alkaline phosphatase conjugate (Promega) used at a 1:7500 dilution. Final signal was developed using nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl-phosphate (Promega) as substrate. The blot membrane was scanned with a GT-9500 flat scanner (Epson, Japan) and analyzed with NIH Image software using a Power Macintosh computer (Apple, Japan). Relative phosphorylation level of MLC20 in MM1 cells was estimated from the signal of P-MLC20 normalized with the signal of MLC20, which was obtained by immunoblotting using MY21 monoclonal Abs on the duplicate blot. The absolute phosphorylation level of MLC20 was estimated using the extracts from cultured human endothelial cells (as internal control) by the comparison of signal density reacted to PMLC20 Abs with in situ MLC20 phosphorylation level determined by the previously reported urea-PAGE immunoblotting method using anti-MLC20 Abs (36).

Immunoblotting
Detection of FLAG-RhoA and Endogenous RhoA-Cells were washed with PBS twice and lysed in a lysis buffer (10 mM Tris-HCl (pH 7.5) containing 50 mM NaCl, 50 mM NaF, 10 mM EDTA, 1 mM DTT, 1% Triton X-100, 1% SDS, 1% sodium deoxycholate, 1 mM phenylmethylsulfonyl fluoride, 5 M leupeptin, and 10 g/ml aprotinin) for 30 min in ice. The lysates were centrifuged to remove insoluble materials, normalized according to their protein content, and loaded onto SDS-12% PAGE. The gel was transferred to a Finetrap NT-31 membrane (Nippon Eido, Tokyo, Japan) with a semidry blotting apparatus and blocked for 1 h at room temperature with 3% BSA (Promega, Madison, WI) in TPBS. Primary antibodies were 1 g/ml anti-RhoA polyclonal Abs (SC-179, Santa Cruz Biotechnology, Inc., Santa Cruz, CA), which could detect human and rat RhoA equally. Secondary antibody was horseradish peroxidase-conjugated donkey anti-rabbit IgG (Amersham) used at a 1:50,000 dilution, and final signal was detected by ECL (Amersham). Next, FLAG-RhoA proteins were detected on the same blots. Membrane was stripped with a solution containing 2% SDS, 100 mM 2-mercaptoethanol, and 62.5 mM Tris-HCl (pH 6.8) for 30 min at 50°C, washed with TPBS, and blocked with 3% BSA in TPBS for 16 h at 4°C. Membrane was then reprobed with anti-FLAG M5 monoclonal Abs (Eastman Kodak Co.) (9 g/ml) for 1 h at room temperature, washed with TPBS, and incubated with a 1:20,000 dilution of horseradish peroxidase-conjugated sheep anti-mouse antibody (Amersham) for 1 h at room temperature following detection by ECL.
Detection of Mitogen-activated Protein Kinase (MAPK) Phosphorylation-Cells were starved in the M-MEM without FCS for 16 h and subsequently treated with LPA at 37°C for various times. The cell lysate was prepared as described under "Estimation of MLC20 Phosphorylation in Situ." The lysates were separated by SDS-8% PAGE. Proteins were transferred to a nitrocellulose membrane with the semidry blotting method and blocked with 3% BSA in TPBS for 1 h at room temperature. Primary antibodies were a 1:1000 dilution of anti-phosphospecific MAPK polyclonal Abs (New England Biolabs, Beverly, MA), and secondary antibody was goat anti-rabbit IgG conjugated to alkaline phosphatase used at a 1:7500 dilution and developed with the same alkaline phosphatase substrate. Membrane was stripped and then reprobed with 1 g/ml anti-p44/42 MAPK polyclonal Abs (New England Biolabs) for 1 h at room temperature, washed with TPBS, and incubated with a 1:7500 dilution of goat anti-rabbit IgG conjugated to alkaline phosphatase for 30 min at room temperature following detection with the alkaline phosphatase substrate.  size of transcripts of endogenous and expressed RhoAs). These plasmids were transfected into MM1 cells, and stably expressed transfectants were isolated by the selection using G418. We obtained a number of clones of stable transfectants and analyzed them with RNA blotting using a RhoA probe as shown in Fig. 2A. The expression levels of active RhoA transcript varied from 10 to 80% of those of endogenous RhoA in these transfectants. Immunoblotting analysis using anti-FLAG monoclonal Abs or anti-RhoA polyclonal Abs (Fig. 2B) revealed that the protein expression level of FLAG-RhoA was 5-10% of that of endogenous RhoA in these cells (Fig. 2, compare A and  B), suggesting posttranscriptional regulation in the expression of RhoA. The mobility difference between V14RhoA and V14/ I41RhoA on SDS-PAGE (Fig. 2B) was likely to be the conformational difference in these RhoA mutants, which has already been reported previously (12). Fig. 2C demonstrates that these transfectants showed 25-100 times higher in vitro invasiveness through the MCL in the absence of FCS in the assay medium than did mock transfectants. It is of note that there was a positive correlation between the expression level of active RhoA and in vitro invasiveness among these transfectant clones (Fig. 2C).

Expression of Active
We next examined the morphology of these cells in the presence of serum. In contrast to mock transfectants, which grew in suspension, the V14RhoA and V14/I41RhoA transfectants adhered to the plastic culture dish with scattered shape (Fig. 3, A,  B, and D). Treatment with 50 g/ml exoenzyme C3 for 24 h made V14RhoA cells completely round up, and the cell morphology was reversed to that of parental MM1 cells (Fig. 3C); however, V14/I41RhoA cells were resistant to the C3 treatment and remained adhered to the culture dish (Fig. 3E).
Active RhoA Promoted Invasiveness in Vivo-To examine the in vivo invasive ability of these transfected cells, 2 ϫ 10 7 cells were implanted in the peritoneal cavity of the syngenic rat. The active RhoA-transfected cells invaded more extensively into the peritoneum and formed many more tumor nodules as compared with mock transfectants (data not shown). The incidence of macroscopic tumor nodule present in the peritoneum of rats implanted with active RhoA transfectants (6 of 10 for V14RhoA and 5 of 7 for V14/I41RhoA) was higher than that of rats implanted with mock transfectants (2 of 8) (see Table I). These results suggested that the activation of RhoA greatly promoted the tumor invasive ability in vivo as well as in vitro. We could not find any macroscopic metastatic legion (lung, liver, spleen, stomach) in these rats.
Active RhoA Enhanced MLC20 Phosphorylation-MM1 cells require the presence of FCS to exert their transcellular migration in vitro, and it has been shown that LPA can be substituted for serum (8). Recently, a series of reports (26,37,38) indicated that stimulation of fibroblasts with LPA activated Rho-induced contractility and stress fiber formation. These processes seem to be linked to the activation of the actomyosin system (26). To ascertain whether the active RhoA-transfected MM1 cells stimulate the actomyosin system, the phosphorylation level of MLC20 was examined with immunoblotting using anti-PMLC20 polyclonal Abs (specific for phosphorylated Ser 19 ) (39). The phosphorylation level of MLC20 was increased 2-5fold in both V14RhoA and V14/I41RhoA transfectants compared with mock transfectants (Fig. 4, A, upper panel, and B, left two columns), but it was completely abolished by the treatment with 15 M KT5926 for 30 min (a specific inhibitor for MLC kinase (40) ; Fig. 4B, column 3). However, this effect of KT5926 was temporal and declined in a time-dependent fashion (Fig. 4, A (bottom) and B). After the removal of 15 M KT5926 and wash out of this inhibitor, the phosphorylation level of MLC20 in the V14/I41RhoA-1 cells gradually elevated to 32.2% (4 h), 37.5% (8 h), 54.2% (12 h), and 84.9% (20 h) of that of the untreated transfectants (percentage of MLC20 phosphorylation was 47.9%). The impaired in vitro invasive ability was also restored in a similar manner (Fig. 4C), suggesting that the Rho signaling pathway and actomyosin system were linked in the transcellular migration of tumor cells through the host cell layer.
Active RhoA Induced Invasiveness in W1 Cells in Vitro and in Vivo-We have previously isolated a number of AH cell sublines showing different in vitro invasive ability from the same parental AH130 cells (6). A clone named W1 showed very low in vitro invasiveness even in the presence of serum or LPA, whereas MM1 demonstrated high invasiveness. Thus, we compared the response to LPA between MM1 and W1 cells. The treatment with 25 M of LPA increased the phosphorylation level of MAPK in a time-dependent manner peaking at 5 min in both cells (Fig. 5, A and B, top). However, W1 cells showed little change in the phosphorylation level of MLC20 by LPA stimulation (Fig. 5B, bottom), while MM1 cells showed increased MLC20 phosphorylation peaking at 30 min after stimulation with LPA (percentage of MLC20 phosphorylation was 11.6% at time 0 and 33.5% at 30 min after LPA stimulation; Fig. 5A, bottom). Moreover, the treatment of MM1 cells with 50 g/ml of C3 for 24 h completely abolished the change in phosphorylation level of MLC20 with LPA stimulation (data not shown). These observations suggested two notions: 1) both cells expressed the putative LPA receptors, and 2) W1 cells lacked Rho-actomyosin stimulation induced by LPA, which resulted in little invasiveness of these cells in vitro.
To try to focus on the second question, we introduced active human RhoA (V14RhoA and V14/I41RhoA) into W1 cells. We obtained several stable transfectants expressing 5-10% of active RhoA protein compared with the endogenous RhoA (Fig.  6A, top and middle). These transfectants also showed an increased phosphorylation level of MLC20 (Fig. 6A, bottom) as well as in vitro invasive ability in the absence of serum (Fig.  6B). Notably, the increased invasiveness of these transfectants was enhanced 3-4-fold by adding the serum into the assay medium, while mock transfectants represented little effect of serum stimulation (Fig. 6B). Furthermore, these active RhoA transfectants showed enhanced invasive ability after implantation into the peritoneal cavity similar to the MM1 transfectants (Table I). The active RhoA-transfected cells invaded more extensively into the peritoneum and formed more tumor nodules compared with the mock transfectants (data not shown). These results suggested that active RhoA expression with coincident MLC20 phosphorylation was sufficient to significantly enhance the invasiveness in vivo as well as in vitro in both cell types.
We also examined the effect of C3 on the morphology of these cells. Both V14RhoA and V14/I41RhoA W1 transfectants adhered to the culture dish with scattered shape. Treatment with C3 made V14RhoA transfectants round up; however, V14/ I41RhoA transfectants were resistant to the C3 treatment  (data not shown).
Constitutively Active RhoA Stimulated LPA-induced Endogenous RhoA Activation-Next, we tested the effect of C3 on in vitro invasiveness of both MM1 and W1 transfectants. As shown in Figs. 7A and 8A, both endogenous RhoA in the cell and expressed C3-sensitive active RhoA (V14RhoA) were effectively ADP-ribosylated 68 -96% by the treatment with 50 g/ml C3 for 24 h, whereas C3-insensitive RhoA (V14/I41RhoA) was not ADP-ribosylated, although it was present in the cell lysates as detected in the immunoblot (Figs. 7B and 8B, lane 3). In the absence of LPA, the C3 treatment greatly reduced the invasiveness of V14RhoA clone (p Ͻ 0.001 for both MM1 and W1), and the invasiveness of the V14/I41RhoA clone was resistant to the C3 treatment (p ϭ 0.04 for MM1; p ϭ 0.11 for W1) as expected (Figs. 7C and 8C, pairs of columns 3 and 5). The invasiveness of both transfectants was enhanced 2-4-fold in the presence of LPA (Figs. 7C and 8C, pairs of columns 4 and 6). Moreover, C3 inhibited the LPA-induced invasiveness in not only V14RhoA (p Ͻ 0.001 for both MM1 and W1) but also V14/I41RhoA clones (p ϭ 0.01 for MM1 and p ϭ 0.004 for W1), suggesting that V14/I41RhoA enhanced the invasiveness of both MM1 and W1 cells in two ways: 1) direct activation of the RhoA downstream cascade including MLC20 phosphorylation (C3-insensitive), and 2) promotion of LPA-induced endogenous RhoA activation (C3-sensitive), for which we postulate a positive feedback mechanism in RhoA activation. DISCUSSION In the present study, we have prepared the MM1 cells (a highly invasive cell line from parental AH130 cells) stably expressing FLAG-V14RhoA or FLAG-V14/I41RhoA. Although several lines of evidence suggested the involvement of Rho in cell proliferation, the active RhoA-transfected MM1 (V14RhoA and V14/I41RhoA) showed little change in growth (estimated doubling times from growth curve were 14.0 h for MM1 cells, 15.4 h for V14RhoA transfectants, and 15.4 h for V14/I41RhoA transfectants, respectively). By using rat fibroblasts (Rat-1) and mouse fibroblasts (NIH 3T3), human RhoA transfectants were found to show an increase in growth and saturation density (41), and the tumorigenicity of human RhoA-transfected Rat-1 cells was correlated with amplification and expression of RhoA (42). In addition, NIH 3T3 cell transfectants overexpressing normal or active RhoA proteins from Aplysia were also reported to have increased cell growth, and tumors were induced when inoculated into nude mice (16). Yamamoto et al. (43) indicated that the treatment with exoenzyme C3 caused inhibition of cell growth and accumulated in the G 1 phase of the cell cycle in Swiss 3T3 cells. Likewise, C3 also stopped the growth and differentiation of PC12 cells (44). In contrast, C3 treatment hardly affected the growth of both active RhoA transfected cells and mock transfectants during the period of in vitro invasion assay in the present study (data not shown). We suggest that the data, taken together, indicate that activation Cell lysates were separated on SDS-8% PAGE followed by immunoblotting with anti-phosphospecific MAPK polyclonal Abs (top). Blots were stripped and reprobed with anti-p44/42 MAPK polyclonal Abs to ensure equal protein loading (middle). At the bottom, cell lysates were separated on SDS-12% PAGE followed by immunoblotting with anti-PMLC20 polyclonal Abs. The horizontal axis of each graph represents the phosphorylation level of MAPK (arbitrary units) using NIH Image software, and the phosphorylation level of MLC20 was normalized using cultured endothelial cells as an internal control (for percentage of phosphorylation, see "Experimental Procedures"). A typical result from three independent experiments is shown.
FIG. 6. Induction of invasiveness in transfected W1 cells with active human RhoA. A, immunoblot analysis. Cell lysates from four different clones (two of V14RhoA and two of V14/I41RhoA) were separated on SDS-12% PAGE followed by immunoblotting with RhoA Abs. Blots were stripped and reprobed with anti-FLAG Abs. The level of MLC20 phosphorylation was semiquantitated using anti-PMLC20 polyclonal Abs. B, in vitro invasion assay. Cells (2 ϫ 10 5 cells/dish) were seeded on MCL in the absence or presence of FCS in the assay medium. In vitro invasion ability was calculated in the manner described for Fig.  2C. Results were representative of three independent experiments. of RhoA largely affects the stimulation of motility but not growth in AH cells. These controversial results of the function of Rho might be attributable to properties of the different cell types.
Using both MM1 and W1 cells (a low invasive counterpart of MM1 cells), we have shown here that the active RhoA transfectants show morphological changes such as adherence to the culture dishes, the ability for migration through MCL even in the absence of FCS or LPA, and the ability of invasiveness in vivo. It is interesting to note that control cells (mock transfectants) developed solid tumors after implantation into the peritoneal cavity, while active RhoA transfectants formed many more small tumor nodules disseminated in the peritoneum using both MM1 and W1 cells. These results strongly suggest that expressed active RhoA plays a pivotal role in the invasion of AH cells. This was further confirmed by the fact that the invasiveness of V14/I41RhoA transfectants showed little response to the exoenzyme C3, indicating that C3-insensitive expressed active V14/I41RhoA could drive the ability of invasiveness.
Next, we focused on how Rho regulates transmigration of AH cells in vitro and hypothesized that the stimulation of the actomyosin system induced by active RhoA leads to the subsequent transmigration of AH cells. The above speculation was supported by the following three findings. 1) Active RhoA transfectants enhanced MLC20 phosphorylation. 2) KT5926, a specific inhibitor of MLCK, greatly suppressed MLC20 phosphorylation as well as the transmigration of active RhoA transfectants. 3) The time-dependent decline of the effect of KT5926 restored the transmigration ability in these cells.
A series of reports support this hypothesis using different cell systems. Jalink et al. (25) reported that LPA-induced neurite retraction of NIE-115 and PC12 cells was abolished by the treatment with C3. They proposed that RhoA regulated LPA induced actin-myosin contractility through the balance between MLCK and MLC20 phosphatase activities employing KT5926 in these cells. Chrzanowska-Wodnicka and Burridge (26) reported that KT5926 blocked contraction, formation of stress fibers and focal adhesions, and tyrosine phosphorylation in LPA-stimulated fibroblasts. They also found that 2,3-butanedione-2-monoxime, an inhibitor of actin-myosin interactions, blocked active RhoA-induced formation of stress fibers, suggesting that activation of Rho results in change of cytoskeletal structures, promoted by MLC20 phosphorylation. In permeabilized pig aortic smooth muscle cells, GTP␥S-induced Ca 2ϩ stimulation of MLC20 phosphorylation has been reported to involve Rho (22). Gong et al. (23,24) reported that GTP␥S induced translocation of RhoA associated with Ca 2ϩ sensitization of rabbit portal vein smooth muscle. Using a cell-free system, Amano et al. (45) reported that GTP␥S-RhoA can activate Rho-kinase, and then activated Rho-kinase directly phosphorylates MLC20 at the site of Ser 19 in vitro. It is important to note that KT5926 could inhibit MLCK; however, this inhibitor had little effect on other kinases. Therefore, complete suppression of active RhoA-induced MLC20 phosphorylation by KT5926 was due to the inhibition of MLCK rather than Rhokinase, and the elevation of MLC20 phosphorylation induced by active RhoA was attributable to either increased MLCK activity or decreased MLC20 phosphatase activity. A number of putative Rho target proteins have been identified (46 -53). Among them, the multiple domains in Rho-associated kinase could interact with reorganization of the cytoskeleton (54,55). Most recently, we made MM1 cells stably expressing the constitutively active form of Rho-kinase. These cells showed a striking phenotype (including spreading and adhering on the plastic dish), increased in vitro invasiveness without LPA stimulation, and enhanced the phosphorylation level of MLC20. 4 LPA was a crucial regulator in the transmigration of MM1 cells through MCL (8). Conversely, W1 cells did not show transmigration upon serum (LPA) stimulation (6). We compared the responses to LPA, such as the phosphorylation levels of MAPK and MLC20 between MM1 and W1 cells. Both cells responded to LPA in respect to MAPK activation, suggesting the presence of putative LPA receptors in parental AH cells. We also detected significant elevation of the phosphorylation level of MLC20 in MM1 cells; however, W1 cells had little response to LPA concerning the phosphorylation level of MLC20. This implies lack of Rho-actomyosin stimulation induced by LPA in W1 cells, resulting in very low invasiveness of these cells. To clarify this possibility, we introduced active RhoA (V14RhoA or V14/I41RhoA) into W1 cells. Consequently, these transfectants showed enhanced MLC20 phosphorylation and a significant increase in invasiveness in the absence of LPA. Furthermore, the invasiveness of V14/I41RhoA transfectants in the absence of LPA showed little response to C3, suggesting that the invasiveness was largely mediated by C3insensitive expressed active RhoA.
It is interesting to note that active RhoA W1 transfectants responded to LPA, showing enhanced extensive invasiveness, while parental W1 cells represented little invasiveness even in the presence of LPA. We speculate that expressed active RhoAenhanced LPA-induced invasiveness via the activation of endogenous RhoA pathway. This hypothesis is supported by the following findings. In V14RhoA W1 transfectants, LPA-enhanced transmigration was C3 sensitive, indicating both endogenous and expressed RhoA activation (Fig. 8, pairs of columns 3 and 4). Conversely, in V14/I41RhoA W1 transfectants, transmigration was C3-insensitive in the absence of LPA, representing expressed RhoA-induced invasion (Fig. 8, pair of  columns 5). LPA enhanced the transmigration of these cells and this enhancement was C3-sensitive (p ϭ 0.004, Fig. 8, pair of columns 6), suggesting that C3-insensitive active RhoA (V14/ I41RhoA) stimulates endogenous RhoA activation, indicating a positive feedback mechanism. The major findings in this study were summarized as the proposed model in Fig. 9.