Serum-soluble factors play a dominant role in the activation of the small GTPase RhoA. Cell adhesion also modulates RhoA activity but the effect is modest in the absence of serum. Here, we show that cell adhesion is required for serum-stimulated Rho signal transduction leading to myosin light chain (MLC) phosphorylation. Characterization of Rho-kinase substrates revealed that diphosphorylation of MLC at Thr-18 and Ser-19 (ppMLCT18/S19) and phosphorylation of the myosin-binding subunit (MBS) of myosin phosphatase at Thr-853 (pMBST853) were mostly Rho and Rho-kinase dependent in attached fibroblasts. MLC monophosphorylation at Ser-19 (pMLCS19) was partially dependent on Rho kinase, whereas phosphorylation of MBS at Thr-696 (pMBST696) and phosphorylation of CPI-17 at Thr-38 (pCPI-17T38) were mostly Rho-kinase independent. Cell detachment caused a significant reduction in pMLCS19 and a more dramatic decrease of ppMLCT18/S19 without inhibiting RhoA. pMBST853, pMBST696 and pCPI-17T38 were not significantly reduced, suggesting that myosin-phosphatase activity was little changed. Cells expressing active RhoA (RhoAV14) or Rho-kinase catalytic domain maintained elevated pMBST853 upon detachment but failed to support ppMLCT18/S19, indicating that the ability of Rho kinase to phosphorylate MLC is impaired. Reattachment to immobilized fibronectin resulted in a gradual recovery of Rho-kinase-induced ppMLCT18/S19 that is absent from the cells attached to poly-L-lysine. The convergence of signals from soluble factors and cell adhesion might therefore occur at the point of MLC phosphorylation, providing an effective mechanism for dynamic control of contractility during cell migration.
The small GTPase RhoA plays an important role in contractile force generation, cytoskeletal organization, cell migration, DNA synthesis and cell growth (reviewed in Hall, 1998; Ridley, 2001). The Rho-associated kinase (Rho kinase; also named ROK or ROCK), one of the RhoA effectors, directly phosphorylates myosin light chain (MLC) primarily at Ser-19 and, to a lesser extent, Thr-18 (Amano et al., 1996), producing both mono- and diphosphorylated MLC (pMLCS19 and ppMLCT18/S19, respectively). Rho kinase also inhibits myosin phosphatase by phosphorylation of its myosin-binding subunit (MBS; also named MYPT1) at Thr-696 and Thr-853 (numbering based on the human sequence) (Kimura et al., 1996; Feng et al., 1999; Kawano et al., 1999; Velasco et al., 2002), and by phosphorylation of CPI-17 (protein-kinase-C-potentiated phosphatase inhibitor of 17 kDa) at Thr-38 (pCPI-17T38) (Eto et al., 1995; Eto et al., 1997; Koyama et al., 2000). Therefore, Rho kinase promotes MLC phosphorylation by direct phosphorylation and by inactivation of myosin phosphatase (reviewed in Hartshorne, 1998; Kaibuchi et al., 1999). Phospho-MLC facilitates activation of the ATPase activity of myosin II, which is crucial for contraction in non-muscle cells (reviewed in Sellers, 1991; Goldman, 1998).
Cell adhesion to extracellular matrix proteins such as fibronectin (FN) not only provides an essential physical support for contraction and migration but also regulates some of the serum-stimulated signaling pathways to control cytoskeletal organization, cell migration and cell growth (reviewed in Schwartz et al., 1995; Miyamoto et al., 1998; Howe et al., 2002; Miranti and Brugge, 2002; Schwartz and Ginsberg, 2002). Specifically, cell adhesion to FN regulates RhoA activity, being inhibitory or stimulatory depending on the spreading stage (Ren et al., 1999; Ren et al., 2000; Arthur and Burridge, 2001). Modulation of Rho GTP loading by adhesion is important because it might provide dynamic control of contractile activity and cytoskeletal reorganization during migration. However, RhoA GTP loading is mainly stimulated by serum-soluble factors, and cell adhesion alone only has a modest effect on RhoA activity. It has been shown that interaction of GTP-Rac with its effector, p21-activated kinase (PAK), is dependent on cell adhesion (del Pozo et al., 2000). This prompted us to test whether cell adhesion could have more impact on RhoA downstream signaling than on GTP loading.
No specific marker of Rho effector activation had, however, been established. The Rho-kinase activity assayed in an in vitro setting, unlike that for PAK, does not change readily in response to Rho GTP loading or hydrolysis. In the presence of excessive GTP-loaded RhoA, the activity of Rho kinase (purified by chromatography or immunoprecipitation) is only increased up to twofold (Ishizaki et al., 1996; Leung et al., 1996; Matsui et al., 1996). In addition, MBS and MLC, two known substrates of Rho kinase, have multiple phosphorylation sites and some of them might not be Rho-kinase specific. Using the recently available phosphoantibodies against several Rho-kinase substrates, we have characterized the phosphorylation status of MLC, myosin phosphatase and CPI-17. We found that ppMLCT18/S19 and pMBST853, but not pMBST696 or pCPI-17T38, correlated well with RhoA and Rho-kinase activation status in attached fibroblasts. Cell detachment showed significant and different effects on ppMLCT18/S19 and pMBST853. Cell-matrix adhesion plays a permissive role in soluble-factor-stimulated Rho signal transduction that controls the contractile activity.
Materials and Methods
Reagents and cell culture
Antibodies against MLC (sc-9449 and sc-15370), ppMLCT18/S19 (sc-12896) and pCPI-17T38 (sc-17560R) were purchased from Santa Cruz Biotechnologies. The pMLCS19 antibody was obtained from the Cell Signaling Technology. Antibodies against pMBST696 (catalog number 07-251), pMBST853 (catalog number 36-003) and CPI-17 (catalog number 07-344) were from the Upstate Group. The rabbit antiserum against MBS (PRB-457C) was purchased from the Covance Research Products. The Y-27632 compound (BioMol) was stored as 10 mM stock solution in water and ML-7 (BioMol) as 25 mM stock solution in dimethylsulfoxide (DMSO). All other chemical reagents were purchased from Sigma unless otherwise stated. The adult human dermal fibroblasts were purchased from BioWhittaker; passages 5-12 were used in this study. Swiss 3T3 cells (CCL-92), NIH 3T3 cells (ATCC CRL-1658) and human HT-1080 fibrosarcoma cells (CCL-121) were purchased from the American Type Culture Collection. Cells were maintained in DMEM supplemented with 10% fetal bovine serum (FBS) (Hyclone). Suspension culture of cells was performed in Petri dishes coated with 1% agarose. FN (20 μg ml–1) and poly-L-lysine (PLL; 20-100 μg ml–1) were immobilized onto the Falcon® tissue culture dishes in PBS at 4°C overnight. The dishes were then blocked with 2% heat-denatured lipid-free bovine serum albumin for 60 minutes at room temperature and rinsed three times with PBS before use.
Plasmid construction and cell transfection
The RhoAV14 insert was isolated from the pcDNA3-RhoAV14 plasmid described previously (Ren et al., 1999) and cloned into the pEGFP-C1 vector (BD BioScience Clontech) to make pEGFP-RhoAV14. The cDNA fragment coding for the Rho-kinase catalytic domain (amino acids 6-553) was isolated by cutting the pEF-Bos-myc-Rho-kinase plasmid (Amano et al., 1997) with BamHI and EcoRV, and cloned into a modified pIRES2EGFP vector (BD BioScience Clontech) that encodes an N-terminal hemagglutinin (HA) tag. The full-length Rho kinase was also subcloned into the same vector. Effectene or SuperFect (Qiagen) was used for transfection of HT-1080 or NIH 3T3 cells, respectively, using the protocols recommended by the manufacturer. Transfection efficiencies were about 40-50% (data not shown).
Purification of recombinant C3 exoenzyme
The cDNA fragment encoding the C3 exoenzyme was isolated from the pGEX2T-C3 (Ren et al., 1999) and cloned into the pQE30 vector (Qiagen), which has an N-terminal six-histidine tag. Recombinant C3 was purified using the Ni-NTA affinity column (Qiagen) according to manufacturer's protocol. After elution, the protein solution was passed through a Sephadex® G25 (Pharmacia) gel-filtration column to change the buffer to PBS. C3 was introduced into the cells using the Lipofectamine reagent (Invitrogen) as described previously (Renshaw et al., 1996).
Western blotting and immunofluorescence microscopy
PVDF membranes (BioRad) were used in western blotting. Antibody binding was detected using the SuperSignal Duro reagents (Pierce-Endogen) and recorded under the ChemiImager-4400 low light digital image system (Alpha Innotech). Images were inverted for better visualization. Densitometry analyses were performed using the AlphaEase Software (Alpha Innotech). For immunofluorescence staining, cells on glass coverslips were fixed with 4% methanol-free formaldehyde for 10 minutes, permeabilized with 0.4% Triton X-100 in PBS for 4 minutes, and blocked with 10% normal serum in PBS. Anti-vinculin antibody (Sigma) was used to stain focal adhesions, followed by Alexa-Fluor-488-labeled anti-mouse IgG (Molecular Probes). Actin filaments were stained with rhodamine-phalloidin (Molecular Probes). Pictures were taken under a Nikon Microphot-FXA microscope using Kodak Elite chrome ASO-400 slide films, and digitized using a Nikon 2900 slide scanner.
RhoA activity assay
The RhoA activity assay was carried out as previously described (Ren et al., 1999; Ren and Schwartz, 2000) using 50-70 μg GST-Rhotekin Rho-binding domain (RBD) per reaction and a milder lysis buffer (50 mM Tris, pH 7.2, 1% Triton X-100, 100 mM NaCl, 5 mM MgCl2, 10 μg ml–1 each of aprotinin and leupeptin, and 1 mM phenylmethylsulfonylfluoride). This is because, under certain conditions, the RIPA lysis buffer described previously (Ren et al., 1999) gave artificially higher readings of GTP-Rho that were inconsistent with Rho effector activation (X.-D.R, unpublished data).
Detection of phosphorylated MLC, MBS and CPI-17
Cells were washed with cold PBS and collected in 12.5% trichloroacetic acid (TCA). The acid precipitates were spun down, washed twice with cold acetone, air dried and dissolved in sodium-dodecyl-sulfate polyacrylamide-gel electrophoresis (SDS-PAGE) sample buffer containing dithiothreitol. Samples were resolved in SDS gels and duplicated membranes were separately blotted with anti-MLC and anti-phospho-MLC antibodies. The MLC phosphorylation levels were normalized with the total MLC levels. The phosphorylation of MBS at Thr-696 and Thr853, and phosphorylation of CPI-17 at Thr-38 were determined in a similar way using the specific antibodies.
In vitro Rho-kinase assay
MLC-encoding cDNA was amplified from a human fibroblast mRNA preparation by reverse-transcription polymerase chain reaction (RT-PCR) and cloned into pGEX2T vector at BamHI and EcoRI sites. The oligonucleotides used for PCR are 5′-CGGGATCCTCGAGCAAAAAGGCAAAGACC-3′ and 5′-GGAATTCAGTCATCTTTGTCTTTGGC-3′, with the cloning sites underlined. The coding sequence was confirmed by DNA sequencing to be identical to the GenBank accession number BC004994. GST-MLC was purified using glutathione beads, and desalted in a Sephadex G25 column. HT-1080 cells transiently expressing HA-tagged full-length Rho kinase was extracted, and Rho kinase precipitated using anti-HA antibody coupled to agarose beads (Santa Cruz Biotechnologies). Kinase reaction was carried out essentially as described using 5 μg GST-MLC per reaction (Amano et al., 1996). The reaction was stopped by adding SDS sample buffer and the proteins resolved by SDS-PAGE. The top portion, containing the Rho kinase, was transferred to PVDF membrane and probed with anti-HA antibody. The lower portion containing the radioactive GST-MLC was stained, dried and exposed to film.
ppMLCT18/S19 is dependent on serum, Rho and Rho-kinase in human fibroblasts
Human dermal fibroblasts were serum starved for various times and collected for the analyses of ppMLCT18/S19 by western blotting. We found that serum starvation induced a dramatic decrease of ppMLCT18/S19. A significant reduction of ppMLCT18/S19 was observed after 5 minutes of starvation. By 15 minutes, ppMLCT18/S19 dropped by ∼80% (Fig. 1A) and was maintained at that level for up to 24 hours (data not shown). Conversely, when the serum-starved cells were stimulated with 10% serum, ppMLCT18/S19 was increased five- to tenfold (Fig. 1B). These data indicate that serum-soluble factors play a central role in maintaining high levels of ppMLCT18/S19.
The change in ppMLCT18/S19 after serum starvation or stimulation was accompanied by a parallel change of RhoA activity (data not shown). To determine how much ppMLCT18/S19 was dependent on the Rho effector Rho kinase in these cells, we used a Rho-kinase inhibitor, Y-27632 (Narumiya et al., 2000), and compared with a MLC-kinase inhibitor, ML-7 (Bain et al., 2003). Treatment of cells with 10 μM Y-27632 in the presence of 10% serum for 30 minutes inhibited ppMLCT18/S19 by approximately 95% (Fig. 1C), accompanied by complete loss of stress fibers and focal adhesions (Narumiya et al., 2000) (Fig. 2D). By contrast, ML-7 had a minimal effect on the level of ppMLCT18/S19 in the presence of serum (Fig. 1C), without noticeable effect on the actin stress fibers (data not shown). The level of ppMLCT18/S19 in ML-7-treated cells was 95.1±5.3% (mean±s.e.m., n=4) of that in DMSO (vehicle control)-treated cells. Similar results were obtained in Swiss 3T3 cells (data not shown). ML-7 inhibited more than 50% of ppMLCT18/S19 in the HT-1080 cells (data not shown), indicating a cell-type variation. The basal level of ppMLCT18/S19 in serum-starved human fibroblasts was also dependent on Rho kinase (Fig. 1C). In serum-starved cells, both ML-7 and DMSO inhibited ppMLCT18/S19 by about 50% (Fig. 1C), accompanied by partial loss of stress fibers and focal adhesions (data not shown).
We also treated human fibroblasts with C3 exoenzyme, a bacterial toxin that specifically inactivates Rho (Aktories et al., 1992). Treatment of cells with C3 exoenzyme abolished the basal and serum-stimulated production of ppMLCT18/S19 (Fig. 1D). Therefore, ppMLCT18/S19 in human fibroblasts is mostly dependent on serum, Rho and Rho kinase.
ppMLCT18/S19 and pMLCS19 are cell-adhesion dependent
The ability of serum to stimulate ppMLCT18/S19 is greatly diminished when cells are held in suspension. We found that the level of ppMLCT18/S19 decreased by about 80% once the cells were detached. Although cells might have been starved during the 5-minute trypsinization, adding 10% serum to the detached cells for up to 10 minutes did not increase ppMLCT18/S19 levels (Fig. 2A). Therefore, the level of ppMLCT18/S19 is both serum and adhesion dependent.
Because pMLCS19 is closely related to ppMLCT18/S19 and is also crucial for contractility, we measured pMLCS19 for comparison. As for ppMLCT18/S19, the pMLCS19 level dropped after either a 30-minute serum starvation or suspension for 3 minutes in the presence of serum (Fig. 2B). However, the decrease was limited to about 50%. This result is inconsistent with the previous report that the level of pMLCS19 was slightly higher in suspended cells (Ren et al., 1999). It is possible that, in the earlier study of Rho activities, more cells were used (to achieve higher assay sensitivity) that sometimes resulted in too high a density in the adherent condition. That might account for the lower Rho activity and pMLCS19 level, because cell-cell adhesion was later found to inhibit Rho (Noren et al., 2000). Alternatively, the variance in the cell types, antibodies and detection methods might also account for the differences.
We further compared the effect of Y-27632 on the levels of pMLCS19 and ppMLCT18/S19. Treatment of cells with Y-27632 inhibited ∼90% of ppMLCT18/S19 within 3 minutes and disrupted the cytoskeleton between 6-13 minutes (Fig. 2C,D), consistent with the notion that Rho-induced contraction drives formation of these structures (Chrzanowska-Wodnicka and Burridge, 1996). pMLCS19 also decreased quickly but the change was not as dramatic, leaving about 35% after 30 minutes of treatment compared with time 0 (Fig. 2C). These results indicate that pMLCS19 is only partially dependent on Rho kinase and cell adhesion, and its regulation possibly involves other kinases.
Minimal inhibition of Rho-kinase upon cell detachment
Because ppMLCT18/S19 is most Rho and Rho-kinase dependent in these cells, we measured RhoA activity before and after detachment. Interestingly, although the ppMLCT18/S19 level decreased significantly upon detachment, RhoA activity was transiently elevated (Fig. 3A). To assess whether Rho kinase is active under this condition, we characterized another Rho-kinase substrate, the MBS of myosin phosphatase. Consistent with the previous reports (Kawano et al., 1999; Velasco et al., 2002), phosphorylation of MBS at Thr853 (pMBST853) is highly sensitive to Y-27632 (Fig. 3B). The pMBST853 level dropped significantly (50-70%) after a 30-minute serum starvation but was marginally decreased (92.9±7.1%, mean±s.e.m.; n=9) when the cells were detached and incubated in 10% serum for 3 minutes (Fig. 3C). The level of ppMLCT18/S19 in the same samples decreased dramatically, as expected (data not shown). Y-27632 inhibited pMBST853 equally well in suspended cells (Fig. 3B). Therefore, the high levels of pMBST853 in detached cells indicate that Rho kinase remains active. We also confirmed that Rho kinase from the detached cells is intrinsically active towards MLC, because the immunoprecipitated Rho kinase efficiently radiolabeled MLC (Fig. 3D). Two possibilities remain: either Rho kinase could not effectively phosphorylate MLCT18/S19 in suspended cells or ppMLCT18/S19 is dephosphorylated more rapidly by myosin phosphatase after detachment.
pMBST696 and pCPI-17T38 levels are not reduced after detachment
To estimate the MLC phosphatase activity, we measured the phosphorylation level of MBS at Thr-696. We found that the pMBST696 level was unchanged after cell detachment and remained high after serum starvation for 30 minutes or Y-27632 treatment for 30 minutes (Fig. 4A). Prolonged treatment with Y-27632 or starvation for 24 hours had no further effect on pMBST696 (data not shown). In the COS-7 cells, however, Y-27632 treatment for 30 minutes decreased the pMBST696 level by 52±4.5% (mean±s.e.m., n=3; Fig. 4A, right), indicating that the antibody is effective.
We also found that the level of phosphorylation of CPI-17 at Thr-38 (pCPI-17T38) was not reduced after either detachment or Y-27632 treatment or serum starvation (Fig. 4B). The pCPI-17T38 was increased about twofold after stimulation of PMA (Fig. 4B, right), consistent with the previous report that protein kinase C increases pCPI-17T38 levels (Eto et al., 1995; Eto et al., 1997). These results suggest that there is no significant increase of myosin phosphatase activity upon detachment.
Reduction of active Rho-kinase-induced ppMLCT18/S19 level upon detachment
To rule out the possibility that a minor decrease in the level of pMBST853 might substantially increase the phosphatase activity, we transiently transfected cells with the active RhoA RhoAV14 and the Rho-kinase catalytic domain (CAT). For better transfection efficiency, we performed experiments in HT-1080 cells. In adherent cells, both RhoAV14 and CAT induced approximately twofold increases of ppMLCT18/S19 levels compared with the pEGFP-transfected control cells (Fig. 5A). When the transfected cells were detached, however, ppMLCT18/S19 levels decreased significantly in the RhoAV14-expressing cells (Fig. 5A). Importantly, RhoV14, full-length Rho kinase and their combination all maintained higher levels of pMBST853 in suspended cells but were unable to sustain ppMLCT18/S19 in detached cells. The CAT-expressing cells had substantially higher pMBST853 after detachment yet ppMLCT18/S19 levels still decreased significantly (Fig. 5B). These data indicate that detachment has different effects on the two Rho-kinase substrates and that Rho kinase appears to be unable to phosphorylate MLC efficiently in detached cells.
Gradual recovery of ppMLCT18/S19 levels during cell spreading on FN
When the cells were allowed to readhere to FN in the presence of 10% serum immediately after detachment, the level of ppMLCT18/S19 was not fully recovered until at least 30 minutes after adhesion (Fig. 6A). During the early stages of attachment and spreading, ppMLCT18/S19 was at low levels even though pMBST853 was relatively high (Fig. 6A). Therefore, the ability of Rho kinase to phosphorylate MLC is recovered slowly during adhesion and spreading. In the absence of serum, the elevation of ppMLCT18/S19 was much slower. Significant increases in the level of ppMLCT18/S19 were not observed until about 60 minutes after attachment (data not shown). To test whether the increase of ppMLCT18/S19 is FN specific, we plated cells onto PLL-coated dishes. To reduce the deposit of FN from serum and to quench the adhesion-dependent signals completely, cells were suspended in 1% serum (instead of 10%) for 60 minutes before reattachment. As shown in Fig. 6B (top), ppMLCT18/S19 levels increased gradually in cells on FN but remained low in cells on PLL throughout the time course. Expression of RhoAV14 or CAT did not bypass the requirement for FN (Fig. 6B, bottom) but did accelerate the increase of ppMLCT18/S19, resulting in premature contraction and reduced spreading (data not shown). These results indicate that Rho-kinase-dependent ppMLCT18/S19 requires integrin-mediated cell adhesion.
We found that the ppMLCT18/S19 level is mostly dependent on Rho and Rho kinase in human fibroblasts. Although MLC kinase, citron kinase and Dlk/ZIP kinase are also able to induce MLC diphosphorylation (Murata-Hori et al., 1999; Somlyo and Somlyo, 2000; Kamm and Stull, 2001; Murata-Hori et al., 2001; Yamashiro et al., 2003), they might be more important in another cell type (e.g. smooth muscle cells or tumor cells) or in special situations (e.g. cell division or apoptosis) (Klemke et al., 1997; Kogel et al., 1998; Madaule et al., 1998; Page et al., 1999). In vitro MLC phosphorylation of Ser-19 by MLC kinase is believed to be 100 times more efficient than at Thr-18 (Ikebe and Hartshorne, 1985; Ikebe et al., 1986). However, the in vivo concentration of diphosphorylated MLC can be as high as the monophosphorylated MLC after RhoA activation by microtubule disruption (Kolodney and Elson, 1995; Ren et al., 1999). This might partly be because Rho kinase phosphorylates Thr-18 more readily (Amano et al., 1996). Diphosphorylated MLC induces higher ATPase activity and is at least as effective as pMLCS19 at inducing in vitro contraction (Ikebe and Hartshorne, 1985; Ikebe et al., 1986; Umemoto et al., 1989). These results underline the significance of ppMLCT18/S19 in contraction regulation.
We also confirmed that pMBST853 is highly specific to Rho kinase and sensitive to Y-27632. However, pMBST696 levels are only slightly reduced by Y-27632 in human fibroblasts. The antibody is able to detect an increase of pMBST696 levels in COS-7 cells expressing active Rho kinase (Riento et al., 2003) and a significant decrease of pMBST696 levels in the same cells after Y-27632 treatment for 30 minutes (this study), indicating that the antibody is effective. That pMBST696 levels are not responsive to Y-27632 is additionally observed in smooth-muscle cells (Takizawa et al., 2002b; Niiro et al., 2003). It is possible that other kinases (such as Dlk/ZIP kinase, PAK, Raf and myotonic dystrophy protein kinase) are more important in MBST696 phosphorylation. Although phosphorylation of MBST696 by Rho kinase and others results in inhibition of the phosphatase activity (Kimura et al., 1996; Feng et al., 1999; MacDonald et al., 2001; Muranyi et al., 2001; Broustas et al., 2002; Takizawa et al., 2002a), pMBST696 levels are unchanged upon detachment. Similarly, we did not observe a decrease in pCPI-17T38 levels after either detachment or Y-27632 treatment. In addition, the PMA-induced increase of pCPI-17T38 levels is accompanied by diminished MLC phosphorylation and disassembly of the actin cytoskeleton, probably the result of Rho inhibition (X.-D.R., unpublished data). This might be due to the low-level expression of CPI-17 in non-muscle cells (Eto et al., 1997). These results indicate that pMBST696 and pCPI-17T38 do not play an important role in regulating rapid changes of contraction downstream of Rho kinase in our fibroblasts.
Because, after detachment, we did not observe a decrease of pMBST696 or pCPI-17T38 levels, two known indicators of myosin phosphatase activity, our data argue that cell detachment impairs the ability of Rho kinase to phosphorylate MLC, and the remaining phospho-MLC is quickly dephosphorylated by the basal phosphatase activity. We do not rule out the possibility that detachment might also increase the rate of dephosphorylation via an unidentified mechanism. Studies of the physical interactions between Rho kinase and its substrates, and between myosin phosphatase and MLC will help to elucidate the mechanism in more depth.
Our data clearly indicate that soluble factors and cell adhesion might converge their signals at MLC phosphorylation. This regulatory mechanism is probably important in the dynamic control of contraction during cell migration. Cell migration is a multistep process that can be divided into membrane-protrusion formation, cell-body contraction and tail detachment (Lauffenburger and Horwitz, 1996; Horwitz and Parsons, 1999; Ridley, 2001). Contractile force, as indicated by the phospho-MLC level, is mainly stimulated by serum-soluble factors. However, contraction must be dynamically regulated during migration. For example, contraction is preferably suppressed at the leading edge to facilitate the formation of membrane protrusions. Once the new adhesions are stabilized in the front, strong contractile force is required to pull the cell body forward. When the tail is detached, excessive contraction needs to be prevented. Previous work indicates that cell adhesion might regulate contraction by modulating Rho GTP loading (Ren et al., 1999; Danen et al., 2000; Ren et al., 2000; Arthur and Burridge, 2001; Cox et al., 2001). In this study, we found that detachment could cause rapid, significant inhibition of Rho-dependent MLC phosphorylation without affecting RhoA activity. The inhibition is strong and effective – the same scale of inhibition in adherent cells would require an ∼80% decrease of RhoA activity. In addition, the ability of Rho kinase to phosphorylate MLCT18/S19 recovers gradually during cell spreading, a membrane-protrusion-formation process. Therefore, this adhesion-dependent mechanism might be involved in preventing excessive contraction after the tail retraction, as well as during membrane protrusion formation at the leading edge.
Although pMBST853 level is not reduced immediately after detachment, prolonged (60 minutes) incubation of cells in suspension caused ∼50% inhibition of pMBST853 in the presence of serum, suggesting a reduced activation of Rho kinase. Expression of RhoAV14 did not rescue the pMBST853 level under this condition (X.-D.R., unpublished data). These results are reminiscent of the adhesion-dependent activation of PAK by Rac (del Pozo et al., 2000). Further investigation is required to assess whether there is a qualitative difference between Rho-to-Rho-kinase and Rac-to-PAK signaling. It is worth mentioning that ppMBST18/S19 levels are decreased much more readily than are pMBST853 levels after cell detachment, so it might be more important in regulating rapid changes of contraction during cell migration.
This work was supported in part by grants from the National Institute of Arthritis and Musculoskeletal and Skin Diseases (AR47894), and the Dermatology Foundation to X.-D. R. We thank D. Bar-Sagi and M. Schwartz for encouragement and discussions.