The Insert Region of RhoA Is Essential for Rho Kinase Activation and Cellular Transformation

Abstract

RhoA is involved in multiple cellular processes, including cytoskeletal organization, gene expression, and transformation. These processes are mediated by a variety of downstream effector proteins. However, which effectors are involved in cellular transformation and how these proteins are activated following interaction with Rho remains to be established. A unique feature that distinguishes the Rho family from other Ras-related GTPases is the insert region, which may confer Rho-specific signaling events. Here we report that deletion of the insert region does not result in impaired effector binding. Instead, this insert deletion mutant (RhoΔRas, in which the insert helix has been replaced with loop 8 of Ras) acted in a dominant inhibitory fashion to block RhoA-induced transformation. Since RhoΔRas failed to promote stress fiber formation, we examined the ability of this mutant to bind to and subsequently activate Rho kinase. Surprisingly, RhoΔRas-GTP coprecipitated with Rho kinase but failed to activate it in vivo. These data suggested that the insert domain is not required for Rho kinase binding but plays a role in its activation. The constitutively active catalytic domain of Rho kinase did not promote focus formation alone or in the presence of Raf(340D) but cooperated with RhoΔRas to induce cellular transformation. This suggests that Rho kinase needs to cooperate with additional Rho effectors to promote transformation. Further, the Rho kinase catalytic domain reversed the inhibitory effect of RhoΔRas on Rho-induced transformation, suggesting that one of the downstream targets of Rho-induced transformation abrogated by RhoΔRas is indeed Rho kinase. In conclusion, we have demonstrated that the insert region of RhoA is required for Rho kinase activation but not for binding and that this kinase activity is required to induce morphologic transformation of NIH 3T3 cells.

Rho proteins form a subgroup of the Ras superfamily of small GTPases that regulate a wide spectrum of cellular functions, including cytoskeletal organization, gene expression, and transformation (26, 43, 44, 54). The most studied Rho function is reorganization of the actin cytoskeleton (11, 28). Among all of the Rho proteins, the best-characterized members are RhoA, Rac1, and Cdc42, which induce stress fibers and focal adhesions, lamellipodia, and filopodia, respectively (30, 36, 37). RhoA also plays a critical role in cellular transformation. Coexpression of constitutively active RhoA together with weakly transforming Raf mutants promoted synergistic focus-forming activity (19, 34). In addition, RhoA possesses weak transforming activity by itself, both in cell culture and in nude mice, and overexpression of Rho has been reported during the advance of human tumors (9, 33).

Similarly to Ras, RhoA exerts its functions through its interaction with downstream effector proteins in a GTP-dependent manner. Since 1994, at least 14 putative RhoA effector proteins have been identified based on their selective binding to Rho-GTP versus Rho-GDP (5, 11, 17, 28, 54). Such a plethora of effectors may enable RhoA to signal through diverse pathways to fulfill its complicated cellular functions. However, it is still a challenge to identify which specific effector(s) is involved in the transformation pathway. Such information is indispensable for designing approaches to block RhoA transforming activity without affecting normal cellular functions.

Among the RhoA-GTP binding proteins, a family of serine/threonine kinases, Rho kinase/ROK/ROCK, was cloned and characterized as downstream effectors (14, 21, 22, 24). Rho kinase has been found to promote stress fiber formation, in cooperation with another RhoA effector protein, p140mDia (2, 28, 46). A Rho kinase inhibitor, Y-27632, specifically blocked the focus-forming ability of RhoA and Rho-GEFs, strongly suggesting an important role for Rho kinase in RhoA-mediated cellular transformation (39). However, constitutively active Rho kinase could only marginally cooperate with an activated Raf mutant in focus formation assays (39). These data suggested that although important, Rho kinase might not be the only mediator of Rho-induced aberrant growth.

Upon activation, Rho-GTP interacts with downstream effector proteins through its switch 1–effector-binding loop (38). However additional regions outside of the switch domains may also contribute to effector binding (10, 55). One domain that has received considerable attention is the insert region, a unique α-helical sequence that replaces loop 8 of other Ras family members (7, 12, 13, 48). The solution-accessible surface of the insert region is rich in charged residues with mobile side chains, making it a candidate region for effector binding (7, 12, 13, 48). Notably, unlike the switch domains that change conformation during GDP-GTP transition, the conformation of the insert region appears to be independent of nucleotide-bound status (13). Thus, interaction of the insert region with other cellular components, if any, may be GTP independent and may only occur after primary RhoA contacts have been established.

Previous studies have shown that the insert region is important for certain functions of Rac and Cdc42. First, the insert region of Rac is essential for activation of the neutrophil-NADPH oxidase complex, although it was not required for interaction between Rac and its target, the oxidase p67 subunit. Therefore, it was speculated that the insert region might interact with some membrane-associated protein (8, 29). Second, the insert region of Cdc42 was required for interaction with the effector protein, IQ-GAP (25). Third, the insert region of Cdc42 is critical for RhoGDI function (50). Finally, the insert regions of both Rac and Cdc42 have been shown to be essential for their transforming abilities but dispensable for other functions, such as JNK activation, actin cytoskeleton rearrangement, and focal adhesion formation (16, 51). Taken together, these data indicate that the insert region is essential for conferring unique functions on both Rac and Cdc42. However, neither the function of the RhoA insert region nor its role in cellular transformation is known.

To understand the mechanism of RhoA-induced transformation, we previously constructed two RhoA mutants, Rho-VA and RhoΔRas, whose transforming abilities were compromised (55). Rho-VA contained mutations in loop 6 that resulted in decreased interaction with the Rho effector proteins, Rho kinase and PRK2. On the other hand, RhoΔRas, in which the insert region of RhoA was replaced by the equivalent loop 8 of Ras, maintained its association with the isolated Rho-binding domains of both effector proteins in vitro. Our present studies were aimed at understanding why the insert region is required for RhoA-induced transformation. We found that deletion of the insert region did not significantly alter the effector-binding profile of Rho-GTP but compromised its ability to activate Rho kinase in vivo. Although the catalytic domain of Rho kinase (ROCK-CAT) failed to promote transformation of NIH 3T3 cells in the presence of Raf, it cooperated with RhoΔRas to increase focus formation. Together these data suggest that for RhoA, effector interaction and activation are separable events; that the insert region of RhoA is apparently involved only in the latter step; and that Rho kinase cooperates with additional Rho effectors to promote transformation.

MATERIALS AND METHODS

Plasmids.

pZIP-RhoA(63L), Rho-VA63L (a 63L 88V 90A triple mutant), RhoΔRas(63L) (Fig. 1), and Raf(340D) were described previously (55). These cDNAs were also subcloned into the hygromycin-resistant mammalian expression vector pCGN-HA (49) as BamHI fragments. Full-length Rho kinase and its isolated catalytic domain (residues 6 to 553) (4) were subcloned into pRK5 (20). pEF-HA-moesin was described previously (31). The moesin cDNA was kindly provided by S. Tsukita, Kyoto, Japan.

FIG. 1.

The insert region of RhoA. (A) Alignment between the RhoA insert region and loop 8 of H-Ras (13). Dashes indicate lack of insert region in Ras. (B) Sequence of RhoΔRas protein in the insert region. Normal type indicates RhoA residues, while outline type represents residues from H-Ras.

Expression of GST fusion proteins.

Glutathione S-transferase (GST)-fused Rho GTPases were expressed in the BL21(DE3)lysE strain of Escherichia coli essentially as described previously (35). Briefly, following a 3-h induction with 0.2 mM isopropyl-1-thio-d-galactopyranoside at 37°C, cells were pelleted and resuspended in lysis buffer (50 mM Tris, pH 7.6, 100 mM NaCl, 5 mM MgCl₂, 100 μM GDP, 0.5% NP-40, 1 mM dithiothreitol, 1.9 mg of aprotinin/ml, and 1 mM phenylmethylsulfonyl fluoride). After sonication to lyse the cells, the cleared supernatant was tumbled with glutathione-agarose beads (Sigma) at 4°C overnight. The beads were then washed three times with lysis buffer and stored at 4°C. The GTP binding capacity of each Rho mutant was determined as described previously, using [α-³²P]GTP (41). For binding experiments, Rho mutants were loaded with GDP or GTPγS as described previously (55).

Pull-down and coimmunoprecipitation assays.

For GST pull-down assays, COS cells were transfected with plasmids encoding Myc-tagged, full-length Rho kinase using LipofectAMINE (Life Technologies). Forty-eight hours posttransfection, the cells were lysed in Rho-binding buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 10% glycerol, 0.5% NP-40, 1 mM dithiothreitol, 1.9 mg of aprotinin/ml, and 1 mM phenylmethylsulfonyl fluoride). The cell lysate was then incubated at 4°C for 2 h with GST-Rho proteins that had been preloaded with either GTP or GDP. Precipitated Rho kinase was detected by Western blotting with anti-Myc antibody (Covance Research Products) using enhanced-chemiluminescence reagents (Amersham Pharmacia Biotech). For coimmunoprecipitation assays, COS cells were cotransfected with plasmids encoding Myc-tagged full-length Rho kinase and hemagglutinin (HA)-tagged Rho proteins. Forty-eight hours posttransfection, the cells were lysed in Rho-binding buffer and immunoprecipitated with anti-HA antibody (Berkeley Antibody Co./Covance) at 4°C overnight. Coimmunoprecipitation of Rho kinase was detected by Western blotting with anti-Myc antibody as described above.

Whole-cell lysate pull-down assay.

Ras-transformed NIH 3T3 cells were labeled with [³⁵S]methionine-cysteine (250 μCi/plate) (Pro-mix; Amersham Pharmacia Biotech) for 16 h in Dulbecco's modified Eagle's medium supplemented with 10% dialyzed (10 mM HEPES, pH 7.4) fetal bovine serum. The cells were lysed in 20 mM Tris, pH 7.4, 50 mM NaCl, 1 mM EDTA, 0.5% NP-40, 10% glycerol, 1 mM dithiothreitol, 1.9 mg of aprotinin/ml, and 1 mM phenylmethylsulfonyl fluoride and cleared by microcentrifugation (16,000 × g; 4°C). Cell lysate (300 μl) supplemented with 15 mM MgCl₂ and 50 μM GTP were incubated at 4°C for 2 h with 10 μg of glutathione-agarose bead-bound GTPases that had been preloaded with GTPγS. The beads were washed three times with 20 mM Tris, pH 7.4, 50 mM NaCl, 10 mM MgCl₂, 0.1% NP-40, 10% glycerol, and 1 mM dithiothreitol. After polyacrylamide gel electrophoresis (8% gel), the binding profile was detected by fluorography using Amplify (Amersham Pharmacia Biotech).

Rho kinase in vivo activity assay.

COS cells were cotransfected with plasmids encoding HA-tagged moesin and the indicated Rho proteins in the presence or absence of full-length Rho kinase. One day posttransfection, the cells were deprived of serum for 24 h. Cell proteins were precipitated with 10% trichloroacetic acid, and pellets were extracted three times with acetone. Following sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE; 8% gel) and Western blotting, the phosphorylation of Thr558 of moesin was detected with anti-pT558 antibody, which recognizes this Rho kinase phosphorylation site of moesin (31). The expression level of HA-moesin in protein precipitates was detected with anti-HA antibody.

NIH 3T3 cell transfection, stable cells, and focus assay.

For transformation assays, NIH 3T3 cells were transfected with various plasmids, using the calcium phosphate precipitation procedure (35). The cells were maintained in regular growth medium (Dulbecco's modified Eagle's medium supplemented with 10% donor calf serum [Life Technologies, Inc.] and penicillin-streptomycin), and transforming foci were fixed and stained with crystal violet after 12 to 16 days of culture. If desired, the Rho kinase inhibitor Y-27632 (Biomol) was added to 10 μM every 24 h from day 3. To create stable cell lines, NIH 3T3 cells were transfected with 0.5 μg of pCGN(HA) plasmids containing the RhoA mutants and selected on growth medium supplemented with 200 μg of hygromycin B (Roche Molecular Biochemicals)/ml. At least 100 colonies were pooled to generate stable cell lines, and focus assays were performed by plating cells at 5 × 10⁵/60-mm-diameter dish and scoring foci as outlined above.

Fluorescence microscopy of stress fibers.

Stable NIH 3T3 cell lines expressing RhoA(63L) or RhoΔRas(63L) were seeded on coverslips and grown in regular growth medium overnight. After 0% serum starvation for 18 h, the cells were washed with 1× phosphate-buffered saline (PBS) once and fixed in 4% paraformaldehyde in PBS for 10 min. Following permeabilization in 0.1% Triton X-100 in PBS, F-actin was stained with 400 nM fluorescein isothiocyanate (FITC)-phalloidin for 1 h and visualized with a fluorescence microscope (60× objective).

RESULTS

Deletion of the insert region impaired RhoA transforming activity.

The structures of many Ras superfamily members have been solved and found to contain very similar folds. However, a feature unique to the Rho subfamily of GTPases is a 13-amino-acid insert coincident with loop 8 of Ras. This insert region forms a highly charged α helix without dramatic disturbance of the global structure shared by other Ras proteins (13, 48). Due to its presence only in Rho family GTPases, the insert region has been proposed to form unique interactions with downstream effector proteins to confer Rac and Cdc42 functions (8, 16, 29, 51). To study the role of the RhoA insert region, we made a mutant, RhoΔRas, in which the insert was replaced by the equivalent Ras loop 8 (Fig. 1) (55), and examined the effect of this modification on RhoA signaling. This mutant was created to delete the insert helix without significantly disrupting Rho structure (55) and is equivalent to a previously characterized Cdc42 mutant (50).

In classical (transient) NIH 3T3 transformation assays, the GTPase-defective mutant, RhoA(63L), cooperated with a Raf(340D) mutant to produce transforming foci while RhoΔRas(63L) was ineffective at inducing NIH 3T3 cell transformation (Fig. 2A and B) (55). Another mutant, Rho-VA(63L), which contains point mutations in loop 6, was also partially defective, as previously described (55). These data suggested that without the insert region Rho is unable to induce morphologic transformation due to an inability to interact with or signal to a downstream target(s).

FIG. 2.

Mutation of the insert region of RhoA disrupts the transformation of NIH 3T3 cells. (A) NIH 3T3 cells were transfected with 1 μg of pZIP-Raf(340D) plus 2 μg of the indicated Rho mutants or control vector. Transforming foci were visualized by staining them with crystal violet after 12 to 16 days in culture. (B) Pooled data from six independent experiments performed in triplicate (mean ± standard error of the mean). (C) Stable NIH 3T3 cell lines expressing the indicated Rho mutants or vector control were cultured for 12 to 16 days. Foci were visualized by staining them with crystal violet. Representative data are shown from six independent experiments performed in triplicate.

To address the ability of our RhoA mutants to induce transformation in the absence of Raf(340D), stable cell lines were established that overexpressed the above-mentioned Rho proteins. Subsequently, these cells were plated at confluence and maintained in growth medium for approximately 14 days. As in the Raf cooperativity assay, cells stably expressing RhoΔRas(63L) formed far fewer foci than were observed with RhoA(63L) (Fig. 2C). Overall, the transforming activity of RhoΔRas(63L) was greatly impaired, either on its own or in cooperation with Raf.

RhoΔRas binds efficiently to putative effector proteins.

The solution-accessible surface of the insert region is rich in charged residues with mobile side chains, making it a candidate region for interacting with cellular proteins (13, 48). Therefore, one possible reason for the reduced transforming activity of RhoΔRas may be the loss of interaction with a specific effector or effectors. We previously demonstrated that RhoΔRas could bind to the isolated Rho-binding domains of Rho kinase and PRK2 or to full-length PRK2 in vitro (55). However, it was still possible that RhoΔRas is unable to bind to another effector protein(s) that is critical for mediating RhoA-induced transformation. If this hypothesis was correct, one or more proteins should be missing from the effector-binding profile of RhoΔRas compared to that of RhoA.

To test this hypothesis, NIH 3T3 cell proteins were metabolically labeled with [³⁵S]Met-Cys, and the ability of GST-Rho fusion proteins, immobilized on glutathione-agarose beads and preloaded with GTPγS, to extract Rho effectors from the cell lysate was examined. As indicated in Fig. 3, RhoA-GTPγS extracted from the cell lysate multiple putative effector proteins that did not associate with GST alone (or GST-Rac1-GTPγS) (H. Zong and L. A. Quilliam, unpublished data). It is clear that the Rho-VA mutant had significantly reduced interaction with the Rho-GTP-binding proteins. In contrast, although some RhoΔRas-precipitated bands appeared to have slightly decreased intensities, overall this mutant had a binding profile similar to that of Rho. The amounts of effector proteins precipitated by both GST-Rho and GST-RhoΔRas were dependent on the GTPase concentration (over a range of 1 to 10 μg of fusion protein), suggesting that the binding assay was not saturated (Zong and Quilliam, unpublished). These data suggested that the partial loss of Rho-VA transforming activity was likely due to a dramatic decrease in effector-binding ability. However, despite RhoΔRas having much weaker transforming ability than Rho-VA, it retained relatively intact effector-binding ability. While we cannot absolutely rule out the possibility that the reduction of interaction of RhoΔRas with one effector is critical for transformation, our data suggested that effector binding may not be the primary cause for the loss of transformation of this insert deletion mutant.

FIG. 3.

Deletion of the insert region did not alter the effector-binding profile of RhoA. NIH 3T3 cells were labeled with [³⁵S]methionine-cysteine for 16 h. The cell lysate was then incubated at 4°C for 2 h with 10 μg of glutathione-agarose bead-bound GTPases preloaded with GTPγS. After extensive washing, binding proteins were separated by SDS-PAGE, and the binding profile was visualized by fluorography. The arrows indicate multiple putative effector proteins that did not associate with GST alone (or GST-Rac1-GTPγS) that were extracted by RhoA-GTPγS from the cell lysate. The data are representative of four independent experiments.

RhoΔRas antagonizes RhoA-induced transformation.

While low-molecular-weight GTPase signaling is initiated by interaction with and subsequent activation of downstream effector proteins, it is possible that the interaction and activation events are separable. Since it appeared that RhoΔRas could still interact efficiently with all detectable effectors, the loss of transformation may instead be due to a lack of effector activation in the absence of the insert region. If so, RhoΔRas will bind to effector proteins without activating them, thereby sequestering their accessibility to wild-type (WT) RhoA. Consequently, RhoΔRas(63L) should function in a dominant-negative fashion to block RhoA signaling.

To test this hypothesis, a focus-forming assay was performed in which RhoA was cotransfected with empty vector or the Rho mutants indicated in Fig. 4A. The ratio between transfected RhoΔRas or Rho-VA and RhoA was 1:1 in the top row and 3:1 in the bottom row. Compared to the vector control, RhoΔRas effectively blocked RhoA focus-forming activity, even when expressed at an equimolar ratio with RhoA. In contrast, the Rho-VA mutant, which is defective in effector binding, showed no dominant-negative effect. Instead, it slightly enhanced focus formation by retaining weak transforming ability (Fig. 4A and B). Therefore, the role of the insert region would appear to be activation of effector proteins during signal transduction. The dominant-negative effect also proved that RhoΔRas(63L) maintained its structural integrity (it loaded with GTP and interacted with effectors) in vivo.

FIG. 4.

RhoΔRas dominantly inhibited RhoA-induced transformation. (A) NIH 3T3 cells were cotransfected with RhoA(63L) and Raf(340D) together with the indicated Rho(63L) mutants. The transfected RhoΔRas/RhoA plasmid ratio was 1:1 for the top row and 3:1 for the bottom row. In contrast to Rho-VA(63L), RhoΔRas(63L) inhibited Rho-induced focus formation in a dose-dependent manner. Representative data are shown from at least four independent experiments performed in triplicate. (B) Quantitation of the number of foci per dish. Colonies larger than 0.1-mm diameter were counted. The results represent the means + standard errors from four experiments.

RhoΔRas is defective at inducing stress fiber formation.

Based on the dominant inhibitory effect of RhoΔRas in the focus assay, it appeared that this mutant was defective in activating effector proteins, at least those involved in transformation. We therefore wished to identify which effector protein(s) cannot be activated by RhoΔRas. To address whether there is a defect in the ability of RhoΔRas to activate Rho kinase or mDia, which cooperate to produce Rho-induced stress fibers (2, 46, 47), we examined the organization of F-actin in serum-starved NIH 3T3 cells stably expressing activated Rho mutants by staining them with FITC-phalloidin. Cells stably expressing RhoA(63L) possessed thick, well-organized stress fibers (Fig. 5). In contrast, more than 90% of both the vector control and RhoΔRas(63L)-expressing cells had very few stress fibers. Therefore, our data strongly suggest that RhoΔRas is defective in activating Rho kinase and/or mDia. The expression level of transfected Rho(63L) or RhoΔRas(63L) was much lower than that of endogenous RhoA; therefore, RhoΔRas(63L) did not suppress basal stress fiber formation.

FIG. 5.

RhoΔRas is defective in promoting stress fiber formation. NIH 3T3 cells stably expressing the indicated Rho mutants or vector control were stained with FITC-phalloidin to visualize F-actin following 18-h serum starvation. Fifteen to 20 fields (30≈40 cells) were examined under 60× magnification on each slide. RhoA-expressing cells formed well-organized stress fibers, while more than 90% of vector control and RhoΔRas-expressing cells had fewer, disorganized stress fibers. No discernible difference in stress fiber quality was noted between control and RhoΔRas-expressing cells. The data are representative of three independent experiments.

RhoΔRas binds efficiently to full-length Rho kinase.

Because of the lack of stress fiber induction by RhoΔRas, we next examined the interaction between Rho kinase and RhoΔRas. We previously showed that both RhoA and RhoΔRas bound to the isolated Rho-binding domain of Rho kinase in vitro (55). However, it was possible that the in vivo interaction with full-length protein may differ from that of the isolated domain in vitro. Therefore, the coimmunoprecipitation of Rho kinase and activated RhoA(63L) proteins was attempted. COS cells were cotransfected with Myc-tagged full-length Rho kinase and HA-tagged RhoA(63L), RhoΔRas(63L), or vector control. As shown in Fig. 6A, Rho kinase can be efficiently coimmunoprecipitated by RhoΔRas as well as RhoA, suggesting that the insert region is dispensable for RhoA binding to Rho kinase in vivo. In our previous studies, we found that the 63L mutant of RhoA may artificially enhance the effector-binding abilities of Rho proteins (55). Therefore, an in vitro pull-down assay using WT RhoA proteins was also performed. Lysates from COS cells transfected with Myc-tagged full-length Rho kinase were incubated with immobilized GST-RhoA or GST-RhoΔRas that had been preloaded with either GDP or GTPγS. The binding of Rho kinase was detected with anti-Myc antibody. GST, GST-RhoA, and GST-RhoΔRas loaded with GDP all failed to pull down full-length Rho kinase (Fig. 6B). On the other hand, GST-RhoA and RhoΔRas loaded with GTPγS both coprecipitated the kinase efficiently. This suggested that the interaction of RhoΔRas with Rho kinase is not due to additional mutations in the Rho molecule. The interaction of RhoΔRas with the isolated Rho-binding domain of mDia, another effector involved in stress fiber formation, was also examined and appeared to be intact (Zong and Quilliam, unpublished).

FIG. 6.

RhoΔRas bound to full-length Rho kinase efficiently. (A) COS cells were cotransfected with plasmids encoding full-length, Myc-tagged Rho kinase and HA-tagged RhoA(63L), RhoΔRas(63L), or the control vector. Rho proteins were immunoprecipitated from cell lysates with anti-HA antibody, and the coimmunoprecipitation (Co-IP) of Rho kinase was detected by blotting with anti-Myc antibody (top). The lower blots show equal expression of Rho kinase or Rho proteins, respectively. (B) COS cells were transfected with full-length (Myc-tagged) Rho kinase. The cell lysate was incubated with immobilized GST, GST-RhoA, or GST-RhoΔRas preloaded with either GDP or GTPγS as indicated. Pull-down of Rho kinase was detected by blotting with anti-Myc antibody following SDS-PAGE. The data are representative of two independent experiments.

RhoΔRas is defective in Rho kinase activation in vivo

After confirming that RhoΔRas could efficiently interact with effector proteins, the ability of RhoΔRas to activate Rho kinase was investigated using an in vivo kinase assay. It has previously been shown that the ERM protein moesin is a substrate for Rho kinase (31). Oshiro et al. used a phospho-moesin-specific antibody, anti-pT558, to demonstrate that introduction of activated RhoA or Rho kinase into COS cells results in increased phosphorylation of moesin at Thr558 (31). Therefore COS cells were cotransfected with RhoA(63L), RhoΔRas(63L), or a vector control along with HA-tagged moesin, with or without WT Rho kinase. Following serum starvation, the cells were harvested in 10% trichloroacetic acid, and the phosphorylation of moesin by Rho kinase was detected with the anti-pT558 antibody described above. Compared to the vector control, RhoA(63L) significantly increased the phosphorylation of HA-moesin in both the presence and absence of cotransfected WT Rho kinase (Fig. 7A and B). In contrast, RhoΔRas(63L) did not significantly stimulate moesin phosphorylation. Therefore, RhoA(63L), but not RhoΔRas(63L), can effectively activate Rho kinase, suggesting that the insert region is indispensable in the Rho kinase activation process.

FIG. 7.

Rho kinase activation in vivo. (A) RhoΔRas is defective in Rho kinase activation in vivo. COS cells were cotransfected with HA-moesin, HA-Rho, and full-length Rho kinase, as indicated. Following 24-h serum starvation, cellular proteins were precipitated with 10% trichloroacetic acid and separated by SDS-PAGE. Phosphorylation of Thr558 of moesin by Rho kinase was detected with a phosphospecific anti-pT558 antibody (top). The lower gels show equal loading of the HA-moesin and HA-Rho proteins, respectively, in each lane. The data are representative of at least three independent experiments. (B) Quantitation of the phosphorylation level of moesin. The density of phospho-moesin bands was divided by that of HA-moesin bands, and the value of RhoA stimulation was arbitrarily set to 100% for comparison of different assays. The results represent the means ± standard errors from three experiments. (C) COS cells were cotransfected as indicated, and assays were performed as described above. Y-27632 was added (+) in the indicated dishes at 10 μM final concentration in the beginning of serum starvation and 4 to 6 h before harvesting. The results represent the means ± standard errors from two experiments.

Although overexpression of WT Rho kinase did not dramatically increase moesin phosphorylation in serum-starved cells (Fig. 7B), ROCK-CAT, the constitutively activated catalytic domain of Rho kinase, efficiently phosphorylated moesin T558 (Fig. 7C). Further, both RhoA- and ROCK-CAT-induced moesin phosphorylation was abolished by pretreating cells with a specific Rho kinase inhibitor, Y-27632 (27). These data confirmed that Rho kinase is responsible for RhoA-induced moesin phosphorylation and that such a system should be valid for measuring in vivo Rho kinase activation by Rho proteins. Attempts to demonstrate differential activation of Rho kinase by Rho versus RhoΔRas in vitro were unsuccessful due to difficulty in activating recombinant, immunoprecipitated Myc-tagged Rho kinase with RhoA-GTPγS (unpublished data).

ROCK-CAT enhances Rho-induced transformation and alleviates the inhibition of RhoA-induced transformation by RhoΔRas.

It was previously reported that the Rho kinase inhibitor Y-27632 could specifically block RhoA- or Rho-GEF-induced transformation (39), an observation that we have independently confirmed (Zong and Quilliam, unpublished). In addition, an activated Rho kinase fragment was found to have very weak transforming activity when coexpressed with Raf (39). Since RhoΔRas is defective in Rho kinase activation (Fig. 7) and transformation (Fig. 2), we sought to determine if activated Rho kinase could enhance the transforming ability of RhoΔRas. In our hands, the constitutively active catalytic domain of Rho kinase (ROCK-CAT) was unable to transform NIH 3T3 cells, either alone or when coexpressed with Raf(340D) (Fig. 8A). However, ROCK-CAT enhanced focus formation when cotransfected with RhoΔRas(63L) in the presence of Raf(340D) (Fig. 8), while Y-27632 was found to totally abolish this effect (Zong and Quilliam, unpublished). This confirmed the positive role of Rho kinase in RhoA-induced transformation. Further, since ROCK-CAT only enhanced transformation in the presence of RhoΔRas(63L), an additional RhoA effector protein(s) (which can be, at least weakly, activated by RhoΔRas) is likely to be involved in the transformation process.

FIG. 8.

The involvement of Rho kinase in RhoA-induced transformation. (A) ROCK-CAT, the constitutively active catalytic domain of Rho kinase, enhances RhoΔRas transformation. NIH 3T3 cells were cotransfected with plasmids encoding RhoΔRas(63L) plus activated Rho kinase (ROCK-CAT) in the presence of Raf(340D). Following 17 days in culture, foci were visualized by staining them with crystal violet. Representative data are shown from at least four independent experiments performed in duplicate. (B) Quantitation of the number of foci per dish. Colonies larger than 0.1-mm diameter were counted. The results represent the means + standard errors of duplicate samples in at least four experiments. *, P < 0.001 compared to mock transfection (paired Student t test).

The above findings suggested that RhoΔRas acts as a dominant inhibitor of Rho-induced transformation at least in part by binding to Rho kinase and preventing its activation by WT RhoA (Fig. 6 and 7). This is consistent with the data of Sahai et al., who demonstrated that Rho kinase activity was required for Rho-induced focus formation (39). If the major effect of RhoΔRas is to block Rho kinase access to RhoA, preventing its subsequent activation, then introduction of the constitutively active Rho kinase catalytic domain (ROCK-CAT) construct should rescue the dominant inhibitory effect of RhoΔRas. To test this hypothesis, NIH 3T3 cells were cotransfected with RhoA(63L) and Raf(340D), along with a vector control; RhoΔRas(63L); or RhoΔRas(63L) plus ROCK-CAT. Compared to the vector control, the expression of RhoΔRas blocked RhoA-induced focus formation (Fig. 9A and B) as seen above (Fig. 4). However, when RhoΔRas and ROCK-CAT were coexpressed, the inhibition of transformation by RhoΔRas was significantly reversed (Fig. 9A and B). These data strongly suggested that RhoΔRas blocks RhoA-induced transformation by sequestering Rho kinase and that the addition of constitutively active Rho kinase can rescue this inhibitory effect, as depicted in Fig. 9C.

FIG. 9.

ROCK-CAT alleviates the inhibition of RhoA-induced transformation by RhoΔRas. (A) NIH 3T3 cells were cotransfected with plasmids encoding RhoA(63L) and Raf(340D), along with the indicated additional proteins. Foci were stained with crystal violet following 12 to 16 days in culture. Compared to the vector control, the expression of RhoΔRas blocked RhoA-induced focus formation. However, when RhoΔRas and ROCK-CAT were cointroduced into the cells, the inhibition of transformation by RhoΔRas was significantly reversed. Representative data are shown from four independent experiments performed in duplicate. +, present; −, absent. (B) Quantitation of the number of foci per dish. Colonies larger than 0.1-mm diameter were counted. The results represent the means + standard errors of duplicate samples from four experiments. *, P < 0.01 compared to mock transfection; **, P < 0.01 compared to the number of foci in the presence of ROCK-CAT (paired Student t test). (C) Diagram to illustrate how ROCK-CAT may alleviate the inhibition of RhoA-induced transformation by RhoΔRas.

DISCUSSION

Rho effector binding and activation are separable events, and the latter requires the insert region.

Similarly to other Ras family proteins, RhoA exerts its biological effects by associating with downstream effector proteins in a GTP-dependent manner. However, it is not well understood how this interaction is translated into effector activation. Considerable research interest has been focused on the role of the insert region, a 13-amino-acid sequence unique to Rho family members, in Rho signaling. The insert region has been proposed to be involved in Rho-effector interaction (7, 12, 13, 48). However, a lack of conformational change during the GDP-GTP transition suggests that interaction between this region and effector proteins, if any, is secondary or constitutive.

Previous studies have shown that the insert region of Rac is required for NADPH oxidase activation but not for interaction with its p67 subunit and that the insert region of Cdc42 is required for RhoGDI function (8, 29, 50). However, the mechanism by which the insert region influences signaling remains elusive. Additionally, the above-mentioned studies were limited to Rac1 and Cdc42, while the function of RhoA's insert region was not investigated. To address this issue, we constructed an insert mutant of RhoA, RhoΔRas, in which the insert region was replaced with the equivalent loop 8 of Ras. Our data demonstrated that loss of the insert region severely impairs the transforming ability of RhoA. This was consistent with previous studies that reported a requirement for the insert domain for Rac- or Cdc42-induced transformation (16, 51). Assuming that loss of transformation was due to a loss of effector interaction, we set out to identify Rho effector proteins that failed to associate with RhoΔRas-GTP. To our surprise, RhoΔRas retained the ability to bind to multiple Rho-interacting proteins in NIH 3T3 cell lysates, suggesting that loss of the insert domain may affect a process distal to effector binding. Therefore, the focus of our investigation shifted to understanding the effector activation process.

RhoΔRas could block RhoA-induced focus formation, suggesting that it competed for but did not activate a Rho effector(s) that is required for transformation. Consistent with this observation, we found that although RhoΔRas could coimmunoprecipitate with Rho kinase, it was an inefficient activator of its kinase activity in vivo. From these findings we concluded that RhoA effector interaction and activation are separable events and demonstrated that the insert region is involved in the activation step but not the interaction step. During the course of our studies, it was reported that the insert region of Cdc42 is similarly not required for phospholipase D binding but is essential for its activation (45). Thus, the insert domain may play a similar role for each of the Rho family GTPases. However, Karnoube et al. have recently suggested that the insert domain of Rac1 may be functionally distinct from those of Rho and Cdc42 (18). Further investigation of Rho family protein effector regulation is therefore required to determine the universality of the role of the insert domain.

The role of Rho kinase in RhoA-induced transformation.

To date, it is not known how many, or indeed which, effector proteins are involved in RhoA-induced transformation. It has been shown that the Rho kinase inhibitor Y-27632 can block Rho- or Rho-GEF-induced focus formation in NIH 3T3 cells (39). However, since the constitutively active catalytic domain of Rho kinase only marginally cooperated with Raf in these focus assays (39), the positive contribution of Rho kinase to transformation demanded further study. Here we demonstrated that one of the defects of RhoΔRas is its inability to activate Rho kinase, supporting the potential role of this kinase in Rho-induced transformation. In our hands, ROCK-CAT, the constitutively active catalytic domain of Rho kinase, failed to cooperate with Raf to cause focus formation. However, it significantly enhanced cellular transformation when coexpressed with RhoΔRas. These data not only confirmed the positive role of Rho kinase in the transformation process but suggest that for Rho kinase to promote focus formation, it needs to cooperate with an additional Rho effector(s) whose activity has been augmented by the cotransfected RhoΔRas, as depicted in Fig. 9C.

Since mDia can cooperate with Rho kinase to induce well-organized stress fibers, we postulated that mDia may also be required to induce transformation. We found that the Dia autoregulatory domain of mDia2, which binds to and activates endogenous mDia (1), did not increase focus formation when transfected either alone or together with ROCK-CAT or Rho mutants (Zong and Quilliam, unpublished). These data suggested that an effector(s) other than mDia might be responsible for mediating Rho-induced transformation. Since the insert region of Cdc42 is important for the activation of phospholipase D activation and RhoA can also activate phospholipase D (42, 45), it will be interesting to examine the ability of RhoΔRas to activate this enzyme. Phospholipase D has mitogenic activity (23), and it would also be interesting to determine if it contributes to RhoA-induced transformation.

Although we and others (39) have found that Rho kinase is essential for RhoA-induced transformation, inactivation of Rho kinase is required for Ras-induced cellular transformation (15). This apparent discrepancy is likely due to the fact that Ras must uncouple Rho from Rho kinase to reduce stress fiber formation and increase cell motility (40). Rho does contribute to Ras-induced transformation by suppressing p21^cip levels, but this event is not mediated by Rho kinase. In contrast, binding to Rho kinase and regulation of the actin cytoskeleton are essential for RhoA to induce transformation (38).

How does the Rho insert region activate effector proteins?

Although our data suggest that the insert region is involved in the effector activation process in vivo, its exact role remains to be established. There are several possible mechanisms by which the insert region could influence Rho kinase activation. To activate many of its targets, RhoA, similarly to other Ras family GTPases, facilitates a conformational change in effector proteins from a closed and autoinhibited to an open and activated state (6). For example, the COOH-terminal portion of Rho kinase containing the Rho-binding domain interacts with the NH₂-terminal catalytic domain to inactivate its kinase activity. Association of Rho-GTP with the COOH-terminal portion overcomes this inhibition (3). The protruding insert region may activate Rho kinase by steric repulsion, separating the autoinhibitory and catalytic domains. Alternatively, transient salt bonds formed between charged residues in the RhoA insert region and Rho kinase may reduce the energy required for the kinase to adopt an active conformation. Finally, the insert region may play a role in the subcellular localization of Rho and its effectors. Since it has been shown that Rho effectors can be activated by arachidonic acid, phosphoinositides, and cardiolipin (32, 52, 53), membrane translocation will also allow the access of effectors to these polar lipid coactivators.

In conclusion, we have shown that although the insert domain is not required for RhoA to bind to Rho kinase, it is essential for its subsequent activation. Thus, effector binding and activation may be separable events mediated by distinct regions of Rho. Our studies also support and extend previous work implicating the involvement of Rho kinase in Rho-induced transformation (38, 39). While Rho kinase did not efficiently cooperate with Raf to promote transformation (reference 39 and this study), we report here that it does cooperate with transformation-defective Rho mutants. Thus, Rho kinase needs to synergize with additional Rho effectors to promote cellular transformation. Due to the unique properties of RhoΔRas, it is likely that this mutant will be a useful tool to gain further understanding of Rho effector activation and to determine which additional effectors are involved in the transformation process. Further, this study suggests that inhibiting Rho effector activation, rather than just Rho binding, might be a fruitful approach for drug design.