p160 RhoA-binding kinase ROKalpha induces neurite retraction.

We previously reported that the activation of prostaglandin E receptor EP3 subtype caused neurite retraction via small GTPase Rho in the EP3B receptor-expressing PC12 cells (Katoh, H., Negishi, M., and Ichikawa, A. (1996) J. Biol. Chem. 271, 29780-29784). However, a potential downstream effector of Rho that induces neurite retraction was not identified. Here we examined the morphological effect of p160 RhoA-binding kinase ROKalpha, a target for RhoA recently identified, on the nerve growth factor-differentiated PC12 cells. Microinjection of the catalytic domain of ROKalpha rapidly induced neurite retraction similar to that induced by microinjection of a constitutively active Rho, RhoV14, whereas microinjection of the kinase-deficient catalytic domain of ROKalpha did not induce neurite retraction. This morphological change was observed even though C3 exoenzyme, which was known to inactivate Rho, had been preinjected. On the other hand, microinjection of the Rho-binding domain or the pleckstrin homology domain of ROKalpha inhibited the EP3 receptor-induced neurite retraction. These results demonstrate that ROKalpha induces neurite retraction acting downstream of Rho in neuronal cells.

We previously reported that the activation of prostaglandin E receptor EP3 subtype caused neurite retraction via small GTPase Rho in the EP3B receptor-expressing PC12 cells ( However, a potential downstream effector of Rho that induces neurite retraction was not identified. Here we examined the morphological effect of p160 RhoA-binding kinase ROK␣, a target for RhoA recently identified, on the nerve growth factor-differentiated PC12 cells. Microinjection of the catalytic domain of ROK␣ rapidly induced neurite retraction similar to that induced by microinjection of a constitutively active Rho, Rho V14 , whereas microinjection of the kinase-deficient catalytic domain of ROK␣ did not induce neurite retraction. This morphological change was observed even though C3 exoenzyme, which was known to inactivate Rho, had been preinjected. On the other hand, microinjection of the Rho-binding domain or the pleckstrin homology domain of ROK␣ inhibited the EP3 receptor-induced neurite retraction. These results demonstrate that ROK␣ induces neurite retraction acting downstream of Rho in neuronal cells. Rho, a member of the Ras superfamily of small GTPases, is implicated in various cellular morphological functions, such as formation of stress fibers and focal adhesion (1), cell motility (2), cytokinesis (3), cell aggregation (4), and smooth muscle contraction (5). When cells are activated by extracellular stimuli, inactive GDP-bound Rho is converted to active GTP-bound Rho. Once activated, Rho probably interacts with its specific targets, leading to a variety of biological functions (6). Recently, several target proteins that interact only with GTP-bound Rho have been identified, including p128 protein kinase N (7,8), p160 RhoA-binding kinase ROK␣ (9) also known as its bovine counterpart Rho-kinase (10) or its mouse counterpart ROCK-II (11), rhophilin (7), rhotekin (12), and p140mDia (13). Among them, ROK␣ has been reported to be involved in several functions of Rho: the regulation of myosin phosphorylation (14,15), the formation of stress fibers and focal adhesions (16,17), and probably the regulation of cytokinesis (18).
Rho has also been implicated in the control of neuronal cell morphologies. The activation of a certain heterotrimeric GTPbinding protein (G-protein)-coupled receptor, 1 such as lysophosphatidic acid and thrombin receptors, caused the rapid retraction of extended neurites in several neuronal cell lines (19 -21). Clostridium botulinum C3 exoenzyme, which specifically ADP-ribosylates Rho and suppresses the actions of Rho (22,23), inhibits the receptor-mediated neurite retraction (24,25), indicating that Rho activity is required for this morphological change. Although this effect appears to be induced by the contractility of the actin-based cytoskeleton (24,26), a downstream effector of Rho that induces neurite retraction has not yet been identified.
We previously reported that the activation of prostaglandin EP3 receptor caused Rho-dependent neurite retraction in the NGF-differentiated PC12 cells expressing the EP3B receptor (27), one of the EP3 receptor isoforms isolated from bovine adrenal medulla (28). In non-neuronal cells, the activation of EP3 receptor stimulates the Rho-mediated formation of stress fibers (29), indicating that EP3 receptor is a potent activator of Rho in various cell types. In this report, we have examined the putative role of ROK␣ in the EP3 receptor-mediated neurite retraction in the NGF-differentiated PC12 cells. We show that ROK␣ is involved in the EP3 receptor-mediated neurite retraction and that the activation of ROK␣ is sufficient for inducing neurite retraction.

EXPERIMENTAL PROCEDURES
Materials-M&B28767 was a generous gift from Dr. M. P. L. Caton of Rhone-Poulene Ltd. NGF 2.5S was purchased from Promega Corporation, and C. botulinum C3 exoenzyme was from Seikagaku Kogyo (Tokyo, Japan). The sources of the other materials are shown in the text.
Expression and Purification of Recombinant Proteins-The coding region of human RhoA was generated by reverse transcriptionpolymerase chain reaction (PCR) from HeLa cells using primers 5Ј-C-TGGACTCGAATTCGTTGCCTGAGCAATGG-3Ј and 5Ј-GCAAGATGA-ATTCTGATTTGTAATCTTAGG-3Ј. The PCR product was digested with EcoRI, cloned into the pBluescript KS(ϩ), and completely sequenced. cDNAs for RhoA V14 and RhoA V14A37 were generated by PCRmediated mutagenesis (30), subcloned into the BamHI/EcoRI sites of pGEX-4T-2 vector, and sequenced. Recombinant RhoA V14 and RhoA-V14A37 were expressed as glutathione S-transferase (GST) fusion proteins in Escherichia coli, and purified on glutathione-Sepharose beads according to the method of Self and Hall (31). The sequence encoding ROK␣ containing the catalytic domain (CD-ROK␣, amino acids 1-543), the Rho-binding domain (RBD-ROK␣, amino acids 932-1065), and the pleckstrin homology (PH) domain (PHD-ROK␣, amino acids 1116 -1379) were generated by reverse transcription-PCR from PC12 cells, using primers 5Ј-ATGAGCGGATCCCCGCCGACGGGGAAA-3Ј and 5Ј-ACCTTCTGAATTCATATCTGAGAGCTCTGG-3Ј for CD-ROK␣, 5Ј-GACGGATCCAAAGAGAAGATCATGAAAGAGC-3Ј and 5Ј-GTTGT-GTGAATTCTTAACGTTCAG-3Ј for RBD-ROK␣, and 5Ј-TCGCAGGGA-TCCGCCTTGCATATTGG-3Ј and 5ЈTCTTGTGGATGGAAGAATTCGA-TCACCTTC3Ј for PHD-ROK␣, respectively. The kinase-deficient mutant of CD-ROK␣ (CD-ROK␣ K112G ) was generated by PCR-mediated mutagenesis. All PCR products for each domain of ROK␣ were cloned into the pCR2.1 vector and sequenced completely. The PCR products for RBD-ROK␣ and PHD-ROK␣ were subcloned into the BamHI/EcoRI sites of pGEX-4T-2 vector, and recombinant proteins were expressed as GST fusion proteins in E. coli and purified on glutathione-Sepharose beads. The PCR products for CD-ROK␣ and CD-ROK␣ K112G were subcloned into the BamHI/EcoRI sites of pAcG2T vector, and recombinant proteins were expressed as GST fusion proteins in Sf9 cells with Bacu-loGold system (PharMingen) and purified on glutathione-Sepharose beads according to the method of Matsui et al. (10). All recombinant proteins were dialyzed with an injection buffer (10 mM Tris-HCl, pH 7.6, 150 mM NaCl, 2 mM MgCl 2 , and 0.1 mM dithiothreitol) at 4°C overnight for microinjection. Protein concentration was determined by comparing with bovine serum albumin standards after electrophoresis on a SDSpolyacrylamide gel and staining with Coomassie Brilliant Blue.
Cell Culture and Microinjection-The EP3B receptor-expressing PC12 cells (27) were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 5% horse serum, 4 mM glutamine, 100 units/ml penicillin, and 0.2 mg/ml streptomycin under humidified conditions in 95% air and 5% CO 2 at 37°C. For microinjection, cells were seeded on poly-D-lysine-coated (Sigma) 35-mm dishes, which were marked with a cross to facilitate the localization of injected cells. After the cells had been differentiated in serum-free Dulbecco's modified Eagle's medium containing 50 ng/ml NGF and 20 M indomethacin for 3 days, microinjection was performed using an IMM-188 microinjection apparatus (Narishige, Tokyo, Japan). During microinjection, the differentiated cells were maintained in Hepes-buffered Dulbecco's modified Eagle's medium, pH 7.4, at 37°C. Cells were photographed at ϫ 200 magnification under a phase contrast microscope. For the quantitative examination shown in Fig. 3, neurite-retracted cells were defined as the cells that retracted by more than 10% of their original length within 30 min of the addition of the agonist or of the microinjection of recombinant proteins. The percentages of neurite-retracted cells were calculated by counting at least 30 cells in the same field. Data were obtained from triplicate experiments.

RESULTS AND DISCUSSION
In our previous study, we obtained evidence that M&B28767, an EP3 agonist, induced neurite retraction in the EP3B receptor-expressing PC12 cells and that this morphological change was completely inhibited when the cells were microinjected with C3 exoenzyme, which ADP-ribosylates and inactivates Rho (22,23), indicating that the EP3B receptor induced neurite retraction through the activation of Rho (27). To determine whether activation of Rho is sufficient for inducing neurite retraction in the PC12 cells, we microinjected a constitutively activated recombinant RhoA, RhoA V14 into the NGF-differentiated PC12 cells and examined its effect. As shown in Fig. 1 (C  and D), microinjection of Rho V14 into the cytoplasm caused retraction of the neurites within 30 min. More than 70% of the injected cells retracted their neurites (Fig. 3). This morphological change was similar to that stimulated by M&B28767 (Fig.  1, A and B). The neurite-retracted cells by microinjection of RhoA V14 was not stained with trypan blue (data not shown), indicating that they did not undergo cell death. On the other hand, RhoA V14 containing a T37A substitution in the effector region, RhoA V14A37 , had no effect on the differentiated cells after microinjection (Fig. 1, E and F, and Fig. 3), suggesting that this mutation blocked the interaction of Rho with its target to induce neurite retraction. This result also indicated that there were not any nonspecific effects due to the microinjection itself.
Previous studies suggested that the generation of actinbased contractile forces was required for neurite retraction (24,26). Among several targets of Rho, ROK␣ appears to participate in Rho-dependent contractile events, such as the formation of stress fibers (16,17) and the regulation of cytokinesis (18). By Northern blot analysis, ROK␣ was expressed in the NGF-differentiated PC12 cells (data not shown). Therefore, we examined whether ROK␣ was involved in the Rho-mediated neurite retraction in the NGF-differentiated PC12 cells. ROK␣ contains the catalytic domain in its amino terminus, the coiledcoil domain, the Rho-binding domain, and the PH domain in its carboxyl terminus (9). It was recently shown that the truncation mutant of ROK␣ containing the catalytic domain displayed constitutive kinase activity without the addition of active form of Rho, whereas the Rho-binding domain and the PH domain of ROK␣ served as dominant negative forms of the kinase (16,17). Based on these characters, we generated recombinant proteins containing these domains: the catalytic domain of ROK␣ (CD-ROK␣, amino acids 1-543), the kinase-deficient mutant of CD-ROK␣ (CD-ROK␣ K112G ), the Rho-binding domain of ROK␣ (RBD-ROK␣, amino acids 932-1065), and the PH domain of ROK␣ (PHD-ROK␣, amino acids 1116 -1379). To examine the effects of these domains of ROK␣ on neurite retraction, we microinjected these recombinant proteins into the NGF-differentiated cells and analyzed their morphologies.
After the cells had been microinjected with CD-ROK␣, they rapidly retracted their neurites within 30 min (Fig. 2, A and B,  and Fig. 3). This morphological change was similar to that induced by microinjection of RhoA V14 . The neurite-retracted cells by microinjection of CD-ROK␣ was not stained with trypan blue (data not shown), indicating that they did not undergo cell death. On the other hand, microinjection of the kinase-deficient mutant of CD-ROK␣ mutant CD-ROK␣ K112G had no effect (Fig. 2, C and D, and Fig. 3), indicating that the kinase activity of CD-ROK␣ was required for inducing neurite retraction. When the cells were microinjected with C3 exoenzyme, the M&B28767-induced neurite retraction was completely inhibited (Fig. 2, E and F). However, the CD-ROK␣induced neurite retraction was not inhibited after the cells had been microinjected with C3 exoenzyme (Fig. 2, G and H), indicating that CD-ROK␣ acted downstream of Rho.
Next we microinjected RBD-ROK␣ or PHD-ROK␣, which served as dominant negative forms of ROK␣, into the differentiated cells and examined each effect on the M&B28767-induced neurite retraction. When the cells had been microinjected with RBD-ROK␣ or PHD-ROK␣, the M&B28767induced neurite retraction was inhibited (Fig. 4). All the cells microinjected with RBD-ROK␣ or PHD-ROK␣ had no response to M&B28767. These results suggest that ROK␣ is involved in the EP3 receptor-mediated neurite retraction in the PC12 cells. Taken together, our results suggest that ROK␣ induces neurite retraction acting downstream of Rho in the NGF-differentiated PC12 cells.
Recently, ROK␣ was shown to be involved in Rho-induced formation of stress fibers and focal adhesion in other cell types such as fibroblasts. However, the organization of stress fibers induced by constitutively active ROK␣ was apparently different from that induced by lysophosphatidic acid or constitutively active Rho (16,17), suggesting that additional signals were required for Rho-induced stress fiber formation. In this study, however, microinjection of CD-ROK␣ sufficiently induced neurite retraction similar to that induced by Rho V14 even though C3 exoenzyme had been preinjected, whereas CD-ROK␣ K112G failed to induce neurite retraction (Figs. 2 and 3), suggesting that the increase in the kinase activity of ROK␣ by Rho appears to be sufficient for inducing neurite retraction. Because myosin-binding subunits of myosin phosphatase and myosin light chain are known to be substrates for ROK␣ and activation of ROK␣ leads to phosphorylation and activation of myosin (14,15), neurite retraction may be induced by ROK␣mediated regulation of myosin phosphorylation. In addition, it was recently reported that glial fibrillary acidic protein, an intermediate filament protein expressed in the cytoplasm of astroglia, was identified as another substrate for ROK␣ (18). Therefore, we will consider substrate(s) of this kinase for neurite retraction in future studies. Until now, we have obtained evidence that the activation of EP3B receptor, coupling to Rho activation, did not affect the NGF-induced mitogen-activated protein kinase activation in the PC12 cells (data not shown), suggesting that the activation of Rho or ROK␣ did not inhibit the NGF-induced signaling to the Ras-mitogen-activated protein kinase pathway. To examine the direct effect of Rho or ROK␣ on the NGF-induced differentiation, we are currently establishing PC12 cell lines that express RhoA V14 or CD-ROK␣ under the control of an inducible promoter.
As shown in Fig. 4, two fragments of ROK␣, the Rho-binding domain and the PH domain, served as dominant negative forms of ROK␣ in the EP3 receptor-mediated neurite retraction, as  3. Quantification of effects of M&B28767, Rho, and ROK␣ on neurite retraction. After the cells had been differentiated with NGF for 3 days, they were exposed to 1 M M&B28767 or microinjected with the recombinant proteins of the indicated mutants of Rho or ROK␣. The percentages of neurite-retracted cells were determined 30 min after the addition of the agonist or the microinjection of the proteins, as described under "Experimental Procedures." Rho V14 or Rho V14A37 was injected at 1 mg/ml, and CD-ROK␣ or CD-ROK␣ K112G was injected at 2 mg/ml. Data are the means Ϯ S.E. of triplicate experiments.

Neurite Retraction Induced by ROK␣ 2491
reported for the formation of stress fibers and focal adhesion (16,17). ROK␣ has been shown to be translocated to peripheral membranes upon transfection with Rho V14 (9). Because PH domains are supposed to play a key role in localization of molecules to the specific target regions in the membranes, the PH domain of ROK␣ may localize this kinase at the specified region in response to the EP3 receptor-induced activation of Rho, and this translocation of ROK␣ to its target region seems to be essential for inducing neurite retraction. On the other hand, RBD-ROK␣ may block the interaction between endogenous Rho and ROK␣. We also showed that RhoA V14A37 , a mutant at the effector region, lost the ability to induce neurite retraction in the differentiated PC12 cells (Fig. 1, E and F, and  Fig. 3). Indeed, RhoA V14 bound to the RBD-ROK␣, but RhoA V14A37 did not (data not shown). This defect in binding to ROK␣ seems to be the reason for the inability of RhoA V14A37 to induce neurite retraction.
In conclusion, we have here shown that ROK␣ is an essential component of Rho-mediated neurite retraction in neuronal cells. Considering that ROK␣ is enriched in the brain (16), ROK␣ may play a critical role in the regulation of neuronal cell morphology in the brain. However, many questions have not yet been elucidated in this field, for example how the G-protein coupled receptor activates Rho. Further investigations are necessary to understand Rho-mediated signal transduction in neuronal cells.