Phosphorylation of Glial Fibrillary Acidic Protein at the Same Sites by Cleavage Furrow Kinase and Rho-associated Kinase*

Site- and phosphorylation state-specific antibodies are useful to analyze spatiotemporal distribution of site-specific phosphorylation of target proteins in vivo . Using several polyclonal and monoclonal antibodies that can specifically recognize four phosphorylated sites on glial fibrillary acidic protein (GFAP), we have previously reported that Thr-7, Ser-13, and Ser-34 on this intermediate filament protein are phosphorylated at the cleavage furrow during cytokinesis. This observation suggests that there exists a protein kinase named cleavage furrow kinase specifically activated at metaphase-anaphase transition (Matsuoka, Y., Nishizawa, K., Yano, T., Shibata, M., Ando, S., Takahashi, T., and Inagaki, M. (1992) EMBO J. 11, 2895–2902; Sekimata, M., Tsujimura, K., Tanaka, J., Takeuchi, Y., Inagaki, N., and Inagaki, M. (1996) J. Cell Biol. 132, 635–641). Here we report that GFAP is phosphorylated specifically at Thr-7, Ser-13, and Ser-34 by Rho-associated kinase (Rho-kinase), which binds to the small GTPase Rho in its GTP-bound active form. The kinase (7:3) contain- ing 0.1% trifluoroacetic acid. Elution was carried out with a 60-min linear gradient of 5–50% 2-propanol/acetonitrile followed by a further 10-min linear gradient of 50–80% 2-propanol/acetonitrile at a flow rate of 0.8 ml/min. Other Procedures— All the procedures for immunoblotting have been described elsewhere in detail (8, 21). Immunofluorescence microscopy was performed as described previously (8). Amino acid sequences were analyzed with an ABI 476A gas-phase sequencer. Two-dimensional phosphoamino acid analysis was performed as described (27). Electron microscopy was carried out as described (28).

Intermediate filaments (IFs), 1 major components of the cy-toskeleton and the nuclear envelope in most eukaryotic cells, undergo dramatic reorganization of their structure during cell signaling and cell cycle (for review, see Refs. [1][2][3]. This IF reorganization is thought to be regulated by site-specific phosphorylation of IF proteins at serine and threonine residues, and several protein kinases have been shown to act as IF kinases in vivo (for review, see Ref. 4). Site-and phosphorylation state-specific antibodies that recognize a phosphorylated serine/threonine residue and its flanking sequence can visualize site-specific IF phosphorylation and thereby IF kinase activities in situ by immunocytochemistry (Ref. 5; for review, see Ref. 6). Recently, we reported two distinct types of mitotic phosphorylation of glial fibrillary acidic protein (GFAP), an IF protein expressed in the cytoplasm of astroglia, using antibodies that react with four distinct phosphorylated sites on GFAP (7,8). One type is the phosphorylation of Ser-8 on GFAP, which appeared at G 2 -M phase transition in the entire cytoplasm. The other type is the phosphorylation of Thr-7, Ser-13, and Ser-34, which appeared at metaphase-anaphase transition at the cleavage furrow. This GFAP phosphorylation specifically localized at the cleavage furrow was observed not only in astroglial cells but also in other cultured cells transfected with GFAP cDNA (8). These findings suggested the existence of a protein kinase specifically activated at the cleavage furrow and its important role in cytokinesis. We tentatively termed this kinase cleavage furrow (CF) kinase (8). However, the molecular identity, regulation, and function of CF kinase remained to be examined.
The small GTP-binding protein Rho is implicated in the control of cytoskeletal structures, cell adhesions, and cell morphology (for review, see Ref. 9). Upon stimulation with certain signals, the GDP-bound inactive form of Rho may be converted to the GTP-bound active form, which binds to specific targets and thereby exerts its biological functions. We have identified three putative targets for Rho, p128 protein kinase N (10, 11), p138 myosin-binding subunit (MBS) of myosin phosphatase (12), and p164 Rho-kinase (13), which is also named ROK (14). Rho-kinase phosphorylates MBS and consequently inactivates myosin phosphatase (12). Rho-kinase also phosphorylates myosin light chain and thereby activates myosin ATPase (15). Other putative targets for Rho include rhophilin (11), p160 Rho-associated coiled-coil containing protein kinase (16), and citron (17).
Recently, Rho was shown to play a critical role in cytokinesis by inducing and maintaining the contractile ring, an actinbased cytoskeletal structure (18,19). In addition, Rho was reported to be translocated from the cytosol to the cleavage furrow during cytokinesis (20). These results raise the possibility that Rho may also be implicated in the regulation of CF kinase and thereby in the efficient separation of IFs to daughter cells. As a first step toward defining this possibility, we have examined whether protein kinase N and/or Rho-kinase can phosphorylate GFAP at the same sites as CF kinase. Protein kinase N was found to phosphorylate GFAP mainly at Ser-8, a site that is not phosphorylated by CF kinase. 2 * This work was supported in part by grants-in-aid for scientific research and cancer research from the Ministry of Education, Science, Sports, and Culture of Japan and by special coordination funds from the Science and Technology Agency of the Government of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed: Laboratory of Biochemistry, Aichi Cancer Center Research Inst., Chikusa-ku, Nagoya, Aichi 464, Japan. Tel.: 81-52-762-6111 (Ext. 8824); Fax: 81-52-763-5233. 1 The In this report, we show that GFAP can serve as an excellent substrate for Rho-kinase and that the GFAP phosphorylation by Rho-kinase prevents its filament formation in vitro. Furthermore, we present evidence that Rho-kinase phosphorylates GFAP at Thr-7, Ser-13, and Ser-34 in vitro, the same sites that are phosphorylated by CF kinase in vivo.

EXPERIMENTAL PROCEDURES
Materials-Recombinant human GFAP was prepared from Escherichia coli as described previously (8). Mouse monoclonal antibodies YC10, KT13, KT34, and MD389 were prepared as described previously (8,21). GST-RhoA was purified and loaded with guanine nucleotides (22). Rho-kinase was purified from bovine brain (13). Constitutively active GST-Rho-kinase, a GST fusion protein of the catalytic fragment of Rho-kinase, was purified from Sf9 cells as described previously (15). The catalytic subunit of cAMP-dependent protein kinase (protein kinase A) was prepared from bovine heart by the method of Beavo et al. (23). Cdc2 kinase was prepared from FM3A cells by the method of Kusubata et al. (24). Protein concentrations were measured according to Bradford (25) using bovine serum albumin as a standard.
Other Procedures-All the procedures for immunoblotting have been described elsewhere in detail (8,21). Immunofluorescence microscopy was performed as described previously (8). Amino acid sequences were analyzed with an ABI 476A gas-phase sequencer. Two-dimensional phosphoamino acid analysis was performed as described (27). Electron microscopy was carried out as described (28).

RESULTS AND DISCUSSION
We recently developed four monoclonal antibodies, YC10 (21), KT13, KT34, and MO389 (8) against four distinct phosphopeptides corresponding to the partial amino acid sequences of porcine GFAP. YC10, KT13, and KT34 react with GFAP phosphorylated at Ser-8, Ser-13, and Ser-34, respectively. MO389 reacts with both the phosphorylated and unphosphorylated GFAPs and stains filamentous structures in both mitotic and interphase cells. MO389 immunostained an intricate mesh of glial filaments in the entire cytoplasm of both metaphase and anaphase cells, but YC10 stained filamentous structures throughout the cytoplasm of metaphase but not anaphase cells (Fig. 1A). In contrast, the immunoreactivities of KT13 and KT34 were observed specifically between the daughter nuclei and at the cleavage furrow of anaphase cells (Fig.  1A). Immunocytochemical studies with KT13 (Fig. 1B) and KT34 (data not shown) using confocal laser scanning microscopy revealed that GFAP phosphorylated at Ser-13 and Ser-34 is associated with the cleavage furrow to form a ring-like structure but not a disc-like structure, such as the telophase disc reported by Andreassen et al. (29).
To search for the putative CF kinase responsible for the cleavage furrow-specific phosphorylation described above, we examined whether Rho-kinase purified from bovine brain can phosphorylate GFAP in vitro. The results indicated clearly that Rho-kinase phosphorylated GFAP in a GST-RhoA-dependent manner ( Fig. 2A). GDP-bound GST-RhoA enhanced the phosphorylation of GFAP by Rho-kinase 13-fold, and GTP␥S-bound GST-RhoA enhanced it 291-fold ( Fig. 2A).
We then examined the phosphorylation sites on GFAP by Rho-kinase using the anti-phosphoGFAP antibodies described above. After the phosphorylation reaction, samples were resolved by SDS-PAGE and immunoblotted with MO389, YC10, KT13, or KT34. As shown in Fig. 2B, Rho-kinase phosphorylated GFAP at Ser-13 and Ser-34 but not at Ser-8 in a GTP␥S⅐GST-RhoA-dependent manner.
We also used the constitutively active GST-Rho-kinase, a fusion protein between GST and the catalytic fragment of Rhokinase produced in Sf9 cells by recombinant baculovirus infection. Analyses with the anti-phosphoGFAP antibodies revealed that GST-Rho-kinase also phosphorylated GFAP at Ser-13 and Ser-34 but not at Ser-8 (Fig. 2C). These results suggest that catalytic characteristics of GST-Rho-kinase are similar to those of native Rho-kinase activated by GTP␥S⅐GST-RhoA. In contrast, the catalytic subunit of cAMP-dependent protein kinase phosphorylated all three serine residues, and Cdc2 kinase weakly phosphorylated only Ser-8 (Fig. 2C).
To confirm phosphorylation sites on GFAP by Rho-kinase, GFAP (130 g) was phosphorylated by GST-Rho-kinase in the presence of [␥-32 P]ATP to approximately 2.7 mol of phosphate/ mol of GFAP (Fig. 3A). The radioactive GFAP was then digested with lysyl endopeptidase to generate about a 6.5-kDa

Phosphorylation of GFAP by CF Kinase and Rho-kinase 10334
fragment consisting mainly of the amino-terminal head domain. Tricine-SDS-PAGE analysis (30) revealed that all radioactivity associated with GFAP was retained in this 6.5-kDa head domain (Fig. 3B). This phosphorylated head domain was isolated by reverse-phase HPLC, digested with trypsin, and then again subjected to reverse-phase HPLC. As shown in Fig.  4A, three major radioactive peaks, R1 to R3, were obtained. Phosphoamino acid analysis showed the presence of phosphothreonine in R1 and phosphoserine in both R2 and R3 (Fig. 4B). Amino acid sequence analysis revealed that R1 was the peptide containing Thr-7, R2 was the peptide containing Ser-34, and R3 was the peptide containing Ser-13 (Fig. 4A). Ethanethiol treatment of R2 and R3, a procedure that converts specifically phosphoserine into S-ethylcysteine (31), suggested that phosphates were located on Ser-34 and Ser-13, respectively (data not shown). Therefore, GFAP was shown to be phosphorylated at Thr-7, Ser-13, and Ser-34 by GST-Rho-kinase. By using a rabbit polyclonal antibody recognizing phosphorylated Thr-7, we have previously reported that Thr-7 is also phosphorylated at the cleavage furrow (7).
We then examined the effect of phosphorylation of GFAP by Rho-kinase on the filament forming ability of GFAP. Soluble GFAP was preincubated with or without GST-Rho-kinase for 30 min, and the samples were incubated under conditions of polymerization (25 mM imidazole-HCl, pH 6.75, and 100 mM NaCl at 37°C) (28) for a further hour. Then the NaCl-and pH-dependent filament formation of GFAP in these samples was analyzed by centrifugation (Fig. 5A) and electron microscopy (Fig. 5B). As shown in Fig. 5, the phosphorylation of GFAP by GST-Rho-kinase resulted in a nearly complete inhibition of its filament formation. These results increase the possibility  4 -6). B, GFAP phosphorylated as in A in the absence of 32 Plabeled ATP was resolved by SDS-PAGE and immunoblotted with MO389, YC10, KT13, or KT34 using the ECL Western blotting detection system (Amersham Corp.). The chemiluminescence was detected by exposure for 1 min. C, GFAP was incubated with GST-Rho-kinase, buffer (control), protein kinase A (A-kinase), or Cdc2 kinase as described under "Experimental Procedures." Samples (each 2 l) were analyzed by immunoblotting with YC10, KT13, or KT34. The chemiluminescence was detected by exposure for 5 s.   4. Phosphorylation of Thr-7, Ser-13, and Ser-34 on GFAP by GST-Rho-kinase. A, the phosphorylated 6.5-kDa fragment derived from GFAP was digested with trypsin and subjected to reverse-phase HPLC as described under "Experimental Procedures." Three radioactive peaks (R1, R2, and R3) were analyzed with a gas-phase sequencer. The determined amino acid sequences of R1 (residues 5-11), R2 (residues 37-41), and R3 (residues 13-29) are indicated on the right. Note that we describe Ser-38 of human GFAP as Ser-34 because Ser-38 of human GFAP corresponds to Ser-34 of porcine GFAP. B, aliquots of R1, R2, and R3 in A were subjected to two-dimensional phosphoamino acid analysis.

Phosphorylation of GFAP by CF Kinase and Rho-kinase 10335
that GFAP phosphorylation at Thr-7, Ser-13, and Ser-34 during cytokinesis may induce the fragmentation of glial filaments at the cleavage furrow.
In the present study, we obtained evidence that GFAP can serve as an excellent substrate for Rho-kinase in a GTP⅐Rhodependent manner. So far, MBS (12) and myosin (15) were the only preferred substrates for Rho-kinase. The phosphorylated GFAP lost its ability to form filaments in vitro. The in vitro phosphorylation sites on GFAP by Rho-kinase were Thr-7, Ser-13, and Ser-34, which are the same sites that CF kinase phosphorylates at the cleavage furrow during cytokinesis. We are considering that Rho-kinase may be CF kinase itself, and if so it may play an important role in the cleavage furrow-specific reorganization of IFs during cytokinesis.
Because Rho-kinase was recently reported to act downstream of Rho in the regulation of myosin phosphorylation (12,15) and the formation of stress fibers and focal adhesion complexes (32,33), Rho-kinase may also mediate the regulation of cytokinesis by Rho (18,19). Whether Rho-kinase is activated during cytokinesis is the subject of ongoing studies. Because Rho-kinase belongs to a family of related serine/threonine kinases including myotonic dystrophy kinase, these kinases may phosphorylate the similar sites on GFAP. Further investigations are necessary to elucidate the relationship between CF kinase and Rho-kinase or its family members.