Protein Kinase C Phosphorylation of Desmin at Four Serine Residues within the Non-a-helical Head Domain*

We reported that phosphorylation by either CAMP- dependent protein kinase or protein kinase C (Ca”+/ phospholipid-dependent enzyme) in vitro induces disassembly of the desmin filaments (Inagaki, M., Gonda, Y., Matsuyama, M., Nishizawa, K., Nishi, Y., and Sato, C. (1988) J. Biol. Chern. 263, 5970-5978). For this subunit protein, Ser-29, Ser-35, and Ser-50 within the non-a-helical head domain were shown to be the sites of phosphorylation for CAMP-dependent protein kinase (Geisler, N., and Weber, K. (1988) EMBO J. 7, 15-20). In the present work, we identified the sites of desmin phosphorylated in vitro by other protein kinase which affects the filament structure. The protein kinase C-phosphorylated desmin was hydrolyzed with trypsin, and the phosphorylated peptides were isolated by reverse-phase chromatography. Sequential analysis of the purified phosphopeptides, together with the known primary sequence, revealed that Ser-12, Ser-29, Ser-38, and Ser-56 were phosphorylated by pro- tein kinase C. All four sites are located within

Phosphorylation, a major modification common to all intermediate filament proteins has been examined in various in vivo systems (1)(2)(3). Intermediate filaments undergo dynamic changes during mitosis, cell locomotion, and in pathologic processes (4)(5)(6). A temporal relationship between changes in intermediate filament organizations and alterations in phosphorylation of their subunit proteins has been demonstrated (7)(8)(9)(10)(11)(12)(13)(14). In particular, intermediate filaments are significantly * This research was supported in part by a grant-in-aid for scientific research and a grant for cancer research from the Ministry of Education, Science and Culture of Japan, by Toray Science and Technology grants, 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 "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
T To whom correspondence should be addressed Laboratory of Experimental Radiology, Aichi Cancer Center Research Institute, Chikusa-ku, Nagoya 464, Japan. reorganized in cells during mitosis, following an increase in filament phosphorylation (9)(10)(11)(12)(13)(14). These observations led to speculation that phosphorylation plays an important role in regulating organization of the intermediate filament component of the cytoskeleton (2,3).
There are reports on in vivo phosphorylation sites and/or domains of the intermediate filament proteins. These proteins contain the amino-terminal head domain, the central a-helical rod domain and the carboxyl-terminal tail domain (3,15). In vertebrate and invertebrate neurofilaments, most if not all of the phosphorylation of serine residues occurs in the extended tail domain (16)(17)(18)(19). These phosphoserine sites mainly contain the sequence Lys-Ser(P)-Pro (18,19). Keratin intermediate filament proteins are significantly phosphorylated on both serine and threonine residues in vivo, and the bulk of this phosphorylation occurs on the amino-and carboxylterminal domains (20)(21)(22)(23). Steinert (24) most recently reported that the major sites of phosphorylation of the keratin 1 chain mainly involve sequences containing Arg-X-Ser(P) within the amino-and carboxyl-terminal domains. Analysis of desmin and vimentin from nonmitotic and mitotic cells has shown that the increased phosphorylation of desmin and vimentin observed during cell division occurs exclusively within the amino-terminal head domains (25). However, the precise effects of phosphorylation on intermediate filaments in intact cells remained to be determined.
We reported the role of phosphorylation of desmin and vimentin, in vitro, to be as follows (26,27): vimentin and desmin are excellent in vitro substrates for protein kinase C (Ca2+/phospholipid-dependent enzyme) and CAMP-dependent protein kinase, but not of several other kinases. Desmin and vimentin phosphorylated by each protein kinase do not polymerize. The filaments that do polymerize tend to depolymerize after phosphorylation. Moreover, dephosphorylation by phosphoprotein phosphatase leads to a reassembly of soluble desmin into filaments.
Geisler and Weber (28) reported that the sites of desmin phosphorylated by CAMP-dependent protein kinase in vitro are restricted to the non-a-helical head domain. We have now obtained findings that not only CAMP-dependent protein kinase but also protein kinase C phosphorylates serine residues within the amino-terminal head domain of desmin in different site recognitions. Together with the reported data (26-28), the present results provide clues to the molecular mechanisms of phosphorylation-dependent disassembly of desmin filament, and the functional role of in vivo phosphorylation of desmin filaments is better understood.

EXPERIMENTAL PROCEDURES
Purification of Proteins-Purified desmin was obtained by extraction of the crude intermediate filament preparation from chicken gizzard with 8 M urea and the subsequent chromatography on DEAEcellulose and CM-cellulose columns in the presence of urea, as described (27). Protein kinase C was prepared from rat brain by the method of Inagaki et al. (29).
Isolation of Phosphorylated Desmin-Two ml each of the phosphorylation mixture containing desmin (0.25 mg/ml) was applied to a Zorbax C8 (0.46 X 20 cm) column attached to a Waters HPLC' system consisting of model 510 pumps, a model 490 detector, and an automated gradient controller. The phosphorylated desmin was eluted at around 35 min, using a linear gradient of 5-90% acetonitrile in 0.1% trifluoroacetic acid over 50 min at a flow rate of 0.8 ml/min. Elution was monitored by UV at 230 nm using a Chromatocorder 11 (System Instruments Corp., Dover, MA) or by radioactivity of each fraction (0.8 ml), using a Beckman scintillation counter LS 5801. This separation was repeated eight times. Fractions containing the radioactive phosphorylated desmin were pooled and lyophilized.
Fragmentation of Phosphorylated Desmin-The phosphorylated desmin isolated by reverse-phase HPLC was dissolved in 50 mM Tris-HCI (pH 7.5) at a concentration of 0.25 mg/ml and was treated with L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Sigma, 1:50 (w/w) of desmin) at 37 "C for 4 h. Fresh trypsin was again added, and the mixture was incubated for an additional 4 h. An aliquot (1.5-2 ml) of the reaction mixture was applied to a Zorbax C8 (0.46 X 20 cm) column and was eluted with 5% acetonitrile, 0.1% trifluoroacetic acid, and subsequently with linear 5-5076 acetonitrile gradient in 0.1% trifluoroacetic acid. The flow rate was 0.8 ml/min, 0.8-ml fractions. Each radioactive fraction was separately lyophilized and stored at 4 'C.
Purification of Fragments-Each radioactive fraction was applied to an anion exchange column of TSK gel QAE-2SW (0.46 X 25 cm) and was eluted with a linear gradient of 0-0.5 M NaCl in 20 mM Tris-HCl (pH 7.5) over 40 min at a flow rate of 0.8 ml/min. Each radioactive fraction obtained was then applied to a Zorbax C8 (0.46 X 20 cm) column and was eluted with the same linear gradient used in the first fractionation of the phosphopeptides at the same flow rate. Radioactive fractions obtained were lyophilized and stored at 4 'C.
Modification of Phosphoserine Residues-The purified radioactive fragments (1-7 nmol) were treated with 100 pl of solution consisting of 9.9 pl of ethanethiol, 33.1 pl of water, 33.1 pl of dimethyl sulfoxide, 13.2 p1 of ethanol, and 10.7 pl of 5 N NaOH, at 50 "C for 1 h, as described by Meyer et al. (30). Prior to sequence analysis, the reaction mixtures were cooled and stored at -20 "C, after adding 10 p1 of acetic acid.
Sequence Analysis-An aliquot of either the purified fragments dissolved in 0.1% trifluoroacetic acid or the chemically modified fragments in the reaction mixture was analyzed with an AB1 gasphase sequenator, model 470 A, using the manufacturer's program. The PTH-derivatives were analyzed on an AB1 PTH-C18 (0.46 X 22 cm) cartridge column attached to an HPLC system consisting of Spectra-Physics 8700 pump, ISCO V4 absorbance detector and Spectra-Physics 4270 integrator.
Phosphoamino Acid Analysis-The radioactive fragments were subjected to acid hydrolysis in 6 N HCl for 1.5 h at 110°C. The phosphoamino acids were resolved by electrophoresis at pH 3.5 on a cellulose thin layer plate as described (31).

RESULTS AND DISCUSSION
Isolation of Phosphopeptides Derived from Protein Kinase C-phosphorylated Desmin-To identify the sites phosphorylated by protein kinase C, desmin (4 mg) was phosphorylated with [y3'P]ATP by protein kinase C to approximately 3.0 mol of phosphate/mol of desmin (see "Experimental Procedures"). The material was directly subjected to HPLC equipped with a reverse-phase column, as described under "Experimental Procedures," and the radioactive desmin (2.4 mg) was eluted as a sharp peak between 61 and 67% acetonitrile. The radioactive desmin was then completely digested The abbreviations used are: HPLC, high performance liquid chromatography; PTH, phenylthiohydantoin. with trypsin (see "Experimental Procedures"). The phosphopeptides were isolated by HPLC equipped with a reversephase column, as described under "Experimental Procedures." As shown in Fig. 1, the HPLC procedure separated several phosphopeptides and each was further purified by anion exchange and reverse-phase HPLC columns, as described under "Experimental Procedures." Each phosphopeptide appeared as a single and symmetric peak, and the amount of the phosphopeptide purified in this manner was in the range of 3-10 nmol.
Phosphorylation Sites Locate to the Head Domain-Each purified radioactive phosphopeptide was analyzed for gasphase Edman degradation, as described under "Experimental Procedures." The amino acid sequences of these phosphopeptides are listed in Table I. By comparison with the reported sequence of chicken gizzard desmin (32), peptides 1, 2, 3, and 4 were located at desmin residues 34-42, 10-14, 28-33, and 49-59, respectively. Thus, all the phosphorylation sites of desmin for protein kinase C apparently locate at the aminoterminal head domain (residues 1-69) (15,33).
The domain location of the protein kinase C phosphorylation sites agrees with certain aspects of intermediate filament structure (3,15). Although the filament wall seems to arise from interaction of the double-stranded coiled-coils provided by the rod domain, the non-a-helical terminal domains may well provide a stabilizing factor. Use of defined proteolytic derivatives of desmin has led to increasing attention directed to the head domain (33). Moreover, Geisler and Weber (28) reported that CAMP-dependent desmin phosphorylation occurs exclusively in the amino-terminal head domain. Thus, our present observations add support to the view that the   The sites of phosphorylation for CAMP-dependent protein kinase (28) were located at the same tryptic peptide obtained in this study. non-a-helical head domain has a strong influence on filament integrity.
Identification of Phosphorylation Sites in the Head Domain-Phosphoamino acid analysis of each phosphopeptide revealed the presence of only phosphoserine (Fig. 2). Since all as shown in Table I, the exact phosphorylation sites had to be defined. For this purpose, each fragment was treated with ethanethiol in alkaline condition to specifically convert the phosphoserine residues to S-ethylcysteine. Normal serine residues were not affected by this treatment (30). The positions the phosphopeptides contained more than 2 sekne residues, of S-ethylcysteine residues within the peptides were identified by gas-phase sequencing (30) . Fig. 3 shows the sequence analysis of each phosphopeptide treated with ethanethiol prior to Edman degradation. A high release of S-ethylcysteine was observed at the fifth cycle for peptide 1, the third cycle for peptide 2, the second cycle for peptide 3, and the eighth cycle for peptide 4, thereby indicating that the phosphate was located on Ser-38, Ser-12, Ser-29, and Ser-56, respectively.
Primary Structure of the Vicinity of the Phosphorylation Sites-The primary structure of the vicinity of the amino acyl residue to be phosphorylated is one important factor for determining substrate recognition, and CAMP-dependent protein kinase reacts normally with serine and threonine residues located at the carboxyl-terminal side close to lysine or arginine (34,35). In the case of desmin, Geisler and Weber (28) identified Ser-29, Ser-35, and Ser-50 as the sites to be phosphorylated by CAMP-dependent protein kinase. We also identified these three sites of desmin phosphorylation for CAMPdependent protein kinase, using the same methods for protein kinase C described in the present work (data not shown). All these serine residues are located at the carboxyl-terminal side close to arginine (Table I) (28). The present analysis revealed Ser-12, Ser-29, Ser-38, and Ser-56 as sites to be phosphorylated by protein kinase C.
Contrary to the sites phosphorylated by CAMP-dependent protein kinase, all 4 serine residues are located at the aminoterminal side close to arginine. Ser-29 phosphorylated by both these kinases has arginine residues at the amino-and carboxyl-terminal sides. These results are summarized in Table  I . These groups noted that the site specificity of protein kinase C was distinct from CAMPdependent protein kinase with regard to the seemingly preferred location of the positively charged residues.
However, sequences around known phosphorylation sites, that is protein kinase C in histone H1 (39), histone H2B (39), glycogen synthase (40), and acetylcholine receptor (41) indicate that the phosphorylation sites are located at the carboxylbut not amino-terminal side close to lysine or arginine. Moreover, studies by House et al. (42) who used as substrates various types of synthetic peptides suggest that protein kinase C can recognize primary specificity determinants on either the amino-or carboxyl-terminal side of the phosphorylatable residue. Some of the protein kinase C phosphorylation sites of desmin do not have an appropriate protein kinase C substrate site, usually arginine. This basic residue is separated from the phosphorylation site by one to two neutral amino acids (37,42,43). In Ser-29 and Ser-38, the carboxyl-terminal arginine is separated from the phosphorylation sites by 3 neutral amino acids.
The Secondary Structure of Phosphorylation Sites and Their Vicinity-Prediction of the secondary structure (44) for the head domain sequence of desmin shows that most of the phosphorylation sites of both the kinases are likely to be located in the @-turn structures (Fig. 4). This would support the observation (27) that these two protein kinases readily phosphorylate not only soluble desmin but also the desmin filament, since the @-turn structures generally locate at the solvent-accessible protein surface (45).
Desmin is phosphorylated in vitro by protein kinase C which in turn leads to desmin filament disassembly in vitro (27). There are 4 serine residues phosphorylated and only one is also phosphorylated by CAMP-dependent protein kinase. However, in these in vitro experiments, phosphorylation of desmin by either protein kinase C or CAMP-dependent protein kinase was performed under conditions of low salt and low MgClz (26)(27)(28). The concentrations of both salt and MgC12 affect the rate of desmin phosphorylation by these protein kinases. In particular, increasing the MgC12 concentrations (>3 mM) dramatically decreased the rate of desmin phosphorylation by the protein kinases through formation of the abnormal desmin filament, as was the case with vimentin (26,27). For desmin, precise locations of the phosphoamino acids within the head domain in vivo have not been determined. Studies are ongoing to examine possible physiological significance of our findings.