Site-specific phosphorylation of synapsin I by mitogen-activated protein kinase and Cdk5 and its effects on physiological functions.

Posttranslational modifications of synapsin I, a major phosphoprotein in synaptic terminals, were studied by mass spectrometry. In addition to a well known phosphorylation site by calmodulin-dependent protein kinase II (CaM kinase II), a hitherto unrecognized site (Ser553) was found phosphorylated in vivo. The phosphorylation site is immediately followed by a proline, suggesting that the protein is an in vivo substrate of so-called proline-directed protein kinase(s). To identify the kinase involved, three proline-directed protein kinases expressed highly in the brain, i.e. mitogen-activated protein (MAP) kinase, Cdk5-p23, and glycogen synthase kinase 3beta, were tested for the in vitro phosphorylation of synapsin I. Only MAP kinase and Cdk5-p23 phosphorylated synapsin I stoichiometrically. The phosphorylation sites were determined to be Ser551 and Ser553 with Cdk5-p23, and Ser62, Ser67, and Ser551 with MAP kinase. Upon phosphorylation with MAP kinase, synapsin I showed reduced F-actin bundling activity, while no significant effect on the interaction was observed with the protein phosphorylated with Cdk5-p23. These results raise the possibility that the so-called proline-directed protein kinases together with CaM kinase II and cAMP-dependent protein kinase play an important role in the regulation of synapsin I function.

Posttranslational modifications of synapsin I, a major phosphoprotein in synaptic terminals, were studied by mass spectrometry. In addition to a well known phosphorylation site by calmodulin-dependent protein kinase II (CaM kinase II), a hitherto unrecognized site (Ser 553 ) was found phosphorylated in vivo. The phosphorylation site is immediately followed by a proline, suggesting that the protein is an in vivo substrate of socalled proline-directed protein kinase(s). To identify the kinase involved, three proline-directed protein kinases expressed highly in the brain, i.e. mitogen-activated protein (MAP) kinase, Cdk5-p23, and glycogen synthase kinase 3␤, were tested for the in vitro phosphorylation of synapsin I. Only MAP kinase and Cdk5-p23 phosphorylated synapsin I stoichiometrically. The phosphorylation sites were determined to be Ser 551 and Ser 553 with Cdk5-p23, and Ser 62 , Ser 67 , and Ser 551 with MAP kinase. Upon phosphorylation with MAP kinase, synapsin I showed reduced F-actin bundling activity, while no significant effect on the interaction was observed with the protein phosphorylated with Cdk5-p23. These results raise the possibility that the so-called proline-directed protein kinases together with CaM kinase II and cAMP-dependent protein kinase play an important role in the regulation of synapsin I function.
Synapsin I has been characterized as one of the major phosphoproteins in nerve terminals and is thought to be involved in the regulation of neurotransmitter release (for reviews see Refs. 1 and 2). Synapsin I cross-links synaptic vesicles and cytoskeleton, and the interactions of the protein with actin filaments and synaptic vesicles are regulated by phosphorylation by calmodulin-dependent protein kinase II (CaM kinase II) 1 and cAMP-dependent protein kinase (3)(4)(5)(6). To understand the regulatory mechanisms of synapsin I function, it is necessary to know the posttranslational modifications of the protein in detail.
Recently we have applied electrospray mass spectrometry to studies on in vivo posttranslational modifications of various phosphoproteins. The high precision (within a few Da) and the high resolution (on the order of 10 Da) achieved by the method has made it possible to analyze protein phosphorylation and myristoylation of isolated proteins directly (7)(8)(9). Liquid chromatography/electrospray mass spectrometry (LC/MS), in which a capillary high performance liquid chromatography is connected online to an electrospray mass spectrometer, was found very useful in analyzing the in vivo posttranslational modifications including protein phosphorylation. Application of the methodology to brain-specific phosphoproteins revealed that prominent in vivo substrate proteins such as myristoylated alanine-rich protein kinase C substrate (MARCKS) or GAP-43 are phosphorylated by proline-directed protein kinases such as mitogen-activated protein (MAP) kinase and Cdk5 (10,11). This was surprising, since these two proteins have been believed to be major and specific substrates of protein kinase C. This prompted us to reexamine in vivo phosphorylation sites of various major phosphoproteins systematically.
In the present study, the posttranslational modifications of synapsin I isolated from bovine brain were studied, and the LC/MS analysis revealed a novel phosphorylation site. To identify the protein kinase(s) involved in the phosphorylation at the novel site, we have examined in vitro phosphorylation of synapsin I by three proline-directed protein kinases expressed highly in the brain, namely MAP kinase, Cdk5-p23 (tau protein kinase II), and glycogen synthase kinase 3␤ (GSK3␤) (tau protein kinase I) (12)(13)(14)(15)(16)(17). Effects of the phosphorylation on physiological functions of synapsin I were further assessed by examining the interaction of the protein with cytoskeletal proteins.

EXPERIMENTAL PROCEDURES
Materials-Synapsin I, purified from bovine brain using acid extraction or detergent extraction under nondenaturing conditions as described (18,19), was stored in 25 mM Tris-HCl buffer (pH 8.0), containing 175 mM NaCl and 0.1 mM EGTA, at Ϫ80°C. Both preparations gave essentially the same results in terms of in vitro phosphorylation. Actin prepared from acetone powder of rabbit skeletal muscle as described (20) was further purified by gel filtration on a Superose 12 column (Pharmacia Biotech Inc.). Tubulin was prepared from bovine brain homogenates by three cycles of temperature-dependent assembly and disassembly in 0.1 M Mes-NaOH buffer (pH 6.8), containing 1 mM EGTA, 0.5 mM MgCl 2 , and 1 mM GTP, followed by phosphocellulose column chromatography (21). MAP kinase was purified from bovine brain as described previously (12), while Cdk5-p23 and GSK3␤ were purified from the same source as tau protein kinase II and I, respectively (22).
Liquid Chromatography/Electrospray Mass Spectrometry Analysis-Purified synapsin I was digested either with L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Worthington) or with lysyl endoprotease from Achromobacter (kindly supplied by Dr. T. Masaki, Ibaraki University, Japan) in 50 mM Tris-HCl (pH 8.9) containing 2 M urea at 35°C for 10 h. The reaction was stopped by adding 0.1% final concentration of trifluoroacetic acid, and the resulting peptide mixtures were directly injected into the LC/MS apparatus as described previously (10,11).
Light-scattering Assay of Actin Filament Bundling Activity-Aliquots of F-actin (0.057 mg/ml) were incubated with various concentrations of purified synapsin I or with synapsin I phosphorylated by MAP kinase in 10 mM Hepes buffer (pH 7.4) containing 100 mM KCl, 2 mM MgCl 2 , 1 mM ATP, and 0.5 mM ␤-mercaptoethanol. Light-scattering intensity of each solution at an angle of 90°was measured in a JASCO FP-777 spectrofluorometer with the excitation and emission wavelengths set at 400 nm as described (23).
Sedimentation Assays-G-actin (3.2 M) and synapsin I (0.55 M, unphosphorylated or phosphorylated form) were incubated under polymerizing conditions of actin for 1 h at room temperature in 70 mM KCl, 25 mM NaCl, 2 mM MgCl 2 , 0.2 mM ATP, 14 M EGTA, 7 M CaCl 2 , 3.5 mM Tris-HCl, and 15 mM Hepes-NaOH (pH 7.4). The samples were then centrifuged at 12,000 rpm for 10 min in a table top centrifuge, and the resulting pellets were analyzed by SDS-polyacrylamide gel electrophoresis.
Binding to Polymerized Microtubules-Tubulin stored as a microtubule pellet at Ϫ80°C was cycled once prior to use. Synapsin I, unphosphorylated or phosphorylated, was mixed with tubulin at 4°C in 0.1 M Mes buffer (pH 6.8), containing 1 mM MgCl 2 , 1 mM GTP, and 1 mM EGTA. Glycerol was added to give a final concentration of 20% (v/v), and the mixture was brought to 37°C for 30 min. Samples were then centrifuged at 100,000 ϫ g for 30 min at 37°C in a Beckman TL-100 table top ultracentrifuge. Supernatants and pellets thus obtained were subjected to SDS-polyacrylamide gel electrophoresis.

LC/MS Analysis of Synapsin I Protease Digests-Synapsin
I isolated from bovine brain using acid extraction as described under "Experimental Procedures" was digested either with trypsin or with lysyl endoprotease, and the resulting peptide mixtures were analyzed with the LC/MS. The total ion current chromatogram obtained with the tryptic digests is shown in Fig. 1. The mass of each peptide eluted from the reversed-phase column was determined with the mass spectrometer and com-pared with the theoretical one calculated from the deduced amino acid sequence (24). Most of the peptides that account for more than 98% of the whole sequences of both synapsin Ia and Ib were easily identified solely from the mass as shown in Table  I. One interesting point to note is that the N-terminal peptide is clearly N ␣ -acetylated.
Two tryptic peptides were found partially phosphorylated as shown in Fig. 2. One peptide corresponds to T52 (from Gln 603 to Arg 612 ), and the other corresponds to T47 (from Gln 532 to Arg 556 ). The former contains one of the two phosphorylation sites by CaM kinase II, i.e. Ser 605 (25). Since the peptide contains only one phosphorylatable amino acid, we conclude that the Ser 605 is the major in vivo phosphorylation site. The latter peptide, T47, contains more than one possible phosphorylatable residue and was subjected to further analysis. A similar Synapsin I isolated from bovine brain was digested with trypsin, and the resulting peptide mixture was subjected to LC/MS analysis as described under "Experimental Procedures." Peaks are numbered according to the elution order, and the peptides identified are shown in Table I. analysis on a synapsin I purified by detergent extraction indicated that these peptides were not phosphorylated, suggesting that dephosphorylation occurred during purification.
Identification of the Phosphorylation Site by Edman Degradation-To determine the phosphorylation sites, the phosphopeptide T47 was isolated by reversed-phase column chromatography and subjected to Edman degradation. Since serine forms both phenylthiohydantoin-derivative of serine (PTH-Ser) and dithiothreitol adduct of dehydroalanine (DTT-Ser), phosphoserine can be detected unambiguously by comparing the yields of the two products (10,11,26). As shown in Table II, the first two of three serine residues in the peptide showed peaks of PTH-Ser and DTT-Ser. The ratios of the two peaks were 0.27 and 0.21, at the second cycle (Ser 533 ) and at the 20th cycle (Ser 551 ), respectively. These values correspond well to those observed with normal serine under the conditions employed. On the contrary, the yield of PTH-Ser peak at the 22nd cycle was under the detection limit, although that of DTT-Ser (2.6 pmol) was comparable with that observed at the 20th cycle (2.9 pmol). Since the phosphopeptide fraction subjected to the analysis contained only singly phosphorylated species as was confirmed by mass spectrometry, these results indicate that Ser 553 is the sole phosphorylation site in the peptide. The phosphorylation site is immediately followed by a proline residue, suggesting that synapsin I is an in vivo substrate of so-called proline-directed protein kinase.
In Vitro Phosphorylation of Synapsin I by Proline-directed Protein Kinases-To identify the protein kinase involved in the in vivo phosphorylation of synapsin I at the novel sites identified, three protein kinases with serine (threonine)-proline specificity that are highly expressed in the brain were tested for their ability to phosphorylate purified synapsin I in vitro. As shown in Fig. 3, significant incorporation of radioactivity was observed with MAP kinase and Cdk5-p23. On the other hand, very little, if any, phosphorylation was observed with GSK3␤. Since phosphorylation of tau protein by Cdk5-p23 is a prerequisite for the subsequent phosphorylation by GSK3␤ (22), a mixture of the two kinases was also tested. The extents of the phosphorylation did not differ appreciably between the sample incubated with Cdk5-p23 alone and that with the mixture. The final level of the phosphorylation reached around 3 mol/mol of synapsin I with MAP kinase, while that obtained with Cdk5-p23 was around 1 mol/mol.
Analysis of the Phosphorylation Site by LC/MS-To determine the in vitro phosphorylation sites, synapsin I phosphorylated by MAP kinase or by Cdk5-p23 was digested with trypsin, and the resulting peptide mixtures were directly analyzed by LC/MS as described above. Since most of the peptides have already been assigned, phosphorylated peptides were easily detected by the increase of their mass by 80 Da. Peptide T47 that contained the newly found in vivo phosphorylation site, Ser 553 , was almost stoichiometrically phosphorylated when MAP kinase was used (Fig. 4, a and b). Interestingly, only singly phosphorylated peptide with a mass of 2516 Da was observed, and no peak corresponding to doubly phosphorylated species was detected. Another peptide found phosphorylated was peptide T5 (from Ala 54 to Lys 85 ) (Fig. 4, c and d). In this case quantitative incorporation of two phosphoryl groups was observed. While the doubly phosphorylated species was the major product, very little, if any, singly phosphorylated or triply phosphorylated species were observed. The peak areas of the deconvoluted mass spectra correlate well with the amounts of the peptides under the conditions employed (10,11). Peptide T4 (positions 8 -53) near the N terminus was also found phosphorylated to varying degrees depending on the incubation conditions (less than 30%; data not shown). Since no other peptides were found phosphorylated to a significant degree, we concluded that the MAP kinase phosphorylates synapsin I at three major sites and probably at one minor site. This corresponds well to the stoichiometry determined from the phosphorylation experiments using [ 32 P]ATP as described above.
When similar experiments were conducted with Cdk5-p23, only one peptide, namely T47, was found phosphorylated to a significant extent (Fig. 5). Judging from the peak intensities of the deconvoluted mass spectra, about 80% of the peptide was found in the singly phosphorylated state. No peak corresponding to doubly phosphorylated species was detected. Peptides T5 and T4 and other peptides identified by LC/MS were not phosphorylated to a significant degree, suggesting that Cdk5-p23 FIG. 2. Deconvoluted mass spectra of phosphopeptides. The electrospray mass spectra between peaks 5 and 7 shown in Fig. 1 were combined and deconvoluted to get a mass spectrum (a). Peptide T52 (1041.4 Da) was accompanied by a monophosphopeptide (1121.3 Da). A similar spectrum was obtained from spectra between peaks 11 and 12 shown in Fig. 1 3. In vitro phosphorylation of synapsin I by proline-directed protein kinases. Synapsin I purified from bovine brain using detergent extraction (19) was incubated with MAP kinase (E), Cdk5-p23 (E), GSK3␤ (ϫ), or a mixture of Cdk5-p23 and GSK3␤ (f) in the presence of [ 32 P]ATP as described under "Experimental Procedures." At the indicated times, aliquots were withdrawn and subjected to SDS-gel electrophoresis. The incorporation of the radioactivity was measured by counting gel slices.
phosphorylates synapsin I at a single site. Table III, the ratios between PTH-Ser and DTT-Ser remained fairly constant during the degradation of the phosphopeptide T5, which was obtained from MAP kinase-phosphorylated synapsin I, except at two positions, i.e. at cycles 9 and 14. Yields of PTH-Ser were reduced to about 10%, while those of DTT-Ser showed expected values. These Ser residues correspond to Ser 62 and Ser 67 , exactly the two Ser residues predicted from the recognition sequence of the kinase. These results demonstrate again how the method works. On the other hand, peptide T47 contains two Ser residues immediately followed by Pro, namely Ser 551 and Ser 553 , the latter being the sole in vivo phosphorylation site found in the peptide. When phosphopeptide T47 obtained from synapsin I phosphorylated with MAP kinase was subjected to the same analysis, only one Ser residue at cycle 20 showed a reduced yield of PTH-Ser (Table IV). The other two Ser residues showed yields corresponding to normal serine. From these results we conclude that the first Ser in the Ser-Pro-Ser-Pro sequence is the sole phosphorylation site in the peptide. This is, in a sense, not surprising, since Ser 553 is in a preferred recognition sequence of the kinase (Pro-Xaa-Ser-Pro) (27,28).

Determination of the in Vitro Phosphorylation Sites by Edman Degradation-As shown in
A similar analysis of the phosphopeptide T47 obtained from synapsin I phosphorylated with Cdk5-p23 gave more complex results (Table IV). The ratios of the yields of PTH-Ser and those of DTT-Ser showed intermediate values between 0.2 (serine) and 0.02 (phosphoserine) both at cycle 20 (Ser 551 ) and cycle 22 (Ser 553 ). Since the isolated phosphopeptide was confirmed by mass spectrometry to contain only singly phosphorylated species free from nonphosphorylated and doubly phosphorylated species (data not shown), the total content of phosphoserine in the peptide should be exactly one mol/mol of peptide. The results obtained indicate that the one phosphoserine is distributed between the two positions. When we assume the PTH-Ser/ DTT-Ser ratios of 0.03 and 0.20 for serine and phosphoserine, respectively, we can roughly estimate the contents of phosphoserine. At cycle 20, roughly 55% is phosphoserine, and at cycle 22 about 35% is phosphoserine. Since the sum (90%) corresponds fairly well to the theoretical value of 100%, such a calculation is feasible. When synapsin I purified by detergent extraction, which lacks the in vivo phosphorylated species, was used, about 50% of Ser 551 and 40% of Ser 553 were found phosphorylated. These results suggest that Cdk5-p23 phosphorylates synapsin I both at Ser 551 and Ser 553 and that the former is slightly favored over the latter. Furthermore, the phosphorylation at the two positions is mutually exclusive, since doubly phosphorylated species was not observed to significant extent as described above (Fig. 5). This was also the case with the MAP kinase-dependent phosphorylation. Although the synapsin I preparation used contained in vivo phosphorylated species, no doubly phosphorylated T47 peptide was observed with the MAP kinase phosphorylated synapsin I as shown in Fig. 4b. Thus, MAP kinase did not phosphorylate Ser 551 when Ser 553 had been already phosphorylated.
Effect of Phosphorylation of Synapsin I on its F-actin Bundling Activity-To understand physiological function of phosphorylation by MAP kinase or Cdk5-p23, we studied the effect of the phosphorylation on the F-actin bundling activity of synapsin I (23). The bundling of actin filaments was measured by the increase in light scattering. As shown in Fig. 6a FIG. 5. Deconvoluted mass spectra of phosphopeptide T47 found in Cdk5-p23-phosphorylated synapsin I. Synapsin I phosphorylated by Cdk5-p23 at 30°C for 5 h was digested with trypsin and subjected to the LC/MS analysis. Only peptide T47 was significantly phosphorylated. About 80% of the peptide was found in singly phosphorylated form. Note that no doubly phosphorylated form was detected. added was also found in the pellet (Fig. 6b). On the other hand, the addition of synapsin I phosphorylated by MAP kinase caused no significant increase of actin found in the pellet. Synapsin I also remained in the supernatant. Note that the mobility of synapsin I in the gel changed after phosphorylation by MAP kinase. These results demonstrated that the phosphorylation of synapsin I by MAP kinase drastically decreases the affinity of the protein to actin filaments. Contrary to the drastic effects caused by the MAP kinase-dependent phosphorylation, synapsin I phosphorylated by Cdk5-p23 retained the ability to bind to actin filaments and to bundle them as shown in Fig. 6b.

Effect of Phosphorylation on Synapsin I-Tubulin
Interaction-Since synapsin I has been reported to bind another major cytoskeletal component, tubulin (29), we then analyzed effects of the phosphorylation on the synapsin I-tubulin interaction. When synapsin I was mixed with polymerized tubulin, incubated at 37°C for 30 min and centrifuged at 100,000 ϫ g for 30 min, most synapsin I cosedimented with tubulin (Fig. 7, lane 6). Similarly, when synapsin I phosphorylated by MAP kinase was mixed with tubulin, most of phosphorylated synapsin I was found in pellets with tubulin (Fig. 7, lane 8). Phosphorylation by Cdk5-p23 did not show any significant effect on the binding either (data not shown). Thus, in contrast to actin bundling activity, phosphorylation of synapsin I by the two proline-directed protein kinases has little influence on the synapsin I-tubulin interaction.

DISCUSSION
Synapsin I, one of the prominent endogenous phosphoproteins in the nerve terminals, has been characterized as a substrate protein of various protein kinases such as CaM kinases I and II and cAMP-dependent protein kinase (2). Physiological functions of synapsin I, i.e. the cross-linking between synaptic vesicles and cytoskeletons seems to be regulated by phosphorylation by these kinases (1, 2). The detailed analysis on the in vivo phosphorylation site described in the present study, however, revealed a novel phosphorylation site. The phosphorylated serine is immediately followed by a proline, suggesting that synapsin I is an in vivo substrate of so-called prolinedirected protein kinase. We have previously shown that the two prominent in vivo substrate proteins of protein kinase C, MARCKS and GAP-43, are also phosphorylated by these kinases in vivo (10,11). These results suggest that the physiological functions of various proteins are regulated by multiple protein kinases in a very complex manner and that cross-talks between various signaling pathways occur not only upstream of the pathways but also at the substrate protein level.
Of the three proline-directed protein kinases tested, only MAP kinase and Cdk5-p23 phosphorylated synapsin I in vitro, and GSK3␤ did not phosphorylate the protein to a significant extent. The phosphorylation by the two kinases was site-specific; only one of two serine residues (Ser 551 and Ser 553 ) was phosphorylated by Cdk5-p23, while three serine residues were phosphorylated by MAP kinase. Since bovine synapsin I contains 11 Ser (The)-Pro motifs, there should be structural determinants other than the adjacent proline in the substrate recognition by the kinases. As for MAP kinase, Ser 551 is within a well known recognition sequence of the kinase, Pro-Xaa-Ser-Pro, where Xaa is usually a small neutral amino acid (27,28). Since the other two sites have a proline either at Ϫ3-position or at Ϫ1-position, the presence of a proline preceding the phosphorylation site serine/threonine may be important for the recognition. It should be noted that peptide T4 (from Leu 8 to Arg 53 ), which was phosphorylated to some extents, contains a single Ser-Pro motif (Ser 39 ), which is preceded by a proline. On the other hand, only one phosphopeptide was observed with synapsin I phosphorylated by Cdk5-p23. This kinase seems to phosphorylate both Ser 551 and Ser 553 , but the phosphorylation at these two sites is mutually exclusive, suggesting that the incorporation of negative charges in the neighborhood changes the substrate specificity. Ser 551 is preferentially phosphorylated, although the difference may not be significant. The recognition sequence of the kinase has yet to be defined, but an arginine at ϩ3-position may be an important determinant.
Only one kinase, so-called proline-directed protein kinase has been so far reported to phosphorylate synapsin I in these regions (30). The kinase has later been identified as Cdc2cyclin A complex (31). Since Cdc2 kinase is not expressed in the brain to a significant extent, this phosphorylation reaction lacks physiological relevance. However, it is of interest to note that the kinase also phosphorylates Ser 551 preferentially (30). Whether the kinase phosphorylates Ser 553 is not clear, because of the limitation of the technique used in determining the phosphorylation site. The radiosequencing employed in the study suffers from a massive carryover, which obscures the determination of successive phosphorylation sites. In any case, it is interesting that the three proline-directed protein kinases so far tested phosphorylate Ser 551 exclusively or preferentially. Only Cdk5-p23 phosphorylates Ser 553 , but the kinase phosphorylates Ser 551 as well, although only the former was found phosphorylated in vivo as has been shown in the present study. One possibility is that a protein kinase or kinases other than the  7. Effects of synapsin I phosphorylation by MAP kinase on the synapsin I-tubulin interaction. Synapsin I was incubated with polymerized tubulin, and the mixture was subjected to ultracentrifugation as described under "Experimental Procedures." The supernatants (S) and the pellets (P) thus obtained were analyzed by SDS-gel electrophoresis. Lanes 1 and 2, tubulin only; lanes 3 and 4, synapsin I only; lanes 5 and 6, tubulin and dephospho-synapsin I; lanes 7 and 8, tubulin and synapsin I phosphorylated by MAP kinase. Note the mobility shift of synapsin I caused by the MAP kinase-dependent phosphorylation. ones tested are responsible for the phosphorylation of Ser 553 . The other explanation is that Ser 551 is preferentially dephosphorylated by protein phosphatase(s). Since the present study on the in vivo phosphorylation state of synapsin I represents only a "snapshot" of the total brain, similar studies conducted with cells and tissues under various physiological stimulations may give an answer to the question. The in vivo and in vitro phosphorylation sites so far identified are summarized in Fig. 8.
Phosphorylation of synapsin I at Ser 568 and at Ser 605 by CaM kinase II abolishes the bundling activity of actin filaments (23). As shown in the present study, phosphorylation by MAP kinase showed a similar effect on the F-actin bundling activity, while Cdk5-p23-dependent phosphorylation had practically no effect. Since MAP kinase and Cdk5-p23 both phosphorylate a similar site in the tail region of synapsin I (Ser 551 or Ser 553 ), the effects caused by the MAP kinase-dependent phosphorylation should be due to the phosphorylation of the two serine residues in the head region (Ser 62 and Ser 67 ). The major actin-binding site has been reported in the globular head domain of the synapsin I molecule (32), and the presence of a second binding site in the tail region has been predicted (33). At the moment it is not clear whether the region around the two phosphorylation sites by MAP kinase is directly involved in the synapsin I-actin binding or if the conformational change caused by the phosphorylation is responsible for the diminished interaction. On the contrary, the binding of synapsin I to tubulin was not affected by the phosphorylation either by MAP kinase or Cdk5-p23. This may suggest that the binding sites for the two major cytoskeletal elements are different, and the interactions with them may be regulated differentially. Physiological function of the phosphorylation at the novel site found in the present study (Ser 553 ), therefore, still remains to be established. Studies on the interaction of synapsin I with other cellular components such as synaptic vesicles and Grb2, a SH3-containing signal transduction protein (34) may give an answer to the question.
Whether the MAP kinase-or Cdk5-p23-dependent phosphorylation occurs in vivo and whether the phosphorylation reactions are involved in the regulation of neurotransmitter release still remains to be seen. However, it should be noted that a mobility shift of synapsin I in SDS gel electrophoresis occurs in PC12 cells after nerve growth factor stimulation (35). According to these authors nerve growth factor induces a novel phosphorylation site of synapsin I at a site other than those by CaM kinase and by cAMP-dependent protein kinase. The phosphorylation has previously been attributed to that by Cdc2-cyclin A (1,30), but the present study demonstrated that the mobility shift is induced only by MAP kinase-dependent phosphorylation but not by Cdk5-p23-dependent phosphorylation. The fact that phosphorylation by Cdk5-p23 or that by Cdc2-cyclin A does not cause any detectable mobility shift (30) suggests that the protein kinase involved in nerve growth factor-dependent phosphorylation of synapsin I in PC12 is MAP kinase. In conclusion, proline-directed protein kinases that include MAP kinase, Cdk5-p23, and (probably) other unknown protein kinases seem to play important roles in the physiological regulation of cytoskeletal components during neurotransmitter release, and there seem to be complex interactions between various protein kinases not only upstream of the signal transduction pathways but also at the substrate protein level (10,11).