Determination of in vivo phosphorylation sites in protein kinase C.

The primary structure of rat protein kinase C βII was probed by high pressure liquid chromatography directly coupled to an electrospray ionization mass spectrometer and by high energy collision-induced dissociation analysis to identify in vivo phosphorylation sites. The N-terminal methionine was found to be cleaved post-translationally and replaced with an acetyl group. Four phosphopeptides were identified. Two peptides, Thr500-Lys520 and Glu490-Lys520, are phosphorylated at Thr500 greater than 90%. Peptide His636-Arg649 is phosphorylated about 75% at Thr641. It is the only site that was previously identified during the in vitro autophosphorylation studies (Flint, A. J., Paladini, R. D., and Koshland, D. E., Jr.(1990) Science 249, 408-411). The fourth peptide Asn650-Lys672 is phosphorylated at Thr660. A discussion of the potential implication of these results follows.

The primary structure of rat protein kinase C ␤II was probed by high pressure liquid chromatography directly coupled to an electrospray ionization mass spectrometer and by high energy collision-induced dissociation analysis to identify in vivo phosphorylation sites. The N-terminal methionine was found to be cleaved post-translationally and replaced with an acetyl group. Four phosphopeptides were identified. Two peptides, Thr 500 -Lys 520 and Glu 490 -Lys 520 , are phosphorylated at Thr 500 greater than 90%. Peptide His 636 -Arg 649 is phosphorylated about 75% at Thr 641 . It is the only site that was previously identified during the in vitro autophosphorylation studies ( Phosphorylation is a rapid and reversible means of regulating protein activity. Its efficiency is evident in the many signal transduction pathways that use cascades of phosphorylation to effect cellular responses (1)(2)(3). Protein kinase C plays a major role in many of these pathways (4 -6). It is a serine/threonine kinase dependent on calcium and phospholipids and activated by diacylglycerols, fatty acids, or phorbol esters at physiological calcium concentrations (7). 12 members of the mammalian protein kinase C family have been identified so far (8). Regions of conservation as well as proteolysis studies indicate that protein kinase C is comprised of two domains, an N-terminal regulatory domain and a C-terminal catalytic domain (9,10).
Protein kinase C autophosphorylates itself in vitro on both its regulatory and catalytic domains (11). Autophosphorylation is particularly intriguing in that it has been shown to be an intramolecular reaction (12), in which regions very distinct in the primary sequence have access to the active site (13). When separated from the regulatory domain by proteolysis, the catalytic domain is no longer able to autophosphorylate, even though it is still fully active against substrates (12).
Six in vitro autophosphorylation sites have been identified in the ␤II isozyme (13). Ser 16 and Thr 17 are located close to the autoinhibitory sequence in the primary structure. Thr 314 and Thr 324 are located in the hinge region between the catalytic and regulatory domains. Thr 634 and Thr 641 are in the C terminus and are the only sites conserved in all the conventional protein kinase C isozymes. These residues are outside the region conserved in most other serine/threonine kinases. Recent studies in vitro and in vivo have elucidated a definite role for phosphorylation of protein kinase C. Phosphorylation by a second kinase is thought to be necessary in the activation of the kinase in vivo (14,15). Mutagenesis studies of Thr 497 and Thr 500 in protein kinase C isozyme ␣ and ␤, respectively, have proposed phosphorylation of those residues as critical for activity in vivo and/or in vitro (15)(16)(17). In addition, mutations of the in vitro autophosphorylation sites in protein kinase C ␤I suggest a role for the C-terminal sites, Thr 635 and Thr 643 in protein kinase C localization, activation, and down-regulation (18).
Since previous results indicated that protein kinase C is phosphorylated in vivo and that phosphorylation is essential for activation, it is important to determine whether these phosphorylation sites are identical to in vitro autophosphorylation sites. The baculovirus expression system was chosen for protein kinase C expression because of the ease of purifying a single isozyme. Previous work has shown that the gel mobility and in vitro autophosphorylation pattern were identical between protein kinase C ␤II overexpressed in insect cells and purified from rat brain (13). In addition, phosphatase-treated protein kinase C from both sources exhibit similar gel shifts, suggesting identical phosphorylation patterns (19).
Mass spectrometry has been successfully used for the determination of phosphorylation sites in various proteins, such as the chemotaxis response regulator protein from Escherichia coli (20), bovine myelin basic protein (21), bovine mitogenactivated protein kinase (22), and bleached bovine rhodopsin (23). HPLC 1 directly coupled with electrospray ionization mass spectrometry (LC/ESIMS) provides means for quick and efficient screening of entire protein digests for covalent modifications (24). LC/ESIMS analysis yields singly or multiply protonated peptide ions, and from the m/z value of these ions, the molecular mass of the corresponding peptide can be determined. Similarly, liquid secondary ionization mass spectrometry (LSIMS) analysis usually yields only molecular weight data in the form of protonated peptide ions. These ions can be activated by collision with inert gas atoms, such as helium. The dissociation induced this way usually reveals the amino acid sequence of the peptide analyzed. Collision-induced dissociation (CID) analysis offers a tool to determine the amino acid sequence of the peptide and the exact site(s) of phosphorylation (25); while under reaction conditions required for Edman degradation, the phosphate group is hydrolyzed or eliminated from phosphorylated serines or threonines. Tandem mass spectrometry permits CID analysis in mixtures by allowing the precursor ion selection for collisional activation. Therefore, we used LC/ESIMS and high energy CID analysis to identify the in vivo phosphorylation sites in protein kinase C ␤II.

MATERIALS AND METHODS
Isolation of Protein Kinase C-The ␤II isozyme of rat protein kinase C (26) was expressed and purified according to the procedures described earlier (13). Sf9 or Sf21 insect cell lines were infected with a recombinant baculovirus, and the enzyme was purified by chromatography on DE-52 anion exchange resin, a phosphatidylserine affinity matrix, and a Mono Q column. Protein kinase C elutes from the Mono Q column in 20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, and 1 mM dithiothreitol at 200 -300 mM KCl. This mixture was dialyzed to equilibrium (19 h) against 500 ml of 0.1 M NH 4 HCO 3 (pH 7.9).
Alkylation-Approximately 750 pmol of protein kinase C were dissolved in 150 l of 6 M guanidine HCl, 200 mM Tris-HCl (pH 8.0). Cysteine residues were reduced with 3 mM dithiothreitol at 60°C for 1 h and alkylated with sodium iodoacetate (6.15 mM) at room temperature for 1.5 h in the dark. The reagent excess was removed by dialysis against approximately 2.5 liters of 100 mM NH 4 HCO 3 buffer (pH 7.8).
Digestion with Trypsin-The carboxymethylated protein was incubated with about 3% (w/w) L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Worthington) at 37°C for 18 h. The enzyme was added in aliquots at the beginning of the digestion and 4 h later.
Reversed-phase HPLC-The tryptic peptides were separated by reverse phase HPLC (Vydac C 18 , 1.0 mm, inner diameter ϫ 250-mm column) using an ABI 140A solvent delivery system. Solvent A was 0.1% trifluoroacetic acid in water; solvent B was 0.08% trifluoroacetic acid in acetonitrile. The column was equilibrated in 2% B, and the gradient was started at 5 min after the injection. A 10% solvent B concentration was reached in 5 min, and then the amount of solvent B was linearly increased to 50% over 100 min. The fractions were manually collected.
Asp-N Subdigest-Phosphopeptide-containing fractions (estimated peptide-content was about 600 pmol) were incubated with 0.2 g of endoproteinase Asp-N (Boehringer Mannheim) in 100 l of 50 mM sodium phosphate buffer (pH 7.2) at 37°C for 20 h. The resulting peptides were analyzed by LC/ESIMS, LSIMS, and CID.
Chymotrypsin Subdigest-Tryptic peptides (490 -520) and (650 -672) were first incubated in 70 mM NH 4 HCO 3 buffer (pH 7.8) with approximately 2% (w/w) chymotrypsin at 37°C for 1.5 h. In the following experiments, the amount of enzyme was increased to 6%, and the incubation time was 5 h. Components of the digests were separated by reversed-phase HPLC and analyzed by LSIMS.
HPLC/ESIMS-A dual syringe pump (Carlo Erba Fisons) was used to deliver mobile phase at a flow rate of 50 l/min. Microbore HPLC separations were performed on an Aquapore 300 C18 microbore column, 1.0 mm, inner diameter ϫ 100 mm (Applied Biosystems) or on the Vydac column mentioned above. Column effluent was monitored by a variable wavelength UV detector (Applied Biosystems) equipped with a high sensitivity capillary flow cell (LC Packings) at 215 nm. Postcolumn addition of 2-methoxyethanol/isopropanol (1:1) (27) was accomplished by a separate syringe pump (Isco) connected to a 3.1-l dead volume PEEK mixing tee (Upchurch Scientific), positioned after the UV detector. After the mixing tee, the column effluent was split at a ratio of 1:20; approximately 5% of the sample entered the mass spectrometer at a flow rate of 3-5 l/min, while the remaining sample was manually collected for subsequent analyses. The microbore HPLC system was interfaced to a VG Biotech/Fisons Bio-Q mass spectrometer equipped with an electrospray source. Typical operating voltages were as follows: probe tip, 4200 V; counter electrode, 550 V; and sampling orifice, 40 -50 V. The source temperature was maintained at 60°C. The mass spec-trometer was scanned in non-continuum mode over a range of m/z 350-2000 at 5 s/scan.
LSIMS Analysis-These experiments were performed on a Kratos MS 50S double focusing mass spectrometer, equipped with a cesium ion LSIMS source (28). Glycerol:thioglycerol 1:1 mixture containing 1% trifluoroacetic acid was used as a liquid matrix.
High Energy CID Analysis-These experiments were carried out using a Kratos Concept IIHH four-sector mass spectrometer of EBEB geometry equipped with an LSIMS source, a continuous flow sample introduction probe, a scanning array, and a charge-coupled device detector (29,30). The C 12 isotope peaks of the MH ϩ ions were selected as precursor ions in the first mass spectrometer. The collision energy was 4 keV. The collision gas was helium. Its pressure was adjusted to attenuate the precursor ion intensity by about 70%. The second mass spectrometer was scanning in B/E mode at a resolution of 1000.
Amino Acid Sequence Analysis by Edman Degradation-This analysis was performed on an ABI 470A gas phase sequencer.

RESULTS
To study the post-translational modifications of protein kinase C, we expressed protein kinase C ␤II in insect cells. The tryptic digest of the carboxymethylated protein was analyzed both by reversed-phase HPLC, followed by LSIMS, and by on-line LC/ESIMS. Tryptic peptides were identified by comparison of the molecular masses observed with those predicted from the published sequence (26); more than 90% of the sequence was identified (Fig. 1). The missing components are small hydrophilic peptides. Post-translational or other covalent modifications can be indicated by discrepancies between the predicted and observed molecular masses. For example, the expected N-terminal tryptic peptide at mass to charge ratio (m/z) 1897.9 was not detected; however, a molecular mass (1809.0 Da) observed in the LC/ESIMS experiment and later measured also by LSIMS (MH ϩ at m/z 1808.8) suggested that the N-terminal methionine had been replaced by an N-acetyl group. This hypothesis has been confirmed by high energy CID High energy CID analysis was used to confirm the peptide sequence and identify the exact site of phosphorylation. High energy CID processes lead to bond cleavages all along the peptide backbone (30). The proton can be retained on either of the newly formed species, yielding an ionic and a neutral fragment; the mass spectrometer only detects ionized molecules. Ions with charge retention at the N terminus are designated as a, b, and c ions, while those at the C terminus are designated as x, y, and z ions, respectively. Fragment ions a and x are the products of a bond cleavage between the ␣-carbon and the carbonyl group. Ions b and y are formed from cleavage of the peptide bond itself. Fragments c and z are generated when the cleavage occurs between the amino group and the ␣-carbon. Fragment ions v, w, and d are formed by a backbone and a side chain cleavage, with charge retention on the C or the N terminus, respectively (30,31). The expected mass values of the fragments can be calculated for peptides with known amino acid sequence.
The peptide of molecular mass of 1677.6 Da was subjected to high energy CID analysis, which confirmed the amino acid sequence as His 636 -Arg 649 and the presence of a phosphate group at Thr 641 (Fig. 3). This peptide was observed without modification as well (rtϳ 37 min). Based on the relative ion abundances of the phosphorylated and non-modified peptides from LC/ESIMS analysis and in vitro autophosphorylation studies (13), it is estimated that this site is phosphorylated at least 75%.
Since both peptides Glu 490 -Lys 520 and Thr 500 -Lys 520 were observed with a molecular mass increase of 80 Da, and peptide Glu 490 -Lys 499 was observed only without the phosphate group, it can be deduced that the modification occurs either on Thr 500 or Thr 504 . Phosphopeptide Glu 490 -Lys 520 was subjected to digestion with various enzymes to produce smaller peptides more suitable for high energy CID experiments. The peptide proved to be resistant to chymotrypsin, and endoproteinase Glu-C removed only the C-terminal nine amino acids. Digestion with endoproteinase Asp-N eventually yielded a phosphopeptide in the desired molecular weight range, D 494 GVTTKTFC*GTP 505 with MH ϩ at m/z 1364.6, which was then subjected to high energy CID analysis (Fig. 4). Fragment ions with charge retention at the N terminus for the first six residues do not indicate the presence of any covalent modification. However, N-terminal fragment ion a 7 (at m/z 755) that results from a cleavage between the ␣ carbon of Thr 500 and its carbonyl group exhibits an 80-Da mass shift, corresponding to a phosphate group. Similarly all the other N-terminal ions containing Thr 500 display this 80-Da mass shift. C-terminal ion y 2 , which is formed via peptide bond cleavage between Gly 503 and Thr 504 with charge retention at the C terminus, was detected at m/z 217, thus indicating no covalent modification at Thr 504 . Thus, the modification occurred at Thr 500 . A peptide for Thr 500 -Lys 520 with no modification was detected as a minor component in the LC/ ESIMS experiment (rtϳ 51 min, Fig. 5). Peptide Glu 490 -Lys 520 was only detected with the modification. The site occupancy for Thr 500 is estimated to be higher than 90%.
The identity of phosphopeptide Asn 650 -Lys 672 was confirmed by Edman degradation (see Table I). The mass is increased by a single 80-Da increment, indicating one phosphate group per peptide. Since the peptide contains three possible phosphorylation sites, Ser 654 , Ser 660 , and Ser 664 , attempts were made to produce peptides containing individual phosphorylation sites. The peptide was resistant to endoproteinases Glu-C and Asp-N; endoproteinase Asp-N was tried since it was reported to cleave at the N terminus of not only aspartic acids but also at the N terminus of other negatively charged residues such as cysteic acids (32) and glutamic acids (24). Chymotryptic digestion yielded a phosphopeptide, Asn 650 -Phe 666 , still containing all three serine residues. Based on the UV and LC/ESIMS data, the occupancy of on peptide Asn 650 -Lys 672 is estimated to be greater than 80%. Keranen, Dutil, and Newton have informed us 2 that the phosphorylation occurs at Ser 660 . This correlates well with the high degree of conservation of Ser 660 in comparison with Ser 654 and Ser 664 (Fig. 6).

DISCUSSION
Three distinct sites of phosphorylation, Thr 500 , Thr 641 , and one of the serines on phosphopeptide, Asn 650 -Lys 672 , were determined by mass spectrometry (Fig. 7). Each site was phosphorylated greater than 75%. Thr 500 lies in the conserved serine/threonine kinase catalytic region. Thr 641 lies outside that conserved region, but the residue itself is conserved in the protein kinase C family. Phosphopeptide Asn 650 -Lys 672 is at the C terminus and lies within a region defined as variable among the major members of the protein kinase C family.
Of the autophosphorylation sites previously identified in vitro (13), only Thr 641 is detected in this analysis of unstimulated sample. The fact that 75% of the sample is already phosphorylated at this residue explains the apparent low labeling level detected in the in vitro autophosphorylation studies. A single mutation to alanine at the corresponding residue in the ␤I isozyme decreases activity in vivo, and the mutant is no longer able to autophosphorylate (33). Recent work using phosphatase treatment and subsequent autophosphorylation of protein kinase C ␤II has suggested that protein kinase C is solely  FIG. 2. HPLC chromatogram of carboxymethylated protein kinase C tryptic digest. The tryptic peptides (ϳ70 pmol) were separated by reversed-phase HPLC on a Vydac C 18 , 1.0 mm, inner diameter ϫ 250-mm column. Solvent A was 0.1% trifluoroacetic acid in water, and solvent B was 0.08% trifluoroacetic acid in acetonitrile. The eluant was monitored at 215 nm. Phosphopeptide His 636 -Arg 649 started to elute in peak 1. Peak 2 contained both the non-modified and phosphorylated peptides for this sequence. Peptides corresponding to non-modified and phosphorylated Thr 500 -Lys 520 eluted in peak 3 (See Fig. 5). Phosphopeptide Glu 490 -Lys 520 eluted in peak 4. Phosphopeptide Asn 650 -Lys 672 eluted in peak 5. These species were not fully separated and coeluted with other tryptic peptides. Peak 6 contained peptide Asn 650 -Lys 672 without any covalent modification. AUFS (absorbance units full scale) gives the relative peak absorbance. responsible for phosphorylation of this residue (19).
It is notable that only one of six in vitro autophosphorylation sites was found to be phosphorylated in this study. The difference between the level of protein kinase C stimulation in vitro (activation by diacylglycerol, Ca 2ϩ , and phosphatidylserine) (13) and in vivo (no artificial stimulation) could explain a lack of autophosphorylation. However, one of the sites, Thr 641 , is phosphorylated, which suggests that protein kinase C was activated in vivo. If protein kinase C autophosphorylates itself at Thr 641 , why are the other autophosphorylation sites also not phosphorylated? Possible explanations are (a) degree of accessibility of Thr 641 relative to the other sites, (b) sensitivity or resistance to phosphatases, (c) specific post-translational processing of protein kinase C (19), (d) a second kinase phosphorylating only Thr 641 (33), or (e) discrepancies between the plasma membrane in vivo and detergent micelles in vitro. Further work will be needed to clarify this issue.
The observed phosphorylation of protein kinase C at Thr 500 is interesting in view of known serine/threonine kinase structures. In protein kinase A (34), a phosphorylated residue at this position is necessary for the integrity of the active site structure. In the modeled protein kinase C structure, a phosphorylated threonine would be able to interact with surrounding residues in a manner very reminiscent to protein kinase A (17). These residues are conserved in many other serine/threonine kinases (35). Complete in vivo phosphorylation at Thr 500 supports the conclusion that it is the activating phosphorylation site conserved in many kinases (36).
Phosphorylation at Thr 500 agrees with biochemical evidence that this residue is critical for activity. Mutagenesis studies demonstrated that Thr 497 and Thr 500 in the ␣ and ␤II isozymes, respectively, were essential for activity (16,17). The authors suggested that a second kinase must be phosphorylating and thus activating protein kinase C. In fact, replacement of Thr 500 in the ␤II isozyme with glutamate restored complete activity and suggests that it is the phosphorylation of the residue that is critical (17). Thorsness and Koshland (37) have shown that an aspartate can mimic the presence and an alanine the absence of an inhibitory phosphorylation in isocitrate dehydrogenase. In protein kinase C, it appears that the larger glutamate is better able to maintain the integrity of the active site.
These studies are in agreement with our identification of greater than 90% in vivo phosphorylation at Thr 500 .
Deletion studies of protein kinase C ␣ (38) suggest that phosphorylation near the C terminus is critical for protein kinase C activity. Truncation of 23 amino acids from the C terminus fully inactivates the kinase (38); Ser 660 corresponds to the 16th residue from the C terminus in the ␣ isozyme (Fig.  6). In addition, the high degree of conservation of Ser 660 suggests a potential family-wide regulation (Fig. 6).
We have determined in vivo phosphorylation sites of unstimulated protein kinase C ␤II. All three of these regions appear to play a strong role in protein kinase C function. Phosphorylation at a particular residue such as Thr 500 may be of structural importance. The other phosphorylated sites may be involved in substrate recognition or activator affinity. The fact that activators of protein kinase C increase the phosphorylation state while epidermal growth factor decreases the FIG. 5. Electrospray mass spectrum of protein kinase C tryptic peptide Thr 500 -Lys 520 non-modified and phosphorylated. This spectrum was recorded in an LC/ESIMS experiment from approximately 70 pmol of the tryptic digest. The average molecular masses of the peptides observed are shown. The calculated average molecular masses are 2405.7 and 2485.7 Da for the non-modified and for the phosphopeptide, respectively.

TABLE I
Edman degradation of a protein kinase C ␤II tryptic peptide with MH ϩ ϭ 2771.0 Serines in parentheses had a low recovery. The sequence matches residues 650 -662 and confirms the mass identification of the peptide as residues 650 -662 with a single phosphate ester group. phosphorylation state (39) suggests that phosphorylation is an important means of regulating protein kinase C activity.