Protein phosphatase 2A is reversibly modified by methyl esterification at its C-terminal leucine residue in bovine brain.

We have recently described a novel protein carboxyl methylation system that results in the reversible modification of a 36-kDa polypeptide component of a 178-kDa protein in the cytosol of a variety of eucaryotic cells. This reaction, catalyzed by a cytosolic 40-kDa methyl-transferase, results in the methyl esterification of the alpha-carboxyl group of the C-terminal leucine residue. We have now purified the major methylated 36-kDa polypeptide from bovine brain. N-terminal sequence analysis of a tryptic fragment of this polypeptide revealed identity to the catalytic subunit of protein phosphatase 2A. This enzyme exists in the cell predominantly as a trimeric 151-kDa native species containing the 36-kDa catalytic polypeptide that terminates in a leucine residue. We then fractionated bovine brain cytosolic extracts to separate the major phosphatase isoforms 2A1 and 2A2 and found that both could be methylated by a partially purified preparation of the methyltransferase. A synthetic C-terminal octapeptide based on the sequence of the 36-kDa catalytic subunit is neither a substrate nor an inhibitor of this methyltransferase, suggesting that this enzyme recognizes aspects of the tertiary and/or quaternary structure of the native phosphatase. Because this modification reaction is readily reversible in extracts, it may represent a novel strategy of the cell to modulate the function of this protein phosphatase.

The reversible modification of protein function by phosphorylation and dephosphorylation reactions has now been clearly established as a major component of both metabolic regulation and signal transduction pathways. While much effort has been focused on the characterization of the large number of protein kinases, especially with regard to their activation by ligands and second messengers (Hunter, 1987;Blackshear et al., 1988;Taylor et al., 1990;Fantl et al., 1993), it has been only more recently appreciated that a variety of protein phosphatases are also subject to cellular regulation (Cohen, 1989;Cohen and Cohen, 1989;Pot and Dixon, 1992;Walton and Dixon, 1993). The level of protein phosphorylation can thus be controlled by modulation of both protein kinases and phosphatases.
We have been interested in the regulation of cell function by another type of post-translational modification involving protein methylation and demethylation reactions. Four types of protein methyltransferases have been described that can modify carboxyl groups on polypeptides (Clarke, 1985; * This work was supported by Grant GM26020 from the National Institutes of Health. The UCLA Protein Microsequencing Facility was supported in part by a BRS Shared Instrumentation Grant RR05554 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This with 18 U.S.C. Section 1734 solely to indicate this fact. article must therefore be hereby marked "advertisement" in accordance 1992; Barten and O'Dea, 1990;Xie and Clarke, 1993). One of these is a widely distributed enzyme that modifies abnormal aspartyl residues on age-damaged proteins and may play a general role in repair (Johnson et al., 1987;Galletti et al., 1988;Li and Clarke, 1992). The other three types of protein carboxyl methylation systems described so far modify specific protein species and can mediate reversible regulation reactions. For example, the output of certain bacterial chemoreceptors is regulated by the degree of methylation of a set of glutamate residues (Stock et al., 1992). In eucaryotic cells, a membranebound enzyme catalyzes the methyl esterification of isoprenylated C-terminal cysteine residues in reactions that may foster interactions of signaling proteins with their upstream and downstream partners (Clarke, 1992;Backlund, 1992;Philips et al., 1993;Hrycyna and Clarke, 1993). Most recently, a cytosolic methylation system has been described in eucaryotic cells that results in the modification of one or more 36-kDa polypeptides in protein(s) of native molecular mass of about 178 kDa (Xie and Clarke, 1993). Here, the methylation reaction results in the formation of a methyl ester on the a-carboxyl group of a C-terminal leucine residue (Xie and Clarke, 1993).
We have focused our efforts on understanding the role of this latter methylation reaction. The methylation of cytosolic 36-kDa polypeptides has been observed previously in a variety of human, mouse, and rat tissues (O'Connor and Clarke, 1984;Siege1 and Wright, 1985;Chelsky et al., 1985;Ladino and O'Connor, 1990;Volker et al., 1991). In this study, we were interested in identifying the nature of the 178-kDa methylaccepting species. We have now purified the methylated component from bovine brain and present evidence that the 36-kDa methylated polypeptide is the catalytic subunit of protein phosphatase 2A. This enzyme has a broad substrate specificity for phosphoproteins that are involved in carbohydrate, amino acid, and lipid metabolism as well as in cell cycle control (Cohen, 1989;Ferrigno et al., 1993). The reversible methylation of this enzyme may play a role in coupling its activity to physiological signals in the cell. EXPERIMENTAL PROCEDURES Materials-Fresh bovine brain was purchased from the Shamrock Meat Co. (Los Angeles, CA). Rabbit muscle phosphorylase b (twice crystallized; 27 unitdmg of protein) and rabbit muscle phosphorylase kinase (170 unitdmg of protein) were obtained from Sigma. [Y-~~PIATP (>lo Cilmrnol) was from Amersham. L3H1AdoMet1 (15 CUmmol) was purchased from DuPont-New England Nuclear. The peptide RRTP-DYFL was synthesized by Dr. Joe Reeves at the UCLA Peptide Synthesis Facility.

1981
and internal sequence analysis was performed using the procedure of Fernandez et al. (1992). Briefly, the portion of the membrane containing the 36-kDa polypeptide (visualized after Amido Black staining) was incubatedat37 "Covernightwithtrypsin(massratioof36-kDapolypeptide: trypsin was 501). The resulting peptides were then separated by HPLC on a C18 reverse phase microbore column with a nonlinear gradient from 0.1% to 0.02% trifluoroacetic acid/80% acetonitrile over 120 min. Edman degradation was then performed using an Applied Biosystems Model 470A gas-phase sequenator with on-line HPLC detection. All of these procedures were performed by Dr. Audree Fowler at the UCIA Protein Microsequencing Facility.
Assay of Protein Phosphatase 2A-The phosphatase substrate [32Plphosphorylase a was prepared from phosphorylase b, [y-32P]ATP, and phosphorylase kinase as described by Brautigan and Shriner (1988) with the following modifications. Norit A-activated carbon (20 mg) was added directly to the solubilized preparation and was then removed by centrifugation. The [32Plphosphorylase a product was precipitated twice by adding an equal volume of saturated ammonium sulfate (pH 6.7) at room temperature, followed by incubation at 0 "C for 15 mi n. The preparation was then dialyzed against 2 liters of 20 n m Tris acetate, pH 6.8,50 m~ P-mercaptoethanol overnight at 4 "C. The resulting crystals of [32Plphosphorylase a were dissolved in glycero1:lOO m~ imidazole HCl, pH 7.2, 10 m~ caffeine, 1 m~ DIT (1:l) and stored at -20 "C. The specific radioactivity of this preparation was 930 cpm/pg of protein.
Protein phosphatase 2A activity was assayed using [32Plphosphorylase a (final concentration, 0.5 mg/ml) as a substrate in the absence or presence of 10 pg/ml protamine sulfate in a 50-pl total volume buffered with 50 m~ bis-Tris acetate, pH 7.0,2 m~ DTT, and 5 m~ caffeine. The reaction was camed out at 30 "C for 10 min. The release of acid-soluble [32Plphosphate was measured aRer precipitation in 10% trichloroacetic acid as described by Brautigan and Shriner (1988).
C-terminal Leucine Protein Methyltransferase-Bovine brain cytosol was prepared as described by Xie and Clarke (1993). This material (50 ml; 20 mg of proteidml) was dialyzed against buffer C (20 nm bis-Tris acetate, pH 6.8, 1 nm DIT, 0.2 nm EDTA) before fractionation on a DEAE-cellulose column (DE52 resin, 1.5-cm diameter x 15-cm length) equilibrated with buffer C at 4 "C. m e r washing the column with 80 ml of buffer C, proteins were eluted at 1 ml/min with a gradient of 0 to 1 M sodium acetate in a volume of 150 ml. The methyltransferase was assayed by mixing 25 pl of column fractions, 5 pl of 37 p~ PHIAdoMet (15 Ci/mmol), and 25 pl of a fraction containing its native polypeptide methyl-acceptor (fraction 46 from a Sepharose CL-4B column as described in Fig. 4 of Xie and Clarke (1993)). The incubation was carried out at 37 "C for 30 min followed by separation of the polypeptides by SDS-PAGE. 3H radioactivity present as methyl esters in the 36-kDa polypeptide was detected by a gel slice vapor phase diffusion assay as described (Xie and Clarke, 1993). The first peak of methyltransferase activity eluting at 0.28 M sodium acetate from the DEAE-cellulose column was concentrated by ultrafiltration and then further purified by gel filtration on Sephacryl S-200 resin under conditions similar to those described in Fig. 2 of Xie and Clarke (1993) to yield a preparation with a specific activity of 0.17 pmol of methyl groups transferred per midml using the assay described above.

Purification of the 36-kDa Methyl-accepting Substrate of the C-terminal Leucine Protein Carboxyl Methyltransferase-To
identify the 36-kDa polypeptide methyl-acceptor, we decided to purify the 3H-methylated protein that was previously characterized in cytosolic extracts of bovine brain (Xie and Clarke, 1993). Initial attempts at purification were limited by the instability of the [3H]methyl ester on the 36-kDa polypeptide. Because we found that the instability persisted under conditions where spontaneous ester hydrolysis would be minimized, we sought conditions where we might inactivate endogenous enzymatic esterase activities. We found that treatment of the cytosolic extract with PMSF, a covalent inhibitor of serine proteases and esterases, stabilized the [3H]methyl esters on the 36-kDa polypeptide and allowed us to follow its purification by monitoring the radioactivity in this polypeptide chain (Xie and Clarke, 1993).
As described in Fig. 1, we fractionated 3H-methylated bovine brain cytosol by DEAE-cellulose chromatography after initial ammonium sulfate precipitation and Sephacryl S-200 gel fil- matography. An extract of cytosolic proteins, prepared as described by ture for 1.5 h. Aportion of this material (14 ml) was then mixed with 1.4 m~ PMSF (prepared from a 0.25 M stock in ethanol) at room temperaml of 37 p~ [SH]AdoMet (15 Cilmmol) and incubated at 37 "C for 30 min. The remainder of the material was incubated with 5 p~ nonisotopicallylabeled AdoMet under similar conditions. The two reaction mixtures were then combined, and solid ammonium sulfate was added to a h a 1 concentration of 10% (w/w). M e r 15 min at room temperature, the precipitate was removed by centrifugation and the supernatant brought to 20% (w/w) ammonium sulfate at 0 "C. The pellet was resuspended in 4 ml of 20 m~ bis-Tris acetate, pH 7.0, 1 m~ D m , 0.2 m~ EDTA, 10 % glycerol, 25 p~ PMSF (buffer A) and desalted on a Sephacryl S-200 column at 4 "C in buffer A. Fractions containing the 3H-methylated 36-kDa polypeptide were identified by SDS-PAGE followed by a gel slice vapor phase diffusion assay (Xie and Clarke, 1993;Xie et al., 1990). These fractions were then loaded on a DEAE-cellulose column (Whatman DE52 resin; 1-em diameter x 15-cm length) equilibrated at 4 "C with buffer A. The column was washed with 50 ml of buffer A before starting a salt gradient (0 to 0.8 M sodium acetate in 300 ml of buffer A) at a flow rate of 15 ml/h. Fractions (3 ml) were collected and 50 pl of every third fraction was analyzed by SDS-PAGWgel slice vapor phase diffusion assays to locate the proteins containing the 3H-methylated 36-kDa polypeptide (filled circles). The absorbance at 280 nm is shown (open circles). tration chromatography. A major peak of radioactivity enriched in this material was found to elute at about 0.38 M salt. These fractions (91-101) were combined and further fractionated by hydrophobic chromatography on a phenyl-Sepharose column ( Fig. 2A). Here, the 3H-methylated 36-kDa polypeptide eluted in a broad peak from fractions 140 to 170 at a salt concentration of about 0.12 M. All of the radioactivity in these fractions is present as [3Hlmethyl esters in the 36-kDa polypeptide ( Fig.  2A). SDS-PAGE analysis of these fractions revealed a prominent 36-kDa band (Fig. 2 B ) . We find that the specific radioactivity of the 36-kDa polypeptide (determined as the ratio of the radioactivity and the silver-staining intensity in the 36-kDa band) is decreased in later-eluting fractions suggesting the partial resolution of methylated and unmethylated proteins containing the 36-kDa polypeptide. Enzymatic remethylation studies are consistent with this picture. For example, we find a 5.9-fold stimulation of methylation of fraction 157 polypeptides but a 95-fold stimulation of methylation of fraction 171 polypeptides (data not shown).
Sequence Analysis of the Purified 36-kDa Polypeptide-lb determine the identity of the 3H-methylated polypeptide purified in Figs. 1 and 2, we combined fractions 161 to 169 of the phenyl-Sepharose column and separated the 36-kDa polypeptide from the other polypeptide components by SDS-PAGE. No N-terminal sequence was found when the intact 36-kDa polypeptide was subjected to automated Edman degradation analysis. The polypeptide was then digested with trypsin, and internal peptides were purified by HPLC as described under "Experimental Procedures." We selected one well-resolved peak eluting at 50 min for N-terminal sequence analysis and obtained the following sequence (picomole yields of each residue are given in subscript): !I'yr14Glyl&n~7Alal&n15Val,Trps-Lys. We then searched the translated Genbank database (re- from the DEAN-cellulose column shown in Fig. 1 were combined, and solid ammonium sulfate was added to a final concentration of 0.5 M before loading on a phenyl-Sepharose column (1-cm diameter x 9-cm length) equilibrated a t 4 "C with 0.8 M ammonium sulfate in buffer A. After washing the column with 50 ml of the equilibration buffer, the proteins were eluted with a 300-ml gradient to buffer A. The ?H-methylated 36-kDa polypeptide was assayed in 50-pl aliquots a s described in Fig. 1 (filled circles); the total radioactivity in similar aliquots of each fraction is also shown (open circles). Panel B, 5O-pl aliquots of fractions were analyzed by SDS-PAGE as described (Xie and Clarke, 1993) and silver-stained. The migration of polypeptide standards ( S t d ) is indicated by small arrows. and these include phosphorylase b (97 m a ) , bovine serum albumin (66 m a ) . ovalbumin (45 kDa), carbonic anhydrase (31 kDa), soybean trypsin inhibitor (21.5 m a ) , and lysozyme (14 kDa) (Rio-Rad). The large arrow indicates the migration position of the 36-kDa polypeptide.
lease 76) using this sequence. We found 100% identity with amino acid residues 137-144 of the catalytic subunit of protein phosphatase 2A from human, pig, rat, rabbit, Drosophila, and the plant Brassica napus (rape). Although this database does not contain the bovine sequence of protein phosphatase 2A, it has been reported to differ only a t residue 55 from that of rabbit and pig (Cohen, 1989). Residues 137-144 are preceded by a lysine residue and would thus be expected to be contained within a single tryptic fragment.
This result suggests that the 36-kDa polypeptide methyl acceptor for the C-terminal leucine methyltransferase is the catalytic subunit of protein phosphatase 2A. This assignment is also consistent with the known features of both the methylated protein (Xie and Clarke, 1993) and this enzyme (reviewed in Cohen, 1989). The polypeptide chain molecular weight of the methylated species and the phosphatase catalytic subunit are both 36 kDa. The product of methylation is a C-terminal leucine methyl ester; the C-terminal residue of protein phosphatase 2A from all species tested is a leucine residue. The molecular weight of the native methylated protein has been estimated a t 178.000 by gel filtration chromatography; the native structure of the major isoform of phosphatase 2A is a 151-kDa trimer. The 36-kDa methylated species is found in the cytosol of a variety of mammalian species and yeast as well a s in wheat germ and nematodes (data not shown); highly conserved protein phosphatase 2A species are found in the cytosolic fraction of all eucaryotic tissues examined to date.
Methylation of Isoforms of Protein Phosphatase 2A in Bovine Brain-Different forms of protein phosphatase 2A, including PP-2A0, PP-2A1, and PP-2A2. can be resolved by DEAE-cellulose chromatography (Cohen, 1989). While isoforms PP-PA1 and PP-2A2 have been purified from several tissues (although not from bovine brain). PP-PA0 has been found only in rabbit skeletal muscle. Isoforms PP-2A1 and PP-2A2 both contain the 36-kDa catalytic subunit and a 60-kDa subunit, while PP-2A1 contains an additional 55-kDa subunit. To determine which isoforms are substrates for the C-terminal leucine methyltransferase, we fractionated bovine brain cytosol on a DEAE-Sephacel column using the approach described by Erickson and Killilea (1992) (Fig. 3). Using ["ZPlphosphorylase a a s a substrate, we found two major peaks of phosphatase activity corresponding to the PP-PA1 and PP-2A2 forms. The activity in both peaks was stimulated by protamine sulfate, a characteristic feature of this enzyme (Pelech and Cohen, 1985). As shown in Fig. 3, fractions containing either PP-SA1 or PP-2A2 are excellent methyl-accepting substrates for the C-terminal leucine protein carboxyl methyltransferase. To show that the methyl-acceptor in the PP-2A1 peak in Fig. 3 was in fact the phosphatase, we further fractionated this material using a phenyl-Sepharose column. As shown in Fig. 4, the methyl-accepting activity co-migrated with the protamine sulfate-stimulatable phosphatase activity.
Specificity of the C-terminal Leucine Methyltmnsferase-We tested whether a synthetic peptide RRTPDYFL. containing the Fraction Number   FIG. 4. Co-migration of protein phosphatase 2A activity and 36-kDa polypeptide methyl-acceptor activity on phenyl-Sepharose chromatography. Fractions 173-203 from the DEAE-Sephacel column shown in Fig. 3 were combined and fractionated by ammonium sulfate precipitation as described by Erickson and Killilea (1992). The precipitated proteins were resuspended in buffer B and loaded on a phenyl-Sepharose column (1-cm diameter x 15-cm length) equilibrated at 4 "C with buffer B containing 1 M ammonium sulfate. The column was eluted with a decreasing salt gradient (1 to 0 M ammonium sulfate in 300 ml followed by a 200-ml wash with buffer B). Protein phosphatase 2A activity in the absence of protamine sulfate (filled circles) and methyl-acceptor activity (open circles) were measured as described in Fig. 3 except for the methylation determination where 20 pl of each fraction was assayed and 2.5 pl of FSH1AdoMet was used.
C-terminal sequence of the PP-2A catalytic subunit from all species examined so far, was recognized by the C-terminal leucine methyltransferase. We did not detect methylation of the peptide by the partially purified enzyme at concentrations up to 1 mM, nor did we observe any inhibitory effect on the methylation of PP-2A in cytosolic extracts at concentrations up to 0.5 m~. Coupled with the previous observation that N-acetyl+ leucine is not recognized by this enzyme (Xie and Clarke, 19931, these results suggest that the methyltransferase requires additional elements of the phosphatase structure for catalysis. Physiological Consequences-Our discovery of this novel modification on the C-terminal leucine residue of the bovine brain protein phosphatase 2A suggests the possibility that this enzyme may be regulated via methylatioddemethylation reactions. The lability of the methyl ester in extracts, coupled with the stabilization by PMSF treatment, suggests the presence of an esterase activity that could work in concert with the methyltransferase to set the level of C-terminal modification. The methylation of the leucine residue described here is not the only modification of the C terminus of the catalytic subunit of PP-PA. A tyrosine residue at position 307, only two amino acids away from the C-terminal leucine residue, is phosphorylated in a reaction that deactivates the phosphatase (Chen et al., 1992). Protein phosphatase 2A2 is also modified (and deactivated) by a serinelthreonine kinase (Guo et al., 1993). The presence of multiple covalent modifications on a single protein can greatly expand the possibilities for precisely controlling metabolic pathways.