Chemical Conversion of Aspartyl Peptides to Isoaspartyl Peptides A METHOD FOR GENERATING NEW METHYL-ACCEPTING SUBSTRATES FOR THE ERYTHROCYTE D-ASPARTYL/L-ISOASPARTYL PROTEIN METHYLTRANSFERASE*

Mammalian protein carboxyl methyltransferases have recently been proposed to recognize atypical con-figurations of aspartic acid and may possibly function in the metabolism of covalently altered cellular proteins. Consistent with this proposal, the tetrapeptide tetragastrin, containing a single “normal” L-aspartyl residue (L-Trp-L-Met-L-Asp-L-Phe-NH2) was found here not to be an in vitro substrate for erythrocyte carboxyl methyltransferase activity. However, chemical treatment of tetragastrin by methyl esterification and then de-esterification of the aspartic acid residue yielded a mixture of peptide products, the major one of which could now be enzymatically methylated. We show here that this new peptide species is the isomeric &aspartyl form of tetragastrin (L-iso-tetragastrin; L-Trp-L-Met-L-A@-L-Phe-NH2), and it appears that iso-merization proceeds via an intramolecular succinimide intermediate during the de-esterification procedure. L-iso-Tetragastrin is stoichiometrically methylated (up to 90% in these experiments) with a K , for the enzyme of 5.0 p ~ . Similar chemical treatment of several other L-aspartyl peptides also resulted in the formation of (12 mg/ml of protein), S-adenosyl-~-[methyl-~H]methi-onine (14 pM; 630 dpm/pmol), and a buffer consisting of 40 mM sodium citrate, 1.5 mM sodium phosphate, 3 mM 0-mercaptoethanol at pH 6.0 in a final volume of 0.1 ml. The enzymatic reactions proceeded linearly to 30 min, within which time each reaction was quenched by an equal volume of 0.2 M NaOH, 1% sodium dodecyl sulfate. The quantity of [3H]methanol derived from enzymatically formed esters was determined by the vapor diffusion assay described under “Materials and Methods.” The reaction velocities were cor-rected by subtracting the background of methylation of protein in the enzyme solution (1 pmol/min/mg) measured when peptide was not added to the incubation mixture. Purified L-iso-tetragastrin trifluoroacetate (350 pmol) was incubated with partially purified erythrocyte carboxyl methyltransferase (10 pg of protein) and 2800 pmol of S-adenosyl-methionine. The final reaction volume (40 pl) also contained S- adenosylhomocysteine hydrolase from rabbit erythrocytes (4 pl; 30 units/pl containing 11 mg/ml enzyme protein and 20 mg/ml bovine serum albumin, Sigma), adenosine deaminase from hog kidney (1 pl of an ammonium sulfate suspension; 2 units/pl, Sigma), and the reaction was buffered by 75 mM sodium citrate, 1.5 mM sodium phosphate, 1.5 mM Na-EDTA, 3% glycerol, 1.5 mM p-mercaptoetha- nol, pH 6. After incubation (37 ‘C) for 60 min, the reaction was quenched with 50 pl of 1 M acetic acid, and 15 pl of the quenched sample (equivalent to 58 pmol of the starting peptide) was analyzed by the standard HPLC conditions described under “Materials and Methods.” Based on UV absorbance, 90% of the starting L-iso-tetragastrin was enzymatically converted to the methylated product, L-isotetragastrin methyl ester.

Mammalian protein carboxyl methyltransferases have recently been proposed to recognize atypical configurations of aspartic acid and may possibly function in the metabolism of covalently altered cellular proteins. Consistent with this proposal, the tetrapeptide tetragastrin, containing a single "normal" L-aspartyl residue (L-Trp-L-Met-L-Asp-L-Phe-NH2) was found here not to be an in vitro substrate for erythrocyte carboxyl methyltransferase activity. However, chemical treatment of tetragastrin by methyl esterification and then de-esterification of the aspartic acid residue yielded a mixture of peptide products, the major one of which could now be enzymatically methylated. We show here that this new peptide species is the isomeric &aspartyl form of tetragastrin (L-iso-tetragastrin; L-Trp-L-Met-L-A@-L-Phe-NH2), and it appears that isomerization proceeds via an intramolecular succinimide intermediate during the de-esterification procedure. Liso-Tetragastrin is stoichiometrically methylated (up to 90% in these experiments) with a K , for the enzyme of 5.0 p~. Similar chemical treatment of several other L-aspartyl peptides also resulted in the formation of new methyltransferase substrates. This general method for converting normal aspartyl peptides to isoaspartyl peptides may have application in the reverse process as well.
Recent studies of the substrate specificities of enzymatic protein carboxyl methylation reactions have helped to define the functions of these reactions in cells (Clarke, 1985). For example, analysis of the site-specific methylation of a set of L-glutamic acid residues in bacterial chemoreceptor proteins has led to a detailed model for the regulation of receptor function through reversible covalent modification of these sites (Terwilliger and Koshland, 1984;Kehry et al., 1983;Boyd et al., 1983;Stock et al., 1985).
A very different type of protein carboxyl methylation re-action, studied mainly in mammalian cells, appears to occur only at aspartic acid sites in nonconventional configurations.
This conclusion is based on the recovery of D-aspartic acid pmethyl ester from proteolytic digestions of radioactively methylated erythrocyte proteins and on the in uitro methylation of a synthetic peptide related to adrenocorticotropin but containing a @-isomerized L-aspartyl residue (L-Tyr-L-Val-L-Pro-L-AwGly-L-Ala) (McFadden and Clarke, 1982;O'Connor andClarke, 1983,1984;Aswad, 1984a;Murray and Clarke, 1984). To explain this unusual enzyme specificity, several related models have been proposed suggesting that this protein carboxyl methyltransferase, which had previously been characterized and purified on the basis of its substoichiometric recognition of a wide range of proteins (Paik and Kim, 1980), is in fact a protein D-aspartyl/L-isoaspartyl carboxyl methyltransferase which recognizes these covalently altered sites that may arise spontaneously in many proteins. The possible cellular functions of this enzymatic methylation reaction have also been discussed (Clarke, 1985). In one such model, the enzymatic methyl esterification of altered protein sites leads to further covalent rearrangements of these sites, resulting in their ultimate conversion back to "normal" L-aspartyl or L-asparaginyl sites in the cell. A proposed intermediate in this process of rearrangement, a five-membered succinimide ring, has been shown in model studies to form rapidly when enzymatically methylated peptides related to ACTH' undergo spontaneous demethylation at physiological pH (Johnson and Aswad, 1985;Murray and Clarke, 1986). Hydrolysis of the succinimide then leads to the formation of a mixture of isoaspartyl and normal aspartyl residues. In a sense then, the enzymatic methylation "activates" the site, lowering the barrier to the formation of the ring intermediate and to further rearrangements.
We considered that this proposed activation and rearrangement of aspartyl sites in cells might be simply modeled by the methylation and demethylation of aspartyl peptides using strictly chemical procedures. We report here that the peptide L-Is0 FIG. 1. Proposed chemical reactions for the covalent rearrangement of L-aspartyl peptides. Tetragastrin, in its normal form, contains the single free @-carboxyl group of Asp-3. This carboxyl group can be methyl esterified by treatment with acidic methanol, and subsequent displacement of the ester group by the neighboring amide nitrogen of Phe-4 results in the formation of an intramolecular five-membered succinimide ring. Ring opening by hydroxide ion attack at the a-carboxyl position leads to an isomerized peptide bond between Asp-3 and Phe-4. Ring opening at the @-position yields a normal peptide configuration identical to the starting material. If the intermediate succinimide peptide is prone to racemization at the aspartyl center (Clarke, 1985), then the final mixture of peptides may also contain 3-[D-aspartyl]tetragastrin in the normal and is0 configurations. carboxyl methyltransferase and in doing so have extended the known range of peptide sequences recognized in vitro by this enzyme.

RESULTS
Esters of aspartyl residues in synthetic polypeptides are characteristically labile due to the favorable intramolecular formation of succinimide derivatives (Bernhard, 1983;Johnson and Aswad, 1985;Murray and Clarke, 1986). This mechanism of intramolecular ester cleavage can also explain the rapid rates of demethylation at physiological pH of enzymatically formed protein methyl esters (Terwilliger and Clarke, 1981;Barber and Clarke, 1985). Hydrolysis of peptide succinimides can occur at both the a-and P-carbonyl groups; this results in the formation of a mixture of normal and isomerized peptides. We decided to study the chemistry of methylation-Portions of this paper (including "Materials and Methods") are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 86M-499, cite the authors, and include a check or money order for $3.20 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press. Enzymatic carboxyl methylation of chemically treated tetragastrin. Commercially available tetragastrin HCl (96% pure as determined by HPLC) was treated as follows: 0, chemical esterification of 4.8 mg by 1.0 ml of 0.1 M HCl in methanol, solvent removal, and then hydrolysis in 1 M ammonium acetate (pH 9.5, 0.8 mg of peptide/ml) for 17 h at 37 'C; 0, pH 9.5 treatment alone for 17 h (2.4 mg of peptide/ml); ., aqueous 0.1 M HC1 treatment alone for 24 h (0.3 mg of peptide/ml); 0, untreated commercial tetragastrin. All samples were lyophilized and redissolved in H20. Enzymatic methylation of the treated peptides was then determined at the indicated concentrations by incubation at 37 "C with erythrocyte cytosol (12 mg/ml of protein), S-adenosyl-~-[methyl-~H]methionine (14 p M ; 630 dpm/pmol), and a buffer consisting of 40 mM sodium citrate, 1.5 mM sodium phosphate, 3 mM 0-mercaptoethanol at pH 6.0 in a final volume of 0.1 ml. The enzymatic reactions proceeded linearly to 30 min, within which time each reaction was quenched by an equal volume of 0.2 M NaOH, 1% sodium dodecyl sulfate. The quantity of [3H]methanol derived from enzymatically formed esters was determined by the vapor diffusion assay described under "Materials and Methods." The reaction velocities were corrected by subtracting the background of methylation of protein in the enzyme solution (1 pmol/min/mg) measured when peptide was not added to the incubation mixture.
induced aspartyl rearrangements using the tetrapeptide tetragastrin as a model peptide (Fig. 1).
Formation of a Methyl-accepting Peptide by Chemical Treatment of Tetragastrin-Protein carboxyl methyltransferases from brain and erythrocytes apparently do not recognize normal L-aspartyl peptides as substrates (Murray and Clarke, 1984;Aswad, 1984a), and this was found to be the case here with normal tetragastrin. Untreated commercial tetragastrin was not detectably methyl esterified by carboxyl methyltransferase activity in crude erythrocyte cytosol (Fig. 2). However, preliminary experiments indicated that simply by chemically esterifying tetragastrin and then treating this methyl ester with mild base, we could form material which could now be methylated by enzyme activity in erythrocyte cytosol (Fig. 2).
We performed control experiments to determine whether both the esterification and de-esterification treatments are necessary to form enzyme substrate. As shown in Fig. 2, acidic treatment (pH 2) similar to that of the esterification step did not lead to the formation of methyl acceptor substrate. Alkaline pH treatment alone, as in the de-esterification step, did generate a small level of enzyme substrate (Fig. 2). However, later experiments indicated that the commercial tetragastrin HCl contained a 1% contaminant of succinimide, and alkaline treatment of this imide may be responsible for the formation of a low amount of enzyme substrate. Mild alkaline treatment of purified tetragastrin produced no detectable enzyme substrate. FIG. 3. Chromatographic separation of purified tetragastrin and tetragastrin methyl ester. The trifluoroacetate salts of tetragastrin (2.2 nmol, upper trace) and tetragastrin methyl ester (1.7 nmol, lower truce) were analyzed by the standard conditions of reversed phase liquid chromatography described under "Materials and Methods." The amount of material was determined by amino acid analysis as in Table I. Analysis of the peak areas allows us to calculate an extinction coefficient at 214 nm of 35,000 M" cm" (+ 2000).

TABLE I
Amino acid composition, D/L-aspartate analysis, and mass spectra of peptides Peptides were purified by HPLC and recovered as lyophilized trifluoroacetate salts. Approximately 1 nmol was acid hydrolyzed, and a portion of the dried residue was analyzed by the orthophthalaldehyde method described under "Materials and Methods." Compositions were normalized to the aspartic acid content. D/L-Aspartate ratios were determined as described under "Materials and Methods." Fast atom bombardment positive ion mass spectrometry was performed on a modified Kratos MS9 instrument operated by Dr. Dilip Sensharma at the UCLA Department of Chemistry and Biochemistry. Samples (approximately 10-50 pg) were suspended in 20 p1 of glycerol and were applied to a room temperature stainless steel electrode that was bombarded by xenon atoms emitted at a gun voltage of 4-6 kV and a current of 1 mA. Parent ion masses are indicated for protonated (M+1) and sodium species (M+23); values in parentheses are the expected masses for these positive ions. Not shown are the masses of Drominent fragment ions which were consistent with the expected peptide sequences (Schafer, 1983). FIG. 4. Hydrolysis of tetragastrin methyl ester at pH 9.5 and enzymatic methylation of the product peptides. Tetragastrin methyl ester (50 p~ in 1 mM HCl) was added to an equal volume of 1 M NH, acetate, pH 9.5, and incubated at 23 "C. At various times, aliquots representing 0.5 nmol of starting ester material were quenched by 1 volume of 1 M acetic acid and then analyzed by the standard conditions of reversed phase HPLC described under "Materials and Methods." Elution positions are indicated for tetragastrin methyl ester, tetragastrin imide, L-iso-tetragastrin, and normal tetragastrin. For the final time point (26 h), an aliquot equivalent to 15 nmol of starting ester material was quenched by 1 M acetic acid, and the solvent was removed by lyophilization before HPLC analysis. The peptides eluted relatively early in this final chromatogram (26 h) due to the larger sample load. Several unidentified peaks and shoulders (*) were reproducibly present and may include the Daspartyl derivatives of iso-tetragastrin and tetragastrin which have been found to migrate after the corresponding L-derivatives (P. N. McFadden and S. Clarke, manuscript in preparation). Fractions (1 ml) from the 26-h chromatogram were lyophilized, and the residual material was tested for the presence of methyl-accepting peptides by incubation (37 "C) with 100 p1 of a mixture containing 0.5 mg of erythrocyte cytosol protein and 2 nmol of [methyl-3H]-S-adenosylmethionine (440 dpm/pmol) in a final buffer composition of 140 mM sodium citrate, 1.5 mM sodium phosphate, 1.5 mM Na,-EDTA, 4.5 mM P-mercaptoethanol, 3% glycerol, pH 6. The enzymatic incubations were quenched after 30 min by 100 pl of 0.2 M NaOH, 1% sodium dodecyl sulfate. The graph ut the top of the figure shows the resulting enzymatic carboxyl methylation, determined as [3H]methanol with the vapor diffusion assay.

~~~~ ~
Tryptophan, destroyed by acid hydrolysis, was determined by UV absorbance measurements of the intact peptides, with calculations based on the extinction coefficient of free tryptophan (see also Table   11).
We next followed a standard protocol (Kim et al., 1983) to partially purify erythrocyte carboxyl methyltransferase activity, using esterified/de-esterified tetragastrin as the methyl acceptor in the enzyme assay. It was found that during ammonium sulfate precipitation and gel filtration chromatography, the peptide-methylating activity behaved identically with protein carboxyl methyltransferase activity (assayed by the methylation of ovalbumin). We conclude that the activity that recognizes esterified/de-esterified tetragastrin is the same as the well-characterized erythrocyte protein carboxyl methyltransferase (Kim, 1974;Kim et al., 1983).
Tetragastrin, purified to homogeneity by reversed phase HPLC (Fig. 3), was converted to the @-methyl ester with acidic methanol. The esterification reaction, performed on a 1.6-pmol scale as described under "Materials and Methods," was 28% complete after 1 h, 48% complete after 2 h, and 98% complete after 19 h as measured by HPLC. No products other than the starting material and the ester were detected a t time Tetragastrin methyl ester (1.1 mM in 1 mM HCl) was added to 4 volumes of 0.1 M sodium phosphate, pH 7.4, and incubated at 37 "C. At various times aliquots were removed and quenched with an equal volume of 1 M acetic acid. Samples equivalent to 2.2 nmol of the starting material were then analyzed by the standard reversed-phase chromatographic conditions described under "Materials and Methods." Left, HPLC traces. The elution positions of tetragastrin methyl ester, tetragastrin imide, and the L-is0 and normal forms of tetragastrin are shown. UV-absorbing buffer components eluted at early times in these runs, and an unidentified reproducibly formed peak is noted by an asterisk. Right, the areas beneath the identified peaks in the chromatograms were integrated and converted to molar quantities by assuming a molar extinction coefficient at 214 nm of 35,000 "' cm" for each identified peptide. The curves were generated as theoretical solutions to the differential equation describing first-order hydrolysis of tetragastrin methyl ester (0) to form tetragastrin imide (0) ( k = 5.6 X rnin"), and first-order hydrolysis of the imide to give the observed ratio of 41 for L-iso-tetragastrin (0) and normal tetragastrin (W) ( k = 6.1 X min-').

Peptide Substrates for Protein Carboxyl MethyEtransferase
points up to 2 h; after 19 h a third unidentified peak, migrating about 10 min later than the ester peak on HPLC, accounted for 9% of the UV absorption. Following the 19-h reaction, the tetragastrin methyl ester was purified by HPLC (Fig. 3), and the structure of the ester was confirmed by its amino acid composition and its molecular mass (Table I).
When tetragastrin methyl ester was incubated in a pH 9.5 buffer a t 23 "C, several new peptide species formed (Fig. 4).
A new peak of material, tentatively identified as peptide succinimide, formed at early times and then gradually disappeared. Two other major new peaks formed and were stable; one of these eluted identically as normal L-tetragastrin, and the other was tentatively identified as L-iso-tetragastrin. The higher yield of the isomerized species, formed at four times the level of the normal peptide, was consistent with previous patterns of succinimide cleavage (Bernhard, 1983;Johnson and Aswad, 1985;Murray andClarke, 1984, 1986). Several minor peaks were also evident by the end of the ester hydrolysis (Fig. 4, asterisks). Importantly, only the isopeptide material was found to act as a substrate for erythrocyte carboxyl methyltransferase activity (Fig. 4).
When the hydrolysis experiment was performed at physio-logical temperature and pH (37 "C, pH 7.4), the chemical hydrolysis of tetragastrin methyl ester was slower than at pH 9.5, with a half-time of ester loss of 124 min (Fig. 5). Again, as at pH 9.5, the imide formed as an intermediate product.
The imide then hydrolyzed with an apparent half-life of 114 min to a mixture of L-is0 and normal peptides (Fig. 5, right).
The imide structure was confirmed by recovering it in pure form and demonstrating its hydrolysis to is0 and normal forms with alkaline treatment (data not shown).
By the end of the pH 7.4 ester hydrolysis (Fig. 5), 80% of the starting ester material was converted to L-iso-tetragastrin. The remaining 20% was recovered as normal tetragastrin. Fewer minor products were formed by pH 7.4 hydrolysis (Fig.  5) compared to the pH 9.5 condition (Fig. 4). It is possible that the pH 9.5 ammonium acetate buffer was reactive, and some of the minor peaks appearing at pH 9.5 are amidated or acylated versions of tetragastrin. In support of this possibility, the de-esterification procedure was also performed at pH 9.5 using a sodium borate buffer, and the minor products were eliminated (data not shown). FIG. 6. Products of tetragastrin methyl ester hydrolysis analyzed after cyanogen bromide cleavage. L-iso-Tetragastrin and normal tetragastrin were purified following the hydrolysis of tetragastrin methyl ester and were then cleaved by cyanogen bromide treatment. The fragments of this cleavage were analyzed by HPLC as described under "Materials and Methods." Standards of L-ASP-L-PheNH2, L-isoAsp-t-PheNH,, and the uncleaved standards of normal tetragastrin and L-iso-tetragastrin were included as internal standards in several analyses, with their elution positions as shown here. The peak marked by the bold arrow was present in each cyanogen bromide reaction mixture, and its UV absorbance was consistent with the expected N-terminal cleavage fragment L-Trp-L-Hse, where Hse is homoserine. Upper trace, cyanogen bromide cleavage of 700 pmol of purified L-iso-tetragastrin. Middle trace, cleavage of 800 pmol of normal tetragastrin. Lower trace, cleavage of a mixture of 400 pmol of normal tetragastrin and 350 pmol of L-iso-tetragastrin. tetragastrin were purified by reversed phase HPLC and were greater than 95% pure when rechromatographed. Several approaches were used to verify the structure of the L-is0 species. First, the amino acid composition and mass of L-isotetragastrin were identical to that of normal tetragastrin (Table I) as expected for an isomerized product. Second, the aspartyl-phenylalanine bond in the L-isopeptide was completely resistant to proteolysis by leucine aminopeptidase (Table 11) as would be expected for either an L-is0 form or a D-aspartyl species. However, the measured D-aSpartiC acid in this peptide (4.2%) was only slightly higher than the background value resulting from the hydrolysis procedure (Table  I). Finally, cleavage of L-iso-tetragastrin by cyanogen bromide resulted in a peptide fragment which co-purified with a standard of L-iso-Asp-L-Phe-NH,, the expected cleavage product (Fig. 6).

Characterization of L-iso-Tetragustrin-Following
The purified L-iso-tetragastrin was a good substrate for partially purified erythrocyte protein carboxyl methyltransferase. This enzyme preparation contained both of the isoe-

Leucine aminopeptidase digestwn of L -~S O -and normal tetragastrin
Chromatographically purified tetragastrin (130 pmol) and L-isotetragastrin (140 pmol) from the pH 9.5 hydrolysis of tetragastrin methyl ester were digested (37 "C) in 40-pl volumes containing 44 pg of hog kidney leucine aminopeptidase (Sigma; 100 units/mg) and buffered by 0.05 M Tris-HC1, 2.5 mM MgCl,, pH 8, for 19 h. Ratios of released amino acids were then quantified by the orthophthalaldehyde derivatization method described under "Materials and Meth- Stoichiometry of enzymatic carboxyl methylation of L-iso-tetragastrin. Purified L-iso-tetragastrin trifluoroacetate (350 pmol) was incubated with partially purified erythrocyte carboxyl methyltransferase (10 pg of protein) and 2800 pmol of S-adenosylmethionine. The final reaction volume (40 pl) also contained Sadenosylhomocysteine hydrolase from rabbit erythrocytes (4 pl; 30 units/pl containing 11 mg/ml enzyme protein and 20 mg/ml bovine serum albumin, Sigma), adenosine deaminase from hog kidney (1 pl of an ammonium sulfate suspension; 2 units/pl, Sigma), and the reaction was buffered by 75 mM sodium citrate, 1.5 mM sodium phosphate, 1.5 mM Na-EDTA, 3% glycerol, 1.5 mM p-mercaptoethanol, pH 6. After incubation (37 'C) for 60 min, the reaction was quenched with 50 pl of 1 M acetic acid, and 15 pl of the quenched sample (equivalent to 58 pmol of the starting peptide) was analyzed by the standard HPLC conditions described under "Materials and Methods." Based on UV absorbance, 90% of the starting L-isotetragastrin was enzymatically converted to the methylated product, L-isotetragastrin methyl ester. lectric variants of the erythrocyte methyltransferase that have been isolated? L-iso-Tetragastrin saturated the erythrocyte activity with a K,,,, measured at pH 7.4, of 5.0 p~ (data not shown). At pH 6, where enzymatically formed esters are relatively stable, the activity methylated the L-isopeptide to essentially stoichiometric levels ( Fig. 7) with 90% of the Lisopeptide methylated within 60 min of enzymatic reaction time. To methylate the peptide as in Fig. 7, it was necessary to decompose the end-product inhibitor S-adenosylhomocysteine by including S-adenosylhomocysteine hydrolase and adenosine deaminase in the incubation mixture. The structure of the enzymatically methylated peptide, L-iso-tetragastrin methyl ester, was verified by chemically synthesizing a standard of this isopeptide ester from purified L-iso-tetragastrin and methanol/HCl and showing that the enzymatic product Commercially available peptides were purified by reversed phase HPLC, and aqueous solutions were prepared of the trifluoroacetate salts. The peptide concentrations and the expected amino acid compositions were verified by phenylisothiocyanate derivatization following acid hydrolysis as described under "Materials and Methods." Samples, containing 0.5-1 mg of peptide, were lyophilized and then methyl esterified in a volume of 200 p1 (0.1 M HCl in methanol, 23 "C, 22 h). The esterification reaction mixture was then lyophilized to dryness and treated under de-esterification conditions (0.1 M sodium phosphate, pH 7.4, 25 "C, 30 h). Peptides without treatment by esterification and de-esterification served as controls and were brought up in this same pH 7.4 phosphate buffer. Both the untreated and treated peptides were assayed for enzymatic methylation with the peptide concentrations referring to equivalents of starting material. The enzyme activity measurements were performed in 20-p1 mixtures containing partially purified erythrocyte carboxyl methyltransferase (5 pg of protein), [methyl-3H]S-adenosyImethionine (280 pmol; 440 dpm/pmol) and buffer consisting of 40 mM sodium phosphate, 3 mM EDTA, 6% glycerol, 10 mM 0-mercaptoethanol, pH 7.4. The assays were quenched after 20 min of incubation (37 "C) by 20 pl of 0.2 M NaOH, 1% sodium dodecyl sulfate, and the resulting [3H]methanol was quantified by the vapor diffusion assay. The background (zero peptide) has been subtracted from the sample measurements. The results are listed as averages of duplicate measurements f the range. <Glu, pyroglutamic acid.
Studies with Other Peptides-Several other aspartyl peptides were chemically esterified and de-esterified a t p H 7.4 and were then tested as substrates for erythrocyte carboxyl methyltransferase. The first four peptides listed in Table I11 were not detectably methylated by the activity until they underwent chemical treatment. Each of these peptides contains an aspartyl residue in a unique sequence and at a variable distance from the amino and carboxyl ends. Peptide 4 (Table 111) also contains an asparagine residue which may possibly deamidate and rearrange spontaneously (Johnson et al., 1985;Clarke, 1985), although it is not known if the present chemical conditions would promote the deamidation of this peptide. The precise structures of the methyl acceptor species obtained from peptides 1 through 4 by chemical esterification/ de-esterification treatment have not been determined, but by analogy to the results with tetragastrin it is expected that L-is0 forms of the peptides are chemically generated. Unlike the situation with the C-terminally amidated tetragastrin, peptides 1 and 2 (Table 111) have a free C terminus that would be esterified during the chemical procedure. We have not assayed for C-terminal methyl esters or for the loss of Cterminal esters in the de-esterification step.
As expected, the last two peptides in Table I11 (peptides 5 and 6) were not enzymatic substrates either before or after chemical treatment. Neither peptide contains an aspartyl group, but each does contain a glutamic acid site. Esters of glutamyl sites in polypeptides are relatively stable groups at mildly alkaline pH (Terwilliger and Clarke, 1981;Kleene et al., 1977) and are probably not easily subject to intramolecular displacement with subsequent glutarimide formation and bond rearrangements. It appears that an aspartyl residue is necessary for the generation of new methyltransferase substrates by this method. DISCUSSION We have developed a general method to convert normal Laspartyl residues to @-isomerized L-aspartyl residues (Fig. 1).  In this study, we have used a two-step procedure to obtain Liso-tetragastrin from normal tetragastrin in 80% yield. This L-isoaspartyl form of tetragastrin is recognized and methylated by erythrocyte protein carboxyl methyltransferase activity, whereas the normal form is These experiments thus demonstrate that normal aspartyl peptides can be rearranged by simple chemical treatments, and this chemical method can be usefully applied to generate new methyltransferase substrates. A parallel chemical method may also be useful to convert isopeptides back to normal aspartyl peptides, although the yield of this reaction may be low since isopeptide formation is generally favored (Bernhard, 1983;Johnson and Aswad, 1985;Murray and Clarke, 1986).
Using this methodology, we have extended the known in vitro range of substrate specificity of erythrocyte protein carboxyl methyltransferase. We find that a wide range of aspartyl peptides can be converted into methyltransferase substrates. For one peptide, L-iso-tetragastrin, the precise structure of the new methyl acceptor species has been determined. The other newly generated substrates (Table 111) may also be L-isoaspartyl peptides, although other structures should be considered. It is possible that the chemical methods developed here can result in the epimerization of normal Laspartate residues to yield D-aspartyl derivatives which may act as enzymatic substrates. In fact, preliminary work with longer reaction times and higher temperatures during the deesterification of tetragastrin methyl ester at pH 9.5 has resulted in product peptides with large D-aspartic acid content. Such D-aspartic acid derivatives of tetragastrin have not yet been found to be enzymatically methylated, but other Daspartic acid-containing peptides might be expected to be It has been reported that unpurified commercial tetragastrin can be methylated by the rabbit erythrocyte protein carboxyl methyltransferase to the extent of 0.009 mol/mol (Gosselin and Liss, 1985). It seems likely that this reaction reflects the methylation of L-isotetragastrin possibly formed by the hydrolysis of contaminating tetragastrin imide (see "Results"). The peptides shown were each tested as enzymatic substrates using preparations of human erythrocyte protein carboxyl methyltransferase. Assays were performed at pH 6.0 except in the case of L-iso-tetragastrin which was tested at pH 7.4. Value given relative to that observed with saturating concentration of ovalbumin as a methyl acceptor. At the indicated concentration of substrate, methylation was not detected.
J. Lowenson and S. Clarke, unpublished data. enzyme substrates, since methylated D-aspartic acid residues have been previously isolated from erythrocyte proteins (McFadden and Clarke, 1982;O'Connor andClarke, 1983, 1984;O'Connor et al., 1984). From the present work, it is evident that a wide variety of residues can surround an enzymatically recognized aspartyl site. In summarizing our findings here along with the earlier work on ACTH-related sequences, it is apparent that internal peptide sequences of the form X-(Asp)-Y can be methylated where (Asp) implies an atypical configuration of an aspartyl site (iso-L-Asp or possibly D-As~) and X has thus far been proline, methionine, phenylalanine, or lysine, and Y has been glycine, phenylalanine, alanine, or valine.
The total number of peptide bonds in a carboxyl methyltransferase substrate can also vary over a wide range. L-iso-Tetragastrin, with three peptide bonds and a C-terminal amide bond, is the shortest substrate for the methyltransferase to have yet been completely characterized (Table IV). To further address the issue of substrate length, we tested several dipeptides at up to millimolar concentrations and found that the enzyme did not detectably recognize the sequences Nacetyl-L-Asp, N-acetyl-D-Asp, N-acetyl-L-Asn, ~-Asp-Gly, D-Asp-Gly, Gly-L-Asn, or Gly-D-Asn (Table IV; data not shown). To our knowledge, tripeptides have not been assayed as methyl acceptor substrates.
Little is now known of the amino acid sequences surrounding methyl acceptor sites in uiuo. Whether the sites defined here and in other in vitro studies resemble in vivo sites will not be known until a methylated site in a cellular protein is fully examined. Our in vitro work indicates that erythrocyte protein carboxyl methyltransferase has broad sequence specificity that may enable it to recognize many altered protein sequences. An important question now is which sequences are most likely to become altered in cellular proteins. The likelihood of rearrangement is presumably dependent on many structural and environmental factors; it is of interest to note that succinimides are likely intermediates in the generation of cellular L-isoaspartyl and D-aSpartyl residues from protein L-aspartyl and L-asparaginyl residues (Johnson and Aswad, 1985;Aswad, 1984a;Johnson et al., 1985;Murray and Clarke, 1984). Protein concentration was determined by uv absorbance a 8 described (Murray and Clarke, 1984). Enzvme Purification. All steps were performed on ice or in a 4-C cold room. Solid ammonium sulfate was added to stirred cytosol to 559 saturation 132.6 g per 100 ml) to precipitate protein carboxyl methyltransferase away from the bulk of the hemglobin (Kim et al., 1983).
The material in the 55% ammonium sulfate pellet was dissolved in Buffer c (above1 and dialyzed overnight against 25 volumes of Buffer C (3500 nw cut-off dialysis tubing, Spectraporl.
Following dialysis, 6 ml (60 mg protein) was layered on top of a column ( 1 . 5 x 90 cml of Sephadex G-75 gel column was eluted (0.5 ml min-ll with Buffer C and fractions ( 1 0 minl were filtration medium (Pharmacia. superfine) equilibrated in Buffer c. The collected and assayed for carboxyl methyltransferase activity as described below. Separate methyltransferase assays were performed using either ovalbumin ( 4 0 mg/mll or ssterified/de-esterified tetragastrin as methyl acceptor substrate. The peak fractions of carboxyl mthyltransferase activity were pooled, and this solution was concentrated approximately ten-fold by osmosis by transferring it to a 3500 nw Cut-off dialysis bag and Packing the outside of the bag for several hours with dry Sephadex G-200 gel.

Vapor Diffusion ASSW for nethyltransferase and nethyl-Accepting substrates.
This assay method is based on the determination of radioactive methanol derived from enzymatically-formed methyl esters after the general procedure of Macfarlane (1984). S-Adenosylmethionine-HSO4 (Boehringer nannheim) was mixed with S-adenosyll3H-methyl]-L-metbionina (15 ci mol-1, Amersham) to achieve the given specific activities of the radiolabel. The measured as described (Murray and Clarke. 1984). Methyl-accepting concentration and radiochemical purity of the S-adenoaylmethionine were mixed and incubated at 37OC as described in the Figure Legends and Tables. substrate, s-adenosylI3H-methyll-L-methionine, and methyltransferas8 were These assays were quenched by an equal volume Of 0.2 M NaOH/lP SDS to hydrolyze radioactive methyl esters. Portions of the quenched mixtures (50 "11 were immediately transferred to filter papers that were then inserted into the necks of polyethylene scintillation vials containing 10 ml of ACS I1 liquid scintillation counting medium (~mershaml as described by Murray and Clarke (1986). The capped vials were allowed to stand for three hours hours (by which time parallel standards of 14C-methanol showed 95s transfer to allow diffusion of 3H-methanol into the counting medium. After three to the counting medium) the filter papers were removed and the with quench correction by measurements of internal radioactive standards. radioactivity in the vials was measured by liquid Bclntillation counting, Peptide Characterization Amino Acid Analysia.
Tetragastrin peptides were acid hydrolyzed with 6 PI HC1 solution at 105 OC for 18 hours in vacuo. Tetrapastrin peptides were enzymatically hydrolyzed by leucine aminopeptidase as described in Table 11. Amino acid residues from these procedures were analyzed by derivatization with orthophthalaldehyde essentially as described (  hydrolyzed as above and the residue was reacted with an optically active and fluprogenic reagent previously described (Aswad, 1984b). The resulting diastereomeric adducts of 0 and I-aspartic acid with N-acetyl-L-cysteine and orthophthalaldehyde were separated and the D-to L-aspartic acid ratio determined as described by Aswad (1984b) and modified by Wurray and Clarke (19841.

D/L AsDsrtic Acid Analysis. Tetragastrin peptides Were acid
Cyanoqen bromide cleavage reaction. The cyanogen bromide cleavage of tetragastrin peptides was performed with one-volume of the peptide (dissolved in water) and one-volume of reagent (0.01 g/ml cyanogen bromide, J.T. Baker, in 888 formic acid. Wallinckrodt) in polyethylene t u b e s and lyophilized. and the residue was resuspended in water for analysis by incubated in a 5 0 T water bath. After one hour, the reactions were frozen HPLC. Samples were injected onto the Waters/Alltech System described in the HPLC section above, and elution was by a linear gradient of 0% to 406 solvent B over 40 minutes. cyanogen bromide cleavage of L-iso-tetragastrin was synthesized by coupling Synthesis of standard L-iso-Asp-Phe-NHZ. This expected product of the N-carbobenzoxy-L-aspartic acid alpha-methyl ester (1 -1, Vega) to L-phenylalanine amide (1 m o l , Vega) in tetrahydrofuran I 6 ml) using dicyclohexylcarbodiimide 11.1 m o l l as coupling agent with 2 m o l removal oE the solid dicyclohexylurea by filtration, the solvent was 1-hydroxybenzOtfiaZole and 1.2 m o l triethylamine also present. After extractions with ethyl acetate, aqueous sodium bicarbonate and aqueous removed by vacuum and neutral peptide material was recovered after solvent citric acid. The neutral material (25% yield) was deprotected by one hour treatment with palladium catalyst on a polyneric support (Pierce Palladium-PEI beads) suspended in 1:4 formic acid (888)/tetrahydrofuran. A single UV214 absorbing product was evident upon puritication by preparative reversed phase HPLC (above) where the product eluted slightly before a standard of normal L-Asp-L-Phe-NH2 (Sigma). The product was judged to be I.-iso-Asp-L-Phe-NH2 by it8 failure to be cleaved by leucine aminopeptidase L-Asp, 1 mole Phe).
(the normal standard was cleaved) and by its amino acid composition ( 1 mole