Protein Carboxyl Methyltransferase Facilitates Conversion of Atypical L-Isoaspartyl Peptides to Normal L-Aspartyl Peptides*

Prolonged incubation of L-isoaspartate-containing forms of lactate dehydrogenase (231-242), sperm activating peptide, and adrenocorticotropin (22-27) at 37 “C, pH 7.4, with S-adenosyl-L-methionine and pro- tein carboxyl methyltransferase from bovine brain leads to extensive conversion of the atypical isopeptide bond to a normal peptide bond. For the lactate dehy- drogenase-related peptide, conversion was 80% complete after 24 h. For the other two peptides, conversion reached a level of -65% after 48 h. The mechanism of conversion involves (i) rapid enzymatic methylation of the a-carboxyl of the L-iso-Asp residue; (ii) nonenzymatic demethylation resulting in formation of an L- aspartyl cyclic imide; and (iii) a slow, nonenzymatic hydrolysis of the cyclic imide to form a mixture of 15- 25% normal L-ASP peptide and 7 5 4 5 % L-iso-Asp peptide. The regenerated L-iso-Asp peptide is remethy- lated and the cycle is repeated. The extent of conversion is limited by a competing

yltransferases having this unusual specificity appear to be present in a wide range of cell types including rat PC12 cells, murine 70Z/3 lymphoma cells, Xenopus oocytes, and the bacterium Salmonella typhimurium (O'Connor and Clarke, 1985). This class of protein methyltransferase is distinct from the methyltransferase which modifies the y-carboxyl group of normal glutamyl residues of methyl-accepting proteins of chemotactic bacteria (Clarke, 1985).
At physiological pH and temperature, the ACTH iso-Asp methyl ester formed by the enzyme undergoes a rapid (tlh = 4-8 min), spontaneous demethylation resulting in formation of an aspartyl cyclic imide (Johnson and Aswad, 1985;Murray and Clarke, 1986). The imide subsequently becomes hydrolyzed (t% = 3-4.2 h) to generate a mixture of isoaspartyl (70-80%) and aspartyl (20-30%) peptides. The overall reaction pathway is shown in Fig. 1. This scheme predicts that prolonged incubation of an isoaspartyl peptide with the methyltransferase and AdoMet should lead to extensive conversion to the normal aspartate-containing form, since the isopeptide will be continually recycled through the pathway by the methyltransferase. The ability of protein carboxyl methyltransferase to facilitate conversion of atypical isoaspartyl peptides to normal aspartyl peptides has important implications for the function of this unusual enzyme. In the present paper, we have investigated the effects of incubating three unrelated synthetic isoaspartyl peptides with bovine brain protein carboxyl methyltransferase and AdoMet for periods up to 48 h under conditions of physiological pH and temperature. A preliminary report of this work has been published previously in abstract form .

RESULTS
General Strategy for Demonstrating Conversion of L-ISO-ASP Peptides to Normal, L-ASP Peptides-Attempts to demonstrate methylation-dependent conversion of L-iso-Asp residues to normal L -A s~ residues were made with three unrelated synthetic L-iso-Asp-containing peptides. We compared these peptides in order to determine possible effects of varying amino acid sequence on the specificity of protein carboxyl methyltransferase for L-iso-Asp and on the overall rate of Portions of this paper (including "Experimental Procedures" and an "Appendix" containing Figs. 9 and 10 and Tables I1 and 111) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full Table 111.
conversion of L-iso-Asp to L-As~. The structures of the L-iso-Asp peptides and their methylation kinetic constants are shown in Table I. The ACTH(22-27) peptide was chosen for study because the synthetic L-iso-Asp, L-As~, and L-cyclic imide peptides were available and because a considerable amount of information on the kinetics of the methylation and demethylation of the L-iso-Asp form had already been obtained in previous studies (Murray and Clarke, 1984;Aswad, 1984a;Murray and Clarke, 1986). Speract was chosen because the L-iso-Asp and L-ASP forms can be prepared easily from the commercially available native peptide (see "Appendix") and because it has a sequence distinct from that of the ACTH peptide except for the iso-Asp-Gly linkage. The LDH peptide was chosen because it was available in both the L-iso-Asp and L -A s~ forms from a previous study (Aswad et al., 1987), because it has a low K , for the methyltransferase, and because it provides additional information regarding sequence dependence, containing Asp or iso-Asp in a carboxyl linkage with Ser rather than Gly.
The general approach used to test for conversion of L-iso-Asp to L -A s~ was to incubate pure L-iso-Asp peptides (6-15 p~) with purified methyltransferase (2.5 p~) and excess AdoMet (200 p~) at 37 "C, pH 7.4, for varying periods of time, up to 48 h. Samples of the reaction mixtures were subjected to reversed-phase HPLC to separate the various intermediates and end products. After determining the identity of the HPLC peaks produced, we plotted the time course of each intermediate or product as a percentage of total peptide. In the following section, we describe in detail our results with the LDH L-iso-Asp peptide because it displayed the most rapid and extensive conversion of the three peptides. We then present a summary of our results with the speract and ACTH(22-27) L-iso-Asp peptides.
Conversion of the L-Iso-As~ LDH Peptide to a Normal, L-Asp Peptide- Fig. 2 shows sample HPLC profiles obtained after the L-iso-Asp LDH peptide (6 p~) was incubated for various times at pH 7.4, 37 "C with 2.5 p~ methyltransferase and 200 p~ AdoMet. Peak 11, which became detectable after ~~ L-Imide D-Imide D -A~P an hour of incubation, was by far the predominant form of the peptide present after 24 h. Because this peak comigrated with the authentic L-Asp-containing LDH peptide, it seemed that extensive conversion of L-iso-Asp to L -A s~ had occurred. In order to verify the identity of peak 11, it was collected from a 24-h reaction mixture and subjected to proteolysis using two proteases of differing specificity, pepsin and the Staphylococcus aureus V8 protease, endoproteinase Glu-C. For comparison, authentic L-iso-Asp-and L-Asp-containing forms of the LDH peptide were incubated with the proteases under identical conditions. Each of the proteolysis mixtures was then subjected to reversed-phase HPLC in order to obtain a profile of the fragments produced. The results obtained using pepsin are shown in Fig. 3. The profiles obtained for peak I1 and for the authentic L-ASP peptide were virtually the same. In contrast, the L-iso-Asp peptide yielded a proteolytic profile clearly different from that of the L -A s~ peptide. Similar results were obtained using endoproteinase Glu-C (not shown). These results indicate that peak I1 is indeed the L-Asp-containing form of the LDH peptide and hence that carboxyl methylation can convert L-iso-Asp residues to normal L -A s~.
Peak I in Fig. 2 consists of a mixture of two forms of LDH(231-242), the methylated and the cyclic imide forms. As shown elsewhere (Aswad et. al., 1987), peak I at early times (-10 min) consists mainly of the methylated isopeptide, since it alone carries the 3H label and the labeling is nearly stoichiometric. As the time of reaction increases, the amount of radiolabel relative to UV-absorbing material in the peak steadily decreases, indicating the presence of another form of the peptide in peak I. Because the hydrolysis of aspartyl carboxyl esters in a variety of Asp-X linkages, including Asp-Ser, are known to generate cyclic imides (Battersby and Robinson, 1955;Bernhard et al., 1962;Bodanszky and Kwei, 1978;Ondetti et al., 1968;Johnson and Aswad, 1985;Murray and Clarke, 1986), it seemed likely that the peptide remaining in peak I after the disappearance of methyl groups was a cyclic imide-containing LDH peptide. The peak I peptide was collected from a 1-h methylation reaction and incubated at

TABLE I Structures and methylation kinetic constants of ~-iso-Asp peptides used in this study
Kinetic constants were all determined using the purified type I isozyme of bovine brain protein carboxyl methyltransferase (Aswad and Deight, 1983  LDH peptides (25 p~) were incubated with 5 pg/ml pepsin at pH 2 and were then subjected to reversed-phase HPLC as described under "Experimental Procedures." The bottom panel is a protease blank, in which water was substituted for peptide. The peaks at the far right of the top three panels elute at the same retention time as the undigested peptides. The correspondence between the profiles obtained with peak I1 and the L-ASP peptide strongly suggests that these two peptides are identical. 37 "C in 50 mM K-HEPES (pH 7.4) for varying periods of time, followed by reversed-phase HPLC. Cyclic imides become hydrolyzed with a half-life of several hours under these conditions into a mixture of L-iso-Asp and L-ASP peptides, the iso-Asp form predominating (Sondheimer and Holley, 1954;Battersby and Robinson, 1955;Bodanszky and Kwei, 1978;Murray andClarke, 1984,1986;Blodgett et al., 1985;Johnson and Aswad, 1985). The peak I peptide became hydrolyzed with a tlh of 3.2 h and resulted in a 5.3:l ratio of isoaspartyl peptide to aspartyl ~e p t i d e .~ Thus, it appeared that the con-Because there is a significant amount of methyl ester in peak I after 1 h of methylation, loss of peak I will involve the serial reactions of demethylation and imide hydrolysis (cf. Fig. 1). Thus, the rate of cyclic imide hydrolysis must be greater than the rate of peak I disappearance obtained here. version of L-iso-Asp to L -A s~ in this peptide occurred via the mechanism outlined in Fig. 1.
After a 24-h methylation of the L-iso-Asp LDH peptide (Fig. 21, it was clear that, in addition to the L-ASP peptide, another form of the peptide (peak 111) had accumulated. Although peak I11 appeared to be a minor product, it seemed important to determine the nature of this peptide, which may represent the product of a physiologically significant side reaction. It has been proposed recently that the formation of a cyclic imide could greatly accelerate racemization at the aspartyl a-carbon (McFadden and Clarke, 1982;Clarke, 1985). This racemization reaction would ultimately lead to the formation of a D-iso-Asp peptide as the major product of D-cyclic imide hydrolysis (Fig. 1). Peak I11 was therefore collected from a 24-h methylation reaction and analyzed for the presence of D -A s~ after acid hydrolysis as described under "Experimental Procedures." D -A s~ was found to comprise 75.3% f 3.8% of the aspartate released. The peptide was also subjected to thin layer electrophoresis at pH 3.5, a method which allows one to distinguish between the presence of iso-Asp and Asp on the basis of the pK, of the free carboxyl group (see "Experimental Procedures"). It comigrated in this system with the authentic L-iso-Asp LDH peptide, indicating that the peptide contained D-iso-Asp rather than D -A s~.
The collected D-iso-Asp LDH peptide was tested for its ability to accept methyl groups from the methyltransferase under conditions which result in stoichiometric methylation of the L-iso-Asp peptide. These conditions included 10 ~L M peptide, 4 p~ methyltransferase, 200 ~L M [met/~yl-~H]AdoMet and were carried out at pH 6 and 30 "C for 1 h. The methylation barely exceeded 0.10 mol of methyl group/mol of peptide, indicating that the D-iso-Asp LDH peptide is, at best, a very poor methyl acceptor. These results provide further support for the stereospecificity of protein carboxyl methyltranferase previously indicated in studies on the methylation of ACTH(22-27) analogues (Murray and Clarke, 1984).
Because significant amounts of D-iso-Asp peptide were formed in the conversion reactions, it was important to determine whether peak 11, previously identified as an L-ASPcontaining peptide by proteolytic mapping, contained significant amounts of D -A s~. The peak therefore was collected from a 24-h methylation and analyzed in the same manner used for peak 111. This hydrolyzed peptide was found to contain 93.8% & 1.0% L-ASP, indicating that very little D -A s~ peptide had been formed.
Thus, the major products of the extended methylation of the LDH L-iso-Asp peptide are the corresponding L-ASPand D-iso-Asp-containing peptides, which are formed in a reaction involving the L-iso-Asp a-carboxyl methyl ester and a cyclic imide as intermediates (Fig. 1). The kinetics of the conversion are shown in Fig. 4. The reaction progressed steadily, approaching completion by 24 h, at which point the L-ASP peptide represented 80% of the total peptide and the D-iso-Asp peptide comprised 13%, the remainder being present as cyclic imide or methyl ester.
Conversion of L-Iso-As~ Speract to L -A s~ S p e r a c t -[~-I~~-Asp5]speract (15 p~) was incubated with methyltransferase and AdoMet in the same manner described for the LDH peptide, and reaction mixtures of varying duration were subjected to reversed-phase HPLC. Fig. 5A shows HPLC profiles obtained upon injection of purified L-iso-Asp speract (top panel) and upon injection of a 48-h methylation of the L-iso-Asp peptide (bottom panel). The major product of the reaction coeluted with the authentic asp^ peptide. Further support for the identity of this peak was obtained by thin layer electrophoresis at pH 3.5 and by determining the stereocon- Panels A and B show conversions of the ACTH and speract L-iso-Asp peptides, respectively. The peaks are labeled as follows: Zso, the L-iso-Asp starting material; Asp, the L -A s~ conversion product; and Imide, the cyclic imide intermediate. The criteria used to identify each peak are described in the text. HPLC profiles of 48-h reaction mixtures lacking methyltransferase or AdoMet were identical to that of the incubated L-iso-Asp peptide. If the L-iso-Asp peptide was omitted from the reaction mixture, no peaks were observed in the retention ranges shown here.
figuration of aspartate released upon acid hydrolysis. The peptide comigrated with authentic [~-Asp~]speract on electrophoresis and the D-/L-Asp ratio was 0.042. Thus, it appears that methylation promotes extensive conversion of L-iso-Asp t o L -A s~ in this peptide as well.
Given our results with the LDH peptide, we suspected that the iso-Asp peptide remaining after the 48-h methylation reaction (Fig. 5A) might contain significant amounts of D-iso-Asp. The peptide in this peak comigrated with L-iso-Asp speract on thin layer electrophoresis at pH 3.6, but it had a D-/L-Asp ratio of 1.0 after acid hydrolysis. [Iso-Asp5]speract contains a normal Asp in position 3 in addition to the iso-Asp residue (see Table I); therefore, this amount of D -A s~ in acid hydrolysates is consistent with complete inversion of stereoconfiguration of t h e ~-i s o -A s p~ residue. This conclusion was further supported by the finding that the 48-h iso-Asp speract peak was almost completely resistant to methylation by the methyltransferase.
At early time points in the conversion of L-iso-Asp speract, two additional peaks representing a significant proportion of the total peptide were observed (retention times given in Table 11). Using the same strategy outlined for the LDH peptide, it was determined that one of the peaks contained the L-iso-Asp a-carboxyl methyl ester and the other contained the corresponding cyclic imide. The minor peak in Fig. 5A is the cyclic imide-containing peptide which is still present after 48 h of reaction. Upon incubation at pH 7.4,37"C, the isolated speract cyclic imide became hydrolyzed with a tlh of 4.4 h into a 5.7:l ratio of iso-Asp speract and Asp speract.
Thus, the conversion of L-iso-Asp speract generated the same intermediates and products that were formed in the conversion of the L-iso-Asp LDH peptide. After 48 h, the speract conversion reaction resulted in 65% L-ASP peptide, 31% D-iso-Asp peptide, and 4% imide.

Conversion of the L-Iso-As~ ACTH Peptide to an L -A s~ Peptide-The [~-iso-Asp'~]ACTH(22-27) peptide (15 PM)
was incubated with methyltransferase and AdoMet as described for the LDH and speract peptides. Fig. 5B shows that a 48-h reaction produced an HPLC profile almost identical to that obtained with the iso-Asp speract peptide. The 22-min peak (70% of the total) was identified as the a asp*^ peptide: it coeluted with authentic L-ASP peptide and the Asp was found to be 90.6% f 1.1% in the L-enantiomer. The 20.5-min peak (26% of the total) was identified as a mixture of about 30% L-iso-Asp and 70% D-iso-Asp peptides. The small peak at 24.5 min (4% of the total) was identified as the imide based on its coelution with authentic cyclic imide and its rate of appearance and disappearance during early time points (not shown). Formation of a cyclic. imide intermediate ha5 been shown previously to occur subsequent to methylation of Liso-Asp-containing forms of ACTH(22-27) (Murray and Clarke, 1986) and ACTH(1-39) (Johnson and Aswad, 1985).

Comparison of the Rates of Conversion of the Three L-ISO-Asp Peptides to Normal, L -A s~ Peptides-
The rates of conversion of L-iso-Asp t o L -A s~ in each of the three peptides are compared in Fig. 6. The LDH peptide displayed the most rapid, as well as the most extensive, conversion. The ACTH L-iso-Asp peptide and L-iso-Asp speract had similar kinetics of conversion, the ACTH peptide conversion proceeding somewhat more rapidly than that of L-iso-Asp speract after 1 2 h of reaction.

DISCUSSION
Enzymatic carboxyl methylation of three unrelated L-isoas-party1 peptides has been shown in each case to promote HOURS FIG. 6. Comparison of the rates of L-ASP formation from three L-iso-Asp peptides. The percentage of peptide present in the L-ASP form was determined after varying periods of methylation of L-iso-Asp speract, L-iso-Asp ACTH peptide, or L-iso-Asp LDH peptide. Each point is the mean of duplicate determinations.
conversion of the atypical L-isoaspartyl residue to a normal L-aspartyl residue with a high degree of efficiency. Conversion occurs through the formation of isoaspartyl methyl ester and cyclic imide intermediates, supporting the mechanism proposed in Fig. 1 (Aswad, 1984a;Murray andClarke, 1984,1986;Johnson and Aswad, 1985). As a side reaction, slow racemization of the L-imide and subsequent hydrolysis of the Dimide causes the formation of significant amounts of D-isoaspartate-containing peptide.
Support for the completeness of the reaction scheme shown in Fig. 1 was obtained by modeling the conversion of the ACTH isopeptide (see "Appendix" for details). This peptide was chosen for modeling because the kinetic constants for each of the individual reactions in the pathway have been determined independently. Fig. 7 shows that the modeled time course of the conversion agrees well with the experimentally determined points. The conversion of L-iso-Asp to L-Asp fails to reach completion primarily because racemization of the imide intermediate occurs. Another factor is inactivation of the methyltransferase caused by accumulation of the product inhibitor AdoHcy and by depletion of AdoMet. Enzyme inactivation was a limitation of our in vitro analysis which presumably does not apply in vivo where AdoMet and AdoHcy levels should each remain near some regulated, steady state level. When enzyme inactivation was eliminated from the model, there was an additional 2% conversion to the L-Asp form by 48 h. Racemization may also occur at a much reduced rate in intact proteins compared with the short peptides used here, since secondary and tertiary structure of proteins will usually provide a large energy barrier to inversion of the aspartyl imide a-carbon. When both enzyme inactivation and racemization are eliminated in the model, the formation of the normal L-ASP peptide reaches 85% by 48 h and 94% by 72 h.
The three L-isoaspartyl peptides which were the subjects of this study differed significantly in amino acid sequence (Table  I), yet they were all excellent substrates for protein carboxyl methyltransferase, and they were all converted to L-aspartyl peptides by the same apparent mechanism. Methylation of HOURS FIG. 7. Modeled kinetics of the methylation-dependent conversion of the L-iso-Asp ACTH peptide. The kinetics of the conversion of this isopeptide were modeled according to the scheme in Fig. 1 using the independently determined rate constants given in Table I11 (see "Appendix"). The model also included an equation approximating the loss of methyltransferase activity which occurred during the incubations (see "Appendix"). The curves represent the results of the reaction modeling. The points are the means of duplicate experimental determinations showing the percentage of peptide present in each form as judged by reversed-phase HPLC analysis performed as in Fig. 58. Curves and points both indicate mixtures of Dand L-stereoisomers. At the 48-h time point of the modeled reaction, 93% of the Asp peptide was in the L-form, as was 84% of the cyclic imide. At the same time point, 86% of the iso-Asp peptide was present as the D-stereoisomer. speract-related peptides has not previously been reported. The fact that only the L-iso-Asp form of this peptide is methylated adds further evidence that the eucaryotic protein carboxyl methyltransferase modifies L-isoaspartyl residues and not normal aspartyl residues. More recently, McFadden and Clarke (1987) have shown that protein carboxyl methyltransferase from bovine erythrocytes promotes conversion of Lisoaspartyl tetragastrin (Trp-Met-iso-Asp-Phe-NHJ to the L-aspartyl form. The LDH isopeptide exhibited a significantly faster conversion to the L -A s~ form than did the other two isopeptides. We have used our kinetic model to assess the possible sources of this enhancement. Fig. 8 shows how independently varying several of the kinetic constants for the ACTH isopeptide conversion affects the amount of L-ASP peptide formed by 24 h. Two-fold changes in K,,, or kl (ester breakdown) have relatively little effect on conversion efficiency. A %fold change in the V , (not shown) also has negligible effect. Decreasing k4 (L-imide to D-imide) by a factor of 2 increases conversion slightly from 53% to about 57%. Variations in the rate of Limide hydrolysis have, by far, the most profound effect. Increasing the rate of hydrolysis by a factor of 2 increases L-Asp formation from 53 to 74%. This effect is expected, since L-imide hydrolysis is the rate-limiting step in the overall conversion cycle. The ratio of L-Asp/L-iso-Asp formed from L-imide hydrolysis (not shown in Fig. 8) also has a very large influence on conversion efficiency. When this ratio is changed from 1/i (the value for the ACTH imide) to 1/5.3 (the value for the LDH imide) conversion drops from 53 to 39%. It appears from this analysis that the enhanced conversion of iso-Asp to Asp observed with the LDH peptide is probably due to a greater instability of the imide form of this peptide. The overall imide hydrolysis must be sufficiently rapid to overcome the unfavorable Asp/iso-Asp ratio produced by imide breakdown. Because the LDH cyclic imide and iso-Asp methyl ester forms could not be separated by HPLC, we have Values for K,,, (0), the methyl ester hydrolysis rate constant (kl, e), the L-imide + D-imide rate constant (kr, m), and the L-imide hydrolysis rate constant (b + k3, A) were individually varied in the reaction model over a 4-fold range centered around the value of the corresponding rate constant which had been determined for the ACTH(22-27) isopeptide. Throughout these modeled reactions, the methyltransferase activity was held constant. The abscissa label k/k' indicates the ratio of the rate constant chosen for the calculation (k) to the corresponding rate constant observed for the ACTH peptide (k'). The latter values are taken from Table 111 (see "Appendix"). Symbols indicate the values of log(k/k') used in the model. For each modeled rate constant, the predicted amount of L-ASP peptide present at 24 h is shown on the ordinate as the percentage of L-iso-Asp starting material. The greater slope observed upon varying the L-imide hydrolysis rate constant indicates that it is a major factor in determining the rate of conversion of L-iso-Asp to L-ASP.

Conversion of L-Isoaspartate to L-Aspartate by Methylation 5627
not accurately determined the imide hydrolysis rate for this peptide; however, our estimates indicate that this imide has a half-life of <3.2 h (see Footnote 3), definitely less than the half-life of 4.2 h determined for the ACTH imide. In the LDH peptide, the iso-Asp residue is linked to a serine rather than a glycine. This may account, at least in part, for the more labile nature of the cyclic imide in the LDH peptide.
In proteins, secondary and tertiary structure might favor the formation of L -A s~ upon imide hydrolysis. As discussed above, the L-Asp/L-iso-Asp ratio has a profound effect on the efficiency of the conversion. Therefore, if the reaction occurs in vivo, the efficiency might be even greater than was observed here.
L-Isoaspartate is apparently formed to a significant extent in vivo. Peptides containing this amino acid have been detected in human urine (Kakimoto and Armstrong, 1961;Buchanan et al., 1962;Lou, 1975;Tanaka and Nakajima, 1978) and in the feces of humans (Welling, 1982) and mice (Welling and Groen, 1978) on antibiotic regimens. The isoaspartyl peptides in human urine are excreted by subjects fasting or adhering to protein-free diets, suggesting a metabolic, rather than dietary, origin . A number of proteins become deamidated i n vivo at asparagine residues in sequences which should be prone to L-isoaspartate formation (De Jong et al., 1975;Moo-Penn et al., 1976;Bornstein and Balian, 1977;Yuan et al., 1981). L-Isoaspartate may also form upon isomerization of L-aspartyl residues (Swallow and Abraham, 1958;Naughton et al., 1960) or possibly as a result of errors in protein translation (cf. Clarke, 1985).
Because the reactions reported in this study occur under physiological conditions, the methylation-dependent conversion of L-isoaspartate should also occur in vivo, unless the methyl ester or cyclic imide intermediates are modified in some other enzymatic reaction. Both of these intermediates were formed when an L-isoaspartyl peptide was incubated in erythrocyte cytosol (Murray and Clarke, 1986), but their ultimate fate in vivo has not yet been determined. Formation of L-isoaspartate in a protein would very likely disrupt its structure and function because it adds an extra carbon to the polypeptide backbone. The presence of this atypical amino acid might also affect the protein's turnover because most proteases are not capable of hydrolyzing the p-carboxyl peptide linkage (Pisano et al., 1966;Haley et al., 1966;Haley and Corcoran, 1967;Dorer et al., 1968). The conversion of Lisoaspartyl residues to L-aspartyl residues initiated by enzymatic carboxyl methylation may therefore be an in vivo reaction which would either prepare the isoaspartyl protein for proteolysis or restore it to a functional state. In support of this latter idea, we have recently found that enzymatic carboxyl methylation at pH 7.4 for 48 h can stimulate the activity of deamidation-inactivated calmodulin by about %fold .
It remains an important goal of future studies to determine the extent and mechanisms of isoaspartate formation in vivo and to ascertain the possible importance of the conversion reported here to the maintenance of normal cellular function.