Partial repair of deamidation-damaged calmodulin by protein carboxyl methyltransferase

Vol. 262, No. 25, Issue of September 5,pp. 12283-12287,1987 Printed in U.S.A.

THEJOURNALOF BIOLOGICAL CHEMISTRY

0 1987 by The American Society for Biochemistry and Molecular Biology, Inc

Partial Repair of Deamidation-damaged Calmodulinby Protein Carboxyl Methyltransferase” (Received for publication, January 30, 1987)

Brett A. Johnson, EstherL. Langmack, and DanaW. Aswad$ From the School of Biological Sciences, University of California, Irvine, Irvine, California 92717

Modification of calmodulin by proteincarboxyl methyltransferase requiresdeamidation of one or more labile asparagine residues (Johnson, B. A., Freitag, N. E., and Aswad, D. W. (1985) J. Biol.Chem. 260, 10913-10916). We now show that deamidation results in the generation of two altered formsof calmodulin, designated A and B, which can be separated by electrophoresis under nondenaturing conditions. The A form is characterizedby a larger apparentmolecular radius, has only 10%the activity of native calmodulin when assayed for its ability to activate a Ca2+/calmodulinas dependent protein kinase fromrat brain, and serves an excellent substrate for the methyltransferase. The B form more closely resembles native calmodulin: it has an apparentmolecular radius more like the native, exhibits about 40% the activity of native calmodulin, andisarelatively poor methyl acceptor. Evidence probably contain isoassuggests that theA and B forms partate (A) and aspartate (B) in place of Asn-60 and/ or Asn-97. Incubation of the A form with methyltransferaseand S-adenosyl-L-methionine converts about half of the A form toan electrophoretic band indistinguishable from the Bform. The activity of this partly converted calmodulin rises to 30-50% that of native calmodulin. These observations imply that the methyltransferase may have a biological role in restoring activity to proteinswhich contain abnormal isoaspartyl peptide bonds resulting from asparagine deamidation.

carboxyl “side chain.” Thus, iso-Asp sites might cause considerable disruption of normal secondary and tertiary protein structure. Prolonged incubation of synthetic L-iso-Asp-containing peptides with methyltransferase and S-adenosyl-L-methionine (AdoMet)’ leads to extensive conversion of the atypical @-linkedisopeptide to a more typical a-linked peptide. For example, incubation of [i~o-Asp*~~]lactate dehydrogenase(231-242) with purified methyltransferase and AdoMet for 24 h at pH 7.4,37 “C resulted in conversion of 80% of the isopeptide to an L-ASPpeptide (3). Similar results have been obtained with L-iso-Asp-containing versions of adrenocorticotropin-(22-27) (3), sperm-activating peptide (3), and tetragastrin (12). The ability of the methyltransferase to catalyze conversion of an iso-Asp-peptide bond to anAsp-peptide bond suggested that the enzyme might be capable of restoring biological activity to a protein damaged by deamidation. In a recent communication, we showed that deamidated calmodulin is an excellent substrate for the methyltransferase (13), suggesting that deamidation results in formation of iso-Asp-Gly sequences in place of the Asn-Gly sequences normally found at positions 60 and 97, each of which occurs in a critical calciumbinding domain (14).In the present paper, we report the effects of deamidation and subsequent methylation on the structure and activity of calmodulin. EXPERIMENTAL PROCEDURES

Purified Proteins-Native calmodulin and protein carboxyl methyltransferase were each purified from bovine brain as described previously (15). Concentrations of protein were determined by the Protein carboxyl methyltransferases from eucaryotic method of Lowry et al. (16) after precipitation by 0.07 g/ml trichlosources exhibit a pronounced specificity for peptides contain- roacetic acid. Nondenaturing Polyacrylamide Gel Electrophoresis (PAGE)-Elecing L-isoaspartate, i.e. aspartyl sequences in which the peptide trophoresis was performed on vertical slab gels using the discontinbond between aspartate and its C-terminalneighbor is made uous gel system of Laemmli and Favre (17), except that sodium through the side chain /?-carboxyl of the aspartate, rather dodecyl sulfate was omitted from the gel and tank buffer solutions, than through the a-carboxyl group (1-6). Deamidation of each of which contained 2mM EDTA. The stacking gel and separating labile asparagine sites, which occurs at neutral and alkaline gel were 0.05 g/ml and 0.15 g/ml acrylamide, respectively. Samples pH, probably constitutes a major source of isoaspartate, both were dissolved in 125 mM Tris-C1, 2 mM EDTA, 10% (v/v) glycerol, in vivo and in vitro (3, 7-9). Asn-Gly sequences appear to be 0.03 g/ml bromphenol blue (pH 7). They were loaded onto the gels prior heating. Electrophoresis was at 80 V until the dye front common sites for this typeof deamidation. L-Isoaspartate can without V until had enteredthe separating gel (about 2h) and then 80-120 at also arise from slowacid-catalyzed isomerization of L-aspartyl the dye front reached the end of the gel (8-10 h). The separated forms peptides (10, 11).Iso-Asp sequences contain an extra carbon of calmodulin were visualized by staining with 1 mg/ml Coomassie in the peptide backbone as well as an abnormal one-carbon Blue in destain solution. Destain solution was 25% (v/v) isopropyl alcohol, 10% (v/v) acetic acid, 65% (v/v) water. Electroelution of Calmodulin Bands-Native calmodulin was deam* This work was supported by United States Public Health Service Grants NS-17269 and AG-00538 (to D. W. A.) and predoctoral idated by a 10-h incubation a t 37 ‘C in 0.1 M NH,OH (pH 11-11.5), National Research Service Award MH-09000 (to B. A. J.). The costs 5% (w/v) glycerol (13). It was then lyophilized, resuspendedin sample of publication of this article were defrayed in part by the payment of buffer, and subjected to nondenaturing PAGE as described above page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18U.S.C. Section 1734 solely to indicate The abbreviations used are: AdoMet, S-adenosyl-L-methionine; this fact. EGTA, [ethylenebis(oxyethylenenitrilo)]tetraaceticacid; HEPES, 4$ To whom correspondence should be addressed School of Biolog- (2-hydroxyethyl)-l-piperazineethanesulfonic acid; HPLC, high-perical Sciences, 231 Steinhaus Hall, University of California, Irvine, formance liquid chromatography; PAGE, polyacrylamide gel electroIrvine CA 92717. phoresis; MES, 2-(N-morpholino)ethanesulfonicacid.

12283

12284

Repair of Deamidated Calmodulin by Enzymatic Methylation

AB-, N)LS Calmodulin

FIG. 1. Characterization of deamidated forms of calmodulin. Panel a, nondenaturing PAGE. Lane I, 5 pg of native calmodulin; &ne 2, 10 pg of calmodulin deamidated by a 3-h, 37 "C incubation in 0.1 M NH40H; lane 3 , 5 pg of band B calmodulin isolated by electroelution; lane 4 , 5 pg of band A calmodulin isolated by electroelution. Electrophoresis was for 8 h a t 120 V after the tracking dye entered the separating gel. The negative electrode was at the top of the gel. Panel b, stoichiometries of methylation. Each form of calmodulin was incubated with [methyl3H]AdoMet and varying concentrations of methyltransferase. In these experiments, the level of methyl incorporation should increase to a plateau representing the true stoichiometry of the reaction (1). Error bars denote the range of duplicate determinations. Panel c, stimulation of calcium/calmodulin-dependent protein kinase. The ability of each calmodulin form to stimulate the phosphorylation of a 54,000-daltonprotein of rat brain membranes was determined and expressed as a percentage of the phosphorylation obtained with 5 pg of native calmodulin. Error bars denote the range of duplicate determinations. 0,A, band A calmodulin; W, B, band B calmodulin; A,N , native calmodulin; PCMT, protein carboxyl methyltransferase. (500 pglgel). After electrophoresis, narrow strips from each side and from the center of each gel were stained for 20 min. The stained strips were used as guides to cut the calmodulin bands from the unstained portions of the gels. The resulting gel sections containing protein were diced,and theprotein was eluted using the electroelution apparatus described by Hunkapiller et al. (18). Elution was carried out at 4 "C for 9-10 h a t 50 V in the nondenaturing electrophoresis tank buffer. The collected protein was dialyzed a t 4 "C for 12-24 h against 2 liters of 2 mM K-HEPES (pH 7), 2% (v/v) glycerol prior to lyophilization. The purity of the eluted bandswas monitored routinely using nondenaturing PAGE. Gel Filtration HPLC-Proteins were separated on a 300 X 7.5-mm Bio-Si1TSK-125 column (Bio-Rad). The solvent was 50 mM Na2S04, 20 mM Na2HP04,1 mM EGTA (pH 7), and theflow rate was 0.5 ml/ min. Samples containing 5-10 pg of calmodulin were injected using a 20-pl sample loop, and protein was detected by absorbance a t 280 nm. For isolation of band A from pH 7.4-deamidated material, 100200pgof the incubated calmodulin were injected using a 100-p1 sample loop. The appropriate peak was collected, concentrated, and reinjected by the same procedure for final purification.

rat brain membrane protein, 10 p~ [y3'P]ATP (2,000-5,000 dpm/ pmol), and varying amounts of the calmodulin forms. Blanks lacked calmodulin. After a 2-min preincubation at 25 "C, reactions were initiated by adding the [y3'P]ATP, and the incubation was continued a t 25 "C for 30 s. Reactions were terminated and subjected to sodium dodecyl sulfate-PAGE. After staining, each gel lane was scanned with a densitometer. Gels were then dried and subjected to autoradiography. Each lane of the autoradiogram was also scanned. The level of phosphorylation of the 54,000-dalton protein was determined by dividing the height of the appropriate peak from the autoradiogram scan by the height of the corresponding peak from the scan of protein staining. RESULTS AND DISCUSSION

Isolation of Deamidated Forms of Calmodulin-To examine the possible effectsof deamidation and carboxyl methylation on the structure and activity of calmodulin, it was important to isolate the deamidated form(s) of the protein. Previous attempts at separation on the basis of charge were unsuccessS-A&nosyl-~-methionine-S-Adenosyl-[methyl-~H]-~-methionine ful, presumably because of the high backgroundof negatively ([methyPHIAdoMet, 15 Ci/mmol) was purchased from Amersham charged residues in calmodulin (13). Subsequently, we have Corp. and was assumed to be present entirelyin the biologically active found that deamidated formscan be separated on the basis of (S,S)diastereomer (19). Unlabeled AdoMet (Sigma) was purified by chromatography on Cm-cellulose (20) prior to use in diluting [methyl- molecular radius in the absence of denaturing agents. Fig. l a shows the results obtained when native (lane 1) and 3H]AdoMet. The purified unlabeled AdoMet contained 82% active ( S , S ) diastereomer as determined by the method of Hoffman (19). partially deamidated (lane 2) calmodulin (obtained by treatSpecific activities of the [ methyl-3H]AdoMetdilutions were corrected ment with 0.1 M NH,OH (pH 11-11.5))were subjected to for the diastereomeric purity of the unlabeled AdoMet. nondenaturing PAGE. The deamidated calmodulincontained Methylation Reactions for Stoichiometry Determinutwns-Incubaaltered forms which migrated more slowly than the native tions were carried out a t 30 "C in a pH 6 phosphate/citrate/EGTA protein. Band A was most abundant, containing 80% of the buffer as previously described (15). Reactions included 10 p~ cal- altered protein. However, this band appears to be heterogemodulin substrate, 1-8 p~ methyltransferase, 200 p~ [methyL3H] AdoMet (50-100 dpm/pmol), and 0.1 mg/ml bovine serum albumin. neous. The slight trailing of the band which isevident in Fig. Blanks lacked methyltransferase. After 1 h of incubation, reactions l a could be resolvedinto at least one additional distinct band were stopped by adding an equal volume of 0.45 M sodium borate (pH when electrophoresis was carried out for longer periods of 10.3), 40 mg/ml sodium dodecyl sulfate, 2% (v/v) methanol. Released time, as in Fig. 3.' [3H]methanol from the hydrolyzed calmodulin methyl esters was assayed by the filter paper diffusion method of MacFarlane (21) as modified by Johnson and Aswad (22). Assays of Calmodulin Actiuity-The ability of calmodulin forms to stimulate the phosphorylation of a 54,000-dalton protein in rat brain membranes was determined by a method similar to thatof Schulman and Greengard (23). Reaction mixtures (100 pltotal) included 50 mM K-MES (pH 7), 5 mM MgC12,0.2 mM EGTA, 0.5 mM CaC12,1mg/ml

The generation of more than two forms of calmodulin upon deamidation is expected. If partial deamidation occurs a t both of the two most labile asparagines, Asn-60 and Asn-97, to produce Asp and iso-Asp, then a total of eight deamidated forms are possible. Someof these forms may be minor products and some may not differ eneugh in molecular radius to separate from one another on nondenaturing PAGE. However, the presence of two subforms in band A (Fig. 3),

Repair of Deamidated Calmodulin by Enzymatic Methylation

12285

native calmodulin was not sufficient to allow resolution by gel filtration. Methylation of Deamidated Calmodulin Forms by Protein Carboxyl Methyltransferme-A common product of asparagine deamidation in peptides is isoaspartate, aspartatelinked to the adjacent amino acid through its @-carboxylgroup (3, 7-9). A peptide chain containing iso-Asp would have altered bond angles and anadditional carbon in its backbone. These changes would almost certainly disrupt aprotein’s secondary and tertiary structure.It therefore seemed likely that iso-Asp was responsible for some of the changes in molecular radius observed in deamidated calmodulin. The specificity of protein carboxyl methyltransferase for Liso-Asp (1-6) provided a simple means of determining which of the isolated bands of deamidated calmodulin was enriched in isoaspartyl residues. Band A, band B, and native calmod, incubated with varying concentraulin, each at 10 p ~ were tions of the methyltransferase andexcess [ meth~l-~HIAdoMet in order to determine stoichiometries of methylation. As shown in Fig. 16, band A incorporated nearly stoichiometric levels of methyl groups, whereas band B and native calmodulin were poor methyl acceptors.’ This suggests that band A contains high levels of L-iso-Asp. Therefore, the presence of Band B this atypical residue in band A maybe responsible for its greater molecular size. Activity of Deamidated Calmodulin Forms in Stimulating Brain Membrane Protein Phosphorylation-Because bands A 4 8 12 16 and B had structures which differed from native calmodulin, Retention Tlme ( m m ) FIG. 2. Gel filtration HPLC of deamidated forms of calmod- it seemed likely that they would bealtered inactivity, as well. d i n . Native calmodulin, calmodulin which had firstbeen deamidated In order to test thispossibility, their ability to stimulate the by incubation in 0.1 M NH,OH for 3 h at 37 “C (incubated), and calcium-dependent phosphorylation of a 54,000-dalton proelectroeluted bands A and B were each subjected to gel filtration tein in rat brain membranes was evaluated. HPLC as described under “Experimental Procedures.” Arrowheads As previously described by Schulman and Greengard (23), denote the elutiontimesobtained for proteinstandards: 670 K, thyroglobulin; 158 K, y-globulin; 44 K , ovalbumin; 31 K, carbonic native calmodulin efficiently stimulated the phosphorylation of this protein (Fig. IC).Under the same conditions, band A, anhydrase; 17 K, myoglobin. which contains the highest level of L-isoaspartate, was markBands A and B were isolated from the gels using a modifi- edly deficient in activity. Band B was intermediate in its cation of the method of Hunkapiller et al. (18).The results of stimulation of phosphorylation. Conversion of Band A into Band B by Prolonged Enzymatic re-electrophoresis of the eluted bands using nondenaturing Carboxyl Methylation-Recent studies have shown that proPAGE are shown in lanes 3 and 4 of Fig.la. In this preparation, the protein doublet making up band A was greater than longed incubation of L-isoaspartyl peptides with protein car97% pure as judged by densitometric scanning. The purity of boxyl methyltransferase and excess AdoMet at pH 7.437“C, band B was 91%, the remainder of the protein being present results inthe conversion of most of the iso-Asp residues to Lin two bands which migrated more slowly than band B on Asp (3, 12). In order to determine whether this conversion nondenaturing PAGE. Thesebandsappear to have been would occur in an intact protein and whether it would result ) in a detectable change in structure, band A (15 p ~ was formed during electroelution.’ 7.4, 37 “ C , with 2 p~ methyltransincubated for 48 h at pH Heterogeneity of deamidated calmodulin could also be deferase and 200 p~ AdoMet. The methylation mixture was tected using gel filtration HPLC performed in the absence of then subjected to nondenaturing PAGE (Fig. 3). This incudenaturing agents. Upon gel filtration, partially deamidated bation converted about half of band A into a form co-migratcalmodulin was resolved into two peaks, peak I eluting earlier ing with band B (lane3, Fig. 3). The conversion did not occur than native calmodulin and peak I1 coeluting with native when the methyltransferase was omitted from the incubation calmodulin (Fig. 2). When subjected to nondenaturing PAGE, (lane 4, Fig. 3). Incubations lacking AdoMet also showed no peak I co-migrated with band A (not shown), and electroeluted evidence of conversion, and, when band A was omitted, no band A coeluted with peak I on gel filtration HPLC (Fig. 2). band was visible in the gel region occupied by the calmodulin This provides further evidence that therelatively slow migra- forms (not shown). tion of band A on nondenaturing PAGE was due primarily to Prolonged enzymatic carboxyl methylation can therefore an increased molecular radius. Nondenaturing PAGE indi- produce a profound structural change in an L-iso-Asp-concated that peak I1 consisted of a mixtureof band B and native taining form of deamidated calmodulin, and this conversion calmodulin (not shown). Electroeluted band B coeluted with returns band A to a form moreclosely resembling native native calmodulin on gel filtration HPLC (Fig. 2). Thus, the calmodulin. Because a form of calmodulin co-migrating with apparent difference in molecular radius between band B and band B is the product of the conversion, band B probably contains an aspartyl form of deamidated calmodulin. the generation of two slower migrating bands upon electroelution of Increase in the Activity of Band A following Conversionband B (Fig. la), and the significant, albeit low,level of methyl incorporation into electroeluted band B (Fig. lb) all suggest their Because band Bwas moreeffective than band A in stimulating existence. protein phosphorylation, conversion of band A t o band B by Y

Y

Y Y Y

I

&1

Repair of Deamidated Calmodulin

12286

12" 3 A-

by Enzymatic Methylation Days at pH 24,37"C

4

c-

0

3

7

14

21

35

" " "

-A

BN-

-0

FIG. 3. Protein carboxyl methyltransferase converts band A into band B. Nondenaturing PAGE was performed as in Fig. l a , except that electrophoresis was for 10 h a t 80 V after the tracking dye entered the separating gel. Lane 1 , 5 pg of band B; lane 2 , 5 pg of band A; lane 3, 5 pg of band A which had first been incubated at a concentration of 15 p M for 48 h a t 37 "C with 2 p M methyltransferase and 200 p~ AdoMet in 50 mM K-HEPES (pH 7.4), 1 mM EGTA; lane 4, 5 pgof band A incubated as for lane 3, except that the methyltransferase was omitted. Arrows denote positions of band A (A), band B ( B ) ,and native ( N ) calmodulin. The negative electrode was at the topof the gel.

-N

0

IO

20

30

40

Days at pH 74,37OC FIG. 5. Deamidation of calmodulinunder physiological conditions. Top panel, calmodulin incubated for varying times at 37 "C FIG. 4. Activation of pH 11-derived band A by protein carboxyl methyltransferase. Unincubated band A (0) or band A which had first been incubated a t 37 "C for 48 h with 2 p M methyltransferase and 200 p~ AdoMet in 50 mM K-HEPES (pH 7.4), 0.1 mM EGTA (0)were assayed for activity as in Fig. IC.Their stimulation of phosphorylation was expressed as a percentage of that obtained using 5 pg of native calmodulin. Error bars denote the range of duplicate determinations.

protein carboxyl methyltransferase wouldbe expected to cause an increase in activity. This was tested by incubating band A with methyltransferase and AdoMet for 48 h under the sameconditions which had resulted initsstructural conversion (Fig. 3). The incubation mixture was then used as a source of calmodulin in an activity assay performed as in Fig. IC. Fig. 4 shows the activity of band A before and after incubation with the methyltransferase and AdoMet. The carboxyl methylation-dependent conversion of band A resulted in a 200% increase in its ability to stimulate calcium-dependent phosphorylation of the 54,000-dalton protein of rat brain membranes. Control experiments verified that the methyltransferase and AdoMet had no effect on basal or native calmodulin-stimulated phosphorylation. Also, control incubations of band Ain theabsence of methyltransferase had no effect on its activity. Similarresults were obtained using different preparations of methyltransferase and pH ll-derived band A (see below), These results suggest that conversion of L-iso-Asp to L - A s by ~ the methyltransferase can result in a significant restoration of calmodulin's activity, presumably by restoring a more normal structure to the peptide backbone. Deamidation of Calmodulin at p H 7.4 and 37 "C-A small population of calmodulin molecules appears to be methylated on carboxyl groups in intact cells (24-27). Thus, isoaspartate

in 50 mM K-HEPES (pH 7.4), 1 mM EGTA, 10% (v/v) glycerol, 2 mg/ml sodium azide. A total of 5 pg from each incubation was then subjected to nondenaturing PAGE, which was performed as in Fig. l a . Arrows denote positions of band A ( A ) , band B ( B ) ,and native ( N ) calmodulin. The positions of bands A and B were determined on the same gel using a deamidation "standard" of calmodulin which had been incubated at 37 "C for 3 h in 0.1 M NH,OH. The negative electrode was at the top of the gel. Bottom panel, stoichiometries of methylation. Calmodulin incubated as in the top panel was subjected to methylation at a concentration of 10 p M using 1 pM methyltransferase. Error bars denote the range of duplicate determinations.

formation and partial structural and functionalrepair by protein carboxyl methyltransferase may bepart of the normal metabolism of calmodulin. To determine whethercalmodulin deamidation and iso-Asp formation might occur at a significant rate in uiuo, native calmodulin was incubated at pH7.4 and 37 "C for extended periods of time. The incubated calmodulin was then either subjected to nondenaturing PAGE or incubated with the methyltransferase and [methyL3H] AdoMet to measure its methyl-accepting capacity. As shown in the top panel of Fig. 5, a form of calmodulin co-migrating with band A on nondenaturing PAGE was detectable by 3 days of incubation and increased in amountwith continued incubation. This suggests that aniso-Asp-containing form of calmodulin might be generated under physiological conditions. A form co-migrating with band B was also generated by the incubations (Fig. 5, top panel), although it was present in a lower relative proportion than when brief incubations at higher pH were used to cause deamidation (see Fig. la). Other altered forms of calmodulin migrating between band A and native calmodulin were apparent afterincubations at pH7.4 and 37 "C. These bands may represent other types of nonenzymatic damage, for example, oxidation of methionine, lysine, or tyrosine residues (28, 29). The methyl-accepting capacity of calmodulin increased with incubation at pH 7.4 and 37 "C coincident with the

Repair of Deamidated Calmodulin by

Enzymatic Methylation

12287

tion to thesame extent as unincubated band A. These experiments suggest that incubation at pH 7.4 may produce deamidation and isomerization of the same sites in calmodulin as does deamidation at pH 11. More importantly, theyshow that calmodulin deamidated under physiological conditions undergoes a significant restoration of activity which probably reflects the methylation-dependent conversion of isoaspartate to aspartate. CONCLUSIONS

"

+

- f PCMT FIG. 6. Activity of calmodulin forms before and after cycles of enzymatic carboxyl methylation. Panel A, native calmodulin or band A calmodulin isolated from either a pH 11 or a pH 7.4 deamidation were tested at 3 pg/assay for their ability to stimulate the calcium-dependent phosphorylation of the 54,000-dalton protein of rat brain membranes (see "Experimental Procedures"). Panel B, native calmodulin, pH 11-derived band A, and pH 7.4-derived band A were incubated for 48 h a t 37 "C,pH 7.4, with 200 p~ AdoMet in the presence or absence of 4 p~ protein carboxyl methyltransferase (PCMT). Samples of these incubations containing3 pg of calmodulin were then used as sources of calmodulin for the activity assay. All phosphorylation reactions were carried out in triplicate,each replicate of which was analyzed on a separate gel along with replicates of the other phosphorylation reactions. For each reaction, the calmodulin activity was expressed as a percentage of the phosphorylation obtained with 3 pg of unincubated native calmodulin in the reaction which was analyzed on the same gel. The values arethe mean percentages, and error bars denote the standard deviations. Because the activity of unincubated calmodulin shown in panel A was defined as 100% on each gel, no error bar is included. -I-

generation of band A (Fig. 5, bottom panel), confirming the formation of L-iso-Asp. Fifty percent of the calmodulin molecules accepted methyl groups after 2 weeks of incubation, and the level of methylation exceeded 1 mol CH3/mol of calmodulin after 5 weeks. Thus, calmodulin deamidation and L-isoaspartate formation occur i n vitro under physiological conditions. To our knowledge, the turnover time of calmodulin in vivo has not been determined. It seems likely, however, that a significant amount of iso-Asp would be formed in a fraction of the calmodulin molecules before they are removed by intracellular proteolysis. If this is the case, restoration of function by enzymatic carboxyl methylation may also occur

i n vivo. In order to determine the effects of methylation on isoAsp-containing calmodulin formed under physiological conditions, the band A material present after a28-day, 37 "C,pH 7.4 incubation was purified by gel filtration HPLC. It was then incubated at pH 7.4, 37 "C, for 48 h with methyltransferase and AdoMet and subjected to nondenaturing PAGE as in Fig. 3. The major product of the methylation migrated at the position of band B (not shown). Incubations performed in the absence of methyltransferase showed no conversion. Fig. 6A shows the activity of the pH 7.4-derived band A compared with native calmodulin and with pH 11-derived band A. Tested at a level of 3 pg/assay, the pH 7.4-derived calmodulin exhibited 18 k 6% of the activity of native calmodulin. In this experiment, 3 pg of the pH 11-derived band A material exhibited an activity of 9 f 1%.Fig. 6B shows the effect of methylation on the activity of these forms. Under methylating conditions (+PCMT),the activity of the pH 7.4derived band A rose to 68 & 12% of native. Activity of the pH 11-derived band A rose to 50 f 2% of native. Control incubations lacking methyltransferasestimulated phosphoryla-

Deamidation may be a major cause of spontaneous protein damage i n vivo (30-32). Deamidation is most likely to occur at labile Asn sites (31) which, bywayof cyclic imide intermediates, lead to the formation of atypical isoaspartyl peptides as the major final products (3, 4, 7-9). Given the specificity of protein carboxyl methyltransferase for L-iso-Asp, it seems likely that the methylation reaction plays some role in the metabolism of proteins damaged by deamidation. As shown here with calmodulin, methylation may serve to restore activity to thedamaged protein and thereby save the considerable metabolic energy which would be required to degrade and replace the damaged protein. The restoration we have observed is admittedly incomplete, and itprobably reflects a conversion of iso-Asp to Asp rather than Asn. Becauseour reactions were carried out in vitro with purified components, it remains possible that a more complete repair reaction might occur i n vivo or that methylation might prepare the damaged protein for subsequent turnover (3, 22). Studies on the fate of deamidated calmodulin in cell extracts or intact cells may provide further insight into the biological function of protein carboxyl methyltransferase. REFERENCES 1. Aswad, D. W. (1984) J. Biol. Chem. 259,10714-10721 2. Aswad, D. W., Johnson, B. A,, and Glass, D. B. (1987) Biochemistry 26, 675-681 3. Johnson, B. A., Murray, E. D., Jr., Clarke, S., Glass, D. B., and Aswad, D. W. (1987) J. Biol. Chern. 262,5622-5629 4. Murray, E. D., Jr., and Clarke, S. (1984) J. Biol. Chem. 259,10722-10732 5. O'Connor, C. M.,and Clarke, S. (1985) Biochem. Biophys. Res. Commun. 132,1144-1150 6. McFadden, P. N., and Clarke, S. (1986) J. Biol. Chem. 261,11503-11511 7. Bornstein, P., and Balian, G. (1977) Methods Enzymol. 4 7 , 132-145 8. Graf, L., Bajusz, S., Patthy, A., Barat, E., and Cseh, G. (1971) Acta Biochirn. Biophys. Acad. Sci. Hung. 6,415-418 9. Geiger, T.,and Clarke, S. (1987) J. Biol. Chern. 262,785-794 10. Swallow, D. L., and Abraham, E. P. (1958) Bioehem. J. 70,364-373 11. Naughton, M. A., Sanger, F., Hartley, B. S., and Shaw, D.C. (1960) Biochem. J. 77,149-163 12. McFadden, P. N., and Clarke, S. (1987) Proc. Natl. Acad. Sci. U. S. A. 84,

9m~-9waa """

l"""

13. Johnson, B. A,, Freitag, N. E., and Aswad,D.W. (1985) J. Biol. Chem. 260,10913-10916 14. Watterson D. M., Bur ess, W. H., Lukas, T. J., Iverson, D., Marshak, D.

R., Schfeicher, M., finkson, B.W., Fok, K.-F., and Van Eldik, L. J.

(1984) Adv. Cyclic Nucleotide Protein Phosphorylation Res. 16,205-226 15. Aswad, D. W., and Deight, E. A. (1983) J. Neurochem. 40,1718-1725 16. Lowry, 0.H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Biol. Chern. 193,265-275 17. Laemmli, U. K., and Favre, M. (1973) J. Mol. Biol. 80,575-599 18. Hunkapiller, M. W., Lujan, E., Ostrander, F., and Hood, L. E. (1983) Methods Enzymol. 91,227-236 19. Hoffman, J. L. (1986) Biochemistry 26,4444-4449 20. Chi icb T. P. (1968) Ph.D. dissertation, University of California, Berkeley 21. Mazardne, D. E. (1984) J. Bid. Chem. 259,1357-1362 22. Johnson, B. A., and Aswad, D. W. (1985) Biochemistry 24,2581-2586 23. Schulman, H., and Greengard, P. (1978) Proc. Natl. Acad. Sci. U. S. A. 75, 5432-5436 24. Gagnon, C., Kelley, S., Manganiello, V., Vaughn, M., Odya, C., Strittmatter, W., Hoffman, A,, and Hirata, F. (1981) Nature 291, 515-516 25. Runte, L., Jurgensmeier, H.-L., Unger, C., and Soling, H. D. (1982) FEES Lett. 147,125-130 26. Brunauer, L. S., and Clarke, S. (1986) Biochem. J. 236,811-820 27. Vincent P. L. and Siegel, F. L. (1986) Fed. Proc. 4 5 1694 28. Hardin;, J. J. '(1985) Adv. Protein Chem. 37,247-33i 29. Malencik, D. A., and Anderson, S. R. (1987) Biochemistry 26, 695-704 30. Ahern, T.J., and Klibanov, A. M. (1985) Science 228,1280-1284 31. Robinson, A. B., and Rudd, C. J. (1974) Curr. Top. Cell. Regul. 8, 247-294 32. Tollefsbol, T.O., and Gracy, R. W. (1983) Bioscience 33, 634-639