Peptides with sequences similar to glycine, arginine-rich motifs in proteins interacting with RNA are efficiently recognized by methyltransferase(s) modifying arginine in numerous proteins.

Several proteins that interact with RNA, e.g. the heterogenous ribonucleoprotein particle A and B proteins, fibrillarin and nucleolin, contain the modified amino acid NG,NG-dimethylarginine. Here, we report that two synthetic peptides, Ac-GGRGGFGGRGGFGGRGGFG-NH2 (R3) and GGFGGRGGFG-NH2 (R1), which are based on methylated sequences in fibrillarin and nucleolin, inhibit the methylation of a large majority of the methyl-accepting proteins observed in extracts of adenosine dialdehyde-treated PC12 cells. Concomitantly, the peptides themselves become methylated, suggesting that they compete for the same enzyme that carries out the bulk of N-methylation in PC12 cells. R3 potently inhibits formation of NG,NG-dimethylarginine in PC12 substrates, with a lesser effect on NG-monomethylarginine and NG,N'G-dimethylarginine. Bovine brain contains an activity that methylates PC12 methyl acceptors. After partial purification, the bovine methyltransferase efficiently modifies R3 and R1, yielding half-maximal rates of methylation at approximately 0.2 and approximately 2 microM peptide, respectively. A search of the GenPept database for the FGGRGGF motif revealed 13 candidate methyl acceptors containing arginine and at most two similar substitutions or one mismatch. Of these, 10 are known or presumed to interact with RNA. These findings are consistent with the hypothesis that a majority of proteins containing NG,NG-dimethylarginine interact with RNA.

The hnRNP proteins are essential in pre-mRNA processing and are believed to serve as an "operating table" for the removal of introns from pre-mRNA (6-9). The hnRNP A1 protein can regulate alternative splicing of pre-mRNA in in uitro systems, where it has been shown that an excess of the hnRNP A1 protein over the essential splicing factor SF2 results in preferential utilization of distal 5' splice sites over the proximal 5' splice sites (10). Fibrillarin is instrumental in pre-rRNA maturation and ribosome biogenesis (11)(12)(13)(14). It is part of the nucleolar U3 small nuclear ribonucleoprotein particle, binds to the Box C sequence of U3 snRNA (15) and functions in the first step of pre-rRNA processing (11). NOP1, a yeast nucleolar protein believed to be the homologue of mammalian fibrillarin, is required for rRNA processing and is essential for cell growth (16). Nucleolin, a major nucleolar protein (13,(17)(18)(19)(20)(21), has been found associated with chromatin and with preribosomal particles (22) and is believed to play a direct role in pre-rRNA transcription and ribosome assembly The W,p-dimethylarginines in fibrillarin and nucleolin occur in clusters embedded in glycine-rich domains (see Table  I). The C-flanking neighbor of dimethylarginines in fibrillarin and nucleolin is always glycine, while the N-flanking neighbor is most often, but not exclusively, glycine. The number of spacer amino acids between two dimethylarginines can vary from a single glycine to several glycines, sometimes interspersed with phenylalanine and/or some other amino acid (29).
When rat PC12 cells are cultured in the presence of micromolar levels of adenosine dialdehyde (AdOx), a compound that raises intracellular levels of S-adenosyl-L-homocysteine (AdoHcy), methylation reactions are inhibited, and N-methylated proteins are synthesized in a hypomethylated state (30). These proteins can be methylated with tritium-labeled methyl groups by incubating extracts of AdOx-treated cells in the presence of [rnethyl-3H]S-adenosyl-~-methionine ([rneth~l-~H] AdoMet). Using this approach, we previously demonstrated the presence of over 300 methyl-accepting proteins, or protein variants, in extracts of . In these studies, 72% of the methyl incorporation was found to occur in the form of p,p-dimethylarginine and 18% was recovered as p-monomethylarginine, a probable precursor of p,P-dimethylarginine (30).
It is not known if a single enzyme is responsible for formation of p,p-dimethylarginine in all proteins where it is present, or whether several distinct enzymes catalyze this (23-28). reaction in different classes of methyl acceptors. In this paper we used two synthetic peptides, GGFGGRGGFG-NH, (R1) and Ac-GGRGGFGGRGGFGGRGGFG-NH2 (R3), designed on the basis of sequences encompassing the known methylation sites in fibrillarin and nucleolin (see Table I) to address two major questions. 1) Is the short sequence GGFGGRGGFG sufficient for recognition by the arginine methyltransferase(s)? 2) If so, how many of the methylated proteins in PC12 cells are methylated by enzyme(s) recognizing the GGFGGRGGFG sequence?
Other Materials-All components of the cell culture medium were purchased from GIBCO/BRL. AdOx (periodate-oxidized adenosine), pOly-D, L-lysine, and unlabeled AdoMet were purchased from Sigma. Unlabeled AdoMet was further purified by chromatography on CMcellulose. [methyl-3H]AdoMet was from Du Pont-New England Nuclear.
Cell Culture and Preparation of Extracts-Rat PC12 cells (American Type Culture Collection) were cultured for 3 days in the presence of 10 p~ AdOx, and extracts were prepared as described earlier (30), except that the homogenization buffer was supplemented with 200 p~ PMSF and 0.5 pg/ml leupeptin. Protein concentrations were determined by the method of Lowry et al. (32) after precipitating the proteins in 7% (w/v) trichloroacetic acid. Bovine serum albumin was used as a standard.
Methylation Reactions-The reactions designed to measure the total methyl incorporation into PC12 cell substrates in the presence of the R3, R1, K1, and ACTH 1-24 peptides were carried out for 40 min at 30 "C in 30 pl containing varying concentrations of peptide, 100 p~ [rnethyl-3H]AdoMet (5,000 dpm/pmol), extract from AdOxtreated PC12 cells (60 pg of protein), 50 mM Na-MOPS buffer, pH 7.2, 2 mM EDTA, 100 p~ PMSF, 0.5 pg/ml leupeptin, and 0.3 M NaC1. After 40 min, the reactions were split into two 15-p1 fractions. To the first fraction, 7.5 pl of SDS-PAGE sample buffer was added, and the sample was frozen at -70 "C until further analysis by SDS-PAGE and fluorography. To the second fraction, 100 pl of 10% (w/ v), ice-cold trichloroacetic acid and 250 pg of y-globulin as carrier protein were added. After 10 min, the samples were centrifuged for 1 min in a microfuge, and the supernatants containing the peptides were saved for further analysis by HPLC as described below. The protein pellets were analyzed for methyl incorporation as described (31). For two-dimensional nonequilibrium pH gradient electrophoresis (NEPHGE)/SDS-PAGE, methylation reactions were carried out similarly, except that the volume was 90 pl containing 180 pg of protein from AdOx-treated PC12 cells and that [meth~l-~HIAdoMet of 10,000 dpm/pmol was used. At the end of the 40-min reaction, the samples were frozen at -70 "C until further analysis by NEPHGEI Methylation reactions destined for acid hydrolysis and two-dimensional thin-layer chromatography were carried out essentially as described above for the total methyl incorporation into PC12 proteins in the presence of varying concentrations of the peptides, except that [meth~l-~HIAdoMet of 10,000 dpm/pmol was used. The reactions were stopped by adding 100 p1 of 10% trichloroacetic acid, the supernatant was aspirated, the pellets were redissolved in 100 p1 of formic acid and reprecipitated twice in 1 ml of 10% trichloroacetic acid. The final pellet was redissolved in 100 pl of concentrated formic acid, dried under vacuum, and subjected to acid hydrolysis in 200 pl of 6 N HC1 for 24 h at 110 "C. The hydrolysates were dried under vacuum, redissolved in 40 pl of 1.6 M acetic acid, dried again, and SDS-PAGE. redissolved in 10 pl of deionized water.
For measuring the enzyme activity in fractions from hydrophobic interaction chromatography, methylation reactions were carried out for 50 min at 30 "C in 25 p1 in the presence of 50 p~ [meth~l-~H] AdoMet (5000 dpm/pmol), 50 mM Na-MOPS buffer, pH 7.2, and 2 m M EDTA. The reactions contained 15 pl of fraction and 5 pl of substrates from AdOx-treated PC12 cells (2.1 pg of protein) depleted of endogenous enzyme. Proteins were precipitated in 10% (w/v) trichloroacetic acid, and methyl incorporation was measured by liquid scintillation counting as described (31).
The reactions aimed at measuring the rate of methylation of the R3, R1, K1, and ACTH 1-24 peptides were carried out for 15 min at 30 "c in 25 p1 containing varying concentrations of peptide, 100 p~ [methyl-3H]AdoMet (2500 dpm/pmol), partially purified bovine brain enzyme (10.2 pg of protein) obtained as described below, 50 mM Na-MOPS buffer, pH 7.2, 2 mM EDTA, 100 p~ PMSF, and 0.5 pg/ml leupeptin. The reactions were stopped by adding 11 pl of 2% (w/v) trifluoroacetic acid. The samples were analyzed by reversed-phase HPLC and liquid scintillation counting as described below.
Assessment of Methyl Incorporation into the Peptides by Reuersedphase HPLC-Methylation reactions containing the peptides were either injected directly (reactions with bovine brain enzyme) or the proteins were removed first by precipitation in trichloroacetic acid, and the supernatants containing the peptides were injected (reactions with PC12 AdOx extracts). The injections were onto a C8 reversedphase guard column (RP-300, 30 X 4.6 mm, 7 pm; Applied Biosystems). Elution was by an acetonitrile gradient in 0.1% trifluoroacetic acid at a flow rate of 2 ml/min (for peptides R3 and ACTH 1-24: 10% acetonitrile for 2 min, 10-30% over 1 min, 30-40% over 3 min, 40-100% over 1 min, 100% for 1 min; for peptides R1 and K1: 5% for 2 min, 5% to 20% over 1 min, 20-35% over 3 min, 35-100% over 1 min, 100% for 1 min. Under these conditions, the peptides separated from unreacted [meth~l-~HIAdoMet, which eluted early in the gradient. The absorbance of the peptides was monitored at 214 nm. The methylated peptide(s) in a single run were collected in one fraction (3-5.5 min), and the radioactivity was counted by liquid scintillation counting.
Two-dimensional Thin Layer Chromatography of Methylated Amino Acids-Approximately 5.5 pg of acid hydrolysate and 10 nmol of each amino acid standard (p,p-dimethylArg, p , N " -dimethylArg, P-monomethylArg, mono-, di-, and trimethyl-Lys and 3-methyl-His) was applied onto a thin layer chromatography sheet and separated as described (33,30). The spots of methylated amino acid standards were visualized by ninhydrin staining. The material within the spots was collected and analyzed by liquid scintillation counting.
One-and Two-dimensional Polyacrylamide Gel Electrophoretic Separation of Methylated Proteins-SDS-PAGE was carried out as described by Laemmli (34), except that the concentrations of the components of the running buffer were halved. NEPHGE was performed in gels containing 4% acrylamide, 9.2 M urea, 2% Nonidet P-40,2% pH 3.5-10 Ampholine and was carried out for 2.5 h at 500 V constant voltage (35). Separation in the second dimension was by SDS-PAGE. The gels were stained with Coomassie R-250, impregnated with sodium salicylate (36), dried, and exposed to preflashed x-ray film at -70 "C (37).
Hypomethylated Substrates from AdOx-treated PC12 Cells-To obtain substrates for detection of enzyme activity during the partial purification of the bovine brain methyltransferase, a soluble extract of AdOx-treated PC12 cells was loaded onto an S-Sepharose cationexchange column in 50 mM sodium phosphate, pH 7.2, 1 mM EDTA, 10 mM 8-mercaptoethanol, 100 pM PMSF. The flow-through, which contained the enzyme and acidic substrates, was discarded. Basic methyl-accepting proteins were then eluted using 1 M NaCl in loading buffer. These substrates were pooled as a single fraction, which was then dialyzed. The basic substrates were then incubated with [methyl-3H]AdoMet in the presence or absence of an extract from untreated PC12 cells as a source of exogenous enzyme, followed by SDS-PAGE and fluorography. Little or no methyl incorporation occurred when the PC12 cell extract was incubated in the absence of the substrates or when the substrates were incubated in the absence of PC12 cell extract (data not shown). However, a number of heavily methylated bands were detected after reactions containing both the extract and the basic substrates. This indicates that the substrates were effectively depleted of enzyme by chromatography on S-Sepharose.
Partial Purification of Bovine Brain Arginine Methyltransferme-In order to separate the bovine brain arginine methyltransferase from basic substrates, cytosol from cerebral cortex was adsorbed onto DE23 anion-exchange resin in 50 mM Tris-HC1, pH 8.0, followed by repeated washes in 50 mM Tris-HC1, pH 8.0, 0.1 M NaCl. The enzyme was then eluted with 50 mM Tris-HC1, pH 8.0, 1 M NaC1. Proteins were concentrated by ammonium sulfate precipitation, dialyzed, and subjected to chromatography on Q-Sepharose using a gradient of 0 to 1 M NaCl in 20 mM Tris-HC1, pH 8.0. Fractions were assayed for the arginine methyltransferase using the basic substrates from AdOxtreated PC12 cells. A broad peak of enzyme activity was revealed in these assays, indicating that bovine brain possesses an activity that is capable of methylating the basic methyl acceptors of PC12 cells (not shown). However, chromatography on Q-Sepharose yielded little purification of the enzyme. We therefore subjected the pooled activity to further purification using hydrophobic interaction chromatography on phenyl-Sepharose. As shown in Fig. 5, a peak of activity methylating the basic substrates from PC12 cells was detected. Fractions containing this activity were pooled as indicated, resulting in a total purification of approximately 7-fold.

Competitive Methylation of FibrillarinlNucleolin-related
Peptides and Endogenous Substrates in Extracts of AdOxtreated PC12 Cells-To determine if the sequences encompassing the methylated arginines in fibrillarin and nucleolin are sufficient for recognition by arginine methyltransferase(s), two synthetic peptides, R3 and R1 (Table I), were incubated a t varying concentrations in methylation reactions containing [meth~l-~HIAdoMet and extracts of AdOx-treated PC12 cells. Methyl incorporation into the peptides and into endogenous PC12 cell substrates was measured using two different assays of each incubation mixture (see "Experimental Procedures"). Both R3 (Fig. lA) and R1 (Fig. 1B) were methylated by an enzyme present in the extracts.
A third peptide (Kl), which contains a lysine in place of the arginine of R1, yielded little or no methyl incorporation (Fig. IC). ACTH 1-24, a highly basic peptide that contains 3 arginines and 4 lysines in sequences unrelated to nucleolin and fibrillarin (Table I), was not methylated when used at concentrations up to 100 p~ (Fig. 1D). These results indicated that PC12 cells contain an arginine methyltransferase that is specific for the fibrillarin/nucleolin-related sequence and that sequences of 10 amino acids or less are sufficient for recognition by this enzyme.
Measurements of methyl incorporation into endogenous substrates revealed that increasing concentrations of the R3 and R1 peptides resulted in increasing inhibition of methylation of endogenous PC12 substrates. A 50% inhibition of endogenous methyl incorporation by the R3 and R1 peptides occurred at -5 and -70 p~, respectively (Fig. 1, A and B ) . In the presence of 1000 pM R3 or R1, the inhibition was 88 and 77%, respectively. This suggests that a large percentage of the total methylation of PC12 proteins under these conditions is catalyzed by the methyltransferase(s) that recognizes the R3 and R1 peptides. The K1 peptide also inhibited the methylation of endogenous substrates (Fig. IC), but much higher concentrations of the peptide were necessary (-1000 pM for 50% inhibition). The ACTH 1-24 peptide did not affect the methylation of endogenous substrates at the concentrations investigated (Fig. 1D).
R3 Preferentially Inhibits Formation of iP,iP-Dimethylarginine in PC12 Cell Proteins-Our previous work showed that the predominant methylated amino acid in extracts from AdOx-treated PC12 cells is P,P-dimethylarginine, followed by p-monomethylarginine and P,WG-dimethylarginine (30). In order to determine which forms of methylated arginine were inhibited by the R3 peptide, we carried out methylation reactions of extracts from AdOx-treated PC12 cells in the presence of varying concentrations of the R3 peptide. These reactions were then subjected to acid hydrolysis and two-dimensional-thin layer chromatography analysis of the methylated amino acids. As shown in Fig. 2, the R3 peptide proved to be most potent in inhibiting the generation of p,P-dimethylarginine (50% inhibition at -1 ~L M peptide), followed by P-monomethylarginine (50% inhibition at -5 p~ peptide). For a 50% inhibition of the generation of w,N"-dimethylarginine, -200 ~L M peptide was necessary, suggesting that this form of methylation might be carried out by an enzyme distinct from that generating p , p -d i m e t h ylarginine. Indeed, a myelin basic protein arginine methyltransferase has been shown to catalyze the formation of Pmonomethylarginine and P,N'G-dimethylarginine but not P,P-dimethylarginine (38). An enzyme forming W,Pdimethylarginine and P-monomethylarginine but not p,NG-dimethylarginine using histone substrates has been partially purified from human placenta (39). An enzyme activity methylating the recombinant hnRNP A1 protein has been detected in rat liver nuclei (40) and is distinct from the enzyme forming p,N"-dimethylarginine in myelin basic protein (40).
Effect of Peptides on the Methylation of Individual Endogenous Methyl Acceptors in PC12 Cells-To determine which endogenous methyl acceptors in AdOx-treated PC12 cells were most sensitive to inhibition by R3, R1, and K1, portions of the methylation reactions used for the measurement of total methyl incorporation into PC12 proteins were analyzed by SDS-PAGE and fluorography. As shown in Fig. 3, the R3, R1, and K1 peptides had a profound effect on the methyl incorporation into PC12 cell methyl acceptors displaying a wide range of M, values. The pattern of methyl acceptors that were inhibited was very similar for all three peptides. The

FIG. 1. Methylation of synthetic peptides and inhibition of methyl incorporation into endogenous substrates in extracts
of AdOx-treated PC12 cells. Extracts from AdOx-treated PC12 cells were incubated in the presence of [methyL3H]AdoMet and varying concentrations of peptides R3 (Ac-GGRGGFGGRGGFGGR-NHZ), and ACTH 1-24. Because only limited quantities of ACTH 1-24 were available, we did not test this peptide at concentrations above 100 p~. Methyl incorporation into the peptides (solid symbols) was assessed by reversed-phase HPLC and liquid scintillation counting as described under "Experimental Procedures." Each point represents the pmol of CH3 incorporated into the peptides in a 15-p1 reaction carried out for 40 min at 30 "C. Methyl incorporation into PC12 cell proteins (open symbols) was analyzed by precipitating the proteins in trichloroacetic acid followed by liquid scintillation counting of the washed pellet. Each point represents the pmol of CH3 incorporated per 30 pg of cell extract protein in the 15-pl reaction. Bars denote the S.D. of triplicate measurements. Where not seen, the bars are contained within the symbols. [methyl-3H]AdoMet, a concentration shown to be optimal for methylation of endogenous substrates under conditions similar to those used here (31). Moreover, the methylation of some proteins was not affected, or was affected only slightly, even in the presence of 1000 p~ R3 peptide. Examples of such methyl acceptors are those with molecular masses of -88, -22, and -20 kDa. These proteins may represent a class of methyl acceptors modified by enzymes distinct from the enzyme catalyzing the majority of methylation in PC12 cells.
In order to reveal more completely the range of methyl acceptors whose methylation becomes inhibited in the presence of these peptides, we performed two-dimensional NEPHGE/SDS-PAGE of extracts from AdOx-treated PC12 cells that had been methylated in the presence or absence of the R3 peptide. As shown in Fig. 4, the methylation of the majority of proteins revealed by NEPHGE/SDS-PAGE was inhibited by the peptide. Two methyl acceptors (-88 and -49 kDa) appeared t o resist the inhibition. The 49-kDa substrate seen in Fig. 4 was not visible on the fluorogram in Fig. 3, where less protein was loaded and a shorter exposure time was employed. The -20-22-kDa methyl acceptors that resist inhibition by R3 and that are seen after SDS-PAGE (Fig. 3) appear to be acidic proteins with large charge heterogeneity, resulting in a smear after NEPHGE/SDS-PAGE that is barely detectable on fluorograms at the exposure time employed for Fig. 4.

Methylation of the FibrillarinlNucleolin-related Peptides by
a Methyltransferme from Bovine Brain-Prior to the synthesis of the R3 and R1 peptides, we partially purified an enzyme from bovine brain that methylates hypomethylated proteins produced in extracts of AdOx-treated PC12 cells. Bovine brain was chosen in anticipation that it would provide a convenient source for subsequent large scale purification of the N-methyltransferase that forms p,p-dimethylarginine.
As described in detail under "Experimental Procedures," an Nmethyltransferase activity from bovine brain cytosol was pu- rified approximately 7-fold, using as substrates a group of hypomethylated proteins from PC12 cells that are retained on S-Sepharose. The profile of enzyme activity obtained by phenyl-Sepharose chromatography is shown in Fig. 5. This enzyme was then tested for its ability to methylate the R3 and R1 peptides. As shown in Fig. 6, both the R3 and R1 peptides served as excellent substrates for the bovine brain methyltransferase, yielding half-maximal rates of methyl incorporation at -0.2 and -2 PM peptide, respectively. The K1 and ACTH 1-24 peptides were not methylated by the bovine brain enzyme (Fig. 6).

DISCUSSION
We have shown here that the methylation of a majority of the endogenous methyl acceptors in AdOx-treated PC12 cells can be inhibited in the presence of low concentrations of peptides R3 and R1. This is consistent with the possibility that there is a single enzyme responsible for the formation of P,p-dimethylarginine in PC12 cell proteins. Alternatively, there may be several, substrate-specific methyltransferases that all recognize the GGFGGRGGFG sequence. Peptides R3 and R1 should prove useful for the purification and characterization of the methyltransferase(s) responsible for the majority of protein arginine methylation.
The efficient methylation of the R1 peptide by enzymes from PC12 cells and bovine brain suggests that this peptide contains a minimal sequence around arginine that is sufficient for recognition by the arginine methyltransferase. It therefore seems likely that proteins containing this sequence would serve as substrates. Assuming that the glycines on the COOH and NH2 termini of the R1 peptide contribute little to recog- FIG. 5. Hydrophobic interaction chromatography of bovine brain arginine methyltransferase. Bovine brain cytosol proteins previously subjected to batch chromatography on DE23 and gradient chromatography on Q-Sepharose were loaded onto a 64-ml column of phenyl-Sepharose in the presence of 4 M NaCl. At fraction 66, a gradient of 4-0 M NaCl was begun, ending at fraction 105, after which the column was washed with buffer lacking NaCl. Fractions (5 ml) were assayed for their ability to methylate basic proteins from AdOx-treated PC12 cells as described under "Methylation Reactions" and for protein content using the method of Bradford (84). 1 Unit = 1 pmol methyl groups transferredlmin.  . 6. Methylation of synthetic peptides by the bovine brain arginine methyltransferase(s). Peptides were incubated at varying concentrations in the presence of partially purified bovine brain methyltransferase (Fig. 5 ) and [methyl-3H]AdoMet. The methyl incorporation into the peptides was analyzed by reversed-phase HPLC and liquid scintillation counting as described under "Experimental Procedures." Bars denote the S.D. of triplicate measurements.
Where not seen, the bars are contained within the symbols. nition, we used the sequence FGGRGGF to search the GenPept database in an attempt to identify candidate substrates. Table I1 shows all sequences of eukaryotic proteins retrieved by the search that contained arginine and that had at most two conservative substitutions or one nonconservative substitution. In addition to the proteins that served as the basis for the design of the peptides, two proteins identified by this search, the hnRNP A1 protein (1,2) and the S2 ribosomal protein (42), are known to be methylated in uiuo, although their sites of methylation have not been reported.
A number of proteins revealed by the search are reported to bind single-stranded nucleic acids. Others contain RNP consensus sequences and are therefore presumed to bind RNA. This suggests that the relationship between arginine methylation and interaction with RNA could extend well beyond the hnRNP A1 protein, nucleolin, and fibrillarin. The reason for such a relationship is not yet clear, but studies on the functional role of the glycine, arginine-rich domains of the hnRNP A1 protein and nucleolin offer a possible explanation. In vitro studies suggest that these domains may interact with nucleic acids (43-47). The glycine, arginine-rich domain of the hnRNP A1 protein has been shown to have DNA and RNA strand-annealing activity (46, 48), while a similar domain of nucleolin has been shown to disrupt the secondary structure of RNA in vitro (49). Because the methylated arginines are present in these domains, the methylation could affect this interaction. It has been suggested that methyl groups in p,p-dimethylarginine might disrupt networks of hydrogen bonds between arginine and RNA (50). These hydrogen bonds appear to mediate specific interactions between certain proteins and RNA (51, 52), e.g. in the interaction of HIV-1 Tat protein with TAR RNA (50, 53). Thus, the domains containing p,p-dimethylarginine might be "locked into nonspecific modes of interaction via electrostatic forces between the arginine side chain and RNA. Another possible role of arginine methylation could be to protect these domains from intracellular proteases (29). The list of candidate methyl acceptors shown in Table I1 is unlikely to be exhaustive because the search failed to identify several known methylation sites in nucleolin and fibrillarin (Table I). These other sites often lack the phenylalanines in the peptide sequences employed here, suggesting that the sequence sufficient for recognition might be more general than that which is present in R1. Further studies with different peptide sequences may narrow down the minimum sequence requirements for recognition by the arginine methyltransferase.
The methyl acceptors resistant to inhibition by the R3 peptide probably represent substrates methylated by different methyltransferases (e.g. Fig. 4, arrowheads). We have shown that these two methyl acceptors are not carboxyl methylated because they do not yield methanol after prolonged treatment at pH 11 at 37 "C (54). There was also no evidence for lysine

Recognition of Peptides by
Arginine Methyltransferase 10507 or histidine methylation in PC12 cells under the conditions of methylation employed here (not shown), suggesting that the methyl groups on these proteins probably do not represent these forms of methylation. Because small quantities of Pmonomethylarginine and P,WG-dimethylarginine remained in the presence of the R3 peptide (Fig. 2), it is possible that the two methyl acceptors are modified by an arginine methyltransferase not recognizing the GGFGGRGGFG sequence.
Other candidates for these two methyl acceptors include proteins methylated on their NH2-terminal amino groups (55) or proteins modified as a result of some other type of methylation. Besides arginine methylation, which seems to be the predominant form of protein methylation in extracts of AdOxtreated PC12 cells, other types of protein methylation have also been detected in these cells. These include methylation of a-carboxyl groups of isoaspartyl sites (54), stable carboxyl methylation akin to methylation of COOH-terminal cysteine residues in the H-ras oncogene product and other GTPbinding proteins (85), and lysine methylation (30). Because the conditions of methylation and/or electrophoretic separation used in this study were suboptimal for detection of these types of methylation and because the levels of these forms of methylation are low relative to arginine methylation, most methylated species in Figs. 3 and 4 most likely represented arginine methylation. For example, the methyl esters of isoas-party1 sites hydrolyze under the conditions of standard SDS-PAGE at pH 8.8 and therefore become undetectable (54). Stable carboxyl methylation in lysates from AdOx-treated PC12 cells is detectable by standard SDS-PAGE; however its abundance is only -2% of arginine methylation and of other forms of methylation that are stable in 6 N HCl or 4 N methanesulfonic acid at 150 "C (85). In an earlier study, we detected low levels of lysine methylation (-7% of total recovered methyl amino acids) in extracts from AdOx-treated PC12 cells (30). In the present study, the level of methyl lysine was below the level of detection, which may have been due to the presence of relatively high concentrations (0.3 M) of NaCl that was included in methylation reactions for optimal detection of arginine methylation. Furthermore, methylation reactions were carried out at pH 7.2, favoring arginine methylation (31) over lysine methylation which has a pH optimum of -8-9 (86).
Finally, it should be noted that the relative abundances of various forms of protein methylation in intact PC12 cells might be different from those in cells treated with AdOx. The number of hypomethylated sites generated by AdOx treatment may depend on, among other factors, the rate of synthesis and/or turnover of methyl-acceptor proteins, possible turnover of the methyl groups themselves and/or varying degrees of inhibition of methyltransferases depending on their K , for AdoHcy generated upon AdOx treatment of cells.