A Novel Methyltransferase Catalyzes the Methyl Esterification of trans-Aconitate in Escherichia coli *

We have identified a new type ofS-adenosyl-l-methionine-dependent methyltransferase in the cytosol of Escherichia coli that is expressed in early stationary phase under the control of the RpoS ς factor. This enzyme catalyzes the monomethyl esterification oftrans-aconitate at high affinity (K m = 0.32 mm) and cis-aconitate, isocitrate, and citrate at lower velocities and affinities. We have purified the enzyme to homogeneity by gel-filtration, anion-exchange, and hydrophobic chromatography. The N-terminal amino acid sequence was found to match that expected for the o252 open reading frame at 34.57 min on the E. coli genomic sequence whose deduced amino acid sequence contains the signature sequence motifs of the major class ofS-adenosyl-l-methionine-dependent methyltransferases. Overexpression of the o252 gene resulted in an overexpression of the methyltransferase activity, and we have now designated it tam for t rans-aconitatemethyltransferase. We have generated a knock-out strain ofE. coli lacking this activity, and we find that its growth and stationary phase survival are similar to that of the parent strain. We demonstrate the endogenous formation of trans-aconitate methyl ester in extracts of wild type but nottam − mutant cells indicating thattrans-aconitate is present in E. coli. Sincetrans-aconitate does not appear to be a metabolic intermediate in these cells but forms spontaneously from the key citric acid cycle intermediate cis-aconitate, we suggest that its methylation may limit its potential interference in normal metabolic pathways. We have detected trans-aconitate methyltransferase activity in extracts of the yeast Saccharomyces cerevisiae, whereas no activity has been found in extracts ofCaenorhabditis elegans or mouse brain.

Our laboratory has been interested in the protein L-isoaspartate (D-aspartate) O-methyltransferase (EC 2.1.1.77), an enzyme that catalyzes the methyl esterification of spontaneously altered residues in a pathway that can lead to the conversion of isomerized aspartyl residues to normal aspartyl residues in a net repair reaction (1). We have postulated that the physiological role of this methyltransferase is to preserve the integrity of the polypeptide chain in the face of age-dependent non-enzymatic reactions that lead to alterations in its configuration (2,3). In the bacterium Escherichia coli, this enzyme is required for optimal survival of stationary phase cells against environ-mental stresses (4,5), presumably functioning to maintain proteins in active configurations under conditions where protein synthesis to replace damaged proteins is limited.
In the course of our studies of the protein L-isoaspartate methyltransferase encoded by the pcm gene in E. coli, we found an activity in cytosolic extracts that appeared to catalyze methyl ester formation but was not dependent upon the pcm gene product. We have now traced this activity to that of a previously undescribed small molecule methyltransferase that is active on trans-aconitate, an apparently non-enzymatically formed derivative of the citric acid cycle intermediate cisaconitate. Since both types of methyltransferase activities are directed to substrates that can be formed by spontaneous agerelated processes, we were interested in characterizing the trans-aconitate methyltransferase. We found that the expression of this activity is dependent on the stationary phase specific factor RpoS. We have purified and characterized this enzyme, identified its gene, and characterized a knock-out strain lacking this activity. We were able to show that transaconitate is an endogenous substrate for the enzyme and that the product is its monomethyl ester.

EXPERIMENTAL PROCEDURES
Preparation of E. coli Cytosol for Methyltransferase Assay E. coli strains and plasmids used in this study are described in Table  I. For analytical studies, E. coli cells were grown to stationary phase (20 h) in 5 ml of Luria-Bertani (LB) broth or M9 medium containing D-glucose supplemented with thiamine (18 g/ml) and leucine (40 g/ ml) (Ref. 10, section A.3). When appropriate, 100 g/ml ampicillin, 50 g/ml kanamycin, 20 g/ml chloramphenicol, or 20 g/ml tetracycline were added. Cells were collected by centrifugation at 5,000 ϫ g at 4°C for 10 min. The cell pellet was resuspended in 0.5 ml of buffer containing 5 mM disodium EDTA, 10% glycerol, 25 M phenylmethylsulfonyl fluoride in 5 mM potassium phosphate buffer at a final pH of 7.0. Cells were lysed by sonication in an ice bath using the microtip of a Branson model W350 instrument at an output control setting of 4 for three sets of 5 pulses separated by 30-s cooling pauses. The extract was centrifuged at 12,000 ϫ g at 4°C for 10 min, and the supernatant was used as a cytosolic fraction. The protein concentration was determined by the method of Lowry et al. (11) after 10% trichloroacetic acid precipitation of the samples.

trans-Aconitate Methyltransferase Assay
reactions were quenched by adding 40 l of freshly prepared 2 M NaOH. Sixty l of this mixture was immediately spotted onto an accordionpleated 1.5 ϫ 8-cm piece of thick filter paper (Bio-Rad 165-0962), and the paper was placed in the neck of a 20-ml vial containing 5 ml of Safety-Solve scintillation fluid (Research Products International Corp.), capped, and incubated for 2 h at room temperature. Radioactivity was determined by liquid scintillation counting in a Beckman LS6500 counter after removal of the filter paper.

Purification of the E. coli trans-Aconitate Methyltransferase
Preparation of E. coli Cytosol-Four flasks each containing 2 liters of LB media were each inoculated with 2 ml of an overnight culture of E. coli strain MC1000 and were grown to stationary phase at 37°C for 20 h with shaking. Cells were collected by centrifugation at 5,000 ϫ g at 23°C for 15 min. The cell pellet (29.1 g) was washed three times with 400 ml of buffer A (50 mM Tris-HCl, 5 mM disodium EDTA, 300 mM NaCl, pH 8.0) and then resuspended in 50 ml of buffer B (50 mM Tris-HCl, 5 mM disodium EDTA, 25 M phenylmethylsulfonyl fluoride, pH 8.0) at 4°C. Cells were disrupted by passing them twice through a French press cell (SLM Aminco) at 20,000 pounds/square inch. The cytosolic fraction was obtained by centrifugation at 23,000 ϫ g for 30 min at 4°C. This supernatant was then further centrifuged at 100,000 ϫ g for 60 min at 4°C to remove any residual membrane material.
Ammonium Sulfate Precipitation-An equal volume of 90% saturated ammonium sulfate (4°C) was gradually added to the cytosol with stirring, followed by additional stirring at 4°C for 30 min. The mixture was then centrifuged at 23,000 ϫ g at 4°C for 30 min. The protein pellet was redissolved in 20 ml of buffer B to a protein concentration of approximately 28 mg/ml. Superdex S-200 Chromatography-An aliquot (5 ml) of the redissolved ammonium sulfate pellet was loaded onto a Superdex-200 (Amersham Pharmacia Biotech) gel filtration column (1.5 cm in diameter ϫ 58 cm in height, 102-ml bed volume), pre-equilibrated at 4°C with buffer C (50 mM Tris-HCl, 5 mM disodium EDTA, pH 8.0). The column was eluted at 18 ml/h, and fractions of 1.2-ml were collected. The activity was eluted at fractions 49 -58, and these fractions were pooled and stored at 4°C. The material from 4 column runs was combined to use in the following step.
DEAE-cellulose Anion-exchange Chromatography-The active pool from the Superdex-200 column (total volume of 48 ml) was loaded onto a DE52 anion-exchange column (Whatman; 2 cm in diameter ϫ 12.7 cm in height, 40-ml bed volume) pre-equilibrated at 4°C with buffer C. After sample loading, the column was washed with 3 column volumes of the equilibration buffer. The enzyme was then eluted with a linear sodium chloride gradient (0 -0.8 M in the equilibration buffer, total of 250 ml) followed by a 5 column volumes of a high salt wash (1.0 M sodium chloride in the equilibration buffer) at 4°C. The flow rate was 19.5 ml/h, and 2.8-ml fractions were collected. Activity was found to elute between sodium chloride concentrations of 300 and 400 mM between fractions 102 and 112. These active fractions were pooled and stored at 4°C.
Hydrophobic Interaction Chromatography--Potassium monobasic phosphate was dissolved in concentrated buffer C, the pH was adjusted to 8.1 with KOH, and the solution was diluted to give a final concentration of 1 M phosphate in buffer C. This solution was added to the active pool from the DE52 anion-exchange column to bring the potassium phosphate concentration to 0.6 M. This material was then applied to a phenyl-Sepharose column (Amersham Pharmacia Biotech; 1 cm diameter ϫ 10 cm height, 7.8-ml bed volume) pre-equilibrated with 0.6 M potassium phosphate in buffer C. The flow rate was 18 ml/h, and 2.5-ml fractions were collected. After loading the sample, the column was washed with 5 column volumes of the equilibration buffer and then eluted with a linear gradient of potassium phosphate (0.6 -0 M in the equilibration buffer). Enzyme activity was found to elute between 0 and 0.1 M potassium phosphate concentrations.

Amino Acid Sequencing
N-terminal amino acid sequence analysis was performed by Dr. Audree Fowler at the UCLA Protein Microsequencing Facility with a Porton 2090E gas-phase sequencer with on-line HPLC detection. The active pool from the DE52 column (fractions 103-110, total volume of 19.6 ml) was added to an equal volume of 25% (w/w) trichloroacetic acid, mixed well by vortexing, and incubated at 4°C while rotating slowly overnight. The mixture was centrifuged at 15,000 ϫ g for 30 min at 4°C, and the protein pellet was then dissolved in 100 l of 1ϫ sample buffer for SDS gel electrophoresis (Ref. 10, section 18.47-18.55). After polyacrylamide gel electrophoresis, the separated polypeptides were electroblotted onto a polyvinylidene difluoride membrane in 25 mM Tris base, 10 mM glycine, 0.5 mM dithiothreitol, in 10% methanol, 90% water (v/v) at pH 9. This membrane was then stained with Coomassie Brilliant Blue R250, and the polypeptide band at 29 kDa corresponding to the trans-aconitate methyltransferase was excised and subjected to automated Edman sequencing.

Cloning and Disruption of the trans-Aconitate
Methyltransferase Gene in E. coli A 3.1-kb DNA fragment containing the entire trans-aconitate methyltransferase gene (o252) and flanking regions was amplified by polymerase chain reaction (PCR) from template DNA in MC1000 cells (12) using Taq polymerase (Promega) at 2.5 mM magnesium chloride and an annealing temperature of 60°C. The primers were KO-5 (5Ј-TATGAC-TACGAAGCGGATCCTAATGGCA, corresponding to bases Ϫ1362 to Ϫ1335 from the translation start site of o252 with the underlined nucleotides changed to create a BamHI site) and KO-3 (5Ј-GCGTATT-GAGAATGGGATCCTAATCACG corresponding to the reverse complement of bases 1714 -1741 with the underlined nucleotides changed to prevent hairpin formation and to create another BamHI site). This fragment was purified by gel electrophoresis using Geneclean II (Bio 101), cut with BamHI, and then ligated into the BamHI site within the multicloning site of the pUC19 vector to generate the plasmid pHC108 (Table I). This plasmid, pHC108, was then used to construct a null mutation in the o252 gene by blunt-end ligation of a 1.5-kb chloramphenicol resistance (Cm r ) cassette (13) at the unique AgeI site within This study the gene to create pHC109. The chromosomal o252 gene was then replaced with the Cm r disrupted gene in pHC109 by homologous recombination in strain JC7623, which does not support plasmid replication (14). pHC109 was transformed into CaCl 2 -competent JC7623 cells, and Cm r colonies were selected on a plate containing 20 g/ml chloramphenicol. The loss of the vector in the recipient strain was confirmed by screening for ampicillin sensitivity. This disrupted o252 gene was subsequently transduced into the MC1000 background by P1 transduction (15). The disruption of the o252 gene was confirmed by the PCR amplification of a 3.96-kb product using a primer (5Ј-GATTCAGTACGC-CAAATGTG) corresponding to genomic sequence upstream of the KO-5 primer described above and a primer OE-3 (described below) corresponding to a sequence downstream of the stop codon for o252 gene. We also confirmed the disruption by the detection of the expected 3.44-kb EcoRV, 6.35-kb HindIII, and 0.60-and 1.35-kb EcoRI fragments using a random-primed probe corresponding to the Cm r gene in Southern blot hybridization (data not shown).

Overexpression of trans-Aconitate Methyltransferase
The trans-aconitate methyltransferase gene was PCR-amplified from colonies of E. coli strain MC1000 as described above using the primers OE-5 (5Ј-CGGGAGTAAACATATGTCTGACTGG; corresponding to bases Ϫ13 to ϩ12 from the translation start site with the underlined bases changed to create a 5Ј NdeI site) and OE-3 (5Ј-ACCACTGGATC-CCATATGCAACGC; corresponding to the reverse complement of bases ϩ848 to ϩ871 with the underlined based changed to create a BamHI site and to prevent hairpin formation; the stop codon is located at bases ϩ757 to ϩ759). The 884-base pair PCR fragment was cleaved with NdeI and BamHI and the large fragment purified as described above. This fragment was then cloned into the corresponding sites in the multicloning site of the pT7-7 vector (16) to generate pHC107. DNA sequence analysis using both oligonucleotides described above as primers showed that no mutations were introduced during the cloning procedure. The plasmid was then transferred into BL21(DE3) cells (Invitrogen) for expression. An aliquot of an overnight culture of the transformed cells (20 l) was diluted into 20 ml of fresh LB medium, incubated with shaking at 37°C, and cultured to an A 600 nm of 1.0. Isopropyl-␤-Dthiogalactopyranoside was added to a final concentration of 1 mM to the culture, and the cells were incubated with shaking for another 2 h. Cytosolic fractions were then prepared and assayed for trans-aconitate methyltransferase activity as described above. The enzymatic activity of this preparation was 100 -150 nmol/min/ml (specific activity 12-22 nmol/min/mg protein), and it was used as a concentrated source of enzyme in the enzyme kinetic assays and in the experiments to characterize the methyl acceptors. A control extract was prepared where the same strain was grown in the absence of plasmid. The specific activity of this extract was 0.020 nmol/min/mg protein, indicating that the degree of overexpression was about 630-fold.

High Performance Liquid Chromatography
Analysis of the substrates and products was carried out with a Waters HPLC system. An Alltech Partisil SAX anion-exchange column (250 mm length ϫ 4.6 mm inner diameter; 10-m resin bead diameter) was used for initial separations. The column was equilibrated with 50 mM potassium phosphate, pH 4.5, and eluted at 1 ml/min at room temperature. After each run, the column was regenerated with 500 mM potassium phosphate, pH 4.5, prior to re-equilibration with the starting buffer. For reverse-phase separations, an Alltech Econosphere C18 column (250 mm length ϫ 4.6 mm inner diameter; 5-m spherical resin bead diameter) was used in a two-solvent system. Solvent A is 0.1% trifluoroacetic acid in water, and solvent B is 0.1% trifluoroacetic acid, 99.5% acetonitrile, and 0.4% water. The column was eluted at room temperature at a flow rate of 1 ml/min for 20 min in solvent A, followed by a linear gradient over 20 min from 100% solvent A to 100% solvent B, followed by 10 min of 100% solvent B. The column was re-equilibrated with 100% buffer A. In each case, 1-min fractions were collected.
The UV profile was monitored by Waters model 441 absorbance detector at 214 nm, and chromatographs were recorded with the Pow-erChrom system from ADInstruments. Radiolabeled products were detected by counting 100 l of each in 5 ml of scintillation fluid.

Thin Layer Chromatography
Thin layer chromatography was also used to analyze the products of the methylation reaction after the method of Otten and Mehltretter (17). Polyester-based 60-Å silica gel-coated plates were used (Whatman PE SIL G, 250 m layer). The solvent is benzene/methanol/acetic acid (45:16:4, v/v/v) and is prepared fresh daily and used to pre-equilibrate the chamber for 2 h before each run at room temperature. Samples (1-5 l) were spotted in each lane in 1-l aliquots, and each spot was allowed to air-dry before further application of sample or chromatography. The solvent front was allowed to migrate approximately 18 cm on the plate, then its position was marked and the plate was air-dried in a hood, followed by baking at 105°C in a vacuum oven for 1 h. Carboxylic acid-containing compounds were detected as yellow spots on a blue background after spraying the plate with 0.04% bromcresol green dissolved in 95% ethanol, pH 8.0.

Synthesis and Characterization of trans-Aconitate Methyl Esters
Chemical Methylation-trans-Aconitic acid (4.8 mg) was incubated with 60 l of methanol and 1 l of concentrated (12 M) HCl for 16 h at room temperature. The sample was vacuum-dried in Speedvac apparatus and dissolved in 138 l of H 2 O to give a final concentration of trans-aconitate derivatives of 0.2 M. An aliquot of the sample (10 l) was chromatographed on the SAX HPLC anion-exchange column as described above. The peaks were collected, and an aliquot of each peak was re-chromatographed on the C18 reverse-phase column as described.
Enzymatic Methylation-The reaction mixture contains 5 l of 0.02 M trans-aconitate in 0.4 M sodium HEPES, pH 7.5, 10 l of 0.4 M sodium HEPES, pH 7.5, 1 l of a preparation of cytosol from BL21 cells overexpressing the trans-aconitate methyltransferase (7.72 g protein, 12.6 pmol/min/g protein), 17.5 l of 8.16 mM S-adenosyl-L-methionine (AdoMet) in water, and 2.5 l [ 14 C]AdoMet to a total volume of 42.5 l made up by H 2 O. The reaction was carried out at 37°C for 24 h. Forty l of the sample was purified as described above. The methylated trans-aconitate was followed by radioactivity.
Mass Spectroscopy-Mass spectroscopy was performed by Dr. Kym Faull at the UCLA Mass Spectrometry Facility. HPLC fractions from the C18 reverse-phase column were collected and dried in a SpeedVac. The dried HPLC samples were redissolved in 20 l of water/acetonitrile/ triethylamine (50:50:0.1, v/v/v), and aliquots were injected into an electrospray ionization source attached to a quadrupole mass spectrometer (Perkin-Elmer, Thornhill, Canada, API III; Ϫ3.5 Kv ion spray voltage, spray nebulization with hydrocarbon-depleted air ("zero" grade air, 40 pounds/square inch, 0.6 liters/min; Zero Air Generator, Peak Scientific, Chicago, IL), curtain gas (0.6 liters/min) from the vapors of liquid nitrogen; mass resolution set so the isotopes of the polypropylene glycol/ NH 4 ϩ singly charged ion at m/z 906 were resolved with 40% valley) scanning from m/z 120 -250 in the negative ion mode. Spectra were collected (step size 0.3 Da, dwell time 20 ms/step, 6.7 s/scan, orifice at Ϫ60 V), and the resulting spectra were summed then background subtracted with software supplied with the instrument.

Identification of a Novel Methyltransferase Activity in E.
coli-In the course of studies quantitating protein L-isoaspartate O-methyltransferase activity in various strains, we measured methyl esterification in cytosolic extracts in the presence and absence of added L-isoaspartyl-containing methylaccepting peptide KASA(isoD)LAKY. [ 14 C]Methyl esters formed when extracts are incubated with [ 14 C]AdoMet in a sodium citrate buffer were hydrolyzed in base to generate volatile [ 14 C]methanol that can be separated from unreacted [ 14 C]AdoMet and other non-volatile species and quantitated. Although we found that the L-isoaspartyl peptide-dependent activity correlated well with the presence of the pcm gene for the isoaspartyl methyltransferase, we were surprised to find that the "endogenous" activity in the absence of added peptide was much higher in extracts from strains with an intact rpoS gene than in extracts from strains mutated in this gene (Table  II). The rpoS gene is located about a kilobase downstream from the pcm gene and codes for a specific factor that is required for the expression of a number of genes in stationary phase cells (for a review, see Ref. 18). This result suggested that either rpoS or one of the genes it regulates might also have a methyl esterification activity.
We then began to investigate the nature of the RpoS-dependent endogenous methyl esterification activity using extracts of the strain HC1011 that lacks the pcm gene so there would be no contribution of the L-isoaspartyl methyltransferase to the methylation activity. We found that about 78% of the total activity was localized in the cytosolic fraction, and 22% of the activity was found in the membrane pellet fraction. The combination of membrane and cytosolic fractions did not increase the activity over that of cytosolic fractions alone (data not shown). The production of the endogenous methylated product from cytosolic extracts was found to be linear with time and with the amount of cytosolic extract used (data not shown). The activity appeared to be a protein because treatment of the extract with proteinase K resulted in a complete loss of activity and because heating the extract greatly reduced the activity. Finally, we showed that the activity could be inhibited by S-adenosyl-L-homocysteine, a product of the reaction and an effective inhibitor of most AdoMet-dependent methyltransferases (19). We found 72% inhibition with 0.38 mM S-adenosyl-L-homocysteine and essentially complete inhibition at 0.94 mM (data not shown).
Initial Characterization of Substrates and Products-The formation of volatile radioactivity by the RpoS-dependent methyltransferase in the assay described above was dependent upon hydrolysis of the reaction products, suggesting that a methyl ester linkage was formed. No radioactivity was detected when water replaced the 2 M NaOH quenching solution. There was no increase in activity when 7 M NaOH was used as a quench; 74% of the activity was detected with 1 M NaOH and only 49% activity found with 0.2 M NaOH. Only 2, 3, and 8% of the maximum activity was found when 1, 2, and 7 M HCl was used as a quench (data not shown).
No loss of activity was seen after dialysis of the extract, initially suggesting that a macromolecule may be the methyl acceptor. However, we found that the endogenous activity was dependent upon the presence of the citrate buffer in the assay mixture. When the citrate was replaced by Tris, phosphate, acetate, or MES buffers of similar pH, no activity was seen (data not shown). The role of citrate in the reaction did not appear to be that of a metal chelator because there was no activity in a buffer containing 5 mM sodium EDTA. These results suggested that citrate could itself be the substrate for the reaction where one or more of its three carboxyl groups could be methyl-esterified. To ask whether compounds structurally related to citrate would be better or worse substrates, we assayed extracts with a variety of tricarboxylates and dicarboxylates. While we found that succinate, fumarate, malate, oxalacetate, and tricarballylate gave no activity when substituted for citrate, even better activity was found with cis-aconitate, DL-isocitrate, (2R,3S)-isocitrate, and trans-aconitate. From the concentration dependence of the activity, we estimated an apparent K m value for trans-aconitate of about 0.3 mM, a value at least 16-fold lower than that measured for citrate (7.1 mM), cis-aconitate (33 mM), (2R,3S)-isocitrate (9.1 mM), or DL-isocitrate (5 mM). We found that the reaction with citrate had a maximal velocity of only 7.5% that of transaconitate, whereas the reaction with cis-aconitate, (2R,3S)isocitrate, DL-isocitrate had comparable maximal velocities (108, 49, and 78%) (data not shown). By using trans-aconitate as a substrate, we found that the apparent K m for [ 14 C]AdoMet is about 5 M, whereas the pH dependence of the enzyme activity demonstrated a broad maximum from pH 6 to 8 decreasing at higher or lower pH values with half-maximal activities at about pH 5.5 and pH 8.5 (data not shown).
trans-Aconitate-dependent Methyltransferase Activity Rises as Cells Enter Stationary Phase and Then Decreases-Because the methyltransferase activity required the presence of the RpoS stationary phase factor (Table II) whose concentration increases near the end of exponential growth phase (20), we assayed this enzyme during various stages of cell growth using trans-aconitate as a substrate. We found that the specific activity of the methyltransferase increases dramatically in late exponential phase and peaks at the transition into stationary phase (Fig. 1A), paralleling the known accumulation of RpoS (21). The specific activity of the methyltransferase then decreases dramatically after 24 h in stationary phase (Fig. 1B), which also correlates with the decrease in the concentration of RpoS (21). There is little or no detectable activity after 72 h in stationary phase. We also observed that the specific activity of trans-aconitate-dependent methyltransferase is 3-4-fold higher in the cytosolic fractions obtained from cells cultured in rich LB media than from cell cultured in minimal (M9) media with glucose (data not shown). This difference may also reflect the higher level of RpoS in rich media compared with minimal media (20,21).
Purification of trans-Aconitate Methyltransferase-From the results presented above, it is possible that either RpoS has a methyltransferase activity or that it is essential for the transcription of the methyltransferase gene. To clarify this issue, the methyltransferase was purified by ammonium sulfate precipitation, gel filtration chromatography on Superdex 200, anion-exchange chromatography on DE52 resin, and hydrophobic chromatography on phenyl-Sepharose resin as described under "Experimental Procedures." A single peak of activity was found in each chromatographic step suggesting that isozymes are not present (Fig. 2). Characterization of the polypeptide composition of each fraction by polyacrylamide gel electrophoresis in sodium dodecyl sulfate (SDS) revealed a polypeptide of about 29 kDa that is progressively enriched during the purification (Fig. 3). In the final step of hydrophobic chromatography, only a single band at 29 kDa was observed whose concentration corresponded directly to the methyltransferase activity (data not shown). The overall purification of 594-fold (Table III) suggested that the enzyme made up approximately 0.2% of the total cytosolic protein in early stationary phase. Additional native gel filtration experiments showed that the methyltrans- ferase eluted between the molecular weight markers ovalbumin (43 kDa) and the human L-isoaspartyl methyltransferase (25 kDa) and thus appears to be composed of a single polypeptide chain.
Identification of the E. coli Gene Encoding the trans-Aconi-tate Methyltransferase-Microsequencing of the purified 29-kDa polypeptide demonstrated an N-terminal amino acid sequence of SDXNKQLYLQFMAEMS, where the assignment of residues 1, 2, and 4 was more certain than residues 5-16, and no assignment could be made for residue 3. This sequence, including an assumed N-terminal methionine residue, was used to search against the GenBank TM protein data base using ungapped BLAST (22). The best match corresponded to the N terminus of the deduced product of the o252 open reading frame in the newly sequenced E. coli genome (23) where identities were found at 11 of the 17 positions (Fig. 4). This previously uncharacterized gene, at 34.57 min on the chromosome, 2 potentially encodes a polypeptide of 28,876.4 Da (lacking the initiator methionine which would be expected to be removed (25)) that corresponds to the 29-kDa polypeptide purified. The deduced sequence also contains the four well conserved motifs (I, post-I, II, and III) in a variety of methyltransferases (Ref. 26; Fig. 4).
Overexpression of the trans-Aconitate Methyltransferase and Methyl-accepting Substrate Characterization-The o252 gene was PCR-amplified and cloned into a pT7-7 expression vector as described under "Experimental Procedures." We found that the specific activity of the trans-aconitate methyltransferase was increased 630-fold in extracts of BL21(DE3) cells compared with extracts of this strain lacking the plasmid. This result suggests that the o252 gene does indeed encode the methyltransferase activity, and we have now named it tam for transaconitate methyltransferase.
The availability of the overexpressed enzyme allowed us to characterize its substrate specificity. Initial velocity measurements at a variety of substrate concentrations confirmed that trans-aconitate was the best substrate with a K m value of 0.32 mM (Table IV). The catalytic efficiencies (V max /K m ) of the enzyme for cis-aconitate, (2R,3S)-isocitrate, and DL-isocitrate were less than 3% that of trans-aconitate, whereas the catalytic efficiency for citrate was only about 0.4% that of trans-aconi-   tate (Table IV). The values in Table IV obtained for the overexpressed enzyme are generally similar to those found when the cytosol of wild type cells was used as a source of enzyme (see above). Finally, we found that the K m for AdoMet with trans-aconitate was 4.8 M, a value also similar to that obtained with the non-overexpressed cytosol.
To exclude the possibility that minor contaminants in the commercial preparations of trans-aconitate, cis-aconitate, (2R,3S)-isocitrate, or citrate could represent all or part of the methyl-accepting activity observed in the experiments described above, we fractionated each these compounds by HPLC high resolution anion-exchange chromatography (Fig. 5). Individual fractions were then assayed as potential methylaccepting substrates for the preparation of overexpressed trans-aconitate methyltransferase. When trans-aconitate was fractionated, we found that the methyl-accepting activity exactly paralleled the elution of trans-aconitate (monitored by its UV absorbance), suggesting that this was in fact the methyl acceptor (Fig. 5A). On the other hand, when the commercial preparation of cis-aconitate was fractionated, a more complex pattern was seen. Here, UV analysis indicated that a small contaminant (approximately 5%) of trans-aconitate was pres- ent. This material was a major methyl acceptor, however, as would be expected from the results described above (Fig. 5B). Significantly, however, there was still a methyl-accepting peak corresponding to the cis-aconitate peak. Thus, cis-aconitate appears to also be recognized by the methyltransferase but at even lower affinity than suggested by the data in Table IV.
When we fractionated (2R,3S)-isocitrate we found that all of the methyl-accepting activity co-eluted with the major peak of isocitrate (Fig. 5C), suggesting that isocitrate is also a methylaccepting substrate of this enzyme. Finally, fractionation of citrate showed that the methyl-accepting activity followed the peak of citrate (data not shown).
Analysis of the Methyltransferase Reaction Products in Vitro-trans-Aconitate can be methyl-esterified to produce one trimethyl-, three structurally distinct dimethyl-, and three structurally distinct monomethyl esters of trans-aconitate. We first analyzed the products of the chemical methyl esterification of trans-aconitate by anion-exchange HPLC at pH 4.5. We found UV-absorbing products eluting at 5.5, 9.9, 11.5, and 15.6 min and a very small peak of residual trans-aconitate was found at 50.4 min (Fig. 6A). Mass spectral analyses demonstrated that the 5.5-min peak contained both mono-and dimethyl derivatives of trans-aconitate and that the 11.5-and 16-min peaks contained monomethyl trans-aconitate derivatives. No trans-aconitate derivatives were detected in the 9.9min peak. In a parallel experiment, we then compared the elution position of the enzymatically generated [ 14 C]methyl esters formed by the E. coli enzyme and trans-aconitate with [ 14 C]AdoMet (Fig. 6A). About 27% of the radioactivity was found in the flow-through fractions at 4 -8 min in the position expected for residual [ 14 C]AdoMet. The remaining 73% of the radioactivity was found to elute in a distinct peak at 16 min in an amount expected for the methyl ester product of the reaction. When the material eluting at 16 min was treated with 2 N NaOH, 95% of the total radioactivity was base-labile, consistent with a methyl ester product (data not shown). This radio- Methyltransferase activity was measured in 0.1 M HEPES buffer at pH 7.5 as described under "Experimental Procedures," except that 5 l of 0.36 mM [ 14 C]AdoMet was used. Substrates were dissolved in 0.1 M HEPES at pH 7.5, and the pH was readjusted to 7.5. The enzyme source was a cytosolic extract of overexpressed enzyme. Substrates were used over a concentration range of 1-64 mM, except for trans-aconitate where a range of 20 M to 2 mM was used. Assays were done in triplicate at both 5-and 10-min incubation times. K m and V max values were calculated by fitting the data to the Michaelis-Menten equation using the DeltaGraph program. The standard deviation of each measurement is indicated. Since cis-aconitate is contaminated with a small amount of transaconitate (Fig. 5), this K m value represents a minimal estimate.
FIG . 5. Methyl acceptor activity with fractionated trans-aconitate, cis-aconitate, and (2R,3S)-isocitrate. Commercial preparations of trans-aconitate, cis-aconitate, and (2R,3S)-isocitrate (10 l of 0.2 M) were each HPLC-purified on an anion-exchange SAX column as described under "Experimental Procedures," and 1-ml fractions were collected. An aliquot of each fraction (0.1 ml) was assayed with 15 l of an E. coli cell cytosol containing recombinant trans-aconitate methyltransferase (36,000 pmol/min/ml; 11,300 pmol/min/mg protein) and 1 l of 365 M [ 14 C]AdoMet (53 mCi/mmol). The reaction mixture was incubated at 37°C for 1 h, followed by addition of 116 l of 2 N NaOH. An aliquot (100 l) of the quenched mixture was spotted on filter paper for the vapor phase determination of [ 14 C]methanol as described under "Experimental Procedures." Assays on each fraction were done in duplicate, and the average is shown in closed circles. The absorbance at 214 nm is given as the continuous line.

FIG. 6. A monomethyl ester of trans-aconitate is the product of the enzymatic reaction.
A, a mixture of the methyl esters of transaconitate was prepared chemically and fractionated by anion-exchange chromatography on a SAX column as described under "Experimental Procedures." The elution position of the products are indicated by the absorbance at 214 nm (solid line). In a parallel experiment, the elution position of the radiolabeled products of the trans-aconitate methyltransferase was determined on the same chromatography system. E. coli HC1011 cytosol (125 g protein) was incubated with 3 l of 0.2 M trans-aconitate and 5 l of [ 14 C]AdoMet in 0.1 M sodium HEPES, pH 7.5, at 37°C for 2 h. The sample was then chromatographed under the same conditions as for the chemically methylated species. B, the major 14 C-methylated product from the column shown in A (150 l of fraction 16) was mixed with the corresponding peak from the fractionation of the chemically synthesized methyl esters and rechromatographed on a C18 reverse-phase column as described under "Experimental Procedures." labeled peak eluted in the same position as one of the peaks of the chemically synthesized methyl esters corresponding to a monomethyl ester (Fig. 6A). To determine the nature of this material, the peaks of material eluting at 16 min from the chromatographic separations of the enzymatically and chemically synthesized material were then mixed and further fractionated on a C18 reverse-phase column (Fig. 6B). Here, the radioactivity was found to elute in a single peak at about 11 min that also corresponded to a peak of absorbance of 214 nm. Analysis of this material by mass spectroscopy indicated that it has an m/e of 187 consistent with one or more of the monomethyl esters of trans-aconitate. Since the analysis of the other UV-absorbing peaks in the experiment shown in Fig. 6A showed that monomethyl esters of trans-aconitate are also present in the 5.5-and 11.5-min peaks, these results suggest that enzymatic reaction may be specific for a single carboxylic acid group of trans-aconitate. In additional experiments, we prepared the enzymatic product on a larger scale with the overexpressed enzyme as described under "Experimental Procedures" and confirmed these results with mass spectral analysis of the product separated by anion-exchange and reversephase chromatography.
Characterization of E. coli Cells Lacking the trans-Aconitate Methyltransferase-To confirm further our identification and to study the physiological function of the methyltransferase, we constructed an E. coli strain where a chloramphenicol resistance element was inserted into the middle of the tam gene as described under "Experimental Procedures." Extracts of this strain, designated HC1014, demonstrated no methyltransferase activity using trans-aconitate, cis-aconitate, citrate, or DL-isocitrate as substrates (data not shown). Because this gene is expressed under the control of the RpoS factor in stationary phase (Table II and Fig. 1), we measured the survival rates of the wild type and mutant strains in stationary phase and under environmental stresses of heat shock, osmotic stress, ethanol treatment, and oxidative stress in experiments similar to those described in Visick et al. (4). We found no detectable differences between the wild type and mutant strains under any of the conditions tested. We also showed that the growth rates were similar in LB media and in minimal media containing either D-glucose or acetate as a carbon source (data not shown). Finally, neither strain was able to grow in a solid or liquid media containing trans-aconitate as the sole carbon source (27), and we could detect no aconitate isomerase activity in wild type or mutant cells using the method of Watanabe et al. (27).
Endogenous Substrates of the trans-Aconitate Methyltransferase in E. coli-To characterize the endogenous substrates of this enzyme, we incubated cytosolic extracts of the parent MC1000 strain (tam ϩ ) and the mutant HC1014 strain (tam Ϫ ) with S-adenosyl-L-[methyl-3 H]methionine ([ 3 H]AdoMet) in the absence of any exogenous methyl acceptors. We analyzed the extracts for radioactivity in compounds present in the parent but not in the mutant strain lacking the trans-aconitate methyltransferase. We first ether-extracted the acidified reaction mixtures and chromatographed the ether-soluble phase on anion-exchange HPLC (Fig. 7A). We found that a peak of radioactivity was present in the position of the in vitro enzymatically formed monomethyl ester of trans-aconitate in the parent strain but no radioactivity at this position in the mutant strain. To confirm that the endogenous material from the parent strain was the methyl ester of trans-aconitate, we pooled the radioactive peak and subjected the material to reverse-phase HPLC. Here we found that the material again chromatographed as the methyl ester of trans-aconitate (Fig. 7B). Finally, we pooled the reverse-phase radioactive peak and sub- The reaction mixtures were then acidified with 20 l 10 N sulfuric acid, and 200 l ether was used to extract the products three times. The ether phase from the three-time extraction was combined and back-extracted with 100 l of water. The resulting ether phase was air-dried and dissolved into 100 l of water. A, 20 l of the ether phase was applied to a SAX anion-exchange column eluted with 50 mM potassium phosphate at pH 4.5. Fractions of 1 ml were collected; 100-l aliquots were mixed with 100 l of 2 N NaOH and base-labile volatile radioactivity determined as described. Radioactivity from the wild-type extract is shown with closed squares and that from the mutant extract with open squares. The enzymatic methylated product of a standard of transaconitate was found to elute at 13.5 min in a parallel experiment. B, fraction 14 (100 l) containing 3 H radioactivity from the wild-type incubation in A above was mixed with a standard of trans-aconitate enzymatically methylated with [ 14 C]AdoMet and the mixture rechromatographed on a C18 reverse-phase column as described under "Experimental Procedures." Fractions of 1 ml were collected, and 0.1 ml was directly added into 5 ml of scintillation fluid and counted. The 3 H radioactivity derived from fraction 14 is shown in closed circles, and the 14 C radioactivity derived from the standard is shown in open circles. C, 0.5 ml of fraction 11 from the reverse-phase column was dried down in SpeedVac and dissolved in 5 l of H 2 O. Thin layer chromatography was carried out as described under "Experimental Procedures." After the run, without staining, the lane was cut into 0.5-cm pieces, and the silica from each piece was scrapped into a microcentrifuge tube. 100 l of H 2 O was added to each tube, and the tube was vortexed vigorously for 30 s. The tube was then centrifuged in a microcentrifuge, and the supernatant was counted for 3 H radioactivity (closed circles) and 14 C radioactivity (open circles).
jected it to thin layer chromatography (Fig. 7C). Once again the radioactivity co-migrated with the methyl ester of trans-aconitate. These experiments demonstrate that trans-aconitate is present in E. coli extracts and is an endogenous substrate of the methyltransferase.
Presence of the Activity and Gene in Other Organisms-We assayed the citrate-, DL-isocitrate-, and trans-aconitate-specific activity in cell extracts of Saccharomyces cerevisiae, Caenorhabditis elegans, and mouse brain. The methyltransferase activity is present in yeast but absent in nematodes and mouse brain ( Table V). The specific activity of the enzyme in the yeast extract used was about half that seen in extracts of E. coli in stationary phase. Analysis of the GenBank TM data base revealed a homolog of the E. coli trans-aconitate methyltransferase gene in Mycobacterium tuberculosis but, surprisingly, not one in the complete genome of S. cerevisiae. However, examination of several unfinished microbial genomes available through the National Center for Biotechnology Information indicated that there are apparent homologs of trans-aconitate methyltransferase in Deinococcus radiodurans, Pseudomonas aeruginosa, and Yersinia pestis (Fig. 4). DISCUSSION We have identified a novel O-methyltransferase in E. coli. It methylates one of the three carboxyl groups of trans-aconitic acid to form a monomethyl ester. The enzyme also recognizes cis-aconitate, (2R,3S)-isocitrate, and citrate but with much higher K m values and/or much lower V max values. No reaction is seen with other related metabolites such as succinate, fumarate, malate, or oxalacetate. The fact that it is not active on tricarballylate (the saturated form of aconitate) suggests the importance of an olefinic or hydroxyl function in the recognition of the substrate for the methyl transfer reaction. Further work will be required to identify which carboxyl group is modified in trans-aconitate.
trans-Aconitate appears to be present in E. coli because we can isolate the radiolabeled methyl ester after incubation of cell extracts with [ 3 H]AdoMet. Its origin, however, is unclear. Wild type E. coli cells are reported to be unable to use trans-aconitate as a carbon source (28), and we are unaware of any pathways where it is a product or substrate for an enzymatic reaction in this organism. We have confirmed that E. coli cells cannot grow on trans-aconitate nor do they contain an aconitate isomerase activity that can convert the citric acid cycle intermediate cis-aconitate to trans-aconitate. Nevertheless, trans-aconitate can be formed spontaneously from cis-aconitate (29 -32).
What advantage might the ability to methylate trans-aconitate give E. coli cells? At least in mammalian systems, transaconitate is an inhibitor of two central enzymes of the citric acid cycle, aconitase (33, 34) and fumarase (35,36). The possibility thus exists that the enzymatic methylation of trans-aconitate can attenuate its inhibition of these crucial enzymatic reactions in energy metabolism. This may occur either because the methyl ester of trans-aconitate is inherently less inhibitory to central metabolic reactions or by a novel type of pathway that might convert trans-aconitate methyl ester to a less toxic species. It is even possible that the methylation of trans-aconitate might initiate a pathway that could result in its net conversion to cis-aconitate and its return to the citric acid cycle. These alternatives are presently under investigation in our laboratory. Significantly, the methyltransferase is expressed in early stationary phase when the cessation of rapid cell division may allow altered metabolites to accumulate (see below). Nevertheless, we have been unable to demonstrate a growth phenotype in the methyltransferase-deficient strain we have constructed in this work.
An interesting aspect of trans-aconitate methyltransferase is that its expression appears to be regulated by the stationary phase specific factor RpoS (20). Upon starvation, E. coli cells embark upon a developmental program resulting in metabolically less active and more resistant cells (37,38). The starvation-induced expression of many genes is controlled by RpoS, and an intact rpoS allele is crucial for maintaining cell shape, resistance to multiple stresses, synthesis of glycogen, and long term survival in stationary phase cells. The fact that the expression of the trans-aconitate methyltransferase is dependent on the presence of an intact rpoS gene and the activity of the enzyme in different growth phases correlates with that of RpoS suggests that the ability to metabolize trans-aconitate is most important when cell division is limited and potential inhibitors might be expected to accumulate. The loss of enzyme activity after extended stationary phase is likely to reflect both the loss of RpoS protein and the instability of the enzyme under these conditions, but it is not clear what the physiological significance of this decrease is.
Although we have detected an active trans-aconitate methyltransferase activity in the yeast S. cerevisiae, no activity has been found in extracts from nematodes or mouse brain. Interestingly, although potential homologous open reading frames have been found in the genomes of a number of procaryotes, there is no clear homolog in the complete genome sequence of yeast. This latter result indicates that this activity may have arisen independently in yeast and bacteria or that the sequence may have diverged rapidly.
In certain plants, trans-aconitate is made in relatively large amounts and can represent up to 12% of tissue dry weight (39,40). The enzyme responsible for this conversion is aconitate isomerase that catalyzes the formation of trans-aconitate from cis-aconitate (27,41). The function of the accumulation of trans-aconitate in plants is not clear. It does present a problem for animals that consume it such as ruminants where it is associated with "grass tetany," a calcium-magnesium defi-TABLE V trans-Aconitate methyltransferase activities in several organisms Cytosolic extracts from various organisms were assayed for citrate, DL-isocitrate, and trans-aconitate-dependent activities as described under "Experimental Procedures." For the endogenous activities, no tricarboxylic acid was added. For substrate-dependent activity, final concentrations of 50 mM citrate, 20 mM DL-isocitrate, or 1 mM trans-aconitate were added, and the endogenous activities were subtracted. Yeast cytosol (23 mg of protein/ml) from strain CH9100 -2 was prepared by Dr. Jonathan Gary, nematode cytosol (10 mg of protein/ml) from strain CB1489 by Dr. Agnieszka Niewmierzycka, and mouse brain cytosol ( ciency linked to the chelation properties of a ruminal bacterial metabolite of trans-aconitate, tricarballylate (24,42).