Purification and characterization of the S-adenosylmethionine:glutamyl methyltransferase that modifies membrane chemoreceptor proteins in bacteria.

The enzyme (EC 2.1.1.24) from Salmonella typhimurium that catalyzes the S-adenosylmethionine-dependent methyl esterification of glutamyl residues in membrane chemoreceptor proteins has been purified to homogeneity, and the nucleotide sequence of the gene coding for this protein, cheR, has been determined. The molecular weight, amino acid composition, and N-terminal amino acid sequence of the purified protein correspond to the values predicted from the sequence of the gene. The pure protein is a 33-kDa monomer. Kinetic studies indicate that, at levels of receptor and S-adenosylmethionine present in wild type cells, the transferase is nearly saturated. The enzyme has a relatively low turnover number, approximately 10 mol of methylester formed per mol of enzyme per min; and there appear to be only approximately 200 methyltransferase monomers per wild type cell.

In bacteria such as Escherichia coli and Salmonella typhimurium a set of membrane chemoreceptor proteins are reversibly methylated at glutamyl residues (for reviews, see Refs. 1-3). Each receptor is composed of a periplasmic Nterminal domain that binds stimulatory ligands, a transmembrane region, and a C-terminal cytoplasmic domain that produces a signal that controls the flagellar motor (4). Receptor signaling is regulated by the methylation and demethylation of specific glutamate residues within the cytoplasmic domain. Two enzymes are involved a methylesterase (5) and an AdoMetl-dependent methyltransferase (6). Although the esterase has been purified to homogeneity and extensively characterized (7), the only characterization of the methyltransferase has been with partially purified enzyme preparations (8,9).
In this paper, we present the purification and characterization of the methyltransferase enzyme together with the nucleotide sequence of the corresponding gene. Previous attempts to purify this enzyme have been hampered by its were replaced with CA. This was accomplished as described previously GATGTCATAGCGCCTTCTTAAT with GATGTCATATGGCCT- (15,16) by replacing the normal cheR coding strand sequence TCTTAAT. To confirm the mutation, phage DNAs were sequenced (17) using the synthetic oligonucleotide GCGCGTCGGACAGCGC, complementary to an adjacent region, as primer. The mutated c h R gene was reintroduced as an EcoRI-Hind111 fragment into a pUC12 pME43.
variant with no NdeI site? The resulting construct was designated

Methyltransferase Assay
Methyltransferase was assayed using the S. typhimurium aspartate receptor, Tar, and AdoMet as substrates. The receptor was produced in an E. coli methyltransferase-deficient strain, RP4080 (12), carrying the S. typhimurium tar gene on a multicopy plasmid, pWK3-55.2.
These cells were grown to a density of 2 X lo9 cells/ml in 6 liters of L broth (18) at 30 "C, harvested by centrifugation for 20 min at 13,000 X g, and washed with phosphate/EDTA buffer (0.10 M potassium phosphate, 1.0 mM EDTA, pH 7.0). The pellet, 15 g, wet weight, was resuspended in 45 ml of phosphate/EDTA buffer and disrupted using a Branson Model 200 Sonifier. Unbroken cells and large debris were removed by centrifugation for 15 min at 16,000 X g. The supernatant was centrifuged an additional 60 min at 100,000 X g. The pellet was resuspended in 2.5 ml of phosphate/EDTA buffer using a Teflon glass homogenizer (protein concentration of 25-35 mg/ml). The receptor fraction was divided into 0.3-ml aliquots and stored at -80 "C until just prior to use. Unless stated otherwise, all procedures were performed at 4 "C. Methyltransferase activity was assayed by incubating 17 pl of methyl-acceptor membranes (see above), 8 pl of 100,000 X g supernatant from a S. typhimurium Fla-strain, ST426 (as a source of an enzyme that cleaves S-adenosylhomocysteine, a product of the reaction and a potent inhibitor of the transferase (9)), and 2.5 nmol of Sadenosyl-~-[methyl-~H]methionine (specific activity, 0.3 Ci/mmol) with methyltransferase in a total volume of 50 pl of 0.10 M potassium phosphate, pH 7.0, at 30 "C. At 5-min intervals, 15-pl aliquots of the assay mixture were placed on 0.3-cm squares of Whatman 3MM paper which were then added to 10% trichloroacetic acid at 25 "C. The paper squares were washed once with 10% trichloroacetic acid, twice with methanol, air-dried, placed in 1.5-ml microfuge tubes, and 0.2 ml of 1.0 M NaOH was added. The tubes were then placed in 10ml liquid scintillation vials with 2.5 ml of Ecoscint scintillation fluid as described previously (9). The vials were incubated overnight at 25 "C, and the amount of [3H]methanol that had diffused into the scintillation fluid was assayed in a Beckman LS-230 liquid scintillation spectrometer. Values were corrected for the less than quantitative transfer of methanol as determined in parallel experiments with standard solutions containing radiolabeled methanol. For each determination of transferase activity, three aliquots were assayed over a 15-min interval, and initial rates were estimated using linear regression analysis. Rates determined separately from time points within a set exhibited less than a 10% standard deviation from the mean.
Purification of the CheR Methyltransferase The following protocol was used to purify the CheR methyltransferase. Unless otherwise indicated, all steps were performed at 0-4 'C.
Step 1: Preparation of Cell-free Extract-E. coli JM109 containing the Salmonella cheR expression vector, pME43, was used as a source for the enzyme. The cells were grown in 6 liters of L broth at 37 "C with vigorous aeration to a density of 2 X lo9 cells/ml, chilled for 5 min at 0 "C, and harvested by centrifugation for 30 min at 10,000 X g. The pellet (11.3 g) was washed by resuspension in 10 ml of phosphate/PMSF buffer (0.10 M potassium phosphate, 1.0 mM EDTA, 1.0 mM 8-mercaptoethanol, 50 mM PMSF, pH 7.0) per g, wet weight, of cells. The suspension was centrifuged for 25 min at 10,000 X g, resuspended in 4 ml of phosphate/PMSF buffer per g, wet weight, of cells and sonicated using a Branson Model 200 Sonifier. The sonicate was centrifuged for 15 min at 16,000 X g, and the supernatant was further clarified by centrifugation for 45 min at 100,000 X g.
Step 2: Ammonium Sulfate Fractionation-The cell-free extract The mutations were introduced to create a NdeI site (CATATG) immediately proximal to cheR. Fortuitously, the change caused a significant increase in cheR expression. This fits the previously reported base preferences at these positions for initiation of translation (45). was brought to 42.3% saturation with respect to ammonium sulfate by adding 29.6 g of finely divided ammonium sulfate/100 ml of solution. During this procedure the pH was maintained constant at 7.0 by addition of ammonium hydroxide. The solution was stirred for 30 min and then centrifuged for 20 min at 14,000 X g. The precipitate was redissolved in 45 ml of phosphate/PMSF buffer, and dialyzed for 4 h against 2 X 4 liters of phosphate/PMSF buffer.
Step 3: DEAE-cellulose Chromatography-The dialyzed ammonium sulfate fraction was diluted 1:lO with Tris buffer (20 mM Tris-HCl, 1.0 mM EDTA, 1.0 mM j3-mercaptoethanol, pH K O ) , and applied to a DEAE-cellulose column (4 X 20 cm) which had been equilibrated in Tris buffer. The column was washed with 460 ml of Tris buffer, and the flow-through, which contained the methyltransferase activity, was brought to 70% saturation with respect to ammonium sulfate, stirred for 30 min, and centrifuged for 25 min at 17,000 X g. The precipitate was dissolved in a minimal volume of phosphate buffer (0.10 M potassium phosphate, 1.0 mM EDTA, 1.0 mM p-mercaptoethanol, pH 7.0).
Step 4: Phenyl-Sephurose Chromatography-The DEAE eluate was brought to 20% saturation with respect to ammonium sulfate before being placed on a phenyl-Sepharose CL-4B column (2 X 15 cm) which had been equilibrated with 25% saturated ammonium sulfate in phosphate buffer. The column was washed with 50 ml of 12.5% saturated ammonium sulfate, 50 ml of phosphate buffer, and 150 ml of 35% ethylene glycol, 50 mM potassium phosphate, 0.5 mM EDTA, 0.5 mM j3-mercaptoethanol, pH 7.0. The methyltransferase was then eluted with 225 ml of 75% ethylene glycol, 25 mM potassium phosphate, 0.25 mM EDTA, 0.25 mM p-mercaptoethanol, pH 7.0. The enzyme solution was dialyzed overnight against 2 X 4 liters of phosphate buffer, brought to 70% saturation with respect to ammonium sulfate, stirred for 30 min, and centrifuged for 25 min at 17,000 X g.
Step 5 : Bio-Gel P-60 Chromatography-The precipitate was dissolved in 10 ml of phosphate buffer and loaded on a 2.8 X 90-cm Bio-Gel P-60 molecular sieve column. The column was eluted with phosphate buffer and fractions with methyltransferase activity were pooled.
Step 6: Hydroxylapatite Chromatography-The protein solution from Step 5 was diluted 1:2 with water, and applied to a 2.0 X 8.5-cm hydroxylapatite column. The column was eluted with a 400-ml linear gradient from 50 to 100 mM potassium phosphate, 1.0 mM EDTA, 1.0 mM j3-mercaptoethano1, pH 7.0, and fractions containing transferase activity were pooled.
Determination of the Nucleotide Sequence of S. typhimurium cheR Gene-The sequence of cheR was determined using a subclone of pMEl (7). A 4.0-kilobase AuaI fragment of pMEl was inserted in both orientations into the AuaI site in the polylinker of M13mpl0, yielding the recombinant phage, Ml3melA and M13melB. Sets of phage with deletions extending processively through the insert were prepared from the two-parent recombinant phage using a modification of the Exonuclease I11 procedure (19, 20). The nucleotide sequence was determined on 100% of both strands by the dideoxynucleotide procedure (17), using DNA polymerase Klenow fragment, [32P]dATP, an M13 universal primer (17-mer), and the deletion phage plus strand DNA as template.

RESULTS
The purification of the CheR methyltransferase was facilitated by its high level production from a genetically engineered cheR expression vector, pME43 (Fig. 1). Since the level of overproduction from this plasmid was approximately 1500fold above wild type (Table I), only about a 10-fold purification of CheR was required to obtain homogeneous protein.
The purification was accomplished in 6 steps (Table I1 and Fig. 2): (i) preparation of the initial cell-free extract; (ii) ammonium sulfate fractionation; (iii) DEAE-cellulose chromatography; (iv) phenyl-Sepharose chromatography; (v) Bio-Gel P-60 chromatography; and (vi) hydroxylapatite chromatography. T h e 13-fold purification that was achieved represents almost a 20,000-fold purification above wild type levels of the enzyme. In the absence of plasmid, the transferase is apparently expressed at very low levels. From these results, the number of CheR molecules per wild type cell was calculated t o be approximately 200.
The purified protein showed only one band when subjected 1. Plasmids used in this study. The S. typhimurium cheR gene is located in the Meche operon which had been cloned into pUC12 to give pMEl (7). A subclone of pMEl was obtained by digestion with NdeI yielding pME4. Further subcloning of cheR was achieved by digestion of pME4 with ChI and ligating the 1.3-kilobase fragment into the AccI site of pUC12 yielding pME5. Another plasmid, pME43, was constructed, essentially from pME5, by changing two bases in front of the cheR initiation methionine codon using sitespecific mutagenesis (see "Experimental Procedures").  47). Cell extracts were prepared and methyltransferase activity was assayed as described under "Experimental Procedures." One unit of methyltransferase activity is that amount of enzyme which will catalyze the formation of 1 nmol of carboxyl methylated membrane protein per min. The values presented are from a single experiment. In three independent determinations of this type the fold overproduction ranged from 1420 to 1550 with an average value of 1480. to SDS-polyacrylamide gel electrophoresis (Fig. 2). The apparent molecular weight of this species was 31,000. A similar value was obtained using gel permeation chromatography under nondenaturing conditions (Fig. 3). The nucleotide sequence of the cheR gene predicts a protein monomer with a molecular weight of 32,900 (see below). A molecular weight of 41,000 has been reported for the enzyme in unfractionated extracts derived from wild type S. typhirnurium (6,8). This relatively high value is presumably due to interactions between the transferase and other components of the extracts. CheR exhibits typical Michaelis-Menten kinetics. As the concentration of AdoMet is increased, the velocity of the reaction approaches a maximum (Fig. 4). Under these condi- was 180 nmol/min/mg enzyme. The kinetics of methyltransferase activity as a function of membrane protein concentration a t nearly saturating AdoMet (Fig. 5)      The nucleotide sequence of the S. typhimurium cheR gene and its flanking regions is shown in Fig. 6 together with the predicted amino acid sequence of the protein. Previous studies have established the approximate location of cheR relative to the other chemotaxis genes (21)(22)(23). Because of this it was possible to tentatively identify an open reading frame corresponding to cheR. The putative gene begins with an ATG translational initiation codon preceded by a possible ribosomal binding site. The latter does not correspond precisely to the Shine-Dalgarno consensus sequence (24, 25). Th' 1s may explain, a t least in part, the relatively low level of expression.

C G C G G A C A C A C G G T G T A T G C G C T M G I A A G G A T A A I \ G C
The nucleotide sequence predicts a protein of 32,900 daltons. This corresponds with the values derived from SDSpolyacrylamide gel electrophoresis and gel permeation chromatography. The amino acid composition of the pure protein was similar to that predicted from the sequence (Table III), and N-terminal analysis showed that the first 24 residues corresponded to the predicted sequence with the first methionine proteolytically cleaved (Table IV). A minor sequence a Levels of amino acids predicted from the nucleotide sequence of the cheR gene given in Fig. 6.
Analysis was performed by Dr. Audree Fowler and Janice Bleibaum at the UCLA Protein Microsequence Laboratory. Samples of purified protein were hydrolyzed in 6 M HCI at 110 "C under vacuum in a Waters Pico-Tag work station. Amino acid derivatives were prepared with phenylisothiocyanate and were quantified by high performance liquid chromatography (with detection at 254 nm) against an amino acid standard using the procedure of Bidlingmeyer was obtained from our preparation of pure transferase protein corresponding to a fragment of CheR resulting from proteolytic cleavage between Ser-3 and Ser-4. These results strongly support the identification of the methyltransferase with the cheR gene.
It has been shown that Salmonella che genes complement  Minor sequence Gln--Ala Picomoles 6 4 19 20 21 22 23 24 Major sequence Leu-Ala-Leu-Ser-Asp-Ala Picomoles 9 8 2 9 3  (20,27,28).  (26), and as expected, the proteins are highly homologous. Of the 5 che genes that have been sequenced in both organisms (20,23,27,28), cheR is by far the least conserved (Table V). A comparison of the amino acid sequence of CheR from Salmonella and E. coli shows 39 differences scattered among a total of 286 residues (Fig. 7). Of these 5 Che proteins in the two species, the transferase is the only one whose length is not conserved. S. typhimurium CheR is longer by 2 amino acids. The lack of conservation of cheR is reflected in a significant difference between the activities of the E. coli and S. typhimurium enzymes (9).

E. coli che mutants and vice versa
In the flanking region just distal to cheR, a base change causes termination of the E. coli gene two codons upstream from the site of termination in S. typhimurium (Fig. 8). In this part of the S. typhimurium sequence, the UGA termination codon of cheR overlaps the AUG initiation codon of the structural gene for the demethylating enzyme, chef?, to give AUGA. It has been suggested that such overlaps may provide a mechanism for translational coupling to insure equivalent expression of two genes (29). One might suppose that the cheR-cheB overlap in Salmonella functions to maintain a balance between the antagonistic methylating and demethylating activities of the two enzymes. The lack of a corresponding overlap in E. coli, and the fact that the transferase is expressed at significantly lower levels than the esterase, argue against this idea.

DISCUSSION
The protein-L-glutamyl methyltransferase (EC 2.1.1.24) from S. typhimurium has been purified and characterized, and its gene, cheR, has been sequenced. This enzyme functions as a 33,000 molecular weight monomer to transfer methyl groups from AdoMet to membrane chemoreceptor proteins.
The K,,, for AdoMet is approximately 17 PM, so that at the concentrations of AdoMet normally present in E. coli or S. typhimurium, approximately 100 NM (30), the enzyme is saturated with this substrate. The enzyme also has a relatively high affinity for its membrane substrate, which in wild type cells is present a t a concentration approximately 5-fold greater than the K, determined with the pure e n~y m e .~ I t has been estimated that there are about 10,000 methylated receptor monomers per bacterial cell, corresponding to a concentration of roughly 20 PM (31). The levels of the enzyme are relatively low, approximately 200 enzyme monomers per cell or about 0.3 mM. It therefore seems likely that essentially all of the transferase in the cell is bound to receptor. From the Vmax of the purified enzyme, one would predict that the maximum possible rate of methylation in vivo would be approximately 2000 methylation events per min. Comparable rates have, in fact, been observed immediately after the addition of saturating concentrations of attractant stimuli (31). The relatively close correlation between these values indicates that in fully stimulated wild type cells the transferase is probably functioning at close to its maximum velocity. This may explain why the methylation reaction functions as the The K , value of 2 mg of protein/ml for the methyl-accepting membranes used in transferase assays reflects a relative concentration of approximately 10 mg of wild type membrane protein/ml since the Tar receptor, comprising about half the total methyl acceptor species, is overexpressed by 10-fold in RP4080pWK3-55.2 membranes. When compared to an estimated concentration of membrane protein in wild type cells of about 50 mg/ml, these data indicate that the concentration of methyl acceptors in wild type cells is approximately 5-fold the apparent K , value obtained in uitro. rate-limiting step for adaptation to large changes in attractant concentration (1,32).
The turnover number of the CheR methyltransferase is approximately 10/min. Why does a bacterial system designed to provide rapid responses to changing extracellular conditions have such a slow enzyme? In considering the chemoreceptor modification reactions it has been assumed that the role of the methyltransferase is essentially equivalent to that of a protein kinase. This seems reasonable until one considers that the metabolic cost of methylation is over 10 times that of phosphorylation; 1 ATP is hydrolyzed per round of protein phosphorylation while approximately 12 ATPs must be hydrolyzed to regenerate AdoMet from S-adenosylhomocysteine (33). It would therefore appear to be inefficient to use methylation in place of phosphorylation in a reversible modification system. In fact, this type of chemistry seems to be unique to the bacterial chemotaxis system. Reversible protein modifications generally involve ATP or cofactors with nearly equivalent metabolic cost to the cell, and most protein methylation reactions are irreversible events associated with protein maturation.
Methylation has been observed in a wide range of proteins in all types of cells (34). A few examples include actin, cytochrome c, rhodopsin, histone proteins, and myelin basic protein in vertebrate tissues; and flagellins, elongation factor Tu, several ribosomal proteins, and pilins in bacteria. Generally the methylation occurs at a side chain nitrogen in arginine, lysine, or histidine, or at an N-terminal CY amino group. None of these modifications appear to be reversed by a corresponding demethylating enzyme. There appear to be two types of protein carboxyl methylations. Whereas methylation at glutamate residues is only known to occur in bacterial chemoreceptors, methylation at abnormal aspartate residues appears to occur in all cells and is not specific for a particular class of proteins (for a review see Ref. 3). Recent evidence suggests that this class of methylation reactions occurs at D-aspartate or isoaspartate groups that arise as proteins age. The resulting methylesters hydrolyze spontaneously by a route which can result in the repair of the D-or iso-residue back to its L-configuration. It is interesting that the turnover numbers for this class of enzymes tend to be even less than that of the CheR methyltransferase (35). In E. coli and S. typhimuriurn each chemoreceptor monomer can eventually be methylated at 4 specific glutamate residues (36,37). The nucleotide sequences of the chemoreceptor genes encode either glqtamate or glutamine at each of these positions (38)(39)(40)(41). The esterase that catalyzes the demethylation reaction functions as an amidase to convert glutamines to glutamates which are then subject to methylation by the transferase (42). The role of amidation and methylesterification has been clarified by studies of the behavior of mutant strains (10,43). Glutamines and methylglutamates appear to have similar effects on receptor activity. Whereas mutants that lack either the methylating or demethylating enzymes are deficient in chemotaxis, mutants lacking both these activities exhibit considerable chemotaxis ability. These and other results strongly support the idea that an intermedipte level of methylation (or amidation) is essential to chemotaxis. In a sense, each receptor monomer contains 4 bits of information in a binary code depending whether a particular glutamate is modified, 1, or not, 0. The genetically encoded residues at each position essentially represent a default value. In the aspartate receptor the four potential sites of methylation correspond to Gln-295, Glu-302, Gln-309, and Glu-491 so the default value is 1 0 1 0. Because of the amidase activity of the demethylating enzyme, transferase mutants have meth-ylation values that approach 0 0 0 0, and because of the transferase, methylesterase mutants have values approaching 1 1 1 1. Transferase/esterase double mutants, however, are fixed at the genetically encoded default value of 1 0 1 0. Since the double mutants retain considerable chemotaxis ability (lo), it is apparently not essential that methylation levels change during chemotaxis, but only that they be maintained in the right range.
Methylation and demethylation at chemoreceptor glutamate residues is the only known instance of a reversible methylation involved in regulating a protein's activity. It appears, however, that the frequency of reversal may be quite low. Once a receptor has attained an optimal level of modification (amidation + methylation) it need not be further adjusted until conditions change significantly. The characteristics of the transferase enzyme reported in the present study make sense in terms of these functional considerations. The relatively low number of enzyme monomers per cell is just sufficient to insure a reasonably equal distribution between daughter cells at each division. The S. typhimurium and E.
coli CheR proteins have diverged more than other components of the chemotaxis apparatus, perhaps this is because there is more latitude in the kinetics of methyltransfer than for other reactions involved in sensory motor regulation.
The transferase appears to provide a mechanism by which cells can reprogram their receptors in response to either different environmental conditions or altered metabolic requirements. The steady state rate of methylation is roughly one-tenth the maximum velocity (1,31). This value corresponds to only about three methylation events per cell per s. If cells experience large changes in their environment, methylation may become rate-limiting. Under these circumstances, swimming is perturbed until a new receptor program is introduced (44). Only after appropriate methylation values have been achieved can chemotaxis proceed.