Sites of covalent modification in Trg, a sensory transducer of Escherichia coli.

The Trg protein mediates chemotactic response of Escherichia coli to the attractants ribose and galactose. Like other transducers, Trg is a transmembrane protein that undergoes post-translational covalent modification. The modifications are hydrolysis (deamidation) of certain glutamine side chains to create glutamate residues and methylation of specific glutamates to form carboxyl methyl esters. Analysis of radiolabeled, tryptic peptides by high performance liquid chromatography and gas-phase sequencing allowed direct identification of the modified residues of Trg. The protein has 5 methyl-accepting residues. Four, at positions 304, 310, 311, and 318, are contained in a 23-residue tryptic peptide ending in lysine. The fifth, at position 500, is within a 25-residue tryptic peptide ending in arginine. At two sites, 311 and 318, glutamines are deamidated to create methyl-accepting glutamates. There is not a required order of modification among the sites. However, there is a substantial preference for methylation on the arginine peptide and, among sites on the lysine peptide, for the middle pair. Comparison of sequences surrounding modified residues identified in this work for Trg and previously for Tsr and Tar suggests a consensus sequence for methyl-accepting sites of Ala/Ser-Xaa-Xaa-Glu-Glu*-Xaa-Ala/OH-Ala-OH/Ala, where OH signifies Ser or Thr and the asterick marks the site of modification.

Chemotaxis is a behavioral response of an organism to changes in the chemical environment, The chemotactic behavior of the closely related bacteria Escherichia coli and Salmonella typhimurium has been investigated extensively (see articles in Eisenbach and Balaban, 1985). In the absence of a chemical gradient, these organisms move through their environment in a random walk, consisting of smooth swimming (corresponding to counter-clockwise rotation of flagella) punctuated by uncoordinated tumbles (corresponding to clockwise rotation) that result in random reorientation of the direction of movement. An increase in attractant concentration or decrease in repellent concentration suppresses tum-* A system for HPLC used in this work was purchased with Biological Instrumentation Grant PCM-8212471 from the National Science Foundation. A gas phase sequenator used in later stages of this work was purchased with Shared Instrumentation Grant RR02677 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$Recipient of Grant GM29963 from the National Institutes of Health, Grant DMB-8416274 from the National Science Foundation, a McKnight Neuroscience Development Award, and an American Cancer Society Faculty Research Award. bles, while either a decrease in attractant or an increase in repellent concentration enhances the tumble frequency. The result of this behavior is net progress of the bacterium up a concentration gradient of attractant or down a gradient of repellent.
There are four chemotactic transducers present in Escherichia coli, Tsr, Tar, Trg, and Tap (see Simon et al., 1985 for a recent review). These integral membrane proteins each bind specific ligands. Trg mediates tactic response to galactose and ribose by recognition of ligand-occupied galactose-and ribosebinding proteins, respectively. Changes in occupancy of ligand-binding sites initiate an excitatory signal, resulting in an alteration of the rotational bias of the flagellar motor. An adaptational process reestablishes rotational bias and thus the pattern of swimming to the pre-stimulus state. Adaptation to a positive stimulus is linked to methylation of specific glutamyl residues by a cytoplasmic methyltransferase which is encoded by the c h R gene. Adaptation to a negative stimulus is characterized by demethylation of these esters, as well as hydrolysis of specific glutaminyl residues (deamidation) and is mediated by a cheB-encoded methylesterase. Transducers contain multiple sites for methylation (Boyd and Simon, 1980;Chelsky and Dahlquist, 1980;DeFranco and Koshland, 1980;Engstrom and Hazelbauer, 1980) and deamidation (Rollins and Dahlquist, 1981;Sherris and Parkinson, 1981;Harayama et al., 1982). These sites have been characterized in detail for Tsr and Tar Dahlquist, 1982a, 1982b;Kehry et aL, 1983a;Terwilliger and Koshland, 1984;Terwilliger et al., 1986aTerwilliger et al., , 1986b. A preliminary study of covalent modification of Trg (Kehry et al., 1983b) indicated that the nature of the tryptic peptides containing sites of modification was similar to those derived from Tsr and Tar. However, the patterns of [3H]methyl-labeled tryptic peptides derived from Trg was considerably more complex than the corresponding peptide maps of Tsr or Tar. In this report, we present a detailed analysis and identification of the sites of covalent modification in the Trg protein.

EXPERIMENTAL PROCEDURES
Bacterial Strains-All strains used are derivatives of E. coli K12.

Chemicals-~-[methyl-~H]Methionine
Preparation of Radiolabeled Trg-Labeling in vivo using L-[methyl-'Hlmethionine was as described (Engstrom and Hazelbauer, 1980). Cells were grown in the presence of 0.4% ribose to insure full induction of ribose-binding protein and the labeling procedure involved stimulation with 10 mM /3-D-allOSe to induce the adaptational increase in methylation of Trg. Preparation of labeled protein for analysis of tryptic peptides by HPLC was done by adding 0.5-1 mCi of 12 Ci/ mmol of ~-[methyl-'H]methionine to 1.25 X lo9 cells in 5 ml to yield a final methionine concentration of 8.5-17 pM. For labeled material with high specific activity, methionine a t 26 Ci/mmol was added to yield a final concentration of 4 pM. Minicells were prepared and radiolabeled as described (Kehry et al., 1983b). Preparation of Trg protein containing a specific radiolabeled amino acid involved adding 40 pCi of ['Hlarginine or 50 pCi of [%]methionine to 50 pl of minicells, harboring a trg-containing plasmid, at an optical density equivalent to 2 X log cells/ml (95 pg of protein in 50 pl). SDSpolyacrylamide gel electrophoresis was performed in conditions optimized for resolution of the electrophoretic forms of Trg (Kehry et al., 1983b). For each 4-mm lane, approximately 40 pg of whole cell protein (1.25 X loR cells) or 95 pg of minicell protein was applied. Usually gels were poured with one, very wide sample lane. Following electrophoresis, gels were stained for 10 min (0.04% Coomassie Brilliant Blue, 25% isopropyl alcohol, 10% acetic acid), destained for 15 min (10% acetic acid) and, for 'H-labeled material, treated with Amplify (Amersham Corp.), then dried a t 60 "C. The positions of radiolabeled proteins were identified by autoradiography or fluorography using Kodak X-AR film.
Analysis of Peptides by HPLC-Strips of the dried polyacrylamide gel containing specific electrophoretic forms of Trg were excised, immersed in 0.2 M N-ethylmorpholine acetate (pH 7.5), digested with diphenylcarbamyl chloride-treated trypsin (Kehry and Dahlquist, 1982a), passed through a 0.45-pm filter (Rainin) to remove particulate matter, and stored a t -20 'C. The trypsin-digested material was applied to a Brownlee Aquapore 300 A, CIS, 10-pm column (0.46 X 25 cm) fitted to a Beckman Model 322 Liquid Chromatograph equipped with a Reodyne injector (2-ml injection loop), a 3-cm guard column, and a Bio-Rad column heater. Analyses were done a t 30 "C. All solvents were filtered through a 0.45-pm membrane (Rainin). Peptides were eluted with a gradient of acetonitrile in 35 mM sodium phosphate buffer (pH 2.2) a t a flow rate of 1 ml/min. The concentration of acetonitrile increased linearly from 0 to 25% over 45 min and then from 25% to 27% over the next 50 min. Fractions of 0.5 ml were collected directly into scintillation vials and scintillation fluid added subsequently. Samples destined for further analysis were collected in vials containing 0.2 ml of N-ethylmorpholine acetate (pH 7.5) so that the pH was brought to 6.8.
Base Hydrolysis-Treatment of peptides to hydrolyze methyl esters was done with a slight modification of the procedure described by Kehry and Dahlquist (1982a). The sample was brought to a pH of approximately 10.5 by gradual addition of 1 M NaOH and then incubated 0.5-5 min at 60 "C. Hydrolysis was terminated by the addition of 5 pl of 50% H3POI.
Chemical Methylation of Methionine-Methylation of methionine residues in vitro with iodomethane was carried out as described by Jones et al. (1976) but with an excess of iodomethane. Isolated tryptic peptides were lyophilized and suspended in 0.4 ml of 0.1 M sodium phosphate buffer (pH 4.5), the pH adjusted to 4.5 with 50% H'PO,, and then 1 pl of iodomethane added. The reaction was terminated by addition of free methionine to a concentration equal to that of iodomethane. Prior to chromatography of the product, 0.2 ml of Nethylmorpholine acetate was added to the sample to solubilize any unreacted peptide.
Gas Phose Sequencing-Isolated, methyl-labeled tryptic peptides were applied to the same column used for the initial separation and eluted with a gradient of 0-40% acetonitrile in 0.01% trifluoroacetic acid. Isolated samples were lyophilized, suspended in 20 pl of 70% aqueous acetonitrile, and applied to a glass-fiber filter impregnated with pretreated Polybrene. Sequence analysis was performed using an Applied Biosystems gas-phase sequenator. Initially, conditions for gas-phase sequencing were modified by lowering the temperature from 45 to 40 "C and eliminating the conversion step, with the goal I The abbreviations used are: HPLC, high performance liquid chromatography; SDS, sodium dodecyl sulfate. of decreasing the probability of methyl ester hydrolysis during sequencing. However, later sequence analysis under standard conditions had improved initial yield and coupling efficiency.

RESULTS
Modifications of Trg-The pattern of covalent modifications of the Trg protein was studied by analysis of radiolabeled, tryptic peptides using HPLC. Peptides that include sites of modification on Trg are eluted from a reversed-phase column much later in a gradient of acetonitrile than most tryptic peptides of the protein and thus are well resolved in the tryptic map generated by HPLC (Kehry et al., 1983b). In the analyses reported here, the shape of the eluting gradient of acetonitrile was designed to compress the early region of the map where many peptides appear and to expand the later region in which peptides carrying modification sites are found (see "Experimental Procedures"). Thus fraction numbers and spacings between peptides do not correspond to those in the previous study (Kehry et al., 1983b). Sufficient Trg protein for analysis was obtained by using cells and minicells harboring multicopy plasmids carrying the trg gene. Methyl-accepting sites were examined using Trg protein that had acquired ['HJmethyl groups in intact, viable cells. Analyses of deamidation and of effects of amino acid substitutions on properties of protein or peptides were performed using Trg protein that had incorporated radiolabeled amino acids during synthesis in intact minicells. Radiochemically pure protein was obtained by excising appropriate regions of SDS-polyacrylamide gels.
Modifications of Trg catalyzed by the CheR and CheB enzymes are reflected in alterations in the mobility of the transducer protein in SDS-polyacrylamide gel electrophoresis (Harayama et al., 1982;Kehry et al., 1983b). Trg synthesized in strains lacking both modification enzymes appears on gels as a single band with an apparent molecular weight of approximately 60,000. An active CheB enzyme results in three slower migrating forms of Trg that have undergone one or two deamidations. Active CheR and CheB enzymes combine to generate seven ['Hlmethyl-labeled, electrophoretic forms of Trg, the four described above and three forms that migrate more rapidly than unmodified Trg (Fig. 1). The forms are numbered in order of increasing mobility, with band 4 at the position of unmodified polypeptide; band 7 is often hardly detectable.

Identity of the KI Peptide-A methionine-and lysine-
containing tryptic peptide, termed K1, that contained two sites of CheB-dependent deamidation had been identified by analysis of peptides from the four electrophoretic forms of Trg observed in cheR-cheB+ strains (Kehry et al., 1983b). In The figure is a fluorograph of an SDS-polyacrylamide gel, optimized for resolution of Trg, containing a sample of ['H]methyl-labeled protein from CP362, which has deletions in the chromosomal copies of tsr, tar, tap, and trg, but harbors the multicopy plasmid, pMG2, that contains the wild type t g gene. Only a segment of the gel, including the Trg bands is shown. The bands are numbered in order of increasing electrophoretic mobility. [35S]methionine during synthesis in minicells derived from a cheR cheB strain and harboring pMG2 or its derivatives carrying a mutated trg. Protein bands excised from SDSpolyacrylamide gels were digested with trypsin and the peptides separated by HPLC. The bottom trace is the complete pattern of [%SI methionine-labeled peptides derived from wild type protein. The peak corresponding to unmodified K1 peptide is labeled (KO) and arrows mark the positions of once-deamidated (K,) and twice-deamidated (K2) forms of the K1 peptide. The two insets show the regions containing a difference from the map of wild trpe protein in the maps of mutant proteins having alanine in place of Gln3'l (top) or G W 8 (middle). Shifted peaks corresponding to unmodified K1 peptide from the mutant proteins are labeled KO and the position of the corresponding wild type peptide is indicated by an arrow. Fig. 2, the elution positions in the modified gradients used in these studies are indicated for the unmodified (Klo), oncedeamidated (Kl,) and twice-deamidated (Kl,) forms of the K1 peptide. The identity of the K1 peptide was established by analyses of proteins containing specific amino acid substitutions that had been generated by oligonucleotide-directed mutagenesis of trg.' Substitutions at residues 311 or 318 shifted the position of the unmodified K1 peptide in the chromatographic pattern (Fig. Z), indicating that the K1 peptide must include those residues and thus, from the deduced amino acid sequence of Trg (Bollinger et al., 1984), must be the 23-residue tryptic peptide from Thr3'' through Lys"' (see below). This peptide corresponds to homologous sequences in Tsr and Tar that include sites of deamidation and methylation (Kehry et al., 1983a;Terwilliger and Koshland, 1984).
PHIMethyl-labeled Tryptic Peptides-As reported previously (Kehry et al., 1983b), analysis by HPLC reveals many peaks of radioactivity after trypsin digestion of [3H]methyllabeled Trg. Fig. 3 provides a representative sample of many analyses. Nine different peaks were identified and were labeled T1 to T 9 in order of elution? Only two peaks of radioactivity, T4 and T7, appeared in chromatographic patterns of digests from electrophoretic forms of Trg in bands 1 through 3. Peak T7 occurred in patterns derived from every electrophoretic form of Trg. It was always the predominant peak in digests of band 1 and was present at relatively low levels in digests of bands 2 and 3. Peaks T1, T2, T5, T6, T8, to T9 in the preliminary study (Kehry et al., 198313). and T9 were observed only in digests of bands 4 through 7.
The T3 peak occurred infrequently and in a strain-dependent fashion. A combination of analytical procedures established that the nine peaks represent multiple forms, differing by the extent of covalent modification, of only two distinct tryptic peptides of Trg (see following sections).
Demethylation Studies-The number of methyl esters on a [3H]methyl-labeled peptide can be determined by analysis with HPLC of the products derived from relatively mild treatment with base (Kehry and Dahlquist, 1982a). Upon hydrolysis of methyl esters that is substantial but not complete, a peptide that originally contained n radiolabeled methyl groups becomes a series of radiolabeled derivatives, containing from 1 to n methyl groups. Each derivative will elute at a characteristic position in HPLC; the loss of each methyl group resulting in earlier elution (Kehry and Dahlquist, 1982a). A monomethylated peptide will not give rise to any other [3H]methyl-labeled species, while a dimethylated peptide will yield a labeled, monomethylated form of the same peptide. Examples of patterns observed in demethylation studies are shown in Fig. 4, and a summary of the data obtained is presented in Table I.
Identification of the Rl Peptide-The data in Table I indicated that the peptide eluting in peak 7 was not a monomethylated form of any multiple methylated peptide and suggested that the peptide represented a distinct fragment of Trg. The distinct nature of the peptide was established by analysis with HPLC of [3H]methyl-labeled peptides that had been treated to modify methionine side chains. In acidic conditions, iodomethane will add a methyl group to the sulfur of methionine to form a cationic sulfonium salt (Jones et al., 1976). Introduction of a positive charge would be expected to reduce the affinity of the altered peptide for a hydrophobic resin and thus shift elution to an earlier position in the gradient of acetonitrile. Earlier eluting forms of peptides in peaks T4 (Fig. 5A) and T6 (data not shown) were generated after

Effects of treatment with iodomethane on [9H]
methyl-labeled peptides. Peptides in specific radioactive peaks from tryptic maps like those shown in Fig. 3 were treated with iodomethane, resulting in formation of cationic sulfonium ions on methionines that might be present. Treated samples were analyzed in the same chromatographic conditions used to generate the original peptide maps. The arrow in each panel marks the position of the particular peak of radioactivity before treatment with iodomethane. treatment with iodomethane, consistent with the two peptides being methylated forms of the methionine-containing K l peptide. In contrast, treatment with iodomethane had no effect on the position of [3H]methyl-labeled T7 (Fig. 5B), indicating that the peptide did not contain methionine and thus represented a fragment of Trg distinct from the K1 peptide. The T7 peak was identified as the monomethylated form of a specific arginine-containingpeptide by analysis with HPLC of tryptic peptides derived from Trg proteins containing particular amino acid substitutions created by oligonucleotide-directed mutagenesis.' Substitution of aspartate or of glutamine for glutamate at residue 5104 resulted in a shift in the elution position of a [3H]arginine-labeled peptide (Fig. 6) and a corresponding shift in the position of [3H]methyllabeled T7 (data not shown). The arginine-containing peptide is not labeled by [35S]methionine. The deduced amino acid sequence of Trg4 (Bollinger et al., 1984) indicates that the arginine-containing tryptic peptide that includes residue 510 consists of the 25 amino acids between Val4w and Arg514 and contains no methionine. This peptide is almost identical in sequence and position in the protein to the arginine-containing, methyl-accepting R1 peptides of Tsr and Tar (Kehry and In the published determination of the sequence for trg (Bollinger et al., 1984), three nucleotides in the region of position 1185 were not detected. The correct sequence, determined by high resolution analwith BssHII has the codon GCG after the codon CGA at position 395.

ysis of a region of compression in the sequencing gels and by analysis
This results in the insertion of alanine in the deduced amino acid sequence of Trg at position 396, and a shift by one number in each of the subsequent residues. Thus G I U~'~ was numbered 509 in Bollinger et al. (1984).  Dahlquist, 1982a;Terwilliger and Koshland, 1984) and thus will be termed the R1 peptide of Trg.
In the preliminary description of modifications of Trg (Kehry et al., 1983b)

, differences between tryptic maps of [3H]
arginine-labeled Trg from band 1 and band 4 suggested the possibility of a CheB-mediated modification on one or more arginine-containing peptides of Trg. This possibility has not been supported by subsequent investigations. Extensive characterization of patterns of tryptic peptides from Trg synthesized in the presence or absence of active CheB has provided no evidence for CheB-mediated modification of R1 or of any other arginine-containing peptide eluting in the vicinity of the K1 and R1 peptides.
Analysis of Peptides by Gas-Phase Sequencing-Selected [3H]methyl-labeled peptides were analyzed using a gas phase sequenator to determine which positions, within the peptide sequence, contained radiolabeled methyl esters. Peptides were purified by HPLC using the same conditions as for analytical experiments. Pooled material from several preparations was rechromatographed using a different solvent system to insure a pure, homogenous preparation of a particular peptide form and to place the peptide in a more volatile solvent. Material released at each cycle of the sequencing procedure was analyzed for radioactivity, not for the identity of the released amino acid. The monomethylated form of the R1 peptide, present in the T7 peak was analyzed by this procedure. A representative pattern is shown in Fig. 7A. The great majority of the radiolabel was released in the cycle corresponding to G I U~~, thus identifying that residue as the single methylaccepting site on the R1 peptide. Approximately 20% of the residue corresponding to a particular cycle is released in the subsequent cycle, thus accounting for radioactivity in the two cycles following G1uSW. Radioactivity in cycle 6 may be due to degraded R1 peptides that have lost the first 5 residues from the amino-terminal by spontaneous cleavage after the asparagine residue. For such degraded peptides, radiolabel from Glu5'"' would be released in cycle 6. Release of rachoactivity in r , ,

T E E O ' A ' A ' A ' I ' E ' O ' T ' A ' A ' S ' M ' E ' O ' L ' ? ' A ' T ' V ' K '
310 320 amino acid

FIG. 7. Release of radioactivity from [SH]methyl-labeled tryptic peptides in cycles of amino acid sequencing reactions.
Tryptic, [3H]methyl-labeled peptides were purified by HPLC as described under "Experimental Procedures" and submitted to analysis with a gas-phase polypeptide sequenator. The quantity of radioactivity released (minus background and corrected for an estimated coupling efficiency of 92%) at each cycle of cleavage is plotted versus the residue position, labeled with the deduced amino acid sequence of the appropriate peptide (see text). Residue numbers corresponding to the entire Trg sequence are indicated. Results are shown for monomethylated, R1 peptide in peak T7 (A), monomethylated, twice-deamidated, K1 peptides in peak T4 ( B ) , dimethylated, once-deamidated K1 peptides in peak T5 (C), and monomethylated, not-deamidated K1 peptide in peak T1 (D).
cycle 5 of sequencing of K1 peptides (Fig. 7, B and C) is consistent with cleavage after Gln305.
The T4 peak consists of a monomethylated (Table I) methionine-containing (Fig. 5) peptide with elution characteristics consistent with identity as a monomethylated, twicedeamidated K1 peptide. Studies of amino acid substitutions within the K1 sequence provide direct evidence that the T4 peak contains forms of the K1 peptide? Analysis of [3H] methyl-labeled T4 material with the sequenator revealed four methyl-accepting sites, corresponding to Glu304, Glu3", Gln3", and Gln318 (Fig. 7B). This pattern implies that two sites are generated by deamidation of glutamines to produce methylaccepting glutamates. The peptides found in the chromatographic peak T4 are all monomethylated as determined by demethylation studies (Table I). Thus the observation of four different methylated sites among the population of molecules indicates that any of the four sites can be the sole methylated site on the K1 peptide and also that the chromatographic system does not resolve monomethylated peptides modified at different sites along the sequence. The predominance of radiolabel at positions 310 and 311 implies that, for the cells used, the steady state levels of methylation are not equal for the four sites. There is a strong preference, in these in vivo conditions, for a single methylation on K1 to be at the central pair of sites.

Sites of Covalent Modification in a Sensory Transducer
The T5 peak consists of dimethylated peptide (Table I) with elution characteristics consistent with identity as oncedeamidated, dimethylated K1 peptide. Analysis with the sequenator revealed that twice-methylated K l peptide contained methyl esters at the same sites utilized in the array of monomethylated species (Fig. 7C). It appears that the first deamidation can occur at either site and that most oncedeamidated, dimethylated K1 peptides are methylated a t position 310.
Sufficient radiolabeled material for sequencing anaIysis of the T1 peak was obtained by use of a mutant protein, Trg-31, that is essentially not deamidated, but is methylated. The protein differs from the wild-type species by a single amino acid residue that is substantially distant from both the K1 and R1 regions of the protein sequence (Park and Hazelbauer, 1986b). Tryptic digestion of [3H]methyl-labeled mutant protein yielded essentially only two radiolabeled, chromatographic species, T7, the monomethylated form of R1, and T1, apparently the monomethylated, not deamidated form of K1. Analysis with the gas phase sequenator revealed that only a single site, GIu304, is methylated in the peptides in the T1 peak (Fig. 7D).

DISCUSSION
Sites of Modification-The data presented here identify five methyl-accepting sites in the Trg transducer protein (Fig. 8). Four sites,residues 304,310,311,and 318, are in the central region of the protein, included in the K1 tryptic peptide. The fifth site is near the carboxyl-terminal at amino acid 500, within the R l tryptic peptide. The sites will be referred to as 1-5 in order of residue number. Sites 3 and 4 correspond to glutamines that must be hydrolyzed to create glutamates before functioning as methyl-accepting residues. Deamidation  Krikos et al., 1983;Russo and Koshland, 1983;Bollinger et al., 1984) are shown for tryptic peptides containing identified sites of modification for Trg (this work), Tsr Dahlquist, 1982a, 1982b;Kehry et al., 1983a), andTar (Tenvilliger andKoshland, 1984;Terwilliger et al., 1986). Methyl-accepting residues are starred. The bracketed starred positions in the Tsr sequence mark candidates for the second methyl-accepting site in the R1 peptide. Starred residues corresponding to glutamines in the deduced sequences are hydrolyzed to glutamates before functioning as methyl-accepting sites. Residue numbers corresponding to the entire Trg sequence are indicated. a t these two sites corresponds precisely to the number of CheB-dependent modifications previously determined for the K1 peptide (Kehry et al., 1983b). As shown in Fig. 8, the pattern of covalent modifications that occur on Trg is quite similar to the patterns documented for Tsr Dahlquist, 1982a, 1982b;Kehry et al., 1983a) and Tar (Terwilliger and Koshland, 1984;Terwilliger et al., 1986a, 198615). The identity of the amino acids in positions near a methylaccepting residue is substantially conserved among the 13 identified sites (five in Trg and four each in Tsr and Tar). Alanine, serine, and threonine are the only amino acids found at position 4 on the amino side of the modified residue and at positions 2, 3, and 4 on the carboxyl side. The conserved positions can be expressed as a concensus sequence of Ala/ Ser-Xaa-Xaa-Glu-Glu*-Xaa-Ala/OH-Ala-OH/Ala, where the asterisk marks the site of modification and the symbol "OH" signifies serine or threonine. If the sequence were contained in an a-helix, then the conserved residues would form a pocket surrounding the methyl-accepting site. The only deviations from the consensus sequence are site 1 in Tar, in which the conserved alanine at carboxyl position 3 is replaced by a serine, and site 4 in Tar and site 2 in Trg, in which the conserved glutamate is replaced by a glutamine and an isoleucine, respectively. The deviating sites in Tar exhibit substantially reduced rates of methylation and demethylation as well as reduced levels of steady state methylation (Terwilliger et al., 1986a). In contrast, there is substantial methylation of the nonconforming site in Trg (Fig. 7). There must also be a methylation site on Tsr that deviates from the concensus sequence. Kehry and Dahlquist (1982a) observed a tetramethylated form of the K1 peptide of Tsr. Thus there is an unidentified, methyl-accepting site on the peptide, in addition to the three shown in Fig. 8. None of the three available glutamates are preceded by another glutamate, thus whatever the position of the fourth methyl-accepting site, it will lack the Glu of the concensus sequence. It seems reasonable that, in analogy with Trg, the fourth site on Tsr is the glutamate corresponding to Trg residue 310.
In the absence of deamidation, only one of the four methylaccepting sites on the K1 peptide is active (Fig. 70). The same observation has been made for Tsr (Kehry and Dahlquist, 1982b) and Tar (Terwilliger and Koshland, 1984), but explanations of the phenomenon were complicated by indications that only a single deamidation occurs on the K1 peptide of those two transducers (Kehry and Dahlquist, 1983b;Terwilliger and Koshland, 1984) and by incorrect identification of methyl-accepting site 1 on Tar (Terwilliger et al., 1986a). For Trg, it is clear that, in the absence of deamidation, the single, active methyl-accepting residue in K1 is Glu304 and that two sites, Gln311 and Gln318, are inactive because neither is deamidated. Among the inactive sites, only Glu310 would be capable of forming a methyl ester, yet is not methylated. It seems reasonable that lack of an adjacent glutamate a t position 311 inactivates the methyl-accepting capacity of GIu3l", perhaps by disrupting productive interaction with the methyltransferase.
Preference among Sites-There is not a strict order in which the four sites on the K1 peptide are methylated and demethylated since among monomethylated peptides, any of the four sites can carry the single methyl group (Fig. 7B). However, in the stimulated and adapted cells used in these studies, there is a substantial preference for the central pair, sites 2 and 3. A similar preference is observed for Tar; in stimulated and adapted cells, the glutamate corresponding to site 3 of Trg exhibits relative rates of methylation and demethylation that maintain the highest level of modification among the four sites (Terwilliger et al., 1986a). Preference for site 2 of Trg would not be predicted from comparison with the concensus sequence, since the conserved glutamate at the adjacent amino position is replaced by an isoleucine. Apparently, activity of a methyl-accepting site is very dependent on the presence of a Glu-Glu pair (see previous paragraph) but the second Glu can follow as well as precede the modified residue.
Peak T7, containing monomethylated R1 peptide, occurred in the tryptic peptide maps of all electrophoretic forms of Trg and generally contained more radioactivity than any other peak (Fig. 3). This suggests that GluSW is a preferred site of methylation, to an extent that exceeds any of the sites on the K1 peptide. A similar preference for methylation on the R l peptide was observed for Tsr (Kehry and Dahlquist, 1982a), but not for Tar (Terwilliger et at., 1986a), where a glutamine in place of the conserved, adjacent glutamate may drastically lower the efficiency of methylation at the residue corresponding to Glu5''.
Complexity of Tryptic Maps of PHIMethyGlabebd Trg-The maps of [3H]methyl-labeled tryptic peptides derived from Trg contain more peptide forms than the maps of Tsr or Tar Dahlquist, 1982a, 1982b). The analyses reported here revealed that the additional peaks occurred because methylated forms of the K1 peptide lacking one or both deamidations appeared as distinct species. The distinction may in part reflect the different origins of the transducer proteins analyzed. The Trg protein was from cells containing multiple, plasmid-carried copies of the transducer gene and single chromosomal copies of the genes for the modification enzymes, while Tsr and Tar were obtained from cells with single chromosomal copies of the genes for the particular transducer and for the enzymes. A relatively high ratio of transducer molecules to CheB deamidase could result in a substantial population of incompletely deamidated Trg proteins. The altered stoichiometry is likely to account for the presence of undeaminated, monomethylated K1 peptide (peak T1) in maps of some electrophoretic forms of Trg (Fig. 3). The corresponding form of K1 from Tsr or Tar was observed only in protein synthesized in cells lacking CheB activity (Kehry and Dahlquist, 1982b). An additional factor in the difference between the peptide maps of Trg and the other transducers may be the lack of resolution of once-deamidated and twice-deamidated forms of K1 peptides from Tsr and Tar. This possibility is consistent with resolution by HPLC of only a single deamidated form of K1 peptide from Tsr or Tar Dahlquist, 1982a, 1982b;Terwilliger and Koshland, 1984) even though the number of methyl groups found on maximally methylated forms of K1 implies that K1 peptide from both transducers can be deamidated twice (Kehry et at., 1983b;Terwilliger et al., 1986a). Differential resolution by HPLC of deamidated forms of Trg but not of Tsr or Tar could be related to the difference in placement of one of the two deamidated glutamines in Trg (Fig. 8). In any case, definition of the specific sites of covalent modification on the Trg transducer provides a basis for structural investigations of the functional significance of multiple covalent modifications of sensory proteins.