Identification of a Possible Nucleotide Binding Site in Chew, a Protein Required for Sensory Transduction in Bacterial Chemotaxis*

Chew is an essential component of the system which mediates chemotaxis in Salmonella typhimurium and Escherichia coli. Here we report the nucleotide se- quence of the chew gene as well as the purification and characterization of the Chew protein. The DNA sequence predicts a protein of 18,000 molecular weight. The pure protein exhibits an apparent molecular weight of 18,000 during sodium dodecyl sulfate-poly-acrylamide gel electrophoresis. Molecular sieve chro- matography under nondenaturing conditions indicates a molecular weight of approximately 36,000, however. This result suggests that Chew is a homodimer. The predicted amino acid sequence between Thr-128 and Asp-160 fits a consensus exhibited by many proteins which bind purine nucleotides.

nucleotide and suggests that Chew may be the locus for the nucleotide requirement in chemotaxis.

EXPERIMENTAL PROCEDURES
Subclones of the S. typhimurium-pUC12 hybrid plasmid, pME1,' were used to sequence the chew gene and to overproduce the Chew protein. This plasmid ( Fig. 1) contains an 8.4-kilobase genomic S. typhimurium PstI fragment which extends from cheA to fhM. For sequencing, M13 subclones of pMEl were constructed by inserting a 3.1-kilobase PstI-SmaI fragment modified with BamHI linkers in both orientations into the BamHI site in the polyliiker region of M13mplO. Processive deletions through the inserts were generated using exonuclease 111 as described elsewhere.' The complete nucleotide sequence of chew was determined on both strands by the dideoxynucleotide method (8), using [cP~*P]~ATP, DNA polymerase I Klenow fragment and the deleted phage plus strands as templates.
The S. typhimurium Chew protein was purified from an E. coli Fla-strain, HBlOl (9), containing a Chew expression vector, pME105. The latter was constructed by inserting a 0.95-kilobase DraI-Sal1 fragment of pMEl into the SmaI and Sal1 sites in the polylinker region of pUC12. The cells were grown at 37 "C in L broth (9) supplemented with 40 pg/ml ampicillin to a density of lo9 cells/ ml, harvested by centrifugation, suspended in 0.10 M sodium phosphate, pH 7.0 (3 ml/g, wet weight), and lysed with a Heat Systems W-225 Sonicator. The resulting cell-free extract was centrifuged at 100,000 X g for 60 min to remove membranes, ammonium sulfate was added to 20% saturation, the precipitate was removed by centrifugation, and the ammonium sulfate concentration was increased to 40% saturation. The 20-40% ammonium sulfate precipitate was collected by centrifugation, dissolved in and dialyzed against 10 mM piperazine HCl, pH 6.0, in 100 mM NaCl, and applied to a 1.4 X 17-cm column of DE52 (Whatman) equilibrated in piperazine NaCl buffer. The column was washed first with 100 ml of the same buffer, and then with a 150-ml linear gradient of 100-250 mM NaCl. Fractions containing Chew were pooled and subjected to high performance liquid chromatography with a 0.75 X 30-cm TSK G2000SW column (LKB) equilibrated with 0.10 M sodium phosphate, pH 6.5. Fractions containing Chew were pooled and concentrated by precipitation in 50% ammonium sulfate, and the precipitate was collected by centrifugation and dissolved in and dialyzed against 0.10 M sodium phosphate, pH 6.5. All steps were performed at 0-4 "C. Protein was assayed by the method of Lowry (10). Chew was qualitatively assayed as a Coomassie Blue staining band of protein migrating with an apparent molecular weight of 18,000 during sodium dodecyl sulfate-polyacrylamide gel electrophoresis (11).

RESULTS
The nucleotide sequence of the S. typhimurium che W gene and flanking regions, together with the deduced amino acid sequence of the Chew protein, is shown in Fig. 2. The predicted product is highly homologous to the corresponding protein in E. coli (12) with only 11 conservative changes among 167 residues. Five of the 11 differences occur between Val-73 and Glu-81, while the remainder are relatively scattered. A Gly-X-Gly-X-X-Gly sequence characteristic of nucleotide binding proteins (13,14) is located between residues 133 and 138. In addition, the sequences flanking this glycinerich region (Thr-128 to Asp-160) fit the consensus described by Wierenga et al. (14) as predictive of the @-a+ motif which is characteristic of these proteins (Fig. 3). The only significant deviation occurs in the length of the predicted loop connecting the a helix to the second &strand. In known structures this loop varies from 2 to 4 residues, while in Chew it would have a length of 6 residues. There is no reason to expect that such an extended loop does not exist, however; and, in fact, it seems likely that NADH dehydrogenase has a loop size of 7 in this region (14, 15). Aside from this difference, Chew matches the Wierenga consensus a t 10 of 11 specified positions. From a search of over 2500 known sequences it has been estimated that the probability of a random protein fitting the consensus this closely is less than 1 in lo6 (14).
As a first step toward investigating Chew function, an expression vector was constructed, and procedures were developed to purify the Chew protein (Fig. 4). N-terminal analysis was performed to establish that the purified protein was in fact the product of chew. The results (Fig. 2) show that, aside from cleavage of the N-terminal methionine, the initial 13 residues of the purified protein correspond to those predicted from the sequence. Chew eluted with an apparent native molecular weight of approximately 35,000 during molecular sieve chromatography (Fig. 5). Since the pure protein exhibited a molecular weight of 18,000 under denaturing conditions and a monomer molecular weight of 18,000 is predicted from the nucleotide sequence of the chew gene, it appears that the purified protein is a homodimer.

DISCUSSION
Over the past few years there has been a rapid accumulation of information relating protein sequences to characteristic features of tertiary structure. Perhaps the best example of this is the @ -w @ motif associated with nucleotide binding  Pooled fractions collected during DE52 chromatography were subjected to high performance liquid chromatography using a TSK G2000SW column (LKB) as described under "Experimental Procedures." Protein was monitored by absorbance at 214 nm. The major peak runs as a single protein species during sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Fig. 4, lane 6 ) and has an N-terminal sequence corresponding to the predicted product of the chew gene (Fig. 2). The column was calibrated with chicken egg albumin (A), a-chymotrypsinogen (B), and cytochrome c (C).

laOp
domains of proteins which interact with purine nucleotides such as NAD, FAD, ATP, and GTP (13,14,16). It is possible to identify this structure within a sequence and thereby discern a probable site of nucleotide interaction. Proteins of this type have frequently been associated with receptor-mediated regulation in vertebrate tissues. Examples include the tyrosine kinase domains of the insulin, epidermal growth factor, and platelet-derived growth factor receptors (17), as well as GTPases such as the transducin, T, (18)(19)(20), or G-proteins, G , (21). In view of the requirement for purine nucleotides (5-7), it seems likely that analogous enzymes are involved in sensory transduction during bacterial chemotaxis. Of the three genes, cheA, cheY, and chew, whose products are required for receptor-mediated motor responses, only the amino acid sequence deduced from che W shows a discernible nucleotide binding fingerprint? Whether or not Chew is actually a locus for nucleotide involvement in chemotaxis, the identity of the nucleotide and the chemistry of its interactions can only be determined by studying the Chew protein.
To address these questions we have purified Chew and have begun to investigate its properties.