Structure and Function of an Unusual Family of Protein Phosphatases: The Bacterial Chemotaxis Proteins CheC and CheX

Transcription

1 Molecular Cell, Vol. 16, , November 19, 2004, Copyright 2004 by Cell Press Structure and Function of an Unusual Family of Protein Phosphatases: The Bacterial Chemotaxis Proteins CheC and CheX Sang-Youn Park, Xingjuan Chao, Gabriela Gonzalez-Bonet, Bryan D. Beel, Alexandrine M. Bilwes, and Brian R. Crane* Department of Chemistry and Chemical Biology Cornell University Ithaca, New York Summary In bacterial chemotaxis, phosphorylated CheY levels control the sense of flagella rotation and thereby de- termine swimming behavior. In E. coli, CheY dephosphorylation by CheZ extinguishes the switching signal. But, instead of CheZ, many chemotactic bacteria con- tain CheC, CheD, and/or CheX. The crystal structures of T. maritima CheC and CheX reveal a common fold unlike that of any other known protein. Unlike CheC, CheX dimerizes via a continuous sheet between subunits. T. maritima CheC, as well as CheX, dephosphorylate CheY, although CheC requires binding of CheD to achieve the activity of CheX. Structural analyses identified one conserved active site in CheX and two in CheC; mutations therein reduce CheY-phosphatase activity, but only mutants of two invariant asparagine residues are completely inactive even in the presence of CheD. Our structures indicate that the flagellar switch components FliY and FliM resemble CheC more closely than CheX, but attribute phosphatase activity only to FliY. Introduction ceptors change methylation state depending on the relative activities of the methyl-transferase CheR and the CheA-regulated methyl-esterase CheB. This feedback mechanism controls sensitivity and prevents saturation as the bacteria swims up chemical gradients. In E. coli, robust chemotaxis requires that the CheY phosphatase CheZ extinguishes the signal provided by CheY-P. However, many chemotactic bacteria, such as B. subtilis and T. maritima do not contain CheZ, but instead contain additional chemotaxis proteins such as CheD, CheC, CheX, and FliY (a flagella switch-complex component) (Figure 1A). In general, E. coli and B. subtilis represent different classes of chemotactic bacteria (to the latter of which T. maritima is more closely related [Nel- son et al., 2001]). For example, E. coli and B. subtilis have contrasted responses to ligands: attractant binding to the receptor increases CheA kinase activity in B. subtilis but decreases CheA activity in E. coli. Counteracting this opposite effect on kinase stimulation, CheY-P promotes CCW flagellar rotation in B. subtilis (Fuhrer and Ordal, 1991; Garrity and Ordal, 1995) instead of CW rotation in E. coli. Adaptation mechanisms also differ in B. subtilis (Garrity and Ordal, 1995; Zimmer et al., 2000) and involve CheC and CheD (Garrity and Ordal, 1995; Rosario et al., 1995), in addition to CheB and CheR (Saulmon et al., 2004). In B. subtilis,achec mutant does not affect the flagel- lar rotational bias (relative degree of CCW versus CW rotation in a chemically stable environment), but does reduce the frequency of switching (Saulmon et al., 2004). The chec mutant also overmethylates receptors and prevents adaptation of the flagellar rotational bias to prestimulus levels (Rosario et al., 1995; Rosario and Ordal, 1996; Saulmon et al., 2004). The ched mutant heavily biases toward the default CW rotation and does not respond well to attractant (Kristich and Ordal, 2002; Saulmon et al., 2004). This may be a consequence of the biochemically characterized activity of CheD to deamidate MCP glutamine residues and thereby mediate receptor maturation (Kristich and Ordal, 2002). Recently, Ordal and colleagues have demonstrated an enzymatic role for CheC and CheD in CheY dephosphorylation: CheC alone acts as a weak CheY-P phosphatase and surprisingly this activity increases in the presence of CheD (Szurmant et al., 2004). Consistent with its similarity to CheC, FliY/N also displays CheY-P phosphatase activity (Szurmant et al., 2004; Szurmant and Ordal, 2004) and has been considered the B. subtilis functional analog of the missing CheZ. A B. subtilis dou- ble mutant chec/ fliy/n 6-15 (expressing no CheC and a FliY/N missing an N-terminal peptide that binds CheY-P) completely biases toward CCW rotation, implying a per- The ability of bacteria to modulate their swimming be- havior in the presence of external chemicals (nutrients and repellants) is one of the most rudimentary behavioral responses known (Falke et al., 1997; Hess et al., 1988). Nevertheless, bacterial chemotaxis has amazing sensi- tivity, robustness, and dynamic range (Alon et al., 1999; Sourjik and Berg, 2002). Bacteria are capable of sensing changes in chemical concentrations of less than a percent over background concentrations ranging five orders of magnitude (Sourjik and Berg, 2002). During bac- terial chemotaxis, ligand binding to transmembrane methyl-accepting chemoreceptor proteins (MCPs) regu- lates the activity of the histidine kinase CheA. CheA phosphorylates the response regulator CheY on a spe- cific aspartate residue. Phosphorylated CheY (CheY-P) binds to the flagellar motor and switches its direction of rotation. The time spent smooth swimming (counterclockwise flagella rotation, CCW) or tumbling (clockwise flagella rotation, CW) determines the direction and rate sistent CheY-P signal (Szurmant et al., 2003, 2004). of movement. In addition to receptor-regulated phos- CheX is probably the least well-characterized chemophotransfer, chemotaxis also incorporates a slower ad- taxis protein; it was identified as a component of chemoaptation response whereby CheA activity results in taxis operons of Trepanoma denticola and Borrelia chemical modification of the MCPs (Parkinson and Ko- burgdorferi (Szurmant and Ordal, 2004). Other than havfoid, 1992). Specifically, the intercellular regions of re- ing homology to CheC, little is known about its biochemical activity and biological function. It is currently found *Correspondence: bc69@cornell.edu in about 20 genomes of which so far none are archaea

2 Molecular Cell 564 Figure 1. The CheC Family (A) Domain organization for the family of protein aspartate-phosphatases and related proteins in three different bacteria, T. maritima, B. subtilis, and E. coli. Purple segments represent the CheC homology region; green segments represent the FliN homology region (PDB code: 1O6A). FliY/N and FliM contain an N-terminal peptide that binds CheY-P (black). CheC, CheX, and FliY/N contain dephosphorylation centers (white stars with conserved residues above), but FliM does not. Most FliY/N proteins follow the domain architecture of bsfliy/n. TmFliY is abnormally short. The structurally unrelated CheY-phosphatase CheZ found in - and -proteobacteria is shown in red. (B) Sequence alignments of T. maritima CheC, FliY, and CheX. Secondary structure elements of CheC (above) and CheX (below) are similar except for the regions (in red) that form helices in CheC ( 2 and 2 ) and strands ( X and X ) in the dimer interface of CheX. Residue conservation (boxes) clusters in the regions of 1 ( 1 ) and 1 ( 1 ). Black boxes highlight conserved residues in the active site regions, and red boxes encircle residues that are markers for the CheX family versus CheC family. and several are human pathogens (spirochaetes). Often (Figure 1A). In B. subtilis, FliY is fused to FliN, generating multiple versions of chex are found in a given organism a protein we will call FliY/N. In T. maritima, FliY is a (see below). separate polypeptide (Figure 1A). Unfortunately, confusion The structures of CheC and CheX are important for can arise because genes for FliY/N fusion proteins understanding the molecular properties of a large pro- can be annotated in the NCBI database as either FliY, tein family that includes the central domains of flagella FliN, or FliY/N. Both FliY/N and FliM proteins contain proteins FliY/N and FliM, whose structures are currently an amino-terminal peptide ( 16 aa) that binds CheY-P unknown (Figure 1A). In E. coli, CheY-P interacts with (Szurmant et al., 2003) (Figure 1A). Unlike FliY/N and the cytoplasmic flagella switch complex consisting of FliM, CheC and CheX do not contain the CheY-P binding FliM, FliN, and FliG (Blair, 1995). FliM has a central domain amino-terminal peptide or the FliN homology region homologous to CheC, but does not have CheY-P (Kirby et al., 2001) (Figure 1A). Genes for CheC/CheX/ phosphatase activity (Figure 1A) (Bischoff and Ordal, FliY/FliM show a clear sequence repetition within the 1992; Bren and Eisenbach, 1998; Kirby et al., 2001; Lee CheC homology domain (Szurmant et al., 2004), suggesting et al., 2001b; Mathews et al., 1998; Szurmant et al., 2003). a common progenitor that evolved by gene duet In many nonenteric bacteria, the flagella switch complex plication. has an additional component that is also homologous to Bacterial response regulator domains homologous to CheC and has CheY-P phosphatase activity (Szurmant CheY belong to a superfamily of proteins found in all et al., 2003, 2004). This domain we will refer to as FliY kingdoms of life. Representatives include the haloacid

3 CheC and CheX Chemotaxis Phosphatases 565 dehalogenase family (HAD) that catalyzes hydrolytic reactions, phosphoserine phosphatases (PSPs) that biosynthesize serine in the brain, P-type ATPases that transport cations across membranes, and GTPases of the p21 ras family that propagate intracellular signals (Aravind et al., 1998; Cho et al., 2001; Wang et al., 2002). The GTPases hydrolyze GTP instead of aspartyl-phosphate, but do conserve the same fold and active center location as response regulators (Cronet et al., 1995; Lowy and Willumsen, 1993). Thus, studies of CheC and CheX provide a new avenue to explore mechanisms of phosphotransfer regulation relevant for a large and varied class of biological processes. Herein, we report the crystal structures of T. maritima CheC and CheX, and identify the common dephosphorylation active sites by structure analysis, sequence comparison, and mutagenesis studies. We show in vitro that CheC has two independent active sites and that two pairs of highly conserved asparagine and glutamate residues are critical for CheY dephosphorylation. We show that CheX has phosphatase activity and a different oligomeric state than CheC that may be related to its enhanced activity. Our structures reveal sequence motifs that distinguish the two subclasses of proteins. Results CheC Has Internal 2-Fold Symmetry CheC is a globular / protein with two long peripheral strands that wrap exposed helices lying atop a sheet platform (Figures 1B and 2A). The structure of T. maritima CheC was determined at 1.75 Å resolution by multiwavelength anomalous diffraction (MAD) of a single-site mercury derivative (Table 1). The final model (R factor of 22.6%, R free of 23.9%) includes 204 (out of 205) CheC residues and 201 water molecules (Table 2). The CheC / / -sandwich fold has six helices surrounding a sixstranded antiparallel sheet (Figure 2A). This topology is unrelated to any other known fold as defined by Dali (Holm and Sander, 1993) (highest Z score 4). An internal 2-fold (C2) symmetry axis perpendicular to the sheet Figure 2. The Folds of CheC and CheX (A) Ribbon diagrams show topologies and secondary structural elements for the CheC monomer (left) and one CheX subunit (right) from two orientations. The approximate locations of important invariant residues (e.g. Glu13, Asn16, and Pro39 for CheC, left active site) are shown by stars on white circles. Pseudo-2-fold axes perpendicular to the page in upper image relate one half of each molecule to the other. The kink in 1 ( 1 ) wraps the center of 1 ( 1 ) where conserved active site residues reside. Lower images show how CheC 2 and 2 replaced X and X that form the dimer interface in CheX. (B) Dimer formation in CheX. Two continuous sheets associate the CheX subunits (blue and gray). Strands X from each subunit (light and dark orange) swap across the dimer interface to form main chain hydrogen bonds with 1 of the adjacent subunit. The flagella proteins, FliY/N and FliM are predicted to have structures more analogous to CheC than CheX. (C) Solvent-accessible surfaces for CheC (left) and CheX (right) rendered transparent to view protein topology and key residue structure within. Conserved proline residues (white bonds) on 1 lie at the center of the CheX dimer interface (blue and gray subunits, yellow and green ribbons) adjacent to the conserved Glu and Asn residues on 1 (white bonds). In CheC, the conserved prolines (white bonds) on 1 form a protrusion at the side of the molecule that may mediate contact with another protein. Conserved residues in the other active site of CheC ( 1-1) also project into solvent (white bonds, lower left).

4 Molecular Cell 566 Table 1. Data Collection and Phasing Statistics CheC CheX Native Hg Peak Hg Inflection Hg Remote Native Wavelength (Å) Resolution (Å) Highest shell ( ) ( ) ( ) ( ) ( ) Completeness (%) 99.9 (99.9) 98.0 (99.6) 97.9 (99.4) 95.0 (77.1) 93.1 (53.9) a R merge (0.368) (0.336) (0.325) (0.311) (0.206) I/ (I) 48 (7) 43 (5) 44 (6) 30 (4) 18 (4) Figure of merit 0.51 ( Å) a R merge j I j I / j I j relates helices 1, 2, 3to 1, 2, 3 and the strands residues and 42 water molecules per asymmetric unit. 1, 2, 3 to 1, 2, 3. Three sets of two helices CheX (155 residues per polypeptide) differs from CheC surround the central sheet: two long helices, 1 and (205 residues) by (1) the loss of 3 (the last helix in 1, run diagonal to the strands on one side; 2 and CheC), (2) a shorter 1, and (3) the replacement of helices 2 pack against the opposite side; and 3 and 3 cap 2 and 2 by two strands X and X (Figures 1B and the edge of the sheet in orientations perpendicular 2A). CheC and CheX have similar pseudosymmetric to- to the other helices (Figure 2A). The strands align pologies, but unlike CheC, CheX is an obligate dimer. antiparallel in the order: Thus, the The latter strand, X, dimerizes CheX by extending the 2-3 and 2-3 hairpins insert into the adjacent re- five-stranded sheet from the adjacent CheX subunit peat (Figures 1B and 2). CheC appears to have arisen by one antiparallel strand. Indeed, in each subunit, X from a gene duplication followed by domain swapping makes four main chain hydrogen bonds with 1 across of the 2-3 hairpin. the dimer interface (Figure 2B). In doing so, these inter- Comparing the two CheC repeats, 1-1 and 1-1 actions create two six-stranded sheets that pack share the most sequence identity and structural relatedness. against each other in the center of the dimer with a In contrast, 2 is significantly shorter than 2, barrel-type arrangement. X is well-oriented to further 3 is longer than 3, and the loop between 2 and 3 extend the sheet but is slightly too far from X of the is longer than the loop between 2 and 3. Given the same subunit for hydrogen bonding. The 1453 Å 2 per divergence of most other elements between the two subunit of surface buried upon CheX dimerization is repeats, the similarity of 1-1 and 1-1 suggests consistent with CheX eluting from a sizing column as a that function constrains these regions. dimer (data not shown). We note that CheC also forms a symmetric dimer in the crystal, although the 3 residues CheX and CheC Differ in Their Quaternary Structure participating in the association create a smaller contact The structure of CheX was determined by molecular region than other lattice interactions. Indeed, CheC does replacement using as a model the RCSB deposited coordinates not dimerize in solution (data not shown). (PDB code: 1SQU) from the structural geno- mics initiative (R. Zhang, A. Savchenko, A. Edwards, Structures Reveal Sequence Markers to Delineate and A. Joachimiak). The model of CheC could not be CheC from CheX used to determine the structure of CheX by molecular Conserved residues differentiate CheX sequences from replacement. Our final CheX model is refined to 2.5 Å those of CheC by marking structural elements important resolution (R factor 24.0%, R free 29.5%) and contains 309 for CheX dimerization (Figure 1B). In addition to CheXs Table 2. Refinement Statistics Protein CheC CheX Space group P P Unit cell a 66.3 Å, c Å a 52.2 Å, b 67.9 Å, c 78.8 Å Resolution range (highest shell) (Å) ( ) ( ) Unique reflections 32,043 9,609 Wilson B (Å 2 ) R factor a (R free ) b (0.239) (0.295) No. of scatters (no. of residues) 1,762 (205) 2,325 (309) No. of water molecules Rmsd bonds (Å) Rmsd angles ( ) Average B factor (main chain) (Å 2 ) Average B factor (side chain) (Å 2 ) Average B factor (waters) (Å 2 ) a R factor ( F obs F calc )/ F obs b R factor for 10% of reflections not included in refinement.

5 CheC and CheX Chemotaxis Phosphatases 567 generally being shorter ( residues) than CheCs ( residues), CheXs conserve a Gly (rarely Ser) residue that mediates a turn important for forming the sheet of the dimer interface (T. maritima Gly 121 at the end of 1, highlighted in Figure 1B). In most CheC sequences, the loop connecting 1 and 2 includes two nonglycine residues, Asp 143 and Met 144. We suggest that the conserved Gly motif important for CheX dimerization should be the basis to distinguish CheC from CheX sequences. With this consideration, the NCBI database has numerous CheX sequences annotated incorrectly as CheC. For instance, five genes of Clostridium thermocellum are assigned as CheC homologs, but only one of the genes has the Asp-Met CheC motif; all the other four genes instead have the CheX-conserved Gly. Similarly, in Geobacter metallireducens, among the four genes assigned as CheC homologs, only one has Asp-Met while the rest have Gly. This clear delineation between dimers and monomers in the CheX/CheC family indicates that oligomeric state may reflect functional differences between the two subclasses. Activities of CheC and CheX toward CheY-P We tested the effects of T. maritima CheC, CheX, and CheD on T. maritima CheA-P and CheY-P. As typical of CheA proteins at 25 C, T. maritima CheA autophosphorylates itself prior to transferring phosphate to CheY (Figures 3A and 3B). Despite reports that B. subtilis CheC and CheA interact in a yeast two-hybrid assay (Kirby et Figure 3. Activities of T. maritima CheC, CheX, and CheD al., 2001), T. maritima CheC, CheX, and CheD had no (A) Flow of phosphate followed in the experiments shown in (B) (C). effect on CheA autophosphorylation or dephophosphor- (B) Autophosphorylated CheA (CheA- 32 P) in the absence or presence ylation. We also saw no evidence for CheC, CheD, or of CheY, CheC, CheD, CheX, and CheC double mutants Glu13Ser CheX phosphorylation by CheA or CheY. However, CheX Glu112Ser (CheC dme) or Asn16Ser Asn115Ser (CheC dmn). Only and CheC (but not CheD) dephosphorylated CheY, with CheY dephosphorylates CheA (lane 2). Both CheX (lane 10) and CheX having greater activity than CheC (Figure 3B). On CheC (lane 6) reduce the amount of CheY-P, although CheX has addition of CheD, CheC activity increased substantially much greater activity as no CheY-P remains in its presence. CheD activates CheC (lane 12) to roughly the same level as CheX (lane and dephosphorylated all of the CheY-P present in our 10). CheC dme and CheC dmn do not noticeably dephosphorylate assay. At the shortest time measured, both CheX and CheY (lanes 7 and 8); however, CheD partially rescues the activity CheC/CheD depleted not only all of the CheY-P but of dme (lane 13), but not dmn (lane 14). also all of the CheA-P, which likely results from a large (C) Effects of CheC mutants on CheY-P in the presence of CheD. increase in the steady-state concentration of unphosphorylated CheA (CheA- 32 P). Lanes 1 7 are controls as designated; Bands corresponding to CheY-P after transfer from autophos- phorylated CheY (Figure 3B). We were unable to detect lanes 8 13 are CheA CheY CheD CheC mutants: Glu13Ser, an effect of CheC or CheX on the T. maritima CheB Glu112Ser, Glu13Ser/Glu112Ser, Asn16Ser, Asn115Ser, and Asn16- methylesterase (which contains a CheY-like domain), Ser/Asn115Ser, respectively. Coomassie-stained SDS-PAGE gel due to the short lifetime of CheB-aspartyl-phosphate showing that the concentrations of the CheC mutants are equivalent. (data not shown). Identification of Putative Dephosphorylation Centers in CheC and CheX 1 helices (Figures 2C, 4A, and 4B). The kinks result In an attempt to identify putative active sites for CheC from a conserved pair of small residues Pro (Val or Thr)- and CheX, we carried out sequence comparisons among Pro that are 21 residues C-terminal to the invariant asparagine the known CheX and CheC homologs in the NCBI database. (Figures 1B, 2C, and 4A 4D). One helix turn These alignments showed that only a few residues N-terminal from the invariant Glu, another conserved were highly conserved in sequence positions corresponding hydrophilic residue projects toward the Glu-Asn pair to the center of helices 1 and 1, and the (Asp9 on 1 and Ser108 on 1 ). Thus, the two CheC adjacent strands 1 and 1 (Figures 1B and 2). In clusters of conserved residues each comprise a protrud- CheC, these motifs expose the invariant hydrophilic resi- ing motif that appears well designed to complement the dues Glu13 and Asn16 on 1 and Glu112 and Asn115 sequestered pocket of CheY containing the aspartyl- on 1. Thus, symmetry suggests the molecule has two phosphate. Notably, these two CheC putative active similar, but distinct active sites. Each Glu-Asn pair sites are similar, but not identical, and hence may have (13 16 and ) flanks a kink/bulge in an adjacent different activities (Figures 4A and 4B). strand (strands 1 or 1 ) that protrudes from the Although the regions surrounding the conserved Asn sheet and forms a ledge with respect to the long 1 and and Glu residues do not resemble a known metal binding

6 Molecular Cell 568 Figure 4. The Dephosphorylation Centers of CheC and CheX (A D) Stereoimages of 2F o F c electron-density maps (contoured at 1, green for 1 and 1 and gray for 1 and 1 ) showing regions surrounding conserved Asn16 (A) or Asn115 (B) for CheC; and surrounding conserved Asn94 for CheX in the A subunit of the dimer (C) and the B subunit of the dimer (D). Conformations of the conserved Asn and Glu residues slightly differ in each case, as does the juxtaposition of the kink in 1 ( 1 ) provided by the Pro (Val)-Pro motif. This variability is most dramatic in CheX where the two subunits differ by a shift in 1 of one helical turn relative to the Pro-Pro repeat (C and D). site, we nevertheless soaked the CheC crystals with 10 mm concentrations of Mg 2 to test for divalent metalion binding and collected a diffraction data set: no metal coordination was observed in difference Fourier electron density maps at either cluster of conserved residues or elsewhere on the molecule (data not shown).

7 CheC and CheX Chemotaxis Phosphatases 569 In CheX, the second putative active site conserves a conserved residue cluster necessary for catalyzing the same residues found in both CheC centers (i.e., dephosphorylation of CheY. In CheC, two pseudosym- Glu91, Asn94, Pro115, Pro116, and Ser87 for T. maritima metric dephosphorylation centers each contain an essential CheX). However, several CheX sequences including that Asn and an important Glu residue. One such of T. maritima lack this motif in the position of the first center is conserved by each CheX monomer, but CheX active center. For example, in CheX, Arg4 replaces the dimerizes, and thereby also generates two centers per invariant Glu and 1 is truncated at the amino terminus molecule. Compared to CheC and CheX, eukaryotic protein (Figures 1B and 2B). Thus, only the second putative phosphatases fall into two very different categories: CheX active site has properties similar to the two CheC serine/threonine protein phosphatases (PPs) and protein sites. In this region, the conserved Pro-Pro protrusion tyrosine phosphatases (PTPs). In PPs, a dinuclear on 1 projects into the dimer interface, and as a result metal center facilitates dephosphorylation reactions by the proline pair lies at the bottom of a cavity created by both binding the target phosphate and activating a water 1 and the adjacent subunit of the CheX dimer (Fig- molecule for nucleophilic attack on the substrate (Jackson ure 2C). and Denu, 2001). In contrast, PTPs catalyze dephosure phorylation by forming a thio-phosphate intermediate Mutant Studies Indicate that CheC between the substrate and an active-center cysteine Has Two Active Centers (Jackson and Denu, 2001; Kolmosin and Aqvist, 2001). CheC single-point mutants of Glu13, Asn16, Glu112, or A water molecule, activated by a PTP general base (Glu Asn115 (all to Ser) and the double mutants, Glu13Ser/ or Asp), then hydrolyzes the thio-phosphate (Jackson Glu112Ser and Asn16Ser/Asn115, all reduced CheY and Denu, 2001; Kolmosin and Aqvist, 2001). In addition phosphatase activity to a nearly undetectable amount to PPs and PTPs, bacteria contain CheY-type response (i.e., increased CheY-P in the assay to a level similar to regulators, which act as phosphatases for proteins conthat seen with no CheC present) (Figure 3). Thus, beaspartyl-phosphatase, taining histidinyl-phosphate (e.g., CheA), and the CheZ cause of low CheC activity in the absence of CheD, it which dephosphorylates CheY. is difficult to distinguish the effects of the single mutants From this point of view, CheY aspartyl-phosphate is from those of the double mutants (data not shown). analogous to the cysteinyl-phosphate intermediate of However, on the addition of CheD, activities of all single PTPs. CheZ, along with CheC and CheX, function to mutants increase to levels greater than that of wildmediate, break down the equivalent of the PTP phosphoryl-inter- type CheC alone, and differential effects of the double an activity intrinsic to the PTPs. Interestingly, mutants become apparent (Figures 3B and 3C). In the in some PTPs, a critical glutamine residue has been presence of CheD, the Asn115Ser mutant displays less shown to facilitate dephosphorylation of the thiophos- activity than wild-type or the Asn16Ser mutant, but more phate by coordinating the hydrolytic water molecule activity than the Asn16Ser/Asn115Ser double mutant, (Pannifer et al., 1988; Zhao et al., 1988). The role of side which shows little or no activity (Figures 3B and 3C). The chain amides in dephosphorylation reactions has also Glu13/Glu112 double mutant also has reduced activity in received considerable attention because of a glutamine the presence of CheD, but exceeds the activity of CheC mutation site in p21 ras that slows GTP hydrolysis and alone (Figures 3B and 3C). Both CheC double mutants, causes cellular transformation (Lowy and Willumsen, Glu13/Glu112 and Asn16/Asn115, have the same affinity 1993). In such GTPases, this glutamine side chain also for CheD as wild-type (dissociation constant K D 0.9 positions a water molecule for in-line reaction with the 1.4 M, and stoichiometry n CheC/CheD by leaving phosphoryl group (Maegley et al., 1996; Scheffisothermal titration calorimetry); thus, these conserved zek et al., 1997). Gln147 in the E. coli CheY phosphatase residues do not solely mediate activation by binding CheZ has been proposed to have a similar catalytic role CheD. Each conserved residue pair, Glu13/Asn16 or (Zhao et al., 2002). Glu112/Asn115, likely participates directly in CheY de- In analogy to PTPs and GTPases, Asn16 and Asn115 phosphorylation, with the Asn residues being more criti- of CheC may also serve to orient a water molecule for cal for activity. Although the two centers reside on the reaction with the CheY phosphoryl group. Glu13/112 same face of CheC (Figure 2A), they are far enough are the only conserved general bases in the vicinity separated to each form distinct binding sites for CheY-P. of Asn16/Asn115 that could activate a hydrolytic water In the presence of CheD, only the Asn16/Asn115 CheC molecule. Although our mutagenesis data indicate that double mutant increases CheY-P levels to those seen Glu13/Glu112 participate in dephosphorylation of CheY-P, in the absence of CheC, whereas any one of the single they are not nearly as critical as Asn16/Asn115, which mutants, with CheD, significantly reduces CheY-P lev- cannot be general bases. However, a general base effect els. Thus, CheC appears to have two pseudosymmetric may not be necessary if the dephosphorylation reaction active sites, each marked by conserved Glu and Asn is largely dissociative (Maegley et al., 1996). Formation residues and each independently capable of dephos- of a covalent phospho-enzyme, as it occurs in the PTPs, is phorylating CheY. However, the mutagenesis studies also unlikely during the CheC dephosphorylation reaction also suggest that the Glu112/Asn115 active site is more because (1) we observe no formation of phospho-chec active than the Glu13/Asn16 site (Figure 3), which is not in our assays, and (2) mutations of the only conserved surprising given the different overall residue composi- nucleophilic residues (Glu13/Glu112) have considerable tions in these two regions. activity in the presence of CheD. Thus, based on analogies to CheZ, PPs, and GTPases, the Asn16/Asn115 Discussion residues may be essential because they orient a water molecule for reaction with the CheY-phosphoryl group. CheC and CheX represent a unique family of protein However, different or additional roles for these residues phosphatases that has a novel folding topology and are also possible. For example, amide residues coordi-

8 Molecular Cell 570 CheZ-CheY complex (model not shown). Thus, the details of CheY recognition and catalysis likely differ for the two classes of CheY phosphatases, even though side chain amide groups are important for both. Furthermore, both proteins apparently use helices to project side chains into the CheY active site. This is a common theme for other proteins that also interact with the phosphorylation sites of response regulators, such as the B. subtilis phosphotransferase Spo0B, the yeast YPD1 phosphotransferase, and the CheA phosphotransferase domain P1 (Varughese, 2002; Xu et al., 2003). Figure 5. Comparing the dephosphorylation centers of CheC and CheZ Comparison between the putative active center residues of CheC and those of CheZ in complex with activated CheY. Superposition of CheZ Gln147 with CheC Glu13 and their bearing helices leads to a clash-free complex between CheY and CheC and aligns CheZ Asp143 with CheC Asp9. However, the catalytically essential CheC Asn16 does not reside in the interface; hence, recognition of CheY-P by CheC must be different than by CheZ. Relative Activities of CheC and CheX The greater activity of CheX compared to CheC and the activation of CheC by CheD indicate that oligomeric state significantly impacts the properties of these phosphatases. CheC contains catalytic residues necessary and sufficient for accelerating dephosphorylation of CheY, but the CheC:CheD complex may have a higher affinity than CheC alone for CheY-P, in much the same way that B. subtilis FliY/N requires a distinct N-terminal region to bind CheY-P (Szurmant et al., 2003). A greater effective concentration of CheY-P would increase the apparent activity of both putative CheC active centers. Consistent with this view, the higher activity of CheX compared to CheC may relate to CheX having a higher affinity to CheY-P because of its dimeric state. Modeling the complex of CheX with T. maritima CheY (PDB code: 4TMY) indicates that, for the essential asparagine resi- due to approach the CheY active center, CheY will invariably interact with the adjacent subunit of the CheX dimer. If such a contact is important for CheY-P recogni- tion, the Pro-Pro motif (residues 115 and 116) in the center of the CheX dimer interface may be conserved by CheC to generate a similar binding surface through interactions with CheD (Figures 2C and 4A 4D). Alternatively, CheD could directly enhance CheC catalysis of dephosphorylation. Our CheX structure indicates that the conserved active center of CheC and CheX can have significantly different conformations in the same molecule. In our CheX dimer, the active center helices do not have the same orientation in each subunit; in one subunit, the 1 helix slides a complete turn to- ward the dimer interface. This motion separates the conserved Asn and Glu on 1 from the two conserved prolines on 1 (Figures 4C, 4D, and 6). Interaction with the adjacent subunit stabilizes the shift. The structural difference between the two CheX subunits derives from the N-terminal methionine residue that inserts between 1 and the sheet in the unshifted subunit of our CheX structure (Figure 6) and both subunits of the CheX struc- ture determined by the structural genomics consortium (PDB code: 1SQU). The N-terminal Met flips out in the shifted subunit and causes the 1 and 1 helices to move away from the sheet and toward the dimer interface. The unique C-terminal helix of CheC ( 3 ) mimics interactions provided by the CheX N-terminal Met (Fig- ure 6). Thus, in both CheC and CheX, the conformation of the putative active site depends on interactions that bridge the central sheet to the two long helices ( 1 and 1 ). In CheC, internal and external residues of 3 have higher conservation than those on other exposed regions of the protein. CheD may modulate CheC activity nate Mg 2 for stabilizing bound phosphates in the CheZ- CheY complex (Zhao et al., 2002) (Figure 5) and in the CheA histidine kinase (Bilwes et al., 2001). Although we found no evidence for Mg 2 coordination by CheC, Mg 2 does stabilize CheY-P (Lee et al., 2001a). Thus, the CheC critical asparagine residues could also facilitate dephosphorylation by interfering with the Mg 2 coordination of CheY-P. Comparison with the CheY Phosphatase CheZ Although CheC differs in overall structure compared to CheZ, which is a long dimeric four helix bundle (Zhao et al., 2002), the proteins share some features in their respective active centers. The structure of the complex between CheZ and BeF 3 -activated CheY shows that CheZ inserts an essential Gln (147) and Asp (143) into the CheY active site (Zhao et al., 2002). Gln147 orients a water molecule for reaction with the CheY phosphorylaspartate (Asp57). Asp143, positioned one turn of helix away from Gln147, contacts the CheY Asp13-Lys109 salt bridge, which stacks against the aspartyl-phosphate. Biochemical studies support a water molecule, positioned by Gln147, as a direct reactant in CheZ-mediated CheY dephosphorylation (Silversmith et al., 2003; Wolanin et al., 2003). Superposition of CheC residue Glu13 (or Glu112) with CheZ Gln147 and superposition of 1 (or 1 ) with the CheZ bearing helix lead to a clashfree model of a CheC-CheY complex that also aligns CheC Asp9 with CheZ Asp143 (Figure 5). However, such a model removes the essential CheC asparagine residues from the vicinity of the CheY aspartyl-phosphate (Figure 5). Superposition of either Asn16 or Asn115 with CheZ Gln147 requires a different orientation of T. maritima CheY relative to the bearing helix than is found in the

9 CheC and CheX Chemotaxis Phosphatases 571 Figure 6. Helical Shifts in CheX and CheC Stereoimage of the superposition of C s traces for the two CheX molecules (red and green) in our structure, the two molecules (gray) in the deposited coordinates of CheX (PDB code: 1SQU) and CheC (blue). The N-terminal Met (Met1) of CheX bridges interactions between 1, 1, and the sheet in all subunits except subunit A of our structure, where the N-terminus of 1 projects into solvent and as a result 1, 1 shift by one helical turn towards the dimer interface. In CheC, 3 structurally mimics the CheX N-terminal Met, resulting in a third conformation for 1 and 1. The side chains for the following conserved residues are represented: Asn115, Pro137, Pro138 (CheC) and Met1, Asn94, Pro115, Pro116 (CheX). by binding to a region that includes 3 and thereby solely binds CheY-P and does not itself participate in propagates structural changes to 1 and 1. the dephosphorylation reaction. The structure of CheC provides insight into the molecular The Flagellar Proteins FliY/N and FliM properties of FliM, one of the remaining flagellar The two flagella switch proteins FliM and FliY/N contain proteins to have its structure determined. Like FliY/N, a domain homologous to CheC and CheX (Figure 1A) FliM has sequence markers that place it in the CheC (Kirby et al., 2001). This domain conserves hydrophobic rather than the CheX subclass. However, FliM is devoid residues in the loop following 1 but not the glycine of phosphatase activity (Szurmant et al., 2003), and it that initiates transition of the polypeptide chain into the indeed does not conserve the catalytically essential Glu CheX dimer interface. The absence of the glycine following and Asn residues of CheC. Thus, the CheC homology 1 indicates that the structures of the central do- region of FliM must have a different function, one that mains of FliM and FliY/N, and hence association properties, includes binding to FliG in the center of the flagella MS are more similar to CheC than to CheX. Unlike most ring (Mathews et al., 1998). FliY/N sequences, T. maritima FliY (tmfliy) is too short to comprise the folding unit found in CheC and CheX. CheC and CheD Are Another Intersection Point tmfliy omits 2, 3, and the second -hairpin 2-3 between Excitation and Adaptation in Chemotaxis (Figure 1B). In the CheC/CheX topology, 1 requires the In E. coli, CheY phosphorylation/dephosphorylation generates 2-3 hairpin to associate with the sheet. Perhaps the chemotaxis excitation response, whereas re- insertion of a -hairpin from another tmfliy subunit, the ceptor methylation mediates adaptation. An intersection N terminus of FliN, or another protein generates a FliY between the two processes occurs at the CheA kinase, domain analogous in structure to that of CheC/CheX. which both transduces the excitation response by phosphorylating Extension of the FliY sheet may thus mediate oligomerization CheY, and initiates adaptation by phosphory- in the flagella rotor. B. subtilis FliY/N and T. mari- lating CheB. In other bacteria and archaea, CheC and CheD tima FliY contain an N-terminal peptide (residues 6 15, also provide a link between excitation and adaptation. Figure 1) that on deletion greatly curtails CheY dephos- CheD activates CheC for dephosphorylating CheY (Szurmant phorylation activity (Szurmant et al., 2003). This peptide et al., 2004), but CheD also deaminates receptors has high homology to the N terminus of FliM where (Kristich and Ordal, 2002). Mutants in CheC (Saulmon et CheY-P binds to switch the direction of flagellar rotation. al., 2004) affect receptor methylation levels, and thus CheC As FliY/N conserves the dephosphorylation centers of may also influence the activity and localization of CheD. CheC and CheX, we suspect that this N-terminal peptide In B. subtilis, it appears that restoration of CheY-P to

10 Molecular Cell 572 prestimulus levels requires action of both CheC and FliY/N The final model (CheC residues 1 204) was then refined against a (Szurmant et al., 2004). The truncated FliY of T. maritima native data set with CNS (Brunger et al., 1998) (final R factor 0.226, R free 0.239). may not be able to dephosphorylate CheY. Thus CheX CheX and/or CheC with CheD may be the dominant CheY phos- The CheX structure was determined by molecular replacement with phatase. Further studies are required to determine AMoRe (Navaza, 1994) using as a model the RCSB deposited coordinates whether and why T. maritima needs two seemingly redundant (PDB code: 1SQU) from the structural genomics initiative (R. CheY phosphatases. In several organisms (Clostridof Zhang, A. Savchenko, A. Edwards, and A. Joachimiak). Several parts ium acetobutylicum, Trichodesmium erythraeum, Bdelfrom the second monomer of CheX (helices 1 and 1 ) were removed the initial model and rebuilt manually. The final model refined lovibrio bacteriovorus, Pseudomonas syringae, most to 2.5 Å (R factor 24.0%, R free 29.5%) contains 42 water molecules, Vibrio, and Shewanella oneidensis), a CheX/C domain 151/155 residues for molecule A, and 155/155 residues for molecule is found fused to response regulators with homology to B plus 3 amino-terminal residues (GSH) introduced by the cloning CheY, perhaps providing an internal off-switch for the vector. The two CheX structures are similar in conformation with phosphorylation response. In one rare instance (Leptoour the exception of a large shift in the helical regions of subunit A for spira interrogans), a CheX/C domain is found fused to structure. a CheA histidine kinase. Genomes that contain multiple Radioactive Dephosphorylation Assays chex homologs, as well as homologs for chec and/or CheA (18 M) was autophosphorylated by incubation with 0.05 M fliy/n suggest that some organisms require many CheY [ - 32 P] ATP (1.5 l of 3000 Ci/mmol, 10 Ci/ l, Perkin-Elmer) and phosphatases for chemotaxis. Whether other response 20 M cold ATP for 15 min in a total volume of 100 l TKM buffer (50 regulators (e.g., CheB) are targets for these phospha- mm Tris [ph 8.5], 50 mm KCl, 5 mm MgCl 2 ). The autophosphorylated tases is an interesting possibility. Thus, emergence of CheA (CheA- 32 P) was then added to premixed protein solutions, the CheC/CheX family of phosphatases reveals surpris- resulting in final concentration of 11.8 M CheA- 32 P, 32.9 M CheY, 4.24 M CheC (native or mutants), 6.1 M CheD, or M ing complexity in chemotaxis signaling pathways. In CheX. After 3 min of incubation, 10 l of2 SDS buffer containing many organisms, the linear signal that takes CheY-P 50 mm EDTA was added to quench each reaction. The proteins from CheA to FliM to CheZ in E. coli, is likely elaborated were separated on a 4% 20% Tris-glycine SDS-PAGE, and then to involve multiple downstream phosphatases and com- transferred to an Immuno-Blot PVDF membrane (blotted for 30 min ponents of the MCP receptor:chea kinase arrays. at 100 V using transfer buffer [25 mm Tris, 192 mm glycine]). The PVDF membrane was exposed to film and the film was later developed. Experimental Procedures Protein concentrations were taken from the calculated extinc- tion coefficients at 280 nm based on aromatic amino acid content Protein Preparation and verified by the RC/DC assay (BioRad). Equal concentrations of The genes encoding T. maritima CheC, CheY, CheA, CheD, and CheC mutants were assured by SDS-PAGE and the RC/DC assay. CheX were PCR cloned into the vector pet28a (Novagen) and expressed with a 6-His tag in E. coli strain BL21 (DE3) (Novagen) in Isothermal Titration Calorimetry Terrific Broth (DIFCO) with kanamycin selection (25 g/ml). The chec BCA assay (Pierce) and RC/DC assay (BioRad) were used with Bopoint mutations were introduced by QuickChange mutagenesis vine Serum Albumin (BSA) and cytochrome c standards to determine (Stratagene) and verified by sequencing. All proteins were purified the total mass concentration (mg/ml) of CheC (MW 22,500 kda), on separate Nickel-NTA columns and their His tags removed by and CheD (MW 16,700 kda). Isothermal titration calorimetry (ITC) thrombin digestion. Proteins were further purified by a Superdex75 measurements were carried out using a VP-ITC titration calorimeter sizing column (Pharmacia) and concentrated by centrifugation (Ami- (MicroCal) at 25 C. Prior to titration of concentrated CheD into CheC, con Centriprep) in GF buffer (50 mm Tris [ph 7.5] and 150 mm NaCl). samples were dialyzed extensively against GF buffer. Crystallization and Data Collection Computer Graphics CheC Figures 2A, 2B, 5, and 6 were made with Molscript (Kraulis, 1991) Initial conditions for growing CheC crystals were found in commer- and rendered with Raster3D (Merritt and Murphy, 1994). Figures cial screening solutions (Hampton) and improved by extensive 4A 4D were made with Bobscript (Esnouf, 1997). Figure 2C was screening with additives (Hampton) for better diffraction quality. made with the software Spock (J.A. Christopher). Optimized crystals of CheC (70 mg/ml) grew by vapor diffusion against a reservoir of M Li 2 SO 4, 0.1 M HEPES (ph 7.5), and Acknowledgments 8% acetonitrile. The crystals belong to the space group P and contain one molecule per asymmetric unit. Mercury-derivatized We are indebted to Kristin Wilson for assistance with phosphorylacrystals were grown in similar conditions with the addition of 1 mm tion assays. We also thank Melvin Simon for his continuing support ethyl mercury bromide. Diffraction data for both native and mercuryand advice; Michael Becker and Shai Vaday for assistance at X25 derivatized CheC crystals were collected under a 100 K nitrogen station in NSLS (Brookhaven National Laboratory); the Cornell High stream at NSLS beamline X-25 on a CCD Quantum detector (Q315). Energy Synchrotron Source for data collection facilities; and Madha- Data at three wavelength (MAD) was collected for mercury-derivavan Buddha for assistance with MAD data sets. S.Y.P. thanks the tized CheC. Patterson analysis revealed one mercury bound per Chemistry and Biology Interface NIH Training Grant. This work was asymmetric unit. Data was processed by the HKL package (Otwisupported by NIH grant GM to B.R.C. nowski and Minor, 1997). CheX Received: August 14, 2004 Optimized crystals of CheX (14 mg/ml) grew by vapor diffusion Revised: October 4, 2004 against a reservoir of 0.1 M Tris (ph 8.5), 22.5% PEG 4k, and 8% Accepted: October 12, 2004 methyl-ethyl ketone. The crystals belong to the space group P Published: November 18, 2004 and contain two molecules per asymmetric unit. Native diffraction data was collected on CHESS beamline A1. References Structure Determination and Refinement CheC Alon, U., Surette, M.G., Barkai, N., and Liebler, S. (1999). Robustness The initial CheC model was built manually in XFIT (McRee, 1992) in in bacterial chemotaxis. Nature 397, a 2.6 Å SOLVE-generated map solvent-flattened in DM (CCP4, 1994). Aravind, L., Galperin, M.Y., and Koonin, E.V. (1998). The catalytic

11 CheC and CheX Chemotaxis Phosphatases 573 domain of the P-type ATPase has the haloacid dehalogenase fold. D.S., Huang, L.S., Kustu, S., Berry, E.A., and Wemmer, D.E. (2001b). Trends Biochem. Sci. 23, Crystal structure of an activated response regulator bound to its Bilwes, A.M., Quezada, C.M., Croal, L.R., Crane, B.R., and Simon, target. Nat. Struct. Biol. 8, M.I. (2001). Nucleotide binding by the histidine kinase CheA. Nat. Lowy, D.R., and Willumsen, B.M. (1993). Function and regulation of Struct. Biol. 8, ras. Annu. Rev. Biochem. 62, Bischoff, D.S., and Ordal, G.W. (1992). Identification and character- Maegley, K.A., Admiraal, S.J., and Herschlag, D. (1996). Ras-cataization of FliY, a novel component of the Bacillus subtilis flagellar lyzed hydrolysis of GTP: a new perspective from model studies. switch complex. Mol. Microbiol. 6, Proc. Natl. Acad. Sci. USA 93, Blair, D.F. (1995). How bacteria sense and swim. Annu. Rev. Micro- Mathews, M.A., Tang, H.L., and Blair, D.F. (1998). Domain analysis biol. 49, of the FliM protein of Escherichia coli. J. Bacteriol. 180, Bren, A., and Eisenbach, M. (1998). The N terminus of the flagellar McRee, D.E. (1992). XtalView: a visual protein crystallographic softswitch protein, FliM, is the binding domain for the chemotactic re- ware system for X11/Xview. J. Mol. Graph. 10, sponse regulator, CheY. J. Mol. Biol. 278, Merritt, E.A., and Murphy, M.E.P. (1994). Raster3D Version 2.0: a Brunger, A.T., Adams, P.D., Clore, G.M., Delano, W.L., Gros, P., program for photorealistic molecular graphics. Acta Crystallogr. D Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Biol. Crystallogr. 50, Pannu, N.S., et al. (1998). Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Navaza, J. (1994). AMoRe: an automated package for molecular Crystallogr. D Biol. Crystallogr. 54, replacement. Acta Crystallogr. A 50, CCP4 (Collaborative Computational Project, Number 4) (1994). The Nelson, K.E., Eisen, J.A., and Fraser, C.M. (2001). Genome of Ther- CCP4 suite: programs for protein crystallography. Acta Crystallogr. motoga maritima MSB8. Methods Enzymol. 330, D Biol. Crystallogr. 50, Otwinowski, A., and Minor, W. (1997). Processing of X-ray diffraction Cho, H., Wang, W., Kim, R., Yokota, H., Damo, S., Kim, S.H., Wemmer, data in oscillation mode. Methods Enzymol. 276, D., Kustu, S., and Yan, D. (2001). BeF(3)( ) acts as a phosphate Pannifer, A.D.B., Flint, A.J., Tonks, N.K., and Barford, D. (1988). analog in proteins phosphorylated on aspartate: structure of a Visualization of the cysteinyl-phosphate intermediate of a protein- BeF(3)( ) complex with phosphoserine phosphatase. Proc. Natl. tyrosine phosphatase by X-ray crystallography. J. Biol. Chem. Acad. Sci. USA 98, , Cronet, P., Bellsolell, L., Sander, C., Coll, M., and Serrano, L. (1995). Parkinson, J.S., and Kofoid, E.C. (1992). Communication modules Investigating the structural determinants of the p21-like triphos- in bacterial signaling proteins. Annu. Rev. Genet. 26, phate and Mg2 binding site. J. Mol. Biol. 249, Rosario, M.M., and Ordal, G.W. (1996). CheC and CheD interact to Esnouf, R.M. (1997). An extensively modified version of Molscript regulate methylation of Bacillus subtilis methyl-accepting chemothat includes greatly enhanced coloring capabilities. J. Mol. Graph. taxis proteins. Mol. Microbiol. 21, , Rosario, M.M., Kirby, J.R., Bochar, D.A., and Ordal, G.W. (1995). Falke, J.J., Bass, R.B., Butler, S.L., Chervitz, S.A., and Danielson, Chemotactic methylation and behavior in Bacillus subtilis: role of M.A. (1997). The two-component signaling pathway of bacterial che- two unique proteins, CheC and CheD. Biochemistry 34, motaxis: a molecular view of signal transduction by receptors, ki- Saulmon, M.M., Karatan, E., and Ordal, G.W. (2004). Effect of loss nases, and adapation enzymes. Annu. Rev. Cell Dev. Biol. 13, of CheC and other adaptational proteins on chemotactic behaviour in Bacillus subtilis. Microbiol. 150, Fuhrer, D.K., and Ordal, G.W. (1991). Bacillus subtilis CheN, a homo- Scheffzek, K., Ahmadian, M.R., Kabsch, W., Wiesmuller, L., Lautlog of CheA, the central regulator of chemotaxis in Escherichia coli. wein, A., Schmitz, F., and Wittinghofer, A. (1997). The Ras-RasGAP J. Bacteriol. 173, complex: structural basis for GTPase activation and its loss in onco- Garrity, L.F., and Ordal, G.W. (1995). Chemotaxis in Bacillus subtilis: genic Ras mutants. Science 277, how bacteria monitor environmental signals. Pharmacol. Ther. 68, Silversmith, R.E., Guanga, G.P., Betts, L., Chu, C., Zhao, R., and Bourret, R.B. (2003). CheZ-mediated dephosphorylation of the Esch- Hess, J.F., Oosawa, K., Kaplan, N., and Simon, M.I. (1988). Phoserichia coli chemotaxis response regulator CheY: role for CheY gluphorylation of three proteins in the signaling pathway of bacterial tamate 89. J. Bacteriol. 185, chemotaxis. Cell 53, Sourjik, V., and Berg, H.C. (2002). Receptor sensitivity in bacterial Holm, L., and Sander, C. (1993). Protein structure comparison by chemotaxis. Proc. Natl. Acad. Sci. USA 99, alignment of distance matrices. J. Mol. Biol. 233, Szurmant, H., and Ordal, G.W. (2004). Diversity in chemotaxis mech- Jackson, M.D., and Denu, J.M. (2001). Molecular reactions of protein anisms among the bacteria and archaea. Microbiol. Mol. Biol. Rev. phosphatases: insights from structure and chemistry. Chem. Rev. 68, , Kirby, J.R., Kristich, C.J., Saulmon, M.M., Zimmer, M.A., Garrity, Szurmant, H., Bunn, M.W., Cannistraro, V.J., and Ordal, G.W. (2003). L.F., Zhulin, I.B., and Ordal, G.W. (2001). CheC is related to the Bacillus subtilis hydrolyzes CheY-P at the location of its action, the family of flagellar switch proteins and acts independently from CheD flagellar switch. J. Biol. Chem. 278, to control chemotaxis in Bacillus subtilis. Mol. Microbiol. 42, Szurmant, H., Muff, T.J., and Ordal, G.W. (2004). Bacillus subtilis CheC and FliY are members of a novel class of CheY-P-hydrolyzing Kolmosin, K., and Aqvist, J. (2001). The catalytic mechanism of proteins in the chemotactic signal transduction cascade. J. Biol. protein tyrosine phosphatases revisited. FEBS Lett. 498, Chem. 279, Kraulis, P.J. (1991). Molscript: a program to produce both detailed Varughese, K.I. (2002). Molecular recognition of bacterial phosand schematic plots of protein structures. J. Appl. Crystallogr. 24, phorelay proteins. Curr. Opin. Microbiol. 5, Wang, W., Cho, H.S., Kim, R., Jancarik, J., Yokota, H., Nguyen, Kristich, C.J., and Ordal, G.W. (2002). Bacillus subtilis CheD is a H.H., Grigoriev, I.V., Wemmer, D.E., and Kim, S.H. (2002). Structural chemoreceptor modification enzyme required for chemotaxis. J. characterization of the reaction pathway in phosphoserine phospha- Biol. Chem. 277, tase: crystallographic snapshots of intermediate states. J. Mol. Lee, S.Y., Cho, H.S., Pelton, J.G., Yan, D.L., Berry, E.A., and Wem- Biol. 319, mer, D.E. (2001a). Crystal structure of activated CheY: comparison Wolanin, P.M., Webre, D.J., and Stock, J.B. (2003). Mechanism of with other activated receiver domains. J. Biol. Chem. 276, phosphatase activity in the chemotaxis response regulator CheY Biochemistry 42, Lee, S.Y., Cho, H.S., Pelton, J.G., Yan, D.L., Henderson, R.K., King, Xu, Q., Porter, S.W., and West, A.H. (2003). The yeast YPD1/SLN1