A Selective Tether Recruits Activated Response Regulator CheB to Its Chemoreceptor Substrate

ABSTRACT Methylesterase/deamidase CheB is a key component of bacterial chemotaxis systems. It is also a prominent example of a two-component response regulator in which the effector domain is an enzyme. Like other response regulators, CheB is activated by phosphorylation of an aspartyl residue in its regulatory domain, creating an open conformation between its two domains. Studies of CheB in Escherichia coli and related organisms have shown that its enzymatic action is also enhanced by a pentapeptide-binding site for the enzyme at the chemoreceptor carboxyl terminus. Related carboxyl-terminal pentapeptides are found on >25,000 chemoreceptor sequences distributed across 11 bacterial phyla and many bacterial species, in which they presumably play similar roles. Yet, little is known about the interrelationship of CheB phosphorylation, pentapeptide binding, and interactions with its substrate methylesters and amides on the body of the chemoreceptor. We investigated by characterizing the binding kinetics of CheB to Nanodisc-inserted chemoreceptor dimers. The resulting kinetic and thermodynamic constants revealed a synergy between CheB phosphorylation and pentapeptide binding in which a phosphorylation mimic enhanced pentapeptide binding, and the pentapeptide served not only as a high-affinity tether for CheB but also selected the activated conformation of the enzyme. The basis of this selection was revealed by molecular modeling that predicted a pentapeptide-binding site on CheB which existed only in the open, activated enzyme. Recruitment of activated enzyme by selective tethering represents a previously unappreciated strategy for regulating response regulator action, one that may well occur in other two-component systems.

IMPORTANCE Two-component signal transduction systems are a primary means by which bacteria sense and respond to their environment. Response regulators are key components of these systems. Phosphorylation of response regulators by cognate histidine kinases generate active conformations which act on specific targets, DNA sequences or proteins. The targets have been considered passive in this process. Our characterization of interaction between response regulator CheB and its target chemoreceptor revealed active participation of the target in response regulator action. We found that a pentapeptide sequence at the carboxyl terminus of Escherichia coli chemoreceptors is a selective tether that binds only phosphorylated CheB, thus selecting the form of this two-component enzyme active for covalent modification of the selecting chemoreceptor. Analogous pentapeptides are found on chemoreceptors in many bacterial species and are presumably also selective tethers. There may well be other, uncharacterized examples of active participation of target molecules in response to regulator action.
T hroughout bacterial diversity, methylesterase/deamidase CheB is a key component in the machinery of bacterial chemotaxis (1)(2)(3). It is also a prominent example of a two-component response regulator in which the effector domain is an enzyme (4). It is one of two enzymes that covalently modify bacterial chemoreceptors. It hydrolyzes glutamyl methyl esters formed on chemoreceptors at specific methyl-accepting glutamyl resides by the action of methyltransferase CheR and creates some of the methylaccepting sites by deamidation of glutaminyl side chains. CheB and CheR adjust the extent of receptor modification to match receptor ligand occupancy and in doing so mediate chemotactic sensory adaptation and the sensing of temporal gradients of attractants and repellents (5). Like other response regulators, CheB is activated by phosphorylation of an aspartyl residue on its regulatory domain (6)(7)(8). In Escherichia coli and related organisms, the action of CheB is also enhanced by a pentapeptide, NWETF, at the carboxyl terminus of some of its chemoreceptors (Fig. 1A) (9). The sequence is connected to the receptor body by a 35-residue unstructured flexible arm (10). The pentapeptide binds CheB (11,12) and, when at its native location at the end of the flexible arm, enhances hydrolysis of chemoreceptor methylesters by the enzyme (13,14). Thus, CheB binds not only to substrate side chains on the helical coil of chemoreceptor cytoplasmic domains but also to the carboxyl-terminal pentapeptide. Its action is enhanced not only by phosphorylation but also by the chemoreceptor pentapeptide. In this study, we investigated the interplay among these features. Related carboxyl-terminal pentapeptides are found on ;11% chemoreceptor sequences (.25,000 at the time of the analysis), distributed among many bacterial species and 11 bacterial phyla (15). These carboxyl-terminal sequences presumably bind to cognate CheBs, an interaction that has been documented in the only other species investigated besides E. coli (16).
CheB is a two-domain protein, with an amino-terminal response regulator domain attached by a linker sequence to a carboxyl-terminal catalytic domain (17) (Fig. 1B). Phosphorylation of the response regulator domain by the chemotaxis histidine kinase creates a high-activity form of the methylesterase (6), shifting interaction between the two domains to an "open" conformation that relieves occlusion of the active site by the regulatory amino-terminal domain (18,19). The enzyme binds its substrate chemoreceptor side chains with the surfaces surrounding its active site (20). In contrast, CheB binds the pentapeptide at the junction of the regulatory domain and the linker, a location removed from the active site ( Fig. 1B) (11). These interactions are incompletely defined. The low-activity, i.e., unphosphorylated, form of CheB from E. coli or Pseudomonas aeruginosa binds to the pentapeptide with weak dissociation constants of 130 to 160 mM (21) and ;90 mM (16), respectively, implying that the pentapeptide could not serve as an effective tether to increase the local concentration of the enzyme. Binding of high-activity, phosphorylated CheB to the pentapeptide has not been measured. In addition, binding to substrate sites on the coiled-coil body of a chemoreceptor has not been characterized for either CheB form but must be significantly weaker than binding to the pentapeptide because it was not detectable by assays that detected binding to the pentapeptide (21). There is no information about how activation of CheB by phosphorylation affects interaction with the pentapeptide or about the binding kinetics of either the lowactivity or high-activity form of CheB to either of its sites of interaction with chemoreceptors. The current study was undertaken to provide this information. It led to new insights into the interactions of methylesterase/deamidase with chemoreceptors and identification of a previously unappreciated role for the pentapeptide.

RESULTS
Experimental design. We determined the kinetics of association and dissociation of methylesterase/deamidase CheB in its low-and high-activity form with the E. coli aspartate chemoreceptor Tar using biolayer interferometry (22,23) and isolated receptor homodimers in Nanodiscs (24,25). Apparent rate constants of association and dissociation were derived from global fits of time courses for binding and release (see Fig. S1 in the supplemental material). Equilibrium dissociation constants were calculated from those apparent rate constants.
CheB interactions with the NWETF. We measured binding of the low-activity, unphosphorylated form of CheB to the NWETF pentapeptide tethered to the sensor surface via a linker or in its native position at the carboxyl terminus of Tar4E or Tar4Q (23). Equilibrium dissociation constants were essentially the same for all conditions, 190 6 85 mM for pentapeptide alone, and 230 6 70 and 160 6 30 for the pentapeptide on Tar4E or Tar4Q, respectively (Table 1). Within the error of the measurements, these values corresponded to those previously determined for free pentapeptide binding to unphosphorylated CheB by isothermal titration calorimetry (160 mM) and equilibrium dialysis (130 mM) and for pentapeptide at the carboxyl terminus of TarQEQE in native membrane binding unphosphorylated CheB determined by a pulldown assay (;150 mM equilibrium dissociation constant [K D ]) (21). These results validated our procedure for measuring CheB binding and indicated that binding of the enzyme to pentapeptide-bearing Tar is essentially binding to its pentapeptide. In addition, they showed that the signaling state of the chemoreceptor coiled-coil body did not influence the binding of its pentapeptide to low-activity CheB, consistent with its separation from the receptor coiled-coil body by an unstructured flexible arm (10).
A high-activity form of the methylesterase/deamidase, designated CheB-P, is created by phosphorylation at a specific aspartyl residue in its N-terminal regulatory domain (6)(7)(8). However, activation is short-lived because of a high rate of autodephosphorylation (7), making it difficult to investigate binding of this activated form of the enzyme to chemoreceptors with confidence. Instead, we used a stable, functionally  (26). In this synthetic mimic, a cysteinyl residue substituted for the native, phosphoryl-accepting aspartyl residue is modified by sodium thiophosphate to introduce a stable negative charge at the site of phosphorylation and thus activate the enzyme to essentially the same extent as native phosphorylation (26). We measured rates of association and dissociation of this activated mimic, CheB-P*, for pentapeptide-bearing Tar4Q and Tar4E. To best illustrate enhancements in enzyme-receptor interactions generated by phosphorylation, Fig. 2 displays parameters that increase with increasing affinity. Thus, the figure compares equilibrium association constants K A s, the inverse of K D s, and dwell times, the inverse of dissociation rate constants, as well as association rate constants. Equilibrium dissociation constants for CheB-P* bound to pentapeptide-bearing receptor were 13 6 3 mM, 12-to 18-fold tighter than low-activity CheB ( Table 1). The stronger binding reflected substantial increases in association rate constants ( Fig. 2; Table 1). As was the case for low-activity CheB, kinetic and equilibrium values for highactivity CheB-P* were not significantly affected by the signaling state of the receptor ( Fig. 2; Table 1). As will be considered in detail below, stronger binding to pentapeptide-bearing chemoreceptors of high-activity CheB-P* versus low-activity CheB can be understood as the consequence of the availability of the pentapeptide-binding site only on the open, active conformation of the enzyme.
CheB-receptor interactions in the absence of the NWETF pentapeptide. It had not been possible by standard methods to detect the weak binding of the methylesterase/deamidase to a chemoreceptor lacking the NWETF pentapeptide (9,21). Detecting such weak interactions required an assay that could be performed at the high protein concentrations necessary to detect low-affinity binding. Biolayer interferometry met that requirement (22, 23). Thus, we were able to measure interactions of low-activity TABLE 1 Kinetic and thermodynamic parameters of CheB binding a a Values for k on and k off are averages from at least three independent experiments with standard deviations. Values for K D , K A , and dwell time were calculated from the mean values for k on and k off with standard deviations calculated as the respective propagated error. Tar modif., Tar modification state; Pentap., pentapeptide. b Synthetic phosphorylation using SPO 3 22 coupled to Cys at phosphorylation site to create synthetically activated enzyme (26). c Synthetically phosphorylated enzyme carrying active site mutation S164A. d Catalytic domain of CheB which lacks the regulatory domain and linker.
CheB and high-activity CheB-P* with the coiled-coil chemoreceptor body. We determined kinetic parameters and calculated equilibrium dissociation constants for TarDpp, which lacks the carboxyl-terminal pentapeptide, in the two extremes of modification, the substrate form Tar4QDpp and product form, Tar4EDpp. Interactions of each form with its substrate and product sites on the chemoreceptor body were weaker than the respective interactions with pentapeptide-containing receptor ( Fig. 3; Table 1). High-activity CheB-P* bound to either form of Tar4Dpp with a K D of ;200 mM, 15-fold weaker than to pentapeptide (Fig. 3A). Binding of CheB-P* to substrate and product sites on the receptor body was distinctly different than binding to the pentapeptide, with apparent association rate constants (k on s) approximately ninefold lower and dwell times almost twofold lower ( Fig. 3A; Table 1). There was no statistically significant preference for binding to the substrate (4Q) or product (4E) form of the chemoreceptor ( Fig. 3A; Table 1). Low-activity CheB bound to its substrate and product sites even more weakly with K D s of 590 6 140 mM to Tar4QDpp and 850 6 80 mM to Tar4EDpp ( Fig. 3B; Table 1), values not significantly different (Fig. 3B). Those affinities were approximately fourfold weaker than the respective binding to pentapeptidebearing receptors, with differences in k on and dwell time ( Fig. 3B; Table 1). Interactions with substrate and product forms of the chemoreceptor body of low-activity and highactivity forms of CheB are compared in Fig. S2.
It was possible that high-activity CheB-P* deamidated Tar4QDpp sufficiently during the binding assay to obscure interactions with the substrate form of the receptor. Thus, we tested binding by an enzymatically inactive form CheB-P*-I, which carried the phosphorylation mimic but also had the catalytic serine replaced by alanine. Within the error of the measurements, the inactivating mutation did not alter the equilibrium binding constant for Tar4QDpp or Tar4EDpp (Fig. 4, left panel), although there were modest and in large part compensatory changes in association rate constants and dwell times (Table 1). We conclude that the altered catalytic side chain is not a major contributor to the energy of binding to substrate and product sites on the receptor body and that any enzymatic deamidation occurring over the few minutes of binding measurements did not significantly influence determinations of affinity of activated enzyme for the substrate form of the receptor. In contrast, substitution of the catalytically crucial serine with alanine reduced ;2.5-fold the affinity (K A ) of the enzyme for either modification state of pentapeptide-bearing Tar (Fig. 4, right panel). The reductions were the result of reduced k on s with no statistically significant effect on dwell times FIG 2 Parameters for binding of forms of the methylesterase/deamidase to pentapeptide-bearing chemoreceptor Tar. The panels compare the equilibrium association constants, K A (left), association rate constants, k on (center) and dwell times, the inverse of the dissociation rate constant k off , (right) for binding of the low-activity, unmodified form of CheB (B) and the high-activity, phosphorylationmimic form (B-P*) to intact (pentapeptide bearing) Tar homodimers in the all-glutamyl (4E) or allglutaminyl (4Q) forms inserted in Nanodisc-enclosed native lipid bilayers. The error bars are standard deviations from the mean for at least three independent measurements. Numerical values for the parameters and errors are shown in Table 1. Statistical significance between pairs of conditions was calculated using a two-sample, t test assuming equal variance (*, P , 0.05; ***, P , 0.001; ns, not significant [P $ 0.05]). Tar modifi., Tar modification state.
Selective Recruitment of Activated Response Regulator ® ( Table 1). Reductions in k on s and correspondingly K A s can be understood as a consequence of the serine-to-alanine substitution at the interface between the two CheB domains, shifting the conformational equilibrium toward the closed state and thus reducing the proportion of enzyme in the open conformation in which the pentapeptide-binding site is available.
Interactions of receptor and CheB lacking the regulatory domain and linker. CheB can be activated by deletion of the regulatory domain (6,27), albeit not as effectively as by phosphorylation (8). To investigate the role binding affinity might have in that activation, we measured binding of CheB C , a form of the methylesterase lacking the amino-terminal regulatory domain and linker, to pentapeptide-bearing and pentapeptide-lacking forms of Tar. This truncated protein bound much more weakly to chemoreceptors than the activated phosphorylation mimic CheB-P* but exhibited an approximately twofold preference for the product form Tar4E (K D of 660 6 150 mM) versus the substrate form Tar4Q (1,200 6 100 mM) ( Fig. S3; Table 1). There was no significant difference in binding in the presence or absence of pentapeptide (Table 1), as would be expected for the response regulator missing the pentapeptide-binding site. We conclude that activation of CheB enzymatic activity by deletion of the regulatory domain is not the result of enhanced receptor binding.
Modeling the CheB pentapeptide-binding site. The difference we observed in pentapeptide binding between unmodified CheB and its phosphorylated mimic ( Fig. 2; Table 1) can be understood by postulating that the pentapeptide-binding site is  Table 1. Statistical significance between pairs of conditions was calculated using a two-sample, t test assuming equal variance (*, P , 0.05; **, P , 0.01; ***, P , 0.001; ns, not significant [P $ 0.05]). Tar modifi., Tar modification state.
available in the open but not closed conformation. That site was localized previously to an 11-residue segment at the carboxyl end of the regulatory domain and the first part of the linker (Fig. 1) (11). Specifically, a proteolytic fragment beginning near the amino terminus of a5 and extending through the catalytic domain bound the pentapeptide, but a fragment beginning at the first residue of the linker did not, pentapeptide protected intact CheB from limited proteolysis at position 132, and a peptide from CheB residues 130 through 140 reduced binding of intact enzyme to pentapeptide (11). However, the specific binding mode of pentapeptide was not determined. Since no structure was available for activated, open conformation of CheB, we turned to molecular modeling, focusing on the 11-residue segment, residues 130 through 140. Eight of those residues are in the linker between the two CheB domains. Linkers in response regulators are conformationally flexible. They can be without regular secondary structure or be partly or entirely helical in different crystallographic or functional forms (16,28,29). The CheB linker has little regular secondary structure in the closed form, but secondary structure prediction (30) suggested that helix a5 could extend past its closed conformation termination at residue 132 to include residues through 138 in the linker, thus incorporating what had been nonhelical residues 133 and 134 plus the single helical turn from residues 135 to 138. We modeled the open form of CheB with this extended a5 helix ( Fig. 5B; see Materials and Methods for details) (31). Extended a5 was probed for pentapeptide binding using the protein-peptide docking program MDockPeP (32,33). The top four scoring models (Fig. S4) were assessed for correspondence with previous biochemical characterization of the importance of each NWETF side chain in binding CheB (12). One model among the top four had the crucial interactions with pentapeptide side chains of W2, F5, and E3 ( Fig. 5C and Fig. S4), the three residues most important for binding of CheB to pentapeptide (12). Analysis of phi/psi angles (34) for residues 130 to 139 in this modeled docking were in highly preferred conformations. To confirm the importance of the aromatic side chain interactions in  Table 1. Statistical significance between pairs of conditions was calculated using a two-sample, t test assuming equal variance (*, P , 0.05; **, P , 0.01; ***, P , 0.001; ns, not significant [P $ 0.05]). Tar modifi., Tar modification state.
Selective Recruitment of Activated Response Regulator ® the modeled docking, we calculated docking scores of peptides with alanine in place of W2, F5, or both and found substantial reductions from 267.4 for the native sequence to 243.2, 244.8, or 220.5, respectively, for the respective alanine-substituted pentapeptides. Furthermore, the way in which the pentapeptide bound to extended a5 was strikingly similar to pentapeptide binding determined by crystallography to its other ligand, methyltransferase CheR (PDB accession no. 1BC5 [35]) ( Fig. 6A and B). For pentapeptide binding to both CheB and CheR from Salmonella, (i) the tryptophan side chain at pentapeptide position 2 inserts into a pocket formed by side chains of an arginine and a histidine, (ii) the negatively charged glutamyl side chain at position 3 forms a salt bridge with a positively charged arginine side chain, and (iii) phenylalanine side chain at position 5 packs in a hydrophobic pocket (Fig. S5). Importantly, this set of concerted pentapeptide interactions cannot occur with the closed unphosphorylated conformation of CheB (PDB accession no. 1A2O [17]). In that  Modeling pentapeptide binding to CheB from E. coli revealed plasticity in the binding mode. The pentapeptide-binding region of E. coli CheB differs from Salmonella CheB at three positions, including one directly involved in binding. However, pentapeptide docked effectively on the extended a5 helix of E. coli CheB (Fig. 6C). The side chain of pentapeptide residue F5 packed in a location and orientation similar to that observed for Salmonella CheB. However, there was an amino-terminal shift in the position of pentapeptide on extended a5 of E. coli CheB, and the other two crucial side chain interactions were different from those observed for Salmonella CheB (compare Fig. 6A and C). The shift resulted in the negatively charged side chain of E3 forming a salt bridge with the CheB R128 side chain, rather than R132 as seen for docking to the Salmonella sequence. In addition, in E. coli CheB, position 134 is a serine, not an arginine as in Salmonella, and thus, there was not a pocket into which the W2 side chain could insert. Instead, the W2 side chain packed on the surface of the helix, between residues L135 and H138, creating T-shaped Pi stacking with the histidine. Thus, it appears that the extended a5 helix of CheB provides an effective platform for pentapeptide binding that accommodates side chain variation. This plasticity implies that there may be considerable variation among CheB pentapeptide-binding sites in other species.
We explored this issue by assessing conservation of key residues in CheBs from other bacterial species. We identified 107 containing at least one chemoreceptor with a carboxyl-terminal NWET/SF pentapeptide and a single CheB (see Table S1 in the supplemental material). As illustrated by a sequence logo of CheB residues corresponding to Salmonella CheB positions 128 through 138 (Fig. 7), there was significant conservation of identity or similarity of key binding residues identified for Salmonella or E. coli CheB. Specifically, there was a predominance of positively charged side chains at positions 128, 132, and 134, hydrophobic side chains at position 135, and large side chains at position 138. We conclude that pentapeptide-binding sites we identified in Salmonella and E. coli CheB are significantly conserved, at least in related organisms.

DISCUSSION
Characterization of binding of the methylesterase/deamidase CheB provided new information and insights into interactions of this key chemotaxis enzyme with the chemoreceptors it modifies. Notably, our studies revealed a crucial synergy between FIG 7 Sequence logo of the modeled pentapeptide-binding region generated from alignment of nonredundant (threshold 96%) CheB sequences with identities of .30% to Salmonella CheB from 107 bacterial species with a single CheB gene and at least one chemoreceptor gene coding for a carboxylterminal NWETF or NWESF. Table S1 in the supplemental material shows the sequence alignment from which the logo is derived.
Selective Recruitment of Activated Response Regulator ® enzyme phosphorylation and binding to the pentapeptide tether on the chemoreceptor carboxyl terminus, thereby documenting a previously unappreciated mechanism for controlling the action of a response regulator. Our quantitative measurements of methylesterase/deamidase binding provided a comprehensive view of the multiple interactions of this two-component enzyme, including establishing that the carboxylterminal pentapeptide serves as a high-affinity tether for the activated methylesterase. The measurements showed that differences in affinity of high-and low-activity CheB were primarily differences in rate constants of association, not dissociation, suggesting that the two differed in the balance between open and closed forms of the two-component enzyme.
CheB interactions with its chemoreceptor substrate. The data summarized in Table 1, Fig. 2 to 4, and Fig. S2 and S3 in the supplemental material provide a comprehensive picture of interactions of CheB with its chemoreceptor substrate. Key to interpreting the data is the notion that the enzyme is in equilibrium between a closed, enzymatically inactive conformation and an open, enzymatically active conformation and that the equilibrium is shifted from predominately closed to predominately open by phosphorylation (8). The differences observed in pentapeptide binding between unmodified CheB and its phosphorylated mimic can be understood as reflecting the absence of the pentapeptide-binding site in the closed conformation but its presence in the open conformation. Thus, the apparently low-affinity binding of pentapeptide by unmodified, low-activity CheB would be the result of a small proportion of unmodified enzyme in the open, binding-competent state. This notion is supported by comparison of kinetic parameters for pentapeptide binding by unmodified, low-activity and phosphorylated, high-activity forms of the enzyme. There is little difference between the forms in dissociation rate constants (and thus corresponding dwell times), but there is a major difference in association rate constants ( Fig. 2; Table 1). If only the open form binds pentapeptide, then dissociation rate constants and corresponding dwell times determined for unmodified and phosphorylated populations of CheB should reflect dissociation of the same complex and thus be similar. This is exactly what we observed. If only the small proportion of unmodified CheB is in the active, binding-competent conformation, then the association rate constant, calculated using total enzyme concentration, would be significantly lower than for phosphorylated enzyme, for which a much larger proportion is in the binding-competent conformation. Correspondingly, the apparent equilibrium binding constant would appear much weaker for unmodified CheB, as it does ( Fig. 3; Table 1). If phosphorylated CheB were 100% open, the proportion of unmodified enzyme in the open state would correspond to the ratio of apparent equilibrium binding constants for the two forms or ;7% of the unmodified enzyme in the open state (Table 1).
Our data identified the pentapeptide as a high-affinity tether for the open form of CheB, as had been previously shown for CheR (36). The association rate constant for binding of CheB-P* to receptor-borne pentapeptide is approximately 10-fold higher than the corresponding constant for binding to the receptor body (Table 1). Thus, in large part, the activated enzyme would bind first to pentapeptide and only subsequently to substrate and product sites on the receptor body. That subsequent binding would occur with high probability because tethering by pentapeptide would increase the local concentration of active CheB to over 1,000 mM (23), approximately fivefold above the equilibrium dissociation constant. The ;3-s dwell times on the pentapeptide would allow sufficient time for diffusional collision of the tethered enzyme to a substrate site on the receptor body.
Synergy between phosphorylation and pentapeptide binding. The notion that only the open CheB has the pentapeptide-binding site is strongly supported by our modeling of the binding side region and of its binding to pentapeptide (Fig. 5). Specifically, the modeled pentapeptide-binding site does not exist in the closed conformation of CheB. A prominent feature of that site is a pocket for the pentapeptide tryptophan. That tryptophan is a major contributor to the energy of interaction between the response regulator and pentapeptide (12). The pocket is formed by the side chains of His138 and Arg134. However, in the closed conformation, the histidine side chain of residue 138 is substantially buried in the interface between the regulatory and catalytic domains and thus not available to be part of the tryptophan-binding pocket (Fig. S6). In addition, in the closed conformation, the side chains of His138 and Arg134 are distant from each other and point in opposite directions (Fig. 5, top center  panel). Thus, the modeled pentapeptide-binding site does not exist in the closed conformation of CheB. Instead, it is formed by the postulated helical extension that is part of the conformational change from closed to open.
Notably Taken together, kinetic and modeling data make a compelling case that the pentapeptide binds only to the open form of CheB, not the closed form. This implies that the pentapeptide at the carboxyl terminus of a chemoreceptor in essence selects the open, enzymatically active conformation of CheB from the population of closed and open molecules. This selective binding ensures that only enzymatically active response regulator is tethered in proximity to its substrate sites on the chemoreceptor coiledcoil body, and thus poised to modify those sites. We characterized this selective binding in vitro, but in vivo CheB localization in chemoreceptor complexes is phosphorylation dependent, implying that the same selection occurs in the cell (37).
A common feature of many signaling systems is "recruitment" of signaling proteins into complexes by specific interaction with another protein in the complex. Our characterization of interactions of CheB and receptor-borne pentapeptide provides an example of such recruitment in a bacterial chemotaxis system. CheB recruitment has the important additional feature of being specific for activated protein. We suggest other chemoreceptors carrying related carboxyl-terminal sequences perform analogous selective tethering of their cognate CheB. Furthermore, selective tethering of activated response regulators may well be a feature of two-component signaling systems in addition to those that mediate chemotaxis. More widely, selection of activated conformations of signaling proteins by selective tethers could be an unappreciated aspect of many signaling systems.

MATERIALS AND METHODS
Strains and plasmids. The host for plasmids carrying cheB or a chemoreceptor gene was E. coli K-12 strain RP3098 (38), which produces neither chemoreceptors nor Che proteins. Genes coding for Tar4E L378C, Tar4Q L378C, Tar4EDpp-6H L378C, and Tar4QDpp-6H L378C were carried on plasmids pML42, pML46, pML36, and pML44, respectively (23). pCW/cheB, carrying E. coli cheB under the control of tandem tac promoters (39) was from F. W. Dahlquist (University of California, Santa Barbara, Santa Barbara, CA). pAL77 is a derivative of pCW/cheB in which the segment of cheB coding for residues 2 to 152 was deleted by PCR-based mutagenesis to code for CheB C . A plasmid carrying the gene coding for CheB D56C/C207S/C309S from Salmonella was from Ann Stock (Rutgers -Robert Wood Johnson Medical School, Piscataway, NJ). PCR-based site-directed mutagenesis of that plasmid generated pML47, coding for Salmonella CheB D56C/S164A/C207S/C309S.
Binding assays. Kinetics of CheB association and dissociation with Nanodisc-embedded Tar biotinylated at L378C were determined using biolayer interferometry (22,23) as measured by a BLItz instrument (ForteBio, Menlo Park, CA, USA) and described in reference 23. Representative time course experiments are shown in Fig. S1 in the supplemental material. Measurements for CheB-P* and CheB-P*-I were in 50 mM Tris-HCl (pH 7.5), 10% (wt/vol) glycerol, 5 mM MgCl 2 , 50 mM KCl, and 2% (wt/vol) bovine serum albumin (BSA). Measurements for CheB and CheBc were in the same solution plus 2 mM dithiothreitol (DTT).
Fitting binding data. Using BLItz software, response curves were fit globally to a 1:1 binding model to yield apparent association rate constants (k on ) and dissociation rate constants (k off ). For analysis of complex patterns of binding, in which more than one binding reaction appeared to be contributing, for instance one that appeared saturable and another nonspecific, nonsaturable, the data were exported and fit globally to nonlinear exponential functions using OriginPro 2017. The best fits to the data were obtained using the following functions.
R ¼ R 0 1 R eq a e 2kd1 t ð Þ 1 b e 2kd2 t ð Þ ½ was used to globally fit dissociation time courses, with R being the response (in nanometers), a 1 b = 1, with a and b being the portion of the signal (in nanometers) for fast and slow phases, respectively, k d1 and k d2 being the respective dissociation rate constants, and R eq being the response at equilibrium. Association data were fit using apparent dissociation rate constants k d1 and k d2 , relevant CheB concentrations (C) and the equation R ¼ R 0 1 R eq a 1 2 e 2ðka1 C1kd1Þt ð Þ 1 b 1 2 e 2ðka2 C1kd2Þt ð Þ ½ Molecular modeling and pentapeptide docking. MODELLER (31) was used to model the CheB regulatory domain with an extended helix a5 from residues 133 through 140, using the crystal structure of response regulator, FixJ (PDB accession no. 5XSO [28]), in which the entire linker sequence is helical and the regulatory domain has a 52% similar sequence to the CheB regulatory domain, thus facilitating superpositioning. The FixJ regulatory domain was superimposed on the regulatory domain of Salmonella CheB (PDB accession no. 1A2O) and CheB a5 extended on the template of FixJ residues 121 to 129. The remainder of the CheB regulatory domain was built using the template of its closed conformation. Figure 5B shows this modeled domain and extended linker placed in an arbitrary open position relative to the catalytic domain.
The NWETF pentapeptide was docked onto the extended a5 structure using MDockPeP (32,33). The initial peptide conformation was extracted from the crystal structure of the pentapeptide-CheR complex (PDB accession no. 1BC5). The center of the docking box was set to the geometric center of the segment from positions 130 to 140 and its size to 25 Å to cover the entire binding region. Other parameters were default values. Constraints maintained the peptide backbone in its initial conformation and pentapeptide side chains W2, E3, and F5 within 5 Å of the segment from positions 130 to 140. Pentapeptide side chains were freely flexible. The top four models for Salmonella CheB (Fig. S4) had ITScorePeP energy scores of 280.8, 276.9, 268.3, and 267.4. Considering the indeterminacy in these scores, we considered these models equally likely.
Sequence alignment. The NCBI-BLAST program version 2.6.01 (45) was used to search the UniProt database release 2021_03 (45)(46)(47) for chemoreceptors with a carboxyl-terminal NWETF or NWESF and to identify bacterial species with at least one such chemoreceptor and a single CheB. Clustering (48) and multiple sequence alignments (49) were performed for Salmonella CheB and CheBs in other species with sequence identity of .30% and a threshold of 96% to avoid redundant sequences. Table S1 in the supplemental material shows a portion of this alignment, including residues aligned with Salmonella CheB residues 132 to 138. Using this alignment, a sequence logo (Fig. 7) was generated by WebLogo (50).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.