ABSTRACT
In bacterial chemotaxis, chemoreceptors in signaling complexes modulate the activity of two-component histidine kinase CheA in response to chemical stimuli. CheA catalyzes phosphoryl transfer from ATP to a histidinyl residue of its P1 domain. That phosphoryl group is transferred to two response regulators. Receptor control is almost exclusively at autophosphorylation, but the aspect of enzyme action on which that control acts is unclear. We investigated this by a kinetic analysis of activated kinase in signaling complexes. We found that phosphoryl transfer from ATP to P1 is an ordered sequential reaction in which the binding of ATP to CheA is the necessary first step; the second substrate, the CheA P1 domain, binds only to an ATP-occupied enzyme; and phosphorylated P1 is released prior to the second product, namely, ADP. We confirmed the crucial features of this kinetically deduced ordered mechanism by assaying P1 binding to the enzyme. In the absence of a bound nucleotide, there was no physiologically significant binding, but the enzyme occupied with a nonhydrolyzable ATP analog bound P1. Previous structural and computational analyses indicated that ATP binding creates the P1-binding site by ordering the “ATP lid.” This process identifies the structural basis for the ordered kinetic mechanism. Recent mathematical modeling of kinetic data identified ATP binding as a focus of receptor-mediated kinase control. The ordered kinetic mechanism provides the biochemical logic of that control. We conclude that chemoreceptors modulate kinase by controlling ATP binding. Structural similarities among two-component kinases, particularly the ATP lid, suggest that ordered mechanisms and control of ATP binding are general features of two-component signaling.
IMPORTANCE Our work provides important new insights into the action of the chemotaxis signaling kinase CheA by identifying the kinetic mechanism of its autophosphorylation as an ordered sequential reaction, in which the required first step is binding of ATP. These insights provide a framework for integrating previous kinetic, mathematical modeling, structural, simulation, and docking observations to conclude that chemoreceptors control the activity of the chemotaxis kinase by regulating binding of the autophosphorylation substrate ATP. Previously observed conformational changes in the ATP lid of the enzyme active site provide a structural basis for the ordered mechanism. Such lids are characteristic of two-component histidine kinases in general, suggesting that ordered sequential mechanisms and regulation by controlling ATP binding are common features of these kinases.
INTRODUCTION
Two-component signaling systems are abundant mediators of environmental sensing and response in bacteria and archaea (1–3). In these systems, a histidine kinase uses ATP to autophosphorylate one of its own histidinyl residues, and the phosphoryl group on the histidinyl side chain is transferred to an aspartyl residue of a cognate response regulator. The modified response regulator then mediates a response. The essence of two-component signaling is control of kinase autophosphorylation and thus of the kinetics of that reaction (1, 4). However, kinetic characterization of histidine kinase autophosphorylation has been limited. For instance, there is no information about the pattern or order in which substrates bind and products are released, information that can provide insights into the control mechanisms. In large part, this lack of information reflects a need in most analyses for enzyme kinetics to vary the concentration of each substrate and each product independent of the enzyme. That requirement is not easily met for the large family of class I sensor histidine kinases because one of the substrates, the phosphoryl-accepting histidine, is located on the dimerization domain of those kinases and dimerization is a prerequisite for enzyme activity (1–3). Thus, severing the connection between the phosphoryl-accepting dimerization domain and the catalytic domain results in an inactive enzyme. However, among class II histidine kinases, those that mediate bacterial chemotaxis, the phosphoryl-accepting histidine is on a domain distinct from the kinase dimerization domain (5). That phosphoryl-accepting domain can be detached from the rest of the kinase and still function as a substrate whose concentration can now be varied (5–7). Using this approach, we were able to utilize the well-characterized chemotaxis kinase CheA from Escherichia coli to investigate the pattern and order of substrate binding and product release in a histidine kinase.
CheA is the two-component histidine kinase that is the signaling enzyme in bacterial chemotaxis (8–11). As in other two-component signaling systems, the central process in chemotactic signaling is the control of the enzymatic activity of that homodimeric enzyme (4). Control is exerted by bacterial chemoreceptors in signaling complexes with the kinase and coupling protein CheW (8–11). The CheA catalytic domain mediates the phosphoryl transfer from ATP to a histidinyl residue of the CheA P1 domain (Fig. 1). In turn, that phosphoryl group is transferred to two response regulators, one that regulates motility and the other sensory adaptation. Almost all chemoreceptor-mediated control is focused on the initial phosphoryl transfer from ATP to P1, not on the subsequent transfer to a response regulator (12).
Chemotactic phosphoryl transfer. The cartoon depicts a core chemotaxis signaling complex consisting of two trimers of chemoreceptor homodimers (blue with black dots marking the positions of methyl-accepting sites), two copies of coupling protein CheW (yellow), and a homodimer of the histidine autokinase CheA (pink). The CheA catalytic domain P4 transfers the γ-phosphoryl group of ATP to the CheA P1 domain. It is subsequently transferred to the response regulator CheY (green). The experiments described in this study characterized the kinetics of the first step by using P1 that was genetically liberated from its usual attachment to the rest of CheA and signaling complexes that contained the P3-P4-P5 portion of CheA. Those components are enclosed in the dashed red line. The components not present in our assays are shown in shadow format outside the line.
CheA is an active kinase only in its dimeric form (13). Each protomer contains five domains connected by linker sequences (7, 14) (Fig. 1). The P1 domain carries the histidinyl residue that is the substrate for the CheA autophosphorylation reaction (15, 16). P2 binds CheY, which conveys signals to the flagellar motor and methylesterase CheB, an enzyme of sensory adaptation, to position the response regulators near phosphoryl-P1 and thus facilitate phosphoryl transfer (17). P1 and P2 are connected by an unstructured linker sequence (18, 19). P3 is an α-helical hairpin that dimerizes with its partner hairpin in the companion protomer of the CheA homodimer to form a four-helix bundle and thus the homodimer (14). It is connected to P2 by a second unstructured linker and to P4 by a short linker (18). P4 is the kinase domain that binds the phosphoryl transfer substrates ATP and P1 and catalyzes P1 phosphorylation (14). Domain P5 is a paralog of CheW (14, 20) that binds to CheW and to chemoreceptors (21). Isolated CheA dimers have low kinase activity but are activated in signaling complexes as much as 1,000-fold (22). The core signaling complexes consist of two trimers of chemoreceptor homodimers, each interacting with a protomer of the dimeric kinase and with a copy of the coupling protein CheW (Fig. 1) (22–24). The core complexes polymerize via CheW-P5 interactions to create hexagonal arrays that can include hundreds of core units (23–26). Small arrays of signaling complexes can be assembled in vitro by mixing CheA and CheW with chemoreceptors in native cytoplasmic membrane vesicles isolated from cells containing high levels of the chemoreceptor (27, 28).
Recent Michaelis-Menten analysis using liberated P1 domains and the P3-P4-P5 portion of CheA incorporated into signaling complexes revealed that signaling-generated changes in kinase activity occurred primarily at the level of the multistep catalytic mechanism and not by changing the Michaelis constant, Km, for ATP or P1 (12). The results indicated that sequestration of P1 at a binding site on P4 distinct from the active site (29) was not a major contributor to kinase control. Mathematical modeling of that kinetic data implicated ATP binding and either P1 binding or phosphoryl transfer as the two targets of kinase control by the respective allosteric effectors, attractant ligand, and receptor adaptational modification (30). The deduced changes in ATP binding involved equal effects on association and dissociation rate constants and thus were consistent with the observed lack of significant effects on Km.
We now report the identification of the nature of the kinetic mechanism of the kinase CheA incorporated into signaling complexes and test crucial implications of that mechanism by direct assays of substrate binding to the enzyme. These results provide biochemical and functional evidence for the central role of ATP binding as a key control point for kinase activity in this histidine kinase and by structural analogy for histidine kinases in general.
RESULTS
Experimental design.We assembled signaling complexes using the P3-P4-P5 portion of CheA and CheW plus Tar, the aspartate chemoreceptor of E. coli, inserted in native cytoplasmic membrane vesicles. The receptor was in its gene-encoded modification state of two glutaminyl and two glutamyl residues at the four methyl-accepting sites (Tar-QEQE) and thus was balanced between the kinase-off and kinase-on signaling states (10, 11). Such preparations form active signaling complexes (12) that likely make small arrays at the membrane, as do similar complexes assembled with full-length CheA (28). An assembly of signaling complexes with the P3-P4-P5 portion of CheA allowed us to perform kinetic experiments in which we could vary each substrate of the phosphoryl transfer reaction, ATP, and the liberated P1 domain (5, 12, 31).
Determining the relationship between substrate binding and product release.For two-substrate, two-product reactions like the one catalyzed by the chemotaxis kinase, there are three possible patterns in which substrates are bound and products are released. In the sequential pattern, both substrates bind prior to any chemical transformations and prior to the release of either product (32). In the ping-pong pattern, one substrate binds, a chemical transformation occurs, and the first product is released, followed by binding of the second substrate, a chemical transformation, and release of the second product (32). In the rapid-equilibrium random bi-bi pattern, each substrate is in rapid equilibrium with the enzyme; once both bind, a chemical transformation can occur and the two products are released in a random order. The three patterns can be distinguished by steady-state kinetic experiments in which initial rates of reactions are measured as a function of the concentration of one substrate at several fixed concentrations of the other substrate (32). We performed such experiments for the phosphoryl transfer from ATP to the liberated CheA P1 domain that was catalyzed by the CheA P3-P4-P5 domain, which was activated by incorporation into signaling complexes. Figure 2A and B display the resulting data and corresponding curves generated by global fits of the entire data set. Transforming the data and the fits to plots of the inverse of initial velocity (1/v) versus the inverse of the concentration of the varied substrate (1/[S]) revealed patterns of straight lines intersecting to the left of the ordinate (Fig. 2C and D), not a set of parallel lines. The intersection of those lines identifies the pattern of substrate binding and product release as sequential, not ping-pong (32). Thus, both ATP and P1 bind to the enzyme prior to the formation and release of either product. For the reciprocal plots, replots of the slopes or the intercepts versus the inverse of the substrate held at several fixed concentrations were linear (see Fig. S1 in the supplemental material), indicating that each substrate adds to only one form of the enzyme (32). The rapid-equilibrium random bi-bi mechanism will be considered below. The fits to the individual kinetic plots shown in Fig. 2A and B are the result of a global fit to Michaelis-Menten equations for two-substrate, two-product reactions for all of the data underlying the two plots. This fit generated values for
Kinetics of phosphoryl transfer as a function of substrate concentrations. (A and B) Mean initial velocities of phosphoryl transfer from ATP to the liberated P1 domain, catalyzed by the P3-P4-P5 portion of CheA incorporated into signaling complexes, as a function of ATP concentration at the indicated fixed concentrations of P1 (A) and as a function of P1 concentration at the indicated fixed concentrations of ATP (B). (C and D) Double-reciprocal plot of the data in panels A and B, respectively. The lines in panels A and B are global fits of the data in both panels to Michaelis-Menten relationships. Those in panels C and D are double reciprocals of the same fits. Error bars are standard deviations of the measured initial velocities from least three independent experiments. The fits in panels C and D converge at points to the left of the ordinate, indicating a sequential rather than a ping-pong reaction mechanism.
Kinetic parameters of phosphoryl transfer from ATP to the P1 domain catalyzed by the P3-P4-P5 domains of CheA incorporated into signaling complexesa
Determining the temporal order of substrate binding and product release.In sequential mechanisms for two-substrate, two-product reactions, substrates and products might bind and be released in a random or ordered manner. The possibilities can be distinguished by patterns of inverse plots of product inhibition generated when the concentration of one substrate is held constant and the concentration of the other substrate is varied in the presence of several concentrations of a product inhibitor (37). In random sequential mechanisms, for each of the four combinations of product inhibitor and varied substrate, linear fits of the data will intersect on the 1/v axis, which is diagnostic of competitive inhibition in which inhibition disappears at an infinite concentration of the varied substrate. For ordered sequential mechanisms, only one combination will exhibit competitive inhibition. The other three will show mixed inhibition in which the linear fits intersect to the left of the 1/v axis. To determine the case for the physiologically relevant state of CheA, we performed a series of experiments characterizing product inhibition for the activated P3-P4-P5 catalytic portion of CheA incorporated into signaling complexes. Initial rates of P1 phosphorylation were measured as a function of the concentration of either ATP or P1, in the presence of a fixed concentration of the other substrate, P1 or ATP, and in the presence of several concentrations of a product inhibitor, either ADP or phosphoryl-P1 (P1-P). Figure 3 shows plots of the resulting four sets of data with respective global fits. Table 1 displays the respective inhibition constants derived from the fits. The nature of the inhibition, competitive, uncompetitive, or mixed, is illustrated graphically in linearized plots of 1/v versus 1/(varied substrate) (Fig. 4). Of the four patterns of inhibition, only inhibition by ADP in experiments in which ATP was the variable substrate showed competitive inhibition, i.e., intersection at a common 1/v value on the ordinate. The other three combinations of product inhibitor and varied substrate exhibited patterns characteristic of mixed inhibition, in which inhibitor changes both the intercept and slope of the linearized fits. These patterns of inhibition identify the phosphoryl transfer from ATP to P1 catalyzed by CheA incorporated into signaling complexes as an ordered sequential reaction, with the mandatory order of substrate binding and release illustrated in Fig. 5. Specifically, of the two substrates and two products, only ATP and ADP bind to the free enzyme, a pattern reflected in the action of ADP but not P1-P as a competitive product inhibitor for ATP. Plots of intercepts and slopes of the respective inverse plots as a function of inhibitor concentration revealed linear relationships (see Fig. S2 in the supplemental material), indicating that each product inhibitor bound to a single form of the enzyme, as would be predicted by the ordered sequential mechanism (37).
Kinetics of product inhibition. The panels show mean initial velocities and standard deviations of phosphoryl transfer as in Fig. 2 as a function of ATP concentration (A and C) or P1 concentration (B and D) at the indicated fixed concentrations of the product inhibitors ADP (A and B) or phosphoryl-P1 (P1P) (B and D). The concentrations of the respective fixed substrates were 200 μM P1 (A and C), 1,000 μM ATP (B and D). The lines reflect respective global fits of the data in each panel. The errors bars are standard deviations from the mean for at least three independent experiments.
Double-reciprocal plots of product inhibition. The panels display double-reciprocal plots of the data and fits shown in Fig. 3. The fits in panel A converge on the ordinate, indicating competitive inhibition of ATP by ADP. The fits in panels B to D converge at points to the left of the ordinate, indicating mixed inhibition.
The deduced ordered sequential mechanism of CheA-catalyzed phosphoryl transfer from ATP to the P1 domain. In this mechanism, only ATP and ADP bind to the same form of P4, the otherwise unoccupied catalytic domain of CheA, and P1 binds only to the ATP-occupied enzyme.
In the mandatory order (Fig. 5), ATP binds to the unoccupied P4 active site, followed by binding of the P1 domain to the P4-ATP complex. After the phosphoryl transfer from ATP to P1, P1-P release from the two-product-enzyme complex necessarily precedes the release of ADP. This mandatory order has two features particularly relevant to control of the kinase, namely, (i) P1 does not bind to the free enzyme and (ii) P1 binds only to the enzyme in complex with ATP.
P1 binding to the P3-P4-P5 portion of CheA.We performed direct biochemical tests of these two key predictions.
In the absence of nucleotides, there was only a minimal association of free P3-P4-P5 to a P1-decorated sensor tip (Fig. 6). This result confirmed the prediction of no significant interaction between P1 and apo-P3-P4-P5. Specifically, exposure of the P1-decorated tip to a solution of P3-P4-P5 at concentrations from 100 to 800 μM resulted in a low-magnitude, gradual increase in signal that did not saturate over a 2-minute time course, a period over which reversible protein-protein interactions would be expected to saturate (39). In addition, the signal diminished only gradually and incompletely over a 2-minute time course following replacement of the P3-P4-P5 solution with buffer alone (Fig. 6). The lack of saturation kinetics and incomplete dissociation are features of a nonspecific protein association, for instance, an interaction with the surface of the sensor tip. They are not features of reversible binding to the protein with which the tip is decorated. Consistent with this notion, removal of the low concentration of surfactant sodium cholate from buffers used in the binding experiments resulted in an increase in the nonsaturating signal and a greater extent of incomplete dissociation. We conclude that the signals observed in solutions lacking nucleotides reflected nonphysiological interactions of P3-P4-P5, not reversible binding relevant to enzyme action. In contrast, in the presence of 10 mM AMPPCP, there was a significant and largely saturable association of P3-P4-P5 to P1-decorated sensor tips, as well as rapid and almost complete dissociation upon changing the surrounding solution to one lacking nucleotides (Fig. 6). Thus, P1 bound to the trinucleotide-occupied form of P3-P4-P5 in a pattern characteristic of physiologically relevant protein-protein interactions, fulfilling the second prediction of the kinetic analysis. Global fitting of the data generated an association rate constant of 340 M−1 s−1, a dissociation rate constant of 0.06 s−1 and a calculated equilibrium dissociation constant of 180 μM. That
Binding of P3-P4-P5 to surface-immobilized P1. The biolayer interferometry traces show time courses of binding of the P3-P4-P5 portion of histidine kinase CheA to a sensor tip on which was coupled the liberated P1 domain of the enzyme. Binding was tested at 100, 200, 600, and 800 μM P3-P4-P5 in the presence (+AMPPCP) or absence (no AMPPCP) of the nonmetabolizable ATP analog AMPPCP at 10 mM. Binding signals below zero at the end of the dissociation phase for the +AMPPCP conditions are artifacts of realignment of dissociation traces that adjusted for a slight jump in signal magnitude at the beginning of the dissociation phase upon shifting the sensor tip from a solution containing P3-P4-P5 to containing buffer alone. The adjustments facilitated global fitting of the data. These artifactual increases in signal are seen in the “no AMPPCP” traces, for which no adjustments were made.
DISCUSSION
Our kinetic analysis found that the phosphoryl transfer from ATP to the CheA P1 domain catalyzed by the activated chemotaxis kinase incorporated into signaling complexes is an ordered sequential reaction, in which binding of ATP is the required first step (Fig. 5). The ordered sequence predicted that the P1 would bind to the ATP-occupied catalytic domain but not to the apo-enzyme. Direct binding assays showed that this prediction was fulfilled (Fig. 6). These results establish that there is a required order of substrate binding to the CheA catalytic domain. Our earlier analysis of the effects of chemoreceptor ligand occupancy and adaptational modification on the kinetics of CheA autophosphorylation (12) demonstrated that almost all of the control exerted on the kinase by chemoreceptors in signaling complexes occurred at the initial, autophosphorylation stage of phosphoryl transfer, not at the later stage of transfer from phosphoryl-P1 to CheY. Mathematical modeling of those data identified ATP binding as a crucial target of this control (30). We have now found that ATP binding is the necessary first step in an ordered sequence of substrate binding for autophosphorylation of CheA. Thus, the control of the chemotaxis kinase by chemoreceptors follows the general pattern for the regulation of biological processes in which control focuses on the first dedicated step. Taken together, our previous and current analyses of the chemotaxis kinase in signaling complexes identify ATP binding as the prime target for kinase control.
A structural basis for an ordered sequential mechanism.Structural and computational observations, summarized in Fig. 7, provide molecular insight into the origin of the ordered sequential mechanism defined by our kinetic characterization. Three-dimensional X-ray structures of a slightly truncated P4 catalytic domain of CheA from Thermotoga maritima revealed that the chemotaxis kinase, like many other histidine kinases, had an “ATP lid” (40, 41). In the absence of nucleotides, the ATP lid is a loop devoid of a regular secondary structure (Fig. 7, top cartoon and inset). When the catalytic domain is occupied by the nonmetabolizable ATP analog AMPPCP, part of the ATP lid becomes a structured alpha helix that closes over the nucleotide and completes its binding site (Fig. 7, top right cartoon and inset). Computational simulations and molecular dynamics analysis of ATP and P1 binding using these structures revealed that the closed, helical form of the ATP loop creates a docking site for P1 in which the phosphoryl-accepting histidinyl side chain is well positioned for phosphoryl transfer (40, 41) (Fig. 7, bottom right cartoon and inset). These observations led to the suggestion that nucleotide binding to the kinase active site generates a concerted conformational change in which ATP binding is completed as the helical ATP lid forms, which in turn creates a functional binding site for P1 (41). Such concerted changes and the absence of a P1-binding interface without nucleotide binding would create a necessary order of substrate binding in which ATP would have to bind to the enzyme before P1 could bind and P1-P would have to be released before ADP could be released (Fig. 7, bottom and top left cartoons and insets). This is precisely the required order of binding and release in the ordered sequential mechanism defined by our kinetic analysis. Conversely, our kinetic and biochemical identification of an ordered sequential mechanism for E. coli CheA activated by incorporation into signaling complexes provides functional relevance to the structural and computational observations made using an isolated portion of a CheA characterized in the absence of its interaction partners.
Structural changes corresponding to the ordered sequential kinetic mechanism of CheA autophosphorylation. In the center of the figure, the ordered sequential mechanism shown in Fig. 5 is displayed as a cycle. Structures corresponding to the states of the P4 catalytic domain and its substrates and products at the respective stages of the mechanism are shown surrounding the cycle, with insets providing a magnified view of the active site. Structures for P4 (PDB accession code 1B3Q), P4-ATP (1I58), and P4-ATP-P1 (1I58 and 1I5N) are based on those provided in references 40 and 41. Structures for P4-ADP-P-P1 and P4-ADP were extrapolated from structures in those references.
Functional correlates of an ordered sequential mechanism.In considering possible mechanisms for the control of kinase activity in chemotaxis signaling complexes, a common suggestion has been that control is exerted by restricting substrate access to the kinase active site (11, 29, 42–45). Restricting access of P1 by binding to a secondary, noncatalytic site on the P4 domain was an attractive possibility since there was evidence for such a site (29). However, simple sequestering of P1 from access to the active site as a central control mechanism was excluded by demonstration of effective kinase control in experiments using a liberated P1 domain at concentrations that would saturate sequestering sites on the P3-P4-P5 catalytic segment incorporated into signaling complexes (12). Instead, if the mechanism of control involves restricted access, it is likely that physical restriction of the active site limits binding of ATP. In fact, recent characterization of the organization of arrays of signaling complexes by electron tomography, computational model building, and molecular dynamics, as well as mutational analysis, has identified alternative orientations of the P4 domain relative to P3 and the membrane-distal tips of chemoreceptor trimers (42–44, 46, 47). These orientations could well represent situations in which the ATP-binding site on P4 is accessible or blocked as a result of direct physical interference or allosteric alteration of the site.
Kinetic mechanisms of sensory histidine kinases.ATP lids are present in the many class I histidine kinases, for which atomic-resolution, three-dimensional structures have been determined (1, 4). Thus, the large family of class I histidine kinases, as well class II histidine kinases, contain the structural features for an ordered sequential kinetic mechanism. We suggest that the kinetic order of substrate binding and product release that we documented for a class II chemotaxis histidine kinase is a general feature of histidine kinases, and thus, the control of kinase activity is likely to focus on binding of the first substrate, ATP.
MATERIALS AND METHODS
Proteins and signaling complexes.CheW, the P1 domain of CheA (residues 1 to 134), with six histidines on its carboxyl terminus, and the P3‐P4‐P5 portion of CheA (residues 261 to 654) (48) from E. coli were produced and purified essentially as described (12). Native cytoplasmic membranes were isolated from cells producing high levels of the E. coli aspartate chemoreceptor Tar with a six‐histidine carboxyl‐terminal extension essentially as described (49). Proteins were dialyzed into 50 mM Tris (pH 7.5), 100 mM NaCl, 0.5 mM EDTA, and 2 mM dithiothreitol (TNED). Protein concentrations were determined by quantitative immunoblots using purified standards for the respective proteins for which concentrations had been determined by quantitative amino acid analysis.
Signaling complexes were formed by incubating 5 μM CheW, 10 μM Tar‐QEQE in native membrane vesicles, and 2 μM P3‐P4‐P5 as described (12, 22). Free P3‐P4‐P5 was removed by two rounds of centrifugation and suspension in TNED (50) to a final volume equal to that of the formation mixture, except for product inhibition experiments for ADP with ATP as the variable substrate and all experiments in which P1-P was the inhibitor, for which the final volume was 20% of the formation mixture volume.
Production and purification of P1-P.P1 with a six-histidine carboxyl-terminal extension was phosphorylated using core signaling complexes formed by mixing 4 μM Nanodisc-inserted Tar-6H in the all-glutamine modification state (Tar-QQQQ) and at approximately five dimers per disc, 2 μM CheA, and 5 μM CheW (12). A total of 800 μM purified P1 in TNED containing 50 mM KCl, 5 mM MgCl2, 2 mM ATP, 7.5 mM phosphocreatine, and 37 U/ml creatine phosphokinase was added. The mixture was incubated 35 min at room temperature and the reaction terminated by the addition of 20 mM EDTA. The buffer was changed to 50 mM Tris (pH 7.5), 100 mM NaCl, and 15 mM imidazole by five rounds of concentration and 10-fold dilutions of the concentrated volume using the Amicon Ultra-4 centrifugal concentrator. The mixture was applied to an Ni-nitrilotriacetic acid (NTA) column, the column was washed with 15 mM imidazole in the same solution, histidine-tagged components were eluted with 250 mM imidazole, and relevant fractions were pooled. The buffer of the pooled material was changed to 20 mM Bis-Tris (pH 6.2) and 2 mM dithiothreitol (DTT) and applied to a quaternary aminoethyl (QAE) ion‐exchange column (Shodex QA‐825) mounted on an high-pressure liquid chromatography (HPLC) apparatus (Gilson, Middleton, WI). The column was washed with 20 mM Bis-Tris (pH 6.2) and 2 mM DTT and eluted with a gradient of 200 to 300 mM NaCl in the same buffer. Fractions were analyzed by polyacrylamide gel electrophoresis for the presence of P1, which was eluted in two peaks. Each peak was pooled and analyzed by mass spectrometry for the proportion of the protein that was phosphorylated by applying a 10-pmol/μl P1 sample in 1% formic acid to a C8 protein chip column (product number G4240-63001 SPQ105; Agilent) coupled to an Agilent 6520 quadrupole time of flight (QTOF) mass spectrometer. In the later-eluting peak, 77% of P1 was phosphorylated, which was over twice the percentage phosphorylated in the earlier-eluting peak. Thus, we used the protein in the later-eluting peak for product inhibition experiments, using the 77% phosphorylated value to calculate the appropriate amount to add for the desired concentration of phosphoryl-P1 and adjusting the amount of unmodified P1 added to reach the desired concentration of P1.
Kinase assays.Steady-state kinetics of P1 phosphorylation catalyzed by P3-P4-P5 was assayed essentially as described previously (12, 22). P1 at the concentration defined by the particular experiment was mixed with P3-P4-P5 signaling complexes previously assembled using Tar-QEQE in isolated native membrane vesicles in TNED containing 50 mM KCl and 5 mM MgCl2. The reaction was initiated by addition of a concentration of [γ‐32P]ATP defined by the experiment and was terminated at 15 s by the addition of EDTA to 20 mM. The reaction mixture was used for SDS polyacrylamide gel electrophoresis. Radioactivity at the migration position of P1 was quantified by phosphorimaging with ImageQuant TL analysis software (GE Healthcare Life Sciences, Pittsburgh, PA) and comparing to a standard curve of the [γ‐32P]ATP used in the experiment. In the experimental conditions, P1 phosphorylation was a linear function of time over periods significantly longer than 15 s (Fig. S2 in reference 12). Thus, initial rates of P1 phosphorylation were calculated using the values determined at 15 s and the relevant specific activity of [γ‐32P]ATP.
Biolayer interferometry.Time courses of the association of the P3-P4-P5 domain of CheA in free solution with the immobilized P1 domain as well as time courses of dissociation of the resulting complex were recorded by biolayer interferometry (38, 39) performed using the BLItz instrument (Fortébio, Fremont, CA). The signals generated by this device, which are recorded in nanometers, are proportional to the mass of immobilized material at the sensor tip and thus can be used to monitor association and dissociation of protein complexes (38, 39). All measurements were in 50 mM Tris (pH 7.5), 100 mM NaCl, 0.5 mM EDTA, 2 mM DTT, 50 mM KCl, 5 mM MgCl2, 1.5% (wt/vol) glycerol, 0.1% (wt/vol) bovine serum albumin (BSA), and 12 mM cholate (BLItz buffer). The presence of BSA and cholate reduced the extent of nonspecific association with the sensor tip but had no significant effect on phosphorylation of P1 by P3-P4-P5 in signaling complexes and thus no significant effect on the operational interaction of P1 and P3-P4-P5. Anti-Penta-HIS (HIS1K) Biosensor tips (Fortébio) were hydrated following the manufacturer’s instructions. Tips were placed sequentially in the following solutions: six-histidine-tagged P1 at 4 μM for 200 s, buffer alone for 30 s, a concentration P3-P4-P5 between 100 and 800 μM for 120 s, and in buffer alone for 120 s. Experiments were performed in the absence or presence of 10 mM β,γ-methyleneadenosine 5′-triphosphate (AMPPCP; Sigma-Aldrich, St. Louis, MO). In the presence of the nucleotide, time courses of association and dissociation exhibited two phases, namely, one representing rapid binding and rapid release and the other slow binding and slow, incomplete release. The slow phase resembled the pattern observed for the apparently nonspecific interactions of apo-P3-P4-P5 with the sensor tip. To analyze the binding of AMPPCP-occupied P3-P4-P5 to immobilized P1, for each concentration of P3-P4-P5 we subtracted the signal observed for apo-P3-P4-P5 from the signal observed for nucleotide-occupied P3-P4-P5. This adjusted data set was fit globally to a single binding-site model provided by the manufacturer (see Fig. S3 in the supplemental material). In this model, first the dissociation time course is fit to y =
ACKNOWLEDGMENTS
We thank Michael Henzl for guidance in performing global fits.
This work was supported in part by National Institute of General Medical Sciences grant GM29963 to G.L.H.
FOOTNOTES
- Received 21 February 2020.
- Accepted 13 April 2020.
- Accepted manuscript posted online 27 April 2020.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
