Core unit of chemotaxis signaling complexes

Abstract

Bacterial chemoreceptors, histidine kinase CheA, and coupling protein CheW form clusters of chemotaxis signaling complexes. In signaling complexes kinase activity is enhanced several hundredfold and placed under receptor control. Activation is necessary to poise enzyme activity such that receptor control has physiologically relevant effects. Thus kinase activation can be considered the underlying core activity of signaling complexes. We defined the minimal physical unit that generates this activity using chemoreceptor Tar from Escherichia coli rendered water soluble by insertion into nanodiscs to (i) measure saturable binding of CheA and CheW to the smallest kinase-activating groups of receptor dimers and (ii) purify and characterize core units of signaling complexes. Purified complexes activated kinase almost as well as signaling complexes formed on arrays of receptors in isolated native membrane. Purified complexes contained two receptor trimers of dimers and two CheW for each CheA dimer, consistent with the approximately 1∶1 CheA∶CheW ratio determined by binding measurements. The 2∶2∶1 stoichiometry implied that CheA dimers, the enzymatically active form, connect two chemoreceptor trimers of dimers by interaction of one CheA protomer and a CheW with each trimer, an organization for which specific molecular interactions have previously been identified. The core unit associates six receptor dimers with a CheA dimer, providing sufficient capacity to account for much of the cooperativity and interdimer influence observed experimentally. We conclude that the 2∶2∶1 organization is the core structural and functional unit of chemotaxis signaling complexes and postulate that hexagonal arrays characteristic of signaling complexes are built from this unit.

Bacterial chemoreceptors, histidine kinase CheA, and coupling protein CheW form clusters of chemotaxis signaling complexes in which the kinase is activated several hundredfold and placed under the control of receptor ligand occupancy and receptor adaptational modification (1, 2). Kinase activation in signaling complexes poises CheA such that control of its activity by receptors can have physiologically relevant effects. Thus kinase activation represents the core activity of signaling complexes. In this study we characterized and purified the minimal physical unit for that activity. In doing so we identified the likely unit from which signaling arrays are constructed.

Chemoreceptors and Signaling Complexes

The fundamental structural unit of a chemoreceptor is a homodimer. Dimers can interact to form trimers of dimers (3, 4) and, upon formation of signaling complexes with CheA and CheW, cluster in groups containing from a few to thousands of receptors (5, 6) (Fig. 1). The notable sensitivity and wide dynamic range exhibited by chemotaxis sensory systems are thought to reflect functional interactions in clusters among multiple receptors and kinases (1). The unit of control of kinase activity does not appear to be fixed, but rather the number or extent of influence of contributing receptors appears to vary as a function of receptor adaptational modification and receptor neighbors from a few to as many as 20 receptor dimers (7–10). Much less is known about units of kinase activation.

Fig. 1.

Chemoreceptor interactions. The cartoon represents a homodimer (Left), a trimer of receptor dimers (Center), and a cluster of signaling complexes consisting of trimers of dimers, CheA and CheW (Right). The region of CheA plus CheW in clusters does not have a known structure and thus is represented by a cloud-like region.

As a first step in defining the structural unit of kinase activation, we recently investigated effects of limiting the number of potentially interacting receptor dimers in the same membrane bilayer (11). We did so utilizing Nanodiscs, soluble, nanoscale (∼10 nm diameter) plugs of lipid bilayer surrounded by a belt of amphipathic membrane scaffold protein (12). Assays of kinase activation by nanodisc-embedded receptors fractionated on size-exclusion columns as a function of receptor dimers per disc revealed low activation by fractions averaging two or fewer per disc and increasing activation as the average number rose until a plateau level of activation occurred at approximately five dimers per disc (11). At this maximum, kinase activation by nanodisc-embedded receptors was equivalent to activation by receptors embedded in native membrane vesicles, an environment with many potentially interacting neighbors. The significance of maximum activation at five dimers per disc becomes clear by considering that dimers incorporate into nanodiscs independently (13, 14), thus only at ≥ five dimers per disc would every disc have at least three dimers oriented in the same direction and thus able to form a trimer of dimers. We concluded that chemoreceptors act to activate kinase as trimers of dimers (11). In the current work, we used nanodiscs carrying such kinase-activating receptor units to characterize and purify core signaling complexes. Two complementary strategies were employed: (i) determination of saturable binding of CheA and CheW to nanodisc-isolated groups of kinase-activating receptors and (ii) purification and characterization of signaling complexes formed with these receptor groups.

Results

Saturable Binding of CheA and CheW to Receptors.

We characterized binding of the water-soluble components of chemotaxis signaling complexes, CheA and CheW, to membrane-inserted Escherichia coli chemoreceptor Tar contained in nanodiscs. The receptor carried a six-histidine tag on its carboxyl terminus (Tar-6H), allowing separation of receptor-bound CheA or CheW from unbound protein through retention by and then elution from a chelated nickel column (Fig. 2). Binding isotherms were determined by varying the concentration of CheA or CheW in the absence of the other protein or in its presence at a constant concentration. To identify binding corresponding to formation of active signaling complexes, we compared binding by nanodisc-inserted receptors that were effective in forming kinase-activating complexes to binding by those which were not. Primary preparations of nanodisc-inserted Tar-6H were fractionated using size-exclusion chromatography to yield purified preparations containing close to one dimer per disc or approximately three dimers per disc (11). The former is capable of only minimal activation of kinase and thus is relatively ineffective at forming functional signaling complexes; the latter activates kinase to more than 70% of maximal activation and thus is effective at forming functional complexes (11). We used purified preparations averaging three Tar dimers per disc, in which 25% of discs would be expected to have three dimers inserted in parallel, rather than five dimers per disc, in which all discs would contain at least three parallel dimers, because the yield of purified material at five dimers per disc is quite low (11) and thus is a significant impediment to obtaining sufficient material for the experiments described here.

Fig. 2.

Assay of binding of the soluble proteins of signaling complexes to nanodisc-embedded Tar-6H using a chelated nickel column. Purified nanodiscs at approximately three Tar-6H dimers/disc were mixed with CheA and CheW, incubated to allow formation of signaling complexes (sample), applied to a Ni-agarose column (FT, flowthrough ), washed with seven bed volumes of buffer (wash), and eluted with six bed volumes of buffer containing 300 mM imidazole (+imidazole). The figure is a Coomassie blue-stained SDS-polyacrylamide gel of samples representing the same proportion of each fraction. Each fraction was a bed volume.

Interaction of CheA or CheW alone with fractionated, nanodisc-embedded Tar-6H yielded patterns of saturable binding for each protein (Fig. 3 and Table 1). However, there was no significant difference in binding to Tar at approximately one dimer versus approximately three dimers per disc. Because binding of either protein alone was independent of effectiveness of the Tar grouping in activating kinase, those patterns of binding did not provide information about interactions specific for functional signaling complexes.

Fig. 3.

Binding of CheA or CheW alone to nanodisc-embedded Tar-6H. (A) CheA binding to nanodisc-embedded Tar-6H at one or three dimers/disc. Values are averages with standard deviation for two independent determinations. (Inset) An expanded plot. (B) CheW binding to nanodisc-embedded Tar-6H at one or three dimers/disc. Values are means with standard deviation for two (one dimer/disc) or three (three dimers/disc) independent determinations. Curves are fits to a Hill relationship as described in Materials and Methods.

Table 1.

Parameters derived from fitting binding data to a Hill relationship

Ligand		Approximately one Tar dimer per disc			Approximately three Tar dimers per disc
Variable	Constant	KD, μM	BT, μM	n	KD, μM	BT, μM	n
CheA	—	1.9 ± 1.5	0.009 ± 0.006	0.6 ± 0.2	1.4 ± 0.4	0.006 ± 0.001	0.64 ± 0.08
CheA	5 μM CheW	1.0 ± 0.4	0.09 ± 0.03	0.6 ± 0.1	0.3 ± 0.1	0.23 ± 0.03	1.0 ± 0.2
CheW	—	7.5 ± 1.9	0.13 ± 0.02	1.6 ± 0.3	3.6 ± 0.6	0.13 ± 0.01	1.6 ± 0.2
CheW	2 μM CheA	1.6 ± 0.4	0.09 ± 0.02	0.8 ± 0.2	1.2 ± 0.3	0.31 ± 0.06	0.6 ± 0.1

In contrast, with both CheA and CheW present, one at a constant concentration and the other varied, there was significantly greater binding of the varied component to Tar in effectively kinase-activating, three-dimer-per-disc preparations than in one-dimer-per disc preparations (Fig. 4 and Table 1). For CheW, the difference reflected enhancement by CheA of CheW binding to three-dimer-per-disc Tar but no enhancement of binding to one-dimer-per-disc receptor (Fig. 5 B and D). For CheA, the difference reflected a greater enhancement by CheW of CheA binding to three-dimer-per-disc Tar than of binding to one-dimer-per-disc Tar (Fig. 5 A and C). Binding of the varied protein to three-dimer-per-disc preparations was accompanied by binding of the protein present at a constant concentration (Fig. 6). As the amount of varied protein bound increased, so did the amount of constant protein bound, implying coupled binding of the two proteins to receptors, albeit with the relative amount of the varied protein increasing slightly as its concentration was increased, consistent with its rising concentration and perhaps a competition for binding by the two proteins that has been documented previously (15–17). Overall, the pattern of coupled binding implied an approximately 1∶1 binding of CheA and CheW to the sets of three Tar dimers that were effective at activating kinase. Furthermore, fits of the data for binding to three-dimers-per-disc Tar preparations for the varied component in the presence of a constant concentration of the other component extrapolated to values of maximal binding capacity equivalent to 1.35∶1 CheW∶CheA (Table 1). Taken together, these observations indicated that coupled binding of CheA and CheW to nanodisc-inserted Tar at approximately three dimers per disc reflected formation of complexes involving Tar, CheA, and CheW and that the interactions among these components generated roughly equal binding of CheA and CheW.

Fig. 4.

Binding of CheA or CheW to nanodisc-embedded Tar in the presence of other soluble components of signaling complexes. (A) Binding of CheA in the presence of 5 μM CheW to one dimer/disc or three dimers/disc Tar-6H. (B) Binding of CheW in the presence of 2 μM CheA to one dimer/disc or three dimers/disc Tar-6H. Values are means with standard deviation of three independent determinations. Curves are fits as for Fig. 3.

Fig. 5.

Effects of the companion soluble component of signaling complexes on binding of CheA or CheW to nanodisc-embedded Tar at 1 or 3 dimers per disc. The data and fits are those shown in Figs 3 and 4 plotted in different combinations to illustrate effects of CheW on binding of CheA to nanodisc-embedded Tar-6H at three dimers (A) or one dimer per disc (C) and the effects of CheA on binding of CheW to the receptor at three dimers (B) or one dimer per disc (D).

Fig. 6.

Coupled binding of one soluble component of signaling complexes as the other component is varied. Binding of CheA and CheW to nanodisc-embedded Tar-6H at three dimers per disc was determined (A) as CheA concentration was varied in the presence of a constant (5 μM) concentration of CheW and (B) as CheW concentration was varied in the presence of a constant (2 μM) concentration of CheA. Data and fit for CheA in A are from Fig. 4A and for CheW in B are from Fig. 4B. Values for binding of the soluble component present at a constant concentration are means with standard deviation of three independent determinations.

Purification of Signaling Complexes.

We utilized nanodisc-embedded receptors and two different affinity tags to purify signaling complexes in a water-soluble form. We used Tar carrying a six-histidine tag, as it did for binding assays, and CheA carrying a biotin tag. Approximately one biotin per CheA was disulfide-coupled to a native cysteine of purified CheA using a commercially available reagent (see Materials and Methods). The modification did not affect the ability of the protein to bind the receptor. Signaling complexes were formed by incubation of CheW, biotin-coupled CheA, and nanodisc-inserted Tar-6H at three dimers per disc. Each soluble protein was at a concentration that our binding curves indicated generated close to maximal binding in the presence of its soluble partner. Mixtures were applied to a chelated nickel column, washed, and eluted with imidazole, separating CheW and biotin-coupled CheA bound to nanodisc-inserted Tar-6H from molecules not bound. The imidazole-eluted material was applied immediately to a biotin-binding column, washed to remove Tar-6H and CheW not in stable complex with biotin-coupled CheA, and treated with a reducing agent to cleave the disulfide bond between biotin and CheA, releasing purified signaling complexes.

Activity of Purified Core Signaling Complexes.

In these purified complexes CheA was activated approximately 735-fold relative to a control in which the nanodiscs lacked chemoreceptors and thus no kinase-activating complexes were formed (Fig. 7). This activation was within a factor of 1.35 as great as the 990-fold activation for signaling complexes formed with receptors in native membrane vesicles relative to a control in which the vesicles lacked receptors (Fig. 7) and was greater than published values for kinase activation (16, 18). Furthermore, purified core signaling complexes not only activated kinase but also placed the activated kinase under effective control of receptor occupancy, exhibiting drastic reduction of kinase activity in the presence of a saturating concentration of the Tar ligand aspartate (Fig. 7).

Fig. 7.

Activation and control of kinase in signaling complexes. Fold kinase activation is the ratio of kinase activity for CheA in a signaling complex divided by its activity in the absence of signaling complexes because receptors are not present but otherwise in the same conditions. Activation ± saturating ligand (1 mM aspartate) is compared for complexes formed with multiple Tar-H6 dimers contained in native membrane vesicles (membrane vesicles) and purified signaling complexes formed with a limited number of receptors contained in nanodiscs (nanodiscs). Ratios are quotients of means of kinase activity determined by three independent experiments (see Materials and Methods for those values). Error bars are standard deviations calculated by propagation of errors.

Stoichiometry of Purified Core Signaling Complexes.

We quantified the proteins in purified functional complexes, expressing the relative amounts of the three component proteins in terms of the CheA dimer (CheA₂), because CheA is an active kinase only in its dimeric form (19). For each CheA₂ there were 2.0 ± 0.2 copies of CheW, consistent with the approximately 1∶1 ratio of soluble components identified by characterization of saturable binding (Figs. 4 and 6 and Table 1) and several previous studies (15, 20–25).

The material eluted from the second column contained 7.9 ± 0.4 nanodisc-borne Tar dimers (Tar₂) per CheA₂. This value required interpretation because individual Tar dimers were not independent soluble units but embedded in nanodiscs containing several dimers. Our previous work showed that nanodisc-inserted Tar dimers participate in signaling complexes as groups of three parallel dimers (11). If such a group were in a disc containing four dimers, the fourth dimer, not necessarily involved in the complex, would still be retained on and eluted from the biotin-binding column and thus distort the apparent ratio of Tar₂ to CheA₂ in purified complexes. We found that nanodiscs in our purified preparations of signaling complexes contained 3.7 ± 0.2 dimers per disc, equivalent to a mixture of discs in which 70% contained four Tar dimers and 30% contained three. This represented enrichment for discs with four dimers over the starting material, which had a mean of 3.0 dimers per disc as the result of pooling fractions from sizing columns that had 2.5 to 3.5 dimers per disc. Such enrichment would be expected because discs containing four randomly inserted dimers have a 62.5% probability of carrying three parallel dimers, the combination required to form kinase-activating signaling complexes, significantly higher than the 25% probability for discs containing only three dimers. Given 3.7 Tar dimers per disc and 7.9 nanodisc-borne dimers per purified signaling complex, each purified core signaling complex involved essentially two nanodiscs (7.9 dimers per complex/3.7 dimers per disc =  ∼ 2.1 discs). Thus each disc would contain three parallel dimers participating in a kinase-activating signaling complex (11) and 70% would have had one additional dimer, not involved in a complex. These calculations indicated that kinase-activating complexes purified by the two-affinity-tag procedure involved two groups of three parallel dimers, each in a different nanodisc, with 70% of the discs containing an additional dimer (Fig. 8). Thus we conclude that the stoichiometry of the components directly participating in purified core units of kinase-activating signaling complexes is 2 CheW∶2 (Tar₂)₃∶1 CheA₂.

Fig. 8.

Model of the core unit of chemotaxis signaling complexes purified using nanodisc-inserted chemoreceptor dimers. The cartoon shows diagrammatically, and in the stoichiometry written (Left), the deduced organization of purified, kinase-activating core units formed with nanodisc-embedded chemoreceptor Tar. See text for details.

Discussion

The core activity of chemotaxis signaling complexes is activation of kinase CheA. We purified and characterized the minimal physical unit that effectively generates this activation. Purification of these minimal units was possible because nanodiscs render intact, membrane-embedded chemoreceptors water soluble in groups containing only a few potentially interacting dimers in the same lipid bilayer. The purified core units, complexes of one CheA dimer, two copies of CheW, and two groups of three parallel Tar dimers in separate nanodiscs (Fig. 8), activated kinase almost as well as signaling complexes assembled on the many neighboring receptors inserted in native membrane vesicles (Fig. 7). Features of this purified core unit have implications for both structure and function of signaling complexes.

Implications for Signaling Complex Structure.

The 2 (Tar₂)₃∶2 CheW∶1 CheA₂ stoichiometry, with each (Tar₂)₃ in a separate bilayer, implies that the core units we purified consisted of two independent chemoreceptor trimers of dimers joined by a CheA dimer, with each protomer of CheA₂ interacting with one trimer of receptor dimers plus a CheW (Fig. 8). Previous studies of signaling complexes assembled on extended patches of many receptors in native membranes had suggested that CheA dimers might connect two receptor trimers (16, 17, 24). Our results indicate that such a unit is sufficiently stable to be assembled in solution and purified. Importantly, the isolated unit was almost as effective in kinase activation as arrays of signaling complexes (Fig. 7). Thus, the isolated complex represents the core unit of signaling complexes. We suggest that hexagons of receptor trimers of dimers characteristic of signaling complexes (26) are constructed by combining three core units and that addition of further core units builds the characteristic extended hexagonal arrays. This suggestion is consistent with electron tomographic images of hexagonal arrays of native signaling complexes in vivo that show one unit density corresponding to CheA/CheW for every two receptor trimers of dimers (27). In addition, specific orientations by which binding of a CheA protomer and a CheW to a receptor dimer in a trimer of dimers would allow CheA₂ to connect two trimers of receptor dimers were recently elucidated in a detailed structural model for interactions in Thermotoga maritima signaling complexes based on X-ray structures and distance measurements using pulsed electron spin resonance (24).

Stoichiometry of Kinase-Activating Core Units of Signaling Complexes.

An E. coli cell wild type for chemotaxis contains approximately 2.9 receptor dimers and 2.4 CheW per 1 CheA₂ (21), i.e., sufficient soluble components for one CheA dimer and two CheW to associate with each receptor trimer of dimers, rather than with a pair of trimers, the ratio we observed for isolated, kinase-activating core units. The two observations are not necessarily contradictory. In vivo there could be significant pools of CheA and CheW not associated with signaling complexes, a possibility supported by estimates of ratios of free and bound protein derived from determinations of rates of protein exchange in vivo (28). Alternatively or additionally, a significant portion of CheA and CheW might bind to receptors in arrays without forming complete kinase-activating signaling complexes or without being crucial for kinase activation. Perhaps such interactions connect core units into hexagonal arrays.

In purified, kinase-activating signaling complexes there were six complex-participating Tar₂ per CheA₂, the ratio observed for signaling complexes assembled on extended patches of receptors (15–17, 24, 29) and for soluble signaling complexes assembled with receptor fragments (25, 30). Purified signaling complexes had 2.0 ± 0.2 CheW per CheA₂. This 1∶1 ratio was consistent with our direct measurements of binding of those proteins to chemoreceptors in conditions that form kinase-activating signaling complexes (Table 1 and Figs. 4 and 6) and several studies of kinase-activating complexes (15, 21, 23–25). However, other studies have observed more CheW than CheA (16, 17). It is possible that more loosely bound CheW would not remain with the core units over the course of our affinity column procedure, in which free protein is removed and thus a slow off rate is required for bound protein to remain with the complex. The notion that more loosely bound CheW is lost in the course of the two-affinity-tag purification is consistent with the difference between the 1∶1 ratio of CheW to CheA for signaling complexes purified by passage over two columns and the 1.35∶1 ratio observed after only one column (Table 1). CheW appears to bind two ways in signaling complexes and thus with two affinities (17, 24). A more weakly binding CheW might not be retained through two affinity columns, particularly in the absence of clustered receptors to enhance retention. The fractionally greater amount of CheW in signaling complexes assembled on multiple adjacent receptors in native membranes might account for the difference in extent of kinase activation by purified core units versus signaling complexes assembled on receptor patches (Fig. 7) or might be involved in amplification of effects of receptor ligand occupancy (8).

Implications for Signaling Complex Function.

Effective kinase activation by purified core signaling complexes involving a few receptors and one CheA dimer indicates that extended arrays are not necessary for this core function of signaling complexes. This is significant because a considerable proportion of total cellular signaling complexes occur as dispersed, small groupings containing only a few constituent proteins (6). Our results indicate that essentially all these groupings would contribute, in proportion to their numbers, to steady-state kinase activity in a cell and thus to its sensory state.

The deduced organization of purified core units of signaling complexes: two trimers of receptor dimers linked to a single CheA dimer (Fig. 8), creates a physical linkage through which ligand occupancy of up to six receptor binding sites (there is one relevant site per receptor dimer) and adaptational modification of as many as 48 receptor methyl-accepting sites (eight sites/dimer × six dimers) could be integrated to control activity of a single dimeric enzyme. This capacity for cooperative and concerted action is sufficiently large to account for much of the cooperativity and most interdimer influences observed experimentally (7, 29, 31–33). However, our model of core units (Fig. 8) does not provide a structural explanation for the amplification that allows occupancy of a single receptor to inhibit kinase activity of approximately 35 kinases (8). Understanding the structural basis of that amplification will require identification of ways in which core units interact in higher-order hexagonal units and arrays. These are subjects for future investigations.

Materials and Methods

Proteins and Nanodiscs.

Tar-6H and membrane scaffold protein MSP1D1E3(-), hereafter MSP (12), were purified as described (14) as were CheA, CheW, and CheY (34). Nanodiscs containing Tar-6H at approximately three receptor dimers per disc were formed, fractionated by size-exclusion chromatography, protein content analyzed, and fractions stored as in ref. 11 except that flow rate for the size-exclusion column was 0.6 mL/ min. For binding studies, size-exclusion column fractions with nominal Tar content between 0.9 and 1.2 or between 2.5 and 3.5 dimers per disc, obtained from one or three primary nanodisc preparationsn respectively, were pooled to produce purified approximately one-dimer and approximately three-dimer-per-disc Tar-H6. For two-affinity-tag purifications of signaling complexes, pooled material at three dimers per disc was prepared in the same way. Prior to coupling of biotin to CheA, reducing agent (2 mM DTT) was removed using a PD SpinTrap G-25 column (GE Healthcare), following the manufacturer’s instructions, from 100 μM purified protein stored at -80 °C in 50 mM Tris-HCl (pH 7.5), 0.5 mM EDTA, 2 mM DTT, 10% (wt/vol) glycerol. N-biotinylaminoethyl methanethiosulfonate (Toronto Research Chemicals Inc.), stored at 50 mM in dimethyl sulfoxide and diluted to 1 or 5 mM in 50 mM Tris-HCl (pH 7.5) immediately prior to use, was added at a molar ratio to CheA of approximately 1∶1, the mixture incubated in the dark 1 h on ice and unreacted reagent removed using a PD SpinTrap G-25 column. For experiments determining stoichiometry of purified signaling complexes, the molar ratio of reagent to protein was 1.3∶1, yielding CheA 65% with one biotin and 34% with two, as determined by mass spectrometry and assuming modification did not significantly affect ionization efficiencies. This preparation was as effective as unmodified CheA in forming signaling complexes. For experiments measuring kinase activation in purified signaling complexes, the molar ratio of reagent to protein was 0.9∶1, yielding CheA 87% with one biotin and 13% unmodified, determined as described above. This preparation was as effective in being activated by Tar embedded in native membrane as unmodified protein.

Binding of CheA and CheW to Nanodisc-Inserted Tar.

Purified nanodiscs containing approximately one or approximately three Tar-6H dimers/disc were mixed at a final concentration of 10 μM receptor protomer with 0 to 5 μM CheA protomer in the absence or presence of 5 μM CheW or with 0 to 15 μM CheW in the absence or presence of 2 μM CheA protomer in 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 10 mM MgCl₂, 50 mM KCl, 10% (wt/vol) glycerol, 30 mM imidazole (buffer A), and incubated 1 to 1.5 h at room temperature (22–24 °C). A 50-μL sample was applied to a 100-μL bed volume column of Ni-Agarose resin (Qiagen) formed in a P200 tip (Rainin) with its narrow end cut approximately 7 mm from the tip, plugged with a small piece of glass wool, and equilibrated with buffer A. The column was washed with seven to nine bed volumes of buffer A and eluted with buffer B (buffer A containing 300 instead of 30 mM imidazole). Over 90% of the Tar-6H eluted after application of buffer B in the second and third 50-μL fractions (Fig. 2). These fractions were pooled and analyzed for content of Tar, MSP, CheA, and/or CheW by SDS polyacrylamide gel electrophoresis and comparison to purified protein standards using staining with Coomassie brilliant blue (Tar-6H and MSP) or immunoblotting (CheA and CheW) (21). Binding data from three or, in a few cases, two independent experiments were fit, using simultaneous least-squares minimization in Origin v.7.5, to the Hill equation:

with X the extent of binding (moles of ligand bound per mole of macromolecule), [B_T] the number of binding sites, [L] the concentration of free ligand, n the Hill coefficient, and K_D the apparent dissociation constant. K_D and n were global fitting parameters applied to respective individual datasets; B_T was a local parameter, allowed to assume different values for individual experiments. For each data point, free ligand concentration was estimated from the known total ligand concentration, [L_T], by employing a bisection strategy to find the value of [L] satisfying

The first term on the right-hand side is the calculated concentration of bound ligand. That value plus [L] should equal the total concentration. In practice, the bisection process was terminated when the difference between calculated and actual total ligand concentrations was < 10^-6[L_T].

Purification of Signaling Complexes.

Signaling complexes were formed by mixing size-exclusion-column-purified Tar-6H contained in nanodiscs at approximately three dimers per disc, biotin-coupled CheA and CheW to provide 10 μM, 5 μM, and 15 μM protein, respectively, in buffer A and incubated 1 to 1.5 h. The mixture was applied to a Ni-agarose column constructed and eluted as described for binding experiments. The two fractions containing the vast majority of eluted Tar, as documented subsequently by SDS polyacrylamide gel electrophoresis, were pooled and immediately applied to a 100-μL column of streptavidin agarose (Thermo Scientific, catalog # 20347) constructed like the Ni column and equilibrated with buffer A or, in later experiments, a column of NeutrAvidin (Thermo Scientific, catalog # 29202). After sample application, the column was washed with seven to nine bed volumes of buffer A and eluted with 50 mM DTT in buffer A. Lower concentrations of reducing agent were not as effective in releasing CheA. Even at this concentration, elution from the streptavidin resin occurred over multiple fractions, but the relative amounts of components of signaling complexes were approximately constant in each fraction. Fractions containing approximately 70% of eluted proteins (five to nine fractions depending on the experiment) were pooled and analyzed for content of Tar, MSP, CheA, and CheW as for binding experiments. There was no significant retention of either Tar-6H or CheW when the protein was applied alone to the streptavadin column. Elution by 50 mM DTT from the NeutrAvidin resin was more focused, with approximately 85% of eluted protein in two fractions. These fractions were pooled and analyzed as above.

Kinase Activation.

Kinase activity was assayed essentially as described (11, 34). Signaling complexes were formed on receptors inserted in native membrane vesicles by incubating 5 μM Tar-6H, 0.25 μM biotin-coupled CheA, and 8 μM CheW 1 h at room temperature. The complexes formed were separated from unincorporated soluble components by two cycles of centrifugation (16,100  × g, Eppendorf 5414D centrifuge, 10 min) plus resuspension and content of the respective proteins in the complexes determined by quantitative gels and immunoblotting as above. Kinase assays at 20 μM CheY, 35 mM DTT (added to match the buffer of purified signaling complexes), and approximately 100 nM CheA contained in signaling complexes were initiated by addition of [γ-³²P-ATP] (63 μCi/μmol) to 0.4 mM and terminated 15 s later by addition of denaturing SDS sample buffer (11, 34). In tests of saturating ligand, 1 mM aspartate was added after isolation of signaling complexes and the mixture incubated 20 min before assay. Purified signaling complexes providing a final CheA concentration of 20 to 30 nM were assayed by adding CheY and, as desired, aspartate so that they were 20 μM and 1.0 mM, respectively, in the final reaction mixture. CheA activity in the absence of receptors, and thus not in signaling complexes, was determined by substituting native membrane vesicles or nanodiscs lacking chemoreceptors in the respective assays with all other components present. Mean kinase activities and standard deviations for three independent experiments, in moles of phospho-CheY per mole CheA, were, respectively, 109 ± 6, 2.4 ± 0.6, 0.11 ± 0.004, 59 ± 9, 0.45 ± 0.05, and 0.08 ± 0.02 for signaling complexes with Tar-6H inserted in native membrane, the same complexes plus saturating aspartate, the same quantity of native membrane lacking Tar-6H, purified complexes formed with nanodisc-inserted Tar-6H, the same complexes plus saturating aspartate, and the same quantity of nanodiscs lacking Tar-6H.