Methyltransferase CheR binds to its chemoreceptor substrates independent of their signaling conformation yet modifies them differentially

Abstract

Methylation of specific chemoreceptor glutamyl residues by methyltransferase CheR mediates sensory adaptation and gradient sensing in bacterial chemotaxis. Enzyme action is a function of chemoreceptor signaling conformation: kinase‐off receptors are more readily methylated than kinase‐on, a feature central to adaptational and gradient‐sensing mechanisms. Differential enzyme action could reflect differential binding, catalysis or both. We investigated by measuring CheR binding to kinase‐off and kinase‐on forms of Escherichia coli aspartate receptor Tar deleted of its CheR‐tethering, carboxyl terminus pentapeptide. This allowed characterization of the low‐affinity binding of enzyme to the substrate receptor body, otherwise masked by high‐affinity interaction with pentapeptide. We quantified the low‐affinity protein–protein interactions by determining kinetic rate constants of association and dissociation using bio‐layer interferometry and from those values calculating equilibrium constants. Whether Tar signaling conformations were shifted by ligand occupancy or adaptational modification, there was little or no difference between the two signaling conformations in kinetic or equilibrium parameters of enzyme‐receptor binding. Thus, differential methyltransferase action does not reflect differential binding. Instead, the predominant determinants of binding must be common to different signaling conformations. Characterization of the dependence of association rate constants on Deybe length, a measure of the influence of electrostatics, implicated electrostatic interactions as a common binding determinant. Taken together, our observations indicate that differential action of methyltransferase on kinase‐off and kinase‐on chemoreceptors is not the result of differential binding and suggest it reflects differential catalytic propensity. Differential catalysis rather than binding could well be central to other enzymes distinguishing alternative conformations of protein substrates.

Abbreviations

DTT

dithiothreitol

EDTA

ethylenediaminetetraacetic acid

PCR

polymerase chain reaction

Tar

chemoreceptor mediating taxis to aspartate and repellents

Tar4E

Tar with 4 glutamyl residues at the sites of adaptational modification

Tar4Q

Tar with 4 glutaminyl residues at the sites of adaptational modification

1. INTRODUCTION

Covalent modification of chemoreceptors is the central mechanism of sensory adaptation and gradient sensing in bacterial chemotaxis.1, 2, 3 The modifications are methylation of specific glutamyl residues and demethylation of the resulting glutamyl methyl esters. Methylation is catalyzed by the dedicated methyltransferase CheR and demethylation by the dedicated methylesterase CheB. CheR preferentially modifies the kinase‐off receptor conformation4, 5, 6, 7 and CheB the kinase‐on conformation,8 thereby shifting the receptor conformational equilibrium to an intermediate signaling state and providing adaptational feedback that restores a balanced state of behavior and biochemical signaling.1, 2, 3

The four chemoreceptors of Escherichia coli that undergo adaptational modification carry four or five modification sites on each protomer of the homodimeric proteins. The eight or 10 sites on a dimer are clustered near each other in the three‐dimensional structure of the chemoreceptor (Figure 1) approximately half‐way between the cytoplasmic membrane and the membrane‐distal tip of the ∼200 Å, rod‐shaped, helical coiled‐coil cytoplasmic domain.13, 14, 15, 16, 17, 18 For each protomer, three or four sites are on the same, solvent‐facing surface of the cytoplasmic domain N‐helix; one or two are on the C‐helix in similarly solvent‐facing orientations. A significant portion of the surface of the chemoreceptor, rod‐shaped body is negatively charged, including the region surrounding the substrate methyl‐accepting sites.18, 19

Figure 1.

Interactions between a pentapeptide‐bearing chemoreceptor and methyltransferase CheR. The figure shows space‐filling models of methyltransferase CheR (orange; PDB ID: 1AF79) and a high‐abundance chemoreceptor with the loci of interaction indicated by double arrows. The protomers of the homodimeric receptor are shown in two shades of gray, the pentapeptide in black and the methyl‐accepting sites in the gene‐coded state of two glutamyl (yellow dots) and two glutaminyl (green dots) residues. Four of the eight methyl‐accepting sites are visible in this view, sites 1–3 on the light‐gray protomer and site 4 on the dark‐gray one. The receptor model is a combination of the X‐ray structure of a HAMP‐cytoplasmic domain fusion (PDB ID: 3ZX610) with a model of the periplasmic plus transmembrane segments of Tar,11 which in turn included the X‐ray structure of a chemoreceptor periplasmic domain (1L1H12). For illustration, a representation of the unstructured flexible arm was added with its origin at the most carboxyl‐terminal residue in the structure of the cytoplasmic domain. The figure was assembled using PyMOL12

In E. coli and related organisms, efficient modification by the methyltransferase or methylesterase requires a five‐residue enzyme‐binding pentapeptide, NWETF or NWESF, located at the extreme carboxyl terminus of the chemoreceptor (Figure 1 20, 21). This enzyme‐interaction site is connected to the structured, coiled‐coil body of the receptor cytoplasmic domain by a ∼35‐residue unstructured sequence.22 Thus, each enzyme binds to chemoreceptors at a minimum of two locations: (a) the pentapeptide at the end of a long unstructured segment and (b) the substrate sidechains on the coiled‐coil body of the chemoreceptor cytoplasmic domain. In the current study, we focus on the interactions of chemoreceptors and CheR. A subsequent study will consider interactions with CheB.

1.1. CheR‐chemoreceptor interactions

CheR binds to the pentapeptide with a dissociation constant of ∼2 μM independent of whether that five‐residue sequence is an isolated peptide or in its native position at the carboxyl terminus of a chemoreceptor.20 In that binding, the five‐residue sequence becomes the fourth beta strand of what is otherwise a three‐stranded beta sheet subdomain.23 CheR binding to the substrate‐containing coiled‐coil body of a chemoreceptor is significantly weaker than to the pentapeptide, too weak to be detected by isothermal titration calorimetry.20 Taken together, these observations suggested that the pentapeptide serves as a high‐affinity binding site that tethers the methyltransferase to the receptor, increasing the local enzyme concentration sufficiently to allow effective binding to the low‐affinity substrate sites and thus catalysis.20, 24 Since CheR binds to receptors carrying the carboxyl‐terminal pentapeptide with the same 2 μM affinity independent of the receptor signaling conformation,20 higher rates of methylation for receptors in the kinase‐off, methylation‐on versus the kinase‐on, methylation‐off conformation must reflect differential action of the enzyme at its substrate sites on the structured receptor body. The basis of this differential action is not known, but has been commonly assumed to be differential binding of the enzyme to receptors in the two conformations, ultimately at the substrate methyl‐accepting sites.1, 2, 3 However, the weak binding of CheR to the chemoreceptor coiled‐coil body has not been characterized. This is the result of the technical challenges in measuring low‐affinity protein–protein interactions, which necessarily involves high concentrations of at least one of the interacting proteins. We overcame that limitation by using bio‐layer interferometry, a technique that can be performed in the presence of the high concentrations of a protein‐binding partner required to characterize low‐affinity interactions.25 The approach generated association and dissociation rate constants for the low‐affinity as well as the high‐affinity interactions between the methyltransferase and chemoreceptors. Equilibrium dissociation constants for the interactions were then calculated using the respective rate constants.

2. RESULTS

2.1. Experimental design

We quantified the kinetics of interactions between methyltransferase CheR and the E. coli aspartate chemoreceptor Tar using bio‐layer interferometry.25 Individual receptor homodimers were reconstituted into bilayers of native E. coli lipids contained in Nanodiscs.26, 27 Such dimers exhibit native activities for ligand binding, adaptational modification, and transmembrane signaling.7 These activities were confirmed for representative samples used in our analyses. Characterizing binding to isolated receptor homodimers avoided potential complications of enzyme trapping by the multiple binding sites24 present in clusters of neighboring receptor dimers in trimers or higher order arrays.28, 29 Nanodisc‐inserted Tar dimers were attached to the surface of avidin‐coated sensor probes via a biotin‐containing linker coupled through a maleimide to an introduced cysteine at chemoreceptor position 378, close to the membrane‐distal tip of the receptor cytoplasmic domain. This substitution is distant from the modification sites and does not significantly perturb function of isolated receptor homodimers.30 Sensor probes decorated with Nanodisc‐embedded chemoreceptors were placed in solutions of CheR at various concentrations and time courses of association and dissociation recorded (Figure S1). Apparent rate constants of association and dissociation were derived from global fits to these time courses. Binding constants were calculated from those apparent rate constants.

We tested intact chemoreceptor Tar and derivatives lacking the CheR‐binding pentapeptide NWETF otherwise present at the receptor carboxyl terminus. For each form, measurements were done for receptors at the two extremes of adaptational modification: all four methyl‐accepting sites negatively charged glutamyl side chains (Tar4E) and all four neutral glutaminyl side chains (Tar4Q). Glutaminyl side chains at methyl‐accepting sites mimic the shape, neutral nature, and signaling properties of the catalytic product, glutamyl carboxyl methylesters.31, 32 The 4E forms of the receptor represented enzymatic substrate and the methylation‐on, kinase‐off conformational signaling state. The 4Q forms represented the enzymatic product and the methylation‐off, kinase‐on conformational state.7, 33

2.2. CheR binding to the NWETF pentapeptide

For intact, pentapeptide‐bearing Tar, rate constants of association and dissociation, and the calculated equilibrium dissociation constants were independent of chemoreceptor modification state and the same as the values for pentapeptide alone (Table 1, compare rows 1, 2, and 5; Figure S2). These results were consistent with previous determinations of equilibrium binding constants for 4E and 4Q forms of a chemoreceptor by isothermal titration calorimetry20, 34 and extended them by documenting that the underlying rate constants were also independent of receptor signaling state. The inverses of the rate constants of dissociation, which we term dwell times, defined average times a complex persisted prior to dissociation. For CheR‐pentapeptide, it was approximately 5 s (Table 1, Figure S2). It represents the average time the pentapeptide tethers CheR in proximity to the structured receptor body and thus to the substrate methyl‐accepting glutamyl side chains (see Section 3).

Table 1.

CheR binding^a

Row	Receptor	Receptor carboxyl terminus	K D (μM)	k on (M−1 s−1)	k off(s−1)	Dwell time (s)
1	None	NWETF	1.9 ± 0.4	114,000 ± 33,000	0.21 ± 0.02	4.7 ± 0.4
2	Tar‐4E	NWETF	2.1 ± 0.2	94,000 ± 4,700	0.20 ± 0.02	5.1 ± 0.4
3	Tar‐4E	NWETFHHHHHH	3.7 ± 0.8	55,100 ± 8,300	0.20 ± 0.02	5.0 ± 0.6
4	Tar‐4E	ΔNWETF	120 ± 50	3,100 ± 180	0.36 ± 0.05	2.8 ± 0.4
5	Tar‐4Q	NWETF	1.9 ± 0.4	96,000 ± 17,000	0.17 ± 0.02	5.8 ± 0.5
6	Tar‐4Q	NWETFHHHHHH	3.1 ± 0.5	74,100 ± 7,800	0.23 ± 0.01	4.3 ± 0.3
7	Tar‐4Q	ΔNWETF	200 ± 20	2,100 ± 310	0.42 ± 0.05	1.2 ± 0.2

^aErrors are standard deviations of at least three independent determinations.

2.3. CheR binding to the receptor body

No interactions between CheR and a chemoreceptor lacking the NWETF pentapeptide had been detected by isothermal titration calorimetry.20 Yet CheR certainly binds to the coiled‐coil body of the receptor dimer, at minimum at the substrate, methyl‐accepting glutamyl side chains. Those interactions must occur even with chemoreceptors lacking pentapeptide since such receptors are methylated, albeit at much lower rates than pentapeptide‐bearing ones.5, 35, 36 Thus, CheR necessarily binds to the substrate receptor body in the absence of the pentapeptide but presumably with a relatively weak affinity. We tested for this binding using Nanodisc‐inserted TarΔpp4Q, in which the carboxyl‐terminal pentapeptide was deleted and all four methyl‐accepting sites contained glutaminyl side chains, the mimic of methyl glutamyl side chains. At sufficiently high CheR concentrations, we observed binding and could record time courses of association and dissociation (Figure S1). Rate constants were extracted from such data and equilibrium constants calculated (Table 1). As expected, CheR binding to the receptor body was substantially weaker than to the pentapeptide (Table 1, Figure 2). The calculated K _D for CheR binding to the body of TarΔpp4Q was 100‐fold weaker than the ∼2 μM dissociation constant for CheR binding to the pentapeptide. The major contributor to the difference was the rate constant of association, ∼50‐fold lower for binding to the receptor body than to the pentapeptide.

Figure 2.

CheR binding to chemoreceptor homodimers lacking the high‐affinity NWETF pentapeptide was weak but detectable. Association (k _on) and dissociation (k _off) rate constants were determined, as described in the text, for CheR binding to functional Tar homodimers inserted into native E. coli lipid bilayers contained in Nanodiscs. Those values were used to calculate the respective equilibrium dissociation constants (K _D). Dwell times, the average time of occupancy before dissociation, were calculated as the inverse of dissociation rate constants. Tar homodimers were in the all‐glutaminyl (4Q) modification state and were deleted of (−) or carried (+) carboxyl‐terminal NWETF. Error bars represent standard deviation for the results of at least three independent experiments

With an assay that could detect binding of CheR to the chemoreceptor body, we investigated effects of chemoreceptor signaling conformation on that binding, comparing enzyme binding to pentapeptide‐deleted Tar shifted toward the kinase‐off or kinase‐on state. We generated the shifts in two ways: occupancy by the attractant ligand aspartate, which shifts the conformational equilibrium toward kinase‐off, methylation‐on, and the extent of adaptational modification with the methyl glutamate mimic glutamine, which shifts the conformational equilibrium toward kinase‐on, methylation‐off. Previous work had shown that saturation with aspartate of Nanodisc‐inserted dimers of Tar4E or Tar1E3Q increased initial rates of methylation an average of 2.1‐fold.7 Similarly, the initial rate of methylation for Tar4E is 2.5‐fold higher than for 3Q1E (the most modified form for which methylation could be assayed).7 If differences in initial rates of methylation were the result of differences in CheR binding to the substrate receptor body, then the affinity of enzyme for TarΔpp should be ∼2‐fold greater for aspartate‐saturated versus aspartate‐free receptor and more than 2.5‐fold greater for enzyme binding to TarΔpp4E versus TarΔpp4Q. We observed no such differences. Binding of CheR to the receptor body was essentially unchanged by shifting the signaling state of Tar4E or Tar4Q toward kinase‐off, methylation‐on via saturation with aspartate (see Table 2 and Figure 3, in which affinities are shown as equilibrium association constants, the inverse of the equilibrium dissociation constants plotted in Figures S2 and S3). Receptor ligand occupancy had no effect on binding of CheR alone. It had no effect on binding of CheR saturated with co‐substrate S‐adenosylmethionine, the condition of the enzyme in the assay showing a twofold increase in initial rate of methylation twofold with ligand occupancy.7 It had no effect on binding of S‐adenosylmethionine‐occupied CheR in the presence of a CheR‐saturating concentration of free pentapeptide. Note that during the assay period of a little more than 30 s over which CheR‐Tar association and dissociation were monitored (Figure S1), only a small proportion of 4E receptor would be methylated and thus the CheR binding detected was predominately to the 4E receptor. In summary, the conformational shift to kinase‐off, methylation‐on generated by aspartate occupancy did not increase enzyme binding but only catalysis.

Table 2.

CheR binding in the presence of enzyme or receptor ligands^a

Receptor		SAM/SAH (1 mM)	NWETF (0.5 mM)	Asp. (1 mM)	K D (μM)	k on (M−1 s−1)	k off (s−1)	Dwell time (s)
Modif.	NWETF
4E	+	−	−	−	2.1 ± 0.2	94,000 ± 4,700	0.20 ± 0.02	5.1 ± 0.4
		SAM	−	−	2.2 ± 0.1	119,000 ± 24,000	0.26 ± 0.04	3.9 ± 0.6
	−	−	−	−	120 ± 20	3,100 ± 180	0.36 ± 0.05	2.8 ± 0.4
		SAM	−	−	210 ± 60	2,000 ± 580	0.41 ± 0.01	2.5 ± 0.1
		SAH	−	−	110 ± 6	3,500 ± 240	0.36 ± 0.02	2.8 ± 0.1
		−	−	+	200 ± 40	2,300 ± 500	0.46 ± 0.02	2.2 ± 0.1
		SAM	−	+	180 ± 58	2,900 ± 640	0.49 ± 0.06	2.1 ± 0.3
		SAM	+	+	160 ± 20	2,400 ± 170	0.38 ± 0.01	2.7 ± 0.1
4Q	+	−	−	−	1.9 ± 0.4	96,000 ± 17,000	0.17 ± 0.02	5.8 ± 0.5
		SAM	−	−	1.6 ± 0.3	165,000 ± 34,000	0.25 ± 0.04	4.0 ± 0.6
	−	−	−	−	200 ± 20	2,100 ± 310	0.42 ± 0.05	1.2 ± 0.2
		SAM	−	−	130 ± 30	3,000 ± 400	0.39 ± 0.04	2.6 ± 0.3
		SAH	−	−	120 ± 20	3,000 ± 360	0.36 ± 0.02	2.8 ± 0.2
		−	−	+	230 ± 90	2,000 ± 800	0.40 ± 0.05	2.3 ± 0.4
		SAM	−	+	160 ± 30	2,500 ± 430	0.40 ± 0.02	2.5 ± 0.2
		SAM	+	+	130 ± 30	2,400 ± 220	0.31 ± 0.03	3.2 ± 0.3

Abbreviations: SAH, S‐adenosylhomocysteine; SAM, S‐adenosylmethionine.

^aErrors are standard deviations of at least three independent determinations.

Figure 3.

Shifting Tar conformational signaling state toward kinase‐off, methylation‐on by ligand occupancy enhances initial rate of CheR‐catalyzed receptor methylation but not affinity of the enzyme for receptor. The leftmost bar (orange) shows the average ratio of initial velocities (V _i ^+Asp/V _i ^−Asp) in the presence (+Asp) or absence (−Asp) of saturating (1 mM) aspartate for CheR‐catalyzed methylation of Tar in the 4E, 3E1Q, 2E2Q, or 1E3Q modification state inserted into native lipid bilayers contained in Nanodiscs at 1 dimer/disc. The value is calculated from data in Table 1 of Amin and Hazelbauer.7 All other bars show ratios of equilibrium association constants (K _A ^+Asp/K _A ^−Asp) from Table 2 in the presence (+Asp) or absence (−Asp) of 1 mM aspartate for binding of CheR to TarΔpp in the 4E (red) or 4Q (green) modification state (Modif.) in the same lipid environment and 1 dimer/disc condition as for the measurements of initial velocities. In the indicated cases, 1 mM S‐adenosylmethionine (SAM), or 1 mM SAM plus 0.5 mM free NWETF pentapeptide were present. Error bars as for Figure 2

Table 3.

CheR binding as a function of salt concentration^a

Receptor		K D (μM)	k on (M−1 s−1)	k off (s−1)	Dwell time (s)
Modification	KAcb (mM)
4E	30	102 ± 18	4,700 ± 840	0.48 ± 0.06	2.1 ± 0.2
	150	120 ± 17	3,100 ± 180	0.36 ± 0.05	2.8 ± 0.4
	750	240 ± 110	1,500 ± 820	0.35 ± 0.03	2.8 ± 0.2
4Q	30	79 ± 35	5,400 ± 1,700	0.43 ± 0.06	2.4 ± 0.3
	150	200 ± 22	2,100 ± 310	0.42 ± 0.05	2.4 ± 0.3
	750	160 ± 69	1,900 ± 860	0.30 ± 0.03	3.4 ± 0.3

^aErrors are standard deviations of at least three independent determinations.

^bKAc, potassium acetate. Experiments in the presence of 10 mM HEPES pH 7.6.

We observed related patterns for experiments in which signaling conformation was manipulated by changing the extent of adaptational modification (Figure 4, Table 2). In the physiologically relevant presence of S‐adenosylmethionine, changing chemoreceptor modification state from kinase‐on, methylation‐off 4Q to kinase‐off, methylation‐on 4E did not alter CheR binding. This was the case for ligand‐free and aspartate‐occupied receptors as well as for aspartate‐occupied receptor binding CheR saturated with free pentapeptide. CheR saturated with its small molecule product, S‐adenosylhomocysteine, bound to the 4E and 4Q forms of Tar with equal affinity. For apo‐CheR, devoid of a small‐molecule ligand, the modification‐determined signaling state of aspartate‐occupied TarΔpp had no significant effect on binding. Apo‐CheR binding to ligand‐free receptor exhibited a modest preference for the modification‐generated kinase‐off, methylation‐on conformation, but it is the only condition in which we observed such a binding preference. We lend more weight to the physiologically most relevant conditions of co‐substrate‐occupied or co‐product‐occupied CheR. Considering all relevant data, we conclude that shifting chemoreceptor conformational signaling state as a function of adaptational modification has little if any effect on binding of CheR to receptor. Thus, the great bulk of the experimental evidence indicates that enhanced initial rates of chemoreceptor methylation in the kinase‐off, methylation‐on conformational signaling state are not the result of enhanced enzyme binding. The data also indicates that the major determinants of CheR binding to the chemoreceptor body must be features common to the two conformational signaling states.

Figure 4.

Shifting Tar conformational signaling state by increasing the number of glutamyl residues at methyl‐accepting sites enhances initial rate of CheR‐catalyzed receptor methylation but not affinity of the enzyme for receptor. The leftmost bar (orange) shows the ratio of average initial velocities (V _i ^4E/V _i ^1E3Q) for Tar inserted into native lipid bilayers contained in Nanodiscs at 1 dimer/disc and carrying glutamyl residues at its four methyl‐accepting sites (4E) or at only one site, with glutaminyl residues at other three (1E3Q). The value is calculated from data in Table 1 of Amin and Hazelbauer using TarQEQQ as 1E3Q.7 All other bars (blue) show ratios of equilibrium association constants (K _A ^4E/K _A ^4Q) from Table 2 for binding of CheR to Tar in the 4E versus 4Q modification state and in the same lipid environment and 1 dimer/disc condition as for the measurements of initial velocities. In the indicated cases, 1 mM S‐adenosylmethionine (SAM), 1 mM SAM and 0.5 mM free NWETF pentapeptide, or 1 mM S‐adenosylhomocysteine (SAH) were present. Error bars as for Figure 2

2.4. Electrostatic interactions of CheR and the chemoreceptor body

A strong candidate for such a common feature is the negative electrostatic potential of the chemoreceptor body, which would attract the positively charged surface near the CheR active site9, 37, 38 (Figure 5). The regions of negative electrostatic potential along the receptor body include many anionic groups in addition to the methyl‐accepting glutamyl side chains and thus the surface of the coiled coil remains substantially negatively charged even with those four groups neutralized by methylation or substitution with glutaminyl residues (Figure 5). We investigated the influence of electrostatics on the binding of CheR to the chemoreceptor body by determining the influence of ionic strength on the kinetics of association. Charge–charge interactions would be expected to be a function of Debye length, the distance from a charged surface at which a point charge of the opposite sign experiences attraction greater than the average kinetic energy of the solution. As shown in Figure 6, over a physiologically relevant range of salt concentrations (40–760 mM), rate constants of association for CheR and pentapeptide‐deleted Tar varied threefold as a function of Debye length, in a roughly linear manner. There was no systematic difference between the values for the 4E and 4Q modification/signaling states. These data suggest that electrostatics contributes significantly to the experimentally observed association of the enzyme with the chemoreceptor body and that contribution is largely independent of substrate versus product side chains and of the conformational difference between the methylation‐on 4E and methylation‐off 4Q modification states.

Figure 5.

Distribution of electrostatic potential on the surface of CheR and the cytoplasmic domain of a chemoreceptor. Space‐filling models of methyltransferase CheR (PDB ID: 1BC523) and a fusion of a HAMP domain and the coiled‐coil region of the cytoplasmic domain of an E. coli chemoreceptor (3ZX610) are colored according to electrostatic surface potential in aqueous solution (dielectric constant = 80). The native QEQE modification state of the receptor was changed to 4E (top) or 4Q (bottom). The positions of methyl‐accepting glutamyl (E) or modification‐mimics glutaminyl residues (Q) are marked by yellow or green dots, respectively. The CheR structure is for the enzyme in complex with NWETF, likely its state when it binds to the receptor body (see text), with the position of the active site marked by a gold dot. Coulombic surface coloring from −0.59 (red) to +0.59 (blue) kcal mol⁻¹ e⁻¹ was generated in UCSF Chimera39 and the figure assembled using Swiss‐PDB Viewer40

Figure 6.

Association rate constants as a function of Debye length. Time courses of binding were performed and parameters determined (Table 3) as for Figures 2, 3, 4, 5 for Tar in the all‐glutamyl (red dots) or all‐glutaminyl (green dots) modification state and at three different salt concentrations. Association rate constants (k _on) are plotted versus Debye length (see text). These lengths were calculated for the 1:1 electrolyte solutions of 10 mM HEPES pH 7.6 plus 30, 150, or 750 mM potassium acetate as $\mathrm{DL}=\frac{0.304}{\sqrt{\left[\text{salt}\right]}}$, yielding Debye lengths of 1.52, 0.76, and 0.35 nm for 0.04, 0.16, and 0.76M monovalent salt, respectively. Error bars as for Figure 2. The dashed line is drawn to aid the eye

3. DISCUSSION

We characterized the weak interactions of methyltransferase CheR with its substrate protein, the structured body of the chemoreceptor dimer and found that those interactions were essentially unaffected by receptor conformational signaling state. However, initial rates of CheR‐catalyzed methylation vary as a function of receptor signaling state and that variation has been commonly assumed to reflect differential binding of the enzyme to receptors in the two conformations.1, 2, 3, 7, 33 Our binding studies revealed that this is not the case. Thus, the aspect of enzyme action altered by changing the receptor signaling conformation must be something other than enzyme binding. We suggest that the conformational signaling differences alter enzyme catalysis, not binding. Such alterations might involve alternative positioning or differential dynamics of the methyl‐accepting side chains relative to the enzymatic active site. In contrast, the energetics of enzyme binding must involve receptor features common to the two conformational signaling states. Our data implicate electrostatic potential as at least one of those features.

3.1. CheR‐receptor binding

The lack of significant discrimination by CheR in binding the two conformational signaling states of the structured chemoreceptor body, whether generated by adaptational modification or ligand occupancy, indicated that the predominant determinant of enzyme binding to the receptor body was common those states rather than different. A strong electrostatic surface potential appears to be at least part of that predominant determinant (Figure 6). A role for electrostatic interaction between the positively charged region of CheR near its active site and the negatively charged surface of the chemoreceptor was first suggested in the paper that reported the enzyme structure.9 Mutational analyses supported this notion.37, 38 Our binding studies provide direct biochemical evidence that electrostatics are important for CheR binding to its protein substrate. In addition, our results indicate that this electrostatics‐driven binding is not significantly sensitive to the conformational signaling state of the structured chemoreceptor body. Thus, interaction of CheR with the chemoreceptors body involves not only stereospecific interactions of the enzyme with methyl‐accepting, substrate glutamyl residues and nearby side chains,41, 42, 43 but also electrostatic interactions.

What do these results and deductions suggest about CheR binding to the structured chemoreceptor body and to the substrate side chains? We suggest two alternatives that are not mutually exclusive. In one, CheR binds to the receptor body at a modification site but the major contributors to the energetics of binding are electrostatic interactions that are independent of whether the side chain is in the substrate or product form and independent of the chemoreceptor signaling conformation. This is plausible, since the positively charged side chains of CheR implicated in chemoreceptor interaction are not thought to bind directly to the substrate side chain on the receptor9, 37, 38 and the consensus sequence surrounding methyl‐accepting sites includes a negatively charged glutamyl residue adjacent to the modified position.41, 42, 43 Available evidence indicates that the conformational differences are subtle between the methylation‐on, kinase‐off and the methylation‐off, kinase‐on signaling states.1, 2, 3 Thus, the negative charges on the chemoreceptor surface are likely available for electrostatic interactions in both conformational states. In a second possibility, the CheR binding measured to a receptor lacking the pentapeptide occurs by electrostatic interactions between the enzyme and negatively charged portions of the receptor not necessarily near substrate sidechains. There are substantial regions of negative electrostatic potential outside the immediate vicinity of methyl‐accepting side chains on the surface of the chemoreceptor cytoplasmic domain (Figure 5). It could well be that electrostatic attraction of the positively charged region near the CheR active site and the negative surface potential of the chemoreceptor surface would result in enzyme binding at a region not directly at modification sites. Subsequent to such binding, the enzyme could diffuse to the substrate side chains by sliding along the negatively charged surface of the chemoreceptor or by hopping from interaction to interaction, kept close by electrostatic attraction.

3.2. Insights into CheR action

The kinetic and thermodynamic parameters determined in this study provide insights into the ways in which the methyltransferase interacts with chemoreceptors. The values indicate that once CheR is bound to a chemoreceptor, it is likely to remain associated with it or its neighbors. The enzyme will remain bound to a receptor‐borne pentapeptide for an average of approximately 5 s (see dwell times in Table 1 for pentapeptide‐bearing Tar4E and Tar4Q). During that time, CheR diffusion is limited by the tether to a volume of approximately 1.2 x 10⁻⁶ fL (see Experimental Procedures for the calculation), approximately one millionth of the 1 fL occupied volume of the cytoplasm. This results in an operational concentration of a single tethered CheR of approximately 1.4 mM in proximity to the chemoreceptor body and the methyl‐accepting sites. Association rate constants for CheR binding receptors lacking the pentapeptide averaged ∼2,500 M⁻¹ s⁻¹ (Table 2). At an operational concentration of 1.4 mM, tethered CheR would bind to the chemoreceptor body with a probability of 3.5 s⁻¹. Once bound, the enzyme would remain bound for an average of 2.5 s (Table 1). As a result, the equilibrium ratio between tethered CheR bound to the chemoreceptor body versus enzyme bound only to the pentapeptide tether would be approximately 9.3 (see Experimental Procedures). Thus, the enzyme would be poised for receptor modification. In addition, these calculations provide a quantitative basis for the observation that in a cell with an occupied volume of ∼1 fL the ∼150 molecules of CheR,44 at a nominal concentration of ∼0.25 μM, are concentrated in the region of chemoreceptor arrays.45

3.3. Concluding remarks

In biological signaling systems, it is often the case that a protein‐modifying enzyme exhibits differential activity for alternative signaling conformations of its protein substrate. The common assumption is that such differential activity reflects differential binding. The data presented here indicates that this is not the case for CheR and bacterial chemoreceptors but instead differential activity reflects differential catalysis. It seems likely that, across the range of biological signaling systems, alterative conformations of other protein substrates are differentially modified as the result of differential catalysis not differential binding.

4. EXPERIMENTAL PROCEDURES

4.1. Strains and plasmids

The host for plasmids carrying cheR or a chemoreceptor gene was E. coli K‐12 strain RP3098,46 which contains a deletion from flhA to flhD that eliminates the presence or expression of all chemoreceptor and che genes. PCR site‐directed mutagenesis of the indicated parental plasmids was used to create the following plasmids coding for the chemoreceptor designated in parentheses after the plasmid name: pML42 (Tar4E L378C) and pML46 (Tar4Q L378C) from pAL177 (Tar4E) and pAL180 (Tar4Q), respectively, both of which are derivatives of pNT201 (Tar)47; pNB12 (Tar4E‐6H L378C) and pNB4 (Tar4Q‐6H L378C) from pAL529 (Tar4E‐6H) and pAL533 (Tar4Q‐6H)33; pML36 (Tar4EΔpp‐6H L378C) and pML44 (Tar4QΔpp‐6H L378C) from pAL686 (Tar4EΔpp‐6H) and pML43 (Tar4QΔpp‐6H), both of which are derivatives of pAL61 (TarΔpp‐6H).21

4.2. Protein purification and Nanodisc formation

CheR was purified as described.48 Membrane scaffold protein MSP1D1(−)49 was purified using a Ni‐NTA column and the six‐histidine tag then cleaved, all as described.27 Tar4EΔpp‐6H L378C and Tar4QΔpp‐6H L378C were purified from isolated membrane in the presence of 2 mM dithiothreitol (DTT) using a Ni‐NTA column.27 DTT was removed using a spin desalting column (PD SpinTrap G‐25, GE Healthcare) and receptors biotinylated by reaction of the cysteine with EZ‐Link maleimide‐PEG₂‐biotin (Thermo Scientific, Rockford, IL) according to the manufacture's instruction. The resulting biotinylated receptors were mixed with MSP1D1(−) and a polar extract of total E. coli lipids (Avanti Polar lipids, Alabaster, AL) to prepare receptor‐containing Nanodiscs at approximately one dimer per disc essentially as described.27, 30

Tar4E L378C and Tar4Q L378C, which lacked a six‐histidine tag, were biotinylated after solubilization of isolated membrane containing the respective receptor using a methanethiosulfonate‐biotin reagent (Toronto Research Chemicals, Inc.) and purified using a column of SoftLink™ Soft Release Avidin Resin (Promega V2012, Madison, WI) and elution with 5 mM d‐biotin essentially as described.50 These biotinylated, purified receptors were mixed with MSP1D1(−) and a polar extract of total E. coli lipids (Avanti Polar lipids, Alabaster, AL) to prepare receptor‐containing Nanodiscs at approximately one dimer per disc essentially as described.27, 30 The mixture was applied to a NeutrAvidin resin (Thermo Scientific, catalog no 29204) column and eluted by 50 mM DTT which cleaved the disulfide linkage between the receptor cysteine and the biotin bound to the avidin resin. The receptors, again carrying an available cysteine at position 378, were then biotinylated by reaction of that cysteine with EZ‐Link maleimide‐PEG₂‐biotin as above after DTT removal using the spin desalting column as above. All preparations of receptor‐containing Nanodiscs were dialyzed into 50 mM Tris–HCl, pH 7.5, 0.5 mM EDTA, 10% glycerol, 100 mM NaCl. Prior to a binding experiment, the solution was changed using a spin desalting column to 10 mM HEPES‐potassium hydroxide, 150 mM potassium acetate, pH 7.6.

To characterize CheR binding to pentapeptide alone, we used a synthesized octapepide, EENWETFC that included the NWETF pentapeptide as well as two amino‐terminal glutamyl residues that conferred useful solubility and corresponded to the sequence of E. coli chemoreceptor Tsr adjacent to the pentapeptide, and a carboxyl‐terminal cysteinyl residue. To provide a means of immobilizing that peptide on a sensor tip, the cysteine was biotinylated by reaction with EZ‐Link maleimide‐PEG₂‐biotin according to the manufacture's instruction after DTT removal using the spin desalting column as above. Free reagent was removed using the same column and the biotinylated peptide coupled to the sensor tips by the same procedure used for biotinylated chemoreceptors.

4.3. Binding assays

The kinetics of interaction between CheR and chemoreceptors inserted in native lipid bilayers contained in Nanodisc was monitored using a BLItz instrument (ForteBio, Menlo Park, CA). The instrument utilizes biolayer interferometry to monitor in real time protein mass attached to the surface of a proprietary sensor tip. The resulting signal is not sensitive to protein in solution but only protein immobilized on the sensor tip. Chemoreceptors were immobilized on a buffer‐equilibrated, high precision streptavidin biosensor (SAX, ForteBio, Menlo Park, CA) by placing the sensor in 250 μl of a solution of biotin‐labeled, Nanodisc‐embedded receptors in 10 mM HEPES‐potassium hydroxide, 150 mM potassium acetate, pH 7.6 plus 2% (wt/vol) BSA (Sigma‐Aldrich, product #A1900). The sensor was washed by replacing the receptor‐containing tube with a tube of buffer alone. A time course of binding to the immobilized receptor was initiated by replacing the tube containing buffer with one containing a particular concentration of CheR. After an empirically determined number of seconds, a time course of dissociation was initiated by replacing that tube with a tube of buffer alone. This sequence was repeated with a new sensor tip for each of several concentrations of CheR. Figure S1 shows the results of a typical experiment and the fits to the data described below.

To test the correspondence between binding parameters determined by bio‐layer interferometry and the well‐characterized technique of isothermal titration calorimetry, we measured the kinetics of association and dissociation for CheR and a NWETF peptide not attached to a chemoreceptor but tethered with a linker to the sensor surface. The 1.9 ± 0.4 μM equilibrium dissociation constant (Table 1, row 1), calculated from the measured rate constants, was indistinguishable from the 1.8 μM value determined in free solution by calorimetry.20

We also tested the sensitivity of the binding assay to a minor change in affinity using pentapeptide‐bearing Tar with six additional residues added after the NWETF sequence. This addition reduces initial rates of CheR‐catalyzed methylation in vivo and in vitro to approximately half the rates for unaltered receptor.51 At the low CheR concentration used in those in vitro experiments, this would correspond approximately a twofold reduction in affinity. We measured binding of CheR to both Tar4E‐6H and Tar4Q‐6H and found the six‐residue extension reduced affinity, that is, increased K _D, close to twofold, in large part reflecting reductions in association rate constants (Table 1, rows 2, 3, 5, and 6; Figure S3). Thus, bio‐layer interferometry was sufficiently sensitive to detect twofold changes in binding parameters.

4.4. Data fitting and calculations

The response curves, collected at several CheR concentrations for each receptor form or condition were fit globally to the single‐site model provided in the instrument software. For each interaction, the fit generated a dissociation rate constant (k _off) and an association rate constant (k _on). Equilibrium dissociation constants (K _D) were calculated as k _off/k _on.

We estimated the effective concentration of pentapeptide‐tethered CheR assuming a Gaussian chain model and utilizing the Jacobson‐Stockmayer factor, which estimates the concentration of one end of a flexible chain in the vicinity of the other. That factor is:

$$j={\left(\frac{3}{2\pi *C_n*n*l^2}\right)}^{3/2}$$

C _n is the Flory characteristic ratio, n the number of links in the chain and l the chain unit length. Using the poly‐l‐alanine C _n of 9.5, n = 39 which represents the 30 residues of the flexible arm between the structured chemoreceptor and NWETF plus the nine residue lengths representing the 35 Å between the first residue of the pentapeptide and the CheR active site, and 3.8 Å for the unit length, we calculated an estimated effective concentration of tethered CheR of ∼1.4 mM.

We estimated an equilibrium association constant for pentapeptide‐tethered CheR binding to a substrate site on the structured body of the chemoreceptor using a model of binding and an analysis, suggested to us by Frederick Dahlquist, University of California, Santa Barbara, which postulated:

$$\mathrm{Tar}+R\stackrel{K_A^{\mathrm{pp}}}{\iff }\mathrm{Tar}\bullet R\stackrel{K_{\mathrm{eq}}}{\iff }{\mathrm{Tar}}_{\leftarrow }R$$

$$\mathrm{Tar}\Delta \mathrm{pp}+R\stackrel{K_A}{\iff }{\text{TarΔpp}}_{\leftarrow }R$$

where Tar is receptor carrying the pentapeptide, TarΔpp is receptor lacking the pentapeptide, R is CheR, Tar ∙ R is the complex of the receptor and CheR bound to the pentapeptide, Tar_←R is the complex of receptor and CheR with the enzyme bound to both a substrate site and the pentapeptide, TarΔpp_←R is the complex of CheR bound to a substrate site of a receptor lacking the pentapeptide and K _A's are equilibrium binding constants. Since CheR appears to bind to TarΔpp essentially independent of modification state or the presence of ligands for the enzyme or receptor, we using the average values for k _on and k _off in Table 1 to calculate K _A for the enzyme and receptor body as 6,700M⁻¹. We assumed that:

$$K_{\mathrm{eq}}={\left[R\right]}_{\mathrm{eff}}\bullet K_{\mathrm{A},}$$

where [R]_eff is the effective concentration calculated above. The resulting K _eq was 9.3.

CONFLICT OF INTEREST

The authors declare no conflicts of interest with this work.

Supporting information

Figure S1 Example time courses of CheR binding to functional Tar

Figure S2 Kinetic and thermodynamic parameters of binding of methyltransferase CheR to chemoreceptor Tar homodimers carrying the native NWETF sequence at the carboxyl terminus of the 35‐residue flexible arm.

Figure S3 A six‐histidine extension at the carboxyl terminus of receptor‐borne NWETF reduced CheR binding.

Li M, Hazelbauer GL. Methyltransferase CheR binds to its chemoreceptor substrates independent of their signaling conformation yet modifies them differentially. Protein Science. 2020;29:443–454. 10.1002/pro.3760