Crystal Structures of Beryllium Fluoride-Free and Beryllium Fluoride-Bound CheY in Complex with the Conserved C-Terminal Peptide of CheZ Reveal Dual Binding Modes Specific to CheY Conformation

Summary

Chemotaxis, the environment-specific swimming behavior of a bacterial cell is controlled by flagellar rotation. The steady-state level of the phosphorylated or activated form of the response regulator CheY dictates the direction of flagellar rotation. CheY phosphorylation is regulated by a fine equilibrium of three phosphotransfer activities: phosphorylation by the kinase CheA, its auto-dephosphorylation and dephosphorylation by its phosphatase CheZ. Efficient dephosphorylation of CheY by CheZ requires two spatially distinct protein-protein contacts: tethering of the two proteins to each other and formation of an active site for dephosphorylation. The latter involves interaction of phosphorylated CheY with the small highly conserved C-terminal helix of CheZ (CheZ_C), an indispensable structural component of the functional CheZ protein. To understand how the CheZ_C helix, representing less than 1% of the full-length protein, ascertains molecular specificity of binding to CheY, we have determined crystal structures of CheY in complex with a synthetic peptide corresponding to 15 C-terminal residues of CheZ (CheZ_200-214) at resolutions ranging from 2.0 Å to 2.3 Å. These structures provide a detailed view of the CheZ_C peptide interaction both in the presence and absence of the phosphoryl analog, BeF₃⁻. Our studies reveal that two different modes of binding the CheZ_200-214 peptide are dictated by the conformational state of CheY in the complex. Our structures suggest that the CheZ_C helix binds to a “meta-active” conformation of inactive CheY and it does so in an orientation that is distinct from the one in which it binds activated CheY. Our dual binding mode hypothesis provides implications for reverse information flow in CheY and extends previous observations on inherent resilience in CheY-like signaling domains.

Introduction

Chemotaxis enables motile eubacteria to regulate their swimming behavior in response to chemical gradients (reviewed in Bren et al.¹, Bourret and Stock² and Wadhams and Armitage³). In enteric species such as Salmonella enterica serovar Typhimurium and Escherichia coli, the key step in chemotaxis is the regulation of phosphorylation of the response regulator CheY in response to chemical signals in the environment. The steady-state level of phosphorylated CheY (P~CheY) determines the direction of flagellar rotation, which regulates the specific swimming behavior of the cell.

CheY is a switch molecule with multiple conformations subject to the phosphorylation state of a conserved active-site residue, ⁵⁷Asp (reviewed in Silversmith and Bourret⁴, Robinson et al.⁵ and Cho et al.⁶). The phosphorylated form of CheY is short-lived (T_1/2 ~ 10⁻¹ sec), owing to both auto-dephosphorylation by CheY as well as dephosphorylation by CheZ, a P~CheY-specific phosphatase. Dephosphorylation of CheY by CheZ requires two interactions: tethering of the two proteins to each other and formation of an active site for dephosphorylation. The former involves interaction of P~CheY with the small C-terminal helix of CheZ (CheZ_C helix).

Blat and Eisenbach⁷, first identified the C-terminal region of CheZ as the locus of this tethering interaction. Schuster et al.⁸ showed that addition of the CheZ_C peptide (CheZ_196-214) in solution enhances the phosphorylation kinetics of CheY but has no effect on its dephosphorylation. Subsequently, Zhou et al.⁹ reported the 2.9 Å crystal structure of CheY, activated with the phosphoryl analog, BeF₃⁻, and bound to CheZ_1-214, which presented a view of the CheZ-bound active site and addressed the structural explanation for the mechanism of CheZ-mediated dephosphorylation of CheY. The structure also provided a description of the CheY-CheZ tethering interaction, albeit limited due to apparent disorder of the region in the crystal.

The CheZ_C helix represents less than 1% of full-length (FL) CheZ (CheZ_1-214) and is an indispensable structural component of the functional CheZ protein. Its binding to P~CheY is essential for CheZ-mediated dephosphorylation of CheY.⁷ The high degree of sequence conservation of the CheZ_C region is consistent with this role (Figure 1). An examination of 39 CheZ proteins revealed that seven of the sixteen strictly conserved residues in the FL protein are localized within the 21 C-terminal residues. Six additional residues of the C-terminus are also highly conserved (Figure 1).

Figure 1.

Sequence conservation in CheZ_C. A ClustalW⁷² sequence alignment of the C-terminal 21 residues of CheZ from 39 bacterial species is shown. Accession codes are indicated in parentheses. Sequence numbering, indicated on the first line, corresponds to S. enterica serovar Typhimurium CheZ. Hydrophobic residues are colored red, polar residues green, and acidic residues blue. The yellow box corresponds to completely conserved residues 202, 206 to 209 and 212 to 213. Strongly conserved residues are highlighted by gray boxes corresponding to hydrophobic residues, residues 198, 199, 205 and 214, a magenta box to the polar and charged residue, residue 204, and a cyan box to the less bulky residue, residue 211.

In order to understand how the small CheZ_C helix specifically binds to CheY and to gain insight into the significance of the CheY-CheZ_C interaction in chemotaxis, we solved crystal structures of wild-type (WT) S. enterica CheY, in the presence of the phosphoryl analog, BeF₃⁻ complex^10-12 (CheY activated) as well as in the absence of BeF ₃⁻ (CheY inactive), each in complex with a synthetic peptide corresponding to 15 residues of the C-terminus of S. enterica CheZ (CheZ_200-214). We determined the structures from two different crystal forms, F432 and P2₁2₁2, at resolutions ranging between 2.0 Å and 2.3 Å.

Comparisons with previously determined structures of inactive CheY (metal-free¹³ and metal-bound¹⁴) and BeF₃⁻-activated CheY, free¹⁵ and bound to CheZ_1-214⁹, revealed distinct differences in peptide binding correlating with differences in CheY conformation. Our crystal structures suggest that the mode of binding of the CheZ_200-214 peptide to activated CheY is different from its binding to inactive CheY. The conformation of CheY in the inactive CheY-CheZ_200-214 complex exhibits both inactive and active features. From these structural studies, we cannot completely rule out that lattice-packing interactions have led to the trapping of an intermediate state in this complex. However, the observed conformation is similar to the “meta-active” sub-state of CheY, first documented by Simonovic and Volz.¹³ Therefore, the structures of the inactive CheY-CheZ_200-214 complex, reported here, suggest a possible functional role for the “meta-active” intermediate of CheY. These structures also argue against the previously postulated two-state model in which long-range conformational changes in CheY are obligatorily coupled.¹⁶

The different CheY-CheZ_200-214 structures illustrate the malleable nature of CheY-like signaling domains. They also suggest that the CheZ_200-214 peptide induces a greater change in CheY conformation upon binding to inactive CheY than upon binding to the activated form of CheY. This might provide an explanation for the phenomenon of peptide-induced acceleration of CheY phosphorylation.⁸ Our structural and mutational studies suggest that this coupling of CheY phosphorylation rate and peptide binding cannot be explained by differences in conformation, bulk and charge of side chains of individual conserved residues and instead involve global changes in the backbone conformation. Based on our observations, we have proposed dual modes of CheZ_C-CheY binding that may be relevant to the modulation of signaling in bacterial chemotaxis.

Results and Discussion

Crystal structures of BeF₃⁻-free and BeF₃⁻-bound CheY-CheZ_200-214 complexes from two different crystal forms

Co-crystals of S. enterica WT CheY and a synthetic peptide corresponding to S. enterica CheZ_200-214 were grown by the hanging drop vapor diffusion method, as described in Materials and Methods. Crystals of both inactive and activated CheY-CheZ_200-214 complexes were generated in the presence of the divalent cation, Mg²⁺, which plays a catalytic role at the active site of CheY.¹⁴ BeF₃⁻, a phosphoryl analog that forms a non-covalent complex with the active-site ⁵⁷Asp, was used to stabilize the active conformation of CheY.^10-12 The BeF₃⁻ species was obtained by using appropriate ratios of BeCl₂ and NaF, as previously described.^10,15 The N-terminus of the synthetic CheZ_200-214 peptide was protected by acetylation to ensure stability in solution, eliminating the charge on the α-amino group and reducing reactivity of the free N-terminus, which is prone to modification.

The CheY-CheZ_200-214 complex crystallized in two different space groups: the F-centered cubic F432 and the primitive orthorhombic P2₁2₁2. The F432 crystals were generated both in the presence as well as in the absence of BeCl₂ and NaF in three different buffer and pH conditions: Hepes (pH 7.5), Tris (pH 8.4) and CAPS (pH 10.5). To obtain the BeF₃⁻-bound F432 crystals, BeCl₂ and NaF were soaked into inactive F432 crystals of the CheY-CheZ_200-214 complex, as detailed in Materials and Methods. These crystals exhibited a range of crystal morphologies with well-defined edges including the most common cubic form (Figure 2(a)). Data from these crystals processed with one molecule per asymmetric unit and a Matthews coefficient of 5.1 ų/Da, yielding a solvent content of 76%. The P2₁2₁2 crystals were grown in MES (pH 6.0), in the presence of BeCl₂ and NaF. Crystals of this form grew as needles with rough edges and with fissures running across a few (Figure 2(a)). These data processed with one molecule per asymmetric unit with a Matthews coefficient of 1.94 ų/Da and a solvent content of only 36.6%, much lower than that of the F432 crystal form.

Figure 2.

Crystals and structures of CheY-CheZ_200-214 complex. (a) Crystal morphologies of ^F432YZ_200-214 (top two panels) and ^P2(1)2(1)2YZ_200-214 (bottom two panels). (b) Structure of BeF₃⁻-free ^F432YZ_200-214 solved from a crystal grown in Tris (pH 8.4). (c) Structure of BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214. (d) Structure of BeF₃⁻-bound ^F432YZ_200-214 solved from a crystal grown in Tris (pH 8.4). CheY molecules in the ^F432YZ_200-214 structures ((b) and (d)) are shown in gray and that in the ^P2(1)2(1)2YZ_200-214 structure ((c)) is shown in wheat. The CheZ_200-214 peptide molecules in BeF₃⁻-free ^F432YZ_200-214 are shown in cyan ((b)), in BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214, in orange ((c)) and in BeF₃⁻-bound ^F432YZ_200-214, in deep blue ((d)). Active-site water molecules are shown in deep red, Mg²⁺ ions are shown in magenta and BeF₃⁻ complexes are shown in yellow and the side chains of key active-site and switch residues are depicted in ball and stick models.

The molecular replacement program Phaser¹⁷ was used to obtain phases for the structures, using as a search molecule, a poly-alanine model of the BeF₃⁻-activated CheY structure¹⁵ (PDB Accession code – 1FQW) lacking the β4-α4 loop (residues 88-92). In most response regulator receiver domains, as is the case in CheY¹⁸, the β4-α4 loop is flexible¹⁹ and undergoes substantial change upon phosphorylation. Six structures in the F432 crystal form, corresponding to different crystallization conditions, and one structure in the P2₁2₁2 crystal form were solved. The statistics for data collection and those for refinement for all seven structures are summarized in Table 1. The structures were refined to resolutions ranging from 2.0 Å to 2.3 Å. All seven final models include one molecule of CheY with residues 2 to 129 and one molecule of the CheZ_200-214 peptide. In case of some of the structures in the F432 crystal form, one to five N-terminal residues of the CheZ_200-214 peptide could not be included because of disorder, possibly arising from the solvent-exposed nature of the peptide in this crystal form. The structures were refined to R-factors ranging between 0.186 and 0.220 and R_free ranging between 0.207 and 0.246 with good chemical geometries, as listed in Table 1.

Table 1.

Summary of data collection and refinement statistics

Lattice	F432	F432	F432	F432	F432	F432	P21212
BeF3−(−/+)	+	+	+	−	−	−	+
pH	10.5	8.4	7.5	10.5	8.4	7.5	6.0
Data collection
Wavelength (Å)	0.979	1.072	1.072	0.979	0.979	0.979	1.072
Cell a=	198.7	197.8	197.9	198.8	198.3	197.1	54.2
(Å) b=	198.7	197.8	197.9	198.8	198.3	197.1	54.2
c=	198.7	197.8	197.9	198.8	198.3	197.1	54.2
α=β=γ =	90°	90°	90°	90°	90°	90°	90°
Resolutiona (Å)	2.0	2.0	2.0	2.1	2.2	2.0	2.0
Rsym b	0.058/	0.068/	0.071/	0.100/	0.097/	0.078/	0.052/
	(0.236)	(0.377)	(0.261)	(0.358)	(0.331)	(0.251)	(0.080)
Completeness (%)	99.9 /	99.9/	96.9 /	99.9 /	99.7/	97.1/	99.4
	(99.5)	(100.0)	(98.5)	(98.7)	(100.0)	(97.0)	(99.3)
I/σI	34.9/	35.7/	35.5/	21.3/	22.8/	22.0/	38.6/
	(7.1)	(7.0)	(8.2)	(6.3)	(8.0)	(6.6)	(21.5)
Refinement
Resolutiona (Å)	2.0	2.0	2.0	2.1	2.3	2.0	2.0
Rworking c	0.210	0.220	0.212	0.197	0.193	0.200	0.186
Rfree d	0.224	0.246	0.228	0.207	0.220	0.213	0.232
No. of reflections	20,589	20,593	20,015	17,947	13,445	19,511	7,864
No. of protein atoms	1058	1067	1090	1083	1090	1078	1097
Small moleculese
Mg2+	1	1	1	−	−	−	1
BeF3−	1	1	1	−	−	−	1
Water	81	53	89	80	87	86	90
CAPS	2	−	−	2	−	−	−
Tris	−	1	−	−	1	−	−
Sulfate	1	1	1	1	1	1	−
Glycerol	−	1	−	−	1	−	−
Acetyl group	−	−	−	−	−	−	1
r.m.s.d. values from ideality
Bond lengths (Å)	0.010	0.013	0.011	0.011	0.013	0.011	0.008
Bond angles (°)	1.250	1.281	1.220	1.184	1.217	1.409	1.116
B-valuef (Å2)	25.60	28.76	24.13	20.04	21.28	27.45	10.92

Values corresponding to highest resolution shells are shown in parentheses.

^aHigh resolution limits for data collection and refinement are indicated.

^b${\mathit{R}}_{\mathit{sym}}=\frac{\Sigma \mid {\mathit{I}}_{\mathit{obs}}-{\mathit{I}}_{\mathit{avg}}\mid }{\Sigma {\mathit{I}}_{\mathit{avg}}}$, where I_obs = observed integrated intensity and I_avg = average integrated intensity from multiple measurements.

^c${\mathit{R}}_{\text{working}}=\frac{\Sigma \mid {\mid {\mathit{F}}_{\mathit{obs}}\mid }_{\left(\mathit{hkl}\right)}-{\mid {\mathit{F}}_{\mathit{calc}}\mid }_{\left(\mathit{hkl}\right)}\mid }{\Sigma {\mid {\mathit{F}}_{\mathit{obs}}\mid }_{\left(\mathit{hkl}\right)}}$, where F_obs and F_calc are the observed and calculated structure factor amplitudes for hkl indices, respectively.

^dR_free is identical to R_working but is calculated from 10% of the reflections set aside as a disjoint set prior to refinement.

^eNumber of small molecules in each case are indicated.

^fOverall B-value includes all atoms in a given model.

The F_O-F_C difference maps contoured at +3σ for all activated models showed the presence of a BeF₃⁻ species and a Mg²⁺ ion exhibiting octahedral coordination geometry at the active site. Following refinement, BeF₃⁻ refined with 100% occupancy in all activated structures, except one. In the structure solved from an F432 crystal grown in Hepes (pH 7.5), which was soaked in BeCl₂ and NaF prior to data collection (see Materials and Methods), the occupancy for BeF₃⁻ refined to only 65%. The BeF₃⁻-free or inactive CheY-CheZ_200-214 complex was crystallized only in the F432 crystal form. At the active site of CheY in all these structures, a water molecule with tetrahedral geometry occupies the position of Mg²⁺. The lack of metal binding in this case, in spite of the presence of 7 mM MgCl₂ during crystallization, likely results from the high ionic strength of the crystallization solutions (see Materials and Methods). Low metal occupancy in the presence of high concentrations of ammonium sulfate has been previously observed.¹⁴ The adverse effect of high ionic strength on metal binding in the F432 crystal form is overcome in the presence of BeF₃⁻, which provides additional ligands for coordination, as observed in the BeF₃⁻-bound CheY-CheZ_200-214 structures solved from the F432 crystals (BeF₃⁻-bound ^F432YZ_200-214). Additional small molecules included in the final models are summarized in Table 1.

CheY is a 129-residue doubly-wound α/β protein with a central sheet of five beta strands, surrounded by five alpha helices. The CheZ_200-214 peptide, in helical configuration, is bound to the α4-β5-α5 face of the CheY molecule in all seven structures, reported in this manuscript, as is the CheZ_C helix in the CheY-CheZ_1-214 structure.⁹ The overall structures of all six models in the F432 crystal form are similar. Superpositions of the CheY molecules from the BeF₃⁻-bound and BeF₃⁻-free ^F432YZ_200-214 complexes yield main chain root mean square deviations (r.m.s.d.) of 0.12 Å to 0.27 Å for CheY and 0.22 Å to 0.28 Å for CheZ_200-214. The CheY-CheZ_200-214 structures in the F432 crystal form (^F432YZ_200-214) differ from that in the P2₁2₁2 crystal form (^P2(1)2(1)2YZ_200-214) by main chain r.m.s.d. values ranging between 0.70 Å and 0.83 Å for CheY and between 0.46 Å and 0.60 Å for CheZ_200-214. The BeF₃⁻-free and BeF₃⁻-bound ^F432YZ_200-214 and the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 structures are shown in Figure 2(b), Figure 2(d) and Figure 2(c), respectively.

Comparison of CheY-CheZ_C contacts – dual binding modes

An examination of the CheY-CheZ_200-214 interface in the ^F432YZ_200-214 and the ^P2(1)2(1)2YZ_200-214 structures revealed that the CheZ_200-214 helix in the two crystal forms is positioned on the α4-β5-α5 face of CheY in two distinct orientations, as illustrated in Figure 3(a) and Figure 3(b). For ease of discussion, the orientation of the CheZ_200-214 helix observed in the ^F432YZ_200-214 structures will be referred to as the ^F432mode and that observed in the ^P2(1)2(1)2YZ_200-214 structure, as the ^P2(1)2(1)2mode.

Figure 3.

The CheZ_200-214 peptide-CheY interface. Ribbon representation of (a) the ^F432YZ_200-214 interface and (b) the ^P2(1)2(1)2YZ_200-214 interface. The side chains of key contacting residues are illustrated as ball and stick models and hydrophobic contacts are shown as light green patches. (c) Relative B-factors of CheZ_200-214 in the CheY-CheZ_200-214 structures. The relative B-factor versus CheZ residue number plot in BeF₃₋-free ^F432YZ_200-214 is shown in cyan, that in BeF₃⁻-bound ^F432YZ_200-214 in deep blue and that in BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 in orange. B_residue is the overall B-factor for each residue and B_CheZ is the overall B-factor for all atoms of CheZ_200-214 included in the final model. (d) Schematic representation of the CheY-CheZ_200-214 contacts. The ^F432Z_200-214 primary sequence in cyan and the ^P2(1)2(1)2Z_200-214 primary sequence in orange are shown on either side of the C-terminal half of CheY, represented in secondary structural elements. Participating residues are highlighted. Hydrophobic contacts are illustrated as solid grey lines, salt bridges as dashed black lines and hydrogen bonds as solid black lines. The BeF₃⁻-free and BeF₃⁻-bound ^F432YZ_200-214 structures solved from crystals grown in Tris (pH 8.4) are used as representatives of ^F432YZ_200-214 structures in this figure.

In both modes, there is some disorder to the termini of the CheZ_200-214 helix, but disorder occurs at opposite ends in the two modes. In the ^F432mode, the CheZ_200-214 helix (^F432Z_200-214) is parallel to the α4 helix of CheY and the N-terminal region of the ^F432Z_200-214 helix is highly disordered and is exposed to solvent (Figure 3(a)). Undefined density in the F_O-F_C difference maps, precluding the placement of one to five N-terminal residues in the final models, and relatively higher B-factors of the N-terminal region of CheZ_200-214 compared to the rest of CheZ_200-214 in the ^F432YZ_200-214 structures (Figure 3(c)) are consistent with this observation. At the C-terminus, the ^F432Z_200-214 peptide is anchored by the phenyl ring of the terminal residue, ²¹⁴Phe that hooks into a hydrophobic pocket on the α4-β5-α5 face of the CheY protein. This pocket is created by the solvent-buried conformation of the side chain of ¹⁰⁶Tyr on the β5 strand of CheY. ¹⁰⁶Tyr is one of the switch residues involved in conformational change that accompanies phosphorylation of CheY (reviewed in Robinson et al.⁵).

In the ^P2(1)2(1)2mode, the CheZ_200-214 helix (^P2(1)2(1)2Z_200-214) is parallel to the α5 helix of CheY (Figure 3(b)). The N-terminus of the ^P2(1)2(1)2Z_200-214 helix is ordered and is buried within the α4-β5-α5 face while the C-terminus is disordered, consistent with the relatively higher B-factors of the C-terminal region of CheZ_200-214 compared to those of the rest of the peptide in the ^P2(1)2(1)2YZ_200-214 structure (Figure 3(c)).

The high-resolution crystal structures of the CheY-CheZ_200-214 complexes provided the opportunity to analyze details of the protein-protein interactions between CheY and the CheZ_200-214 peptide. The two different interfaces involve similar types of interactions. However, different residues mediate the interactions in the two binding modes (Figure 3(d) and Table 2). Both interfaces are predominantly hydrophobic. In the two modes of binding, the hydrophobic interface is overlapping but it is flanked by distinct hydrogen bond and salt bridge contacts. Buried surface analyses revealed that the C-terminal residues, ²¹²Leu, ²¹³Gly and ²¹⁴Phe of the ^F432Z_200-214 peptide contribute to 30% of the surface buried at the ^F432YZ_200-214 interface. In contrast, contribution to buried surface at the ^P2(1)2(1)2YZ_200-214 interface is substantially higher from the ^P2(1)2(1)2Z_200-214 N-terminus than the C-terminus. The C-terminal residues, ²¹²Leu and ²¹⁴Phe contribute to only 17% of the buried surface at the ^P2(1)2(1)2YZ_200-214 interface.

Table 2.

Comparison of CheZ_C peptide contacts with CheY

Hydrophobic interactionsa
CheY	CheZC	F432YZ200-214c	P2(1)2(1)2YZ200-214	CheY-CheZ1-2149
95Ile	200Ala	−	+d	− d
95Ile	208Leu, 209Leu	+	+	+
95Ile	212Leu	+	−	−
95Ile	205Val	+	+	+
96Ile	208Leu	+	−	−
99Ala	209Leu	−	+	+
99Ala	212Leu	+	−	−
99Ala	214Phe	−	+	− e
106Tyr	205Val	−	+	+
106Tyr	209Leu	+	−	−
106Tyr	214Phe	+	−	−
108Val	205Val	−	+	+

Hydrogen bonds b
CheY		CheZC		F432YZ200-214c	P2(1)2(1)2YZ200-214	CheY-CheZ1-2149
90Ala	O	200Ala	N	−	2.80d	− d
106Tyr	O	202Gln	Nε2	−	2.99	2.85
108Val	N	202Gln	Oε1	−	3.02	2.42
119Lys	Nζ	214Phe	O	2.78f	−	− e

Salt bridges b
CheY		CheZC		F432YZ200-214c	P2(1)2(1)2YZ200-214	CheY-CheZ1-2149
119Lys	Nζ	206Asp	Oσ1	−	3.16	2.73
119Lys	Nζ	206Asp	Oσ2	−	2.78	2.91
119Lys	Nζ	214Phe	OXT	2.78f	−	−

^aCarbon-carbon distances in the hydrophobic interactions are within 4.0 Å.

^bHydrogen bonds and salt bridge contact distances are between 2.6 Å and 3.2 Å.

^cValues are given for the BeF₃⁻-free ^F432YZ_200-214 structure solved from a crystal grown in Tris (pH 8.4). Contact distances differ marginally in the rest of the ^F432YZ_200-214 models.

^dIt is not clear if the contacts involving ²⁰⁰Ala of CheZ, observed in the ^P2(1)2(1)2YZ_200-214 structure, is physiological or an artifact of acetylating the N-terminus of CheZ_200-214 peptide. The backbone NH group of ²⁰⁰Ala should still be available for hydrogen bond interaction if the chain is not truncated at this residue, as in CheZ_1-214. However, in contrast to the other contacting residues in both CheY and CheZ, only ²⁰⁰Ala is not conserved (Figure 1). The absence of this contact in the CheY-CheZ_1-214 structure is due to the absence of ²⁰⁰Ala from the final CheY-CheZ_1-214 model⁹.

^e214Phe is not included in the final CheY-CheZ_1-214 model due to high disorder⁹.

^fThe lone electron pair on the C-terminal carboxylate group of CheZ_200-214 should exist in resonance with the carbonyl group of the C-terminal residue ²¹⁴Phe. Hence, in all the ^F432YZ_200-214 structures, the salt bridge between the positively-charged ¹¹⁹Lys sidechain and the negatively-charged C-terminus of the peptide would predominate over the hydrogen bond between the ¹¹⁹Lys sidechain and the backbone carbonyl of ²¹⁴Phe of the CheZ_200-214 peptide.

Comparison of binding modes of the CheZ_C helices in the CheY-CheZ structures was initiated by structural superposition of the CheY molecules in the CheY complexes. Figure 4 depicts such a structural alignment of the three CheY-CheZ_C models (^F432YZ_200-214, ^P2(1)2(1)2YZ_200-214 and CheY-CheZ_1-214⁹). In the CheY-CheZ_1-214 structure, the CheY active site contacts the coiled coil region of CheZ and the α4-β5-α5 face binds to the CheZ C-terminal tail, located at the end of a long flexible linker (CheZ_169-200) that is disordered and absent from the final model. Thus, the region of the CheY-CheZ_1-214 model with the C-terminus of CheZ contacting CheY is analogous to the models of CheY bound to a synthetic peptide corresponding to the 15 C-terminal residues of CheZ, described in this work.

Figure 4.

Ribbon diagrams of CheY-CheZ_C structures upon superposition of CheY showing different orientations of CheZ_C. CheY molecules in ^F432YZ_200-214, ^P2(1)2(1)2YZ_200-214 and CheY-CheZ_1-214⁹ structures are shown in light gray and the respective CheZ_C helices are shown in cyan, orange and magenta, respectively. The BeF₃⁻-free ^F432YZ_200-214 structure solved from a crystal grown in Tris (pH 8.4) is used in this figure as a representative of all six ^F432YZ_200-214 structures.

It is apparent from Figure 4 that the orientation of the CheZ_C helix in CheY-CheZ_1-214 is similar to that in ^P2(1)2(1)2YZ_200-214. In both these cases, the interacting CheZ_C helix is positioned at the α4-β5-α5 face in the ^P2(1)2(1)2mode and is parallel to the α5 helix of CheY, tilted by ~30° with respect to the axis of the CheZ_C helix in the ^F432mode. Comparison of the details of the interacting residues in the three CheY-CheZ_C models (^F432YZ_200-214, ^P2(1)2(1)2YZ_200-214 and CheY-CheZ_1-214) reveal that the specific hydrophobic interactions, hydrogen bonds and salt bridges observed for the ^P2(1)2(1)2mode in both the CheY-CheZ_1-214 and the ^P2(1)2(1)2YZ_200-214 models are almost identical, but distinct from those observed for the ^F432mode (Table 2).

Comparison of the crystal structures of CheY bound to CheZ_200-214 with previously published crystal structures of CheY bound to the N-terminal peptide of FliM^20,21, a component of the flagellar motor and to the P2 domain of the histidine kinase CheA^22-24 reveal that the binding interfaces of CheY with its binding partners are overlapping but not identical, a paradigm that was formerly pointed out in mutation²⁵ and NMR²⁶ studies. Additionally, the residues involved in the CheY-CheZ_200-214 binding interface, as revealed in the CheY-CheZ_200-214 structures, reported here, were also implicated in previous studies.^25,26 Two different protein interfaces have been previously observed in crystal structures of CheY bound to the P2 domain of CheA.²² This study provides the first evidence for dual modes of binding of CheY to CheZ.

Comparison of binding interfaces, model quality, experimental conditions and crystal lattice influences

An understanding of the functional significance of dual binding modes of CheY and CheZ, the ^F432mode and the ^P2(1)2(1)2mode, necessitates a thorough comparison of all available crystal structures of CheY-CheZ_C complex (^F432YZ_200-214, ^P2(1)2(1)2YZ_200-214 and CheY-CheZ_1-214⁹). A detailed analysis of interface characteristics, structural model features, crystallization conditions and influences of the crystal lattice in these structures is presented in Table 3. For simplicity, the BeF₃⁻-free ^F432YZ_200-214 structure solved from a crystal grown in Tris (pH 8.4) will be used as a representative of all six ^F432YZ_200-214 models.

Table 3.

Comparison of interface characteristics, model quality, experimental conditions and crystal lattice characteristics

Models	F432YZ200-214a	P2(1)2(1)2YZ200-214	CheY-CheZ1-2149
Binding modes	F432Mode	P2(1)2(1)2Mode	P2(1)2(1)2Mode
Interface characteristics
Buried surface area (Å2)	789.6 (762 – 821)	1044	677
H-bondsb	1	2	1
Salt bridgesb	1	2	2
Hydrophobic contactsb	8	8	5
Gap volume indexc	2.14 (1.39 – 2.16)	1.25	2.70
Model quality
Resolution (Å)	2.3 (2.0 – 2.3)	2.0	2.9
Rworking	0.193 (0.193 – 0.220)	0.186	0.279
Rfree	0.220 (0.207 – 0.246)	0.232	0.298
CheZC B-valuesd (Å2)	38.8 (29.5 – 46.0)	17.67	164.8
No. of CheZC residuese	14 (10 – 14)	15	13
Experimental conditions
CheZ FL/peptidef	peptide	peptide	FL
Crystallization conditions
[Protein] (mM)g	0.9/ 3.6	0.8/ 1.2	0.264/ 0.264
Chemical conditions
[BeCl2] (mM)	5.0	35.0	3.6
[NaF] (mM)	30.0	35.0	10.0
[MgCl2] (mM)	7.0	7.0	10.0
Buffer (0.1 M)	Hepes (pH 7.5)/ Tris (pH 8.4)/ CAPS (pH 10.5)	MES (pH 6.0)	Bicine (pH 8.5)
Precipitant	2 M ammonium	0.05 M sodium	0.2 M ammonium
	sulfate	phosphate	acetate
	(1.8 M – 2.3 M)	(monobasic)	30% isopropanol
	0.2M lithium sulfate	37.5% PEG-8000
Cryoprotectanth	20% glycerol/	18% PEG-400 2% sucrose	50% sucrose
Temperature	25°C	25°C	4°C
Crystal lattice	F432	P21212	P43212

^aValues are given for the BeF₃⁻-free ^F432YZ_200-214 structure solved from a crystal at Tris (pH 8.4). Range of values within parenthesis pertains to all six ^F432YZ_200-214 structures.

^bHydrogen bonds and salt bridges are enumerated as atom-atom contacts. Hydrophobic contacts are enumerated as residue-residue contacts.

^cGap volume index is defined by the volume of gaps between two interacting molecules per Ų of interface accessible surface area.

^dAverage B-value indicates the averaged B-value for all modeled backbone and side-chain atoms of the CheZ_C region in a given model.

^eNumber of CheZ_C residues included in the models is given in each case.

^fCheZ full-length (FL) protein or peptide is indicated in each case.

^gConcentrations of CheY/ CheZ_200-214 peptide are given for the CheY-CheZ_200-214 structures. Concentrations of CheY/ CheZ_1-214 protein are given for the CheY-CheZ_1-214 structure.

^hFinal concentrations of cryoprotectant solutions are given.

Analyses of buried surface at the CheY-CheZ_C interfaces revealed that the specific interactions in the ^F432mode provide ~790 Ų of buried surface area while those in ^P2(1)2(1)2mode contribute 1044 Ų and 680 Ų at the interfaces in the ^P2(1)2(1)2YZ_200-214 and CheY-CheZ_1-214 structures, respectively. The difference in buried surface between ^P2(1)2(1)2YZ_200-214 and CheY-CheZ_1-214, in spite of an identical mode of binding, is due to the fact that two residues, ²⁰⁰Ala and ²¹⁴Phe in the CheZ_C region are absent from the CheY-CheZ_1-214 model due to disorder but are well-ordered in the ^P2(1)2(1)2YZ_200-214 model. The surface of CheY buried by the peptide in the two binding modes is largely overlapping but not identical. This contrast arises from the differences in the specific contacts, as discussed in the previous section and as detailed in Table 2. Together these moderately distinct interactions contribute to comparable buried surface areas.

The solvent-exposed face of the CheZ_C helix is similar in both binding modes (data not shown). Difference in gap volume index, a measure of complementarity of interacting surfaces²⁷, is only moderate and this parameter does not strongly favor one binding mode over the other. Notably, all the residues in CheY that contact CheZ in either binding mode are strongly conserved among the proteobacters that synthesize CheZ (data not shown). Similarly, in the C-terminus of CheZ, most residues that contact CheY (Table 2) are completely conserved (Figure 1). This conservation supports the biological significance of the protein-protein interactions observed in both modes of binding.

The quality of the models of the CheY-CheZ_200-214 and CheY-CheZ_1-214 complexes differs substantially. The F_o-F_c difference maps contoured at similar sigma levels reveal complete backbone and discernible side chain density for the CheY-CheZ_200-214 peptide models determined at ~2.0 Å from both crystal forms, F432 and P2₁2₁2, in contrast to limited backbone and a complete absence of side chain density for the CheY-CheZ_1-214 model determined at 2.9 Å resolution (data not shown). Both the ^F432YZ_200-214 and ^P2(1)2(1)2YZ_200-214 structures have substantially lower R_free values than the CheY-CheZ_1-214 structure, indicating better agreement of the CheY-CheZ_200-214 models to the measured data (Table 3). The average crystallographic B-factors of all the CheZ_C atoms in the ^F432YZ_200-214 and ^P2(1)2(1)2YZ_200-214 structures are 38.8 Ų and 17.7 Ų, respectively, while that in the CheY-CheZ_1-214 structure is 164.8 Ų. The lower B-factors indicate a more defined spread of electron density, less dynamic mobility and less errors in model building for the structural models of the CheY-CheZ_200-214 peptide complex (Table 3). However, while the poor model quality of the CheY-CheZ_1-214 model raises questions regarding details of the CheY-CheZ_C interface in this structure, the F_o-F_c maps suggest no ambiguity in the orientation of the CheZ_C helix axis relative to the α4-β5-α5 face of CheY.⁹ This is further strengthened by the fact that contact analyses show that both the orientation and the specific interactions at the CheY-CheZ_C interface in the CheY-CheZ_1-214 structure are similar to those in ^P2(1)2(1)2YZ_200-214, which exhibits better model characteristics.

The comparative analyses presented above indicate that neither of the two modes of binding of the CheZ_200-214 peptide to CheY appears to be an obviously better interface. Differences in experimental conditions (Table 3) are likely to have influenced the conformations observed, though none provides a readily discernable explanation for the different modes of binding. Crystals of the CheY-CheZ_1-214 complex were generated by co-crystallization of E. coli CheY and FL E. coli CheZ (CheZ_1-214) whereas crystals of CheY-CheZ_200-214 peptide complex, reported in this work, were generated by co-crystallization of S. enterica CheY and a synthetic peptide corresponding to the 15 C-terminal residues of S. enterica CheZ (CheZ_200-214). The C-terminal residues 200 to 214 in E. coli and S. enterica CheZ are identical (Figure 1) and the three minor amino acid differences between E. coli and S. enterica CheY proteins map outside the α4-β5-α5 binding face. Likewise, contacts of CheY with the N-terminal regions of CheZ_1-214 occur near the CheY active site and do not appear to impart long-range perturbations to the α4-β5-α5 binding face that interacts with the CheZ C-terminal peptide. In the ^F432YZ_200-214 structures solved from crystals grown in CAPS (pH 10.5), a CAPS buffer molecule interacts with both CheY and CheZ_200-214. However, the absence of this molecule in the structures of the complex solved from F432 crystals grown at different buffer and pH conditions suggests that the orientation of the CheZ_200-214 peptide in the ^F432mode is not influenced by the specific crystallization buffer.

Contributions of lattice contacts are more difficult to assess. In the F432 crystal lattice, each asymmetric unit is subject to three symmetry operations, one three-fold and two two-fold operators (Figure 5(a) and Figure 5(b)). The CheZ_200-214 peptide in this lattice is part of the 3-fold interface (Figure 5(a)). The CheZ_200-214 peptide makes nonspecific van der Waals interactions with parts of the α2 helix of a symmetry-related CheY molecule with contact distances >3.5 Å, but there are no specific hydrogen bonds or salt bridges that appear to constrain the position of the peptide. On the other hand, the P2₁2₁2 crystal lattice is characterized by three mutually perpendicular two-fold symmetry operators, two of which are associated with 2₁ translational symmetry functions (Figure 5(c)). The CheZ_200-214 peptide in the P2₁2₁2 lattice is involved in a few specific electrostatic contacts in addition to nonspecific van der Waals interactions with the α1 and α2 helices of a CheY molecule in the neighbouring asymmetric unit.

Figure 5.

Symmetry-related molecules of CheY-CheZ_200-214 in the F432 lattice shown at the 3-fold axis (a) and at the 4-fold axis (b) and in the P2₁2₁2 crystal lattice (c). CheY molecules in the F432 lattice are shown in gray and those in the P2₁2₁2 lattice are shown in wheat. The CheZ_200-214 peptide molecules in the F432 lattice are shown in cyan and those in the P2₁2₁2 lattice are shown in orange. The BeF₃⁻-free ^F432YZ_200-214 structure solved from a crystal grown in Tris (pH 8.4) is used as a representative of all six ^F432YZ_200-214 structures in (a) and (b).

Notably, the CheZ_200-214 peptide in both the F432 and the P2₁2₁2 lattices is positioned at interfaces of symmetry elements and contributes to 40% of the total buried surface due to lattice contacts (data not shown). Nonetheless, it is difficult to argue that the binding modes are solely a consequence of crystal packing. Although it cannot be ruled out that the peptide orientation in the ^F432mode is influenced by the van der Waals interactions imposed by the cubic F432 lattice, analyses of the CheY-CheZ_200-214 interfaces in these structures suggest that the contact displays all the characteristics of bonafide biological interfaces. Furthermore, any assumption of bias imposed on the orientation of the peptide in the P2₁2₁2 lattice is undermined by the fact that the peptide orientation in the P2₁2₁2 lattice is identical to that in the CheY-CheZ_1-214 P4₃2₁2 lattice, in which the CheZ_C helix is not part of any crystal contacts.⁹

Comparison of CheY conformational states

CheY is a switch protein with different conformations dictating distinct signaling states (reviewed in Robinson et al.⁵ and Cho et al.⁶). The ability of CheY to signal in bacterial chemotaxis is regulated by its phosphorylation state. Phosphorylation at the active site stabilizes an active conformation, in which the α4-β5-α5 signaling face of CheY, distant from the active site, is altered relative to the inactive conformation. Structural changes promoted by phosphorylation are conserved in most bacterial response regulator receiver domains. Structural studies of N-terminal receiver domains activated by phosphorylation or BeF₃⁻ are the basis of the present understanding of these conserved changes and have defined signatures for the active conformation.^15,18,28-35

In activated CheY, one of the phosphoryl oxygens (or fluorines of BeF₃⁻) serves as one of the six conserved ligands for the octahedrally coordinated divalent cation at the active site. Based on previous mutational and structural evidence, CheY activation is associated with three major changes: 1) displacement of the protein backbone of the conserved residue ⁸⁷Thr at the tip of the β4 strand towards the active site, allowing a strong hydrogen bond interaction with the phosphoryl group (or BeF₃⁻), 2) repositioning of the conserved residue ¹⁰⁹Lys on the β5-α5 loop, enabling a salt bridge interaction with the phosphoryl group (or BeF₃⁻) and ¹²Asp, one of the three conserved aspartate residues at the active site, and 3) rotameric conversion of the conserved aromatic switch residue ¹⁰⁶Tyr on the β5 strand from a solvent-exposed to a solvent-buried conformation.^{15,16,18,36-39} According to previous dynamic studies and NMR analysis of inactive CheY-Mg²⁺, the β4-α4 loop in the inactive protein is flexible and the active-site interactions involving ⁸⁷Thr and ¹⁰⁹Lys, promoted by phosphorylation or binding of the BeF₃⁻ ligand, influence the protein backbone in the β4-α4 and the β5-α5 loop regions, respectively, thereby stabilizing the active conformation.^15,18,40 Whether these structural changes occur in conjunction with, or as a consequence of, one another is not clear. Although a mechanism of activation in FixJ_N was suggested based on a molecular dynamics study, further experimental evidence for such a mechanism is lacking.¹⁹

The conformational states of CheY in the CheZ_200-214 peptide-bound complexes (^F432YZ_200-214 and ^P2(1)2(1)2YZ_200-214) were compared with previously determined structures of the Mg²⁺-bound inactive state¹⁴ (PDB Accession code – 2CHE) and the BeF₃⁻-activated state¹⁵ (PDB Accession code – 1FQW). Least-squares superpositions of the main chain atoms of CheY molecules in the ^F432YZ_200-214 and ^P2(1)2(1)2YZ_200-214 structures with those of Mg²⁺-bound inactive and BeF₃⁻-activated CheY structures were performed. Side chain conformations of the conserved active-site (Figure 6(a) and Figure 6(c)) and switch residues (Figure 6(b) and Figure 6(d)) are illustrated in superpositions with inactive Mg²⁺-bound CheY (Figure 6(a) and Figure 6(b)) and with BeF₃⁻-activated CheY (Figure 6(c) and Figure 6(d)).

Figure 6.

Comparison of active-site and switch residues in the CheY-CheZ_200-214 complexes with those in inactive and active CheY. CheY molecules in BeF₃⁻-free ^F432YZ_200-214 (key features shown in cyan), BeF₃⁻-bound ^F432YZ_200-214 (key features shown in deep blue) and BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 (key features shown in orange) are superposed on inactive Mg²⁺-bound CheY (2CHE)¹⁴ (key features shown in red) in (a) and (b) and on BeF₃⁻-activated CheY (1FQW)¹⁵ (key features shown in green) in (c) and (d), focusing on the active site in (a) and (c), and on the activation-sensitive α4β4α5 region in (b) and (d). The side chains of key active-site and switch residues are illustrated as ball and stick models. The Mg²⁺ ions (magenta), coordinating water molecules (deep red) and the BeF₃⁻-complexes (yellow) at the active site in BeF₃⁻-bound ^F432YZ_200-214 are shown in (a) and those in BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 are shown in (c). The BeF₃⁻-free ^F432YZ_200-214 and the “metaactive” conformer of the BeF₃⁻-bound ^F432YZ_200-214 structures solved from crystals grown in Tris (pH 8.4) are used as representatives of ^F432YZ_200-214 structures in this figure.

Following a least-squares superposition procedure, ${\overrightarrow{\delta }}_3$ analysis was performed to fully assess the extent of differences in the tertiary structures of the CheY-CheZ_200-214 complexes and the previously determined structures of CheY (see Materials and Methods). $\mid {\overrightarrow{\delta }}_3\mid$ reflects the magnitude of a main chain displacement vector that is a measure of individual atomic displacement vectors with summations carried out over a short peptide of three contiguous residues in the protein.⁴¹ Concerted structural changes result in constructive addition of these vectors while uncorrelated differences lower the magnitude of these vectors. The magnitudes of displacement vectors versus CheY residue number upon least-squares superposition of backbone atoms of CheY molecules in the CheY-CheZ_200-214 peptide structures with those of inactive Mg²⁺-bound CheY and BeF₃⁻-activated CheY are shown in Figure 7. Using low r.m.s.d. values and small magnitudes of the displacement vectors as indications of structural similarity, both the overall r.m.s.d. and ${\overrightarrow{\delta }}_3$ analyses suggest that CheY molecules in the BeF₃⁻-free ^F432YZ_200-214 structures are similar to inactive Mg²⁺-CheY while the CheY molecule in the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 structure is similar to BeF₃⁻-activated CheY. Intriguingly, CheY molecules in the BeF₃⁻-bound ^F432YZ_200-214 structures are similar to inactive Mg²⁺-CheY, rather than to BeF₃⁻-activated CheY.

Figure 7.

Comparison of the CheY backbone conformations in the CheY-CheZ_200-214 complexes with those in inactive and active CheY. ${\overrightarrow{\delta }}_3$ plots are shown for main chain atoms of CheY in BeF₃⁻-free ^F432YZ_200-214 (cyan), BeF₃⁻-bound ^F432YZ_200-214 (deep blue) and BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 (orange) following “sieve-fit” superposition with inactive Mg²⁺-bound CheY (2CHE)¹⁴ (upper panel), resulting in overall r.m.s.d. values of 0.50 Å, 0.48 Å and 0.98 Å, respectively and with BeF₃⁻-activated CheY (1FQW)¹⁵ (lower panel), resulting in overall r.m.s.d. values of 0.80 Å, 0.77 Å and 0.35 Å, respectively. The BeF₃⁻-free ^F432YZ_200-214 and the “meta-active” conformer of the BeF₃⁻-bound ^F432YZ_200-214 structures solved from crystals grown in Tris (pH 8.4) were used for ${\overrightarrow{\delta }}_3$ and r.m.s.d. calculations, shown here. Key active-site and switch residues are highlighted on the plots by square symbols. The secondary structure of CheY is plotted for reference.

Partial occupancy of BeF₃⁻ in BeF₃⁻-bound ^F432YZ_200-214 could likely explain the absence of CheY activation in these structures, as is the case in the BeF₃⁻-bound ^F432YZ_200-214 structure solved from a crystal grown in Hepes (pH 7.5), where the BeF₃⁻ occupancy refined to only 65%. However, in the remaining two BeF₃⁻-bound ^F432YZ_200-214 structures, one solved from a crystal grown in CAPS (pH 10.5) and the other solved from a crystal grown in Tris (pH 8.4), the BeF₃⁻ species is fully occupied. To resolve this paradox, structural features at the active site and at the activation-sensitive α4-β5-α5 face of CheY were compared in the CheY-CheZ_200-214 structures and the structures of the inactive metal-free apoCheY¹³ (PDB Accession code – 1JBE), inactive Mg²⁺-bound and BeF₃⁻-activated CheY (Table 4).

Table 4.

Comparison of CheY conformational signatures and pockets on signaling surface

Models	F432YZC(−)a	F432YZC(+)b	P2(1)2(1)2YZC(+)c	Actived	Inactivee	Apof
BeF3−(+/−)	−	+	+	+	−	−
M2+/H2O at active siteg	H2O	M2+	M2+	M2+	M2+
H2O
Active-site features
Coordination tetrahedral	tetrahedral	octahedral	octahedral	octahedral	octahedral
No. of ligands 4	4	6	6	6	6
M2+/H2O contact distancesg (Å)
57Asp Oσ	2.6	2.0	2.0	2.2	2.1
3.0
59Asn CO	2.8	2.2	2.1	2.3	2.2
2.8
13Asp Oσ	2.6	2.0	2.2	2.3	2.1
2.8
H2O # (range)h	1 (>3.0)	2 (2.1 – 2.3)	2 (2.1 – 2.2)	2 (2.4)	3 (2.1 – 2.2)
SO4 (2.8)
BeF3−	n/a	2.2	2.0	2.1	n/a
n/a
BeF3− contact distances (Å)
Be – 57Asp Oσ	n/a	1.6	1.7	1.5	n/a
n/a
F− – 59Asn NH	n/a	2.9	2.9	3.0	n/a
n/a
F− – 58Trp NH	n/a	3.2	3.2	2.9	n/a
n/a
F− – 88Ala NH	n/a	4.2/ 3.2i	2.9	2.9	n/a
n/a
F− – 109Lys Nζ	n/a	2.5	2.9	2.9	n/a
n/a
12Asp Oσ – H2O (Å)	2.9	2.7	2.6	2.6	2.2
2.8
109Lys Nζ – 12Asp Oσ (Å)	4.5	3.1	2.8	2.8	6.2
4.5
Naturej	indirect	direct	direct	direct	none
indirect
109Lys Nζ – 57Asp Oσ (Å)	2.8	3.3	3.3	3.3	5.0
2.7
57Asp χ1 (°)	−157	−177	−175	178	165
−151
109Lys χ4 (°)	−177	−176	173	172	37
−165
87Thr position
57Asp Oσ – 87Thr Oγ (Å)	6.9	6.9/ 4.2i	4.2	4.1	6.5	6.0
F− – 87Thr Oγ	n/a	5.2/ 2.6i	2.6	2.5	n/a	n/a
106Tyr features
106Tyr OH	87Thr Oγ/	87Thr Oγ/	89Glu NH	89Glu NH	n/a	87Thr
Oγ/
	94Asn Nσ	94Asn Nσ				94Asn
Nσ
Naturej	indirect	indirect	direct	direct	n/a
indirect
106Tyr conformation	buried	buried	buried	buried	exposed
dual
ϕ (°)	−147	−152	−155	−145	−136
−145
Ψ (°)	−52	−54	−60	−57	−31
−52
χ1 (°)	−178	−178	−155	−164	31
179/ 72k
χ2 (°)	−94	99	105	110	35
−71
Volume of α4-β4-α5 surface pockets (Å3)
α4-β5	−	−	12.0	15.7	285.0	−/
170.0k
β5-α5	55.5	61.5	119.1	96.0	−
35.0/ 40.0k

^aValues are given for the BeF₃⁻-free ^F432YZ_200-214 structure solved from a crystal grown in Tris (pH 8.4).

^bValues are given for the BeF₃⁻-bound ^F432YZ_200-214 structure solved from a crystal grown in Tris (pH 8.4).

^cValues are given for the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 structure.

^dValues are given for BeF₃⁻-activated CheY¹⁵.

^eValues are given for inactive Mg²⁺-bound CheY¹⁴

^fValues are given for apoCheY¹³.

^gM²⁺ indicates a divalent cation

^hNumber of water molecules at the active site are indicated and the range of contact distances are given in parentheses.

ⁱValues correspond to “A” conformer/ “B” conformer of BeF₃⁻-bound ^F432YZ_200-214 (pH 8.4).

^jIndirect contacts are water or solvent-mediated hydrogen bonds; direct contacts are direct hydrogen bonds.

Comparison of active sites

Comparison of the active-site coordination geometry revealed that in both the BeF₃⁻-bound ^F432YZ_200-214 and the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 structures, the active site of CheY is occupied by an octahedrally coordinated Mg²⁺ with coordination geometry similar to that in BeF₃⁻-activated CheY (Figure 6(c) and Table 4). In the BeF₃⁻-free ^F432YZ_200-214 structure, the CheY active site is occupied by a tetrahedrally coordinated water molecule in place of the metal ion, similar to the active site of CheY in the inactive metal-free apoCheY structure (Table 4).

Examination of the contact geometries of the conserved residue ¹⁰⁹Lys showed that in the BeF₃⁻-bound ^F432YZ_200-214 structures, in which the BeF₃⁻ ligand is fully occupied, the side chain of the conserved residue ¹⁰⁹Lys on the β5-α5 loop of CheY extends towards the active site and makes ionic interactions with BeF₃⁻ and ¹²Asp. This is similar to the conformation of ¹⁰⁹Lys in BeF₃⁻-activated CheY as well as in BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 (Figure 6(c)). In the BeF₃⁻-free ^F432YZ_200-214 structure, while the ¹⁰⁹Lys C_α position is similar to that in BeF₃⁻-bound ^F432YZ_200-214 (Figure 7) and the ¹⁰⁹Lys side chain also extends into the active site, ¹⁰⁹Lys is involved in a salt bridge interaction with the active-site ⁵⁷Asp residue, similar to that in inactive metal-free apoCheY (Table 4).

Comparison of ⁸⁷Thr and β4-α4 loops

Interpretation of differences in the conserved residue ⁸⁷Thr and the β4-α4 loop is complex. In BeF₃⁻-free ^F432YZ_200-214, the position of ⁸⁷Thr is similar to that in the inactive Mg²⁺-bound structure, but the β4-α4 loop is slightly displaced from the inactive position (Figure 6(a), Figure 6(b) and Table 4) while in the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 structure, ⁸⁷Thr and the β4-α4 loop are in the active conformation, as in BeF₃⁻-activated CheY (Figure 6(c), Figure 6(d) and Table 4). The differences in the corresponding ${\overrightarrow{\delta }}_3$ plots are consistent with these observations (Figure 7).

In the three different BeF₃⁻-bound ^F432YZ_200-214 structures, ⁸⁷Thr and the β4-α4 loop are found in different positions. In the structure solved from a crystal grown in Hepes (pH 7.5), in which BeF₃⁻ occupancy is partial, ⁸⁷Thr is in the inactive position, as in inactive Mg²⁺-CheY, while the β4-α4 loop is slightly displaced, as in BeF₃⁻-free ^F432YZ_200-214. In the BeF₃⁻-bound ^F432YZ_200-214 structure solved from a crystal grown in CAPS (pH 10.5), ⁸⁷Thr is displaced towards the active site by only 0.3 Å and the β4-α4 loop is positioned intermediate to that in Mg²⁺-bound inactive and BeF₃⁻-activated CheY.

In the BeF₃⁻-bound ^F432YZ_200-214 structure solved from a crystal grown in Tris (pH 8.4), ⁸⁷Thr and the β4-α4 loop show two distinct positions, each with 50% occupancy. In one conformation, ⁸⁷Thr is similar to that in inactive Mg²⁺-CheY and the β4-α4 loop is slightly displaced, as in the BeF₃⁻-free ^F432YZ_200-214 structures and the BeF₃⁻-bound ^F432YZ_200-214 structure solved from a crystal grown in Hepes (pH 7.5). In the alternate conformation, the position of ⁸⁷Thr is identical to that in BeF₃⁻-activated CheY but the β4-α4 loop is placed intermediate to that in inactive Mg²⁺-bound CheY and BeF₃⁻-activated CheY (Figures 6(a), Figure 6(b) and Table 4).

Differences in the positions of ⁸⁷Thr and the β4-α4 loop observed in the BeF₃⁻-bound ^F432YZ_200-214 structures are not likely to be related to differences in buffer and pH conditions during crystallization but rather to non-specific crystal to crystal variation in characteristics such as size and morphology that might influence diffusion rates of the BeF₃⁻ species, resulting in varying degrees of propagative change in ⁸⁷Thr and the β4-α4 loop.

Comparison of ¹⁰⁶Tyr

¹⁰⁶Tyr, the key aromatic switch residue in CheY is located on the β5 strand that is part of the signaling surface.¹⁶ The rotameric conformation of ¹⁰⁶Tyr is highly correlated with the signaling state of the protein. The two rotamers have large differences in solvent accessibility of the bulky aromatic phenol group with the solvent-exposed position of ¹⁰⁶Tyr corresponding to the inactive signaling surface and the solvent-buried conformer to the active α4-β5-α5 face. In all seven structures of the CheY-CheZ_200-214 complex from the two crystal forms, ¹⁰⁶Tyr is in the solvent-buried or active conformation irrespective of the presence of BeF₃⁻ and Mg²⁺ at the active site (Figure 6(b) and Figure 6(d) and Table 4). The solvent-exposed conformation of ¹⁰⁶Tyr is not competent to bind to the CheZ_C helix in either of the two orientations observed because of steric clashes. This has also been noted in the FliM_N-bound CheY structures.^20,21

Y-T coupling

Previous structural studies of CheY activation established that the solvent-buried conformation of ¹⁰⁶Tyr and the movement of ⁸⁷Thr towards the active site and the consequent repositioning of the β4-α4 loop are obligatorily coupled, a phenomenon termed “Y-T coupling”.¹⁶ A molecular dynamics study of the FixJ receiver domain further suggested that during the mechanism of activation, the movement of the β4-α4 loop occurs prior to rotation of the aromatic switch residue (¹⁰¹Phe in FixJ), and the rearranged β4-α4 loop is stabilized by a hydrogen bond between the conserved Thr (⁸²Thr in FixJ) and the phosphoryl oxygen at the active site.¹⁹ Based on previous studies supporting “Y-T coupling”, the solvent-buried “active” conformation of ¹⁰⁶Tyr is not compatible with the inactive conformation of ⁸⁷Thr and the β4-α4 loop because of steric hindrance imposed by the short distance (1.5 Å) between the hydroxyl oxygen of the solvent-buried ¹⁰⁶Tyr side chain and the C atom of the ⁸⁷Thr side chain.¹⁶

A direct consequence of “Y-T coupling” during CheY activation is the perturbation of the backbone of the β4-α4 loop. This is the case in the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 structure as well as the previously determined BeF₃⁻-CheY structure.¹⁵ In this conformation, the solvent-buried rotamer of ¹⁰⁶Tyr is stabilized by a hydrogen bond between its side chain hydroxyl and the backbone amide NH of the repositioned Glu⁸⁹ of the β4-α4 loop (Table 4).

However, in all six ^F432YZ_200-214 structures, the solvent-buried conformation of ¹⁰⁶Tyr is not associated with the complete repositioning of the β4-α4 loop and ⁸⁷Thr, as detailed above. In this intermediate state, the side chain hydroxyl of ¹⁰⁶Tyr is stabilized by a water-mediated contact with the side chain hydroxyl of ⁸⁷Thr and the side chain amide NH of ⁹⁴Asn (Table 4). Two additional structures of CheY have been previously reported in which ¹⁰⁶Tyr is buried from solvent but ⁸⁷Thr and the β4-α4 loop are not repositioned.^13,28 In all these cases as well as in the ^F432YZ_200-214 structures, reported here, the solvent-buried conformation of ¹⁰⁶Tyr does not impose steric clashes with ⁸⁷Thr because of moderate shifts in C_α positions (~0.3Å), ~10-20° rotation of the ⁸⁷Thr C-C_α bond as well as minor changes in ¹⁰⁶Tyr torsion angles (Table 4). These subtle changes position the ¹⁰⁶Tyr side chain OH group ~3.0 Å away from the side chain C_γ of ⁸⁷Thr, allowing such a conformation to exist. These observations provide evidence for plasticity and are contrary to a two-state hypothesis in which long-range conformational changes are obligatorily coupled.

Summary of CheY conformations

The comparisons described above allow categorizations of the CheY conformations in the different CheY-CheZ_200-214 structures. CheY in BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 has all the characteristic signatures of a fully activated receiver domain. CheY in the ^F432YZ_200-214 structures is not as easily categorized. In the BeF₃⁻-free ^F432YZ_200-214 structures, ⁸⁷Thr is in an inactive position, ¹⁰⁶Tyr is in an active conformation, albeit with small differences in χ1 from that of active CheY, while the β4-α4 and the β5-α5 loops are slightly different from those in inactive CheY. The ${\overrightarrow{\delta }}_3$ analysis shows that the difference between the CheY conformations in the peptide-bound and peptide-free structures in the absence of BeF₃⁻ activation, i.e., between ^F432YZ_200-214 and Mg²⁺-CheY, is greater than the difference in CheY conformations in the peptide-bound and peptide-free structures in the presence of BeF₃⁻ activation, i.e., between ^P2(1)2(1)2YZ_200-214 and BeF₃⁻-CheY (Figure 7), suggesting that CheZ_C peptide binding to inactive CheY is associated with a greater change in CheY conformation than that associated with CheZ_C binding to fully activated CheY. The greater similarity of conformation signatures of CheY in the ^F432YZ_200-214 complex to those in the active state relative to those in the inactive state provides the basis for designating this CheY conformation as the “meta-active” state. Although not identical, this conformation of CheY closely resembles the “meta-active” state, first introduced by Simonovic and Volz¹³ (Figure 8).

Figure 8.

Comparison of the CheY conformation in ^F432YZ_200-214 with “meta-active” apoCheY. The activation-sensitive α4β4α5 region of CheY molecules in BeF₃⁻-free ^F432YZ_200-214 (key features shown in cyan) and BeF₃⁻-bound ^F432YZ_200-214 (key features shown in deep blue) are superposed on “meta-active” apoCheY (1JBE)¹³ (key features shown in magenta) with side chains of key switch residues depicted in ball and stick models.

In the BeF₃⁻-bound ^F432YZ_200-214 structures, ⁸⁷Thr and the β4-α4 loop are in multiple positions while ¹⁰⁶Tyr is similar to the active conformation, albeit with small differences in χ1 (Table 4). Conformational analysis revealed that CheY molecules in these structures have the signatures of a partially activated structure. The multiple conformations of CheY in the BeF₃⁻-bound ^F432YZ_200-214 structures likely represent intermediates in the pathway from a “meta-active” to a fully active conformation. The lack of complete activation in these structures might result from restrictions imposed by lattice contacts although the regions of the CheY protein that define its conformational state are not directly associated with symmetry interfaces. Modeling the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 structure in the F432 cell showed that the ^P2(1)2(1)2mode cannot be accommodated in the F432 cell because of steric clashes with a symmetry-related CheY molecule (Figure 9). The idea that lattice contacts constrain the conformation of BeF₃⁻-bound CheY is further supported by relatively higher disorder in the CheZ_200-214 peptide (Figure 3(c)) as well as a higher degree of variation in the “meta-active” conformation of CheY in the BeF₃⁻-bound ^F432YZ_200-214 structures compared to those in the BeF₃⁻-free ^F432YZ_200-214 structures.

Figure 9.

Incompatibility of the ^P2(1)2(1)2mode of CheZ_200-214 peptide binding in the F432 lattice. CheY molecules in ^P2(1)2(1)2YZ_200-214, (CheY shown in wheat and CheZ_200-214 shown in orange) are superimposed on the symmetry-related CheY molecules in ^F432YZ_200-214 (CheY shown in gray and CheZ_200-214 shown in cyan) in the F432 lattice. The BeF₃⁻-free ^F432YZ_200-214 structure solved from a crystal grown in Tris (pH 8.4) is used as a representative of all ^F432YZ_200-214 structures in this figure.

Binding model

Two inferences can be drawn from analyses of the modes of CheZ_200-214 peptide binding and the conformational states of CheY in the structures of the CheY-CheZ_200-214 peptide complexes: 1) in the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 complex, in which CheY is in the fully activated conformation, the CheZ_200-214 peptide binds in the ^P2(1)2(1)2mode, and 2) in the ^F432YZ_200-214 complexes, in which CheY exists in the “meta-active” state, the CheZ_200-214 peptide binds in the ^F432mode. Based on the correlation between dual CheZ_C binding modes and the CheY conformational state, we propose that the CheZ_200-214 peptide binds to CheY in two different modes that are reflective of the activation state of the CheY protein. The 20-fold lower binding affinity of the CheZ_C peptide to inactive versus active CheY is likely to be reflective of the different modes of binding.²⁶ Whether a similar difference in K_d’s for binding of the FliM_N peptide to active and inactive CheY might correspond to dual modes of binding is unknown as FliM_N-bound inactive CheY structures have not been reported.

Superpositions of CheY molecules bound to the CheZ_C peptide in the two distinct orientations did not reveal any steric clashes that might explain why one conformation binds in one mode and not the other. Two surface pockets on the α4-β5-α5 signaling face of CheY, one between α4 and β5 (α4-β5) and the other between β5 and α5 (β5-α5), were compared (see Materials and Methods). The volume (Table 4), hydrophobicity and electrostatic characteristics (data not shown) of the two pockets differ in different conformations of CheY.

In the fully activated protein, the α4-β5 pocket is smaller in volume and more hydrophobic than the β5-α5 pocket, which has a higher concentration of positively-charged residues, as in the BeF₃⁻-activated CheY and the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 structures. In the inactive protein, the β5-α5 pocket is absent while the α4-β5 pocket is ~20-fold greater in volume, because of the solvent-exposed conformation of ¹⁰⁶Tyr coupled with the positions of the β4-α4 and β5-α5 loops, as in the CheY-Mg²⁺ structure. However, when ¹⁰⁶Tyr is buried with only small changes in the β4-α4 and β5-α5 loops, as in the “meta-active” state, the α4-β5 pocket disappears while part of the β5-α5 pocket exhibits a hydrophobic characteristic because of slightly altered torsions of the solvent-buried ¹⁰⁶Tyr residue, as in “meta-active” apoCheY and CheY in the ^F432YZ_200-214 structures. It is possible that these conformation-dependent differences in surface profiles of the α4-β5-α5 face favor binding of the CheZ_C helix in different orientations.

Implications of CheZ binding to inactive CheY

CheZ binding to inactive CheY offers interesting scenarios for discussion although its physiological significance is yet to be established. Binding to inactive CheY might allow CheZ to present inactive CheY to CheA for phosphotransfer. Although the binding affinity of inactive CheY to CheA-P2 is two orders of magnitude higher than that for CheZ^26,42, the relative intracellular concentrations of chemotaxis proteins might allow for such a situation. A comprehensive quantitative determination of chemotaxis components in bacterial cells, reported by Li and Hazelbauer⁴³, suggests that 70% of unphosphorylated CheY molecules could exist in complex with P2 domains of CheA. It follows that the remaining 30% of unphosphorylated CheY might exist as soluble cytoplasmic protein, potentially available for binding to CheZ. Based on the calculations of Li and Hazelbauer, such a scenario would be possible only within the 44% of receptor signaling complexes that contain both CheA_L, proficient in phosphorylating CheY, and CheA_S, a shorter form of CheA that cannot transfer phosphoryl groups to CheY but localizes CheZ to the receptor complex.^44,45

Localization of the source (kinase CheA) and the sink (phosphatase CheZ) of phosphoryl groups for CheY to the same site in the cell (polar chemoreceptor patches) creates a uniform distribution of P~CheY throughout the cytoplasm, ensuring rapid responses to stimuli.^46,47 It can be speculated that binding of unphosphorylated CheY to CheZ at the core signaling complex would result in an increase in the rate or number of futile cycles of phosphorylation and dephosphorylation of CheY, competing with the productive transfer of phosphoryl groups from CheA to CheY and CheB.

In a number of different response regulators, it has been reported that mutations or ligands that stabilize the active conformation increase rates of phosphorylation, suggesting that phosphorylation occurs preferentially in a sub-population that exists in an active state.^{13,18,30,48-50} A “meta-active” conformation of CheY in the BeF₃⁻-free ^F432YZ_200-214 structures suggests that the CheZ_200-214 peptide binds preferentially to the “meta-active” sub-population of inactive CheY. This provides the evidence for the functional nature of the proposed “meta-active” state of the CheY protein. This might provide the structural explanation for the coupling of CheY phosphorylation and peptide binding⁸ and suggest a functional role for the “meta-active” state of CheY. It was shown that the CheZ_C and FliM_N peptides cause a 10-fold increase and CheA-P2 causes a 6-fold decrease in the rate of phosphoryl transfer to CheY from small molecule phosphodonors.⁸ The observation that CheY exists in a “meta-active” state in the ^F432YZ_200-214 complex, provides evidence that the increased rate phosphorylation upon peptide binding results from a shift in equilibrium towards a more active conformation.

It was previously shown that ⁸⁷Thr, ¹⁰⁹Lys and ¹⁰⁶Tyr, key conserved residues involved in the conformational switch are not important in the coupling of peptide binding to increased rates of phosphorylation.⁸ Two additional residues near the active site, ¹⁴Phe and ⁵⁹Asn, display completely different rotamer conformations in the CheZ_200-214 peptide-bound structures as compared to inactive and active CheY structures not bound to peptide (Figure 10(a) and Figure 10(b)). ⁵⁹Asn is highly conserved in all species within the proteobacter group and ¹⁴Phe is highly conserved in γ-proteobacteriaceae, Enterobacteriaceae and β-proteobacteriaceae families (data not shown). ¹⁴Phe has been previously implicated in regulating the solvent accessibility of the active-site pocket in structures of CheY bound to CheA-P2.^22,24 It has been reasoned that when CheY is bound to P2, the phenyl group of ¹⁴Phe is rotated away from the active site, increasing accessibility of the phospho-histidine. A similar configuration of ¹⁴Phe is observed in the CheZ_C peptide-bound structure but not in the FliM_N-bound structure. The change in rotamer conformation of ⁵⁹Asn is also observed in the FliM_N-bound^20,21 as well as CheA-P2-bound^22-24 structures of CheY proteins.

Figure 10.

Role of key residues in peptide-induced acceleration of CheY phoshorylation. (a) Active site features of CheY in BeF₃⁻-free ^F432YZ_200-214 superimposed on inactive Mg²⁺-bound CheY. (b) Active site features of CheY in BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 superimposed on BeF₃⁻-activated CheY. Key features in BeF₃⁻-free ^F432YZ_200-214 are shown in cyan, in BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 in orange, in inactive Mg²⁺-bound CheY in red and in BeF₃⁻-activated CheY in green. The Mg²⁺ ion (magenta), water molecules (deep red) and the BeF₃⁻species (yellow) at the active site are shown. The BeF₃⁻-free ^F432YZ_200-214 structure solved from a crystal grown in Tris (pH 8.4) is used as a representative of all ^F432YZ_200-214 structures in (a). (c) Role of ¹⁴Phe and ⁵⁹Asn. Rates of phosphotransfer to WT CheY, CheY¹⁴F→A and CheY⁵⁹N→A proteins from ammonium phosphoramidate in the presence (+) and absence (−) of the CheZ_200-214 peptide are shown as bars (see Materials and Methods). Each rate corresponds to an average from three independent experiments with standard errors indicated.

In order to assess the significance of the different rotameric conformations of ¹⁴Phe and ⁵⁹Asn in the acceleration of CheY phosphorylation upon CheZ_C binding, phosphorylation kinetics of CheY proteins containing alanine substitutions at ¹⁴Phe and ⁵⁹Asn (CheY¹⁴F→A and CheY⁵⁹N→A) were measured in the presence and absence of the CheZ_200-214 peptide by following changes in ⁵⁸Trp fluorescence (see Materials and Methods). Extending previous observations, we found that changes in charge, bulk and rotameric conformations of key residues have little effect on the enhancement of phosphorylation upon peptide binding (Figure 10(c)). Consistent with earlier results, the presence of the CheZ_200-214 peptide increased the phosphorylation rate of the WT protein by 10-fold.⁸ While both mutant proteins alone have phosphorylation rates 1.5- to 2-fold less than the WT protein, both mutant proteins exhibit 10-fold greater rates in the presence of CheZ_200-214 peptide, exactly similar to the WT protein.

Implications of dual binding modes

Structures of the CheY-CheZ_200-214 complexes indicate that CheY is a flexible molecule capable of accommodating dual modes of binding to the CheZ_200-214 peptide. Previous evidence of such flexibility in the α4-β5-α5 face of CheY was found in crystal structures displaying two distinct modes of binding of this region to the P2 domain of the histidine kinase CheA.²² The physiological significance of each of the CheZ peptide binding modes remains to be determined. However, a distinct mode of binding of inactive CheY to CheZ offers interesting implications.

The consequence of CheY activation is the elevated binding affinity for CheZ and FliM and reduced binding affinity for CheA.^26,42 Peptide-induced enhancement of CheY phosphorylation suggests that relay of information in the reverse direction is also possible.⁸ Reverse flow of signaling information has also been reported between effector and receiver domains of OmpR upon DNA binding.⁵¹ The different binding modes observed in the CheY-CheZ_200-214 structures may provide different mechanisms of information flow in the forward and reverse directions, allowing bacteria to fine-tune signaling along the two-way pathway of signal transduction to CheY.

Although the CheZ protein is not a canonical phosphatase, analogies can be made with 4-amino-5-hydroxymethyl-2-methylpyrimidine (HMP) phosphate kinase, which catalyzes two independent ATP-dependent phosphotransfer reactions of the substrate, HMP, in which the product of the first reaction, HMP phosphate acts as the substrate for the second reaction.⁵² The HMP phosphate kinase is able to catalyze the two reactions by allowing two different binding modes for HMP phosphate, corresponding to its role as product or substrate. This mechanism is in some ways analogous to the two binding modes of the CheZ_C helix observed in the CheY-CheZ_200-214 structures where one mode corresponds to binding of the primary substrate of the CheZ protein, the activated form of CheY, and the other mode corresponds to binding of the secondary substrate for CheZ phosphatase for reverse flow of signaling information, the inactive or “meta-active” forms of CheY.

Conclusions

The high-resolution models of the CheZ_200-214 peptide bound to inactive and active CheY provide evidence for dual binding modes subject to the conformational state of the CheY protein. Our studies suggest that the CheZ_200-214 peptide binds to the “meta-active” sub-population of inactive CheY molecules. This provides a potential structural explanation for the reverse information flow from peptide binding to increased phosphotransfer at the active site. Our studies also provide evidence against the proposed obligatory “Y-T coupling”, and indicate that repositioning of ⁸⁷Thr and the β4-α4 loop and the rotameric conversion of ¹⁰⁶Tyr need not be obligatorily coupled. The dual modes of binding displayed by our CheY-CheZ_200-214 complexes extend previous conclusions about the inherent malleability of CheY. Such plasticity allows for different binding modes of CheY partners, which may be relevant to the fine-tuning of signaling responses.

Materials and Methods

Materials

The CheZ_200-214 peptide, corresponding to the C-terminal 15 residues of S. enterica serovar Typhimurium CheZ, bearing the sequence ASQDQVDDLLDSLGF with the N-terminus protected by acetylation, was obtained from the peptide synthesis core facility of Massachusetts General Hospital (Boston, MA) as lyophilized powder. Oligonucleotides for site-directed mutagenesis, bearing sequences 5′-TTT TTG GTT GTG GAT GAC GCT TCG ACC ATG CGT CGT ATC-3′ and 5′-GAT ACG ACG CAT GGT CGA AGC GTC ATC CAC AAC CAA AAA-3′ for CheY¹⁴Phe→Ala mutagenesis and 5′-ATT ATC TCC GAC TGG GCC ATG CCG AAC CTG-3′ and 5′-CAT GTT CGG CAT GGC CCA GTC GGA GAT AAT-3′ for CheY⁵⁹Asn→Ala mutagenesis, were obtained from the DNA core facility of the University of Medicine and Dentistry of New Jersey (Piscataway, NJ). The ammonium salt of phosphoramidate was synthesized by the method of Sheridan et al.⁵³

Proteins

The S. enterica serovar Typhimurium WT cheY gene was previously cloned within a 0.8 kilobase SnaI-SmaI fragment in a pUC12-based CheY expression vector (pME124).^54,55 Plasmids, pEF41 and pJG1, carrying the ¹⁴Phe→Ala and ⁵⁹Asn→Ala mutations, respectively, were constructed using the plasmid pME124 as template and respective oligonucleotides using the site-directed mutagenesis kit (Stratagene).

Protein expression and purification

The pME124 plasmids bearing WT and mutant cheY genes were transformed into HB101 cells.^14,56 The WT and mutant CheY proteins were purified by a modification of previously described procedures.⁵⁷ The previously used ion-exchange and gel filtration columns were substituted with a HiTrap Q Fast Flow column (GE Healthcare) and a Superdex 26/60 column (GE Healthcare) in fast performance liquid chromatography using an AKTA system (GE Healthcare). The purified protein was quantitated by measuring UV absorbance at 280 nm, using extinction coefficients calculated from amino acid composition.⁵⁸ The ε₂₈₀ value in ml mg⁻¹ cm⁻¹ 280 calculated for WT CheY is 0.493, for CheY¹⁴F→A is 0.496 and for CheY⁵⁹N→A is 0.495.

Crystallization

Purified WT CheY protein was dialyzed into a buffer containing 50 mM Hepes and 7 mM MgCl₂ (pH 7.0) (buffer A) and concentrated to 25 mg/ml (1.8 mM) using a Centriprep-10 (Millipore). A 20 mM stock solution of CheZ_200-214 peptide was prepared in buffer A. Both purified protein and dissolved peptide were filtered through 0.22 μm-poresize cellulose acetate filters (Corning Costar).

For crystallization of the BeF₃⁻-free ^F432YZ_200-214 complex, WT CheY in buffer A was mixed with the CheZ_200-214 peptide to a final concentration of 0.9 mM CheY and 3.6 mM CheZ_200-214. Crystals were grown by the hanging drop vapor diffusion method at room temperature by mixing equal volumes of the protein/peptide mixture and reservoir solutions. The BeF₃⁻-free ^F432YZ_200-214 crystals grew to ~100 μm × 100 μm × 50 μm in ~3 days in the presence of reservoir solutions containing 1.8 M to 2.3 M ammonium sulfate and 0.2 M lithium sulfate (precipitant A) in three different buffer and pH conditions: 0.1 M Hepes (pH 7.5) (buffer B), 0.1 M Tris (pH 8.4) (buffer C), and 0.1 M CAPS (pH 10.5) (buffer D). A few BeF₃⁻-free ^F432YZ_200-214 crystals from each buffer and pH condition were incubated in gradually increasing concentrations of BeCl₂ and NaF in precipitant A to a final concentration of 5 mM BeCl₂ and 30 mM NaF, at least 4-5 hours prior to data collection to obtain BeF₃⁻-bound ^F432YZ_200-214 crystals. Cryoprotection was achieved by serially transferring the ^F432YZ_200-214 crystals into their respective reservoir solutions, supplemented with glycerol at a final concentration of 20% (v/v) at 5% steps, soaking for 1 min at every step. Crystals were placed in a 100 K nitrogen cryostream for data collection.

For crystallization of the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 complex, WT CheY in buffer C was activated with 5 mM BeCl₂ and 30 mM NaF, prior to mixing with the CheZ_200-214 peptide, dissolved in buffer C to a final concentration of 0.8 mM CheY and 1.2 mM CheZ_200-214. These crystals were also grown by the hanging drop vapor diffusion method at room temperature by mixing equal volumes of the protein/peptide mixture and the reservoir solution, containing 0.05 M sodium phosphate (monobasic) and 37.5% PEG-8000 (precipitant B) in 0.1 M MES (pH 6.0) (buffer E). These crystals at first grew as fuzzy needles in the absence of BeCl₂ and NaF at the high throughput crystallization lab at the Hauptman Woodward Medical Research Institute, Buffalo, NY and were subsequently optimized to grow in the presence of BeCl₂ and NaF to ~100 μm × 10 μm × 10 μm by repeated microseeding procedures. For crypotection, the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 crystals were serially transferred into reservoir solution, containing PEG-400 and sucrose to a final concentration of 18% (w/v) PEG-400 at 4.5% steps and 2% (w/v) sucrose at 0.5% steps, soaked for 1 min at each step and frozen in a 100 K nitrogen cryostream for data collection.

Data collection and processing

Data were collected at beamline X4A at the National Synchrotron Light Source at Brookhaven National Laboratory, Upton, NY. For the ^F432YZ_200-214 crystals, 100° of data with 0.5° oscillations and for the ^P2(1)2(1)2YZ_200-214 crystals, 200° of data with 1° oscillations were collected, processed with DENZO and SCALEPACK.⁵⁹ Data collection and processing statistics are listed in Table 1.

Structure determination and refinement

All seven structures were solved using the molecular replacement program, Phaser.¹⁷ Following rigid-body refinement using Refmac 5.1.24⁶⁰, as part of the CCP4 suite of programs⁶¹, interative cycles of maximum likelihood and isotropic temperature factor and model building using O⁶² were performed, in which the data were extended gradually from 2.4 Å to the highest resolution shell until convergence. Water molecules were initially modeled using the ARP-WARP routine⁶³ and subsequently only waters and other small molecules with Fourier difference peaks greater than 3σ were included in the final models. Refinement statistics of the seven final models are listed in Table 1.

Quality of structural models

In Ramachandran plots, generated by the PROCHECK program⁶⁴, 91.3% to 94.4% of the residues occupy the most favored regions. Similar to all previously solved CheY structures, only ⁶²Asn is disallowed because of its location within a γ-turn.⁶⁵ The average B-factors for all atoms of the structures range between 10.9 Ų and 27.4 Ų.

Structural analyses

Superpositions of atomic models were accomplished by a previously described least-squares based iterative superposition procedure⁴¹ that is similar to the “sievefit” program⁶⁶ to define a “static core” of main chain atoms for CheY molecules, achieved by omitting residues with atomic displacements greater than twice the r.m.s.d. value. “Sievefit” superpositions of CheY coordinates in the ^F432YZ_200-214 structures with those in apoCheY¹³, Mg²⁺-bound CheY¹⁴ and BeF₃⁻-activated CheY¹⁵ structures converged within 9 to 12, 3 to 6 and 20 to 35 cycles, respectively, yielding r.m.s. residuals of 0.33 to 0.36, 0.31 to 0.37 and 0.37 to 0.39 Å, respectively, defining a static core of 392 to 412, 456 to 476 and 364 to 400 main chain atoms, respectively. “Sievefit” superpositions of CheY coordinates in the ^P2(1)2(1)2YZ_200-214 structures with those in Mg²⁺-bound CheY and BeF₃⁻-activated CheY structures converged within 25 and 6 cycles, yielding r.m.s. residuals of 0.29 and 0.26 Å, and defining a static core of 364 and 464 main chain atoms, respectively. Overall r.m.s.d. values were obtained using the r.m.s.d. calculation algorithm within CNS 1.1.⁶⁷

To estimate the extent of difference in tertiary structure in the superimposed CheY molecules, the magnitude of the main chain displacement vector, ${\overrightarrow{\delta }}_3$ was plotted versus residue number as in Kavanaugh et al.⁴¹ $\mid {\overrightarrow{\delta }}_n\mid$ is defined as,

$$\mid {\overrightarrow{\delta }}_n\mid ={\delta }_n=\frac{1}{4n}\mid \sum _{j=1}^{4n}\Delta {\overrightarrow{r}}_j\mid$$

where summation is over n contiguous residues and $\Delta {\overrightarrow{r}}_j$ denotes the atomic displacement vector corresponding to the j^th main chain atom of the n-residue peptide. Summations were carried out over 3 contiguous residues.

CNS 1.1⁶⁷, iMOLTALK ver. 2.0⁶⁸ and the protein-protein interaction server (http://www.biochem.ucl.ac.uk/bsm/PP/server/) were used for the peptide-CheY contact analyses and the WHATIF⁶⁹ web interface (http://swift.cmbi.kun.nl/WIWWWI/) was used for the lattice contact analyses. Buried surface areas were calculated using CNS 1.1⁶⁷ and cavity/pocket searches and analyses were carried out using the CASTp v 1.1⁷⁰ server. Graphical representations were generated using Microsoft Excel 2004 version 11.1.1 (Microsoft Corporation). All structural figures were generated using Pymol.⁷¹

CheY phosphorylation kinetics

To investigate the role of the residues, ¹⁴Phe and ⁵⁹Asn on peptide-induced acceleration of phosphorylation, time courses of pre steady-state phosphotransfer from the small molecule phosphodonor ammonium phosphoramidate to WT CheY versus mutant CheY proteins (CheY¹⁴F→A and CheY⁵⁹N→A) were recorded by following the quenching of intrinsic fluorescence of ⁵⁸Trp as a probe of phosphorylation. Fluoresence measurements were carried out using an SLM-Aminco Series 2 luminescence spectrometer with a stopped-flow attachment, with an excitation wavelength of 295 nm and emission wavelength of 345 nm and 4 nm band passes for both. The dead time on the stopped-flow attachment was 4 msec. The pre steady-state CheY phosphorylation data were collected and analyzed as previously detailed⁸ using SIGMAPLOT 8.0 (Systat Software, Inc.).

Protein data bank accession numbers

The coordinates and structure factors have been deposited in the RCSB Protein Data Bank. The PDB ID codes for the BeF₃⁻-free ^F432YZ_200-214 structures solved from crystals grown in CAPS (pH 10.5), Tris (pH 8.4) and Hepes (pH 7.5) are 2FLK, 2FMI and 2FMF, respectively. The corresponding PDB ID codes for the BeF₃⁻-bound ^F432YZ_200-214 structures solved from crystals grown in CAPS (pH 10.5), Tris (pH 8.4) and Hepes (pH 7.5) are 2FKA, 2FMH and 2FLW, respectively. The PDB ID code for the BeF₃⁻-bound ^P2(1)2(1)2YZ_200-214 structure is 2FMK.

Abbreviations used

P~CheY

phosphorylated CheY

CheZ_C

C-terminal region of CheZ

FL

full-length

WT

wild-type

^F432YZ_200-214

the CheY-CheZ_200-214 complex in the F432 crystal form

^P2(1)2(1)2YZ_200-214

the CheY-CheZ_200-214 complex in the P2₁2₁2 crystal form

r.m.s.d.

root mean square deviation

^F432mode

binding mode observed in the F432 crystal form

^P2(1)2(1)2mode

binding mode observed in the P2₁2₁2 crystal form

^F432Z_200-214

the CheZ_200-214 helix in the F432 crystal form

^P2(1)2(1)2Z_200-214

the CheZ_200-214 helix in the P2₁2₁2 crystal form

HMP

4-amino-5-hydroxymethyl-2-methylpyrimidine