The Crystal Structure of Beryllofluoride Spo0F in Complex with the Phosphotransferase Spo0B Represents a Phosphotransfer Pretransition State

Abstract

A number of regulatory circuits in biological systems function through the exchange of phosphoryl groups from one protein to another. Spo0F and Spo0B are components of a phosphorelay that control sporulation in the bacterium Bacillus subtilis through the exchange of a phosphoryl group. Using beryllofluoride as a mimic for phosphorylation, we trapped the interaction of the phosphorylated Spo0F with Spo0B in the crystal lattice. The transition state of phosphoryl transfer continues to be a highly debated issue, as to whether it is associative or dissociative in nature. The geometry of Spo0F binding to Spo0B favors an associative mechanism for phosphoryl transfer. In order to visualize the autophosphorylation of the histidine kinase, KinA, and the subsequent phosphoryl transfer to Spo0F, we generated in silico models representing these reaction steps.

Bacteria use two-component systems to monitor the environment and adapt to new challenges (14, 30). When conditions for growth become unfavorable, Bacillus subtilis forms spores. The initiation of sporulation is controlled by an expanded version of the two-component system called a phosphorelay (6), consisting of four main components: a histidine kinase (KinA), a secondary messenger (Spo0F), a phosphotransferase (Spo0B), and a transcription factor (Spo0A) (Fig. 1). In addition to KinA, four other histidine kinases are known to be involved in sporulation, albeit with lower levels of signal input into the phosphorelay (15). The activities of kinases are modulated by variety of specific signals, but the nature of the signals is unknown. The activation of the signaling pathway involves four phosphotransfer reactions, and the first one is the autophosphorylation of a conserved histidine residue in the kinase, which is dependent upon the conversion of a bound ATP to ADP. The kinase then transfers the phosphoryl group to an aspartate on Spo0F, which transfers it to a histidine residue on Spo0B. Subsequently Spo0B transfers it to an aspartate on Spo0A. Therefore, the order of the phosphoryl flow is ATP→His→Asp→His→Asp (Fig. 1) (13). The initiation of sporulation is dictated by the degree of phosphorylation of the transcription factor Spo0A (37).

FIG. 1.

Phosphorelay that controls the initiation of sporulation in Bacillus subtilis. Five different histidine kinases are known to be involved in processing the sporulation signals. The flow of phosphoryl groups is His→Asp→ His→Asp.

We have been involved in an effort to explore the mechanism of phosphoryl transfer through the structural characterization of the components of the phosphorelay. We determined the crystal structures of Spo0F in the metal-bound and metal-free states (20, 21, 29). Other investigators have studied the solution structure of Spo0F by nuclear magnetic resonance techniques in the unliganded state (9) and in complex with beryllofluoride (BeF₃⁻) (10). Spo0F is a single-domain protein like CheY. It is made up of a central β-sheet surrounded by five α-helices. The site of phosphorylation, Asp54, is situated in a shallow pocket on the top of the β-sheet. Crystallographic analysis of Spo0B showed that it is homodimer with a twofold symmetry (40). The monomer consists of an N-terminal helical hairpin and a C-terminal domain. The helical hairpins associate to form a stable four-helix bundle. In Spo0B, His30 is the site of phosphorylation. The His30 residues in the two monomers are situated on opposite sides of the four-helix bundle. In order to understand how Spo0F and Spo0B interact, we analyzed the crystal structure of Spo0F in complex with Spo0B (43). This structure showed the environment for the exchange of the phosphoryl group between the two proteins; however, a crucial question remained unanswered: how does phosphorylated Spo0F (Spo0F-P) interact with Spo0B?

The obstacles to studying the association of Spo0F-P with Spo0B by crystallographic techniques are twofold. First, the phoshorylated forms of response regulators are intrinsically unstable. Although Spo0F-P is relatively stable, the half-life is only on the order of hours (44). Second, the association of Spo0F-P with Spo0B may lead to transfer of the phosphoryl group to Spo0B during the formation of the complex. It was pointed out that the binding of aluminofluoride or beryllofluoride mimics the state of phosphorylation (8). Studies by Wemmer and associates on NtrC and CheY have indicated that the interaction of beryllofluoride with response regulators activates the signaling pathway and that it produces the same structural changes in response regulators as phosphorylation (11, 17, 42). Beryllofluoride is tetrahedral and does not have a pentavalent state similar to the phosphate transition states (8), and therefore it can produce a stable “phosphorylated state” for structural studies. Here we report the crystal structure of beryllofluoride Spo0F in complex with Spo0B. When Spo0F-P interacts with the phosphotransferase, Spo0B, it undergoes a transition state before the phosphoryl group is transferred to Spo0B. The present structure represents a pretransition state interaction between Spo0F-P and Spo0B frozen in the crystal lattice. The structure of the unliganded Spo0F · Spo0B complex (43) represents the conformation of the phosphotransfer transition state. The differences between the two structures are mostly confined to one of the loop regions of Spo0F, showing that the changeover from one state to the other does not involve any change in the mode of association of the two molecules.

MATERIALS AND METHODS

Spo0B was expressed and purified following the same procedure as described earlier by Zhou et al. (45) and was concentrated to 64 mg/ml in 25 mM Tris-HCl buffer containing 1 mM dithiothreitol. The phosphatase-resistant Y13S mutant of Spo0F was expressed and purified as described earlier (22) and was concentrated to 2 mg/ml in 10 mM Bis-Tris (pH 7.3) containing 50 mM KCl. The beryllofluoride complex of Spo0F was prepared by adding BeCl₃ and NaF₃ to a solution containing Y13S mutant Spo0F protein. After incubation of the solution for 30 minutes, it was concentrated to 24 mg/ml and Spo0B protein solution was added to it. The mixture had Spo0F and Spo0B concentrations of 9 mg/ml and 10.3 mg/ml, respectively. The concentrations of BeCl₃, NaF₃, and MgCl₂ in the solution were 5.3 mM, 35 mM, and 7 mM, respectively.

Crystals were grown under condition identical to those used for the Spo0F · Spo0B complex (43) by the hanging-drop method with polyethylene glycol 2000 (PEG 2000) as a precipitant. The drop contained 1 μl of protein and 1 μl of reservoir solution, which contained 0.5 M KCl, 25% PEG 2000, and 100 mM Tris HCl at pH 8.1. The crystals grew in 4 days to a maximum size of 0.3 by 0.3 by 0.2 mm³.

Data Collection.

A crystal was dipped in a cryoprotectant containing 40% PEG 400, 5% PEG 3350, 200 mM KCl, 100 mM Tris-HCl (pH 8.1), and 2 mM MgCl₂ and was “flash frozen” by being put into a cold nitrogen stream at 90 K. The crystals are orthorhombic and of space group P2₁2₁2₁, and the diffraction data were collected at the Stanford Synchrotron Radiation Laboratory at beam line 9-1 using an ADSC Q315 charge-coupled-device detector. The crystal data and the diffraction statistics are provided in Table 1. The crystal diffracted to around 3 Å, and 360 images were collected with 0.5° rotation. The data were processed using the XDS program package (16).

TABLE 1.

Data collection and refinement statistics

Parameter (unit)	Value
Data collection
Space group	P212121
Cell dimensions (Å)
a	73.467 (71.97)a
b	118.331 (117.77)
c	168.235(170.34)
Wavelength for data collection (Å)	1.07809
Maximum resolution (Å)	3.05
No. of measured reflections	153,399
No. of unique reflections	28,073
Overall completeness (%)	98.3
Completeness in the outer shell (3.14 Å-3.05 Å) (%)	93.0
Rsym (%)	4.7
Rsym at the highest shell (%)	23.3
I/σ at the highest shell	6.1
Refinement
Resolution range (Å)	20-3.05
No. of reflections used in refinement	26,669
R factor	23.1
No. of reflections used in Rfree	1,404
Rfree (%)	28.2
RMSDb from ideal bond lengths (Å)	0.009
RMSD from ideal bond angles (°)	1.3

^aValues within parentheses are for the Spo0F · Spo0B complex (43).

^bRMSD, root mean square deviation.

Structure analysis.

The cell constants of the present crystal deviated only by 2% or less from those of the Spo0F · Spo0B complex (43), indicating that the two structures are similar, and therefore the unliganded structure was used as a starting model. There are two Spo0B dimers and four Spo0F molecules in the asymmetric unit. Treating them as six different rigid bodies, a rigid-body refinement was carried out using the CNS program package (5), which brought down the R factor from 42% to 32% for 4-Å data. Positional refinement using 3.2-Å data resulted in R and R_free values of 24.2% and 33.1%, respectively. Difference electron density clearly showed the density for beryllofluoride moieties and the cation Mg²⁺ on all four Spo0F molecules. Additionally, the electron density map clearly indicated that the conformation of the β4-α4 loop, which carries the Thr82 residue, has changed, and these regions were rebuilt in all four Spo0F molecules. For His101, no change in orientation was detected in three molecules, while in the fourth molecule His101 was found to take an alternate conformation in addition to the earlier conformation. After several cycles of simulated annealing refinements, positional refinements, and one cycle of individual B-factor refinement, the final R and R_free values for 3.05Å data are 23.1% and 28.2%, respectively. Figure 2 shows the electron density map of the active site region.

FIG. 2.

A 2Fo-Fc map of the active-site region contoured at the 1σ level. The Spo0F residues Asp10, Asp11, Asp54, and Thr82 are labeled D10, D11, D54, and T82. The site of phosphorylation His30 in Spo0B is labeled H30. BeF₃⁻ is attached to OD1 of Asp54. The Be atom and Mg atoms are shown in cyan. The fluorine atoms and oxygen atoms are shown in red and the nitrogen atoms in blue.

Protein structure accession number.

The atomic coordinates and structure factors have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org) under accession number 2FTK.

RESULTS AND DISCUSSION

Beryllofluoride binds at Asp54, the site of phosphorylation, and produces conformational changes in Spo0F. The complex formation results from the binding of two beryllofluoride Spo0F molecules to the Spo0B dimer (Fig. 3). The asymmetric unit contains two such units. The mode of association of the two proteins is the same as observed in the unliganded Spo0F · Spo0B complex (43); however, the binding appears to be stronger in the present case.

FIG. 3.

Diagram showing the association of the two Spo0F molecules with a Spo0B dimer. The two protomers of Spo0B are shaded blue and green. The two Spo0F molecules are shaded magenta. The sites of phosphorylation, Asp54 of Spo0F and His30 of Spo0B, are shown in red. The beryllofluoride binds at Asp54. The His101 residue lies outside the interacting interface, and it orients towards the solvent.

Overall view of the complex.

Figure 3 shows the symmetrical association of the two Spo0F molecules with Spo0B. We used the Y13S mutant of Spo0F to form the complex, as in the previous analysis (43), because the mutant protein was found to be more amenable to crystallization. In the phosphotransfer reactions, the mutant protein functions in the same manner as the wild-type protein, as the side chain of residue 13 does not interact with Spo0B.

The phosphotransferase, Spo0B, appears to be built specifically for interacting with Spo0F and the receiver domain of Spo0A. The conformation of Spo0B is the same as that in the noninteracting state (40). Spo0B interacts with Spo0F mainly through the four-helix bundle. The majority of the interactions are from helix α1, which carries the active histidine, while helix α2′ of the second protomer also participates in the interactions, underscoring the functional relevance of the dimerization. Spo0F makes extensive use of helix α1 and the loop regions around the active site for interacting with Spo0B, and these loops undergo certain adjustments upon binding to Spo0B (43). In the present case, there are additional changes in Spo0F due to its “phosphorylated state.” The conformational changes of Spo0F seen the present complex are discussed below.

Conformation of the active site and orientations of Thr82 and His101.

Figure 4 shows the superposition of the Spo0F in the current crystal structure with a Spo0F molecule in the unliganded Spo0F · Spo0B complex (43), the crystal structure of wild-type Spo0F (20), and the solution structure of Spo0F · BeF₃⁻ (10). The main differences between the BeF₃⁻-bound and the unliganded Spo0F · Spo0B complexes are in the β4-α4 loop of Spo0F. In the BeF₃⁻-bound complex, the Thr82 residue moves towards the active site to interact with a fluorine atom, causing a shift in the loop region. Spo0F in the unliganded complex is very similar to noninteracting Spo0F (20), with some variations in the β3-α3 loop and the β4-α4 loop. Surprisingly, the solution structure of BeF₃⁻-bound Spo0F (10) is very different from the molecular structure in the BeF₃⁻-bound Spo0F · Spo0B complex. All the helices in the solution structure are shifted by 2 to 3 Å. These differences could be a reflection of the flexibility of the molecule.

FIG. 4.

Superposition of the Spo0F molecule in the BeF₃⁻-bound complex (red) with the unliganded Spo0F · Spo0B complex (43) (blue), the solution structure of BeF₃⁻-bound Spo0F (10) (green), and the crystal structure of wild-type Spo0F (20) (black). The residues Asp10, Asp11, Asp54, Thr82, and His101 are indicated by the residue numbers. In the presence of beryllofluoride, Thr82 orients towards the active site. The His101 side chain is positioned on the surface in all the molecules. Superposition was carried out using the Cα atoms of the β-strand residues 5 to 10, 29 to 33, 50 to 54, 77 to 82, and 100 to 104.

The structural analysis of the phosphorylated response regulators and beryllofluoride complexes of response regulators shows that the major changes that occur upon phosphorylation involve residues at positions 82 and 101 (Spo0F numbering). In response regulators the residues at these two positions are conserved, with a hydroxyl-containing Ser/Thr and an aromatic Phe/Tyr, respectively (41). The former residue is located in the β4-α4 loop, and the latter residue is located on the β5 strand at the α4-β5-α5 surface. In the unphosphorylated state, the side chain of the aromatic residue is positioned on the surface and the side chain of Ser/Thr points away from the active site. Upon phosphorylation, the Thr/Ser residue moves towards the active site to interact with a phosphoryl oxygen and the aromatic residue moves in a correlated fashion to an internal orientation (2, 4, 19). The movement of the aromatic residue from the α4-β5-α5 surface makes it ready for interactions with other proteins or for dimerization. Crystal structure analysis of response regulators from the OmpR family in complex with BeF₃⁻ shows that the beryllofluoride also produces the same movements for the two residues, paving the way for the formation of homodimers (36). Transcription factors of the OmpR family bind DNA in a tandem mode. Stock and associates have hypothesized that phosphorylation-induced dimerization is a common mechanism of regulation for the OmpR/PhoB subfamily (36).

The situation in Spo0F, however, is different, because the residue at position 101 is a histidine and the α4-β5-α5 surface is not involved in dimerization or interactions with Spo0B (39, 43). In the current structure, the His101 residue does not take up an internal orientation in any of the four Spo0F molecules in the asymmetric unit. In one of the molecules, His101 is disordered, but both conformations correspond to external orientations. Therefore, it appears that His101 may not have the same role as the Phe/Tyr residues in other response regulators, and it does not participate in the coupled movement with the Thr82. This is in keeping with the fact that the α4-β5-α5 surface of Spo0F is not part of the surface for interaction with Spo0B. Thr82 is critical for its function, and mutation of this residue to alanine completely shuts down sporulation. The mutation of His101 to alanine, on the other hand, caused only milder changes in sporulation (38). The mutations of residues in the interacting interface produce severely defective phenotypes, and many of them even shut down sporulation completely.

Mechanism of phosphoryl transfer.

The transfer of the phosphoryl group can be accomplished through either an associative mechanism or a dissociative mechanism, analogous to S_N1 or S_N2 mechanisms (Fig. 5). In the former case, the phosphoryl group takes up a pentavalent intermediate state bound to both proteins, en route to binding totally to the second protein. In the latter case, the phosphoryl group dissociates from one protein to form an unstable metaphosphate intermediate before binding to the other protein. There has been considerable debate regarding the nature of the transition state of the enzyme-catalyzed phosphoryl transfer (26). Some investigators favor an associative mechanism (3, 18, 27, 28, 33, 34), while others favor a dissociative mechanism (1, 12, 23). In the midst of strong arguments in favor of both models, we consider the crystal structures of the Spo0F · Spo0B complex to be evidence for the associative mechanism of phosphoryl transfer, because the geometry of association makes the dissociative mechanism energetically unfavorable. If the phosphoryl group dissociates to form the metaphosphate, the phosphorus atom should position itself 3.3 Å from the carboxyl oxygen, the sum of van der Waals radii of the phosphorus and oxygen. In the same way, the phosphorus atom should position itself 3.3 Å from the N atom of histidine. The geometry of the active site is such that the O and N are only 4.9 Å apart (43). Hence, there is insufficient room in the active site for the formation of the detached metaphosphate without rearrangement of the residues in the active site. The corresponding distance between N and O in the BeF₃⁻-bound Spo0F · Spo0B complex, 5.27 Å, is greater yet is inadequate to accommodate a detached metaphosphate.

FIG. 5.

Two possible modes for phosphoryl transfer from an aspartic acid residue to a histidine residue. (a) Associative mechanism. (b) Dissociative mechanism.

Environment of the active site.

Figure 6a shows a stereo diagram depicting the environment of His30 and Asp54. All three fluorine atoms are involved in favorable interactions similar to those of phosphoryl oxygens in Spo0A-P (19) and FixJ-P (4). The fluorine atom F1 interacts with Mg²⁺, F2 interacts with Lys104, and F3 interacts with Thr82. We had proposed a model for the transition state intermediate based on our unliganded Spo0F · Spo0B structure and the model is shown in Fig. 6b. It represents a pentacoordinated state where the phosphoryl group is covalently bonded to both proteins (43). When the covalent bond between Asp and the phosphorus atom is broken, the transfer from Spo0F to Spo0B is complete.

FIG. 6.

Active-site geometry when Spo0F binds to Spo0B. (a) Geometry of the active site as beryllofluoride Spo0F associates with Spo0B. The interactions of the three fluorine atoms with Nζ atom of Lys104, the carboxyl oxygen of Asp11, and Mg²⁺ are denoted by dashed lines. Mg²⁺, in addition, interacts with the carbonyl of Lys56 and the side chain oxygen atoms of Asp54 and Asp11. (b) Model for the transition state intermediate, created by placing a phosphoryl group between the active His30 and Asp54. The phosphorus atom forms partial covalent bonds with O^δ of Asp and N^ɛ of His and is in a pentacoordinated state. (Reprinted from reference 43 with permission of the publisher.)

Binding affinities of phosphorylated and nonphosphorylated states.

The phosphotransfer reactions are freely reversible, yet there is a preferential flow of the phosphoryl group from kinases downstream to transcription factors to ensure a signal response. Therefore, phosphorylation of Spo0F might reduce its affinity for kinases and increase the affinity towards Spo0B. In the crystal structure of the unliganded Spo0F · Spo0B complex (43), one of the four Spo0F molecules in the asymmetric unit was found to be loosely bound to the Spo0B molecule, making fewer interactions with Spo0B, while in the present BeF₃⁻-bound complex, the corresponding molecule binds in a manner similar to that of the other three Spo0F molecules. We interpret this observation as an indication for tighter binding of the Spo0F-P to Spo0B, compared with the unphosphorylated Spo0F.

The main interaction site of Spo0F on Spo0B is the four-helix bundle. As the four-helix bundle is a fairly rigid structure and the active histidines point to the solvent, it seems unlikely that phosphorylation causes major changes on the helix bundle of Spo0B. Hence, the proposed difference in affinities of Spo0F and Spo0F-P towards Spo0B would appear to arise from the phosphorylation-induced changes of Spo0F. As in other response regulators, in Spo0F the most significant changes upon phosphorylation also are on the β4-α4 loop. Interestingly, most of its interactions are with helix α2′ of the second protomer. Therefore, it is likely that discrimination between the phosphorylated and unphosphorylated states arises mainly from the interactions with helix α2′.

An overall view of the signal transmission in the phosphorelay.

In the phosphorelay, the journey of the phosphoryl group from ATP towards its destination on Spo0A involves four phosphotransfer reactions, and in this section, we will attempt to provide a structural view of these four steps, which are as follows: step 1, KinA + ATP ↔ KinA-P + ADP (phosphotransfer reaction 1 [autophosphorylation]); step 2, KinA∼P + Spo0F ↔ KinA + Spo0F∼P (phosphotransfer reaction 2); step 3, Spo0F∼P + Spo0B ↔ Spo0F + Spo0B∼P (phosphotransfer reaction 3); and step 4, Spo0B∼P + Spo0A ↔ Spo0B + Spo0A∼P (phosphotransfer reaction 4). The first two steps involve two distinct conformational states of KinA, and in the absence of crystallographic structures for KinA, we will use the tools of model building to generate pictures for the first two steps.

KinA has an N-terminal sensor domain and a C-terminal autokinase domain (Fig. 7a). The sensor domain comprises three PAS (Per-Arnt-Sim) domains. The autokinase domain consists of a phosphotransferase subdomain and an ATP binding subdomain. The four-helix bundle of Spo0B is formed through the dimerization of two helical hairpins from two monomers, and it is a prototype for the phosphotransferase domains of histidine kinases. A recent report of the crystal structure of the entire cytoplasmic portion of a histidine kinase supports this concept (24). The C-terminal domain of Spo0B is very similar to the ATP binding domains of histidine kinases, although it does not bind to ATP. These similarities make Spo0B a model for the autokinase domain of KinA.

FIG. 7.

KinA domain organization and proposed models. (a) Domain organization of KinA. KinA has 606 amino acid residues, and His405 is the site of phosphorylation. The N-terminal sensor domain is made up of three PAS domains. The PAS domains are involved in a variety of functions, but they appear to share a common fold. The autokinase domain consists of a phosphotransferase subdomain (PT) and an ATP binding subdomain. (b) Model for autophosphorylation. The two protomers are colored green and blue. In this model, the ATP domain associate with the four-helix bundle, bringing ATP (red) of the green protomer in proximity to the site of phosphorylation, H405 (red), of the blue protomer. The region from position 446 to 452 at the C-terminal end of helix α2 is hydrophobic. The side chains of Leu446, Leu449, Leu450, Met451, and Leu452 are shown in red. (c) Model for phosphotransfer from KinA to Spo0F developed on the basis of the Spo0F · Spo0B structure (43).

Modeling of the autokinase domain of KinA.

The four-helix bundle of KinA was modeled on the basis of the Spo0B bundle, and the ATP binding domain was modeled using the Spo0B structure and the ATP domains of EnvZ (32) and PhoR (25). The autophosphorylation reaction (step 1) requires the association of the ATP domain with the phosphotransferase domain, but the phosphotransfer from KinA to Spo0F (step 2) would require a displacement of the ATP domain to allow Spo0F to bind at the site of phosphorylation. This conformational isomerism has to be a general feature of all histidine kinases, although we do not know what triggers the conformational changes.

All the molecular modeling procedures were carried out with the “Homology” module of the program suite InsightII (Accelrys, 2005). The energy minimization calculations were done using the “DISCOVER” module from the same package. The models were energy minimized for 10,000 iterations of steepest descent, with the distance-dependent dielectric constant initially assigned a value of 2.

(i) Step 1.

The first step in the signal transmission is the autophosphorylation where the histidine residue in one protomer receives the γ-phosphoryl group from the ATP bound to the other protomer. In order for the autophosphorylation to take place, it is necessary for the ATP domain to orient suitably to bring the γ-phosphorus atom in close proximity to the active histidine. Figure 7b shows the conformation of the autokinase domain of KinA, which lends itself for autophosphorylation, and in this model the distance between the γ-phosphorus atom of ATP and the N^ɛ atom of His 405 is around 4.3 Å.

(ii) Step 2.

Figure 7c shows a model for Spo0F binding to KinA. After phosphorylating the histidine residue, the ATP domain is displaced, allowing Spo0F to bind to the four-helix bundle. We have modeled the KinA-Spo0F interactions based on the Spo0F · Spo0B structure. The interface between KinA and Spo0F is very similar to that between Spo0B and Spo0F (Table 2). In fact, only very small adjustments of the side chains of KinA were required to generate a favorable interface between KinA and Spo0F. In addition, alanine-scanning studies had shown that Spo0F makes use of the same surface for interacting with Spo0B and KinA (38).

TABLE 2.

Similarity of Spo0F interactions with Spo0B and KinA

Type of interaction	Residue of:
	Spo0F	Spo0B	KinA
Electrostatic	E21	K47	K422
Hydrophobic	V22	L45	L420
Electrostatic	D107	K40	K415
Possible hydrogen bond	K104	N34	N409
	Q12	N34	N409

Rowland et al. (31) have carried out the structural characterization of the antikinase Sda, and they have proposed a mechanism for KinA inhibition by Sda. Through mutational studies, they identified a hydrophobic patch on the N-terminal portion of Sda which binds to KinA. As Sda inhibits autophosphorylation without hindering phosphoryl transfer from KinA-P to Spo0F, the binding sites for Sda and Spo0F cannot overlap. Therefore, they propose that Sda binds to KinA either at the bottom of the four-helix bundle or at the adjacent region, the linker between the four-helix bundle and the ATP domain. An examination of our model shows that the C-terminal end of helix α2 comprises residues Leu446, Leu449, Leu450, Met451, and Leu452, making it predominantly hydrophobic. Therefore, this region could be a likely site for Sda binding (Fig. 7b). The binding of Sda at this region would block the association of the ATP domain with the four-helix bundle.

(iii) Step 3.

In step 3, the phosphoryl group is transferred from Spo0F-P to Spo0B. The current crystal structures of the BeF₃⁻-bound Spo0F · Spo0B complex and the unliganded Spo0F · Spo0B structure (43) are models for a pretransition state interaction and a transition state interaction.

Step 4.

In step 4, the phosphoryl group is transferred from Spo0B-P to Spo0A, and as the phosphoryl group moves from a histidine to an aspartate, the reaction may appear to be very different from the previous step. However, because the phosphotransfer reactions are freely reversible and the receiver domain of Spo0A is very similar to that of Spo0F, the mode of association between Spo0A and Spo0B will be very similar to the association between Spo0F and Spo0B.

Other models.

Marina et al. (24) have used the crystal structure of the cytoplasmic region of a sensor kinase from Thermotoga maritime to predict models for autophosphorylation and phosphoryl transfer to a response regulator. They proposed a model for the phosphotransfer reaction is based on the crystal structure of the Spo0F · Spo0B complex, and it is very similar to our model for step 2. On the other hand, their model for autophosphorylation differs from our model in the orientation of the ATP domain in our model for step 1. We consider both configurations to be possible models for autophosphorylation. Cai et al. (7) had proposed a model for autophosphorylation of the kinase EnvZ based on targeted disulfide cross-linking studies. The model was constructed using the solution structures of the two individual domains (32, 35). As the arrangement of helices in the solution structure of the EnvZ phosphotransferase domain is very different from the four-helix bundle of Spo0B, their model has large differences from our model. The crystal structure of the entire cytoplasmic portion of a histidine kinase (24) indicates that the helices in the phosphotransferase domains have a Spo0B-like arrangement.

Conclusion.

The crystal structure of the BeF₃⁻-bound Spo0F · Spo0B complex depicts the interactions of a phosphorylated response regulator with a phosphotransferase domain. It represents the initial stage of interaction before it enters into the transition state intermediate. The mode of molecular association seen in the present structure provides a basis for understanding two-component/phosphorelay signal transduction mechanisms.