HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 44, 42003-42010, November 1, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/44/42003    most recent M204789200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Guillet, V. Articles by Samama, J.-P. Search for Related Content PubMed PubMed Citation Articles by Guillet, V. Articles by Samama, J.-P. Social Bookmarking                      What's this?

Crystallographic and Biochemical Studies of DivK Reveal Novel Features of an Essential Response Regulator in Caulobacter crescentus^*

Valérie Guillet, Noriko Ohta^§, Stéphanie Cabantous, Austin Newton^§, and Jean-Pierre Samama^¶

From the  Groupe de Cristallographie Biologique, IPBS-CNRS, 205 route de Narbonne, 31077 Toulouse, France and the ^§ Department of Molecular Biology, Lewis Thomas Laboratory, Princeton University, Princeton, New Jersey 08544

Received for publication, May 16, 2002, and in revised form, August 2, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

DivK is an essential response regulator in the Gram-negative bacterium Caulobacter crescentus and functions in a complex phosphorelay system that precisely controls the sequence of developmental events during the cell division cycle. Structure determinations of this single domain response regulator at different pH values demonstrated that the five-stranded / fold of the DivK protein is fully defined only at acidic pH. The crystal structures of the apoprotein and of metal-bound DivK complexes at higher pH values revealed a synergistic pH- and cation binding-induced flexibility of the 4-4 loop and of the 4 helix. This motion increases the solvent accessibility of the single cysteine residue in the protein. Solution state studies demonstrated a 200-fold pH-dependent increase in the affinity of manganese for the protein between pH 6.0 and 8.5 that seems to involve deprotonation of an acido-basic couple. Taken together, these results suggest that flexibility of critical regions of the protein, ionization of the cysteine 99 residue and improved K_D values for the catalytic metal ion are coupled events. We propose that the molecular events observed in the isolated protein may be required for DivK activation and that they may be achieved in vivo through the specific protein-protein interactions between the response regulator and its cognate kinases.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The so-called two-component systems constitute widespread signaling cascades in microorganisms. These systems are generally represented by a sensor protein with histidine autokinase activity and a cognate response regulator (1, 2). The sensor protein is autophosphorylated on a conserved histidine residue, and the response regulator is activated by phosphorylation on the conserved aspartate residue in a Mg²⁺-dependent transphosphorylation reaction from the phosphohistidine of the cognate kinase. Phosphorylation of the response regulator elicits the appropriate cellular response through protein-protein interactions with downstream partners, or by activating or repressing gene(s) expression (3, 4). These His-Asp phosphorelay pathways mediate a wide array of physiological processes, most commonly in response to environmental conditions and external stimuli (5).

There is now evidence that the His-Asp family of proteins also plays essential roles in the regulation of the cell division cycle and cell differentiation in the Gram-negative aquatic bacterium Caulobacter crescentus. Cell cycle progression in these bacteria proceeds by a series of discrete morphogenic events and culminates in an asymmetric cell division to produce an new mother stalked cell plus a new motile swarmer cell (reviewed in Refs. 6 and 7). The two progeny cells inherit identical genomes but express different genetic programs. After cell division, the stalked cell immediately initiates DNA synthesis and develops into a predivisional asymmetric cell. The sibling swarmer cell, by contrast, is silent for DNA synthesis and does not initiate chromosome replication until undergoing a series of developmental events, including loss of motility, shedding of the flagellum, and formation of the cellular stalk (8). In this bacterium, multiple histidine kinases and response regulators make up a network of signal transduction pathways that couple the execution of successive developmental events to cell cycle progression (reviewed in Refs. 6 and 9).

The essential, single domain response regulator DivK plays a central role in the phosphorelay pathways controlling both cell division and motility (10). DivK belongs to the same protein subfamily as the response regulators CheY, which is involved in chemotaxis (11) and spo0F, which is required for sporulation in Bacillus subtilis (12). These regulatory domains generally contain ~130 amino acids and share a doubly wound / fold (5). Genetic results indicate that DivK functions downstream of histidine kinases PleC and DivJ (10, 13, 14). The purified cytoplasmic domains of PleC and DivJ efficiently phosphorylate DivK in the presence of ATP and dephosphorylate phospho-DivK (10). Consistent with the latter findings is the observation that DivK phosphorylation in whole cells is largely dependent on the DivJ kinase (15). Recent results have also shown that the PleC and DivJ kinases, as well as the DivK protein, are dynamically localized within C. crescentus cells during the cell cycle (15, 16). Although not yet demonstrated experimentally, the subcellular distribution of His-Asp proteins, along with modulation of the kinase and phosphatase activities of these signal transduction components, may be critical for the control of the activation pathways (Ref. 7, reviewed in Ref. 17).

In this report, we present several three-dimensional structures of DivK in its metal-free and metal-bound forms at different pH values. The metal-free response regulator adopts a stable fold only at pH 6.0, and the x-ray structures of the protein at higher pH values document atomic motions at critical regions of the response regulator. Coordination of the catalytic metal ion in the active site was unusual and increased the flexibility of the protein. The observations provided by the structures of the metal-bound species are consistent with the biochemical investigations which revealed a major pH dependence of the K_D value of the protein for the metal ion. The results collectively suggest a direct coupling between metal ion binding, flexibility of helix 4, and ionization of the single cysteine residue in the protein. These data provide new insights into the possible modulation of DivK activation throughout the cell cycle.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Chemicals-- Buffers were purchased from Sigma, Fischer, or Euromedex. N-(maleimidylethyl)-5-(4-methoxyphenyl)-oxazol-2-yl) pyridinium methanesulfonate (PyMPO-maleimide)¹ was purchased from Molecular Probes. All solutions (unless stated) were supplemented with fresh dithiothreitol in order to prevent the formation of disulfide-mediated DivK dimers.

Crystallization-- The purification and crystallization of the protein was described (18). Briefly, crystals were obtained at pH 6.0 using the vapor-diffusion method and polyethylene glycol monomethylether (PEG MME 550 32% v/v) as precipitating agent. They belong to the orthorhombic space group P2₁2₁2₁ with cell dimensions a = 37.2 Å, b = 40.5 Å, and c = 67.1 Å. The structure of the protein at pH 7.0, pH 8.0, and pH 8.5 was obtained from crystals transferred in reservoir solutions whose pH was increased from 6.0 to the desired value by steps of 0.5 pH units (the soaking time in each solution was 12 h). Manganese or magnesium derivatives were obtained by soaking for 24 h the protein crystals in the following solutions: 20 mM MnCl₂ (pH 6.0); 20 mM MnCl₂ (pH 7.0); 20 mM MgCl₂, 1 mM and 10 mM MnCl₂ (pH 8.0); 20 mM MgCl₂ and 10 mM MnCl₂ (pH 8.5).

Data Collection-- Crystals were cryocooled in liquid propane after soaking for a few seconds in a cryoprotecting solution made of 50% v/v PEG MME 550 in the reservoir. The data sets were collected at 100 K on the synchrotron beam lines W32 at LURE (Orsay, France), BM30A, ID14-EH3, and ID14-EH4 at ESRF (Grenoble, France), X11 and BW7B at DESY (Hamburg, Germany). Data processing and scaling were performed using MOSFLM (19) and SCALA (20). Statistical parameters are given in Table I.

Structure Determinations, Model Building, and Refinement-- The binding of heavy atoms did not occur in crystals grown at pH 6.0, but isomorphous substitutions were obtained when the crystals were brought to pH 7.0. Data sets were collected for crystals soaked in 5 mM mercury acetate, 0.5 mM samarium nitrate, and 1 mM uranyl nitrate for 79 h. Anomalous data were collected with a crystal soaked in 1 mM mercury acetate for 27 h. The heavy atom sites were located by inspection of the difference Patterson maps, and from Fourier difference maps using an initial set phases obtained from molecular replacement (18). Phasing was performed using SHARP (21) including the anomalous signal from the mercury derivative (Table I). The overall figure of merit for the MIRAS phases was 0.64 between 20 and 2.8 Å. The 2.8 Å electron density map was improved by several cycles of solvent flattening using SOLOMON (20). The mask was automatically calculated assuming a solvent content of 26% in the unit cell. The resulting electron density allowed chain tracing for most parts of the polypeptide chain, except for a stretch of 15 residues corresponding to the fourth helix and the preceding loop.

A polyalanine model was built in the experimental (3F_o  2F_c) electron density map displayed on a Silicon Graphics work station using TURBO-FRODO (22). The initial experimental phases, and those derived from this model, were combined using the program SIGMAA (23) and improved the quality of electron density map. This process was repeated until 72% of the residues (60% of the atoms) were built. This model was refined using CNS (24) including anisotropic B factor and bulk-solvent correction. About 10% of the data set was excluded from the refinement and used for the calculation of the free R factor (25). The high resolution data were progressively introduced during refinement and solvent molecules, with density higher than 4, were included in the last stages. In the protein structure at pH 7.0 solved to 1.8 Å resolution, there was no electron density for residues 84-90, and for the side chain of residues 101, 102, and 105. The final R and R_free values were 21.5% and 25.7%, respectively, and the average temperature factor for all protein atoms was 18.0 Å² (Table II). The refinement of the structure of the protein at pH 6.0 was performed using CNS (24) using as starting model the structure of the protein at pH 7.0. The final R and R_free values at 1.6 Å resolution were 19.1 and 21.2%, respectively (Table II). This monomer is fully defined from residues 2-125. The average temperature factor for all protein atoms was 12.3 Å². The mean coordinate error was estimated to be less than 0.18 Å (26). All residues belong to the most favored regions of the Ramachandran plot, except Gln-55. The side chains of residues Lys-3, Ser-37, Met-52, Asp-71, and Ile-116 were modeled as two-state conformers. The native DivK structures at pH 8.0 and pH 8.5 were refined to 1.87 and 1.65 Å, respectively, using the structure at pH 6.0 as starting model. There was no electron density for the region 84-93 and 84-89 at pH 8.0 and 8.5, respectively.

The Mg²⁺ or Mn²⁺ derivatives at pH 6.0, pH 7.0, pH 8.0, and pH 8.5 were refined using the same procedures. The electron density of the manganese ion was the highest positive peak in the F_o  F_c SIGMAA weighted map (Table II). The metal was introduced in the last stages of refinement. Statistics of the structure refinements are presented in Table II. The figures illustrating the protein structure were produced by BOBSCRIPT (27). The coordinates have been deposited in the Protein Data Bank (accession numbers: 1M5T, 1M5U, 1MAV, 1MB0, and 1MB3).

Fluorescence Experiments-- Steady-state fluorescence measurements were made using a P.T.I. (Photon Technology International) fluorescence spectrophotometer. The experiments were performed at pH 6.0, 7.0, 7.5, 8.0, and 8.5. Manganese (MnCl₂) was added stepwise to a 2-ml cuvette containing DivK (2 µM) in buffer. After each addition, the cuvette was incubated for 5 min at 4 °C using a thermostated stirred cell holder and the fluorescence of the single tryptophan in the protein (Trp-67) was measured (excitation wavelength = 295 nm, emission wavelength = 348 nm, bandwidth = 4 nm).

The fluorescence data were fitted to the following Equation 1,

(Eq. 1)

where F_corr is the value of F corrected for the dilution, F the fluorescence at infinite quencher concentration, [Q] the metal ion concentration in mM and K_SV the Stern-Volmer constant in mM¹, the reciprocal of the dissociation constant. The plots of the modified Stern-Volmer data provided the dissociation constant (K_D) at each pH value (Table III).

Fluorescence Labeling of DivK with PyMPO-- A monomer-dimer mixture of DivK was obtained by dialysis of the protein (60 µM) in 20 mM Tris-HCl, pH 8.0, 50 mM NaCl. This mixture was reacted with a 5-fold molar excess of the fluorescent probe PyMPO for 45 min at room temperature. The concentration of the stock solution of PyMPO (10 mM in Me₂SO) was determined spectrophotometrically (₄₁₂ = 23,000 M¹·cm¹). The products of the reaction were analyzed on non-reducing 15% SDS-PAGE. The fluorescence of free and DivK-bound PyMPO was measured using a STORM 840 gel imaging system (Molecular Dynamics, San Jose, CA) and the blue light as the excitation source. The digitized data were analyzed using ImageQuant software (Molecular Dynamics). Proteins were revealed by silver staining.

Circular Dichroism-- CD studies were performed using a Jobin-Yvon CD6 dichrograph at 20 °C. Spectra from 195 to 260 nm were recorded (0.2 nm·s¹) using a quartz cell with an optical pathlength of 1 mm. Solutions of DivK in 20 mM MES pH 6.0 or 20 mM Tris-HCl, pH 7.0, 50 mM NaCl, 1 mM dithiothreitol, at concentrations varying from 120 to 500 µg/ml, were used. The spectra were smoothed using the software supplied by the manufacturer (CD6 software). Each spectrum was the result of 5 scans using a bandwidth of 4 nm.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Overview of the DivK Crystal Structure-- The DivK protein crystallized in the orthorhombic P2₁2₁2₁ space group with one molecule in the asymmetric unit. Crystallization occurred only at pH 6.0 but once formed, the crystals could be manipulated at pH values as high as 8.5 without alteration of the diffraction pattern. The structure was solved using multiple isomorphous replacement with anomalous scattering (MIRAS) phasing (Table I). The structures presented in this report were refined at resolutions ranging between 2.3 and 1.35 Å, with standard values for the R, R_free, and mean temperature factors (Table II). There are two major crystal packing contacts. The first one is provided by one direct and three water-mediated hydrogen bonds between strand 2 from one molecule and residues 57-60 from the -loop of a symmetry-related molecule. The second one involves van der Waals contacts between the indole ring of Trp-67 in one molecule and the side chains of Glu-12 and Arg-33 of an adjacent molecule. These features were observed at all pH values.

The atomic positions in the metal-free protein were all defined only at pH 6.0. The 125 residues of DivK adopt a doubly wound 5-stranded / fold, with helices 1 and 5 on one side of the sheet and helices 2, 3, and 4 on the other side (Fig. 1). Helices 1, 2, 3, and 5 are N-capped by the side chains of residues Asn-11, Glu-58, Ser-60, and Ser-108, respectively. The cleft defined by loops 1-1, 3-3, 4-4, and 5-5 constitutes the active site of DivK (Fig. 1). It contains Glu-9, a residue that is commonly an aspartic acid in this protein family, Asp-10 and Asp-53 (the site of phosphorylation). The carboxylate groups of Glu-9 and Asp-53 are hydrogen bonded to the main chain nitrogen atoms of Asn-11 from loop 1-1 and Gln-55 from the 3-3 loop, respectively (Fig. 2). The side chain of Asp-10 points outside the active site and forms a salt-bridge interaction with Arg-33. Three water molecules are bound in the acidic pocket (Fig. 2). Water1 (Wat1) is coordinated to the carboxylate group of Glu-9, to the main chain nitrogen atom of Asp-10 and to Wat2. The Wat2 molecule is hydrogen bonded to the main chain carbonyl oxygen of Gln-55. The third water molecule (Wat3) bridges the side chains of Asp-53, Gln-55, and Lys-105. A salt-bridge interaction links Asp-53 and Lys-105, two invariant residues in response regulators. The hydroxyl group of Thr-83, a key response regulator residue in signal transduction (28), is hydrogen bonded to the amide side chain of Gln-55 (Fig. 2). Finally, the side chain of Tyr-102, another key residue in signal transduction (29-31) displays a solvent-exposed conformation.

View larger version (60K): [in this window] [in a new window]   Fig. 1.   DivK structure. A, ribbon diagram of DivK at pH 6.0 colored according to temperature factors (blue, low B; red, high B). B, structure-based sequence alignment. The solvent accessibility (acc) of the residues in DivK was scaled from white (low) to purple (high). The figure was created using ESPript (45). Secondary structure elements were calculated using the program DSSP (46).

View larger version (41K): [in this window] [in a new window]   Fig. 2.   DivK structure at pH 6.0. A, stereoview of the active site. The atoms are color-coded, and water molecules are represented as red spheres. B, stereoview of the interface between 4 and the protein core.

Very few interactions secure helix 4 (residues 88-96) to the core of the protein. They arise from van der Waals contacts between the side chain of Ile-94 and the side chains of Ala-81 (3.8 Å) and Thr-83 (3.8 Å) from strand 4, and of Cys-99 (3.3 Å) from the 4-5 loop (Fig. 2). Residues from the 4 helix have otherwise high accessibilities to solvent (Fig. 1).

Effect of pH on the Metal-free Protein-- At all pH values other than 6.0, i.e. pH 7.0, 8.0, and 8.5, there was no electron density for the 4-4 loop (residues 84-87) and for the N-terminal part of helix 4 (residues 88-96). In the active site area, Asp-10 was partially reoriented toward the active site, and only Wat3 was always observed. All other residues in the acidic pocket displayed identical atomic positions but the Tyr-102 and Lys-105 side chains were not visible. The loss of the salt-bridge interaction between Lys-105 and Asp-53 had no influence on the positions of the residues in the acidic pocket, suggesting that this geometry was independent of Lys-105. Excluding the flexible, non-modeled regions (residues 84-90 at pH 7.0, 84-93 at pH 8.0, and 84-89 at pH 8.5), the r.m.s. deviation between the C atoms in all metal-free DivK structures ranged from 0.45 to 0.55 Å. High r.m.s. deviations were found in the C-terminal half of the protein and were correlated with higher temperature factors (Fig. 3). The flexibility of the C-terminal moiety of the protein from the 4-4 loop suggested by these crystal structures was confirmed by the biochemical properties described later in this report.

View larger version (63K): [in this window] [in a new window]   Fig. 3.   Main chain temperature factors in the DivK structure solved at pH 8.0 (gray). The B factors were higher in the C-terminal part of the protein at all pH values. The r.m.s. deviation (black) between the structures solved at pH 6.0 and 8.0 indicates that most of the positional differences also occur in the C-terminal part. The figure was created with PRISM (GraphPad software).

Metal Binding in the Active Site-- At pH 6.0 and 7.0, there was no evidence for Mg²⁺ binding in the active site in crystals obtained by cocrystallization with this ion or in crystals soaked with 20 mM Mg²⁺ (data not shown). Binding of Mg²⁺ occurred at pH 8.0 and 8.5 with occupancies increasing from 0.3 to 0.8 (Table II). In contrast, using 20 mM Mn²⁺, metal binding occurred at all examined pH values with occupancies increasing with pH (0.5, 0.6, 0.7, 0.9 at pH 6.0, 7.0, 8.0, and 8.5, respectively, Table II). At much lower ion concentration (1 mM) and at pH 8.0, Mn²⁺ binds to the DivK protein with about the same occupancy. These observations were in good agreement with the pH dependence of the dissociation constants for these ions measured by fluorescence experiments (see next paragraph).

The metal ion corresponded to the highest positive peak in difference Fourier maps (Fig. 4). It induced no conformational change in the acidic pocket except for the side chain of Asp-10, which rotated in the inward position, and for the side chain of Asp-53, which flipped by about 80°. The ₁ and ₂ values for Asp-53 in the metal-free and metal-bound structures were (165°, 55°) and (165°, 25°), respectively, very similar to the corresponding values in the metal-free and metal-bound CheY structures (11, 32). The metal ion binding is not canonical in that it displays a distorted octahedral coordination that is not observed in the active sites of other response regulators and is not suitable for catalysis (Fig. 5). The ligands of the Mn²⁺ or Mg²⁺ ions were the oxygen atoms from the carboxylate groups of Glu-9, Asp-10, and Asp-53, and two water molecules. One of the water molecules was hydrogen bonded to the main chain carbonyl oxygen of Gln-55. The coordination of the metal to the third acidic residue (Glu-9) and the absence of direct coordination to the main chain carbonyl oxygen of Gln-55 differs from what has been reported for all metal-bound receiver domains (32-36).

View larger version (74K): [in this window] [in a new window]   Fig. 4.   Stereoview of the 2FoFc electron density map in the active site region for the Mn2+-DivK complex at pH 8.0.

View larger version (34K): [in this window] [in a new window]   Fig. 5.   Stereoview of metal binding in Mn2+-DivK complex at pH 8.0 (A) and Mn2+-DivK or Mg2+-DivK complexes at pH 8.5 (B). The metal is represented with a magenta sphere in the non-canonical position and as a green sphere in the canonical position. The second rotamer of Asp-53 is also represented in green.

Evidence for canonical metal ion binding was only observed at pH 8.5 in both the Mn²⁺-DivK and Mg²⁺-DivK complexes. In these structures, which were solved at 1.41 and 1.34 Å resolutions, respectively, the electron density for the metal ion was elongated and not spherical, as typically observed. We interpreted this observation as an indication that metal binding occurs at two alternate positions, separated by 1.5 Å (Fig. 5). Only the second position corresponded to the canonical location of the metal in the acidic pocket. The electron density also suggested the presence of two rotamers of the Asp-53 side chain.

Structural Effects of Metal Binding-- Metal binding had a destabilizing effect on the protein structure. The magnitude of this effect was correlated with the binding efficiency of the metal ion, estimated by the occupancy of the metal after refinement of the x-ray structures (Table II). At low occupancies (<0.5), metal binding induced flexibility of the 4-4 loop and of the N-terminal part of the 4 helix. This effect was revealed by the structure of Mn²⁺-DivK at pH 6.0, the single pH value at which all atomic positions were defined in the metal-free protein (the destabilizing effect of the metal ion could not be determined for the Mg²⁺-DivK complex at pH 8.0 because the flexibility of these regions was already induced by pH in the metal-free protein). Metal binding at occupancies higher than 0.6 increased the number of undefined atomic positions. In addition to the 4-4 loop, none of the 4 helix (residues 88-96) was visible (Fig. 6), and the electron density was weak for the C-terminal part of the 4-strand. These features were conveniently observed in the Mn²⁺-DivK complexes because metal binding occurred at all pH values. In all metal-bound structures, there was no electron density for Lys-105 and Tyr-102.

View larger version (22K): [in this window] [in a new window]   Fig. 6.   CA representation of the DivK structure illustrating by dotted line the non visible region (residues 84-97) when the metal binds with high affinity.

Dissociation Constants of the Metal Ions from Fluorescence Experiments-- The affinity of the protein for magnesium and manganese was evaluated by fluorescence measurements (37). The addition of Mn²⁺ to monomeric DivK resulted in saturation kinetics and induced a significant quenching of the fluorescence signal from the single tryptophan 67 residue in the protein, located in the middle of the 3 helix. The emission spectrum maximum at 348 nm is consistent with the solvent-exposed fluorophore and was unaffected by pH variations and addition of metal ions.

We observed a 200-fold pH-dependent variation of the K_D value for Mn²⁺. The dissociation constant decreased from 40 mM at pH 6.0 to 0.2 mM at pH 8.5 (Table III), following a pattern that suggested that the decrease in K_D values involves deprotonation of an acido-basic couple of about pK_a 7.4 (Fig. 7). This pK_a value could imply the involvement of histidine or cysteine residues. All three histidine residues (His-19, His-76, His-111) in the protein were fully exposed to solvent, and a possible relationship between their protonation states and the Trp-67 fluorescence would not be expected from the three-dimensional structure of the protein. The other possible candidate was the single cysteine 99 residue, which is located in the loop following helix 4 (4-5). In fact, solvent accessibility of Cys-99 increased from 10 Å² in the metal-free protein to 40 Å² in the Mn²⁺-DivK complexes in which the metal ion was bound with high occupancies. This increased accessibility to solvent results from the flexibility of helix 4 and likely favors ionization of the thiol group.

View larger version (9K): [in this window] [in a new window]   Fig. 7.   Variation of the KD values of manganese as a function of pH.

The decrease in fluorescence intensity upon addition of magnesium ion was at least 2-fold lower than the one observed with manganese, and the K_D values of Mg²⁺ for DivK could only be reliably measured at pH 8.0 and 8.5 (47 and 26 mM, respectively). These values are consistent with the lower occupancies of Mg²⁺ in the Mg²⁺-DivK complexes and the absence of a titration curve agrees with the restricted flexibility of helix 4 and smaller increase of the solvent accessibility of Cys-99 observed in the corresponding structures.

Accessibility of Cys-99 and Flexibility of Helix 4 in Solution-- In the absence of reducing agent, DivK undergoes dimerization. The monomeric and dimeric species could be clearly distinguished on non-reducing SDS-PAGE and on native gels. Dimerization in the absence of the reducing agent occurred within hours and at rates that increased with pH, leading to a dimer to monomer ratio of ~1 at pH 8.0. Addition of a reducing agent to this mixture led to 100% monomer, suggesting that dimerization of the protein was mediated by disulfide bridge formation through Cys-99. This conclusion was confirmed by measuring the reactivity of the cysteine residues in the monomer and in the dimer. Labeling by the fluorescent PyMPO label of the Cys-99 residue in the DivK dimer was only 10% of that observed in the monomeric protein (Fig. 8). Although purified monomeric DivK is readily phosphorylated by the DivJ kinase while the disulfide-mediated DivK dimer is not,² we have no evidence that DivK dimerization occurs in vivo or that dimerization plays a role in the regulation of DivK activity.

View larger version (26K): [in this window] [in a new window]   Fig. 8.   PyMPO labeling of Cys-99 in the DivK monomer and dimer. The mixture of monomers and dimers was incubated with the fluorescent probe and loaded on a non-reducing SDS-PAGE gel.

The finding that PyMPO labeled Cys-99 in monomeric DivK (Fig. 8) suggested that there is a significant population of solvent-exposed Cys-99 conformers. This is consistent with the ability of this residue to mediate dimer formation. It also indicates that the buried orientation of Cys-99 and the conformation of the region encompassing this residue in the crystal structure at pH 6.0 (Fig. 2) only represents a local energy minimum. This conformation must be altered for a solvent-exposed Cys-99. This conclusion was supported by the CD spectrum of a purified sample of the DivK dimer (Fig. 9), which significantly differed from the CD spectrum of the monomeric protein. The decreased molar ellipticity at pH 7.0 compared with pH 6.0 for the monomeric protein (Fig. 9) may be related to the pH-induced flexibility observed in the x-ray structures of the metal-free protein.

View larger version (23K): [in this window] [in a new window]   Fig. 9.   Circular dichroism spectra. CD of the DivK monomer at pH 6.0 and 7.0 (circles and squares gray lines, respectively) and of a protein solution enriched by gel filtration to 80% of disulfide-mediated dimer at pH 8.0 (black line).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The crystallographic and biochemical results reported in this paper shed light on several remarkable features of the DivK protein, an essential single domain response regulator required for cell cycle progression and polar morphogenesis in C. crescentus. These results lead us to conclude that the conformation of residues 84-100, which include the 4-4 loop, the 4-helix, and the 4-5 loop is critical for the regulation of the DivK activity. We also suggest that the dynamic changes in conformation of these regions may be coupled to the acido-basic properties of the single cysteine 99 residue.

The metal-free DivK protein is to our knowledge the first crystallized member of the response regulator family that exhibits a pH-dependent stability of the 4-4 loop, of the N-terminal half of the 4 helix, and of several residues and side chains, including those of the invariant Lys-105 and highly conserved Tyr-102. The flexibility of these regions is reminiscent of the microsecond time motions that have been reported from NMR studies for unphosphorylated CheY (38), Spo0F (39), and NtrC (40). The same regions were previously shown to undergo structural changes in activated receiver domains and to propagate the response to phosphorylation (41, 42)

In solution, the isolated DivK protein displays a higher affinity for Mn²⁺ than for Mg²⁺. This property has been documented for other response regulators, including CheY, whose catalytic properties are independent of which of the two metal ions are present (37). The binding constants for Mn²⁺ exhibit a pH dependence that follows an apparent acido-basic titration profile and suggest that deprotonation of a group with pK_a of about 7.4 may be involved in metal ion binding affinities in the millimolar range (Fig. 7). The molecular events underlying the biochemical results were revealed by the crystal structures of Mn²⁺-bound DivK. The flexibility of the 4-4 loop and of helix 4, and the solvent accessibility of Cys-99 from the 4-5 loop increase in direct relationship with metal binding occupancy. The biochemical and structural data collectively suggest that increased metal ion affinity, motion of residues 84-100, and deprotonation of Cys-99 are coupled events. It should be noted that the 4-5 loop is unusual among response regulators in containing the residues Gly-Gly-Cys (see Fig. 1B). This distinctive sequence may contribute to the dynamic properties of the 84-100 region of DivK and be responsible for the formation of the disulfide-mediated dimers in the purified protein.

The affinity of the protein for Mg²⁺ is severalfold lower than for Mn²⁺ and accordingly, the x-ray structures indicated reduced flexibility of helix 4 and solvent exposure of Cys-99 in the Mg²⁺-DivK structure. This low affinity seems unrelated to the occurrence of a glutamic acid at position 9 instead of the aspartic acid often found at this position. Indeed, the Escherichia coli PhoB protein, which also contains a glutamate residue at this position, displays a 2 mM K_D value for Mg²⁺ (43) and an active site geometry (44) very similar to that of DivK. The apparent contradiction between the K_D value of Mg²⁺ for DivK that seems too high to be physiologically relevant and the activity of DivK in transphosphorylation assays with this metal ion (10) suggests that other factors are required to stabilize a conformation of DivK in which the active site geometry is suitable for phosphotransfer and activation of the response regulator. This hypothesis is supported by the following observations.

First, the 30-fold improved K_D value of Mn²⁺ at pH 8.0 (1.3 mM) compared with pH 7.0 (41 mM) was not reflected in the structural details of metal binding in the acidic pocket. The structures of these two Mn²⁺-DivK complexes were essentially identical and the binding of the metal ion remained non-canonical and not suitable for catalysis. Second, the non-canonical coordination of the metal shifted partly, but not entirely, to a canonical one in both the Mg²⁺-DivK and Mn²⁺-DivK complexes at pH 8.5 (Fig. 5), although there is a 100-fold difference in affinity for these two ions (26 and 0.21 mM, respectively).

In vivo, DivK has at least three protein partners: the two histidine kinases PleC and DivJ and a putative phosphotransfer protein (Hpt or histidine phosphotransferase) that relays the phosphoryl group to the global regulator CtrA (DivJ ; DivK ; Hpt ; CtrA; Ref. 14). At the beginning of the swarmer cell cycle, DivK is evenly distributed in the cytoplasm of the swarmer cell. It then co-localizes with DivJ at the new stalked cell pole after the swarmer to stalked cell transition. Later in the cell cycle, DivK also co-localizes with PleC in the predivisional cell at the flagellated cell pole, which is opposite to the stalked cell pole (15, 16). These results are striking in conjunction with the finding of protein-protein interactions between DivK and the PleC and DivJ kinases that have been identified recently in a yeast two-hybrid screen (cited in Ref. 6). Taken together, these data raise the possibility that DivK associates with its cognate kinases to form stage-specific complexes during the cell cycle. In view of the results presented in this report, we speculate that the molecular interactions between the DivK response regulator and the cognate PleC and DivJ kinases may remodel the fold of the 4-4 and the 4 regions, decrease the K_D value of the metal ion, and favor a single conformation of the protein where coordination of the metal allows trans-phosphorylation. Such a mechanism would explain the in vitro activity of DivK at Mg²⁺ concentrations below the K_D value measured for the isolated protein and would provide a strong control for modulating DivK activity in vivo. It offers an explanation to the two observations mentioned in the preceding paragraph. Namely, that the biochemical and structural observations reported in this study, mediated by pH variations and cysteine ionization, only partly induce the conformational changes that are required for canonical metal ion binding in the acidic pocket and that would presumably occur in the DivK-kinase complexes.

	FOOTNOTES

^* Work carried out at Princeton University was supported in part by Public Health Service Grant GM58794 from National Institutes of Health. Work in France was supported by CNRS and the French Ministry of Education and Research.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

 The atomic coordinates and the structure factors (code 1M5T, 1M5U, 1MAV, 1MB0, and 1MB3) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).

^¶ To whom correspondence should be addressed. Tel.: 33-3-88-65-35-45; Fax: 33-3-88-65-32-76; E-mail: Jean-Pierre.Samama@igbmc. u-strasbg.fr.

Published, JBC Papers in Press, August 10, 2002, DOI 10.1074/jbc.M204789200

² T. S. Hofmeister, V. Guillet, N. Ohta, J.-P. Samama, and A. Newton, unpublished data.

	ABBREVIATIONS

The abbreviations used are: PyMPO, N-(maleimidylethyl)-5-(4-methoxyphenyl)-oxazol-2-yl) pyridinium methanesulfonate; MES, 4-morpholineethanesulfonic acid; r.m.s., root mean square.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				E. Nowak, S. Panjikar, P. Konarev, D. I. Svergun, and P. A. Tucker The Structural Basis of Signal Transduction for the Response Regulator PrrA from Mycobacterium tuberculosis J. Biol. Chem., April 7, 2006; 281(14): 9659 - 9666. [Abstract] [Full Text] [PDF]

				N. Ohta and A. Newton The Core Dimerization Domains of Histidine Kinases Contain Recognition Specificity for the Cognate Response Regulator J. Bacteriol., August 1, 2003; 185(15): 4424 - 4431. [Abstract] [Full Text] [PDF]

				L. Kroos and J. R. Maddock Prokaryotic Development: Emerging Insights J. Bacteriol., February 15, 2003; 185(4): 1128 - 1146. [Full Text] [PDF]