The Structure of a Full-length Response Regulator from Mycobacterium tuberculosis in a Stabilized Three-dimensional Domain-swapped, Activated State*

The full-length, two-domain response regulator RegX3 from Mycobacterium tuberculosis is a dimer stabilized by three-dimensional domain swapping. Dimerization is known to occur in the OmpR/PhoB subfamily of response regulators upon activation but has previously only been structurally characterized for isolated receiver domains. The RegX3 dimer has a bipartite intermolecular interface, which buries 2357 Å2 per monomer. The two parts of the interface are between the two receiver domains (dimerization interface) and between a composite receiver domain and the effector domain of the second molecule (interdomain interface). The structure provides support for the importance of threonine and tyrosine residues in the signal transduction mechanism. These residues occur in an active-like conformation stabilized by lanthanum ions. In solution, RegX3 exists as both a monomer and a dimer in a concentration-dependent equilibrium. The dimer in solution differs from the active form observed in the crystal, resembling instead the model of the inactive full-length response regulator PhoB.

The full-length, two-domain response regulator RegX3 from Mycobacterium tuberculosis is a dimer stabilized by three-dimensional domain swapping. Dimerization is known to occur in the OmpR/PhoB subfamily of response regulators upon activation but has previously only been structurally characterized for isolated receiver domains. The RegX3 dimer has a bipartite intermolecular interface, which buries 2357 Å 2 per monomer. The two parts of the interface are between the two receiver domains (dimerization interface) and between a composite receiver domain and the effector domain of the second molecule (interdomain interface). The structure provides support for the importance of threonine and tyrosine residues in the signal transduction mechanism. These residues occur in an active-like conformation stabilized by lanthanum ions. In solution, RegX3 exists as both a monomer and a dimer in a concentration-dependent equilibrium. The dimer in solution differs from the active form observed in the crystal, resembling instead the model of the inactive full-length response regulator PhoB.
Tuberculosis is a global threat to human health, with onethird of the world's population infected with the causative agent of tuberculosis, Mycobacterium tuberculosis. With the emergence of multidrug-resistant strains of M. tuberculosis, novel therapeutic agents are urgently required. This will be best achieved by developing a better understanding of the molecular biology of this pathogen. In this context, signaling systems play an important role in the continued survival of the organism and are central to the way M. tuberculosis responds to environmental stress, especially that generated by the host immune system. The predominant signaling system for adaptive gene expression changes in bacteria is called a two-component system (TCS). 3 TCS are found in eubacteria, archaea, and eukarya; however, they are rare in eukarya. As the name implies, the system typically consists of two proteins: a sensor histidine kinase (SK) and a response regulator (RR) (1). SKs are usually membrane-anchored proteins with a characteristic core consisting of a histidine-containing dimerization domain and a catalytic domain with ATPase activity. In response to extracellular conditions, and in some cases intracellular conditions, the SK autophosphorylates at a histidine residue within the dimerization domain. This autophosphorylation occurs upon dimerization through a phosphate transfer from the catalytic domain (annotated as a HATPase_c domain by the SMART server (2,3)) of the adjacent protomer to the conserved histidine in the dimerization domain (HisKA domain). The activated SK may then function as a phosphate donor to a universally conserved aspartic acid residue in the RR. RRs typically consist of two domains, a receiver and an effector domain, although a large family of 54 activators also contain a central AAAϩ domain. The N-terminal receiver domain acts as the phosphoacceptor, whereas the C-terminal effector domain is usually a DNA-binding domain involved in the transcriptional regulation of genes required to respond to the sensed environment. Phosphorylation of the receiver domain results in the activation of the effector domain in a manner that is not yet completely understood, in part due to the lack of structural information on full-length RRs in the activated state. Dimerization is a proposed method of regulation (4). This is based upon structures of the receiver domain alone in both active and inactive states (4).
Some TCS appear to be essential for the survival of bacteria, and to date homologous TCS have not been identified in the animal kingdom. TCS are therefore considered attractive targets for the development of novel agents that could alter the response of the bacterium to its detriment. More than 180 different TCS have been identified in bacteria (5); however, the number varies enormously between species. M. tuberculosis contains 30 genes encoding TCS proteins (6), 12 complete TCS, and several orphan TCS proteins possibly belonging to the SK and RR families (6,7). Of these 12 complete TCS, all but five have been found deleted or exist as pseudogenes in the closely related intracellular pathogen Mycobacterium leprae (8). The TCS SenX3-RegX3 is one of the five conserved TCS, emphasizing its importance for the survival of mycobacteria (not just the virulent strains) in hostile environments. Previous studies carried out on this TCS have revealed that both SenX3 and RegX3 communicate in the normal fashion, with His 167 of SenX3 being the active histidine residue that is phosphorylated and Asp 52 being the corresponding phosphorylation target in RegX3, the cognate RR (9). The production of M. tuberculosis strains with a compromised senX3-regX3 operon (through the removal of the 3Ј end of the senX3 gene, the intergenic repeats, and the 5Ј end of the regX3 gene) has shown that an intact SenX3-RegX3 TCS is required for a progressive infection (10). Microarray experiments comparing M. tuberculosis strains with and without the senX3-regX3 operon revealed 50 possible regulon members (10). One operon in particular, Rv0096-Rv0101, appears to be directly, negatively regulated with expression found to be 5-17-fold increased in the absence of the SenX3-RegX3 TCS (10). Further experiments using M. tuberculosis strains with a compromised senX3-regX3 operon displayed significant attenuation in active and resting macrophages as well as in immunocompromised and immunocompetent mice, thus demonstrating that the senX3-regX3 operon is involved in the virulence of M. tuberculosis (10). Recently, experiments in Mycobacterium smegmatis revealed that the SenX3-RegX3 system responds to limiting environmental phosphate concentrations by inducing the phoA and pstS genes (11). In E. coli, these genes are regulated by another TCS, PhoR-PhoB (12,13). Through DNAbinding experiments, it was shown that phosphorylated RegX3 directly regulates phoA and pstS, in M. smegmatis, by binding to their promoter regions. Previously, RegX3 has only been demonstrated to have DNA-binding properties with the promoter region of senX3 (9). A putative "pho-box" for RR binding has been derived for M. smegmatis consisting of an inverted repeat (GTGAAC) separated by seven nonconserved nucleotides (11).
Most RRs are multidomain proteins. All RRs contain structurally similar receiver domains with a characteristic ␤1-␣1-␤2-␣2-␤3-␣3-␤4-␣4-␤5-␣5 topology and a conserved aspartic acid that is phosphorylated. They can be subdivided into families based upon similarities in their (output) effector domains. The three largest families are named after their founding members: OmpR/PhoB, FixJ/NarL, and NtrC/DctD. The RR studied here belongs to the OmpR/PhoB family. This family is characterized by a winged helix-turn-helix (wHTH) DNA-binding domain (14 -16). This domain has been experimentally shown to bind direct tandem repeat half-sites (17)(18)(19) and inverted repeats (11) (20) and PrrA (21) exist as very compact structures with the recognition helix completely inaccessible to DNA. By comparison, the structures of DrrD (22) and DrrB (23) exhibit very open conformations with the DNA recognition helix of the effector domain completely accessible. In other RR families, the recognition helix is occluded in the unphosphorylated state, becoming accessible only upon phosphorylation (24,25). This was recently suggested to be the effect of signal transduction by phosphorylation in PrrA (21). PrrA is one of the closed, compact structures of the OmpR/PhoB subfamily. Dimerization has been suggested as important for signal trans-duction in other OmpR/PhoB members. This is based upon several biochemical experiments and the two full-length crystal structures of DrrD (22) and DrrB (23), which are both open structures with accessible recognition helices. The ␣4-␤5-␣5 face of the receiver domain of a RR is the region most affected by phosphorylation and is the proposed site of dimerization in the OmpR/PhoB family. This site in this family displays high residue conservation as compared with other RR families (4), consistent with the importance of this region as a dimerization interface. However, dimerization is not only restricted to activated RRs. ArcA, for instance, exists as a dimer in the inactive state and forms an octamer upon activation (26). PmrA is also known to exist as a dimer in both states (27), and PhoB has been shown to undergo a transition that involves changing the dimerization interface from the ␣1-␣5 region in the inactive form to the ␣4-␤5-␣5 region in the active form (28,29).
Three-dimensional domain swapping is becoming increasingly more evident in the Protein Data Bank as more structures are deposited. Three-dimensional domain swapping describes the formation of a symmetry interaction between two molecules via the exchange of identical regions. The region that is swapped can be a single element of secondary structure or an entire globular domain (30). Although many proteins display this phenomenon, the functional and physiological relevance is often not understood. Some potential advantages attained through domain swapping include higher local concentrations of active sites, larger binding surfaces, new active sites at subunit interfaces, the possibility of allosteric control, and an economic way to produce a large protein interaction network (31). Here we present the structure of the full-length RR, RegX3, stabilized in what we believe to be an active form that exhibits three-dimensional domain swapping. As such, it would represent the first full-length RR to be captured in an activated dimeric state and to display three-dimensional domain swapping. We also present an analysis of the solution structure of RegX3 in an inactive form as both a monomer and dimer.

EXPERIMENTAL PROCEDURES
Cloning and Expression-The gene sequence encoding RegX3 was cloned by PCR from genomic Mycobacterium tuberculosis (H37Rv strain) DNA. Flanking DNA primers 5Ј-TATACCATG-GCAACCAGTGTGTTGATTGTGGAGG-3Ј and 5Ј-CAGAA-GCTTATTACTAGCCCTCGAGTTTGTAGCCCAGCCCG-CGC-3Ј were engineered to provide melting temperatures of ϳ65°C with extensions encoding NcoI and HindIII (New England Biolabs, Frankfurt, Germany) restriction sites. The corresponding PCR product was ligated with T4 DNA ligase (New England Biolabs, Frankfurt, Germany) into a pETM11 (EMBL) expression vector with an N-terminal His 6 tag. The forward primers inserted a GCA codon immediately following the start codon to preserve the reading frame. Following sequence confirmation, the plasmid was transformed into Rosetta(DE3)pLysS cells. Cells were induced with 1 mM isopropyl 1-thio-␤-D-galactopyranoside at A 600 ϭ 0.7 and grown for 5 h at 25°C. The cells were harvested by centrifugation and lysed by sonication in a buffer containing 50 mM Tris⅐HCl, pH 8.0, 200 mM NaCl, 5 mM ␤-mercaptoethanol (buffer A), and two tablets of protease inhibitor mixture (Roche Applied Science).
The lysate was spun down and loaded onto a 5-ml nickel-nitrilotriacetic acid affinity column (Qiagen) equilibrated in buffer A. The protein was eluted with a 100-ml linear gradient of 10 -250 mM imidazole in buffer A. The fractions containing RegX3 were pooled and diluted to a final imidazole concentration of 50 mM. The protein was then mixed with (His-tagged) tobacco etch virus protease in a 1:20 ratio and incubated at 4°C overnight in order to cleave the His 6 tag. The digested protein was applied to the nickel-nitrilotriacetic acid affinity column previously equilibrated in buffer A containing 50 mM imidazole. The flow-through fractions were collected, concentrated to a final volume of 5 ml, and further purified on a HiLoad 16/60 Superdex 75 column (Amersham Biosciences) equilibrated with 20 mM Tris⅐HCl, pH 8.0, 100 mM NaCl, and 2 mM dithiothreitol (buffer B). The peak fractions were concentrated using a 10-kDa cut-off Centricon (VIVASPIN) to a final concentration of 10 mg⅐ml Ϫ1 as measured by Bradford's assay (Bio-Rad) (32).
Crystallization, Data Collection, and Processing-Crystallization of RegX3 was carried out at 19°C using the hanging drop vapor diffusion method. The initial conditions (0.2 M calcium acetate, 18% polyethylene glycol 8000, 0.1 M sodium cacodylate at pH 6.0) were identified using the Crystal Screen from Hampton Research.
Following optimization, the best crystals were obtained after 5 days in a mixture containing 2 l of protein solution with 2 l of reservoir buffer. The reservoir buffer consisted of 0.2 M calcium acetate, 2-4% polyethylene glycol 4000, and 0.1 M sodium cacodylate at pH 6.5. These crystals diffracted to a maximum resolution of 6.0 Å. To obtain better quality crystals, it was necessary to mix the protein drop with LaCl 3 to obtain a final concentration in the crystallization drop of 0.01 M La 3ϩ . For data collection, the crystals were transferred to a drop-containing reservoir solution mixed with 2-methyl-2,4-pentendiol to a final concentration of 20% and then flash-cooled to 100 K in a cold nitrogen gas stream. Two data sets were collected: one at ϭ 1.5 Å, to a resolution of 2.6 Å, on the BW7A beamline in order to increase the anomalous signal from the La 3ϩ ions and the second at shorter wavelength and higher resolution (2.03 Å) on the X11 beamline. All measurements were carried out on EMBL beamlines at DESY, Hamburg, using MAR165 CCD detectors.
The data were indexed and integrated using DENZO (33) and scaled using SCALEPACK (33). The precision-indicating merging R factor was calculated using the program RMERGE (34). Intensities were converted to structure amplitudes using the program TRUNCATE within the CCP4 program suite (35). The crystals belong to the space group P42 1 2 with one molecule in the asymmetric unit. Data collection details are summarized in Table 1.
Structure Determination-The structure was solved to a resolution of 2.6 Å in the P42 1 2 space group using the single anomalous diffraction protocol of AUTO-RICKSHAW, an automated crystal structure determination platform (36). Within this procedure, the structure factors were calculated from the measured intensities by employing TRUNCATE, and the five lanthanide positions were located using SHELXD (37). The positions of these ions were refined using the program MLPHARE (35), and the resultant phases were subjected to density modification using DM (38). Although the maximum resolution was 2.6 Å, the solvent content (63%) was relatively high, and model building was performed using the program ARP/wARP (39). The resultant model contained 184 residues of 228. In the last cycle of the ARP/wARP, 144 residues were docked with the correct sequence. The model was completed with manual building using COOT (40) alternating with additional REFMAC5 (41, 42) refinement cycles. The refinement cycles included a bulk solvent correction and anisotropic scaling and used the entire molecule as a single rigid body for TLS refinement. The resultant refined coordinates were then used to refine against the 2.03 Å data set using REFMAC5. The overall geometric quality of the model was assessed using the program PROCHECK (43). The refined coordinates and the structure factors for RegX3 have been deposited in the Protein Data Bank (Protein Data Bank code 2OQR). Data collection and refinement statistics are shown in Table 1. The superimpositions and r.m.s. deviation calculations were carried out using the programs COOT and SSM (44). All molecular images were generated using PYMOL (45).
Small Angle Scattering-Synchrotron x-ray scattering data from solutions of RegX3 were collected at the X33 EMBL beamline using an MAR345 image plate detector (46). The scattering patterns were measured with a 3-min exposure time for multiple solute concentrations ranging from 2.7 to 15.3 mg⅐ml Ϫ1 . No indication of radiation damage was detected. This was checked by the comparison of duplicate scattering patterns from 2-min exposures. Using a sample-detector distance of 2.7 m, the momentum transfer range, 0.009 Ͻ s Ͻ 0.5 Å Ϫ1 , was covered (where s ϭ 4sin()/, 2 is the scattering angle, and ϭ 1.5 Å is the x-ray wavelength).
The data were processed using standard procedures by the program package PRIMUS (47). The forward scattering I(0) and the radii of gyration R g were evaluated using the Guinier approximation (48), assuming that at very small angles (s Ͻ 1.3/R g ), the intensity is represented as I(s) ϭ I(0)exp(Ϫ(sR g ) 2 /3). The maximum particle dimensions (D max ) and the interatomic distance distribution function p(r) were computed using the program GNOM (49). The molecular mass (MM) of the solute at various concentrations was evaluated by a comparison of the forward scattering with that from a reference solution of bovine serum albumin (MM ϭ 66 kDa). Particle shape at low resolution was reconstructed ab initio using the program DAMMIN (50), which represents the particle as a collection of M Ͼ Ͼ 1 densely packed beads inside a sphere with the diameter D max . Each bead belongs either to the particle or to the solvent, and the shape is described by a binary string of length M. Starting from a random string, simulated annealing is employed to search for a compact model that fits the experimental data I exp (s) to minimize the discrepancy , where N is the number of experimental points, c is a scaling factor, I exp (s j ) and I calc (s j ) are the experimentally determined and calculated intensities, respectively, and (s j ) is the experimental error of I exp (s j ) at the momentum transfer s j . The program GASBOR (51) was also used to create ab initio models consisting of dummy residues instead of beads. Simulated annealing is employed to construct a model with protein-like distribution of beads providing the best fit to the experimental data. For the ab initio and rigid body analyses, multiple runs were performed to verify the stability of the solution, and typical three-dimensional reconstructions are presented below. The scattering patterns from atomic models were computed using the program CRYSOL (52). Rigid body modeling was performed by the global refinement program SASREF (53). The full-length RR, DrrD, was used as a template and was broken into two domains: the receiver (residues 1-118) and the effector (residues 119 -217). Based upon the dimerization interface reported for the inactive PhoB structure (Protein Data Bank code 1B00), a model was generated that represented the optimal spatial configuration of the domains against the scattering data recorded at the highest protein concentration (15.3 mg⅐ml Ϫ1 ). The program OLIGOMER (47) was used to calculate the ratio of monomeric and dimeric species present in solutes at intermediate concentrations (between 4.8 and 11.3 mg⅐ml Ϫ1 ). The ratios were determined based upon the scattering intensities of monomeric DrrD (Protein Data Bank code 1KGS) calculated by CRYSOL and the model of the dimeric protein constructed by SASREF.

RESULTS
The structure of the full-length RegX3 RR was determined using experimental phases obtained from single wavelength ( ϭ 1.5 Å) anomalous dispersion data from a single crystal co-crystallized in the tetragonal space group P42 1 2 with lanthanum chloride. Phasing information from the five lanthanum ions in the asymmetric unit was used to calculate the initial experimental electron density map. Following manual building of the initial model, the structure was refined to 2.6 Å. The resulting coordinates were then used to refine against 2.0 Å data collected at ϭ 0.813 Å from a second crystal co-crystallized with lanthanum chloride. The final model was refined with REFMAC5 to a final R work of 0.178 and R free of 0.213 (Table 1). All residues were visible except for the initial methionine, the final glycine, and the first two residues prior to the methionine, which were introduced during cloning. The resulting electron density map shows strong electron density for the five La 3ϩ ions.
Overall Structure-The overall structure, shown in Fig. 1, consists of two domains: the receiver domain, which at first sight has the expected secondary structural elements although in a totally new spatial arrangement, and a wHTH DNA binding domain characteristic of the OmpR/PhoB RR subfamily (14 -16). The structure contains five lanthanum ions that were identified and distinguished from other possible metal ions by their anomalous signal. Of the five ions, one occupies the active site, two interact between the domain-swapped receiver domain and the effector domain, and the remaining two stabilize the crystal lattice (see below). The largest contact area between molecules is mediated by the ␣4-␤5-␣5 face and forms a twofold symmetric crystallographic dimer burying a total surface area of 2357 Å 2 /monomer (Fig. 2). An examination of the crys- where N represents the redundancy, I i is the intensity of the ith reflection, and I is the average intensity.
where F o and F c are the observed and calculated structure factors, respectively. e R free was calculated as for R work but from a randomly selected subset of the data (5%) excluded from refinement. tal packing reveals three other contact areas that stabilize the crystal lattice. One contact occurs between the C-terminal end of ␣2, the ␣2␤3-loop, the N-terminal end of ␣4, and the ␣3␤4loop of the protomer in the asymmetric unit with the N-terminal end of ␣8 and the ␤11␤12-loop of a symmetry-related mol-ecule (Ϫx ϩ 1 ⁄ 2, y Ϫ 1 ⁄ 2, Ϫz) burying an area of 591 Å 2 /monomer. The next interface occurs between the N-terminal region of ␣1 and the linker region between ␣5 and ␤6 of a symmetry-related molecule (Ϫy, Ϫx, Ϫz ϩ 1). The smallest interface, burying 208 Å 2 , occurs between the ␤11␤12-loop and the C-terminal region of ␣5 from a symmetry-related molecule (y ϩ 1 ⁄ 2, x Ϫ 1 ⁄ 2, z).
A typical receiver domain has an ␣/␤ topology that consists of a central five-stranded ␤-sheet (␤2-␤1-␤3-␤4-␤5), which forms a hydrophobic core surrounded by two helices (␣1 and ␣5) on one side and three (␣2-␣4) on the other side. The monomer of RegX3 in the asymmetric unit has an apparently incomplete receiver domain with only a four-stranded parallel ␤-sheet (␤2-␤1-␤3-␤4) and three ␣-helices, of which ␣1 is positioned on one side of the ␤-sheet and ␣2-␣3 are positioned on the other side (Fig. 1). The linker region connecting the two domains is made up of ␣4 and ␤5, which, as stated above, form a strong dimerization interface. The three-dimensional domain-swapped element, namely the incomplete receiver domain (␤1, ␣1, ␤2, ␣2, ␤3, ␣3, and ␤4) from molecule 2 of the crystallographic dimer, interacts with ␣4, ␤5, and ␣5 from molecule 1. Strand ␤5 of molecule 1 aligns parallel to the domain-swapped, four-stranded ␤-sheet of molecule 2, forming a composite five-stranded ␤-sheet. Helix ␣4 of molecule 1 lies just outside of ␤5 on one side of the composite ␤-sheet with ␣1 and ␣2 from the domain-swapped element, whereas ␣5 from molecule 1 lies on the other side of the composite ␤-sheet (Fig. 2). The new composite receiver domain is now complete, with three ␣-helices on the other side of a five-stranded ␤-sheet. When the composite receiver domain is superimposed with other previously characterized receiver domains from the OmpR/PhoB family (Fig. 3A), such as the active PhoB receiver domain (29), the inactive PhoB receiver domain (28), DrrD (22), DrrB (23), PrrA (21), and MtrA (20), the secondary structures occupy similar positions with an overall r.m.s. deviation between C ␣ atoms ranging from 1.45 to 2.19 Å ( Table  2). Not surprisingly, the region of highest variance is the dimerization interface composed of ␣4␤5␣5.
The effector domain of RegX3 has a typical wHTH fold, consisting of a four-stranded antiparallel ␤-sheet followed by a three-helix bundle and a C-terminal ␤-hairpin (Fig. 2). The DNA recognition helix (␣8) is fully exposed to the solvent, as is also the case for the OmpR/ PhoB family members DrrB (23) and DrrD (22) but unlike the other structurally characterized fulllength family members PrrA (21) and MtrA (20).
The Receiver Domain-The active site of all RRs is in a crevice in the receiver domain. In the case of RegX3, this is formed by the ␤5␣5loop from molecule 1 and by the FIGURE 1. The full-length RR, RegX3, represented by the C ␣ trace of the single polypeptide chain in the asymmetric unit (ASU). Every 10th residue is labeled. Five La 3ϩ ions appear in the asymmetric unit and are colored magenta. The main C ␣ chain is colored from blue to red going from the N to C terminus.

FIGURE 2.
A schematic representation of the three-dimensional domain-swapped RegX3 dimer. The dimer consists of molecule 1 and molecule 2, colored red and green and colored blue and magenta, respectively. The three-dimensional domain-swapped elements from each molecule consist of ␤1, ␣1, ␤2, ␣2, ␤3, ␣3, and ␤4 and are colored green and magenta, respectively. The three-dimensional domain-swapped element from molecule 2 (magenta) completes the receiver domain of molecule 1 (red) by forming a composite receiver domain with ␣4, ␤5, and ␣5 from molecule 1. The same occurs in molecule 2 (blue) with the three-dimensional domain-swapped element from molecule 1 (green). Glu 84 is displayed in a stick representation. This residue from ␣4 is part of the active site from the same protomer.
␤1␣1-loop, ␤3␣3-loop, and ␤4␣4-loop from the domainswapped element of molecule 2. The active site of the complete, but composite, receiver domain contains all of the residues essential for phosphorylation. These are Glu 8 , Asp 9 , Glu 10 , Asp 52 , Leu 53 , Met 54 , Thr 79 , and Glu 84 from the three-dimensional domain-swapped element and Lys 101 . In some RRs, Glu 8 and Glu 10 are replaced by aspartate. It is common to find a divalent metal ion in the active site of an RR, since RRs are phosphorylated by their cognate SK in a magnesium-dependent reaction. La 3ϩ is found in the active site of RegX3, where it is predominantly coordinated by acidic side chains. The ion displays nonadentate coordination involving both carboxylate oxygen atoms from Asp 9 and Asp 52 , a single carboxylate group from Glu 84 , and a carbonyl oxygen atom from Met 54 (Fig. 4D). Water molecules occupy the three remaining coordination sites. Nonadentate coordination is sterically facilitated, because two residues (Asp 9 and Asp 52 ) exhibit bifurcated coordination through both carboxylate groups. The distance between La 3ϩ and the coordinating atoms lies in the range of 2.23-2.85 Å. Glu 8 and Glu 10 , although essential for phosphorylation, are not directly involved in the coordination to La 3ϩ and instead form strong hydrogen bonds with one and three water molecules, respectively. The water molecule interacting with Glu 8 is also hydrogen-bonded to Glu 10 bridging the two residues. Asp 54 , Thr 81 , Lys 103 , Ile 55 , Asn 56 , and Gly 82 have been shown, in the structure of the activated RR ArcA, to be important in the stabilization of the phosphate mimic BeF 3 Ϫ (4). In RegX3, the active site residues Asp 52 , Thr 79 , Lys 101 , Leu 53 , and Glu 84 are required for the formation and stabilization of the phosphorylated species. Of particular interest is the residue Lys 101 on ␤5, since it is the only residue that is not from the three-dimensional domain-swapped element. It forms a hydrogen bond (2.84 Å) with a carboxylate oxygen atom from Asp 52 . In the phosphorylated molecule, this bond would be broken, and an alternative hydrogen bond would be formed between the phosphate moiety and Lys 101 , as has been observed in ArcA (4).
The signature switch residues known to be crucial in signal transduction by the rearrangements resulting from phosphorylation are serine/threonine and phenylalanine/tyrosine located in the ␤4␣4-loop and in ␤5, respectively (54,55). In inactivated RRs the tyrosine or phenylalanine occupy positions facing away from the active site in gauche Ϫ (OH group points outward from the domain, exposed to solvent) and gauche ϩ (OH group points toward the C terminus of ␣4) positions. This is believed to be due to the outward orientation of the threonine/serine residue on the ␤4␣4-loop. In RegX3, these residues are Thr 79 and Tyr 98 , respectively. Thr 79 is from the three-dimensional domain-swapped element, whereas Tyr 98 , like Lys 101 , is not. In the RegX3 structure, Thr 79 is orientated toward the active site in a position that would coordinate phosphate, if it were there, whereas Tyr 98 adopts an inward transposition, occupying the space freed by the inward rotation of Thr 79 (Fig. 5). La-1 is coordinated by the carbonyl oxygen atom of Tyr 98 . This La 3ϩ ion is positioned between ␣4 and ␤5. It also displays nonadentate coordination, interacting directly with both carboxylate oxygen atoms of Glu 86 , a single carboxylate oxygen atom from Asp 97 , and the carbonyl oxygen atom from  Tyr 98 , the switch residue. Five water molecules complete the coordination (Fig. 4A). The coordination of the carbonyl oxygen atom of Tyr 98 forces the aromatic ring to be orientated inward in the active state position. Dimerization Interface-The total buried surface area created as a result of dimerization is 2357 Å 2 /monomer shared equally between two distinct subinterfaces. The first, which we refer to as the dimerization interface, buries 1169 Å 2 /monomer and is formed between the two composite receiver domains. It is centered on the helices that pack about the crystallographic two-fold axis (Fig.  6A). The second interface is formed between the composite receiver domain (consisting of the N-terminal, three-dimensional domainswapped element, ␣4, ␤5, and ␣5) and the effector domain (Fig. 6B). The dimerization interface contains four salt bridges per monomer with a forked salt bridge within the active site between E8 (␤1␣1-loop) and D52 (C terminus of ␤3) from the domain-swapped element of molecule 2 and Lys 101 (␣4␤5-loop) from molecule 1. A salt bridge is also formed between Asp 96 (␣4␤5-loop) from molecule 2 and Asp 68 (␣3␤4loop) from molecule 1. The extended five-stranded ␤-sheet created by the three-dimensional domain swapping is stabilized by an extensive network of hydrogen bonds, which further strengthen the interface. Asp 68 (C terminus of ␣3) from the domain-swapped fragment of molecule 2, already participating in a salt bridge, is further stabilized by Gly 94 (␣4␤4-loop) from molecule 1. Ile 76 (␤4), also from the domain-swapped element, stabilizes Asp 96 (␤5) from molecule 1 by forming two hydrogen bonds. One hydrogen bond is formed between the carbonyl oxygen atom of Ile 76 and the main-chain nitrogen atom of Asp 96 . The other bond is formed between the main-chain nitrogen atom of Ile 76 and one of the carboxylate oxygen atoms of Asp 96 that is already involved in a salt bridge. Two hydrogen bonds are formed between both the carbonyl oxygen atom and main-chain nitrogen atom of Val 78 (␤4) from molecule 2 with Asp 97 and Val 99 (␤5) from molecule 1, respectively. The last two hydrogen bonds are formed between the main-chain nitrogen atom and carbonyl oxygen atom of Ala 80 (␤4) from molecule 2 and the carbonyl oxygen atom of Val 99 (␤5) and the main-chain nitrogen atom of Lys 101 (␣4␤5-loop) from molecule 1, respectively. The last salt bridge involved in the dimerization interface is formed between Asp 97 (␤5) and Arg 111 (␣5) both from molecule 1, ensuring the two secondary structure elements maintain close contact as they complete the composite receiver domain. The hydrophobic surface formed by Leu 108 , Ile 109 , Ile 112 , Val 115 , and Leu 116 of ␣5 protects an otherwise exposed hydrophobic surface of the receiver domain formed by ␣1 and the central ␤-sheets. In other RRs, this hydrophobic surface is protected by both ␣4 and ␣5; however, in RegX3, ␣4 acts as a linker packing against the ␣4 of the dimeric partner. The two helices pack against each other on a hydrophobic interface made by Ile 85 , Val 88 , Val 89 , and Leu 93 . The N-terminal resi-  RegX3 (three-dimensional domain-swapped elements in red, the remainder in blue) displays a geometry representing the activated state and is similar to the activated structure of PhoB (magenta). The inactive conformation of PhoB (green) differs in the orientation of the threonine and tyrosine residues, which adopt a conformation pointing away from the active site indicated by the conserved aspartate (Asp 52 in RegX3). For better visualization, the domainswapped element has been clipped from residue 82 onward.
dues of each ␣4 helix play important roles in completing the active site of their own molecule.
Interdomain Interface-We suggest that the model of RegX3 provides the first high resolution structural data of the interdomain contacts occurring between the regulatory and the effector domains for a member of the OmpR/PhoB subfamily in an activated state. The interaction is formed by ␣5 from the composite receiver domain with the ␤7␤8-loop and C terminus of ␣7 from the effector domain. This interface is the second subinterface of the three-dimensional domain-swapped dimer and buries a total surface area of 1188 Å 2 /monomer. This is significantly larger than that in previously characterized full-length RRs, PrrA (820 Å 2 ), MtrA (738 Å 2 ), DrrB (751 Å 2 ), and DrrD (245 Å 2 ).
The RegX3 interdomain interface is most similar to the interface of DrrD in that it involves the face of ␣5. However, unlike DrrD and the other structures, RegX3 exhibits a large linker region of 11 residues between ␣5 and ␤6, which is involved in many interactions that strengthen the interface. Two hydrogen bonds are formed from the carbonyl oxygen atoms of Asp 121 and Ser 124 to the N atoms of Arg 167 (␣6). The other two hydrogen bonds occur between the carboxyl oxygen atoms of Glu 125 and the N atoms of Arg 181 (␣7). This extended linker region may function as a "clasp" to fix the orientation of the effector domain by coordinating with two of the three helixes that form the DNA-binding domain. However, in the RegX3 crystal structure, the "clasp" (residues 123-129) is distorted by two La 3ϩ ions at a crystal contact (Fig. 4C). The ions La-3 and La-4 are located as a bimetallic cluster proximal to ␣1, which interacts with the "clasp" region from the effector domain of a symmetry-related monomer (Ϫy, Ϫx, Ϫz ϩ 1). Both ions display nonadentate geometry and are coordinated by a single residue each from ␣1. La-3 is coordinated by a carboxylate oxygen atom from Asp 15 , and La-4 is coordinated by both carboxylate oxygen atoms from Glu 11 . Five residues from the "clasp" region between ␣5 and ␤-sheets 6, 7, and 8 of the symmetry-related monomer complete the coordination. Two residues, Asp 122 and Asp 123 , bridge the two ions with both carboxylate oxygen atoms coordinating individually with each ion. Asp 128 and three waters complete the coordination sphere for La-3. La-4 is additionally coordinated by a carboxylate oxygen atom from both Asp 121 and Glu 141 and three water molecules (Fig. 4C).
Another contact point between the two domains occurs at the position of La-2, which lies between ␣7 from the effector domain of one molecule and the end of ␣1 from the receiver domain of the second molecule of the dimer. La-2 also displays nonadentate coordination geometry. It is coordinated by a single carboxylate oxygen atom from Asp 186 of the ␣7␣8-loop, both carboxylate oxygen atoms from Asp 180 at the N terminus of ␣8, and both carboxylate oxygen atoms from Glu 24 in the C terminus of ␣1 from the symmetryrelated molecule. Four water molecules complete the coordination sphere (Fig. 4B). As a result of the stabilization of the ␣7␣8loop, a forked salt bridge is formed between Gly 184 from the loop and Arg 113 from ␣5.
The Effector Domain-The wHTH fold contains the structural and sequence elements responsible for DNA recognition. OmpR and PhoB are the founding members of this family. The isolated structures of the OmpR (15) and PhoB (18) effector domains complexed with DNA as well as the uncomplexed but full-length structures of DrrB (22), DrrD (23), and PrrA (21) have all been described in the literature. The RR subfamily differs from other wHTH non-RR proteins in that they have an N-terminal, four-stranded, antiparallel ␤-sheet platform and a large loop between the positioning helix (␣7) and the recognition helix (␣8) called the transactivation loop (14 -16). In RegX3, the ␤-sheet platform is positioned to the side, and the recognition helix is completely accessible (Fig. 7). The isolated effector domains from all of the above structures superimpose quite well on the isolated effector domain from a single monomer of RegX3 ( Table 2). The largest backbone deviations occur in the transactivation loop regions, the ␣7␣8-loop and the ␣8␤11-loop (Fig. 3B).
A single residue in the effector domain (Asn 148 ) is found in the disallowed region of the Ramachandran plot (Table 1). This residue is located at a ␤-hairpin on the ␤8␤9-loop and is involved in main-chain hydrogen bonding but in a region with poor electron density.
Small Angle X-ray Scattering-We believe that the crystal structure of RegX3 reflects an activated RR dimer generated by the presence of lanthanum ions rather than through phosphorylation. Small angle x-ray scattering (SAXS) was used to analyze the protein in solution in the inactive state without the presence of metal ions. The addition of even minimal amounts of lanthanum to the protein solution led to a precipitation of the solute within minutes, and, moreover, the buffer containing La 3ϩ yielded an extremely strong background. These factors made SAXS measurements on lanthanum-containing proteins impossible. RegX3 was therefore analyzed in lanthanum-free solutions at five different concentrations ranging from 2.7 to 15.3 mg⅐ml Ϫ1 (see Fig. 9A). The overall parameters of the solutes (R g , MM, and D max ) were determined for each measurement and showed a clear tendency to increase with concentration (see Fig. 9B). Based upon a comparison of the experimental MM with that determined from the scattering curves and an increasing R g , it is evident that inactivated RegX3 exists as both a monomer and a dimer in solution. A monomer was observed at the lowest concentration (R g ϭ 25.6 Ϯ 0.5 Å, MM ϭ 22.2 Ϯ 2 kDa, D max ϭ 82.0 Ϯ 5 Å) with a dimer evident at the highest concentration (R g ϭ 34.8 Ϯ 0.5 Å, MM ϭ 47 Ϯ 5 kDa, D max ϭ 135 Ϯ 10 Å). Indeed, the latter MM value matches that of dimeric RegX3 (49.7 kDa). Ab initio models of the dimer were constructed from the scattering curve of RegX3 at the highest concentration measured using different programs (DAMMIN and GASBOR) with and without twofold symmetry imposed. The overall shape in solution, derived by GAS-BOR with two-fold symmetry applied, appears linear in a "transshaped" dimer ( Fig. 8B) compared with the "cis-shaped" crystal dimer (Fig. 8A). The elongated transshape is strikingly similar to the proposed model of the inactive fulllength PhoB structure (29), where the positioning of the effector domains in opposite directions provides an additional means of inhibition. We then used, as prior knowledge, the inactive PhoB dimerization interface contacts and the full-length RR DrrD for rigid body modeling against the 15.3 mg⅐ml Ϫ1 RegX3 SAXS data. Several runs of the program SASREF, assuming two-fold symmetry, produced consistent results and a good fit ( ϭ 2.05) to the scattering data (Fig. 9). This model is superimposed with the ab initio envelope in Fig.  8B.
The crystal dimer has two-fold symmetry, which is inconsistent with binding to both parts of a DNA direct repeat, but could conceivably bind to both elements of a DNA inverted repeat. This agrees with recent work on the SenX3-RegX3 system in M. smegmatis (11). Note that the crystal dimer fails to provide a reasonable fit to the SAXS data ( ϭ 9.6) (Fig. 9A).
The scattering data from 4.8 mg ml Ϫ1 solutions of RegX3 yield a molecular mass of 22.2 Ϯ 2.2 kDa, consistent with a monomer (24.8 kDa). Moreover, the scattering curve at this concentration agrees well ( ϭ 0.748) with the computed scattering pattern of DrrD (22) (Fig. 9A). DrrD is a monomeric RR but has a more compact shape when compared with the structure of a single protomer from the crystallographic RegX3 dimer, which does not fit the SAXS data. RegX3, although existing as a monomer at low concentrations, is driven into a dimeric state with increasing concentration. No higher oligomers are present, and the scattering at intermediate concentrations can be fitted well by linear combinations of monomeric and dimeric species ( Table 3). The overall shape determined from the SAXS measurements is dominated by the more accurately measured lower resolution data, and the deviations at higher angles (s Ͼ 0.3 Å Ϫ1 , corresponding to resolutions better than 20 Å) between the experimental data and some of the fits in Fig. 9A may be explained by small scale conformational flexibility or possibly by detailed differences between the monomeric DrrD The Crystal Structure of RegX3 DECEMBER 28, 2007 • VOLUME 282 • NUMBER 52 crystal structure and the RegX3 solution structure. In Table 3, the concentration-dependent equilibrium is quantitatively characterized in terms of the volume fractions of monomers and dimers using the program OLIGOMER. We note that the observed monomer-dimer equilibrium is fully reversible. If a high concentration sample is diluted, the volume fraction of monomers increases accordingly.

DISCUSSION
The crystallization of multidomain RRs has proven difficult, probably due to conformation variability. Based on information from the full-length structures that have been determined, it is evident that (at least) two distinct structural subclasses exist. These subclasses are based upon the accessibility of the DNA recognition helix. The structures of DrrD, DrrB, and now RegX3 all exhibit accessible recognition helices and hence belong to the "open" structural subclass (Fig. 7A). PrrA and MtrA, on the other hand, are very compact structures with inaccessible recognition helices and therefore belong to the "closed" subclass (Fig. 7B). To date, all of the known full-length RRs have been structurally characterized in an inactive state, and the only structural information about the activation mechanism has come from isolated activated receiver domains. In this paper, we have presented the first full-length structure of a response regulator that not only supports the dimerization hypothesis proposed by Toro-Roman et al. (56) but also exhibits three-dimensional domain swapping. Although crystals were not grown in the presence of any activating substances, such as BeF 3 Ϫ , we believe, based upon the position of a number of important residues surrounding the active site, the length of helix ␣4, and the existence of a dimer, that the structure we observe represents RegX3 driven to an active state by the presence of lanthanum ions.
Three-dimensional Domain Swapping-The formation of a three-dimensional domain-swapped dimer requires the breaking of many interactions within the monomer in order for the domains to be swapped. In the case of RegX3, it also requires the partial unfolding of the receiver domain, which must be a process with a relatively high energy of activation. This leads to speculation as to what conditions would be conducive to the formation of a three-dimensional domain-swapped dimer. The majority of three-dimensional domain-swapped dimers in the past have been formed in vitro at high protein concentrations under environmental conditions that favor unfolding. These conditions exist in many crystallization experiments and are a likely reason for the observation of this and other three-dimensional domain-swapped dimers. In RegX3, the partial unfolding may be due to the LaCl 3 present in the crystallization conditions, which is subsequently stabilized in the crystal lattice. The in vivo conditions that lead to three-dimensional domain swapping are still largely unknown.  RRs are DNA-binding proteins, and it is possible that the domain-swapped dimer represents a form able to recognize alternative regulons. Three-dimensional domain swapping has previously been observed in the RR Spo0A (Protein Data Bank code 1DZ3) (57), albeit only for the isolated receiver domain. The result of the domain swapping, however, was a nonfunctional receiver domain with an incomplete active site. In contrast, RegX3 appears to be structurally complete. This observation is based upon the formation of a complete composite receiver domain, a complete active site with all of the residues necessary to stabilize a phosphate group, the formation of the dimer on a highly conserved interface that is strongly linked to dimer formation, and the presentation of the DNA recognition helices in an orientation capable of interacting with DNA. As a result of the three-dimensional domain swapping, two new functional units are created, each consisting of a partly domain-swapped, composite receiver domain and an effector domain. The buried surface area per monomer consists of two parts of roughly equal size and is much greater than that seen in other full-length RRs or in isolated receiver domains. One interface region, the dimerization interface, consists of the elements ␣4 and ␤5, which are exchanged about the two-fold crystallographic axis facilitated by the orientation of ␣4. Strand ␤5 is tightly associated with the domain-swapped four-stranded ␤-sheet from the other protomer, forming an extended fivestranded ␤-sheet. The second interface region, mediated by ␣5, forms the interdomain interface between the composite receiver domain and the effector domain.
Dimerization Interface-Although three-dimensional domain swapping upon activation may not occur in other OmpR/ PhoB subfamily members, the fact that the same structural elements (␣4, ␤5, and ␣5) are also involved in the formation of dimeric interfaces in non-domain-swapped active receiver domains reflects a similar primary structure within the subfamily.
Multiple sequence alignments of members of the OmpR/ PhoB subfamily as well as other RR families reveal that the key residues involved in the ␣4␤5␣5 interface interactions are highly conserved only within this subfamily (4). This suggests that the ␣4, ␤5, and ␣5 are of specific importance to the OmpR/ PhoB subfamily.
The largest variation seen when aligning the composite receiver domain with other characterized receiver domains is in the orientation of ␣4. Its unique orientation positions ␤5 and ␣5 near the protomer's incomplete receiver domain, allowing the formation of the composite receiver domain.
Changes in the length of ␣4 have previously been associated with the activation of the receiver domain. In the active form of PhoB, for instance, the helix extends from Glu 87 to Thr 97 , as opposed to Gly 86 to Gly 94 in the inactive form. Consequently, it adopts a position closer to the other activated receiver domains (29).
This helix in all receiver domain structures, apart from DrrD and RegX3, is positioned parallel to the central five-stranded ␤-sheets. It is proposed that the helix packs alongside ␣5 in the inactive state, thereby shielding the hydrophobic face of the ␤-sheets. In the active state, the helix extends, generating a rotational shift that exposes a hydrophobic face. This newly exposed hydrophobic face forms the active dimerization interface.
In RegX3, the role of ␣4 is slightly different. Although the helix has the same length as an active helix (12 residues), it is mainly involved as a linker between the two receiver domains formed about the crystallographic two-fold axis. Its hydrophobic face packs against the hydrophobic face of the other protomer's ␣4. These differences lead to an alternative dimerization interface, which is evident when the composite receiver domain is superimposed upon other "activated" receiver domains (Fig. 7). The dimerization interface is significantly larger than in other (isolated) active receiver domains due to the existence of two subinterfaces. The ␣4 helix from RegX3 is positioned at an angle of ϳ70°from the face of the central ␤-sheets of the same protomer such that the N-terminus of this helix donates a residue (Glu 84 ) to the active site of that protomer (Fig. 2). The only other RR with an ␣4 helix orientated differently is DrrD, and this is reported to be due to the crystal packing (22).
Switch Residues-The key switch residues in RegX3, as mentioned previously, are Thr 78 and Tyr 98 . Although structures of receiver domains alone in an activated state exist, such as those of PhoB (29), ArcA (4), KdpE (56), TorR (56), and MicA (60), RegX3 represents the first full-length RR structure from the OmpR/PhoB family in what we believe represents an activated state. RegX3 is forced into an active state through its interaction with one of the La 3ϩ ions. La-1 is coordinated by the carbonyl oxygen atom of Tyr 98 , twisting the side chain of the residue into the active state position. The repositioning of Tyr 98 forces Thr 78 to shift inward also into an active state position. This mechanism of a geometric switch is in the reverse order to the normally proposed activation mechanism, where the threonine residue shifts in response to phosphorylation followed by a shift in the orientation of the tyrosine residue.
In other active receiver domains, it has been noted that the tyrosine, in its active position, coordinates through the side chain's hydroxyl group to a carbonyl oxygen atom of an arginine or lysine residue. This coordination is proposed to stabilize the active site (29). In RegX3, Tyr 98 is oriented so that the side-chain hydroxyl group points toward the carbonyl oxygen atom of Arg 81 from the dimeric partner, located toward the N terminus of ␣4. However, the distance (3.9 Å) between the two residues is too large for direct hydrogen bonding. Instead, the hydroxyl group hydrogen bonds a water molecule, which in turn hydrogen-bonds with Thr 100 from ␤5 of the same protomer. In the inactivated full-length RR structures, PrrA, MtrA, and DrrB, the switch tyrosine is oriented away from the receiver domain in an inactive state and forms hydrogen bonds with the effector domain. In the "closed" models (PrrA and MtrA), these bonds occur with the loop region between ␣7 and ␣8 (the DNA recognition helix). In the "open" model, represented by DrrB, the interaction occurs with the ␤7␤8-loop from the ␤-scaffold. DrrD is an exception, since the orientation of the switch tyrosine is apparently influenced by the effect of crystal packing on ␣4 (22). It has been proposed that the inward orientation of the tyrosine would destabilize the interdomain interface between the receiver and effector domains, thereby activating the RR by exposing the DNA recognition helix in preparation for DNA binding (21). The idea of dimerization is strongly attached to RRs of the "open" subclass. This type of RR does not have the option of opening up upon activation, since they are apparently already open and in the case of RegX3 already bound to DNA in an inactive state. Dimerization may allow the recognition of other DNA consensus sequences, and although this speculation still requires investigation, it is consistent with the report of an inverted repeat present in three RegX3-regulated operons in M. smegmatis (11). The RegX3 structure would support this idea, because the switch residue is distant from the interdomain interface and is positioned on the dimerization interface (Fig.  6A). This suggests that the reorientation of the switch tyrosine is associated with a destabilization of the interdomain interface, leading to a reorientation of the effector domain and the creation of the dimerization interface (where the tyrosine is now positioned).
Interdomain Interface-All of the current structures of inactive full-length OmpR/PhoB members exhibit different interdomain interfaces formed by some of the secondary structure elements ␣4, ␤5, and ␣5. The type of interdomain interface appears to be dependent on the subclass of RR. For the "closed" subclass, the interdomain interface involves all three components of the ␣4␤5␣5 surface. However, members of the "open" class consisting of RegX3, DrrD, and DrrB form interfaces on only a single component. In DrrB, the interface involves ␣4, whereas in DrrD and RegX3 the interface involves only ␣5 (Fig.  7A). Interestingly, although either domain from all of the known full-length OmpR/PhoB RRs can be superimposed individually on one another with small r.m.s. deviations, when the full-length proteins are superimposed, it becomes apparent that the two domains have very different spatial orientations (Fig. 7A). In some instances, of course, this could result from crystal packing, although in the case of (inactivated) PrrA, the crystal packing does not determine the interdomain interface (21).
The most noticeable difference between RegX3 and the other "open" inactive RRs is the orientation of the effector domain. Although all of them have an accessible ␣8 DNA recognition helix, DrrD and DrrB mediate the interdomain interface through the ␤-scaffold. In the effector domains of the active, three-dimensional domain-swapped RegX3 dimer, the ␤-scaffold does not interact with the interdomain interface but instead is orientated toward the side in a similar position to the orientation of the scaffold in the "closed" subclass, where the whole effector domain is rotated so as to bury ␣8. The different orientation of the ␤-scaffold in relation to the interdomain interface in the RegX3 structure may represent the active state for the "open" subclass of RR, allowing the two ␣8 helices from the RR dimer to correctly interact with DNA.
The Inactive Dimer-Inactive dimers of OmpR/PhoB response regulators have been reported in the past. Although their function remains largely unknown, it is assumed that they serve an inhibitory function. The dimer in solution was detected by SAXS at a high concentration (15.3 mg⅐ml Ϫ1 ), suggesting that it is a low affinity dimer. Although this concentration may appear unrealistic inside a cell, it may be influenced through intracellular macromolecular crowding. As suggested by Bachhawat et al. (29), the inactive dimer may play a role in regulation. Initially, RegX3 exists as an inactive monomer. When activated, it autoregulates itself, and expression increases until there is an abundance of RegX3 in vivo, whereupon it dimerizes, allowing other regulatory functions. It is possible that when the stress signal (possibly low environmental phosphate concentration (11)) that the SenX3-RegX3 system is responding to subsides and the active kinase is no longer present, the active RegX3 dimers, which are at high concentration, switch to an inactive dimeric form, thus providing a quick attenuation of the response.
The formation of an inactive dimer may inhibit the RR activities in many ways, such as disruption of DNA binding by orientating the effector domains in opposite directions. The dimer may also serve to disrupt or enhance the interaction interface with the SK, or the new dimer may facilitate rapid dephosphorylation by further exposing the active site containing the unstable phosphoaspartate.
We also tried to fit the experimental scattering data from the dimeric species by keeping the three-dimensional domainswapped interface and allowing for the rigid body movements of the peripheral portions of the effector domain. This modeling was able to give extended trans-shapes, which could almost fit the experimental data as well as the inactive PhoB-like dimer shown in Fig. 8B. The latter dimer, however, seems more likely, since the monomer dimer equilibrium is reversible.
Conclusions-Based upon the position of the switch residues and the dimerization interface, we believe that the RegX3 structure represents the RR in an active state. The cis-dimer seen in the crystal structure is the first full-length OmpR/PhoB member RR to be crystallized as a dimer in an apparently active state. The RR dimer formed by three-dimensional domain swapping retains a complete active site as is necessary for phosphorylation. This structure provides the first detailed analysis of the interdomain interface for an active dimeric RR. However, SAXS measurements show that the orientation of the effector domains may not be representative of the dimer in vivo in an inactive state. The SAXS studies revealed that the inactive protein exists as a monomer and a dimer in solution. The inactive trans-dimer formed in solution provides the first structural evidence for a previously hypothesized regulatory mechanism. A structural comparison of the active and inactive RegX3 bound to DNA would provide answers regarding conformational changes of the effector domain and interdomain interface.