Structural Analysis and Solution Studies of the Activated Regulatory Domain of the Response Regulator ArcA: A Symmetric Dimer Mediated by the α4-β5-α5 Face

SUMMARY

Escherichia coli react to changes from aerobic to anaerobic conditions of growth using the ArcA-ArcB two-component signal transduction system. This system, in conjunction with other proteins, regulates the respiratory metabolic pathways in the organism. ArcA is a member of the OmpR/PhoB subfamily of response regulator transcription factors that are known to regulate transcription by binding in tandem to target DNA direct repeats. It is still unclear in this subfamily how activation by phosphorylation of the regulatory domain of response regulators stimulates DNA binding by the effector domain and how dimerization and domain orientation, as well as intra- and intermolecular interactions, affect this process. In order to address these questions we have solved the crystal structure of the regulatory domain of ArcA in the presence and absence of the phosphoryl analog, BeF₃⁻. In the crystal structures, the regulatory domain of ArcA forms a symmetric dimer mediated by the α4-β5-α5 face of the protein and involving a number of residues that are highly conserved in the OmpR/PhoB subfamily. It is hypothesized that members of this subfamily use a common mechanism of regulation by dimerization. Additional biophysical studies were employed to probe the oligomerization state of ArcA, as well as its individual domains, in solution. The solution studies show the propensity of the individual domains to associate into oligomers larger than the dimer observed for the intact protein, and suggest that the C-terminal DNA-binding domain also plays a role in oligomerization.

INTRODUCTION

Rapid responses to changes in the environment are crucial for bacterial competitiveness and survival, as well as for virulence. Many of these responses are coordinated through signaling systems known as two-component signal transduction systems.^1,2 A typical two-component system (TCS) is comprised of a sensor histidine protein kinase (HK) and a response regulator (RR) protein. The HK senses environmental signals that regulate the equilibrium between its kinase and phosphatase activities. This in turn modulates the activity of the RR, facilitating transfer or removal of a phosphoryl group at a conserved aspartate in its regulatory domain. Phosphorylation of the RR induces conformational changes that result in an output response specific to the system. TCSs have not yet been identified in the animal kingdom and some have been shown to be essential for the survival of Gram-positive and Gram-negative pathogens. Therefore they are considered attractive targets for the development of antimicrobial drugs.^3–5

In Escherichia coli, as in other facultative anaerobic bacteria, a complex regulatory network involving several transcription factors mediates adaptation to changes in various respiratory growth conditions.^6,7 The ArcA-ArcB TCS is an important member of these regulatory systems involved in cell growth during anaerobiosis.^8,9 The sensor ArcB is a member of the group of so-called unorthodox HKs that mediate multiple phosphoryl transfer reactions via several cytoplasmic His-Asp-containing domains.^10,11 ArcB is comprised of a histidine kinase core, a central receiver domain and a C-terminal phosphotransfer domain. These catalytic domains are connected to the N-terminal transmembrane region by a putative leucine zipper and a sensing PAS (Per-Arnt-Sim) domain.¹² Quinones have been identified as the redox signal sensed by this TCS¹³ and more recently a model for a quinone-sensitive redox switch in the HK ArcB has been proposed.¹² The RR of this TCS is ArcA¹⁴, a global regulator known to control more than 30 operons involved in redox regulation (the Arc modulon).¹⁵ More recently its been reported that ArcA is involved in the direct regulation of at least 55 other operons implicated in numerous cell functions and that 9% of all open reading frames in E. coli are affected directly or indirectly by phosphorylated ArcA.¹⁶ ArcA, like the majority of RRs, is a transcription factor that, in addition to its N-terminal regulatory domain, contains a C-terminal helix-turn-helix DNA-binding domain. In recent years the ArcA-ArcB TCS has been implicated in the direct or indirect regulation of virulence and resistance in a number of clinically important human pathogens including Salmonella enterica serovar Enteritidis,¹⁷ Haemophilus influenzae¹⁸ and Vibrio cholerae.¹⁹

ArcA is a member of the OmpR/PhoB subfamily of RR transcription factors, by far the largest subfamily of RRs, accounting for 15 of the 34 RRs presently identified in E. coli.^20,21 RRs of this subfamily are distinguished by a C-terminal winged helix-turn-helix DNA-binding motif^22,23 and the most extensively characterized members of this subfamily are known to bind as tandem dimers to DNA direct repeat recognition sequences.^24–26 Phosphorylation of the regulatory domain enhances DNA binding, ultimately resulting in regulation of transcription, but a detailed mechanism of this process has not been elucidated.

The crystal structures of two full-length RRs from other subfamilies, transcription factor NarL²⁷ and methylesterase CheB,²⁸ revealed the recognition helix of NarL and the active site of CheB to be occluded by their regulatory domains in their unphosphorylated states. This suggested a mechanism of regulation in which phosphorylation of the regulatory domain relieved an inhibitory interaction with the effector domain. This has been supported by subsequent biochemical and biophysical studies.^29–32 In contrast, the crystal structures of two full-length RRs of the OmpR/PhoB subfamily from Thermotoga maritima, DrrD³³ and DrrB,³⁴ showed their recognition helices to be completely exposed, making them readily available for DNA binding. The structural data, together with several biochemical studies that point to the significance of dimerization for members of the OmpR/PhoB subfamily,^34–38 are suggestive of a different mechanism for transcriptional regulation.

Details of the phosphorylation-mediated activation of regulatory domains are well understood.³⁹ For the OmpR/PhoB subfamily, as for most RRs, questions remain as to how phosphorylation affects the DNA-binding domain, specifically with regard to domain orientations upon dimer and oligomer formation, and the contributions of both intra- and intermolecular interactions resulting in tandem DNA binding and ultimately in transcriptional regulation. Crystallographic studies of activated full-length RRs might provide insight to some of these questions. However, crystallization of such proteins has so far been unsuccessful, possibly due to intrinsic flexibility of the linker region between the regulatory and effector domains in the active state. Additionally, even though the OmpR/PhoB subfamily is the most represented subfamily in E. coli and other bacteria, no structures of phosphorylated isolated regulatory domains have been reported to date. This has prompted us to pursue structures of isolated regulatory domains in their active forms.

In this article, we report an analysis of the crystal structures of the N-terminal regulatory domain of E. coli ArcA (ArcA_N) in the presence and absence of the previously described phosphoryl analog, BeF₃^−.40–42 This is the first report of the structure of an activated regulatory domain for a member of the OmpR/PhoB subfamily, providing insight into how the regulatory domain regulates the activities of the effector domain, and consequently, activation of transcription. The structures show a symmetric dimer mediated by the α4-β5-α5 face of the protein via the combination of a hydrophobic patch and a network of salt bridges which are both highly conserved exclusively in the OmpR/PhoB subfamily. We propose that most OmpR/PhoB subfamily members are activated by a common mode of dimerization. Additionally, we present solution studies showing that ArcA dimerizes using its N-terminal regulatory domain. The results also indicate a role for the C-terminal DNA-binding domain (ArcA_C) in multimerization and regulation of the oligomerization state of the intact protein.

RESULTS

Structures of ArcA_N and BeF₃⁻-bound ArcA_N

The structure of the regulatory domain of ArcA was determined using unbiased experimental phases obtained from single-wavelength anomalous dispersion data from a single crystal of selenomethionine-derivatized ArcA_N. Phasing information from 8 selenium atoms, corresponding to two molecules in the asymmetric unit, was used to calculate an initial experimental electron density map taking advantage of density modification with two-fold non-crystallographic symmetry (NCS) averaging. After manual building of the initial model the structure was refined using NCS restraints to a final resolution of 2.5 Å. The final model includes residues 2–122 of monomer A, residues 2–123 of monomer B and 62 solvent molecules with a final R_cryst of 0.213 and R_free of 0.256 (Table 1). For ArcA_N-BeF₃⁻, crystals of ArcA_N were soaked for 9 h in a solution of mother liquor containing appropriate amounts of BeCl₂, NaF and MgCl₂. The structure was solved by molecular replacement using an alanine model of ArcA_N lacking the β4-α4 and β5-α5 loops. After density modification, using NCS averaging, inspection of the resulting solvent-flattened electron density map showed clear strong density for the BeF₃⁻ and Mg²⁺ ions at the active site in both monomers. This was later corroborated by inspection of difference maps during the initial stages of refinement. The final model was refined using NCS restraints to a resolution of 2.15 Å and includes residues 2–122 of monomer A, residues 1–122 of monomer B and 124 solvent molecules with a final R_cryst of 0.187 and R_free of 0.212 (Table 1).

Table 1.

Data collection and refinement statistics

Data set	Native ArcAN	SeMeta ArcAN	SeMet ArcAN-BeF3−
Data collection
Space group		P6522
Resolution limits (Å)	30.0-2.50 (2.59-2.50)	30.0-2.50 (2.59-2.50)	30.0-2.15 (2.23-2.15)
Wavelength (Å)	0.97139	0.97896	0.97176
Unit-cell parameters:
a = b, c (Å)	86.70, 214.71	87.04, 214.09	86.17, 213.08
No. of reflections	17,085	16,949	26,280
Completeness (%)	98.3 (95.6)	97.4 (82.9)	100 (100)
Rsymb	0.10 (0.54)	0.10 (0.45)	0.11 (0.53)
I/σ (I)	19.1 (2.9)	26.7 (5.2)	18.8 (3.7)
Refinement
Resolution limits (Å)	30.0-2.50 (2.56-2.50)	-	30.0-2.15 (2.21-2.15)
No. of protein monomersc	2	-	2
No. of protein atoms	1,921	-	1,923
No. of solvent molecules	69	-	124
No. of reflections (work/test)	16,212/864	-	23,565/2,641
Rcryst/Rfreed	0.213/0.256	-	0.187/0.212
rmsd bond length (Å)	0.017	-	0.017
rmsd bond angle (°)	1.615	-	1.445
Mean temperature factor (Å2)	42.62	-	29.17

Values in parenthesis correspond to the highest resolution shell.

^aSelenomethionine-substituted protein.

^bR_sym = Σ|I_obs − I_avg|/ΣI_avg

^cProtein monomers in the asymmetric unit.

^dR factor = Σ||F_obs|_(hkl) − |F_calc|_(hkl)|/Σ|F_obs|_(hkl)

The overall structure of ArcA_N follows the signature CheY-like α/β fold architecture of regulatory domains shared by all RRs.⁴³ The two molecules in the asymmetric unit form a symmetric dimer mediated by the α4-β5-α5 face of the protein (Fig. 1). In the crystal structures, ArcA_N and ArcA_N-BeF₃⁻ are almost identical, except for differences at the active site and the β4-α4 loop, with the two dimers having an rmsd of 0.553 Å for all Cα atoms (Fig. 2(a)).

Figure 1.

Ribbon diagram and surface representation of the BeF₃⁻-activated regulatory domain of ArcA. The protomers that make up the dimer are shown in gold and teal. The two-fold rotationally symmetric dimer is mediated by the α4-β5-α5 face, which buries 850 Ų of surface area at the interface. Side chains of Asp54 and Tyr100 (gray and red), BeF₃⁻ (magenta) and the catalytic Mg²⁺ ion (green) are shown as ball-and-stick models. Figures were created using Pymol (http://www.pymol.org/).

Figure 2.

(a) Superposition of the Cα backbone of ArcA_N-BeF₃⁻ (teal) versus that of ArcA_N (yellow). Circles identify the β4-α4 loops, which are found to be in different conformations in the two dimers. (b) Superposition of ArcA_N-BeF₃⁻ versus MicA_N, an OmpR/PhoB subfamily member from S. pneumoniae (red, PDB accession code 1NXW).⁴⁹

Although crystals grown in the presence of BeF₃⁻ were obtained, we faced difficulties with successfully reproducing crystals of this form sufficiently large for data collection. To overcome this problem we soaked apo-ArcA_N crystals in solutions containing BeF₃⁻. Based on several factors, we consider both crystal structures to be mimicking the active dimer conformation of ArcA_N. First, crystallization screens using protein solutions containing either inactive or BeF₃⁻-activated ArcA_N were set up in parallel and, in most cases, crystals of identical morphology grew with either sample. Preliminary X-ray diffraction analysis showed that crystals from both samples had the same unit cell parameters. Second, crystals soaked in BeF₃⁻ did not crack, as might be expected in the case of a domain undergoing structural changes upon activation. This suggested that no marked changes occurred in the overall conformation of the individual domains, and that the dimer interface observed between the two molecules present in the asymmetric unit was not disrupted upon addition of BeF₃⁻. Finally, addition of BeF₃⁻ to the crystals improved resolution markedly and it appeared to stabilize the molecules as evident from a drop in mean B values for all atoms from 42 Ų to 29 Ų. We attribute this active-like state to a combination of the high protein concentrations and chemical conditions used for crystallization from which a few nucleation events, driven by phase transition, could drive the equilibrium towards an active dimer conformation without the need for phosphorylation. In other words, the conditions required for crystal formation artificially created an active dimer conformation that would otherwise not be prevalent in solution under physiological conditions. Because we believe that both structures represent the active conformational state of the regulatory domain, the ArcA_N-BeF₃⁻ structure will be used for subsequent discussions, unless indicated otherwise.

Coordination at the active site and differences in the two structures

ArcA is phosphorylated at Asp54, the residue that corresponds to the conserved phosphorylation site shared by the regulatory domains of RRs.³⁹ In the crystal structure, the BeF₃⁻ is non-covalently bound to Asp54 centered within the active site (Fig. 3(a)). The BeF₃⁻ and Mg²⁺ ions are extensively coordinated to residues within the active site (Fig. 3(b)). In addition to binding Asp54, BeF₃⁻ also participates in a hydrogen bond and salt bridge with Thr81 and Lys103, respectively. Other atoms that further stabilize the BeF₃⁻ include the catalytic Mg²⁺ ion and the backbone nitrogens of Ile55, Asn56 and Gly82. The octahedrally coordinated Mg²⁺ ion binds to a fluorine of BeF₃⁻, the side chain oxygens of Asp11 and Asp54, the carbonyl oxygen of Asn56 and two water molecules which are further stabilized by interactions with the side chain oxygens of Glu10, Asp11 and Glu12. Other signature residues involved in rearrangements associated with phosphorylation involve the movement of the conserved residues Ser/Thr and Phe/Tyr located in the β4-α4 loop and β5, respectively (Thr81 and Tyr100 in ArcA).^2,44 In the ArcA_N-BeF₃⁻ structure, Thr81 is oriented toward the active site to coordinate the BeF₃⁻ while Tyr100 adopts an inward position (Fig. 1) forming a hydrogen bond with the main chain carbonyl oxygen of Arg83 situated at the middle of the β4-α4 loop. Similar interactions have been reported for many structures of the activated regulatory domains, such as of Spo0A,⁴⁵ FixJ⁴⁶ and CheY.^47,48

Figure 3.

Stereo views of the active site of ArcA_N-BeF₃⁻. (a) Close up view of the active site showing the Fo-Fc electron density for the BeF₃⁻ (magenta), Mg²⁺ (green) and two water molecules (red spheres). The Mg²⁺ ion is octahedrally coordinated to Asp54, Asn56, Asp11, BeF₃⁻ and the two water molecules. The difference density map was calculated with the occupancies of the BeF₃⁻, Mg²⁺ and two water molecules set to zero and contoured at 3 σ. (b) Extended view of the active site. The BeF₃⁻ moiety, in addition to serving as a ligand for Mg²⁺, makes contacts with the side chain oxygen of Thr81, the Nζ of Lys103 and the backbone nitrogens of Gly82, Ile55 and Asn56. Tyr100 stabilizes the β4-α4 loop by forming a hydrogen bond with the carbonyl oxygen of Arg83. Three side chain atoms of Asn56 have been removed for clarity. Carbon, oxygen and nitrogen atoms from the protein chain are shown in gray, red and blue, respectively.

The area that differs most between the two structures is the β4-α4 loop, specifically residues Thr81, Gly82 and Arg83 (Fig. 2(a)). In the ArcA_N structure Tyr100 is in an inward position, but Arg83 is too far away to make the hydrogen bond seen in the presence of BeF₃⁻. When comparing ArcA_N to ArcA_N-BeF₃⁻ it is apparent that binding of BeF₃⁻ promotes movement of Thr81 and Gly82 into the active site to hydrogen bond to BeF₃⁻. This in turn brings the backbone carbonyl oxygen of Arg83 to within hydrogen bond distance of the side chain hydroxyl group of Tyr100, further stabilizing the β4-α4 loop in a fixed conformation. In the absence of interactions with Tyr100 and BeF₃⁻ the β4-α4 loop appears to be very flexible as evident from the poor density for these residues in the ArcA_N structure. Average B values for all atoms in residues Thr81, Gly82 and Arg83 in the ArcA_N and ArcA_N-BeF₃⁻ structures are 69 and 31 Ų, respectively, further indicating stabilization of the β4-α4 loop upon BeF₃⁻ binding.

Dimer interface

The asymmetric unit of the crystal is composed of a two-fold rotationally symmetric dimer mediated by the α4-β5-α5 face of the protein (Fig. 1). The total surface area buried at the dimer interface is 850 Ų. Analysis of the interface reveals the presence of a large number of intra- and intermolecular interactions involving hydrophobic and polar residues. The α4 helix of one monomer packs against the α5 helix of the other monomer through a hydrophobic patch formed between Ile90 (α4), Leu93 (α4), Ile112 (α5) and the aliphatic portions of side chains Lys89 (α4), Glu109 (α5) and Arg113 (α5) (Fig. 4(a)). The interface is further stabilized by a number of side chains involved in bifurcated salt bridges, creating an extensive network that surrounds the hydrophobic packing between the α4 and α5 helices, especially toward the center of the interface. The intermolecular salt bridges are formed between Lys89 (α4)-Glu109 (α5) at the top, Asp99 (β5)-Arg113 (α5) at the center, Glu94 (α4)-Arg115 (α5) at the outer sides and Asp98 (α4-β5 loop)-Arg120 (α5) at the bottom of the interface as depicted in Figure 4(b). Additionally, Lys89 (α4) is stabilized by an intramolecular salt bridge with Glu86 (α4). In the ArcA_N structure, this interaction is disrupted at one of the Lys89 (α4) residues by a hydrogen bond to a nearby water molecule. Similarly, the guanidinium group of Arg120 (α4) is further stabilized by two hydrogen bonds to the carbonyl oxygen of Asn74 (α3-β4 loop). Another intermolecular polar interaction is seen at the bottom of the interface between the amide side chain nitrogen of Asn116 (α5) and the carbonyl oxygens of Leu93 (α4) and Ala97 (α4-β5 loop). These polar interactions are coupled to the Glu94 (α4)-Arg115 (α5) salt bridge by the amide side chain oxygen of Asn116 (α5) which interacts with the guanidinium group of Arg115 (α5). Because of the two-fold symmetric nature of the packing, each of these interactions mentioned above occurs twice, resulting in an extensively coordinated interface.

Figure 4.

Intermolecular interactions at the α4-β5-α5 dimer interface. (a) The core of a hydrophobic patch (spheres) that brings helices α4 and α5 together is formed between Ile90 (α4), Leu93 (α4) and Ile112 (α5). (b) The interface is further stabilized by an extensive network of salt bridges (ball-and-stick models, yellow dotted lines) formed between Lys89 (α4)-Glu109 (α5), Asp99 (β5)-Arg113 (α5), Glu94 (α4)-Arg115 (α5) and Asp98 (α4-β5 loop)-Arg120 (α5). Protomers are distinguished by colors gold and teal.

The structure of ArcA_N is strikingly similar to that of the regulatory domain of MicA (RR02), an OmpR/PhoB subfamily RR from Streptococcus pneumoniae, recently reported by Bent et al.⁴⁹ In the unphosphorylated apo-structure, they found an extensive dimer interface created by a two-fold crystallographic axis and suggested that it represented of the active dimer conformation. The ArcA_N and MicA_N dimers superimpose very well with an rmsd of 1.265 Å for all Cα atoms (Fig. 2(b)). The interface of MicA_N contains analogous interactions found in ArcA_N such as the intermolecular salt bridges formed by Asp99-Arg113 and Asp98-Arg120, the intramolecular pairing between Glu109 and the Asp99-Arg113 salt bridge, as well as the hydrophobic patch. However, the other salt bridges seen in ArcA_N between Lys89-Glu109 and Glu94-Arg115 do not occur between the analogous residues of MicA_N. This is not surprising giving the fact that these intermolecular salt bridges, unlike the other interactions in the interface, occur at the periphery of the protein making these residues prone to interact with symmetry related molecules formed by crystal packing or susceptible to solvation. In fact, the analogous residues of the Glu94-Arg115 salt bridge are involved in lattice contacts in MicA_N, while a water molecule is found bridging the interaction between the analogous residues of the Lys89-Glu109 salt bridge. In addition to the structure of ArcA_N reported here, X-ray structures have been determined for the activated regulatory domains of E. coli PhoB, PhoP, TorR, KdpE, and T. maritima DrrB and DrrD, all members of the OmpR/PhoB subfamily (manuscripts in preparation). In all cases, the regulatory domains exist as dimers with the interface mediated by the α4-β5-α5 face suggesting a common dimerization mechanism upon activation.

The recurrence of this dimer interface reflects similar primary structure within the OmpR/PhoB subfamily. A multiple sequence alignment of the regulatory domains of all members of this subfamily from E. coli,²⁰ in addition to subfamily members from other organisms for which structures of the regulatory domain are available,^33,34,49,50 revealed that the residues involved in key dimer interface interactions are highly conserved within this subfamily (Fig. 5). Moreover, an alignment of the α4-β5-α5 sequences from members of other subfamilies of RR transcription factors from E. coli revealed a lack of similarity with the α4-β5-α5 sequences of the OmpR/PhoB subfamily (data not shown). These sequence analyses of E. coli RRs provide evidence for a subfamily-specific common mechanism of dimerization that appears to be unique to members of the OmpR/PhoB subfamily.

Figure 5.

Sequence alignment of the α4-β5-α5 face of members from the OmpR/PhoB subfamily from E. coli and other organisms for which structures of the regulatory domain are available. Sequence numbering is for ArcA. Abbreviations are as follow: E.c., E. coli; T.m., T. maritima; S.p., S. pneumoniae; B.s., Bacillus subtilis. Residues mediating the dimer interface of ArcA_N-BeF₃⁻ are highly conserved in this subfamily. Residues involved in formation of the hydrophobic patch and the intermolecular salt bridges are highlighted in blue and gold, respectively. A red highlight represents residue pairs that are not conserved but could complement each other at the interface. The alignment was performed using sequence data corresponding to the complete regulatory domains. Members used for the alignment are E. coli ArcA (sp: P03026), TorR (sp: P38684), KdpE (sp: P21866), PhoB (sp: P08402), PhoP (sp: P23836), OmpR (sp: P03025), CpxR (sp: P16244), RstA (gb: AAC74680), BasR (sp: P30843), QseB (gb: AAC76061), CusR (gb: AAC73672), PcoR (sp: Q47456), YedW (gb: AAC75035), BaeR (sp: P30846), and CreB (sp: P08368); T. maritima DrrD (gb: AAD35484) and DrrB (gb: AAD35220); S. pneumoniae MicA (emb: CAB54568); B. subtilis PhoP (sp: P13792). Accession numbers are from GenBank (gb), Swiss-Prot (sp), NCBI Reference Sequence (ref) and EMBL Data Library (emb). Members for which a structure is available are E. coli PhoB (PDB accession code 1B00),⁶⁴ T. maritima DrrD and DrrB (1KGS and 1P2F, respectively),^33,34 S. pneumoniae MicA (1NXW),⁴⁹ and B. subtilis PhoP (1MVO).⁵⁰

ArcA phosphorylation and oligomerization analyzed by liquid chromatography

Members of the OmpR/PhoB subfamily are known to form dimers or oligomers upon activation or DNA binding. It has been reported that full-length ArcA exists as a dimer in solution in its inactive state and forms an octamer upon phosphorylation, with a 1:1 ratio of unphosphorylated to phosphorylated protein.⁵¹ We have further examined the behavior of ArcA in solution by analyzing the contributions of the individual domains to oligomerization. First, we examined the phosphorylation state of ArcA and ArcA_N in the presence of the small molecule phosphodonor phosphoramidate⁵² by using reverse-phase high performance liquid chromatography (HPLC) to separate the unphosphorylated and phosphorylated proteins. Figures 6(a) and (b) show the HPLC profiles of ArcA and ArcA_N, respectively. In both cases, most of the protein in the sample is phosphorylated in the presence of phosphoramidate with a small percentage of ArcA (11%) remaining unphosphorylated, suggesting that ArcA and ArcA_N can be fully phosphorylated in vitro. This contrasts with the 1:1 ratio of unphosphorylated and phosphorylated ArcA reported previously. ArcA and ArcA_N were subjected to analytical size-exclusion chromatography in the presence and absence of phosphoramidate to detect changes in their oligomerization states resulting from phosphorylation (Fig. 6(c) and (d), respectively). The analyses show that both ArcA and ArcA_N change their oligomeric states as a result of phosphorylation, migrating with earlier retention times suggestive of either a monomer to dimer or dimer to tetramer transition. The results also suggest that the phosphorylated regulatory domain alone is sufficient to promote dimerization.

Figure 6.

Analysis of ArcA phosphorylation and oligomerization. Unphosphorylated and phosphorylated proteins are represented by solid and dotted lines, respectively. Phosphoramidate was used as a small molecule phosphodonor. (a and b) Reverse-phase HPLC analysis of the phosphorylation of ArcA (a) and ArcA_N (b) at loading concentrations of 128.7 μM and 3.7 μM, respectively. The ability of ArcA and ArcA_N to completely phosphorylate in vitro was evident from the observed shifts in migration in the presence of phosphodonor. (c and d) Analytical size-exclusion chromatography of unphosphorylated and phosphorylated ArcA (c) and ArcA_N (d). Earlier retention times of the phosphorylated species are consistent with either a monomer to dimer or dimer to tetramer transition. The following molecular weight standards were used for calibration (c) of the size-exclusion column as described in the Materials and Methods: albumin (66 KDa), ovalbumin (44 KDa), trypsin inhibitor (21 KDa), and RNAse A (13.7 KDa). Loading concentrations of 2.2 μM and 128.7 μM were used for ArcA and ArcA_N, respectively.

Analytical ultracentrifugation analysis of ArcA, ArcA_N and ArcA_C

ArcA_N was analyzed by both sedimentation velocity (SV) and sedimentation equilibrium (SE) at concentrations ranging from 36 to 143 μM. Continuous sedimentation (c(s)) distributions of ArcA_N show a single feature at a sedimentation coefficient of ~1.7 S (Fig. 7(a)). Furthermore, there is no change in the position of this peak as a function of concentration, suggesting that the 1.7 S species does not exist in a reversible equilibrium with a higher order species. In addition to SV analysis, ArcA_N was also analyzed by SE at concentrations of 36, 71 and 143 μM (Fig. 7(a), inset). ArcA_N was spun at 27 and 33 K rpm resulting in a total of six data sets that were simultaneously fit to a single species model using the program NONLIN.⁵³ The quality of the fit was good with an rmsd value of 5.9577E-3 and a converged molecular weight value of 13,541 [12,996, 14,072] which agrees well with the theoretical value of 13,989 g/mol (bracketed values represent 95% confidence intervals).

Figure 7.

Continuous sedimentation coefficient (c(s)) distributions and sedimentation equilibrium (SE) profiles of ArcA_N, ArcA_C, and ArcA. (a) c(s) distributions of ArcA_N run at loading concentrations of 36 μM (solid), 71 μM (dashed), 143 μM (dotted) collected using an interference optical system. The c(s) distributions were generated as described in the Methods section. The inset shows the SE distributions, collected using absorbance optics (280 nm), for the 36 μM concentration spun at 27 K (circles) and 33 K (triangles) rpm and the residual plots of the fits to a single species model. (b) c(s) distributions for ArcA at 18 μM (solid), 37 μM (dashed), and 73 μM (dotted) collected using interference optics. The inset shows the SE profiles for the 9 μM concentration spun at 27 K (circles) and 33 K (triangles) rpm and the residual plot of the fits to a monomer-dimer model. (c) ArcA_C c(s) distributions run at loading concentrations of 20 μM (solid), 41 μM (dashed), and 81 μM (dotted) collected using absorbance optics. (d) ArcA_N-P c(s) distributions for ArcA_N-P at 36 μM (solid), 71 μM (dashed), and 143 μM (dotted) collected using absorbance optics. The buffer for the SV experiments was 25 mM Tris-Cl, 100 mM NaCl, 1 mM DTT and 1 mM EDTA, pH 7.5 while for the SE experiments it was 25 mM Tris-Cl, 100 mM NaCl and 1 mM DTT, pH 7.5. All sedimentation coefficients have been corrected to the density and viscosity of water at 20 °C.

Full-length ArcA was also analyzed by SV and SE at concentrations ranging from 9 to 73 μM (Fig. 7(b)). In contrast to ArcA_N, which shows a single feature on the c(s) distribution that does not vary with concentration, ArcA shows larger features that appear with increasing concentration. This suggests that the ArcA monomer exists in a reversible equilibrium with a higher order structure. These larger features only become apparent at loading concentrations above 20 μM suggesting that the K_d is greater than 20 μM. Because SV does not give a definitive answer to the oligomeric state of ArcA, SE analysis was also conducted at concentrations ranging from 9 to 37 μM. Three different concentrations were run at three different speeds, with the nine data sets fitting well to a monomer–dimer model with a K_d of 435 [297, 657] μM.

ArcA_C was then subjected to SV at loading concentrations ranging from 20 to 80 μM. As can be seen from Figure 7(c) the c(s) distributions of ArcA_C are quite broad, ranging from ~2 S to ~6 S, indicating the presence of multiple species. The concentration dependence of the c(s) profiles suggest that these species are in reversible equilibrium with each other and that irreversible aggregation is not occurring. Given an ArcA_C molecular weight of 12,566 g/mol the largest sedimentation coefficient that an ArcA_C monomer can have, that of a perfect sphere, is ~2.0 S. The corresponding values for dimer, tetramer and octamer are ~3.1, ~5.0 and ~7.9 S, respectively. It is clear from Figure 7(c) that even at loading concentrations as low as 20 μM there are features above the maximum S value for a monomer and a dimer, and that at higher concentrations there are features greater than a tetramer. These data suggest that at high concentrations ArcA_C can exist at least as a tetramer of monomers.

Since there is a great deal of evidence that phosphorylation of the regulatory domain alters the oligomeric state of many RRs, both ArcA and ArcA_N were activated with phosphoramidate and subjected to SV analysis. The half-life of phosphorylation of ArcA is short but can be significantly increased by the removal of divalent cations.⁵¹ Specifically, HPLC analysis of phosphorylated ArcA_N (ArcA_N-P) treated with 50 mM ethylene diamine tetraacetic acid (EDTA) subsequent to phosphorylation indicated >95% phosphorylation after 24 h (unpublished data), longer than the 12 h required for SV analysis. Figure 7(d) shows a c(s) distribution of ArcA_N-P at concentrations ranging from 36 to 143 μM. As in the case of ArcA and ArcA_C there is a shift in the position of the peaks as a function of concentration, suggesting the existence of a reversible equilibrium. Surprisingly, however, the distribution extends to S values greater than a dimer. The SE profiles are complex and cannot be modeled by a simple monomer-dimer-tetramer equilibrium. Phosphorylated ArcA was also subjected to SV analysis, but no conclusive results were obtained. Significant precipitation occurred during the run precluding conclusive interpretation of the results.

DISCUSSION

ArcA oligomerization and functional implications

Modulation of the phosphorylated state of ArcA by ArcB during transition from aerobic to anaerobic growth regulates the activation of genes needed for respiration during anaerobiosis and the repression of genes encoding elements of aerobic metabolism. Previous studies using electrophoretic mobility shift and DNAse I protection assays have shown that phosphorylation of ArcA is required for specific binding to target sites, that broad regions of DNA are protected by ArcA-P suggesting formation of multimers and that this multimerization can occur independently of DNA binding.^51,54,55

Solution studies, both size-exclusion chromatography and sedimentation, indicate that the isolated regulatory domain of ArcA exits exclusively as a monomer. Yet ArcA_N crystallizes as a dimer in both the absence and presence of BeF₃⁻. As described earlier, the two structures of ArcA_N are almost identical to each other and in both of them the residues that have been previously identified to participate in the activation switch assume conformations that have been associated with the activated state of RRs. Thus both crystal structures reflect the active state of ArcA_N. It has been shown that some RRs, such as CheY^56,57 and the regulatory domain of NtrC,⁵⁸ exist in conformational equilibrium between inactive and active states even in the absence of phosphorylation. If this conformational equilibrium exists in ArcA as well then it is possible that the combination of high concentrations, the removal of the C-terminal DNA-binding domain and the conditions required for the formation of crystals might trap the molecules in an active state in the absence of phosphorylation or phosphoryl analogs.

The oligomerization of ArcA that we observe differs at least quantitatively from that reported previously.⁵¹ Based on gel-filtration and glycerol gradient sedimentation analyses, Jeon et al. concluded that unphosphorylated ArcA exists as a dimer and active ArcA exists as an octamer. Our SE and SV results also indicate that unphosphorylated ArcA exists as a dimer, but in equilibrium with a monomer with a K_d ~400 μM. Given this low affinity, the monomeric form of ArcA would be expected to predominate at the concentrations examined by Jeon et al. It is possible that the presence of 10% glycerol in Jeon’s experiments promoted dimerization.

The DNA-binding domain of ArcA also participates in oligomerization. While unphosphorylated ArcA_N exists as a monomer in solution, intact ArcA has some propensity for dimerization. The observation that the DNA-binding domain associates into oligomers greater than tetramers suggests that dimerization of full-length ArcA may at least in part be mediated by the DNA-binding domain and that the unphosphorylated regulatory domain may inhibit higher order oligomerization. Both size-exclusion chromatography and sedimentation data indicate that phosphorylation promotes formation of large oligomers. Aggregation of phosphorylated intact ArcA has precluded sedimentation analysis of its oligomerization state. While in vitro studies suggest that oligomerization is likely to play an important role in transcriptional regulation by ArcA, the details of ArcA associations in vivo are likely to be at least quantitatively different and substantially influenced by intracellular conditions, localization, and the presence of DNA and other protein components of transcription complexes.

The α4-β5-α5 face

The α4-β5-α5 face of the regulatory domain of RRs is the region of greatest phosphorylation-induced conformational perturbation. It is not surprising therefore that many RRs utilize the α4-β5-α5 face for regulatory protein-protein interactions, the affinities of which are altered by phosphorylation. These interactions form the basis for a variety of different inter- and intramolecular regulatory mechanisms such as the binding of CheY-P to FliM⁵⁹ and CheZ,⁶⁰ the formation of homodimers of phosphorylated FixJ,⁴⁶ the alternation between different dimeric states in DctD,^61,62 and the inhibitory contacts between the unphosphorylated regulatory and effector domains of NarL²⁷ and CheB.²⁸

An analysis of residue conservation at the α4-β5-α5 face of the three major subfamilies of RR transcription factors (OmpR/PhoB, NarL/FixJ, and NtrC/DctD) revealed important differences that distinguish the OmpR/PhoB subfamily from the others. The Consurf Server⁶³ (http://consurf.tau.ac.il) was used to map the conservation of residues identified in a multiple sequence alignment of members of each subfamily onto the three-dimensional structure of one of its representative members in an activated state (Fig. 8). The analysis shows that the α4-β5-α5 face is highly conserved in the OmpR/PhoB subfamily whereas it is not in the NarL/FixJ and NtrC/DctD subfamilies. The side chains of these conserved residues are exposed, making them available for intermolecular interactions as are observed in the active structures of ArcA_N. The other face of these proteins, the α1-β2-α2 face, is not conserved in any of the three subfamilies. Residues at the variable α1-β2-α2 face in the OmpR/PhoB subfamily, as well as other nonconserved residues in the regulatory domain, might be involved in other protein-protein interactions specific to individual subfamily members. Such interactions might include, for example, the formation of active higher order oligomers, or the formation of alternate inactive dimer interfaces as seen in E. coli PhoB⁶⁴ and B. subtilis PhoP.⁵⁰

Figure 8.

Consurf⁶³ (http://consurf.tau.ac.il) analysis of the conservation of surface residues of the regulatory domain for members of the three major subfamilies of RR transcription factors (OmpR/PhoB, NarL/FixJ, and NtrC/DctD). Results from ClustalW multiple sequence alignments were used to map the conservation of residues, using the Consurf maximum likelihood method for calculating conservation scores, onto the structures of ArcA_N-BeF₃⁻, Sinorhizobium meliloti FixJ_N-P (PDB accession code 1D5W)⁴⁶ and S. enterica NtrC_N-P (1DC7)⁸² as model members of the OmpR/PhoB, NarL/FixJ and NtrC/DctD subfamilies, respectively. Ribbon diagrams of ArcA_N at the far left represent the face of the protein that is shown (gold) for each model. The coloring scheme for conservation ranges from cyan (variable) to dark red (conserved). The results show a highly conserved α4-β5-α5 face for members of the OmpR/PhoB subfamily, in contrast to the other two subfamilies, consistent with a common mechanism of dimerization upon activation. Aside from the active site, marked by orange ellipses, the α1-β2-α2 face is not conserved in any of the three subfamilies. Members of the OmpR/PhoB subfamily used for the sequence alignment are E. coli ArcA (sp: P03026), PhoP (sp: P23836), PhoB (sp: P08402), TorR (sp: P38684), KdpE (sp: P21866), CpxR (sp: P16244), OmpR (sp: P03025), BaeR (sp: P30846), RstA (gb: AAC74680), CusR (gb: AAC73672), BasR (sp: P30843), CreB (sp: P08368), YedW (gb: AAC75035) and QseB (gb: AAC76061); T. maritima DrrB (gb: AAD35220) and DrrD (gb: AAD35484); S. pneumoniae MicA (emb: CAB54568); Shewanella oneidensis YgiX (ref: NP_717707); Staphylococcus epidermidis SrrA (ref: NP_764731); B. subtilis PhoP (sp: P13792). Members of the NarL/FixJ subfamily are E. coli NarL (emb: CAA33023), FimZ (emb: CAA35973), UvrY (gb: AAC74981), RcsB (gb: AAC75277), EvgA (ref: NP_416870), NarP (gb: AAA16411) and UhpA (gb: AAC76692); S. meliloti FixJ (ref: NP_435915); Agrobacterium tumefaciens TraR (ref: NP_059701); Bacillus licheniformis ComA (gb: AAU24816); Listeria monocytogenes DegU (gb: AAT74538); Bordetella bronchiseptica NodW (ref: NP_887648). Members of the NtrC/DctD subfamily are E. coli AtoC (gb: AAC75280), YfhA (gb: AAC75607) and HydG (gb: AAC76978); S. enterica NtrC (gb: AAV79614); Rhodopirellula baltica SasR (ref: NP_869833); Brucella suis NtrC (gb: AAN30037); Xanthomonas axonopodis NtrC (ref: NP_644040); Helicobacter hepaticus FlgR (ref: NP_861166); Burkholderia pseudomallei DctD (emb: CAH37507); A. tumefaciens NtrC (ref: NP_532136). Accession numbers are from GenBank (gb), Swiss-Prot (sp), NCBI Reference Sequence (ref) and EMBL Data Library (emb).

A common mechanism of regulation by dimerization

A notable feature of the OmpR/PhoB subfamily of winged-helix transcription factors is that the majority of the members that have been experimentally characterized have been shown to bind in tandem to DNA direct repeat half-sites.^24–26 Other RR transcription factors, such as members of the NarL/FixJ subfamily, bind to DNA in a variety of arrangements including inverted repeats with two-fold dyad symmetry.⁶⁵ In E. coli, the OmpR/PhoB subfamily consists of 15 of the 34 RRs encoded by the genome,^20,21 representing 45% of the total. In the OmpR/PhoB subfamily, the α4-β5-α5 face of the receiver domain, used in dimer formation upon activation, shows a high degree of conservation not seen in any of the other subfamilies (see previous section). Moreover, this conservation comes from residues involved in the formation of important intermolecular contacts. Interestingly, in the NtrC/DctD subfamily, which lacks conservation in the α4-β5-α5 face, different members have been shown to use distinctly different surfaces of the regulatory domain for multimerization.⁶¹ Another notable feature of the OmpR/PhoB subfamily is that in the inactive state the DNA recognition helix of the effector domain is freely exposed to the solvent^33,34 making it readily available for DNA binding in contrast to other RRs in which the functional regions of the effector domains are occluded by the unphosphorylated regulatory domains.^27,28 Taking into account these marked differences, it is hypothesized that members of the OmpR/PhoB subfamily use a common mechanism of regulation by dimerization.

It is important to note that the rotationally symmetric orientation of the ArcA_N active dimer is different than the expected head to tail orientation adopted by the C-terminal effector domains bound to direct repeat DNA recognition sites. Furthermore, the α4-β5-α5 face involved in the active dimer interface is the same surface that forms the regulatory–effector domain interface in inactive DrrB.³⁴ If the structure of DrrB is representative of OmpR/PhoB subfamily members, then the tight domain interface mediated by the α4-β5-α5 face of the regulatory domain must be disrupted upon phosphorylation to allow for the separation of the two domains tethered by a flexible linker. This would leave the α4-β5-α5 face of the regulatory domain available for formation of an active symmetric dimer and the effector domain free to bind DNA with a different symmetry. Additionally, the formation of the α4-β5-α5 dimer interface juxtaposes the linkers of the RRs on the same plane, as opposed to the dimers formed in inactive E. coli PhoB⁶⁴ and B. subtilis PhoP,⁵⁰ facilitating association of the effector domains into dimers for DNA binding. While this model is compatible with tandem binding to direct repeats, it does not exclude the possibility of other binding modes such as symmetric binding to inverted repeats as has been postulated for some OmpR/PhoB subfamily members.⁶⁶ If the regulatory domain exerts no constrains on the orientation of the effector domains then their positioning will be determined by the arrangement of DNA recognition sequences and/or the specificity of protein-protein interactions at the interface of effector domain dimers.

MATERIALS AND METHODS

Cloning, expression and purification of ArcA

The gene encoding ArcA was amplified from E. coli DH5α genomic DNA using the polymerase chain reaction (PCR) and subsequently inserted into the pT7 vector pJES307⁶⁷ at the NdeI and BamHI restriction sites to create pEF23. Plasmid pEF23 was then transformed into E. coli strain BL21(DE3) (Novagen). Cells were grown at 37 °C in Luria-Bertani medium supplemented with ampicillin at a final concentration of 100 μg/ml. After reaching mid-log phase, induction of ArcA expression was achieved by addition of 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG) followed by incubation for an additional 3–5 h. All subsequent steps were carried out at 4 °C. Cells were harvested by centrifugation for 20 min at 4,000 × g, washed with 50 mM Tris-Cl, 300 mM NaCl, 1 mM dithiothreitol (DTT), 1 mM EDTA and 1 mM phenylmethylsulfonyl fluoride, pH 8.0, and lysed by sonication in the same buffer. The lysate was clarified by ultracentrifugation for 60 min at 80,000 × g. The soluble portion was retained and a saturated solution of (NH₄)₂SO₄ was added to 55% saturation. The (NH₄)₂SO₄ pellet was resuspended and dialyzed overnight in 20 mM Tris-Cl, 200 mM NaCl and 2 mM β-mercaptoethanol (βME), pH 7.5 (buffer A). The dialyzed ArcA sample was loaded onto two tandem 5-ml HiTrap Blue dye affinity columns (Amersham Biosciences) and eluted with a 100-ml gradient of 0 to 2.0 M NaCl in buffer A. Fractions containing ArcA were pooled and a saturated (NH₄)₂SO₄ solution was slowly added to the solution to ~45% saturation. The sample was then filtered through a 0.45-μm filter and loaded onto a HiLoad phenyl Sepharose 16/10 column (Amersham Biosciences) equilibrated in 20 mM Tris-Cl, 1.0 M (NH₄)₂SO₄ and 2 mM βME, pH 7.8 (buffer B). ArcA was eluted with a 300-ml gradient of 1.0 to 0 M (NH₄)₂SO₄ in buffer B. Fractions containing ArcA were pooled, precipitated with a saturated solution of (NH₄)₂SO₄ to 60% saturation, resuspended in 25 mM Tris-Cl, 100 mM NaCl and 1 mM DTT, pH 7.5 (buffer C) and finally subjected to gel filtration chromatography using a Superdex75 26/60 column (Amersham Biosciences) equilibrated in buffer C.

Cloning, expression and purification of ArcA_N

ArcA_N, encompassing residues 1–123 (N123Q), was PCR-amplified and subcloned from pEF23 into pJES307 at the NdeI and BamHI restriction sites to create pEF29. Plasmid pEF29 was transformed into E. coli BL21(DE3). Cells were grown and lysed as described above for ArcA except that after induction with IPTG the cells were incubated overnight at 20 °C. The soluble portion of the lysate was precipitated with a saturated solution of (NH₄)₂SO₄ to 30% saturation. The (NH₄)₂SO₄ pellet was resuspended and dialyzed overnight in 25 mM Tris-Cl, 200 mM NaCl and 2 mM βME, pH 8.0 (buffer D). The dialyzed protein sample was loaded onto two tandem 5-ml HiTrap Q Sepharose anion exchange columns (Amersham Biosciences) and eluted with a 150-ml gradient of 0 to 0.35 M NaCl in buffer D. Fractions containing ArcA_N were pooled and a saturated solution of (NH₄)₂SO₄ was added until ~20% saturation. The sample was then loaded onto a HiLoad phenyl Sepharose 16/10 column equilibrated in 20 mM potassium phosphate, 0.8 M (NH₄)₂SO₄ and 2 mM βME, pH 7.0 (buffer E). ArcA_N was eluted with a 200-ml gradient of 0.8 to 0 M (NH₄)₂SO₄ in buffer E. Fractions containing ArcA_N were pooled, precipitated with a saturated solution of (NH₄)₂SO₄ to 40% saturation, resuspended in buffer C and finally chromatographed on a Superdex75 26/60 column equilibrated in buffer C.

For preparation of selenomethionine-derivatized ArcA_N (SeMet ArcA_N) plasmid pEF29 was transformed into the methionine auxotroph strain B834(DE3)-pLysS (Novagen). The protocol for cell growth was adapted from Hendrickson et al.⁶⁸ as described elsewhere⁶⁹ with the exception that after induction with 0.5 mM IPTG cells were incubated overnight at 20 °C. SeMet ArcA_N was subsequently isolated and purified following the procedure for the wild-type protein except for the inclusion of 10 mM βME in all solutions.

Cloning, expression and purification of ArcA_C

The effector domain of ArcA, residues 132–238 (R132Q), was PCR-amplified and subcloned from pEF23 into pJES307 at the NdeI and BamHI restriction sites to create pEF34. Plasmid pEF34 was transformed into E. coli BL21(DE3). Cells were grown and lysed as described above for ArcA followed by fractionation with a saturated solution of (NH₄)₂SO₄ to 60% saturation. The (NH₄)₂SO₄ pellet was resuspended and dialyzed overnight in 25 mM Tris-Cl and 2 mM βME, pH 8.0 (buffer F). The dialyzed protein sample was loaded onto two tandem 5-ml HiTrap Blue columns and eluted with a 200-ml gradient of 0 to 1.3 M NaCl in buffer F. Fractions containing ArcA_C were pooled and concentrated and chromatographed on a Superdex75 26/60 column equilibrated in buffer C.

Analytical gel filtration and HPLC

Size exclusion chromatography was performed on a fast performance liquid chromatography AKTA system (Amersham Biosciences) using a Superdex75 10/30 column. Experiments were carried out at 4 °C at a flow rate of 0.5 ml/min using 25 mM Tris-Cl, 100 mM NaCl and 1 mM DTT, pH 7.5 as the mobile phase while monitoring absorbance at 280 nm. Each analysis was performed with a 200-μl aliquot. ArcA_N samples were prepared by incubating 360 μg of protein in mobile phase buffer for 25 min at room temperature in the presence or absence of 50 mM ammonium hydrogen phosphoramidate,^52,70 and 10 mM MgCl₂. Prior to injection, samples were filtered through 0.22-μm filters. Limited solubility of phosphorylated ArcA has been reported before⁵⁵ and this restricted its useful concentration limit to 0.1 mg/ml. Samples containing 12 μg of ArcA in the presence and absence of phosphoramidate and MgCl₂ were prepared as described above for ArcA_N.

Reverse-phase HPLC analysis was carried out using a Hitachi-7000 Series instrument and a 214TP5415 C-4 column (Grace-Vydac). Eluent A contained 0.1% trifluoroacetic acid and 99.9% H₂O; eluent B contained 0.1% trifluoroacetic acid and 99.9% acetonitrile. ArcA (10 μg) and ArcA_N (180 μg) samples in the presence and absence of phosphoramidate were prepared as described above, using the same buffer, in a volume of 100 μl. Chromatography was performed at room temperature with a 40-ml gradient from 35 to 65% of eluent B at a flow rate of 1.0 ml/min. Injected sample volumes for ArcA and ArcA_N were 80 and 40 μl, respectively.

Analytical ultracentrifugation

Sedimentation velocity (SV) and sedimentation equilibrium (SE) experiments were carried out in a Beckman XL-I analytical ultracentrifuge utilizing both interference and absorbance optical systems. The SV experiments were conducted in epon charcoal-filled double sector centerpieces with quartz or sapphire windows (quartz for absorbance collection and sapphire for interference collection) while the SE experiments were done in epon charcoal-filled 6-channel centerpieces with quartz windows. All experiments were done at 4 °C in 25 mM Tris-Cl, 100 mM NaCl and 1 mM DTT, pH 7.5 either with or without 1 mM EDTA (Fig. 7 legend). The buffer density and viscosity, as well as the partial specific volumes for each of the proteins analyzed were determined using the software SEDNTERP.⁷¹ The density and viscosity for the buffer containing 1 mM EDTA are 1.00503 g/ml and 1.5946E-2 poise, respectively, while the density and viscosity of the EDTA free buffer are 1.00485 g/ml and 1.5928E-2 poise, respectively. The partial specific volume at 4 °C for ArcA is 0.7273 ml/g, for ArcA_N is 0.7348 ml/g and for ArcA_C is 0.7203 ml/g. Analysis of the SV data was accomplished using the program SEDFIT⁷² which allowed for the generation of c(s) distributions. Fixing the frictional ratio at 1.2 and floating the meniscus position generated c(s) distributions. The resolution setting for the continuous distributions is 200 and the regularization setting is 0.9. For the SE experiments the data were analyzed using the program WinNonLin.⁵³

Crystallization of ArcA_N

Prior to use, ArcA_N was dialyzed into 50 mM Tris-Cl, pH 8.4, with the addition of 10 mM βME in the case of the SeMet-derivatized protein. After dialysis the protein was concentrated at 4 °C with a Biomax-10 concentrator to about 20–35 mg/ml as determined by absorbance at 280 nm. For BeF₃⁻ activation the concentrated protein sample was mixed with 5.3 mM BeCl₂, 33 mM NaF and 6.7 mM MgCl₂ in a total volume of 200 μl.^40,59 Crystals were grown at 4 °C by the hanging drop vapor diffusion method by mixing equal amounts of protein sample and reservoir solution. Native and SeMet-derivatized protein crystals were grown side by side under the same conditions. Identical crystals of ArcA_N and BeF₃⁻-activated ArcA_N, with the same unit cell parameters, grew in 0.2 M ammonium acetate, 0.1 M tri-sodium citrate dihydrate (pH 5.5) and 15% v/v 2-methyl-2,4-pentanediol. Crystals were difficult to reproduce using protein from different preparations and for data collection the inactive SeMet-derivatized crystals were soaked for ~9 h in a fresh solution of reservoir solution containing BeF₃⁻ in which the citrate buffer was substituted with sodium acetate trihydrate (pH 5.5). Inactive crystals remained stable after transferring them to the BeF₃⁻-containing solutions, even improving resolution. Crystals were cryoprotected by increasing the 2-methyl-2,4-pentanediol concentration to 20% and transferring them to a 100 K nitrogen stream. Native and multiwavelength anomalous dispersion datasets were collected with crystals grown in the absence of BeF₃⁻. A non-anomalous dataset was collected using a SeMet-derivatized ArcA_N crystal soaked in BeF₃⁻. Crystals belonged to space group P6₅22 with average cell constants of a = b = 87.0 Å, and c = 214.0 Å containing 2 molecules per asymmetric unit and an average solvent content of 70%. Data were collected at beamline X4A at the National Synchrotron Light Source at Brookhaven National Laboratory, Upton, N.Y. All data were processed and scaled with DENZO and SCALEPACK.⁷³

Structure determination and refinement

The structure of ArcA_N was solved by the single-wavelength anomalous dispersion technique using the program CNS 1.1.⁷⁴ Eight out of ten possible selenium atom sites for the two molecules in the asymmetric unit were found. Initial phase estimates to 2.5 Å resolution in the form of Hendrickson-Lattman coefficients from CNS in addition to amplitudes from the native data set were input into DM⁷⁵ from the CCP4 4.2.2 program suite⁷⁶ and density modification was performed using solvent flattening, histogram matching and two-fold NCS with matrices derived from the heavy atom sites. The resulting electron density map was of excellent quality and permitted almost a complete model to be manually traced using XtalView.⁷⁷ Multiple rounds of model building and restrained refinement were performed against the native data set with Refmac 5.1.24⁷⁸ using the maximum likelihood target function. Tight NCS restraints were used at the initial stages of refinement and were gradually released until convergence. Waters were initially placed using the ARP_WATERS⁷⁹ routine integrated in Refmac and later added or removed by manual inspection. Electron density for residues Met1-A, Met1-B and Gln123-A was not observed, as well as for side chains of residues Gln2-A, Glu45-A and Arg83-B.

The structure of ArcA_N-BeF₃⁻ was solved by molecular replacement with Phaser 1.2⁸⁰ using a poly-alanine model of ArcA_N lacking the β4-α4 and β5-α5 loops. After rigid body refinement in Refmac, initial iterative cycles of density modification and refinement (solvent flattening, histogram matching and two-fold NCS averaging) were performed automatically to a resolution of 2.15 Å with Refmac and DM as implemented in NCSref from CCP4. The resulting averaged map was of excellent quality, showing clear density for the BeF₃⁻ and Mg²⁺ atoms at the active site, and permitted easy manual reconstruction of the side chains and loops that were removed from the starting model. Iterative rounds of model building and refinement in XtalView and Refmac were performed until convergence. NCS restraints were treated and water molecules were added as in ArcA_N. Electron density for residues Met1-A, Gln123-A and Gln123-B was not observed, as well as for side chains of residues Gln2-A, Met1-B and Glu71-B.

Both structures have good stereochemistry and show no residues in disallowed regions of the Ramachandran plot as defined by PROCHECK.⁸¹ Statistics of structure determination and refinement are shown in Table 1.

Protein Data Bank accession codes

The coordinates and structure factors of ArcA_N and ArcA_N-BeF₃⁻ have been deposited in the Protein Data Bank with ID codes 1XHE and 1XHF respectively.

Abbreviations used

TCS

two-component system

RR

response regulator

HK

histidine kinase

ArcA_N

ArcA regulatory domain

ArcA_C

ArcA effector domain

NCS

non-crystallographic symmetry

HPLC

high performance liquid chromatography

SV

sedimentation velocity

SE

sedimentation equilibrium

PCR

polymerase chain reaction

IPTG

isopropyl-β-D-thiogalactopyranoside

EDTA

ethylene diamine tetraacetic acid

βME

β-mercaptoethanol

DTT

dithiothreitol