Structure of the response regulator VicR DNA-binding domain

The structure of the DNA-binding domain of the response regulator VicR from E. faecalis has been solved at 1.9 Å resolution. It is very similar to the related domains from PhoB and OmpR, but differs in two loops that may affect transcription activation or DNA–protein interactions.


Introduction
VicR is the response-regulator component of the VicKR (YycFG) two-component signal transduction system (TCS) of the infectious agent Enterococcus faecalis (Hancock & Perego, 2002. TCSs are the main mechanism by which bacteria sense and respond to environmental change and they are found in almost all species so far examined (Stock et al., 2000). Typically, they comprise a membranelocated histidine protein kinase (HPK, responsible for environmental signal sensing) and a partner response-regulator protein (RR) that effects the appropriate response (usually activation or repression of specific genes). Extracellular stimuli are sensed by and serve to modulate the phosphorylation activities of the sensor HPK. Under appropriate conditions, the phosphoryl 'signal' is transferred to the cytoplasmic RR protein, which usually results in the activation of a downstream effector domain that elicits the specific adaptive response; for example, increased affinity for target promoter binding and hence gene transcription.
The VicKR system is a member of the YycFG/Vic subfamily that is essential for cell viability (Fabret & Hoch, 1998;Hancock & Perego, 2002;. Only a few members of this important subfamily have been identified and characterized to date, although genome database searches reveal further putative homologues in a large number of Gram-positive species, all of which exhibit high homology with known members (79-90% identity with E. faecalis VicR). Their essentiality makes them attractive targets for novel classes of antibacterial agents. The first member of the YycFG/ Vic subfamily was reported in Bacillus subtilis (Fabret & Hoch, 1998). Phylogenetically, YycFG is similar to PhoRP (Fabret et al., 1999) and these two systems have recently been shown to have regulatory links in this bacterium (Howell et al., 2006). YycFG (Vic) homologues have subsequently been identified in pathogenic Staphylococcus aureus (Martin et al., 1999), Streptococcus pneumoniae (Wagner et al., 2002) and Strep. mutans (Senadheera et al., 2005), as well as in E. faecalis (Hancock & Perego, 2002. In all cases, it is the RR proteins that have been shown to be essential, whilst the HPK components are not essential for all members of the YycFG/Vic TCS subfamily (Lange et al., 1999;Throup et al., 2000).
The target genes regulated by YycF/VicR include those critical for cell-wall biosynthesis or cell-membrane composition (Martin et al., 1999;Howell et al., 2003;Ng et al., 2003Ng et al., , 2005Dubrac & Msadek, 2004;Mohedano et al., 2005), cell division (ftsAZ; Fukuchi et al., 2000), virulence and genetic competence (Wagner et al., 2002;Senadheera et al., 2005;Ng et al., 2005) and biofilms (Senadheera et al., 2005). The DNA sequence recognized by this subfamily of 'essential' regulators is highly conserved amongst species (Dubrac & Msadek, 2004;Ng et al., 2005;Senadheera et al., 2005;Howell et al., 2006). Despite an expanding list of target genes identified for these 'essential' RRs and the elucidation of the crystal structure of the highly conserved N-terminal receiver domain of one homologue, RR02rec of Strep. pneumoniae YycF (Bent et al., 2004), no other structural information is available, including none at all on the effector domains of this essential group of DNA-binding proteins. Here, we present the crystal structure of the putative DNA-binding effector domain of VicR, VicR c .
Overexpression and purification of recombinant VicR c was carried out as described previously (Potter et al., 2002), except that (i) protease-inhibitor cocktail (Sigma-Aldrich, UK) was added prior to cell breakage according to the manufacturer's instructions and (ii) nickel-affinity chromatography included an additional single wash with 100 mM imidazole prior to elution. VicR c was eluted with 200 mM sodium acetate buffer pH 4.0. Purified VicR c was confirmed by N-terminal sequencing (obtaining the expected GSSHHH), SDS-PAGE, Western blotting using HisProbe-HRP (Perbio Science UK Ltd) that detects the hexa-His tag, electrospray mass spectrometry (obtaining 14 031.54 Da, matching the expected 14 031.79 Da) and CD spectroscopy, as described previously (Potter et al., 2002). Protein was purified to 95% as estimated by densitometry and concentrated to 8 mg ml À1 prior to crystallization. Crystals of VicR c were grown by vapour diffusion using the sitting-drop method in 1.8 M ammonium sulfate, 0.1 M sodium citrate pH 4.5, 3% dimethyl sulfoxide. Crystals were transferred to a cryoprotectant buffer consisting of the crystallization buffer with 25% glycerol prior to flash-cooling in liquid nitrogen.

Data collection, structure determination and refinement
Data were recorded at 100 K with an R-AXIS IV ++ image-plate detector mounted on a Rigaku RU-H3R rotating-anode X-ray generator and integrated using MOSFLM (Leslie, 1992) to 1.9 Å resolution. Data reduction and subsequent calculations were carried out using the CCP4 progam suite (Collaborative Computational Project, Number 4, 1994). The crystals belong to space group P4 3 , with unit-cell parameters a = b = 36.24, c = 73.94 Å and one molecule of VicR c per asymmetric unit (Table 1). The structure was determined by molecular replacement using the program Phaser (Read, 2001). Initial search models were based on the C-terminal DNA-binding domains of the response regulatory proteins OmpR from E. coli (Martinez-Hackert & Stock, 1997; PDB code 1opc), PrrA from Mycobacterium tuberculosis (Nowak et al., 2006; PDB code 1ys6), DrrB from Thermotoga maritima (Robinson et al., 2003; PDB code 1p2f) and PhoB from E. coli (Blanco et al., 2002; PDB code 1gxq), with sequences having approximately 25, 35, 33 and 34% identity to VicR c , respectively. Each of these initial models individually failed to produce a clear molecular-replacement solution, which was only found from Phaser when the molecular-replacement model was constructed using an ensemble of these four structures. Prior to the molecular-replacement search, the four homologous structures were superimposed using SSM superposition (Krissinel & Henrick, 2004) in Coot (Emsley & Cowtan, 2004) and regions showing large deviations were removed. Model building and refinement were carried out using Coot and REFMAC5 (Murshudov et al., 1997), allowing the deleted regions to be rebuilt. In the later stages of refinement, TLS parameters (Winn et al., 2001) based on a single-group TLS model calculated from the TLS Motion Determination server (http:// skuld.bmsc.washington.edu/~tlsmd/) were refined in REFMAC5, leading to a final R factor and R free of 17.9 and 23.6%, respectively.

Results and discussion
In the refined structure, the 2F o À F c electron-density maps revealed poorly ordered density at the N-and C-termini of VicR c . 18 residues at the N-terminus were not visible in the density [including the sixresidue His tag and five of the six residues (LVPRG) of a thrombin cleavage site originating from the plasmid vector] and the two C-terminal residues Gln233 and Glu234. The model of VicR c contains 101 residues, starting with Ser-His-Met from the plasmid vector at the N-terminus, followed by Thr135-Glu232 of VicR c . The electrospray mass spectrum of the VicR c sample used for crystallization gives a molecular weight of the protein that is consistent with the expected 121 amino-acid residues.  Residues in most favoured regions 87.5 Residues in additional allowed regions 12.5 † R free was calculated with 5% of the reflections set aside randomly.

Overview of the VicR c structure
The structure of VicR c is highly similar to the DNA-binding domains of other members of the OmpR/PhoB subfamily. It comprises an N-terminal four-stranded antiparallel -sheet (residues 134-136, 139-141, 147-150 and 153-156), a central three -helical bundle (1, 2 and 3; residues 159-170, 178-185 and 194-208, respectively) and a C-terminal -hairpin turn (residues 210-215) (Fig. 1) and belongs to the winged helix-turn-helix family of DNAbinding motifs. On DNA binding, the recognition helix 3 would be expected to occupy the major groove and to interact with bases of the consensus sequence. The loop linking 2 and 3 has been designated the -loop in OmpR and the transactivation loop in PhoB, where it probably interacts with the 70 subunit of RNA polymerase and is essential for transcription activation (Kim et al., 1995;Makino et al., 1996).

Comparison with the OmpR/PhoB subfamily
VicR c superimposes closely on representative members of the OmpR/PhoB family, OmpR, PhoB, PrrA, DrrB and DrrD (Buckler et al., 2002; PDB code 1kgs), with r.m.s. d. values of 1.5, 1.1, 1.2, 1.3 and 1.4 Å , respectively, when using C atoms only. Fig. 2 only shows the superposition of VicR c on OmpR and PhoB, with the other structures omitted for clarity. Although the sequence identities of VicR c with OmpR and PhoB are low (25 and 32%, respectively), the overall structures are almost identical. This is perhaps not surprising as the conserved residues play a key role in linking the secondary-structure elements responsible for DNA recognition (Blanco et al., 2002). The differences arise mainly for two loop regions: the -loop and the loop region following 3 and linking it to the C-terminal -hairpin (Fig. 2). Both these loops were removed when the ensemble model was prepared for molecular replacement.
The positioning of the -loop of VicR c closely matches that of the transactivation loop of PhoB, but not the equivalent -loop of OmpR, which is positioned differently (Okamura et al., 2000) and which, in common with B. subtilis PhoP, has a longer loop and shorter 3 than PhoB (Chen et al., 2004). Both the OmpR and PhoB loops are postulated to interact with RNA polymerase for transcription activation. Evidence suggests that PhoB interacts with the 70 subunit of RNA polymerase (Kim et al., 1995;Makino et al., 1996), whilst OmpR might associate with the subunit (Sharif & Igo, 1993;Pratt & Silhavy, 1994;Kondo et al., 1997). Therefore, our structural data here may suggest possible interactions of VicR c with the 70 rather than Structural comparison of the C backbones of VicR c (red), PhoB (green) and OmpR (blue). The -loops and the loops linking 3 to the C-terminal -hairpin are labelled.

Figure 1
Ribbon diagram of VicR c . Helices 2 and 3 form the helix-turn-helix motif, linked by the -loop. All figures were created with PyMOL (DeLano, 2002).

Figure 3
Ribbon representation showing two molecules of PhoB (green) in complex with DNA shown in space-filling representation (Blanco et al., 2002;PDB code 1gxp). VicR c is shown in red, superimposed onto one PhoB monomer. The consensus sequences for YycF/VicR homologues and PhoB targets are also shown, with the C to A/T substitution mentioned in the text indicated by red arrows in each repeat. The filled red arrow indicates the corresponding base pair in the model, which has been coloured magenta and lies close to the recognition helix. The loop region of PhoB does not reach the DNA, while the VicR c loop could make contact with it. the subunit of RNA polymerase in E. faecalis. PhoB molecules bind DNA head-to-tail in a tandem arrangement (Okamura et al., 2000;Blanco et al., 2002), whilst for global OmpR binding is head-to-tail (Harlocker et al., 1995) or head-to-head (Maris et al., 2005). The orientation of VicR c binding to DNA has not yet been investigated. The position of the loop following 3 in VicR c is different in PhoB, but superimposes well with that of OmpR, although its role in PhoB and OmpR is uncertain.

Modelling of the VicR c -DNA complex
The specific DNA target sequence for PhoB, the pho box, was identified as two seven-base-pair direct repeats separated by an ATrich region of four base pairs (Mizuno, 1997). The corresponding consensus sequence for YycF (and presumably also VicR c ) is almost identical, with one base change from C in PhoB to A or T in VicR c at position 5 in the repeat (with respect to the PhoB consensus; Fig. 3). This suggests that the protein-DNA interactions made by PhoB with its consensus DNA should be very similar to those in the VicR c -DNA complex. Fig. 3 shows a model built by superimposing VicR c on the PhoB-DNA complex (Blanco et al., 2002). In PhoB, residues Trp184 (at the C-terminal end of 2) and Arg193, Thr194, His198 and Arg200 (from 3) were all shown to be involved in specific DNA recognition (Okamura et al., 2000;Blanco et al., 2002) and all except His198 are conserved in VicR. In OmpR, Val203, Ser206 and Arg210 of 3 have been shown to be important for target specificity . The equivalent residues in PhoB are Val197, which packs against the CG base pair at position 5, Arg200, which is nearby but not in contact, interacting with the phosphate backbone, and Lys204, which lies at the end of 3 but does not contact the DNA. The corresponding residues in VicR c are Val199, Arg202 and Glu206. The first two are unchanged and Glu206 is too far away from the DNA to make contact, although it is close enough to influence the conformation of Arg202, which lies near the single base pair that differs between the consensus sequences. The following loop in PhoB is too short to contact the DNA, but in VicR c and OmpR it is longer (Figs. 2 and 3) and might play a role in DNA-protein interactions, although this loop has no previously reported association with DNA and whether the positions of these loops differ upon binding to DNA is not yet known.