Crystal Structure of the Pseudomonas aeruginosa Virulence Factor Regulator ^,

Abstract

Virulence factor regulator (Vfr) enhances Pseudomonas aeruginosa pathogenicity through its role as a global transcriptional regulator. The crystal structure of Vfr shows that it is a winged-helix DNA-binding protein like its homologue cyclic AMP receptor protein (CRP). In addition to an expected primary cyclic AMP-binding site, a second ligand-binding site is nestled between the N-terminal domain and the C-terminal helix-turn-helix domain. Unlike CRP, Vfr is a symmetric dimer in the absence of DNA. Removal of seven disordered N-terminal residues of Vfr prevents the growth of P. aeruginosa.

INTRODUCTION

Pseudomonas aeruginosa is a ubiquitous, Gram-negative opportunistic pathogen that is a leading source of morbidity and mortality for individuals with compromised immune systems or cystic fibrosis. P. aeruginosa switches from an environmental organism to a pathogen through the regulation and elaboration of multiple virulence factors. Virulence factor regulator (Vfr) is a central player in a transcriptional cascade that controls this transition.

Vfr is a 24-kDa protein belonging to the winged-helix family of transcription regulators (40). Vfr positively regulates quorum sensing by promoting the transcription of lasR, upregulates transcription of the type III secretion system, represses flagellar gene transcription, and positively regulates twitching motility (1, 2, 9, 41). Vfr binds cyclic adenosine 3′5′ monophosphate (cyclic AMP [cAMP]) in vitro with a dissociation constant of 1.6 μM (39). Deletion of both cytoplasmic P. aeruginosa adenyl cyclases results in a transcription profile similar to a vfr deletion (41). Furthermore, like the strain carrying the vfr deletion, the double adenyl cyclase mutant has reduced virulence compared to the wild-type strain in a mouse pneumonia model (38). Thus, cAMP is a biologically relevant ligand for Vfr in vivo.

The amino acid sequence of Vfr is 67% identical to that of cAMP receptor protein (CRP) from Escherichia coli (40), a long-standing model for the study of transcription regulation and a structurally well-characterized protein (22). When overexpressed, Vfr complements the β-galactosidase and tryptophanase-deficient phenotypes of a crp deletion in E. coli (40). Conversely, CRP expressed in a vfr deletion strain of P. aeruginosa promotes sub-wild-type levels of lasR transcription. However, CRP is unable to rescue the protease- or exotoxin A-deficient phenotypes of the vfr mutant (40). Recent work suggests that this discrepancy might be due at least in part to cAMP-independent regulation at the lasR promoter (15).

Mounting evidence supports a model for CRP function in which a dimer of CRP binds a cAMP ligand in the N-terminal domain of each monomer to achieve the protein conformation needed for high-affinity specific DNA binding and DNA bending that precedes transcriptional activation. CRP demonstrates low affinity for specific DNA sequences, with a K_d of ∼10⁻⁴ M, in the absence of cAMP (16). Its specificity and affinity for its promoter sites increase to 10⁻⁷ to 10⁻⁸ M in cAMP concentrations as low as 3 μM (34) and up 1.0 mM (35). Cytoplasmic cAMP levels in E. coli of ∼10 μM (12) fall within this range. The crystal structure of apo CRP (37) indicates both cis- and trans-interactions with cAMP induce large protein domain rearrangements that are needed to expose DNA-binding helices. The crystal structure of a CRP-DNA complex, including one subunit of RNA polymerase, and a high-resolution electron microscopy reconstruction of CRP in the context of DNA and holo-RNA polymerase indicate two cAMP molecules in the active, DNA-bound CRP dimer (3, 19). In addition to the high-affinity cAMP sites in the N-terminal domain of CRP, a distinct low-affinity cAMP site has been identified in the C-terminal domain. The biological role, if any, of this secondary site remains enigmatic (17, 25, 29, 33, 42). In Vfr, the number of cAMP binding sites and their specific role(s) in regulation is unknown.

We carried out X-ray crystallography of Vfr in order to examine the structure of an additional member of the cAMP-dependent CRP family and to understand how functional differences from CRP might be structurally encoded. Our results indicate that while the tertiary structure of the monomer and interactions with cAMP are similar from Vfr to CRP, the quaternary assembly of Vfr is poised in a symmetric DNA-binding form in the absence of DNA. In addition, the N-terminal residues may play a unique role in Vfr they do not play in CRP. The importance of these studies to pathogenesis extends beyond P. aeruginosa and the gammaproteobacteria as diverse organisms use cAMP-dependent CRP homologues as global virulence regulators and regulators of natural competence (6, 24, 36).

MATERIALS AND METHODS

Cloning and site-directed mutagenesis of vfr and crp.

The coding sequence for vfr from a P. aeruginosa strain O1 (PA01) genomic library was inserted into pET15b (Novagen), using BamHI and NcoI restriction sites. Plasmid pvfr-Ndel7, encoding the gene for an N-terminal 7-amino-acid truncation of Vfr (Vfr-Ndel7), was prepared according to manufacturer's instructions via QuikChange mutagenesis (Stratagene).

Expression and purification of Vfr and Vfr-Ndel7.

Vfr for crystallization and rabbit polyclonal antibody production was overexpressed from pET15b-vfr in BL21(DE3) cells carrying a crp deletion (13). One liter of cells was grown to an optical density at 600 nm (OD₆₀₀) of 0.6 to 0.7 in lysogeny broth (LB) (4) and induced with 1 mM isopropyl-β-d-thiogalactopyranoside for 3 h at 37°C. After centrifugation, cell pellets were frozen in liquid nitrogen. Thawed cells were resuspended in 50 ml of 80 mM KCl, 50 mM potassium phosphate, and 5% glycerol (pH 7.2) and lysed by incubation with lysozyme, followed by passage through a French pressure cell. The lysate was cleared by centrifugation at 75,000 to 95,000 × g for 90 min, applied to a Biorex-70 cation-exchange resin (Bio-Rad) column equilibrated with the above buffer, and eluted with a 0.08 to 1.0 M linear gradient of KCl. Vfr-containing fractions were pooled and applied to a cAMP agarose resin (Sigma) column and eluted with the buffer described above containing 500 mM KCl and 2 mM cAMP.

Vfr-Ndel7 and Vfr used for biochemical assays were purified by using a similar protocol except that induction was at 20°C for 16 h, no lysozyme was used, and cleared supernatant was applied to PolycatA resin (The Nest Group) column rather than Biorex-70 in the first chromatography step. For Vfr-Ndel7, pooled peak fractions were applied to a heparin affinity column (GE Healthcare) and eluted with a 0 to 1.0 M linear KCl gradient, while full-length Vfr was further purified using cAMP agarose as described above. All purified proteins were dialyzed against two to four changes of a 500- to 1,000-volume excess of 50 mM potassium phosphate, 300 mM KCl, and 1 mM DTT (pH 7.0). Protein concentrations were determined by using the OD₂₈₀ and the calculated extinction coefficient for Vfr or Vfr-Ndel7. Removal of cAMP was assessed by comparing the OD₂₈₀ to the OD₂₉₆.

Crystal growth, data collection, and structure determination.

Crystals of Vfr-Ndel7 were grown by vapor diffusion from drops containing a 1:1 ratio of Vfr at 7 mg/ml in 50 mM HEPES, 300 mM KCl, 1 mM DTT, and 2 mM cAMP (pH 6.5) and a well solution of 1.5 M sodium malonate and 37.5 mM sodium acetate (pH 5.0) (27). Crystals were vitrified in 3.4 M sodium malonate (18). 120° of data were collected on a Bruker Proteum charge-coupled device using a Microstar X-ray source and processed with HKL2000 (31) (Table 1). The crystals belong to space group I4₁ with unit cell dimensions 70.0 by 70.0 by 147.5 Å. Initial phases were determined by molecular replacement using monomer A from CRP structure 1g6n as the search model in Molrep (7, 32). Subsequent iterative rounds of refinement with CNS and manual fitting using Xfit (5, 28) produced the final high-quality structure (Table 1). The seven N-terminal residues, residues 80 to 84, and the C-terminal amino acid were not well ordered and were not included in the final model. Structure analysis was carried out using the tonal molecular modeling program TimMol (8) and PyMOL (10). Protein interface areas were calculated using the PISA web server (21).

Table 1.

Data collection and structure refinement statistics for Vfr-Ndel7

Parameter	Valuea
Data collection
No. of unique reflections	8,735 (876)
Resolution range (Å)	32–2.8 (2.9–2.8)
Completeness (%)	99.8 (99.5)
Redundancy	5.0 (4.9)
Rsymm (%)	0.095 (0.377)
I/σI	16.5 (3.8)
Wilson B (Å2)	54.0
Space group	I41
Refinement
Resolution range (Å)	30.0–2.8 (2.93–2.8)
R-factor (%)	0.190 (0.251)
Free R-factor (%)	0.271 (0.341)
No. of protein atoms	1,596
No. of hetero atoms	57
Avg B, protein (Å2)	41.0
Avg B, hetero (Å2)	35.8
RMSD bond (Å)	0.009
RMSD angle (°)	1.4
Estimated coordinate error (Å)	0.30
Ramachandran plot most
favorable/outlier (%)	96.5/0

^aValues in parentheses correspond to the high-resolution shell.

cAMP binding.

Binding of cAMP to Vfr was monitored by fluorescence titration with slight modifications of a published protocol in which fluorescence of the dye 8-anilino-1-naphthalene sulfonic acid (ANS) complexed noncovalently with CRP is quenched by binding of cAMP (25). Details regarding these methods are provided in the supplemental material.

Fluorescence anisotropy.

Fluorescence anisotropy experiments were performed in KK200 buffer (50 mM potassium phosphate, 200 mM KCl [pH 7.0]) with 5 nM Texas Red (TexRd)-labeled lasR probe (Integrated DNA Technologies), 80 μg of sheared salmon sperm DNA/ml, and Vfr ranging from 0 nM to 1 μM (prepared via serial dilutions in KK200). The lasR probe is a 30-bp double-stranded oligonucleotide formed by heating and annealing a 5′-end-labeled oligonucleotide (5′ TexRd-CATAAAATGTGATCTAGATCACATTTAAAC) with its unlabeled reverse complement. Samples were incubated for 30 min at 25°C. The data were collected in a Beacon 2000 fluorescence polarization system (Panvera Corp.) and analyzed with nonlinear regression according to the calculations of Lundblad et al. (26).

Electrophoretic mobility shift assay (EMSA).

A double-stranded 30-mer representing the lasR Vfr binding site was formed by annealing single-stranded oligonucleotides (with the same sequence used for fluorescence anisotropy experiments). The probe was end labeled with ³²P using T4 Polykinase (Invitrogen) at 37°C for 1 h and then diluted to 50 counts per s (cps)/μl. Various concentrations of protein were combined with the radioactive DNA, 100 ng of bovine serum albumin/μl, 10% glycerol, 65 μg of sheared salmon sperm DNA/ml, 0.8 mM dithiothreitol, cAMP, and buffer (200 mM KCl, 50 mM Tris [pH 8], 1 mM EDTA). Each 20 μl of protein-DNA mixture contained 0.21 cps of radioactivity/μl. The samples were incubated at 25°C for 45 min and loaded onto 10% polyacrylamide gels, which were run at 4°C and 220 V for 2 h. Running buffers contained the same cAMP concentration used in the binding reaction. Gels were dried and incubated overnight on a Kodak Storage Phosphor Screen sd230 and scanned by using a Typhoon phosphorimager. ImageQuant (GE Healthcare Lifescience) was used to determine the K_d of the protein for the lasR oligonucleotide.

PDB accession number.

Structure factors and coordinates for Vfr are available at the Protein Data Bank under accession number 2OZ6.

RESULTS AND DISCUSSION

Structure of Vfr.

Full-length Vfr yielded crystals unsuitable for high-quality structure refinement, although a 2.8-Å resolution model with an R-value of 31.4% was obtained (see Table S1 and further details regarding methods in the supplemental material). This crystal form contains four independent Vfr monomers, which create two symmetric dimers. In an effort to improve the diffraction through directed protein engineering, we took advantage of the observation that the seven N-terminal amino acids were disordered in three of the four Vfr monomers. Crystals of a truncated Vfr-Ndel7 protein lacking these residues formed under unique conditions with improved diffraction quality (Table 1). A crystallographic 2-fold axis generates a perfectly symmetric dimer of Vfr-Ndel7 in the final well-refined structure (Table 1 and Fig. 1 A). Although our description focuses on the superior Vfr-Ndel7 model, all of the features in the Vfr-Ndel7 structure, except where specifically mentioned below, are consistent with the preliminary full-length Vfr model.

Fig. 1.

X-ray crystal structure of Vfr-Ndel7. (A) Overall, structure of a Vfr-Ndel7 dimer, highlighting the ligand-binding N-terminal domain (blue) and DNA-binding C-terminal domain (red) of the left monomer and showing the relative positions of the cAMP binding sites. Missing residues 80 to 84 in the right monomer (purple dashes) would cross in front of the left monomer. (B) Close-up of the primary cAMP-binding pocket, with cAMP and neighboring residues in stick representation (C, green and gray for ligand and side chains, respectively; N, blue; O, red; P, orange; S, yellow). (C) Close-up of the secondary cAMP binding pocket (coloring as in panel B). F_obs-F_calc omit electron density calculated without ligand in the model, contoured three standard deviations above the mean, reveals cAMP in the secondary pocket. (D) Stabilizing interactions among Vfr structural components.

The N-terminal domain of Vfr, extending from residues 8 to 140, consists of an 8-stranded jelly roll β-sheet motif capped by two short antiparallel helices, αA and αB, and flanked by the long helix αC (Fig. 1A). In the center of this N-terminal domain is the primary bound cAMP in the anti conformation. αC extends from residues 114 to 140, thus connecting the N-terminal domain to the C-terminal helix-turn-helix (HtH) DNA-binding domain (Fig. 1A). As in CRP, this long helix provides the majority of the dimer interface contacts. Dimerization buries 1,571 A² of solvent accessible surface area per subunit. Following αC and αD are strands β9 and β10 before the canonical HtH motif, comprised of αE and αF. Finally, two C-terminal strands β11 and β12 finish a small antiparallel β-sheet begun by β9 and β10. Nestled between the N-terminal domain and the DNA-binding domain is an unexpected but well-ordered second cAMP molecule in the syn conformation. As expected, Vfr and CRP (2CGP) are structurally very similar, with a root mean square deviation (RMSD) of 0.9 Å over 1,120 atoms within the monomer.

The primary cAMP-binding pocket is located in the center of the N-terminal domain (Fig. 1B). A salt bridge from Arg87 and a hydrogen bond from Ser88 form to cyclic phosphate oxygens, and Arg128 packs across the ribose ring. The Thr132 −OH hydrogen bonds to the adenine exocyclic N, and the side chain of Thr133 extends across the dimer interface to form a bifurcated hydrogen bond with the same exocyclic N and adenine N7, providing a mechanism to communicate information on the occupancy of one primary cAMP site with the other Vfr monomer. In CRP this residue is a serine, but it forms the same two interactions. Clear electron density was not detectable for residues 80 to 84, a region that includes a three-residue insertion in Vfr relative to CRP. Thus, this extended β6-β7 hairpin, which links the two segments of the cAMP-binding pocket, is flexible and surface exposed rather than participating directly in cAMP binding.

The secondary ligand-binding pocket is formed largely by residues 57 to 61 at the base of the β4-β5 hairpin on one side and residues 175 to 185 of the αE-αF turn within the HtH motif on the other (Fig. 1C). From the β4-β5 hairpin, the aliphatic portion of Arg59 lies across the adenine ring of cAMP, and the backbone N of Glu60 forms a hydrogen bond with the ribose ring O2. From the HtH, the aliphatic portion of Arg179 extends across the ribose ring and Arg185 forms a salt bridge with the cyclic phosphate. Residues 140 and 141 from the base of the dimerization helix αC of the opposite monomer close off one end of this relatively exposed gully. Although there are many fewer polar bonds to the cAMP in the secondary pocket than in the primary one (4 versus 14), the solvent accessible surface area lost to the cAMP due to binding is nearly the same (340 Ų versus 381 Ų). The secondary pocket is less well conserved in CRP, although it is seen in some structures (33). In particular, equivalent residues to Arg59 and Arg179 in Vfr are Lys57 and Glu174 in CRP.

A network of interactions stabilizes the relative positioning of the Vfr structural elements (Fig. 1D). For example, Met61 in the β4-β5 hairpin is near the backbone atoms of residues 180 to 182 in the HtH and also approaches the side chain of Leu139 at the base of the dimerization helix. Glu60 from the β4-β5 hairpin forms a salt bridge with Arg92 in the N-terminal nucleotide-binding domain. The hydroxyl of Tyr65 from the N-terminal domain hydrogen bonds with the Glu176 carboxylate, which in turn hydrogen bonds to the secondary pocket residue Arg179. These interactions help position the β4-β5 hairpin and provide structural connections among the primary and secondary cAMP pockets, HtH, and dimerization domain. Indeed, a spontaneous mutation in vfr which substitutes His for Tyr65 prevents Vfr function in vivo (14), highlighting the importance of this network for maintaining the activated conformation of Vfr.

cAMP binding.

Since the crystal structure reveals two distinct cAMP binding sites in each monomer, we measured the affinities for cAMP in Vfr via displacement of the fluorescent dye ANS upon cAMP binding (Fig. 2A). When a one-site cAMP-binding model is used to fit the titration data (e.g., cAMP binds either to one primary cAMP site per dimer or to one primary cAMP site per monomer with the monomers having identical affinities at that site), a dissociation constant of 5.5 μM was calculated for cAMP binding to Vfr. A single isothermal titration calorimetry (ITC) experiment was used as an independent verification, and a K_d value of 1.0 μM was obtained for a single binding site model (data not shown). These values are comparable to the 1.6 μM determined by equilibrium dialysis (39). A two-site model was also used to fit the ANS titration data, assuming that a dimer of Vfr has two nonidentical cAMP binding sites (Fig. 2B). Vfr displays positive cooperativity, with a 2.35-fold decrease in the K_d for the second versus the first binding site. Refinements using three or four site models that could account for the binding of cAMP to the secondary ligand-binding pockets in one or both C-terminal domains independently were unstable, or the equilibrium constants refined to 0, possibly due to strong dilution effects in the fluorescence signal at cAMP levels greater than 1 mM. Thus, our in vitro titration data do not provide binding affinities for the lower-affinity cAMP sites identified in the crystal structure nor do these data confirm their functional role. An intriguing but unconfirmed hypothesis is that the secondary ligand-binding site recognizes an alternative small molecule in vivo. We tested binding of cyclic-di-GMP to Vfr by using ITC but found no evidence of interaction either in the presence or absence of cAMP (data not shown), although other CRP homologues are directly regulated by this signaling molecule (23). In keeping with this lack of interaction, there is no effect of c-di-GMP on VFR binding to DNA (15, 23).

Fig. 2.

Vfr binding to cAMP. (A) Fluorescence decreases as ANS is displaced from Vfr by the binding of cAMP. Points are the means of three replicates displayed with standard errors. The curve is the best fit of the data to a two-site cAMP binding model. (B) The predicted abundance of VFR dimers bound to one (□) or two (■) cAMP molecules as the cAMP concentration increases, based on equilibria determined in panel A.

DNA binding.

The Vfr dimer is symmetric in the absence of DNA (Fig. 1) in contrast to published forms of crystalline wild-type CRP which adopt an asymmetric arrangement (32) until bound to DNA (30, 33). The HtH orientation in the Vfr dimer closely matches the HtH position in the activated CRP-DNA structure. In a more distant CRP homologue, PrfA from Listeria monocytogenes, stabilization of the HtH domain in a conformation similar to that of activated CRP increases DNA-binding affinity and in vivo activity (11). Thus, Vfr is poised in a DNA-binding conformation and may require less induced fit than does CRP. To test this model, we used two methods to measure the affinity of Vfr for its 30-bp binding sequence in the lasR promoter (1) in the absence or presence of cAMP. Fluorescence anisotropy binding curves were obtained at several cAMP concentrations by varying the Vfr concentration. Vfr showed cAMP-dependent DNA binding (Table 2), with the affinity of Vfr for the lasR promoter increasing 8- to 25-fold in the presence of 2 μM cAMP. Comparable 10- to 20-fold differences in DNA-binding affinity between liganded and unliganded Vfr were measured by EMSA (Table 2, Fig. 3). These modest affinity changes differ substantially from the several orders of magnitude in DNA affinity increase observed for CRP upon cAMP binding. Possibly, this biochemical difference correlates to the abrupt lifestyle shift that Vfr is poised to mediate as an on/off switch, as opposed to the graduated regulatory roles CRP has. It has been shown that the interaction of Vfr with the perfect consensus sequence in the lasR promoter (20) has unusual features compared to general Vfr-promoter interactions (15). For example, two complexes are seen in EMSA experiments, with the higher mobility band predominating (Fig. 3) (15).

Table 2.

Binding of Vfr and Vfr-Ndel7 proteins to the lasR promoter

[cAMP] (μM)	Vfr Kd (nM) ± SDa	Kd (nM)b
		Vfr	Vfr-Ndel7
0c	>400	>140	>100
2	54.4 ± 2.8
20	50 ± 2.8	5.0 (n = 8)	20 (n = 2)
200	45 ± 2.9
2,000	54.6 ± 3.6	4.6
6,000	57.5 ± 3.7

^aValues are based on fluorescence polarization measurements.

^bValues are based on EMSA analyses.

^cDue to uncertainty in the true [cAMP] at low values of [cAMP], these findings should be considered estimates only.

Fig. 3.

Vfr and Vfr-Ndel7 bind in a cAMP-dependent manner to the synthetic 30-bp lasR promoter in an EMSA. (A) 20 μM cAMP. The protein concentrations over each of the sets of 10 lanes are as follows: 0 nM, 0.1 nM, 0.33 nM, 1.0 nM, 3.33 nM, 10 nM, 33.3 nM, 100 nM, 333 nM, and 1 μM. (B) No cAMP. The protein concentrations for Vfr are the same as in panel A, whereas for Vfr-Ndel7 they are 0 nM, 38 nM, 58 nM, 86 nM, 130 nM, 196 nM, 296 nM, 444 nM, 666 nM, and 1 μM. The three bands are marked with hypothetical structural models. For the lower shifted band representing the symmetric Vfr:DNA complex, Vfr was superpositioned onto a DNA-bound CRP dimer (2CGP), and the DNA was displayed with Vfr. For the upper band representing interaction of a Vfr monomer with one DNA half site and for free DNA, the DNA model is from 1OSL.

Role of the N-terminal region of Vfr and implications of full-length Vfr model.

We sought to determine whether the truncated Vfr-Ndel7 used for structure determination retained its function in vivo. Vfr-Ndel7 was expressed from its endogenous promoter on a plasmid in P. aeruginosa strain K (PAK) or PAK Δvfr. This plasmid led to poor or no growth of both strains in broth or on solid media. Rare colonies were small and without twitching motility, and we did not analyze whether these were due to second site reversions or plasmid loss. This dominant-negative growth defect is specific to P. aeruginosa, since E. coli BL21(DE3) Δcrp cells expressing Vfr-Ndel7 had a growth rate equal to cells expressing Vfr. Introduction of plasmids expressing N-terminally truncated CRP into E. coli BL21(DE3) Δcrp cells (13) created no growth defects in LB compared to the parent strain expressing CRP from the chromosome (data not shown).

To rule out that Vfr-Ndel7 has lost DNA-binding properties, EMSA was used to test the ability of Vfr-Ndel7 to bind specifically to DNA. Like Vfr, Vfr-Ndel7 demonstrates cAMP-dependent DNA binding to the lasR oligonucleotide (Table 2, Fig. 3).

In analyzing the striking growth-deficient phenotype for P. aeruginosa strains expressing Vfr-Ndel7, we noticed the synthetic N-terminal methionine occupies a position within the small hydrophobic core of the N-terminal domain that it would not be able to reach in the full-length protein. Met8 is tightly packed with residues 13, 16, 43, 71, 107, 117, and 121 (Fig. 4A). In contrast, in our initial full-length Vfr structure, the electron density provides a stereochemically plausible model that includes the solvent-exposed N-terminal seven amino acids for one monomer. Curiously, these residues approach the now ordered amino acids 80 to 84 from the subunit across the dimer interface (Fig. 4B). This loop packs against the αC helix, potentially reinforcing the dimer interaction. Whether the proximity of the N terminus to 79-EKEGS-83 in full-length Vfr contributes to the phenotype of the strains expressing Vfr-Ndel7 remains to be tested experimentally, but it is noteworthy that the deletion of codons for a second quasi-repeat 84-EQERS-88 disrupts a subset of Vfr-mediated functions (2).

Fig. 4.

Structure of the Vfr N terminus. (A) The N-terminal Met side chain of Vfr-Ndel7 packs tightly into the hydrophobic core of the ligand-binding domain. The side chains closest to Met8 are contributed by αA, αB, αC, β2, and β3 (reverse rainbow color-coding, with the N-terminal Met in dark blue, accentuates the topology). (B) The full-length Vfr model indicates the N-terminal seven amino acids of one monomer (cyan) approach the 78-EKEGS-82 region (magenta) of the dimer partner (green).

Despite decades of work on CRP, a Vfr homologue, and the paradigm for bacterial transcriptional regulation, the X-ray crystal structures of Vfr and Vfr-Ndel7 provide several unexpected observations and open new avenues for research.