The DNA-binding domain of OmpR: crystal structures of a winged helix transcription factor

Research Article

109

The DNA-binding domain of OmpR: crystal structure of a winged helix transcription factor Erik Martínez-Hackert1,2 and Ann M Stock1,3,4* Background: The differential expression of the ompF and ompC genes is regulated by two proteins that belong to the two component family of signal transduction proteins: the histidine kinase, EnvZ, and the response regulator, OmpR. OmpR belongs to a subfamily of at least 50 response regulators with homologous C-terminal DNA-binding domains of approximately 98 amino acids. Sequence homology with DNA-binding proteins of known structure cannot be detected, and the lack of structural information has prevented understanding of many of this family’s functional properties. Results: We have determined the crystal structure of the Escherichia coli OmpR C-terminal domain at 1.95 Å resolution. The structure consists of three a helices packed against two antiparallel b sheets. Two helices, a2 and a3, and the ten residue loop connecting them constitute a variation of the helix-turn-helix (HTH) motif. Helix a3 and the loop connecting the two C-terminal b strands, b6 and b7, are probable DNA-recognition sites. Previous mutagenesis studies indicate that the large loop connecting helices a2 and a3 is the site of interaction with the a subunit of RNA polymerase. Conclusions: OmpRc belongs to the family of ‘winged helix-turn-helix’ DNA-binding proteins. This relationship, and the results from numerous published mutagenesis studies, have helped us to interpret the functions of most of the structural elements present in this protein domain. The structure of OmpRc could be useful in helping to define the positioning of the a subunit of RNA polymerase in relation to transcriptional activators that are bound to DNA.

Introduction The bacterial response regulator OmpR is a transcription factor that differentially regulates the expression of genes encoding the major outer membrane porin proteins, OmpF and OmpC [1]. OmpR promotes the transcription of ompF in conditions of low osmolarity; at high osmolarity, OmpR represses transcription of ompF and activates transcription of ompC [2,3]. This complex regulation involves multiple OmpR-binding sites upstream of the ompF and ompC genes to which OmpR has been shown to bind in a hierarchical fashion [4–6]. Phosphorylation of OmpR enhances its DNA-binding activity [7]. Biochemical and mutagenesis data are consistent with a model in which the concentration of phosphorylated OmpR increases with increasing osmolarity; this results in increased occupancy of the different OmpR-binding sites. Depending on the specific sites bound by OmpR, OmpR functions as a transcriptional activator or repressor. The response regulator OmpR and the osmosensor histidine protein kinase EnvZ function as a two component phosphotransfer signal transduction system [8,9]. OmpR is a 27 kDa protein containing an N-terminal domain of approximately 125 amino acids joined by a small linker region to a C-terminal domain of approximately 98 residues.

Addresses: 1Center for Advanced Biotechnology and Medicine, 679 Hoes Lane, Piscataway, NJ 08854, USA, 2Department of Chemistry, Busch Campus, Rutgers University, Piscataway, NJ 08854, USA, 3University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, Piscataway, NJ 08854, USA and 4Howard Hughes Medical Institute, Piscataway, NJ 08854, USA. *Corresponding author. E-mail: [email protected] Key words: DNA-binding protein, osmoregulation, response regulator, winged helix-turn-helix, X-ray structure Received: 21 October 1996 Revisions requested: 18 November 1996 Revisions received: 27 November 1996 Accepted: 27 November 1996 Electronic identifier: 0969-2126-005-00109 Structure 15 January 1997, 5:109–124 © Current Biology Ltd ISSN 0969-2126

The N-terminal domain contains the aspartate residue which is phosphorylated and other active-site residues that are highly conserved within the family of response regulator proteins. Genetic studies and in vitro DNA-binding assays have localized the DNA-binding activity to the C-terminal domain of OmpR [10,11]. The isolated C-terminal domain is capable of binding DNA, although with significantly lower affinity than is observed for intact phosphorylated OmpR. Both the C-terminal domain and phosphorylated OmpR have been shown to bind as dimers to a twenty base pair region of DNA containing a tandem ten base pair repeat [4,6,12–14]. In addition to binding DNA, OmpR must also interact productively with RNA polymerase to effectively regulate transcription [15–18]. Mutational analyses suggest that interaction with the C-terminal domain of the a subunit of RNA polymerase is mediated by the C-terminal domain of OmpR [19–21]. Thus the relatively small C-terminal domain of OmpR must accommodate numerous inter- and intramolecular interactions: a regulatory interaction with the N-terminal domain, dimerization or perhaps oligomerization with itself, DNA binding, and interaction with the a subunit of RNA polymerase. OmpR is one of the most extensively studied members of a subfamily of more than 50 response regulator proteins

110

Structure 1997, Vol 5 No 1

characterized by sequence similarity in their C-terminal DNA-binding domains. Members of this family include such diverse transcriptional regulators as Escherichia coli PhoB (of the phosphate assimilation pathway), Enterococcus faecium VanR (which controls resistance to the antibiotic vancomycin), and Agrobacterium tumefaciens VirG (involved in the establishment of crown gall tumors in plants). This DNA-binding domain has also been found in proteins other than response regulators, such as Vibrio cholerae ToxR, a transmembrane protein involved in cholera toxin expression. Despite extensive genetic analyses in several different systems, a detailed mechanism for DNA binding by these domains is unknown. The structure of the C-terminal domain of E. coli OmpR (OmpRc), reported here, provides the first structural information for the OmpR/PhoB family of DNA-binding domains. The region of OmpRc predicted from mutational analyses to be involved in DNA binding has a helixturn-helix fold similar to that of some previously characterized DNA-binding proteins (such as the biotin operon repressor BirA, catabolite gene activator protein CAP, and hepatocyte nuclear factor HNF-3g). The pattern of sequence conservation within the OmpR/PhoB family indicates a common fold for all of the DNA-binding domains, and the OmpRc fold can be used as a model to correlate mutational analyses in different proteins. Regions involved in DNA binding and interaction with the a subunit of RNA polymerase can now be mapped onto the OmpRc structure.

Results and discussion Structure determination

The structure of OmpRc, a fragment of OmpR corresponding to residues Ala130–Ala239, was determined by the multiwavelength anomalous diffraction (MAD) method using crystals of the selenomethionine substituted protein. Wild-type or selenomethionyl containing OmpRc crystals that belong to the trigonal spacegroup P3221 with cell dimensions a = 59.9 Å, c = 58.9 Å and g= 120.0° were grown by the hanging drop vapor-diffusion method as described in the Materials and methods section. The diffraction data from four wavelengths were measured at and about the selenium K-shell edge to maximize Bijvoet and dispersive differences. Experimental phases were calculated by the procedures implemented in the MAD phasing applications developed by Hendrickson and coworkers [22]. Selenium sites were located using both direct methods [23] and Patterson search methods [24] and were confirmed by calculating difference Fourier maps while omitting one site. The refined sites were used to calculate phases. A clear solvent boundary was discernible in a Fourier synthesis calculated from phases computed for space group P3221, consequently determining the handedness. Density modification [25] was required to generate interpretable maps. Even following density modification

the map was difficult to interpret, in many places showing stronger connectivity between the sidechains than for the mainchain. Using the density modified map it was possible to trace the mainchain, and the knowledge of the five selenium positions aided significantly in the assignment of amino acid positions. The model was subjected to several iterations of simulated annealing refinement and manual rebuilding, and simulated annealing omit maps were used to confirm the correctness of the tracing. In the current electron-density maps there is continuous well defined density (Fig. 1) for the entire polypeptide with the exception of a few surface sidechains and the region encompassing residues 194–200. The final model consists of 806 non-hydrogen atoms, corresponding to residues 137–235, and 86 solvent molecules. The protein model has good stereochemistry with a root mean square (rms) deviation of 0.017 Å for bond lengths and 2.07° for bond angles. The Ramachandran plot showed that 88.1 % of the residues have favored f and ψ angles, and no outliers are found in the disallowed regions. The crystallographic R factor for this model is 22.8 % for 7807 reflections and the free R factor is 26.9 % for 388 reflections between 5.0 Å and 1.95 Å resolution. The average temperature factor for the mainchain is 20.7 Å2 with all values lying in a range of 14.6–44.9 Å2. Higher than average B factors can be found in the loop regions of amino acid residues 190–200 and at the N terminus with an average value of 31.2 Å2. Two crystallization reports have preceded this publication [26,27]. Our laboratory has grown three different trigonal crystal forms of OmpRc. The crystal form used for the structure determination (space group P3221) is similar to the crystal form published by Kondo et al. [26] where P3n21 crystals with cell dimensions a=60.4 Å, c=58.8 Å were reported to diffract to 3.0 Å. It is not clear that the forms are identical as the specific OmpR fragment used by Kondo et al. is not described. Crystals that belong to the trigonal space group P3n12 with cell dimensions a = 53.5 Å, c = 131.4 Å and g = 120.0°, and which diffracted to at least 2.3 Å resolution at a synchrotron light source, were the major focus of a report published by our laboratory [27]. The P3n12 crystal form could not be solved by conventional phasing methods; heavy-atom derivatization failed to give useful derivatives. Selenomethionyl MAD data were collected at the selenium edge for this crystal form. Although a measurable, wavelength dependent, anomalous signal was observed, it was not possible to locate any one of the five selenium atoms per polypeptide chain by either Patterson methods, direct methods or cross-phasing using a single mercury derivative. Single-site cysteine residues were introduced by site-directed mutagenesis to provide a mercury-binding site. One crystallized mutant could be derivatized by mercury acetate with the heavy-atom position located very close to the threefold axis. Heavy-atom

Research Article DNA-binding domain of OmpR Martínez-Hackert and Stock

111

Figure 1 Stereoview showing a region of the final 1.95 Å |2Fo|–|Fc| electron-density map, contoured at 1s. Amino acid residues 151–153 of the refined model are superimposed. (Figure generated with MOLSCRIPT [70].)

data and mercury MAD data were collected, but the computed Fourier syntheses were uninterpretable. This was also reflected in the weak phasing power cal-culated for this mercury derivative. We now believe the P3n12 crystal form to be a trilled pseudo-hexagonal orthorhombic crystal. Three orthorhombic individuals can be intergrown so that these individuals are related by symmetry. It is in this fashion that orthorhombic crystals of pseudo-hexagonal cell dimensions commonly form trilled intergrowths to mimic hexagonal crystals [28]. The orthorhombic crystals would be related by threefold screw symmetry (31) with the longest orthorhombic axis providing the twofold rotation axis in P3n12. The pure orthorhombic crystal was grown when using one of the cysteine mutants designed for heavy-atom derivatization, thus supporting this hypothesis. Both orthorhombic and P3n12 crystal forms will be investigated in a separate context. Structure and tertiary fold

The 12 kDa DNA-binding domain of E. coli OmpR used in this study consists of the C-terminal residues Ala130– Ala239 and an initiator methionine residue (the numbering used in the text refers to intact OmpR). The structure of OmpRc consists of three a helices packed against two antiparallel b sheets, an N-terminal four-stranded antiparallel b sheet and a C-terminal hairpin. This hairpin interacts with a short stretch of b strand, that connects helices a1 and a2, to generate a three-stranded antiparallel b sheet (Fig. 2a,b). The topology for OmpRc is b1-b2-b3b4-a1-b5-a2-a3-b6-b7 (Fig. 3). The hydrophobic core of the protein is formed by sidechains contributed by each of the seven b strands and three a helices. A long ten-residue loop from Arg190–Ser200 connects helices a2 and a3. This a2-loop-a3 segment is similar to the helix-turn-helix (HTH) DNA-binding motif. The program DALI [29] was used to find structural homologs

(Fig. 4). The superposition of the OmpRc HTH region with the corresponding region of either the biotin repressor BirA [30] (Fig. 4a) or hepatocyte nuclear factor HNF3g [31] (Fig. 4b) clearly indicates a structural relationship between OmpRc and other winged HTH DNA-binding proteins. The length of the turn in the HTH motif of OmpRc differs from that of other winged HTH proteins, where the usual connection between the helices is four to five residues. The structure of the Mu transposase (MuA) provides a precedent for a winged HTH protein with an unusually large turn, albeit three amino acids shorter than is the case with OmpRc [32]. This large turn in OmpRc will be henceforth referred to in the text as the a loop because of its proposed interaction with the a subunit of RNA polymerase. The b strand that follows the HTH, b6, pairs with the C-terminal strand, b7. Strand b7 also pairs with strand b5 to form a three-stranded antiparallel b sheet. The loop connecting strands b6 and b7 extends away from the core to form a shaft with the N-terminal region of helix a3. An analogous protruding loop connecting the two b strands that follow the HTH motif has been referred to elsewhere as a ‘wing’ [33,34]. The wing, referred to in the text as W1, includes amino acid residues 225–229. This wing is involved in the formation of a classic antiparallel b bulge. Comparison with other winged HTH structures

The structures of several HTH proteins have been described [35]. All of these proteins have three helices, two of which are positioned in a similar fashion to form the HTH motif. The second helix in the HTH motif binds DNA, thus it is named the recognition helix. In some of the HTH transcription factors, such as BirA [30], CAP [36,37], histone H5 [38], HNF-3g [31], MuA [32] and the heat shock transcription factor (HSF) [39], an additional hairpin immediately follows the recognition helix. The

112

Structure 1997, Vol 5 No 1

Figure 2 Stereoview showing the overall fold of OmpRc. (a) A Ca tracing of OmpRc. The N and C termini and every twentieth residue are labeled; Ca positions are marked by open circles. (b) A ribbon diagram of OmpRc in the same orientation as (a). Secondary structure elements and biologically relevant elements are labeled. (Figure generated with MOLSCRIPT [70].)

(a)

C

140 160

N

160

180

200

C

140

220

220

N

180

200

(b)

β4 β3 β2

β4 β3 β2 β1

β1 α3

α3 β6

α loop

β7 β5

α1

W1

β6 α loop

β7 β5

α1

α2

loop connecting the two b strands that form the hairpin has been shown to interact with DNA [31,36,37]. This structural family has been referred to as the ‘winged-helix’ protein family to emphasize the importance of the recognition loop or wing [33,34]. The topology of winged HTH proteins can be generalized as H1-B1-H2-T-H3-B2-W1B3 (with the exceptions of CAP, where a second b strand is inserted between H1 and H2, and MuA where the first helix, H1, is transplanted to the C terminus of the DNAbinding domain). Winged HTH proteins have been structurally classified by the orientation of the three helices with respect to each other, or more specifically by the angle formed between helices H2 and H3 (the HTH angle) [40,41]. With respect to this angle, it is possible to subdivide winged HTH proteins into three groups: histone H5 and HNF-3g have HTH angles of approximately 50°; CAP, LexA (a bacterial repressor involved in the SOS response) and BirA have angles between 84° and 92°; HSF and MuA have angles of about 105° to 110°. Overall, OmpRc is most closely related to BirA. The rms deviation between the three helices of OmpRc and BirA is 1.05 Å, placing it within the CAP/ LexA/BirA subfamily [41]. The three helices, a1, a2 and a3, as well as the C-terminal

W1

α2

b hairpin are practically superimposable (Fig. 4a). The loops that connect helices a2 with a3 and helix a3 with the C-terminal hairpin, however, are significantly larger in the case of OmpRc, with four and two amino acids in the loops of BirA versus ten and eight amino acids in the loops of OmpRc. OmpRc is more distantly related to the HNF-3g/histone H5 subfamily of winged HTH proteins. The smaller interhelix angle observed in this subfamily results in a substantially different positioning of the a loop and the wing relative to that observed in OmpRc (Fig. 4b). Suzuki and Brenner [40] had previously predicted a close structural similarity between OmpRc and histone H5, based on primary sequence analysis. In retrospect, this prediction can be understood, as the turn present in the OmpRc HTH motif, the a loop, is rather large and comparable in size to the turn present in the histone H5 structure. The one significant structural difference between the OmpRc domain and other winged HTH proteins is its N-terminal four-stranded antiparallel b sheet. The antiparallel b sheet is an integral part of the OmpRc DNAbinding fold. This b sheet packs against helices a1 and a3, contributing six amino acids to the hydrophobic core of OmpRc. The b sheet is unique to the OmpR subfamily,

Research Article DNA-binding domain of OmpR Martínez-Hackert and Stock

Figure 3

Schematic representation of the secondary structure of OmpRc. Residues in b strands are enclosed in rectangular boxes and residues in a helices are encircled by ovals. Hydrogen bonds for mainchain atoms are represented by arrows from the hydrogen-bond donor to the acceptor and arrowheads mark hydrogen bonds involving loop residues. (Figure generated with PROMOTIF [71].)

and provides a basis for defining a new subclass of winged HTH proteins. The relationship of OmpRc with the winged HTH family of proteins has helped to define, by analogy, the function of some of the structural elements present in the OmpRc structure. Helix a3 is the DNA-recognition helix and the loop that connects b strands b6 and b7 functions as the DNA-recognition wing.

113

identify regions of the protein that affect the three independent functions necessary for its intact activity: activation, repression and DNA binding [19,21,42]. These mutants are summarized in Table 1. Only mutations that lie within the crystallized fragment and alter activation or DNA binding will be discussed here. These mutational analyses have aided in interpreting the separate functions of the structural components found in OmpRc. OmpR mutants that fail to activate transcription of ompC, but that can repress transcription of ompF, belong to the class of activation mutants. The ability to repress ompF implies that the protein can bind DNA, yet somehow fails to activate transcription. These mutant proteins are thought to lack the ability to interact with the a subunit of RNA polymerase. Mutations cluster into two regions of the amino acid sequence and tertiary structure: Pro179→Leu and Ser181→Pro; and Gly191→Ser, Glu193→Lys, Ala196→ Val and Glu198→Lys (Fig. 5a). The first set of mutations precedes helix a2, while the latter set forms part of the a loop. The a loop is remarkable in that its length establishes a departure from the typical winged HTH motif. It should be noted that high temperature factors were observed for residues in this region of the OmpRc model, implying the possibility of conformational flexibility. Multiple conformations of this loop may be relevant to physiological function. The a loop mutations all lie on one surface of the protein and thus are likely to be part of an OmpRc surface that interacts with RNA polymerase. The Ser181→ Pro mutation, though classified as an activation mutant, might be expected from the structure to directly affect DNA binding. Ser181 functions as a classic helix N-cap residue with the Og accepting a hydrogen bond from the backbone nitrogen of Lys184 to stabilize the first turn in helix a2. A possible consequence of this hydrogen bond is the specific positioning of Arg182, a residue shown to affect DNA binding. Furthermore, the Ser181→Pro mutation might destabilize helix a2 and, as a result, modify the conformation and positioning of the a loop. The residue corresponding to Ser181 is highly conserved throughout the OmpR/PhoB family as either a serine or a threonine. Pro179 and the a loop lie on different faces of the protein surface, therefore, it is questionable whether they interact with the same region of the a subunit. It seems more likely that the Pro179→Leu mutation indirectly influences transcription activation, perhaps by perturbing the structure of b5 and hence the position of Ser181.

Correlation with mutagenesis data

Transcription of the genes encoding the OmpF and OmpC proteins is regulated by OmpR. Under conditions of low osmolarity, OmpR activates transcription of ompF, while under high osmolarity it represses transcription of ompF and activates transcription of ompC. OmpR controlled transcription is thought to function through a direct interaction with the a subunit of RNA polymerase [19–21]. OmpR mutants have been widely studied to

A second class of OmpR mutants was shown to be defective in DNA binding. Mutations cluster in several regions of the amino acid sequence: Arg150→Ser, Thr162→Ile, Lys170→ Glu, Arg182→Leu, Ser200→Phe, Val203→Met, Met211→ Val, Val212→Ala, Arg220→Cys, Thr224→Ile and Gly229→ Ser (Fig. 5b). Because of their location in the structure and/or involvement in specific hydrogen-bond interactions, four of these mutations (Arg150→Ser, Lys170→Glu,

114

Structure 1997, Vol 5 No 1

Figure 4 Stereoview of the helix-HTH-wing motifs of the biotin repressor BirA and hepatocyte nuclear factor HNF3-g superimposed on the structure of OmpRc. A ribbon diagram of OmpRc is displayed in blue and cyan. (a) Superposition of OmpRc with the BirA DNA-binding domain [30]; the BirA domain is shown in red. The relative positions of the three helices and the C-terminal b hairpin are very similar, while the conformations of the loops preceding and following the recognition helix are dissimilar. The OmpRc loops are significantly larger than the respective loops found in other bacterial transcription factors. (b) Superposition of OmpRc with the HNF3-g DNA-binding domain [31]; the HNF3-g domain is shown in red. The simultaneous superposition of the three helices leaves little doubt of the more distant relationship between OmpRc and the HNF3-g/histone H5 family of winged HTH proteins. (Figure generated with RIBBONS [72]; the structures were aligned using DALI [29] and SUPERPOSE [63].)

Val212→Ala and Arg220→Cys) might be expected to disrupt DNA binding through a structural perturbation. The remaining mutations probably correspond to residues involved in interaction with the DNA. Three of the mutations, Ser200→Phe, Val203→Met and Met211→Val, map onto the recognition helix while two of the mutations, Thr224→Ile and Gly229→Ser, lie within the W1 loop. In all cases where a winged HTH protein has been co-crystallized with DNA (e.g. CAP, HNF-3g and ETS-PU.1 [43]; ETS transcription factors regulate gene expression during growth and development) the wing has been found to directly interact with the DNA backbone. By analogy, and considering the supporting mutagenesis data, it might seem reasonable to suggest that the W1 loop of OmpRc also contacts DNA. The other two sites of mutation, Thr162 and Arg182, precede the N termini of helices a1 and a2 and are both positioned in proximity to the recognition helix a3. Therefore, these sites are also potential candidates for protein–DNA interactions. Amino acids in analogous positions, preceding the N termini of helices a1 and a2, have been found to contact both the DNA backbone and bases in the co-crystal structures of CAP, HNF-3g and ETSPU.1, respectively.

A mutational analysis of PhoB, an OmpR homolog, has been carried out previously [44]. This extensive characterization parallels and corroborates the OmpR mutagenesis data analysis presented here. Surface interactions

OmpRc interacts with three, if not four, distinct surfaces. Two interactions, with DNA and the a subunit of RNA polymerase, have already been discussed. Previous studies have shown that the DNA-binding activity of intact OmpR is greatly enhanced when the N-terminal domain is phosphorylated [7,45], thus the N-terminal domain somehow modulates the ability of the C-terminal domain to bind DNA. Furthermore, experiments using limited trypsin digestion have shown that the proteolytic susceptibility of the linker region, that connects the N- and C-terminal domains, changes upon phosphorylation [46]. These two independent findings both support a direct interaction between the two domains. A fourth interaction of OmpRc, dimerization upon DNA binding, has been proposed [47,48]. The electrostatic surface potential of OmpRc reveals three interesting regions (Fig. 6). The DNA-binding surface

Research Article DNA-binding domain of OmpR Martínez-Hackert and Stock

115

Table 1 Mutants of OmpR. Phenotype Activation

Mutation position

Amino acid substitution

Secondary structure

Reference

179 181 191 193 196 198

Pro→Leu Ser→Pro Gly→Ser Glu→Lys Ala→ Val Glu→Lys

b5 b5 a loop a loop a loop a loop

Pratt et al. [20] Kato et al. [21] Russo et al. [19] Russo et al. [19] Pratt et al. [20] Pratt et al. [20]

150 162 170 182 200 203 211 212 220 224 229

Arg→ Ser Thr→ Ile Lys→Glu Arg→Leu Ser→Phe Val→Met Met→Val Val→Ala Arg→Cys Thr→Ile Gly→Ser

b3 a1 a1 a2 a3 a3 a3 a3 b6 W1 W1

Kato et al. [21] Kato et al. [21] Kato et al. [21] Kato et al. [21] Kato et al. [21] Russo et al. [19] Aiba et al. [42] Aiba et al. [42] Russo et al. [19] Kato et al. [21] Kato et al. [21]

DNA binding

No No No No No No Weak No Weak No No No No No No No No

probably involved in the OmpRc–RNA polymerase interaction. Although the a loop surface is formed primarily by hydrophobic residues, it also includes several charged amino acids. The aliphatic components of these charged amino acids, two glutamate and two arginine residues, are likely to account for the hydrophobicity of the surface. A

has, as would be expected, a positive electrostatic potential, ideal for association with the negatively charged phosphodiester backbone of DNA (Fig. 6a). Three arginine sidechains along the recognition helix provide the majority of the positive charge. The a loop surface is rather neutral and hydrophobic in nature (Fig. 6b); this surface is Figure 5 Stereoview showing residues corresponding to mutation sites in OmpRc that affect transcriptional activation and DNA binding. The wild type amino acid sidechains are shown, with carbon atoms colored green, oxygen atoms red, nitrogen atoms blue and sulfur atoms yellow. (a) OmpR mutations Pro179→Leu, Ser181→Pro, Gly191→Ser, Glu193→Lys, Ala196→Val and Glu198→Lys fail to activate transcription. These mutant proteins are thought to lack the ability to interact with the a subunit of RNA polymerase. The a loop is highlighted in pink. (b) OmpR mutations Arg150→Ser, Thr162→Ile, Lys170→Glu, Arg182→Leu, Ser200→Phe, Val203→Met, Met211→Val, Val212→Ala, Arg220→Cys, Thr224→Ile and Gly229→Ser fail to bind DNA. The recognition helix, the C-terminal b hairpin and the W1 loop are highlighted in orange (for references see Table 1). (Figure generated using MOLSCRIPT [70].)

(a)

191 196

193

191

179

196

193

181

182

182

229

229

200

200

203

203 224

162 170 150

181

198

198

(b)

179

150

211 212

220

224

162 170

211 212

220

116

Structure 1997, Vol 5 No 1

third surface stands out most intriguingly in this electrostatic potential diagram (Fig. 6c). This highly negative face of the protein lies at the C-terminal end of the recognition helix and is generated by amino acids Glu213, Glu214, Asp215, Asp155, Glu156 and Asp157. The function of this acidic cluster is unclear, but its strikingly negative potential possibly indicates some important function for this region. A similar, strongly negative electrostatic potential has been previously noticed on the DNA-binding surface of the Trp repressor [49]. Extensive analysis of TrpR led the authors to propose a model in which the negative potential serves to orient the protein, with the N-terminal region of the recognition helix directed towards the major groove of the DNA. The negative electrostatic potential of OmpRc could presumably have a similar function.

Figure 6

Relationship to other members of the OmpR/PhoB family of transcriptional regulators

The DNA-binding domain of OmpR exhibits sequence similarity over its entire length with a family of over 50 different transcriptional regulators. Most of these proteins are response regulator proteins that function as part of two component signal transduction systems. These proteins are characterized by conserved N-terminal regulatory domains that control the activities of their C-terminal DNA-binding domains in a phosphorylation-dependent manner [50,51]. The DNA-binding domain, however, is apparently a modular unit that can be integrated into other types of proteins as well. For instance, V. cholerae ToxR is a transmembrane protein that consists of an N-terminal DNA-binding domain and a C-terminal periplasmic domain. There is significant variation in the degree of sequence similarity among different DNA-binding domains, with members of the family exhibiting from 20 % to 65 % sequence identity with the 98 amino acid DNA-binding domain of OmpR. A representative alignment of a few of the most extensively characterized family members is shown in Figure 7. Patterns of sequence conservation leave little doubt that all members of the family can be interpreted in terms of a common fold. Specifically, the amino acid residues that form the hydrophobic core of the OmpRc domain are conserved as hydrophobic residues throughout the family (Fig. 7). The N-terminal fourstranded antiparallel b sheet that distinguishes OmpRc from other winged HTH proteins is conserved in the OmpR/PhoB subfamily. Many of the residues in OmpR that have been identified from mutagenesis studies as being important for DNA binding are conserved or are conservatively substituted in other family members. Some of these residues are expected to serve structural roles, while others may be directly involved in recognition of specific DNA sequences (see discussion below). The relatively low level of sequence identity in the OmpR family of DNA-binding domains is not altogether

Three views of the electrostatic surface potential of OmpRc. Red indicates a negatively charged region, blue indicates a positively charged region and white indicates a neutral or hydrophobic region. (a) The molecule is oriented similarly to Figure 5b, displaying the positive electrostatic surface potential of the DNA-binding surface. (b) The molecule is oriented similarly to Figure 5a. The a loop surface, that may interact with the a subunit of RNA polymerase, is rather neutral or hydrophobic. (c) This remarkable, negatively charged face of the protein lies at the C-terminal end of the recognition helix. (Figure produced using GRASP [73].)

Research Article DNA-binding domain of OmpR Martínez-Hackert and Stock

117

Figure 7

β1

OmpR(136) VirG(142) PhoB(130) VanR(142) ToxR( 19)

β2

β3

β4

β5

α2

* * * * AVIAFGKFKLNLGTREMFREDEP...MPLTSGEFAVLKALVSHPREPLSRDKLM RSFSFADWTLNLRRRRLISEEGSE..VKLTAGEFNLLVAFLEKPRDVLSREQLL EVIEMQGLSLDPTSHRVMAGEEP...LEMGPTEFKLLHFFMTHPERVYSREQLL NVIVHSGLVINVNTHECYLNEKQ...LSLTPTEFSILRILCENKGNVVSSELLF KFILAEKFTFDPLSNTLIDKEDSEEIIRLGSNESRILWLLAQRPNEVISRNDLH α3

OmpR(187) VirG(194) PhoB(181) VanR(183) ToxR( 73)

α1

β6

β7

* * ** * * * .NLARGREYSAMERSIDVQISRLRRMVEEDPAHPRYIQTVWGLGYVFVPDGSKA .IASRVREEEVYDRSIDVLILRLRRKLEGDPTTPQLIKTARGAGYFFDADVDVS NHVWGTNVYVE.DRTVDVHIRRLRKAL.EPGGHDRMVQTVRGTGYRFSTRF... HEIWGDEYFSKSNNTITVHIRHLREKMNDTIDNPKYIKTVWGVGYKIEK..... DFVWREQGFEVDDSSLTQAISTLRKMLKDSTKSPQYVKTVPKRGYQLIARVETV

Sequence alignment of the DNA-binding domains of OmpR and several family members. The sequences of E. coli OmpR [74], Agrobacterium tumefaciens VirG [75], E. coli PhoB [76], Enterococcus faecium VanR [77] and Vibrio cholerae ToxR [78] DNA-binding domains are shown. The amino acid positions are

indicated by the numbers in parentheses and a schematic diagram of the secondary structure of OmpRc is indicated above. Residues corresponding to the hydrophobic core of OmpRc are shaded, and positions of mutations that affect DNA binding of OmpR are indicated by asterisks above the OmpRc sequence.

unexpected, given the significant variations in DNA binding and activation mechanisms of these proteins. Each DNA-binding domain mediates unique interactions, such as the protein–protein interaction with a specific regulatory domain and the recognition of a specific DNA sequence. There also appear to be differences in the extent to which protein dimerization, inhibition by the regulatory domain, and activation by phosphorylation play a role in DNA binding in different proteins. While all members of the family, for which DNA-recognition sites have been characterized, appear to recognize tandemly oriented DNA sequences, there is a variation in the arrangement of sites, both with respect to the number of recognition sites and the spacing between them.

been defined and bases that are important in OmpR– DNA interactions have been identified [13,14] (Fig. 8a).

DNA binding

The structural similarity between the DNA-binding domains of CAP and OmpRc has provided a structural model that can be qualitatively interpreted with respect to OmpR–DNA interactions. Nevertheless, both OmpRc and the DNA may exist in different conformations when bound together in a complex. The recognition helix a3 should, by correspondence with CAP, bind to the major groove of the DNA. Three OmpRc–DNA models were examined: one complex was built using the bent DNA found in the CAP co-crystal structure; a second model was examined using the less bent DNA found in the HNF-3g co-crystal structure; and a third model was built using idealized B-form DNA.

OmpR has been shown to bind in a hierarchical fashion to multiple sites: the –100 to –40 base pair region of ompC [4,11,52]; and the –100 to –40 and –384 to –351 base pair regions of ompF [5,6,11]. The nature of the OmpRbinding site was until recently poorly understood, and the bases specifically recognized by OmpR were not accurately defined. A monomer of OmpR is thought to bind to a DNA region that spans ten base pairs. Recognition sequences recur in a tandem arrangement and conserved bases are separated from each other by ten base pairs, or roughly one helical turn [14]. Two adjacent recognition sites are necessary for DNA binding and are recognized by two OmpR molecules [13,48]. A consensus sequence has

The models agree qualitatively with the previously discussed mutagenesis data. However, mutagenesis data suggests that amino acids at both the N- and C-terminal ends of the recognition helix are important for DNA binding, but is not possible to simultaneously position amino acids at both ends of the recognition helix near the DNA. Furthermore, it is not possible to accommodate either bent CAP DNA, HNF-3g DNA or B-form DNA with a lockand-key fit to OmpRc. The latter suggests several possibilities: firstly, the conformation of DNA is different in an OmpRc–DNA complex; secondly, OmpRc undergoes a conformational change required for DNA binding; and

118

Structure 1997, Vol 5 No 1

Figure 8 (a)

Promoter

Nucleotide sequence

F1 C1 F2 F3 F4

T T T T G

Mutated fragment Base substitution

No. of mutations

T T T T T

T T T A T

A A T T A

C C C C C

T A T T G

T T T T G

T T T T A

T T T G A

G-G T-G T-G T-A T-A

T A A G T

T A A C T

T T A C A T T T T-G A A T T T G G G G C A A 4 3 10 7 2 2

A A A A A

C C C C C

A A C T A

T T A T T

A C A T T

T T A C G

T A T A C

A C A T C T T T T G G C C C 8 8 8

1

Random selection T T T A C T T T T G-G T T A C A T N T N (b)

Helix α1

*

*

**

OmpR 163 S G E F A V L K A L V S H P R E P L S R 182 CAP 138 D V T G R I A Q T L L N L A-Q I K I T R 169 + + (152-163)+ Helix α2

*

OmpR 183 D K L M N L A R G R E Y S A M E R S I D 202 CAP 170 Q E I G Q I V G . .(178) . . . . S R E T V G 184 + + + + + Helix α3

*

* *

*

OmpR 203 V Q I S R L R R M V E E D P A H P R Y I 222 CAP 185 R I L K M L E D Q N L . . . . . . . I S 197 + + β6 β7

*

OmpR DNA recognition. (a) OmpR-binding sequences. The sequences of OmpR-binding sites located upstream of the ompF and ompC genes, designated F1–F4 and C1, are aligned with a dash separating the tandem ten base pair recognition sequences that comprise the twenty base pair site bound by two OmpR molecules. Strictly conserved bases are boxed. A mutated fragment studied by Pratt and Silhavy is aligned below [14]. The base pair substitutions that affected DNA binding and the number of mutations isolated at each site are indicated. At the bottom, is an alignment of the conserved sequences isolated by Harlocker et al. in an OmpR binding selection from a pool of oligonucleotides in which the bases shown in bold had been randomized [13]. (b) Correlation between residues in OmpRc and CAP. The amino acid sequences of the DNA-binding motifs of OmpRc and CAP are aligned according to the superposition of the tertiary structures as output by the program DALI [29]. The larger loop regions of OmpRc necessitate the insertion of gaps within the CAP sequence; residues that were not aligned by DALI are indicated in parentheses. The amino acid positions in OmpRc and CAP are indicated by numbers at the ends of the sequences and the secondary structure elements of OmpRc are diagrammed above. Asterisks above the OmpRc sequence mark positions at which mutations affect DNA binding and crosses below the CAP sequence indicate residues that are observed to directly interact with DNA in the crystal structure [36,37].

*

OmpR 223 Q T V W G L G Y V F V P D 235 CAP 198 A H G . .(201) . . . . T I V V 205 +

thirdly, both DNA and OmpRc conformations are altered in forming the complex. While a local change in the structure of OmpRc is a distinct possibility, it is not clear if, and to what degree, OmpR/OmpRc binding causes the DNA to bend. Despite the lack of a perfect fit between OmpRc and DNA, several features of the model agree qualitatively with the existing data. Amino acids that are important for DNA binding, Arg182, Ser200, Val203 and Thr224, are positioned so as to interact with DNA. If CAP and OmpRc are superimposed, these residues of OmpRc correspond to CAP residues Arg169, Thr182, Arg185 and His199, which directly contact DNA in the co-crystal structure. In CAP, residues Arg180 and Glu181 also play an important role in recognition of specific bases (Fig. 8b).

Interaction with the N-terminal regulatory domain

As has been found for many other response regulator proteins, the activity of the C-terminal domain of OmpR is regulated by the N-terminal domain. In some response regulators, the unphosphorylated N-terminal domain inhibits the activity of the C-terminal effector domain. Phosphorylation of the regulatory domain relieves this inhibition, and additionally, in some proteins contributes to activation of the C-terminal effector function. In vitro DNA-binding studies suggest that this is the case for OmpR. Gel shift and footprinting analyses have demonstrated the binding of OmpR to a variety of DNA sequences. Unphosphorylated OmpR, the C-terminal domain and phosphorylated OmpR have increasing binding affinities, respectively [7,13,48,53].

Research Article DNA-binding domain of OmpR Martínez-Hackert and Stock

119

Figure 9 Stereoview superposition of the DNA-binding domains of the response regulators NarL and OmpRc. The three a helices of OmpRc were aligned with helices a7, a8 and a9 of the DNA-binding domain of NarL [57] using the program SUPERPOSE [63]. Residues 143–154, corresponding to the linker region of NarL, were not visible in the electrondensity map and are not displayed. When the C-terminal DNA-binding domains of OmpRc and NarL (shown in blue and red, respectively) are aligned in this manner, the N-terminal regulatory domain of NarL (shown in yellow and orange) collides with the C-terminal b hairpin of OmpRc. The arrangement of the N-terminal domains, with respect to the recognition helices, must differ in the two families. (The figure was generated with RIBBONS [72].)

The structure of OmpRc, together with structural knowledge of the N-terminal domain by analogy with its homolog CheY (the response regulator that controls flagellar rotation in bacterial chemotaxis) [54,55], permits definition of the boundaries of the linker region that joins the two domains. The linker consists of approximately 12 residues if the endpoints of the N-terminal and C-terminal domains are considered to be residues 124 and 137, respectively. This is substantially shorter than the 24 or 31 residue linker regions determined from structural studies of other response regulator proteins. There is a 24 residue linker in CheB (the chemotaxis receptor methylesterase) and a 31 residue linker in NarL (a transcriptional regulator belonging to a different subfamily) [56,57]. Still, the linker is sufficiently large to provide for a significant interface between the two domains. Consistent with this, no detectable protein complex is formed from isolated Nand C-terminal domains of OmpR (A West, unpublished data). The linker region of OmpR is highly susceptible to proteolysis [46], as has been observed for other response regulators [58]. It is likely that the linker is exposed and/or perhaps disordered, as is found in the crystal structure of NarL [57]. The recently determined structure of intact NarL provides a starting point for analyzing regulatory domain– effector domain interactions [57]. The DNA-binding domain of NarL consists of four helices with a classic HTH motif. The N-terminal regulatory domain of NarL packs directly against one face of the recognition helix, effectively precluding DNA binding. Thus, activation of NarL must involve disruption of the interaction between the Nand C-terminal domains, allowing access of DNA to the

recognition helix. It was of interest to explore whether a similar strategy could be applicable to the OmpRc domain. Although there is limited similarity between the folds of the NarL and OmpR DNA-binding domains, each contains a HTH motif and the structures can be aligned by superimposing the HTH helices. However, when aligned in this manner, the N-terminal domain of NarL collides with the C-terminal b hairpin of OmpRc (Fig. 9). Thus it is clear that the N-terminal domains are not positioned similarly, with respect to the recognition helices, in these two subfamilies of response regulator proteins. It still seems likely, based on biochemical characterization of OmpR, that the unphosphorylated N-terminal domain interferes with DNA binding, but the surfaces through which this inhibition occurs remain to be defined.

Biological implications A wide variety of organisms use a common phosphotransfer signaling strategy involving a histidine protein kinase and a response regulator. The histidine protein kinase autophosphorylates at a histidine residue, providing a high energy phosphoryl group available for transfer to an aspartate residue within the conserved regulatory domain of the response regulator. Phosphorylation of the regulatory domain controls the activity of an associated or downstream effector domain. In most response regulators, the effector domain is a DNA-binding domain that activates or represses transcription. Response regulators can be subclassified by sequence similarity in their effector domains. OmpR, a response regulator that regulates the transcription of genes encoding the major outer membrane porins, is representative

120

Structure 1997, Vol 5 No 1

of a subfamily of over 50 proteins with homologous DNA-binding domains. Despite significant biochemical and mutagenesis analyses, this class of transcription factors has remained poorly understood due to the lack of structural information for the 98 amino acid DNAbinding domain. The crystal structure of the Escherichia coli C-terminal DNA-binding domain of OmpR (OmpRc), determined at 1.95 Å resolution, has revealed that OmpRc belongs to the ‘winged helix-turn-helix’ family of DNA-binding proteins. OmpRc contains three a helices and an antiparallel b sheet, common to all winged HTH proteins. OmpRc, however, also contains an additional four-stranded antiparallel b sheet that distinguishes the OmpR family from all previously defined classes of winged HTH proteins. Analogy with other winged HTH proteins and correlation with mutagenesis data have allowed the identification of important structural features: a recognition helix and a wing structure involved in DNA binding, and an extensive loop preceding the recognition helix involved in interaction with the a subunit of RNA polymerase. The structure of OmpRc provides a scaffold for designing and interpreting future experiments aimed at describing the detailed interactions between OmpRc and the N-terminal regulatory domain, DNA and RNA polymerase.

Materials and methods Purification of wild-type and selenomethionyl protein The 12 kDa DNA-binding domain of OmpR from E. coli, OmpRc, composed of the C-terminal residues Ala130–Ala239 and an initiator methionine residue, was overexpressed using a T7 expression vector and purified from E. coli BL21 cells. The method used was a minor modification of the procedure described previously [27] and summarized briefly here. Approximately 6.0 g of cells from 3 L of culture were lysed by sonic disruption, resuspended in 20 mM Tris⋅HCl, 2 mM 2-b-mercaptoethanol (bME), pH 8.6, and membranes were removed by ultracentrifugation. The OmpRc protein was purified using a 2.5 × 20 cm Q Sepharose fast flow (Pharmacia) column equilibrated in 20 mM Tris⋅HCl, 2 mM bME, pH 8.6. The protein was eluted with a 400 ml linear gradient of 0–200 mM NaCl in 20 mM Tris⋅HCl, 2 mM bME, pH 8.6. The flow-through and fractions from the gradient containing OmpRc were pooled and concentrated by precipitation with 60 % saturated ammonium sulfate. The precipitate was collected by centrifugation and resuspended in 10 ml of 10 mM sodium phosphate, 2 mM bME, pH 6.0. The dialyzed protein was loaded onto a 5 ml HiTrap S Sepharose column (Pharmacia) and eluted with a 200 ml linear gradient of 0–200 mM NaCl in 10 mM sodium phosphate, 2 mM bME, pH 6.0. The fractions containing OmpRc were pooled and concentrated by precipitation with 60 % saturated ammonium sulfate. The precipitate was collected by centrifugation and resuspended in 4 ml of 20 mM Tris⋅HCl, 200 mM KCl, 2 mM bME, pH 7.5. The filtered sample was applied to a HiLoad 26/60 Superdex G75 pg FPLC column (Pharmacia) equilibrated in 20 mM Tris⋅HCl, 200 mM KCl, 2 mM bME, pH 7.5. Selenomethionyl protein was overexpressed from E. coli DL41, a methionine auxotroph strain [59], and purified using the same protocol.

protein and reservoir solutions. It was critical to significantly slow down crystallization to grow usable crystals. Both wild-type and selenomethionyl OmpRc were crystallized at 4°C from a reservoir buffer containing 3 % polyethylene glycol monomethylether (PMME) 5000, 17.5 % ethyleneglycol, 2.5 % 2-methyl-2,4-pentanediol (MPD), 30 mM 2-(N-morpholino) ethanesulfonic acid (MES), pH 6.5, using freshly purified protein concentrated with Centricon 10 units (Amicon) to 20–25 mg ml–1 in 2 mM bME. Crystals were grown in hanging drop vapor-diffusion setups, in which 2 ml of protein solution were mixed with 2 ml of reservoir solution. Large crystals with nice morphology appeared after two days and eventually dissolved after approximately five weeks. Crystals grown under these conditions belong to the trigonal spacegroup P3221 with cell dimensions a = 59.9 Å, c = 58.9 Å and g = 120.0°. For data collection at 100 K, crystals often have to be soaked in cryoprotecting solutions. P3221 OmpRc crystals, however, are extremely sensitive to changes of the mother liquor, and start to dissolve instantly upon perturbation. The original rationale to overcome this sensitivity problem was to grow crystals in a solution that can serve as cryoprotectant. The crystallization solution could serve as cryobuffer, but was far from ideal. Interestingly, this sensitivity problem was solved by adding 200 mM ammonium sulfate to the cryoprotecting solution. Adding salts to a cryoprotectant for crystal stabilization is rather unusual when salts are not present in the crystallization buffer. Prior to freezing, crystals were slowly overlaid and soaked with a solution containing 10 % PMME 5000, 20 % ethyleneglycol, 10 % PEG 200, 200 mM ammonium sulfate and 30 mM MES, pH 6.5. Crystals were equilibrated for at least 30 min in the cryobuffer, mounted on a nylon loop, and directly frozen in a cold nitrogen stream at a temperature of 100K. Cell dimensions of frozen crystals are slightly smaller (a = 59.1 Å, c = 58.1 Å) than those of unfrozen crystals. We have previously described several crystal forms of OmpRc [27], focusing on a trigonal crystal that belongs to spacegroup P3n12 with cell dimensions a = 54.4 Å, c = 135.5 Å and g = 120.0°. We reported selenomethionyl and heavy-atom data collected for this crystal form. However, the structure for these trigonal crystals could not be solved by either selenomethionyl MAD, mercury MAD or single isomorphous replacement and anomalous scattering (SIRAS) methods. For a number of reasons we believe this trigonal crystal form to have a lattice defect called ‘trilling’. The phase problem of a trilled crystal is not solvable by either MAD or isomorphous replacement methods. We will further investigate the OmpRc P3n12 crystal form, but will refrain from more detailed discussion in this paper.

Data collection MAD data were collected from one crystal at the Brookhaven National Synchrotron Light Source (NSLS), HHMI beamline X4A. The NSLS source was operated at 2.5 GeV with an initial current of approximately 310 mA. The selenium K-shell X-ray absorption spectrum was measured with a scintillation counter using the frozen selenomethionine containing crystal from which data were taken. The monochromator settings were chosen to be 0.9794 Å for the edge, 0.9792 Å for the peak, 0.9876 Å for the low energy remote and 0.9686 Å for the high energy remote wavelengths, respectively. Data were collected using inverse beam geometry with an oscillation range of 2.0° and no overlap. Oscillation images were indexed and integrated using the program DENZO and images were scaled using the program SCALEPACK [62]. Data were scaled within each wavelength. Unmerged scaled data were further processed using the MADSYS set of programs [22]. For refinement, the data of the low energy remote wavelength were processed to the data collection limit of 1.95 Å, merged and converted to structure factors using the program TRUNCATE (Table 2).

Crystallization and freezing Initial crystallization conditions were found using the sparse matrix approach in a hanging drop vapor-diffusion experiment [60,61]. Microneedles grow in a variety of PEG-containing solutions upon mixing the

Data from an unfrozen crystal have been collected on an R-AxisIIc detector using mirror-focused and Ni-filtered CuKa radiation from a Rigaku RU-200 rotating-anode source operated at 5 kW. The crystal

Research Article DNA-binding domain of OmpR Martínez-Hackert and Stock

121

Table 2 Data collection statistics. Wavelength (Å)

0.9876* Remote l1

Cell axis a (Å) 59.1 Cell axis c (Å) 58.1 Temperature (K) 100 Resolution (Å) 1.95 (2.00–1.95) 0.046 (0.086) Rsym†† 10.8 (6.9) No. of unique reflections 8604 Completeness (%) 97.1 (98.7) Multiplicity 7.1 (5.6)

0.9794* Edge l2

0.9792* Peak l3

0.9686* Remote l4

1.5418†

59.1 58.1 100 2.0 (2.05–2.00) 0.051 (0.107) 10.3 (6.0) 7956 99.0 (99.0) 7.4 (6.0)

59.1 58.1 100 2.0 (2.05–2.00) 0.047 (0.107) 11.1 (6.3) 7977 99.0 (99.0) 6.9 (5.4)

59.1 58.1 100 2.0 (2.05–2.00) 0.046 (0.095) 10.9 (6.7) 7988 99.0 (99.0) 7.5 (6.2)

59.9 58.9 273 2.35 (2.41–2.35) 0.055 (0.182) 12.3 (4.1) 4672 88.8 (85.6) 2.0 (1.9)

*Data collected at the National Synchrotron Light Source beamline X4A. †Data collected on an R-AxisIIc detector system. ††R sym=SS(|Ij–|)/SS, where Ij is the intensity of observed reflection j and is the mean intensity for reflection j. Values in parentheses correspond to the highest resolution shell. was cooled to approximately 273K using an FTS cool air system. Data were integrated using the program DENZO, scaled and merged with the programs ROTAVATA/AGROVATA and converted to structure factors with TRUNCATE [63] (Table 2).

Phase determination MAD phasing was carried out using both the algebraic approach implemented in the MADSYS programs [22] and the probabilistic approach implemented in MLPHARE [64]. The two procedures were carried out as a confirmation of the correctness of the solvent boundary in experimental maps of low quality. Scaled and unmerged data were used in MADSYS phasing and were subjected to local scaling to reduce noise in the Bijvoet signal and in the dispersive signal among the four wavelengths. Scaled data were fitted in MADLSQ to approximate f′ and f′′ values obtained from previously published experiments. Unique data and phasing statistics were obtained from MERGIT. The positions of five selenium atoms were determined independently using the Bijvoet differences of the peak wavelength in combination with the direct methods program MULTAN [23] (using reflections between 4.0 and 2.2 Å) and with SHELX [24] using the |°FA| Patterson

map calculated with reflections between 5.0 and 2.5 Å. |°FA| is the normal scattering factor of the anomalous scatterers as determined by MADLSQ. The selenium sites were confirmed in cross-phased difference Fourier experiments omitting one selenium site at a time. The refined selenium sites were input into MADABCD to obtain phases and figures of merit to 2.2 Å resolution (Table 3). A solvent boundary could be observed in the electron-density map in space group P3221. Surprisingly, the solvent boundary could only be detected when calculating the electron-density map to better than 3.0 Å resolution, implying that phasing at low resolution was rather poor. The experimental phases of the selenomethionine data set were improved by density modification using DM [25]. Phases were applied to the native data set and also improved by density modification.

Model building and crystallographic refinement The experimental maps with phases computed directly from the MAD analysis were virtually uninterpretable. The density modified maps were clearer, but showed strong connectivity between sidechains and many breaks in the mainchain. Chain tracing and model building were carried out with computer graphics using the program O [65]. A combination of two density modified maps from the selenomethionine data and native data, calculated between 6.0 Å and 2.5 Å, had to be inspected for clear interpretation. The full-atom model was built in O

Table 3 MAD phasing statistics.* MAD diffraction ratios† 9.0–3.0 Å data

l1 l2

Scattering factors (e-)

3.0–2.2 Å data

l1

l2

l3

l4

l1

l2

l3

l4

0.025 (0.029) –

0.048

0.034

0.028

0.049

0.039

0.072 (0.030) –

0.032

0.051

0.035 (0.034) –

0.077 (0.030) –

0.036

–

0.075 (0.035) –

0.056 (0.028)

–

l3

–

l4

–

–

*MAD phasing: R(°FT(h)) = 0.047, R(°FA(h)) = 0.366 (R = SS||Fi(h)|–<|F(h)| >|/SS<|F(h)|>, for each reflection h, Fi(h) is the ith determination); = 32.18, = 17.80 (average difference between independent determinations of and ); = 0.767, R(°FAmodel) = 0.425 (R = S||°FA(h)|–|°FA,calc(h)||/S|°FA(h)|, where |°FA, calc(h)| is the structure factor calculated from the anomalous

–

f′(e-)

f′′(e-)

0.040

–4.5

0.56

0.036

0.053

–9.1

4.2

0.078 (0.037) –

0.042

–6.8

4.6

0.059 (0.035)

–3.9

3.2

scatterers). R(°FT(h)), R(°FA(h)), and are merging statistics for 6046 unique reflections with 34 648 good redundant measurements. †Bijvoet ratios: the diagonal elements (in parentheses) are values for centric reflections, indicating the level of noise in the anomalous signal, dispersive ratios between different wavelengths are the off-diagonal elements.

122

Structure 1997, Vol 5 No 1

References

Table 4 Refinement statistics. Data set

Selenomethionine

Native

5.0–1.95 7807 22.8 26.9 0.017 2.06 806 86

8.0–2.4 4011 21.6 28.5 0.015 2.40 811 38

Resolution range (Å) Number of reflections R factor (%) Rfree (%) Rmsd bond lengths (Å) Rmsd bond angles (°) Non-hydrogen atoms Water molecules

from approximate Ca positions and fitted to the electron density using real space refinement and geometry refinement. For refinement the remote low energy wavelength data collected to 1.95 Å resolution were used. The model was subjected to several cycles of molecular dynamics and restrained refinement with X-PLOR [66], PROLSQ [67] and manual rebuilding. Simulated annealing omit |Fo|–|Fc| maps, calculated for the entire polypeptide chain, confirmed the correctness of the model. The model includes amino acids 137– 235 and contains 806 protein atoms. Water molecules were placed in a subsequent round of restrained ARP refinement [68]. The water molecules were inspected with respect to B values, real space correlation and hydrogen bonds. The final model includes 86 water molecules and was refined with REFMAC [63] using 7807 reflections between 5.0 Å and 1.95 Å Bragg spacings to an R factor of 22.8 % with a free R factor of 26.9 % for 5 % of the data (Table 4). The refinement was carried out using data between 5.0–1.95 Å. This resolution range was chosen because data below 5.0 Å resolution were strongly overloaded and relatively incomplete. The agreement between model and low resolution data was not satisfactory, therefore these data were omitted. The program PROCHECK was used for assessment of model quality [69]. The protein model was independently refined against the native data set. 4011 reflections between 8.0–2.4 Å resolution were used in the refinement. The model was subjected to simulated annealing and restrained refinement with X-PLOR and PROLSQ and was rebuilt in O. The final model includes amino acids 136–235 and consists of 811 protein atoms and 38 water molecules and was refined to an R factor of 21.6 % with a free R factor of 28.5 % for 10 % of data (Table 4). The two refined models are very similar and the Ca carbons are essentially superimposable, having an average rms deviation of less than 0.3 Å. The structures are, however, translated by approximately 0.6 Å with respect to each other. This would explain a comparatively large isomorphous difference between the two data sets and was reflected in an uninterpretable difference Patterson map.

Accession numbers Coordinates for the refined model and structure factors have been deposited with the Protein Data Bank, Brookhaven, with accession code 1OPC.

Acknowledgements We thank C Ogata, R Abramowitz, G Anand, P Goudreau, A Santa Cruz and A West for assistance with data collection at the National Synchrotron Light Source (NSLS); H M Berman, M Georgiadis, S Harlocker and M Inouye for helpful discussions; E Arnold and P Goudreau for comments on the manuscript; and R Leidich and G Listner for X-ray technical assistance. This work was supported by grants from the NIH (GM47958), the NSF, and the Sinsheimer Foundation to AMS, and funding from the WM Keck Foundation for support of structural biology computing at the Center for Advanced Biotechnology and Medicine. AMS is an assistant investigator of the Howard Hughes Medical Institute.

1. Csonka, L.N. & Hanson, A.D. (1991). Prokaryotic osmoregulation: genetics and physiology. Annu. Rev. Microbiol. 45, 569–606. 2. Hall, M.N. & Silhavy, T.J. (1981). The ompB locus and the regulation of the major outer membrane porin proteins of Escherichia coli K12. J. Mol. Biol. 146, 23–43. 3. Forst, S.A., Delgado, J. & Inouye, M. (1989). Phosphorylation of OmpR by the osmosensor EnvZ modulates expression of the ompF and ompC genes in Escherichia coli. Proc. Natl. Acad. Sci. USA 86, 6052–6056. 4. Maeda, S. & Mizuno, T. (1990). Evidence for multiple OmpR-binding sites in the upstream activation sequence of the ompC promoter in E. coli: a single OmpR-binding site is capable of activating the promoter. J. Bacteriol. 172, 501–503. 5. Huang, K.J., Schieberl, J.L. & Igo, M.M. (1994). A distant upstream site involved in the negative regulation of the Escherichia coli ompF gene. J. Bacteriol. 176, 1309–1315. 6. Rampersaud, A., Harlocker, S.L. & Inouye, M. (1994). The OmpR protein of Escherichia coli binds to sites in the ompF promoter region in a hierarchical manner determined by its degree of phosphorylation. J. Biol. Chem. 269, 12559–12566. 7. Aiba, H., Nakasai, F., Mizushima, S. & Mizuno, T. (1989). Phosphorylation of a bacterial activator protein, OmpR, by a protein kinase, EnvZ, results in a stimulation of its DNA-binding ability. J. Biochem. 106, 5–7. 8. Forst, S. & Inouye, M. (1988). Environmentally regulated gene expression for membrane proteins in Escherichia coli. Annu. Rev. Cell Biol. 4, 21–42. 9. Pratt, L.A. & Silhavy, T.J. (1995). Porin regulon of Escherichia coli. In Two-Component Signal Transduction. (Hoch, J.A. & Silhavy, T.J., eds), pp. 105–127, American Society for Microbiology, Washington, DC, USA. 10. Kato, M., Aiba, H., Tate, S., Nishimura, Y. & Mizuno, T. (1989). Location of phosphorylation site and DNA-binding site of a positive regulator, OmpR, involved in activation of the osmoregulatory genes of Escherichia coli. FEBS Lett. 249, 168–172. 11. Tsung, K., Brissette, R.E. & Inouye, M. (1989). Identification of the DNA-binding domain of the OmpR protein required for transcriptional activation of the ompF and ompC genes of Escherichia coli by in vivo DNA footprinting. J. Biol. Chem. 264, 10104–10109. 12. Forst, S., Kalve, I. & Durski, W. (1995). Molecular analysis of OmpR binding sequences involved in the regulation of ompF in Escherichia coli. FEMS Microbiol. Lett. 131, 147–151. 13. Harlocker, S.L., Bergstrom, L. & Inouye, M. (1995). Tandem binding of six OmpR proteins to the ompF upstream regulatory sequence of Escherichia coli. J. Biol. Chem. 270, 26849–26856. 14. Pratt, L.A. & Silhavy, T.J. (1995). Identification of base pairs important for OmpR–DNA interaction. Mol. Microbiol. 17, 565–573. 15. Slauch, J.M., Russo, F.D. & Silhavy, T.J. (1991). Suppressor mutations in rpoA suggest that OmpR controls transcription by direct interaction with the alpha subunit of RNA polymerase. J. Bacteriol. 173, 7501–7510. 16. Sharif, T.R. & Igo, M.M. (1993). Mutations in the alpha subunit of RNA polymerase that affect the regulation of porin gene transcription in Escherichia coli K-12. J. Bacteriol. 175, 5460–5468. 17. Bowrin, V., Brissette, R., Tsung, K. & Inouye, M. (1994). The alpha subunit of RNA polymerase specifically inhibits expression of the porin genes ompF and ompC in vivo and in vitro in Escherichia coli. FEMS Microbiol. Lett. 115, 1–6. 18. Kato, N., Aiba, H. & Mizuno, T. (1996). Suppressor mutations in alphasubunit of RNA polymerase for a mutant of the positive regulator, OmpR, in Escherichia coli. FEMS Microbiol. Lett. 139, 175–180. 19. Russo, F.D., Slauch, J.M. & Silhavy, T.J. (1993). Mutations that affect separate functions of OmpR the phosphorylated regulator of porin transcription in Escherichia coli. J. Mol. Biol. 231, 261–273. 20. Pratt, L.A. & Silhavy, T.J. (1994). OmpR mutants specifically defective for transcriptional activation. J. Mol. Biol. 243, 579–594. 21. Kato, N., Tsuzuki, M., Aiba, H. & Mizuno, T. (1995). Gene activation by the Escherichia coli positive regulator OmpR: a mutational study of the DNA-binding domain of OmpR. Mol. Gen. Genet. 248, 399–406. 22. Hendrickson, W.A. (1991). Determination of macromolecular structures from anomalous diffraction of synchrotron radiation. Science 254, 51–58. 23. Main, P. (1985). MULTAN - a program for the determination of crystal structures. In Crystallographic Computing, 3: Data Collection, Structure Determination, Proteins, and Databases. (Sheldrick, G.M., Krüger, C. & Goddard, R., eds), pp. 206–215, Clarendon Press, Oxford, UK.

Research Article DNA-binding domain of OmpR Martínez-Hackert and Stock

24. Sheldrick, G.M. (1991). Heavy atom location using SHELXS-90. In Isomorphous Replacement and Anomalous Scattering: Proceedings of the CCP4 Study Weekend, 25–26 January 1991. (Wolf, W., Evans, P. & Leslie, A.G.W., eds), SERC Daresbury Laboratory, Warrington, UK. 25. Cowtan, K. (1994). ‘dm’: An automated procedure for phase improvement by density modification. In Joint CCP4 and ESFEACBM Newsletter Protein Crystallography 31, 34–38. 26. Kondo, H., et al., & Tanaka, I. (1994). Crystallization and X-ray studies of the DNA-binding domain of OmpR protein, a positive regulator involved in activation of osmoregulatory genes in Escherichia coli. J. Mol. Biol. 235, 780–782. 27. Martínez-Hackert, E., Harlocker, S., Inouye, M., Berman, H.M. & Stock, A.M. (1996). Crystallization, X-ray studies, and site-directed cysteine mutagenesis of the DNA-binding domain of OmpR. Protein Sci. 5, 1429–1433. 28. Buerger, M.J. (1960). Selection of material suitable for crystal structure analysis. In Crystal Structure Analysis. pp. 53–67, Wiley, NY, USA. 29. Holm, L. & Sander, C. (1994). The FSSP database of structurally aligned protein fold families. Nucl. Acids Res. 22, 3600–3609. 30. Wilson, K.P., Shewchuk, L.M., Brennan, R.G., Otsuka, A.J. & Matthews, B.W. (1992). Escherichia coli biotin holoenzyme synthetase/bio repressor crystal structure delineates the biotin- and DNA-binding domains. Proc. Natl. Acad. Sci. USA 89, 9257–9261. 31. Clark, K.L., Halay, E.D., Lai, E. & Burley, S.K. (1993). Co-crystal structure of the HNF-3/fork head DNA-recognition motif resembles histone H5. Nature. 364, 412–420. 32. Clubb, R.T., Omichinski, J.G., Savilahti, H., Mizuuchi, K., Gronenborn, A.M. & Clore, G.M. (1994). A novel class of winged helix-turn-helix protein: the DNA-binding domain of Mu transposase. Structure 2, 1041–1048. 33. Brennan, R.G. (1993). The winged-helix DNA-binding motif: another helix-turn-helix takeoff. Cell 74, 773–776. 34. Lai, E., Clark, K.L., Burley, S.K. & Darnell, J.E., Jr. (1993). Hepatocyte nuclear factor 3 / fork head or ‘winged helix’ proteins: a family of transcription factors of diverse biological function. Proc. Natl. Acad. Sci. USA 90, 10421–10423. 35. Harrison, S.C. (1991). A structural taxonomy of DNA-binding domains. Nature 353, 715–719. 36. Schultz, S.C., Shields, G.C. & Steitz, T.A. (1991). Crystal structure of a CAP–DNA complex: the DNA is bent by 90°. Science 253, 1001–1007. 37. Parkinson, G., Wilson, C., Gunasekera, A., Ebright, Y.W., Ebright, R.E. & Berman, H.M. (1996). Structure of the CAP–DNA complex at 2.5 Å resolution: a complete picture of the protein–DNA interface. J. Mol. Biol. 260, 395–408. 38. Ramakrishnan, V., Finch, J.T., Graziano, V., Lee, P.L. & Sweet, R.M. (1993). Crystal structure of globular domain of histone H5 and its implications for nucleosome binding. Nature 362, 219–223. 39. Harrison, C.J., Bohm, A.A. & Nelson, H.C.M. (1994). Crystal structure of the DNA-binding domain of the heat shock transcription factor. Science 263, 224–227. 40. Suzuki, M. & Brenner, S.E. (1995). Classification of multi-helical DNAbinding domains and application to predict the DBD structures of s factor, LysR, OmpR/PhoB, CENP-B, Rapl, and XylS/Ada/AraC. FEBS Lett. 372, 215–221. 41. Wintjens, R. & Rooman, M. (1996). Structural classification of HTH DNA-binding domains and protein–DNA interaction modes. J. Mol. Biol. 262, 294–313. 42. Aiba, H., Kato, N., Tsuzuki, M. & Mizuno, T. (1994). Mechanism of gene activation by the Escherichia coli positive regulator, OmpR: a mutant defective in transcriptional activation. FEBS Lett. 351, 303–307. 43. Kodandapani, R., et al., & Ely, K.R. (1996). A new pattern for helix-turnhelix recognition revealed by the PU.1 ETS-domain–DNA complex. Nature 380, 456–460. 44. Makino, K., et al., & Suzuki, M. (1996). DNA binding of PhoB and its interaction with RNA polymerase. J. Mol. Biol. 259, 15–26. 45. Aiba, H. & Mizuno, T. (1990). Phosphorylation of a bacterial activator protein, OmpR, by a protein kinase, EnvZ, stimulates the transcription of the ompF and ompC genes in Escherichia coli. FEBS Lett. 261, 19–22. 46. Kenney, L.J., Bauer, M.D. & Silhavy, T.J. (1995). Phosphorylationdependent conformational changes in OmpR, an osmoregulatory DNA-binding protein of Escherichia coli. Proc. Natl. Acad. Sci. USA 92, 8866–8870.

123

47. Tsuzuki, M., Aiba, H. & Mizuno, T. (1994). Gene activation by the Escherichia coli positive regulator, OmpR: phosphorylationindependent mechanism of activation by an OmpR mutant. J. Mol. Biol. 242, 607–613. 48. Harlocker, S. (1996). Characterization of the binding of OmpR, a transcription factor in Escherichia coli, to the upstream regulatory sequences of ompF. PhD thesis, Rutgers University, New Brunswick, NJ. 49. Guenot, J., Fletterick, R.J. & Kollman, P.A. (1994). A negative electrostatic determinant mediates the association between Escherichia coli Trp repressor and its operator DNA. Protein Sci. 3, 1276–1285. 50. Bourret, R.B., Hess, J.F. & Simon, M.I. (1990). Conserved aspartate residues and phosphorylation in signal transduction by the chemotaxis protein CheY. Proc. Natl. Acad. Sci. USA 87, 41–45. 51. Stock, J.B., Stock, A.M. & Mottonen, J.M. (1990). Signal transduction in bacteria. Nature 344, 395–400. 52. Mizuno, T., Kato, M., Jo, Y. & Mizushima, S. (1988). Interaction of OmpR, a positive regulator, with the osmoregulated ompC and ompF genes of Escherichia coli. Studies with wild-type and mutant OmpR proteins. J. Biol. Chem. 263, 1008–1012. 53. Tate, S., Kato, M., Nishimura, Y., Arata, Y. & Mizuno, T. (1988). Location of a DNA-binding segment of a positive regulator, OmpR, involved in activation of the ompF and ompC genes of Escherichia coli. FEBS Lett. 242, 27–30. 54. Stock, A.M., Mottonen, J.M., Stock, J.B. & Schutt, C.E. (1989). Threedimensional structure of CheY, the response regulator of bacterial chemotaxis. Nature 337, 745–749. 55. Volz, K. & Matsumura, P. (1991). Crystal structure of Escherichia coli CheY refined at 1.7 Å resolution. J. Biol. Chem. 266, 15511–15519. 56. West, A.H., Martínez-Hackert, E. & Stock, A.M. (1995). Crystal structure of the catalytic domain of the chemotaxis receptor methylesterase, CheB. J. Mol. Biol. 250, 276–290. 57. Baikalov, I., Schröder, I., Kaczor-Grzeskowiak, M., Grzeskowiak, K., Gunsalus, R. & Dickerson, R.E. (1996). Structure of the Escherichia coli response regulator NarL. Biochemistry 35, 11035–11061. 58. Simms, S.A., Keane, M.G. & Stock, J.B. (1985). Multiple forms of the CheB methylesterase in bacterial chemosensing. J. Biol. Chem. 260, 10161–10168. 59. Hendrickson, W.A., Horton, J.R. & LeMaster, D.M. (1990). Selenomethionyl proteins produced for analysis by multiwavelength anomalous diffraction (MAD): a vehicle for direct determination of three-dimensional structure. EMBO J. 9, 1665–1672. 60. Jancarik, J. & Kim, S.-H. (1991). Sparse matrix sampling: a screening method for crystallization of proteins. J. Appl. Cryst. 24, 409–411. 61. Cudney, R., Patel, S., Weisgraber, K., Newhouse, Y. & McPherson, A. (1994). Screening and optimization strategy for macromolecular crystal growth. Acta Cryst. D 50, 414–423. 62. Otwinowski, Z. (1993). Oscillation data reduction program. In Proceedings of the CCP4 Study Weekend: Data Collection and Processing. (Sawyer, L., Isaacs, N. & Bayley, S., eds), pp. 56–62, SERC Daresbury Laboratory, Warrington, UK. 63. SERC (UK) Collaborative Computational Project, No.4. (1994). The CCP4 suite: programs for protein crystallography. Acta Cryst. D 50, 760–763. 64. Otwinowski, Z. (1991). Isomorphous replacement and anomalous scattering. In Proceedings of the CCP4 Study Weekend. (Wolf, W., Evans, P.R. & Leslie, A.G.W., eds), pp. 80–86, SERC Daresbury Laboratory, Warrington, U.K. 65. Jones, T.A., Zou, J.Y. & Cowan, S.W. (1991). Improved methods for building protein models in electron-density maps and the location of errors in these models. Acta Cryst. A 47, 110–119. 66. Brünger, A.T., Kuriyan, J. & Karplus, M. (1987). Crystallographic R factor refinement by molecular dynamics. Science 235, 458–460. 67. Konnert, J.H. & Hendrickson, W.A. (1980). A restrained-parameter thermal-factor refinement procedure. Acta Cryst. A 36, 344–350. 68. Lamzin, V.S. & Wilson, K.S. (1993). Automated refinement of protein models. Acta Cryst. D 49, 129–147. 69. Laskowski, R.A., McArthur, M.W., Moss, D.S. & Thornton, J.M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Cryst. 26, 282–291. 70. Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Cryst. 24, 946–950. 71. Hutchinson, E.G. & Thornton, J.M. (1996). PROMOTIF — a program to identify and analyze structural motifs in proteins. Protein Sci. 5, 212–220.

124

Structure 1997, Vol 5 No 1

72. Carson, M. (1991). Ribbons 2.0. J. Appl. Cryst. 24, 958–961. 73. Nicholls, A., Sharp, K. & Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins 4, 281–296. 74. Comeau, D.E., Ikenaka, K., Tsung, K. & Inouye, M. (1985). Primary characterization of the protein products of the Escherichia coli OmpB locus: structure and regulation of synthesis of the OmpR and EnvZ proteins. J. Bacteriol. 164, 578–584. 75. Powell, B.S., Powell, G.K., Morris, R.O., Rogowsky, P.M. & Kado, C.I. (1987). Nucleotide sequence of the virG locus of the Agrobacterium tumefaciens plasmid pTiC58. Mol. Microbiol. 1, 309–316. 76. Makino, K., Shinagawa, H., Amemura, M. & Nakata, A. (1986). Nucleotide sequence of the phoR gene, a regulatory gene for the phosphate regulon of Escherichia coli. J. Mol. Biol. 192, 549–556. 77. Arthur, M., Molinas, C., Depardieu, F. & Courvalin, P. (1993). Characterization of Tn1546, a Tn3-related transposon conferring glycopeptide resistance by synthesis of depsipeptide peptidoglycan precursors in Enterococcus faecium BM4147. J. Bacteriol. 175, 117–127. 78. Miller, V.L., Taylor, R.K. & Mekalanos, J.J. (1987). Cholera toxin transcriptional activator ToxR is a transmembrane DNA-binding protein. Cell 48, 271–279.