The high-resolution crystal structure of the molybdate-dependent transcriptional regulator (ModE) from Escherichia coli: a novel combination of domain folds

Article15 March 1999free access The high-resolution crystal structure of the molybdate-dependent transcriptional regulator (ModE) from Escherichia coli: a novel combination of domain folds David R. Hall David R. Hall The Wellcome Trust Building, Department of Biochemistry, University of Dundee, Dundee, DD1 5EH UK Search for more papers by this author David G. Gourley David G. Gourley The Wellcome Trust Building, Department of Biochemistry, University of Dundee, Dundee, DD1 5EH UK Search for more papers by this author Gordon A. Leonard Gordon A. Leonard Joint Structural Biology Group, ESRF, BP 220, F-38043 Grenoble cedex, France Search for more papers by this author Elizabeth M.H. Duke Elizabeth M.H. Duke CCLRC-DL, Daresbury Laboratory, Warrington, Cheshire, WA4 4AD UK Search for more papers by this author Lisa A. Anderson Lisa A. Anderson Department of Biochemistry, University of Dundee, Dundee, DD1 5HN UK Search for more papers by this author David H. Boxer David H. Boxer Department of Biochemistry, University of Dundee, Dundee, DD1 5HN UK Search for more papers by this author William N. Hunter Corresponding Author William N. Hunter The Wellcome Trust Building, Department of Biochemistry, University of Dundee, Dundee, DD1 5EH UK Search for more papers by this author David R. Hall David R. Hall The Wellcome Trust Building, Department of Biochemistry, University of Dundee, Dundee, DD1 5EH UK Search for more papers by this author David G. Gourley David G. Gourley The Wellcome Trust Building, Department of Biochemistry, University of Dundee, Dundee, DD1 5EH UK Search for more papers by this author Gordon A. Leonard Gordon A. Leonard Joint Structural Biology Group, ESRF, BP 220, F-38043 Grenoble cedex, France Search for more papers by this author Elizabeth M.H. Duke Elizabeth M.H. Duke CCLRC-DL, Daresbury Laboratory, Warrington, Cheshire, WA4 4AD UK Search for more papers by this author Lisa A. Anderson Lisa A. Anderson Department of Biochemistry, University of Dundee, Dundee, DD1 5HN UK Search for more papers by this author David H. Boxer David H. Boxer Department of Biochemistry, University of Dundee, Dundee, DD1 5HN UK Search for more papers by this author William N. Hunter Corresponding Author William N. Hunter The Wellcome Trust Building, Department of Biochemistry, University of Dundee, Dundee, DD1 5EH UK Search for more papers by this author Author Information David R. Hall1,‡, David G. Gourley1,‡, Gordon A. Leonard2, Elizabeth M.H. Duke3, Lisa A. Anderson4, David H. Boxer4 and William N. Hunter 1 1The Wellcome Trust Building, Department of Biochemistry, University of Dundee, Dundee, DD1 5EH UK 2Joint Structural Biology Group, ESRF, BP 220, F-38043 Grenoble cedex, France 3CCLRC-DL, Daresbury Laboratory, Warrington, Cheshire, WA4 4AD UK 4Department of Biochemistry, University of Dundee, Dundee, DD1 5HN UK ‡D.R.Hall and D.G.Gourley contributed equally to experimental work *Corresponding author. E-mail: [email protected] The EMBO Journal (1999)18:1435-1446https://doi.org/10.1093/emboj/18.6.1435 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info The molybdate-dependent transcriptional regulator (ModE) from Escherichia coli functions as a sensor of molybdate concentration and a regulator for transcription of operons involved in the uptake and utilization of the essential element, molybdenum. We have determined the structure of ModE using multi-wavelength anomalous dispersion. Selenomethionyl and native ModE models are refined to 1.75 and 2.1 Å, respectively and describe the architecture and structural detail of a complete transcriptional regulator. ModE is a homodimer and each subunit comprises N- and C-terminal domains. The N-terminal domain carries a winged helix–turn–helix motif for binding to DNA and is primarily responsible for ModE dimerization. The C-terminal domain contains the molybdate-binding site and residues implicated in binding the oxyanion are identified. This domain is divided into sub-domains a and b which have similar folds, although the organization of secondary structure elements varies. The sub-domain fold is related to the oligomer binding-fold and similar to that of the subunits of several toxins which are involved in extensive protein–protein interactions. This suggests a role for the C-terminal domain in the formation of the ModE–protein–DNA complexes necessary to regulate transcription. Modelling of ModE interacting with DNA suggests that a large distortion of DNA is not necessary for complex formation. Introduction Molybdenum is an essential trace element used to provide a redox active centre in the molybdopterin cofactor of a number of enzymes (Rajagopalan, 1996; Kisker et al., 1997). This second row transition metal is bio-available as the water soluble molybdate, (MoO42−), and specific systems have evolved to regulate its uptake and to utilize its complex chemical properties in biochemical reactions. The modABCD operon of Escherichia coli codes for a high-affinity molybdate transport system that is a member of the ABC transporter family (Higgins, 1992; Pau et al., 1998), and in addition some molybdate may leach through the sulfate transporter. ModA encodes a periplasmic molybdate binding protein, the structure of which has recently been determined (Hu et al., 1997; Lawson et al., 1997). ModB and ModC are integral membrane proteins, the latter of which binds ATP. Little is known about ModD. Transcription of this operon is negatively controlled by the molybdate-responsive regulatory protein encoded by the modE gene. This gene is positioned immediately upstream and transcribed divergently from the modABCD operon (Grunden et al., 1996). A schematic depiction of MoO42− uptake and the regulation of the high-affinity transport system is given in Figure 1. ModE is a pleiotropic regulator of transcription since it appears to act as a positive regulator of the moaABCDE operon which encodes enzymes involved in molybdopterin biosynthesis (Walkenhorst et al., 1995; McNicholas et al., 1997, 1998a) and dmsABC, the structural operon for the molybdoenzyme DMSO reductase (McNicholas et al., 1998b). Figure 1.A schematic depiction of molybdate-dependent gene regulation in E.coli. Molybdate enters the bacterium via a high-affinity transporter system which comprises three proteins (ModA, ModB and ModC), and also, less efficiently, through the sulfate transporter which consists of CysP, T, W and A. Once in the cytoplasm the anion binds to the dimeric molybdate-dependent transcriptional activator (ModE), which in turn binds to and represses the modABCD operon. ModE also regulates the moaABCDE and dmsABC operons. Modified from Pau et al. (1998). Download figure Download PowerPoint ModE has been cloned, overexpressed and biochemically characterized (Anderson et al., 1997). The protein functions as a homodimer of subunits consisting of 262 amino acids. It displays a high affinity for MoO42− and WO42− (Kd = 0.8 μM), and the binding of the anion is accompanied by a large quench (50%) of the intrinsic protein fluorescence, producing a blue shift of the emission maximum. This suggests that a significant conformational change accompanies anion binding (Anderson et al., 1997). With anion bound, ModE binds near to the transcription start of the modABCD promoter and functions as a repressor. The intracellular availability of molybdate determines the conformation of ModE, regulating transcription of the transport system operon and thereby controlling the acquisition and subsequent use of the element. Analysis of the ModE sequence suggested two domains (Grunden et al., 1996). The N-terminal domain sequence carries a putative helix–turn–helix (HTH) motif frequently observed in DNA binding proteins (Brennan, 1993). The C-terminal residues present significant homologies with domains found in a range of bacterial proteins, in particular with a small (∼7 kDa molecular mass) molybdenum-containing protein from Clostridium pasteurianum termed Mop (Hinton and Freyer, 1986; Hinton and Merritt, 1986). Mop was originally thought to bind molybdopterin although this has not been confirmed. Studies with ModE strongly indicate that Mop and homologous domains share the ability to bind MoO42−. Such domains are of wide significance in the biology of molybdenum (Grunden et al., 1996; Anderson et al., 1997). The C-terminal domain of E.coli ModE contains a tandem repeat of the Mop sequence and we refer to this as the Di-Mop domain. A model for this C-terminal molybdate binding domain has been proposed based on secondary structure predictions (Lawson et al., 1997). It has been postulated that ModE is related to one of the most common families of prokaryotic transcription regulators, the LysR family (Anderson et al., 1997). This group of proteins controls the divergent transcription of linked genes or unlinked regulons encoding a range of functions such as nitrogen fixation, amino acid biosynthesis and bacterial virulence (Schell, 1993). The family has a DNA binding N-terminal region of ∼65 residues that includes a HTH motif and a co-inducer recognition domain. The co-inducers that stimulate LysR-type transcriptional regulators (LTTRs) are diverse molecular entities, including aromatic and aliphatic compounds, amino acid derivatives and ions. Binding of the co-inducer leads to LTTR–DNA association near the RNA polymerase binding site, DNA bending and transcription activation (Schell, 1993; Ansari et al., 1995). The structure of a complete LTTR has not yet been published although crystal structures of the co-inducer domain of CysB (Tyrrell et al., 1997) and the DNA binding domain of OmpR are available (Martinez-Hackert and Stock, 1997). We embarked upon a structural analysis of E.coli ModE to define an accurate three-dimensional model of the complete transcriptional regulator, and now report the details of native and a selenomethionine-derived protein at resolutions of 2.1 and 1.75 Å, respectively. The model illustrates the structure and organization of the subunits in the dimer, the basis for the dimerization, the orientation of DNA binding domains with respect to each other and the architecture of the Mop and Di-Mop domains. The putative MoO42− binding site and residues that might be important for interactions with DNA are also identified. Results and Discussion The ModE subunit The ModE monomer consists of two domains, the N-terminal domain (I) comprising all residues to 121, which interacts directly with DNA, and the C-terminal domain (II) which is the putative molybdate binding component. Domain II can be delineated into two sub-domains (a and b), corresponding to the two Mop sequence similarity regions. Sub-domain a consists of residues 122–183 and 256–262, whilst sub-domain b is formed from the contiguous stretch of residues 184–255. The folding of each domain and the assembly of each subunit of the dimer are illustrated in Figure 2. Secondary structure assignments have been made with a combination of automated methods (DSSP, Kabsch and Sander, 1983; PROMOTIF, Hutchinson and Thornton, 1996) and by visual inspection. Secondary structure elements are mapped onto the amino acid sequence in Figure 2D. Figure 2.Ribbon diagrams to depict the architecture of ModE. (A) Domains I and II. The helices of the HTH motif of domain I are coloured green. Sub-domains a and b of domain II are coloured blue and yellow. Residues Arg128, Lys183 and Trp186 are shown as sticks. (B) A ModE monomer. (C) The dimer with elements of secondary structure at the C-terminal domain assigned to aid orientation of this domain with respect to (A). Figures 2, 3, 4, 5 and 6 were generated with MOLSCRIPT (Kraulis, 1991) and RASTER3D (Merritt and Murphy, 1994). (D) The amino acid sequence with elements of secondary structure that have been assigned. Download figure Download PowerPoint The N-terminal DNA binding domain (I) is composed mainly of α-helices (60% α-helix, 20% β-strand). There are five α-helices and four β-strands, with the DNA binding region being a winged HTH motif comprising the secondary structure elements α2, α3, β3 and β4. Helix α4 (residues 80–108) forms the backbone of this domain. The C-terminal domain (II or Di-Mop) consists mainly of β-sheet (25% α-helix, 60% β-strand) and starts at strand β5. It consists of two β-barrels, the first of which is constructed from β6–β9 and β15. The second β-barrel is formed by β-strands 10–14 and this domain also contains a short segment of helix, α8. A turn of 310 helix and a short α helix (α9) are positioned at the junction of the two barrels. The ModE dimer A homodimer constitutes the asymmetric unit but there are significant deviations from non-crystallographic symmetry in the arrangement of monomers with respect to each other. There are two distinct non-crystallographic 2-fold axes relating the N-terminal domains with each other and the C-terminal domains with each other. The relationship of these axes to each other is shown in Figure 3. The reorientation of the domains occurs through the linker region (residues 115–126) where few interactions occur between domain I and domain II. There is a considerable dimer interface involving interactions between domain I subunits that constrains the dimer in this region in a rigid structure. McNicholas et al. (1998b) explored the role of the N- and C-terminal domains of ModE with respect to dimerization and DNA binding. They concluded that dimerization of ModE was required for efficient DNA binding, and that the C-terminal domain was primarily responsible for dimer formation. The crystallographic results indicate that it is the N-terminal domain that plays the major role in dimerization although the C-terminal domain does make a contribution (see below). The relationship of domain II to domain I, which is partially responsible for asymmetry in the homodimer, is discussed later in terms of the dimer interface and the crystal lattice. Figure 3.Stereoview Cα trace of the ModE dimer. Monomer A is red, monomer B is cyan and the black lines represent the non-crystallographic 2-fold axes relating each type of domain. Download figure Download PowerPoint Similar domain folds occur in other proteins The ModE structure presents a new combination of domains with previously observed folds. The model was submitted to the DALI server (http://www2.ebi.ac.uk/d…; Holm and Sander, 1993) to search for structural homologues. None were found for the ModE dimer or the intact monomer; however, fold similarities were observed separately for the C-terminal (Mop) sub-domains and the N-terminal DNA binding domain. As expected structures containing the winged and non-winged HTH moiety were identified. The N-terminal monomeric DNA binding region has greatest structural similarity to the diphtheria toxin repressor (DTR) monomer (Ding et al., 1996; White et al., 1998), which also has a winged HTH fold for DNA recognition and binding. For all Cα atoms, an r.m.s. deviation of 3.5 Å was observed, whilst for 47 Cα atoms between residues 34 and 100 (ModE numbering), an r.m.s. fit of 1.7 Å was observed. This region corresponds to the winged HTH motif plus the long backbone helix α4. Figure 4A illustrates the overlay of the monomeric DNA binding domain of ModE with the intact DTR molecule. Figure 4.Least-squares overlaps of components of the ModE structure with other protein fragments. (A) Domain I of ModE (red) with DTR (blue). (B) Domain II of ModE (red) with PT–ATP S4 (green) overlayed on sub-domain a, and PT S5 (black) overlayed on sub-domain b. Download figure Download PowerPoint The Di-Mop C-terminal domain has a novel assembly with no homologous structure being observed for this entire domain. Each of the two sub-domains aligned best with two chains of pertussis toxin (PT) and Cα traces of the superimposed structures are depicted in Figure 4B. These sub-domains display the oligonucleotide/oligosaccharide binding fold (OB-fold), which is based on a five-stranded Greek-key β-barrel capped by an α-helix located between the third and fourth strands. This fold has been observed in a number of proteins with little or no significant sequence homology (Murzin, 1993). Most of the binding site residues are from the loops protruding from the barrel axis and these loops differ in length, sequence and conformation. The OB-fold for sub-domain a of domain II of ModE is not formed in the same ordering of β-sheets/α-helix as that described for typical OB-folds (Murzin, 1993). Subunit S4 of the PT–ATP complex (Hazes et al., 1996) aligned best with sub-domain a (residues 127–182, 256–262) with all Cα atoms having an r.m.s. fit of 3.1 Å. Subunit S5 of PT (Stein et al., 1994) aligned best with sub-domain b (residues 183–255), with all Cα atoms having an r.m.s. fit of 3.0 Å. These PT subunit structures are similar six-stranded anti-parallel β-sheets capped by an α-helix with an OB-fold. The strand arrangement of sub-domain b of ModE closely matches that of subunit S5 of PT, whilst sub-domain a has five of the six strands of subunit S4 of PT. The residues 136–144 of ModE form a loop instead of the related β-strand of subunit S4 of PT. Both sub-domains of the Di-Mop region of ModE have the capping helix although it is shorter than those observed in S4 and S5 of PT. The spatial relationship of S4 with S5 subunits in PT does not match that of the sub-domains in the Di-Mop region of ModE. However, that both S4 and S5 are involved in extensive protein–protein interactions within the PT structure suggests that the Di–Mop fold may interact with other proteins in order to carry out its function, a point that is discussed later. The OB-fold was found for other structural matches using DALI. These homologues included the bacterial enterotoxins such as verotoxin-1 (Stein et al., 1992), toxic shock syndrome toxin-1 (Papageorgiou et al., 1996), heat-labile enterotoxin (van den Akker et al., 1996) and the bacterial superantigen Streptoccocus pyrogenic exotoxin C (Roussel et al., 1997). All these proteins have the OB-fold either as a subunit of the biological unit or as part of a larger monomeric chain. The dimer interface There are extensive contacts between the two subunits of the ModE dimer but these are mainly due to interactions between the N-terminal domains which account for almost 70% of the dimer interface. This suggests that the HTH motifs are held in a relatively fixed conformation in the ModE dimer. The dimer presents a solvent accessible area of 23 795 Å2 and the buried surface interactions between monomers occupies 6017 Å2, with some 44 residues from subunit A and 46 residues from subunit B closer than 3.6 Å to their partner subunit. Of these, 38 residues are common to each subunit and form 41 inter-subunit hydrogen bonds using nitrogen and oxygen atoms from both main chains and side chains. The interactions are localized in two regions of the N-terminal domain I and at one region in the Di-Mop domain II (Figures 2C and 3). In domain I they are found between strand β1 of both subunits, resulting in an antiparallel two-stranded β-sheet, and helix α4 of both subunits, which has an aliphatic hydrophobic core. Most of the helix–helix contacts are formed at the opposing ends of these long helices. Differences in hydrogen bonding patterns between subunits are observed at these helix–helix contact regions. An inter-subunit salt bridge is observed between Asp106A (A and B are used to identify the subunits where a distinction is necessary) and Arg80B, and this contributes to a strained conformation for the acidic residue. Due to the asymmetry of the dimer, Asp106B has no such interaction. Almost all of the inter-subunit contacts formed by the residues in the C-terminal Di-Mop domain are formed between strand β5 of one subunit and β14 of the partner subunit. This association occurs between two anti-parallel β-strands which, in conjunction with another from each monomer (strand β8), creates two three-stranded antiparallel β-sheets that form a β-barrel at the centre of the dimer interface. Domain I–domain II interactions and the influence of the crystal lattice There are only a few associations between domain I and domain II, and these consist of nine hydrogen bonds and one salt bridge, all of which occur between residues of the same subunit. Thus, there are no interactions between subunit A domain I and subunit B domain II or between subunit B domain I and subunit A domain II. The lack of contacts between domains permits a degree of flexibility in the organization of the dimer which may be necessary for the function of ModE. The orientation of domain II with respect to domain I is influenced by crystal lattice contacts between symmetry-related molecules. Each dimer contacts 10 symmetry-related dimers using five different interfaces. There are 56 potential hydrogen bonds between symmetry-related molecules as well as solvent mediated interactions and van der Waals contacts. There are however, only a few interactions between domain I with its symmetry partners and these contacts mainly involve sub-domain b of domain II. Helix α3, strands β3 and β4 of chain A interact with strands β11 and β14 of a symmetry-related chain B. Helix α3 of chain B interacts with helices α7, α8 and α9 of a symmetry-related chain B, and helix α7 of chain A interacts with residues immediately following strand β1 in chain A of a symmetry-related molecule. The average temperature factor per residue distribution for each subunit is similar in the DNA binding N-terminal domain, and is reasonably similar in the Di-Mop region although subunit B has slightly higher average temperature factors overall. This is borne out by the average temperature factors for domain I of subunits A and B (35 and 39 Å2, respectively) and for domain II of subunits A and B (43 and 44 Å2, respectively). The Di-Mop domain The OB-like fold of the sub-domain of this region has already been discussed and the arrangement of the sub-domains with respect to each other and their organization in the dimer are presented in Figure 2. Residues that are strictly conserved in the Di-Mop domain of ModE and the Mop domains of other proteins are Val147, Ile162, Ser166, Glu177, Lys183, Val187, Asn201, Glu218 and Gly244. The polar residues are worthy of further comment. Ser166A Oγ has no interactions but Ser166B Oγ forms a hydrogen bond (3.15 Å) with Nη1 of Arg169B. In turn, Arg169B forms three hydrogen bonds to strand β5 of subunit A. Glu177 has differing interactions in each subunit and is involved in interaction between domains II and I. Glu177A Oϵ1 hydrogen bonds to Oγ of Ser113A whilst Glu177B Oϵ1 forms a salt bridge (2.57 Å) with the Nϵ1 of Arg120B and Glu177B Oϵ2 forms a hydrogen bond (2.99 Å) with Nη2 of Arg120B. This is another example of the asymmetry observed between the domains of the two subunits. Lys183 is the only other conserved positively charged side chain found in the Di-Mop domain. The Nζ atom of this residue interacts with the carbonyl of Asp254 of the opposing subunit and is therefore involved in dimerization. Asn201, in both subunits, forms hydrogen bonds with Lys159 and Phe251 of the same subunit. The interactions with Lys159 are between sub-domain a and sub-domain b. Asn201 is located in strand β11 and Phe251 in strand β14. Hydrogen bonds between the main-chain atoms of these residues form part of the β-barrel of sub-domain b. As mentioned earlier, the two Mop-like sub-domains (a and b) are interdependent for their structure. This is a result of sub-domain a being formed from two stretches of polypeptide chain (residues 128–183 and 256–262). This contrasts with the assignment of the Mop-like domains from sequence analysis (Walkenhorst et al., 1995; Lawson et al., 1997), whereby each sub-domain was assigned to continuous sequence. Sub-domain b is formed from continuous sequence (residues 184–255). The contacts between the sub-domains in each subunit are similar and include the residues Arg128 and Asn129 from the conserved sequence SARN. This tetrapeptide sequence (residues 126–129) has been implicated in dimer formation (McNicholas et al., 1998b) and molybdate binding (Grunden et al., 1996; Lawson et al., 1997). The putative molybdate binding site The ModE dimer binds two molybdates (Anderson et al., 1997), and by consideration of residues that are conserved in ModE and Mop proteins, their position in the structure and their chemistry, we can assign a molybdate binding site. The residues of interest are Arg128, which is contributed from the SARN segment, Ser166, Lys183 and Glu218. These residues are strictly conserved in ModE and Mop domain sequences. Arg169 is conserved between E.coli ModE and a ModE homologue identified in Haemophilus influenzae ORF7 (Fleischmann et al., 1995), but in Mop sequences is changed to either a glutamic acid or asparagine. Thr232 is present in all ModEs and is homologous to a serine found in C.pasteurianum Mop. The positions of Arg128 and Lys183 are shown in Figure 2A to indicate their position at the junction of subunits a and b. The asymmetry of the ModE structure results in distinct arrangements of the two putative MoO42− binding sites in the homodimer (Figure 5). The only positively charged and conserved residues are found near the interface of the sub-domains. In the periplasmic molybdate binding protein, ModA, the oxyanion binding site is created in a deep cleft at the junction of four helices oriented to use the helix macrodipole (Hol et al., 1978). There are seven hydrogen bonds donated from uncharged donor groups to the oxyanion provided by four main-chain amides, hydroxyl groups from two serines and finally a tyrosine Nδ (Hu et al., 1997; Lawson et al., 1997). The binding site in ModA is different from that of ModE although the possible conservation of serine/threonine–MoO42− interactions is noted. This is not unexpected since this residue type is typically found in binding sites for another oxyanion, namely phosphate (Copley and Barton, 1994). Figure 5.(A) Stereoview to show the sub-domain IIa interface with sub-domain IIb and residues postulated to bind molybdate or to be influenced by oxyanion-induced conformational changes. Dashed lines represent hydrogen bonding interactions. (B) The same view as (A) but illustrating secondary structure with elements labelled as for Figure 2C. Download figure Download PowerPoint Ser126 is positioned at the end of strand β5 and the amide forms a hydrogen bond with Ile258 O at the beginning of strand β15 of the opposing subunit. The sheet formed by these two strands is responsible for dimer interactions in domain II. Asn129 as found in both subunits forms hydrogen bonds with Leu182 and Cys230 of the same subunit. Glu218A forms a hydrogen bond (2.96 Å) to the symmetry-related Arg138A. This constrains Glu218A, permitting its Oϵ2 atom to form intramolecular salt bridges to Nη1 (2.52 Å) and Nη2 (2.46 Å) of Arg128A, a residue implicated in molybdate binding (Lawson et al., 1997). The related site in the partner subunit has no such lattice contact. If molybdate binds to Arg128 and to Lys183 then this would disrupt the interactions between sub-domains. The movement of sub-domains with respect to each other would in turn influence the environment of Trp186 given its location at the interface of the sub-domains. The side chain of Trp186A is present in a double conformation (not shown). One side chain conformer has its Cη2 atom 3.2 Å from the Cη2 atom of Trp186B. Alterations in the environment of this particular residue following molybdate binding would account for the large fluorescence quenching observed on the addition of the oxyanion (Anderson et al., 1997). How does ModE bind to DNA? DNA recognition and binding of ModE to the modA promoter involves recognition of a palindromic sequence (Anderson et al., 1997). The dimerization of the protein provides the 2-fold related residues necessary for interaction with such a DNA motif. Ansari et al. (1995) surmised that the LTTR MerR, when in repressor mode or closed conformation was able to bend DNA towards itself in a similar fashion to the bacterial catabolite activator protein, CAP. The activated or open form induced unbending and provided the template for transcription. This was proposed as a paradigm for how the LTTR family operate. A comparison of residues conserved, in terms of structural position and chemistry, on the HTH motifs of ModE, CAP and DTR gives some insight into components of ModE that may be involved in DNA binding. The structures of CAP (Schultz et al., 1991; Parkinson et al., 1996) and DTR (White et al., 1998) have been solved in complex with oligonucleotides and indicate that DNA–protein interactions occur between helix α3 and the major groove, and a conserved serine of the turn binds to the phosphate backbone. From the CAP, DTR and ModE sequence alignments and by visual inspection of the ModE model (Hall, 1999) we can predict which ModE residues might interact with DNA. These are Ser35, Gln36, Lys39, Ser44 (conserved in all examples considered here), Tyr45 and Lys46 (CAP has interactions with DNA at the equivalent residues whilst DTR has glycine and proline which can have no side chain interactions), Ser47, Trp49 and Asp50. When CAP binds to its target, the DNA is distorted significantly (reviewed by Kolb et al., 1993). Such a large distortion is not observed when DTR is complexed to DNA (White et al., 1998). A 30 bp B-DNA duplex model was generated with the program INSIGHT based on the promoter region of modA (Anderson et al., 1997). A ModE–DNA model was assembled using knowledge of DTR– an