The solution configurations of inactive and activated DntR have implications for the sliding dimer mechanism of LysR transcription factors

LysR Type Transcriptional Regulators (LTTRs) regulate basic metabolic pathways or virulence gene expression in prokaryotes. Evidence suggests that the activation of LTTRs involves a conformational change from an inactive compact apo- configuration that represses transcription to an active, expanded holo- form that promotes it. However, no LTTR has yet been observed to adopt both configurations. Here, we report the results of structural studies of various forms of the LTTR DntR. Crystal structures of apo-DntR and of a partially autoinducing mutant H169T-DntR suggest that active and inactive DntR maintain a compact homotetrameric configuration. However, Small Angle X-ray Scattering (SAXS) studies on solutions of apo-, H169T- and inducer-bound holo-DntR indicate a different behaviour, suggesting that while apo-DntR maintains a compact configuration in solution both H169T- and holo-DntR adopt an expanded conformation. Models of the SAXS-obtained solution conformations of apo- and holo-DntR homotetramers in complex with promoter-operator region DNA are consistent with previous observations of a shifting of LTTR DNA binding sites upon activation and a consequent relaxation in the bend of the promoter-operator region DNA. Our results thus provide clear evidence at the molecular level which strongly supports the ‘sliding dimer’ hypothesis concerning LTTR activation mechanisms.

(d) The current hypothesis is that activation of LTTRs leads to a change in conformation from compact to expanded homotetrameric configurations.
Scientific RepoRts | 6:19988 | DOI: 10.1038/srep19988 morph from a compact to an expanded homotetrameric configuration (Fig. 1b-d)) thus allowing the DBD dimers to bind to different DNA sites. However, while the crystal structures (Fig. 1b,c) of various full-length LTTR homotetramers adopt either compact (i.e. CbnR 14 , DntR 5 , BenM 15 ) or expanded configurations (i.e. TsaR 7 , ArgP 12 , AphB 16 , OxyR 13 ) no full-length LTTR homotetramer has yet been observed to adopt both. Moreover, in the only case where the crystal structures of a LTTR are available in both apo-and inducer-bound forms (TsaR 7 ) both exhibit the same expanded configuration.
To provide further insights into the sliding dimer hypothesis and the conformational changes this might involve, we carried out structural studies of the LTTR DntR, obtaining the crystal structures of inactive apo-DntR and of a partially autoinducing H169T mutant (H169T-DntR). Furthermore, we determined, using Small Angle X-ray Scattering (SAXS), models of the solution states of apo-, H169T-and inducer-bound (i.e. fully activated, salicylate-bound) holo-DntR homotetramers. The crystal structures of apo-and H169T-DntR suggest that both inactive and activated DntR homotetramers have the same compact configuration and that no large conformational changes are required for DntR activation. However, SAXS studies of apo-, H169T-and holo-DntR reveal a completely different picture, suggesting that in solution apo-DntR maintains a compact quaternary configuration but that both H169T-DntR and holo-DntR homotetramers adopt an expanded conformation. Moreover, putative models bound to promoter DNA regions appear to support the physiological relevance of the solution conformations of apo-and holo-DntR obtained. These models suggest that a shift of binding sites, accompanied by a relaxation of DNA bend, would occur upon activation of DntR. Our results show that a single LTTR can adopt different conformations, depending on activation state, providing compelling structural evidence to support the sliding dimer mechanism for the activation of homotetrameric LTTRs and confirming that this involves a change in quaternary structure from compact to expanded conformations.

Results and Discussion
The crystal structures of apo-DntR and H169T-DntR. The crystal structure of apo-DntR is isostructural with those of DntR in complex with either acetate or thiocyanate (Table 1, Fig. 2) 5 . As for acetate-or thiocyanate-bound DntR, the asymmetric unit (a.u) contains two DntR molecules with a homotetramer adopting a compact configuration (Fig. 2a) being constructed by the association of two symmetry-related dimers. However, in contrast to the crystal structures of acetate/thiocyanate-bound DntR the crystal structure reported here shows well-defined electron density for the wHTH regions. The crystal structure of apo-DntR thus represents the first of a full length DntR homotetramer, although this is extremely similar to the model of a full-length thiocyanate-bound DntR homotetramer 5 produced by Smirnova and colleagues (r.m.s. deviation in C α positions of 1.04 Å for 1190 residues aligned).
The IBCs in the crystal structure of apo-DntR are devoid of any ligands (Fig. 2b), confirming the apo-nature of the homotetramers. Indeed, a comparison of the apo-DntR IBC conformation with those of the IBCs of 'open-hinge region' salicylate-bound DntR RDs 6 ( Fig. 2c) shows that if DntR IBCs maintain the conformation seen in the crystal structure of apo-DntR they cannot bind salicylate and DntR cannot be activated. The crystal structure of apo-DntR thus seems to confirm that inactive apo-DntR homotetramers adopt a compact homotetrameric configuration. Interestingly, a comparison of the configuration of apo-IBCs with those of the IBCs seen in the crystal structures of acetate-or thiocyanate-bound DntR (Fig. 2d) shows them to be very similar. This suggests that acetate/thiocyanate molecules bound in IBCs of DntR homotetramers serve to reinforce the apo-IBC configuration rather than, as previously suggested 5 , acting as mimics of bound salicylate.
The crystals of H169T-DntR we obtained show very similar unit cell dimensions and the same space group to those of apo-DntR (Table 1). Similar to apo-DntR, in the crystal structure of H169T-DntR contains a dimer in the a.u. with a homotetramer adopting a compact configuration formed by the association of two symmetry-related dimers. The quality of the electron density map for H169T-DntR did not allow the building of the wHTH regions, hence the final model obtained essentially includes only that of a thiocyanate-bound ( Fig. 2e) H169T-DntR RD homotetramer. This adopts an identical quaternary structure to the central RD homotetramer seen in crystals of apo-DntR (r.m.s.d. in C α positions, 0.36 Å; 428 atoms superposed in the RD dimers seen in the a.u.). However, a superposition of the crystal structure of H169T-DntR with that of 'open-hinge' inducer-bound DntR RDs 6 ( Fig. 2f) shows, in contrast to what we report for apo-DntR above, the conformations of their IBCs to be very similar. This observation seems to provide a rational explanation as to why the H169T mutation autoinduces transcription (Fig. 2g) and suggests that upon activation DntR homotetramers could maintain a compact configuration.
The solution conformations of apo-and H169T-DntR. The conclusion, based on the crystal structures described above, that both inactive and active forms of DntR homotetramers maintain a compact configuration is inconsistent with the results of thermal stability analyses (TSAs) of apo-, H169T-and holo-DntR (Table 2). These show the melting temperature (T m ) of H169T-DntR is increased by 9 °C relative to that of apo-DntR, suggesting that, in solution, there are larger conformational differences between apo-and H169T-DntR homotetramers than are seen in the crystal structures described above.
It is generally accepted that, in many cases, the packing interactions required for the formation of crystals can lock a particular macromolecule in a single conformation and there are an increasing number of examples where solution structures reveal physiologically relevant conformations of biological macromolecules that are unobtainable in crystals [17][18][19][20] . We therefore examined the crystal packing interactions for both apo- (Fig. 3a) and H169T-DntR and found these interactions constrain the H169T-DntR tetramer to adopt a compact configuration similar to that adopted by apo-DntR. In order to verify whether these compact configurations represent the physiological states of apo-DntR and H169T-DntR homotetramers, we carried out SAXS experiments ( For apo-DntR homotetramers in solution, the Guinier region of the scattering curve obtained yields a radius of gyration of 39 Å and analysis of the pair distribution function (P(r)) derived from the scattering curve obtained suggests a maximum intramolecular distance of 118.2 Å. This latter is significantly shorter than the maximum distance (130.5 Å) observed in the apo-DntR crystal structure and suggests, as does the poor agreement (χ 2 = 2.1) of a fit (Fig. 4a) of the experimentally measured scattering curve and a scattering curve calculated using the homotetramer seen in the crystal structure, that the conformation of apo-DntR homotetramers in solution is different to that observed in the crystal. To test whether, in solution, apo-DntR homotetramers adopt an expanded configuration form similar to that seen in the crystal structure of TsaR 7 (Fig. 1b) we constructed a homology model of such a homotetramer using Swiss-Model 21 . A fit of the calculated scattering curve based on this model and the experimental scattering curve also results in a rather poor agreement (χ 2 = 1.9, Fig. 4a). Therefore, in order to obtain a model of the conformation of apo-DntR homotetramers in solution we employed rigid body refinement procedures. The resulting model provides very good fits to both the ab-initio molecular envelope obtained from our SAXS experiments (Fig. 5a) and to the experimental scattering curve (χ 2 = 1.0; Fig. 4a) and shows that in solution apo-DntR maintains a compact tetrameric conformation. However, and as expected from the value of D max derived from the P(r) function, the solution shape of apo-DntR is less elongated than is seen in the crystal with the DBD dimers -the positions of which may also be influenced by lattice constraints in the crystal -packed closer to the tetrameric core (Fig. 5a,b).

apo-wtDntR
H169T-DntR    In SAXS analysis of solutions of H169T-DntR the Guinier region of the scattering curve obtained yields a radius of gyration value of 42 Å, larger than the 39 Å observed in solution for apo-DntR. D max (135 Å) as derived from the pair distribution (P(r)) function is also significantly larger than that seen in solution for apo-DntR (118 Å). SAXS experiments thus indicate that the solution shape of H169T-DntR homotetramers is different to that of inactive apo-DntR homotetramers. This observation is reinforced by the results of fits of the experimental scattering curve of H169T-DntR to curves calculated from various structural models (Fig. 4b). Here, calculated curves based on either the solution (χ 2 = 3.4) or crystal (χ 2 = 2.4) structures of apo-DntR reported here result in very poor matches. However, a scattering curve based on our homology model of a TsaR-type expanded apo-DntR homotetramer provides an improved fit (χ 2 = 1.5). This suggests that in solution H169T-DntR adopts a conformation similar to that of the expanded homotetramers seen in the crystal structure of TsaR. As H169T-DntR activates transcription even in the absence of its inducer molecule salicylate, implying that it more easily adopts an activated conformation than does apo-DntR, these results indicate that an expanded conformation is the activated state of DntR and, as previously hypothesised 7 , of LTTR family homotetramers in general.
The solution model of holo-DntR. We have been unable to crystallise full-length DntR in the presence of salicylate and therefore examined the solution conformation of holo-(i.e. active) DntR in a series of SAXS experiments. The invariant parameters (R g , I (0), D max , Porod Volume) derived from scattering curves obtained and the resulting P(r) functions are shown in Table 2. Incubation of apo-DntR with 20 μ M sodium salicylate produces no significant change compared to apo-DntR while incubation with either 100 μ M or 5000 μ M sodium salicylate yields increased values of R g and D max . However, values of I(0) and Porod volume in the presence of 5000 μ M sodium salicylate increase markedly compared to those obtained for apo-DntR, indicating the presence of sufficient inter-particle effects to make this scattering curve unreliable for further analysis. This is in contrast to values obtained following incubation with 100 μ M sodium salicylate. The scattering curve (Fig. 3a) obtained for apo-DntR incubated with 100 μ M sodium salicylate was therefore used in further analysis of the solution conformation of holo-DntR.
For holo-DntR a fit of the Guinier region (Fig. 3b) of the scattering curve yields a radius of gyration of 41.4 Å, similar to that obtained for H169T-DntR in solution but larger than that seen for apo-DntR. D max (142 Å) is also similar to that obtained for H169T-DntR in solution and significantly larger than is seen for apo-DntR (Fig. 3c). As for H169T-DntR, comparisons of the experimental scattering curve of holo-DntR with scattering curves based on the solution (χ 2 = 2.2) and crystal structures (χ 2 = 2.1) of apo-DntR homotetramers show poor fits while that with a curve calculated using the TsaR type homology model of a DntR tetramer is clearly improved (χ 2 = 1.6) (Fig. 4c). This suggests, again as for H169T-DntR, that the solution conformation of holo-DntR homotetramers is closer to the expanded configuration observed for the crystal structure of TsaR than to either of the solution or crystal structures of apo-DntR described here.
To obtain a clearer idea of the solution structure of activated holo-DntR homotetramers rigid body modelling was employed. While the scattering curve calculated from the resulting model provides a good fit (χ 2 = 1.2) to the experimental curve (Fig. 4d), poor agreement at higher q values indicates local differences between this model and the true solution structure. A plausible explanation is that the RDs treated as rigid bodies in the modelling procedure derive from the crystal structure of acetate-/thiocyanate bound DntR which, as has been shown above (Fig. 2c,d), cannot bind salicylate. In an attempt to obtain a better fit these were replaced with RDs observed in the crystal structure of doubly-salicylate-bound holo-Δ N90DntR (PDB 2Y7K, chain A and B) 6 . Here, the IBCs are expanded to accommodate salicylate and the RD is slightly closed around the ligand. This results in a model for the solution structure of holo-DntR which provides an excellent fit to the ab-initio molecular envelope obtained (Fig. 5b) and for which the calculated scattering curve provides a very good fit to the experimental scattering curve over the entire q range (χ 2 = 1.0, Fig. 4d). Our SAXS experiments thus indicate that, in solution, both H169T-DntR (autoinducing) and holo-DntR (fully activated) homotetramers do not adopt the compact conformation seen in solution for apo-DntR homotetramers but that both adopt similar, TsaR-type expanded configurations. This observation is consistent with TSA measurements showing that both H169T-DntR and holo-DntR have similar T m s which are significantly higher than that seen for apo-DntR (Table 2). Our SAXS experiments thus show the solution conformation of activated DntR to be that of an expanded form homotetramer with an open central cavity and very reminiscent in nature to the conformation seen in the crystal structure of TsaR.

Models of the binding of the solution conformations of apo-and holo-DntR to promoter region DNA.
DntR, and other members of the LTTR family, are bound to DNA in both inactive and active states 5,11 . Activation of transcription by LTTRs must therefore be coupled to conformational differences between active and inactive states prompted by the binding of inducer molecules. The results of our SAXS studies (Fig. 5) suggest, as previously hypothesised 6,7 , that this conformational change involves a transformation from a compact homotetrameric configuration which represses expression to an expanded configuration that promotes it.
Our SAXS studies of holo-DntR also suggest, as we have previously postulated 6 , that the driving force for the transition from compact to expanded homotetrameric configurations is a closure upon inducer binding of the RDs making up the core of the tetramer which, if the compact form of the homotetramer were maintained, would result in steric clashes between RD2 subdomains at the tetramer interface. Our SAXS studies also indicate that upon activation, as might be expected from an analysis of the crystal structure of TsaR 7 which the SAXS-obtained solution conformation of holo-DntR closely resembles, DntR homotetramers adopt a much flatter configuration than is seen for apo-DntR (Fig. 5). In this activated conformation, the separation between the recognition helices in the HTH dimers flanking the central RD tetrameric core is significantly increased compared to that seen in solution for apo-DntR, suggesting both different promoter region DNA binding sites for inactive and activated DntR and a relaxation in the bend of promoter region DNA upon the transition from the inactive to active conformations of DntR.
In the absence of bound inducer molecules, inactive tetramers of the LTTR BenM binds to two palindromic sites (RBS and ABS') on p benA promoter region DNA repressing expression. Upon activation, DNA binding sites shift to the RBS and ABS" sites of p benA and transcription occurs 11 . A similar 'sliding' of the DNA binding sites of inactive and active LTTRS has also been observed for OccR 22 and OxyR 23 . While the exact nature of the DntR RBS and ABS' and ABS" binding sites in its p dnt promoter region DNA have yet to be determined, alignment of p dnt with the nucleotide sequences of p benA and p nahr (the latter the promoter-operator region of another homologous LTTR, NahR) suggests that these will be found in similar regions in all three promoter regions (Fig. 6a). In order to verify that the models obtained here for the solution conformations of apo-and holo-DntR are physiologically relevant, we constructed models of both species bound to p benA .
Here, the 3D-DART server (http://haddock.science.uu.nl/dna/dna.php) 24 was used to generate series of models of p benA (75bp in length) each with a total DNA bending angle distributed homogenously along the DNA sequence. The most convincing DNA model allowing interaction between the solution structure of apo-DntR and both the RBS and ABS' binding sites of p ben A was obtained with a DNA displaying a bend angle of 240° (Fig. 6b). In this model, as seen in the crystal structure of wHTH DNA binding domain (DBDs) of BenM in complex with RBS of p benA 25 , the recognition helices and wings of both wHTH dimers interact with consecutive major and minor grooves of each binding site.
A similar approach was performed for holo-DntR (Fig. 6b) with the resulting model suggesting that a 94° bending of the p benA region is required to satisfy the simultaneous interaction of its RBS and ABS" binding sites with the two wHTH dimers flanking the heterotetrameric core. Here, however, our model predicts that the ABS" binding region of activated DntR is shifted more towards the − 10 region than is predicted for BenM. This observation may explain why no conserved ABS" binding motif has yet been deduced for LTTRs 26 .
The models described above are consistent both with a sliding of DntR promoter region binding sites and a significant relaxation of promoter region DNA bend upon the activation of LTTRs [27][28][29] and suggest the solution conformations obtained here for apo-and holo-DntR are physiologically relevant. However, while the bending angles of bound promoter-operator region DNA (94 o for active DntR, 240° for its inactivated counterpart) deduced from our models are not inconsistent with relatively high degrees of DNA bend that can induced in short DNA sequences 30-32 they are much more pronounced than those measured for LTTRs in circular permutation experiments (0-50° and 50-100° for active and inactivated LTTRs, respectively) 8 . This is intriguing and we therefore constructed a second series of models (Fig. 6c) to try to predict how LTTR homotetramers might bind to DNA promoter-operator regions with lower bending angles. Our model in which the p benA DNA promoter-operator region is bent by 50 o (i.e. close to that measured for active LTTR homotetramers) shows that simultaneous binding of RBS and ABS" can be achieved by a homotetrameric LTTR although this will have a conformation very different to those described here for activated or inactivated DntR. Our model of LTTR binding to p benA with a DNA bending angle in the region of 100° (i.e. close to that measured for inactive LTTRs) shows that simultaneous binding of RBS and ABS' regions cannot be achieved if DntR, and thus other LTTRs, maintain a homotetrameric conformation. Indeed, this model predicts that in their inactive states LTTRs will bind to RBS and ABS' as dimers and implies that the activation process involves changes in the affinity of a LTTR dimer such that ABS" is preferred for binding over the ABS' and that binding to the ABS" results in the formation of a LTTR homotetramer. Given that LTTRs generally associate as homotetramers 4,8 when not bound to DNA this scenario appears unlikely and our 'high-angle' DNA models (Fig. 6b) thus seem both to confirm the physiological relevance of the solution structures of inactive and activated DntR described here and to indicate, as has been previously suggested 33 , that the bending angles reported for LTTR promoter-operator regions based on circular permutation experiments are underestimated.
Previous studies have shown small basal (i.e. in the absence of inducer molecules) levels of transcription in genes placed upstream of the promoter region controlled by DntR 34 . This might suggest that in vivo apo-DntR exists in equilibrium between the compact configurations shown in Fig. 5a and the expanded configuration shown in Fig. 5b. Both our modelling of the DNA binding of the solution conformations we have obtained for the repressor form of DntR (apo-state) and activator-form form of DntR (holo-form) (see above) and DNA footprinting/binding assays accumulated on LTTRs 11,22,35 do not seem consistent with this hypothesis. However, we do not rule out that apo-DntR exists in an equilibrium of inactive and active conformations with the fraction of the latter too small to be detected in our SAXS experiments. We also do not rule out that apo-DntR can exist in an equilibrium of conformations between a form that completely represses transcription and a form that incompletely represses it. A possibility for a conformation that incompletely represses transcription is that seen in the crystal structures of apo-DntR and H169T-DntR reported here. Again, however, we see no evidence for such an equilibrium of conformations in our SAXS analysis of apo-DntR.

Conclusions
The current consensus is that LTTRs regulate transcription through large conformational changes which modify LTTR DNA binding sites promoter region DNA. Previous structural studies 6,7,12 have inferred that the change in question is a transition from compact to expanded form tetramers but real structural evidence for this idea has thus far been scarce. In particular, no full-length LTTR homotetramer has yet been observed to adopt both conformations. The results presented here provide the first clear evidence both that a single LTTR homotetramer can adopt both conformations and that the conformation adopted by the homotetramer is a function of activation state. In solution, inactive apo-DntR homotetramers adopt a compact configuration in which DBD dimers pack closely against a compact RD tetrameric core (Fig. 5a) while activated DntR homotetramers adopt a more open quaternary configuration (Fig. 5b) highly reminiscent of the configuration seen in the crystal structure of TsaR. That these configurations of active and inactive DntR are physiologically relevant is supported by models of the binding of these two distinct homotetrameric arrangements to promoter region DNA which are consistent both with a sliding of DntR binding sites and a significant relaxation of DNA bend upon activation. Given the level of structural similarity of LTTR family members this work appears to confirm, as previously proposed 12 , that a switch from compact to expanded configurations is likely to be a general activation mechanism for the largest family of transcription factors found in prokaryotes. However, a number of fundamental questions remain to be answered. In particular, the high angles of the bending LTTR promoter region DNA predicted in the models presented here are incommensurate with those measured for other LTTRs in permutation experiments. Further studies will clearly be required to fully validate our conclusions.

Methods
Cloning, expression and purification. DntR-His 6 (DntR) from Burkholderia sp. was expressed and purified as previously described 5 with the exceptions that two pellets of cOmplete Protease Inhibitor Cocktail Tablets (Roche; http://www.roche-applied-science.com) per litre of culture were added during cell lysis and that purification was conducted at 4 o C. To produce H169T-DntR-His 6 (H169T-DntR) site-directed mutagenesis using PCR primers containing the mutation was carried out (Stratagene, La Jolla, USA; forward primer 5′ cggcgcctctttcg-cacccgctacgtatgcat3′ ; reverse primer 5′ atgcatacgtagcgggtgcgaaagaggcgccg3′ ). H169T-DntR was then expressed and purified as for wild-type DntR. Once purified, both proteins were stored at various concentrations in a buffer comprising 1 M NaCl, 2 mM MgSO 4 , 1 mM DTT, 17% (v/v) glycerol, 25 mM NaH 2 PO 4 -NaOH (pH 8.0). These solutions were then used as stocks in both crystallisation protocols and SAXS experiments.  (Table 1) were collected on ID29 (apo-DntR) and ID23-1 (H169TDntR) of the European Synchrotron Radiation Facility (ESRF 36,37 ). Diffraction images were integrated and scaled using XDS 38 , symmetry-related intensities merged using SCALA 39 and structure factors derived using TRUNCATE 40 . For both apo-DntR and H169T-DntR structure solution was carried out using Molecular Replacement (MR) in the program PHASER 41 using the crystal structure of acetate-bound DntR (PBD ID, 1UTB 5 ) stripped of water molecules and other ligands as a search model. Structure refinements (Table 1) were carried out in REFMAC5 42 and PHENIX 43 interspersed with rounds of manual rebuilding in COOT 44 . The crystal structure of apo-DntR was refined to d min = 2.64 Å resulting in a model providing R work = 19.9% and R free = 22.7%. The crystal structure of H169T-DntR was refined to d min = 3.30 Å and resulted in a model providing R work = 18.8% and R free = 24.0%.
Small-angle X-ray Scattering (SAXS). SAXS measurements were carried out on ESRF beamline ID14-3 45 at λ = 0.931 Å or ESRF beamline BM29 46 at λ = 0.992 Å. Scattering curves were recorded in the momentum transfer range 0.04 < q < 0.61 Å −1 (q = 4π sin (θ )/λ , 2θ is the scattering angle) using a Pilatus 1M detector (Dectris Ltd., Baden, Switzerland). Prior to experiments all solutions were centrifuged at 16,000 xg for 10 minutes to remove aggregated particles. In static SAXS experiments 30 μ L of protein solution loaded into a sample capillary using a liquid handling robot were exposed to X-rays and scattering data collected using multiple exposures (see below for details of number and length). Matched buffer measurements taken before and after every sample measurement were averaged and used for background subtraction. Individual exposures were processed automatically and independently using PyFAI 47 as implemented in the EDNA framework yielding radially averaged curves of normalized intensity versus scattering angle. For each exposure series additional data reduction used the automatic data processing tools of EMBL-Hamburg ATSAS package 48 to combine individual measurements, exclude any data points affected by aggregation induced by radiation damage, and yield an average scattering curve. Merging of scattering curves obtained at different sample concentrations and downstream analysis were performed manually as described in the literature 49 . Here, the forward scattering I(0) and radius of gyration (Rg) were calculated from the Guinier approximation 50 . Particle volume and the maximum particle size (D max ) were determined from the pair distribution function P(r) as computed by GNOM 51 using PRIMUS 52 . Scattering curves from solutions of apo-DntR (buffer as for the stock solution described above) were measured at sample concentrations of 3, 1 and 0.5 mg/mL. For all concentrations, images were recorded using 10 exposures each of 10 seconds and the scattering curves obtained at each concentration scaled and averaged to produce the final solution scattering curve. Ab initio models of the solution shape of apo-DntR were derived from the experimental scattering curve using DAMMIF 53 , imposing P22 symmetry as well as standard P1. To produce a final molecular envelope 20 independently generated ab initio models were aligned, averaged and filtered using DAMAVER 54 . Theoretical scattering curves of apo-DntR and the TsaR type homology model were calculated and used to generate fits against experimentally-obtained scattering curves using CRYSOL 55 . Rigid body modelling of the model of the solution conformation of apo-DntR was carried out in the program SASREF 56 . Here, using the previously postulated model of the structure of a full-length DntR tetramer 5 as a basis, the four RDs and two DBD dimers that make up a LTTR tetramer (Fig. 1) were treated as separate rigid entities and their positions refined against the experimental scattering curve employing a connectivity restraint of 3Å between residues T85 and A86 of each DntR monomer. Scattering curves from solutions of apo-H169T-DntR (same buffer as for the stock solution described above) were measured at a sample concentration of 0.7 mg/mL. 10 × 10 frames each of 2 seconds exposure time were recorded and averaged. Given the limited resolution of the scattering curve obtained rigid body modelling of the solution conformation of apo-H169T-DntR was not attempted. The solution shape of holo-DntR was elucidated from a scattering curve obtained from a solution of apo-DntR pre-incubated with 100 μ M sodium salicylate. Ab initio models of the solution conformation of holo-DntR were obtained as for apo-DntR with the exception that no symmetry restrictions were imposed. Rigid body modelling of the solution conformation of holo-DntR was carried out as for apo-DntR. To verify that results obtained from static SAXS experiments using solutions of apo-DntR and of holo-DntR were not artefacts due to interparticle interactions, experiments were repeated, this time coupled with size exclusion chromatography (SEC, Viscotek RImax, Malvern; Superdex 200 column) immediately prior to sample injection 57 . Solutions were applied to a gel filtration column and the eluent exposed directly to the X-ray beam. For each solution applied to the column 3500 frames were collected. Individual scattering curves yielding similar values of R g were then processed and averaged. The scattering curves obtained were rather noisy as the concentration of proteins in the eluent is low. Nevertheless, it was possible to obtain satisfactory Guinier regions and P(r) plots from which invariant parameters were derived ( Table 2). These are near identical to those obtained in our static SAXS measurements, confirming that models for the solution conformations of apo-DntR and of holo-DntR derived from higher q-range experiments are valid.