Article

Sandra P. San

0022-2836/© 2017 Elsevi

SAXS Structural Studies of Dps fromDeinococcus radiodurans Highlightsthe Conformation of the MobileN-Terminal Extensions

tos1, Maxime G. Cuypers2

, Adam Round3, 4, Stephanie Finet 5,Theyencheri Narayanan2, Edward P. Mitchell 2 and Célia V. Romão1

1 - ITQB-NOVA, Instituto de Tecnologia Química e Biológica António Xavier, Universidade Nova de Lisboa, Avenida da República,2780-157 Oeiras, Portugal2 - ESRF- The European Synchrotron, CS40220, 38043 Grenoble Cedex 9, France3 - European Molecular Biology Laboratory, Grenoble Outstation, 38042 Grenoble, France4 - Faculty of Natural Sciences, Keele University, Staffordshire ST5 5BG, UK5 - Institut de Minéralogie, de Physique des Matériaux et de Cosmochimie, CNRS-UPMC-IRD-MNHN, 4 place Jussieu,75252 Paris CEDEX 5, France

Correspondence to Célia V. Romão: ITQB-NOVA, Instituto de Tecnologia Química e Biológica António Xavier,Universidade Nova de Lisboa, Avenida da República, Estação Agronómica Nacional, 2780-157 Oeiras, Portugal.cmromao@itqb.unl.pthttp://dx.doi.org/10.1016/j.jmb.2017.01.008Edited by Helene Hodak

Abstract

The radiation-resistant bacterium Deinococcus radiodurans contains two DNA-binding proteins from starvedcells (Dps): Dps1 (DR2263) and Dps2 (DRB0092). These are suggested to play a role in DNA interaction andmanganese and iron storage. The proteins assemble as a conserved dodecameric structure with structurallyuncharacterised N-terminal extensions. In the case of DrDps1, these extensions have been proposed to beinvolved in DNA interactions, while in DrDps2, their function has yet to be established.The reported data reveal the relative position of the N-terminal extensions to the dodecameric sphere in

solution for both Dps. The low-resolution small angle X-ray scattering (SAXS) results show that the N-terminalextensions protrude from the spherical shell of both proteins. The SAXS envelope of a truncated form ofDrDps1 without the N-terminal extensions appears as a dodecameric sphere, contrasting strongly with theprotrusions observed in the full-length models. The effect of iron incorporation into DrDps2 was investigatedby static and stopped-flow SAXS measurements, revealing dynamic structural changes upon iron binding andcore formation, as reflected by a quick alteration of its radius of gyration.The truncated and full-length versions of DrDps were also compared on the basis of their interaction with

DNA to analyse functional roles of the N-terminal extensions. DrDps1 N-terminal protrusions appear to bedirectly involved with DNA, whilst those fromDrDps2 are indirectly associated with DNA binding. Furthermore,detection of DrDps2 in the D. radiodurans membrane fraction suggests that the N-terminus of the proteininteracts with the membrane.

© 2017 Elsevier Ltd. All rights reserved.

Introduction

DNA-binding proteins from starved cells (Dps)belong to the ferritin family and have been identifiedonly in prokaryotes. The function of these proteins hasbeen associated mainly with iron storage and DNAprotection against degradation promoted by reactiveoxygen species (e.g., Refs. [1–3]). The structure ofthese proteins is quite conserved. They are composed

er Ltd. All rights reserved.

of 12 identical monomers, assembling together to forma roughly spherical dodecamer with 23 symmetry. Thishollow shell has around 90 Å and 50 Å external andinternal diameters, respectively. Each subunit is afour-helix bundle, as in ferritins, but contains a furthershort helix between helix B and C and lacks the fifthhelix at the C-terminus, which is responsible for the4-fold symmetry in ferritins (e.g., see reviews in Refs.[1–5]).

J Mol Biol (2017) 429, 667–687

668 SAXS Structural Studies of Dps from D. radiodurans

Although the dodecameric core is structurallyconserved, the Dps family members have a varietyof either shorter or longer extensions of the aminoacid chain at the N- or C-termini. These extensionshave been proposed to be involved in DNA inter-action and condensation and are thought to beimportant for the DNA protection mechanism ofDps. For example, the Dps from Escherichia coli(EcDps) presents an N-terminal extension of 20 aaresidues with four positively charged residues:three lysine residues (Lys5, Lys8, and Lys10) andone arginine (Arg18). Truncating the first 18 aaprevents DNA condensation [6]. It has beentherefore proposed that this region is implicated inDNA interaction. The model of EcDps resulting fromelucidation of its crystal structure does not includethe full N-terminus, suggesting a highly flexibleregion that could not be modelled [7]. In the Dpsfrom Mycobacterium smegmatis (MsDps1), theextension occurs at the C-terminus, which contains26 aa residues. This region has three positivelycharged lysine residues (Lys172, Lys177, andLys181), which have also been suggested to beinvolved in DNA binding [8]. Again, the crystalstructure model of this protein does not include theextension; this disorder was suggested to be relatedwith local flexibility [9]. As for EcDps, the presenceof the positively charged amino acid residues andthe flexibility of this region could contribute to theinteraction and condensation of DNA. On the otherhand, for both Dps from Lactococcus lactis, whichhave an N-termini of 26 aa residues, it was possibleto model this region in the corresponding crystalstructures, since it forms a stable helix, and thedeletion of this region in DpsA impairs DNA binding[10].Interestingly, both Dps from Deinococcus radio-

durans (DrDps) have longer N-terminal extensionscompared to any other Dps family members studiedso far. Before the helix A of the four-helix bundle,DrDps1 and DrDps2 contain a total of 54 and 41residues, respectively (Fig. 1).The N-terminus of DrDps1 can be divided into two

regions (Fig. 1): a positively charged region (resi-dues 1–29), which contains six lysine residues (Lys3,Lys4, Lys7, Lys13, Lys15, and Lys16) and one arginine(Arg28) and has been proposed to be involved in theassociation with DNA [11]. The second region(residues 30–55) contains a metal binding site withthe motif Asp36x2His39x10His50x4Glu55, which islocated on the external surface of the dodecameraccessible to the solvent. The disruption of this metalsite affects the dodecamer assembly and alsoreduces DNA binding ability [12], but its nature isnot yet fully established, since different metals havebeen found to bind to this site, including zinc andcobalt [13,14].DrDps2 contains a signal peptide with 30 aa

residues, which makes this protein unique since the

presence of a signal peptide has not been reportedfor other Dps homologues. It has been suggestedthat under in vivo conditions, this signal peptidedirects DrDps2 to a non-cytoplasmic localisation[15]. This region is probably cleaved, and the matureprotein has 41 aa residues before the start of the firsthelix (the amino acid residue numbering does notaccount for the signal peptide) (Fig. 1). The functionof the N-terminal extension has yet to be addressed,but since it only harbours one positively chargedresidue, Arg33, it is unlikely to be the DNA-bindingregion. However, at the C-terminus, this proteincontains an extension of 24 aa residues, with threepositively charged residues: one lysine and twoarginines (Lys204, Arg209, and Arg211). ThisC-terminal extension has been proposed to beinvolved in dodecamer formation [15]. Although itsrole in DNA binding has not been addressed to date,the presence of positively charged residues in thisregion strongly hints that it could be involved in DNAinteraction as in MsDps1.As mentioned above, the N-terminal extension of

DrDps1 is suggested to be involved in the interactionof Dps with DNA, whilst in the case of DrDps2, thefunction remains unknown. However, a significantpart of these N-terminal extensions could not bemodelled from the DrDps crystal structures obtainedto date, probably due to protein degradation,disorder, or flexibility of these regions; DrDps1lacks the first 29 aa residues, and DrDps2 crystalstructure lacks the totality of its 41 residues [14,16]. Itis crucial to determine the structural conformationand localisation relative to the dodecamer sphere ofthese regions in solution. Solution small angle X-rayscattering (SAXS) is part of the suite of structuraltechniques available to explore the properties ofbiological macromolecules and provides unique,low-resolution conformational information, whichsometimes cannot be obtained from other tech-niques (in this case, crystallography).Therefore, solution scattering studies were per-

formed on both DrDps, aiming to explore thestructural conformation of these regions in solution.SAXS studies have been previously described forthe Dps from Porphyromonas gingivalis [17], butonly to characterise the spherical dodecamershape. This current work describes the locationand shape of the N-terminal flexible extensions ofDrDps proteins, which protrude outside from thedodecameric sphere. The aim was to obtain the abinitio shape from SAXS data and contribute to theunravelling of the function of these extensions.Since previous studies have shown that DrDps2behaves as a stable iron-storage dodecamerprotein [18], the short-term structural effect of theaddition of iron to DrDps2 was explored by staticand stopped-flow time-resolved SAXS, revealingstructural changes upon iron binding and iron coreformation.

Fig. 1. Structural alignment of Dps1 and Dps2 from D. radiodurans. DrDps secondary structure is shown above the alignment, in which the hN represents a helixpresent in the N-terminal region of DrDps1, and the hC represents an helix present in the C-terminal region of DrDps2. The strictly conserved amino acids arerepresented as black boxes, ferroxidase ligands are represented as red boxes, zinc metal ligands are represented in grey boxes, and positive charged residuespresented in the N- or C-termini are represented as blue boxes. For DrDps1, the N-terminal region that was truncated is represented in pink. The two regions from theseN-terminal extensions, the positively charged region and metal binding site region, are represented in blue and grey, respectively. For DrDps2, the signal peptide isrepresented in black, and the truncated region is shown in green.

669SAXSStructuralS

tudiesof

Dps

fromD.radiodurans

670 SAXS Structural Studies of Dps from D. radiodurans

Results

SAXS data from DrDps usingonline chromatography

SAXS data were measured at the ESRF (Grenoble,France) BioSAXS beamline, BM29 [19], using anonline size-exclusion chromatography (SEC) columncoupled with the standard experimental setup [20].Full-length DrDps1 and DrDps2 and their truncatedconstructs were analysed in order to confirm the dataanalysis and interpretations obtained from the full-length proteins. The DrDps1 truncated form lacks thefirst 50 aa residues (DrDps1tΔ1–50), while the DrDps2truncated form (DrDps2tΔ1–39) lacks the initial 39 aaresidues (numbering does not take in considerationthe 30 residues from the signal peptide; Fig. 1).Analysis of the elution profile obtained from the SEC

indicates that DrDps1 elutes as two oligomeric forms,corresponding to dodecamer and trimer, and DrDps2elutes as a stable dodecamer, as previously described[18]. The truncated form DrDps1tΔ1–50 presented asimilar elution profile as the full-length protein.DrDps2t-Δ1–39 showed thepresenceof several oligomeric states;however, the dodecameric assembly was only presentin small quantities, making it impossible to determine astructural model for this oligomeric state. Therefore, thesamples carried forward for in-depth analysis wereDrDps1, DrDps1tΔ1–50, and DrDps2, using SAXS dataobtained from the dodecamer fraction.The three proteins show similar SAXS scattering

profiles with typical features for a spherical particle(Fig. 2A). Analysis of the Kratky plot indicates that theproteins were correctly folded (data not shown). TheGuinier plots of each scattering profile provide a linearfit for amonodisperse solution (Fig. 2B), and the radiusof gyration (Rg) was calculated from this approxima-tion. The value of Rg is similar for both DrDps1 andDrDps2, 4.15 nmand 4.25 nm, respectively, whilst thevalue forDrDps1tΔ1–50 is smaller at 3.88 nm (Table 1).Based on the pair distance distribution function

[P(r)] determined by the indirect Fourier transformfrom the scattering profile curves using GNOM [21],it was possible to determine the molecular mass foreach sample and compare it with the calculatedtheoretical molecular mass determined from theknown sequences (Table 1). The P(r) was deter-mined using the data of the q range b 2 nm−1, sinceabove q N 2 nm−1, the data contain significant noise(Fig. 2A). The P(r) profile of both DrDps1 andDrDps1tΔ1–50 showed a single maximum at a radius(r) of ~5.5 nmwith a shoulder at around 2.2 nm, whileDrDps2 showed a single maximum at ~5.0 nm with aless pronounced shoulder at ~2.0 nm (Fig. 2C).These profiles are characteristic of a sphericalshape with a hollow core [22]. The value of 5.5–5.0 nm corresponds to the external radius of theproteins,while the value of 2.0–2.2 nmcorresponds to

the internal radius. The values are consistent with thecrystal structures, in which the external radius for bothproteins is approximately 4.5 nm and the internalradius is around 2.0 nm. Furthermore, theP(r) profilesfor the two full-length proteins present an extended tailwith maximal particle size (Dmax) of 12.75 nm and12.74 nm for DrDps1 and DrDps2, respectively,probably related to the presence of the longN-terminal tails in these two proteins (Fig. 2C). Thisfeature is notably absent from the P(r) profile forDrDps1tΔ1–50, whichhasaDmax of 10.00 nm (Fig. 2C).The Rg values determined using the P(r) are similar

to those determined using the Guinier approximation,which demonstrates consistency on the results (Table1). Moreover, the Dmax obtained for DrDps1tΔ1–50 issimilar to the diameter obtained in the crystalstructure, which is approximately 9 nm [14].

DrDpsSAXSab initioderivedmolecular envelopes

The molecular envelopes for each protein wereobtained by ab initio modelling using the programsDAMMIN and GASBOR (dummy atoms: 5858 ±51for DrDps1, 5895 ± 32 for DrDps1tΔ1–50, 5858 ±14 for DrDps2) using default parameters andimposing P23 symmetry, consistent with the sym-metry observed for the oligomer in solution, that is,the dodecameric hollow sphere with 23 symmetry.The knowledge of the oligomer symmetry is critical,as trials using P1 symmetry did not sufficientlyrestrain the modelling and did not yield an interpret-able model. The q range used to generate themodels was q b 2 nm−1. The results using bothprograms were identical, and only the modelsobtained with GASBOR are presented. A total of10 ab initio reconstructed models were averagedand filtered to obtain the final model for each sample.The normalised spatial discrepancy value of a set ofGASBOR models for DrDps1, DrDps1tΔ1–50, andDrDps2 were 0.77 ± 0.043 (Rf = 0.02 ± 0.002),0.90 ± 0.10 (Rf = 0.008 ± 0.001), and 0.87 ± 0.08(Rf = 0.01 ± 0.002), respectively (Fig. 3A1, B1,and C), with final χ2 values against corrected dataof 11.0, 10.6, and 3.0 for DrDps1, DrDps1tΔ1–50, andDrDps2, respectively (Table 1).The crystal structure of DrDps docks well into the

SAXS molecular envelope generated from the corre-sponding experimental scattering data. However, theSAXS envelopes determined for DrDps1 and DrDps2present extra areas (Fig. 3A1 and B1) protruding fromthe sphere, which are attributed to the N-terminalregions and were not observed in the crystal structure.Importantly, these protrusions are not observed in theSAXS envelope corresponding to DrDps1tΔ1–50, con-sistent with the absence of the first 50 aa residues inthis protein construct. The crystal structure of DrDps1docks into the spherical envelopeofDrDps1tΔ1–50 (Fig.3C). The models obtained for DrDps1, DrDps2, andDrDps1tΔ1–50 are superimposed in the central region of

Fig. 2. SAXS profile of DrDps1, DrDps1tΔ1–50, and DrDps2. (A) Experimental (dotted line) and calculated (solid line)scattering curves. (B) Guinier plot and linear fit (solid line) to the data for the Guinier region, qmax*Rg b 1.3. (C) P(r) profile.In all panels, DrDps1 is represented in blue, DrDps1t Δ1–50 in green, and DrDps2 in red.

671SAXS Structural Studies of Dps from D. radiodurans

the dodecamer, but the extra areas protruding thesphere are only observed in the full-length proteins,DrDps1 and DrDps2.The N-terminal regions from Dps are known to be

flexible regions but in the sense that they changeconformation upon different conditions; for thisreason, it has not been possible to structurallycharacterise these regions by X-ray crystallographywhere they were disordered, perhaps due to crystalpacking. The low-resolution models obtained showin solution that these regions protrude from thedodecamer sphere as observed in both modelsobtained from GASBOR/DAMMIN for DrDps1 andDrDps2. These models correspond to the averageshape attributed to the N-terminal regions that couldnot be observed in the crystal structure.It is important to mention that using GNOM

program followed by GASBOR/DAMMIN to obtainthe final ab initio models from scattering curves of a

flexible sample may lead to some possible errors.This is due to the over-background adjustment of thescattering curve, since GNOM tries to force intensitydecay to q−4 (globular shape) using a constant [21].Thus, analysis using only GNOM should be takenwith caution. In order to complement our studies,molecular envelopes for each protein were obtainedusing ensemble optimisation method (EOM), anapproach that was developed to characterise proteinmobility [23] (described below).

EOM to flexible systems

The results showed that by using EOM, it generatesspecific structural assemblies of the N-terminal tailsdemonstrating that these regions have a high degreeof mobility, in which the 12 individual regionsprotruding from the dodecamer hollow sphere havedifferent conformations (Fig. 3A2 and B2).

Table 1. Data collection and scattering-derived parameters from the DrDps1, DrDps1tΔ1–50, and DrDps2 using the SECcoupled to the SAXS detector

DrDps1 DrDps1tΔ1–50 DrDps2

Data collection parametersBeamline BM29, ESRFBeam geometry 700 × 700 μmWavelength (Å) 0.9919q range (nm−1) 0.08–4.5 nm−1

Exposure time (s) 1Structural parameters

I(0) (from Guinier) 50.55 52.25 123.88Rg (nm) (from Guinier) 4.15 3.88 4.25I(0) (cm−1) [from P(r)]⁎ 50.75 52.14 121.90Rg (nm) [from P(r)]⁎ 4.14 3.83 4.17Dmax (nm) [from P(r)]⁎ 12.75 10.00 12.74Porod volume estimate (nm3; +/− 10%) [from P(r)]⁎ 447.74 290.96 497.67Rg (nm) (from EOM)⁎ 4.14 3.70 4.88Dmax (nm) (EOM)⁎ 14.16 10.50 18.84Porod volume estimate (nm3;EOM)⁎ 436.85 339.46 445.08

Molecular mass estimationMolecular mass Mr. (from Porod volume;+/− 10%;kDa) 279.8+/−8 181.8+/−18 311.1+/−21Molecular mass Mr. (from EOM;kDa) 273.1 213.3 278.1Calculated dodecamer Mr. from sequence (kDa) 276.2 216.4 279.9

CRYSOL (χ2)experimental scattering data 12.82 10.24 21.44EOM scattering data (EOM) 0.64 − 0.32

Model (χ2)GASBOR 11.0 14.6 3.03EOM 11.90 − 12.39

SoftwarePrimary data reduction (circular averaging) PRIMUSData processing PRIMUSAb initio analysis GASBOR, DAMMIN, and EOMComputation of pdb model intensities CRYSOLThree-dimensional graphics representation Pymol

Accession Codes (SASBDB) SASDBG7 SASDBH7 SASDBF7

The qRg range used was less than 1.3.⁎ The data range used was q b 2 nm−1.

672 SAXS Structural Studies of Dps from D. radiodurans

The different parameters were also obtained usingEOM, namely Rg, Dmax, and molecular mass estima-tion (Table 1). The Rg determined for DrDps1,DrDps1tΔ1–50, and DrDps2 are similar to thoseobtained from Guinier plots and P(r). However, theDmax obtained using EOM for both DrDps1 andDrDps2 is higher than those determined by P(r),14.16 and 18.84, respectively, in contrast to those forDrDps1tΔ1–50 (Dmax = 10.50;Table 1). The Dmax dif-ference corresponds to the different conformations ofthe N-terminal tail observed in the models generatedby EOM, showing 12 individual regions protrudingfrom the dodecamer hollow sphere (Fig. 3A2 and B2).The estimation of the molecular mass for each proteinis similar to the theoretical molecular mass (Table 1).

Comparison of SAXS scattering curves andcrystal structure models

The comparison of scattering experimental datafor the full-length DrDps proteins with the corre-sponding crystal structure curves using CRYSOLgives a fit (χ2 value) of 12.8 and 21.4 for DrDps1 and

DrDps2, respectively (Table 1). Similar χ2 valueswere obtained using the program FoXS [24]. Thesehigh χ2 values are associated with structuraldifferences between the crystal structure and theproteins in solution, which could be related to thepresence of the N-terminal regions. The χ2 value forDrDps1tΔ1–50 at 10.2 is lower when compared withthe DrDps1 crystal structure (modelled without the29 residues). Further comparison with a truncatedmodel of DrDps1, the deletion of the total region ofthe N-terminus (residues 30–50) in the crystalstructure, yields an even lower χ2 of 9.7. Thedecrease from 10.2 to 9.7 is due to the fact that thecrystal structure lacks the first 29 residues, while thetruncated model used to obtain the SAXS envelopelacks 50 residues.Moreover, the scattering curves obtained from the

crystal structures were compared using CRYSOLwith the scattering curves of each DrDps generatedby EOM (fit for the best ensemble). In this case, theχ2 values obtained are lower than those presentedabove, 0.64 and 0.32 for DrDps1 and DrDps2,respectively (Fig. 4).

Fig. 3. DrDps ab initio models superimposed with the corresponding crystal structures. (A1 and B1) Ab initio shapegenerated by DAMAVER from GASBOR models for (A1) DrDps1 and (B1) DrDps2, represented as blue mesh and redmesh, respectively. (A2 and B2) Molecular envelope obtained by using EOM for (A2) DrDps1 and for (B2) DrDps2,represented through the 3-fold N-terminus (left) and 3-fold C-terminus (right). (C) Ab initio shape generated by DAMAVERfrom GASBOR models for DrDps1tΔ1–50 represented as green mesh. DrDps crystal structures are represented in cartoonwith each monomer in a different colour.

673SAXS Structural Studies of Dps from D. radiodurans

SAXS data from DrDps2 in the presence of iron

DrDps2 is a stable dodecamer and it behaves as atypical iron-storage protein being able to incorporateup to 500 iron atoms per dodecamer (Fe/dodecamer)in its cavity, which is highly negatively charged [18]. Inorder to address the question of the iron coreformation in DrDps2, SAXS data were measured onthe ID02 beamline at ESRF [25]. The experimentalconditions used include 50% (wt/vol) glycerol as

described in the Experimental Procedures, in orderto increase the contrast of the iron core versus protein[26] and to reduce the radiation damage.However, it isimportant to mention that under these conditions, thedegree of hydration shell (thickness and contrast)may have been altered, but the Rg values and overallshape of the scattering curves remained similar.The experimental scattering pattern for apo-DrDps2contains several fringes, of which the first two arewell defined at momentum transfer (q) values of

Fig. 4. CRYSOL of DrDps scattering curves and thecorresponding crystal structures. (A) Superposition ofscattering curve obtained from EOM SAXS data (dottedline) and from the crystal structure of DrDps1 (solid line,blue). (B) Superposition of scattering curve obtained fromEOM SAXS data (dotted line, green) and from the crystalstructure of DrDps1tΔ1–50 (solid line). (C) Superposition ofscattering curve obtained fromEOMSAXSdata (dotted line,red) and from the crystal structure of DrDps2 (solid line).

674 SAXS Structural Studies of Dps from D. radiodurans

0.89 nm−1 and 1.74 nm−1 (Fig. 5). The position ofthese fringes can be used to calculate the approxi-mate particle radius. Assuming a particle of uniformdensity, as would be the case for uniform sphericalparticles, the mean radius (Rparticle) can be approxi-mated using the relation q × Rparticle ≈ 4.50 and 7.73,respectively, for the first and second minima of theBessel function [27].

Note that the scattering function of a dodecahe-dron is identical to an equal volume sphere for the qrange used for the analysis [28]. The sphere radiiobtained from fringes 1 and 2 are consistent withthose observed for the protein shell of the DrDps2crystal structure at around 4.5 nm [16]. However, thenoticeable difference in the Rparticle obtained from thefirst and second minima indicates a more complexcore-shell structure than a simple sphere. The finerdetails such as the 41 N-terminal residues protrud-ing from the protein shell were not included in theseestimations, and the entire scattering contribution isassumed to be only from the dense protein core.The experimental scattering curves for DrDps2 in

the presence of iron also display these fringes, whichbecame less pronounced as the iron contentincreased (Fig. 5). Iron addition to DrDps2 with aratio of 24 to 400 Fe/dodecamer induces a modifi-cation to the scattering patterns characterised bysmearing and shifting of the fringe minima q values(Fig. 5). Due to an apparent increase in polydisper-sity or shape asymmetry observed in the sampleswith higher amounts of iron (Fig. 5), it was onlypossible to determine the intensity minima q valuesof the scattering functions up to 100 Fe/dodecamer.The visibility of fringes decreases as the number ofFe/dodecamer increases. This could reflect anincrease in polydispersity due to differences in ironloading or the protein shell becoming asymmetricupon iron addition (Table 2). Furthermore, fringe 1shifts slightly towards higher q values with theloading of iron, while fringe 2 shifts towards lower qvalues, which could be an indication of variation inthe contrast of the core-shell assembly [29].The entrapment of an iron core within the dodeca-

meric protein shell can be simplified and seen as aspherical core-shell model that takes into account theparticle morphology with an inner iron core and anouter shell of protein. In addition, to describe the highqregion of iron-loaded samples, it was necessary toinclude the scattering contribution from iron clusters(mean radius ≈ 1.3 nm), which could be formed withinthe protein core or free in the solution. Analysis interms of a polydisperse spherical core-shell model[30] including free iron “nano-clusters” [27] is present-ed in Fig. 5B and Table 2. The model assumes aspherical core shell with homogeneous density withinthe core (iron) and shell (protein). The nano-clustersare presumably embedded in the core, and the proteinshell becomes invisible due to the high contrast of thecore. Indeed, the protein is not spherical but more anelongated dodecameric form, which is indicated bythe deviation from the model at low q values. Theapparently high polydispersity is also a signature ofthis non-sphericity (it is well known that slightlyelongated objects can be described by a sphericalobject with an apparently high polydispersity). Neitherthe core nor the shell is homogeneous and that isdemonstrated by the deviations at high values.

Fig. 5. Effect of iron in the SAXS patterns of DrDps2. (A) Scattering curves of DrDps2 in apo-form and in the presenceof iron (equivalents of Fe/dodecamer). (B) Spherical core-shell model of DrDps2 in apo-form and in the presence of iron(Fe/dodecamer). The high q part of the scattering curve is described by including the iron clusters of mean radius of about1.3 nm.

675SAXS Structural Studies of Dps from D. radiodurans

Nevertheless, this simple colloidal model captures thebroad structural features and their trend upon theuptakeof iron. For the apo-form, themeancore radius,shell thickness, and polydispersity were 2.9 nm,1.6 nm, and 0.17, respectively, which changed to2.2 nm, 1.5 nm, and 0.3, respectively, in the presenceof 400 Fe/dodecamer (Table 2). This corresponds tothe mean radius decreasing from 4.5 nm to 3.7 nmwith iron uptake.As mentioned above, the apparent high polydis-

persity is likely to be due to shape fluctuations froman idealised sphere geometry, as can be visualisedin the ab initio envelopes (Fig. 6). The values of theradius of gyration were determined for apo-DrDps2and compared with those computed using the twoDrDps2 crystal structures, PDB codes 2C2J and2C6R, which correspond to the “as-isolated form”and iron-soaked crystal structures, respectively. TheRg calculated from the experimental SAXS data isconsistently higher than those determined from thecrystallographic models missing the 12 N-termini(Table 3). However, when the missing residues inthe N-terminus were added to the crystal model[Native (2C2J) + N-terminus; Table 3], the Rg value

Table 2. Structural parameters derived from core-shell model

Fe/protein RC (nm) tSh (nm) Polydispersity SLD c

0 2.9 1.6 0.17 9.424 2.8 1.6 0.20 9.448 2.7 1.5 0.22 9.4100 2.4 1.5 0.30 9.5200 2.2 1.5 0.30 9.5400 2.2 1.5 0.30 9.6

Note: RC is the mean core radius, tSh is the thickness of the shell, SLD isRNC is the mean radius of spherical nano-clusters and INC is their preprotein = RC + tSh.

is similar to those from experimental SAXS data byGuinier and GASBOR analysis (Table 3).The radius of gyration calculated from the ex-

perimental scattering patterns, using the Guinierapproximation, of the iron-loaded samples alsodecreases upon the addition of iron, consistent withthe core-shell model analysis described above. Thisdecrease could be due to the shrinkage of the coreregion with iron incorporation or due to a change incontrast with the uptake of iron.

Effect of iron on the DrDps2 shape

The scattering curve simulated using CRYSOLwiththe dodecameric assembly of the DrDps2 crystallo-graphic structure (PDB code 2C2J) shows somedifferences likely to be mostly due to the presence ofthe N-terminal extensions in solution (Fig. 6A), as alsodescribed above. The calculated curves in the q range1–6 nm−1 show only small differences near the fringeminima. The SAXS scattering patterns were used toperform molecular envelope calculations using theprogram GASBOR (Fig. 6). In this case, a molecularenvelope was obtained from the experimental

analysis with iron nano-clusters included

ore (nm−2) SLD shell (nm−2) RNC (nm) INC

6 × 10−4 9.54 × 10−4 0 07 × 10−4 9.54 × 10−4 0 08 × 10−4 9.55 × 10−4 1.4 2 × 10−4

× 10−4 9.56 × 10−4 1.2 6 × 10−4

8 × 10−4 9.56 × 10−4 1.28 15 × 10−4

2 × 10−4 9.56 × 10−4 1.32 33 × 10−4

the scattering length density (electron density x Thomson radius),factor (concentration x square of volume). The total radius of the

Fig. 6. DrDps2 envelope modelling from the averaged dummy residue models built from the SAXS scattering function.(A) DrDps2 in apo-form, (B) in the presence of 24 Fe/dodecamer, and (C) with 48 Fe/dodecamer. Panel (A) (dotted line)presented the X-ray solution scattering curve computed by CRYSOL from the native DrDps2 crystal structure (without theN-terminal extensions). Each panel presented the modelled (black line) and experimental scattering functions (colouredcircles) and the DrDps2 crystal structure docked within the modelled map. The DrDps2 atomic model (PDB code 2C2J) isdisplaced in cartoon represented as a dodecamer oriented along the C-terminal 3-fold axis, in which each monomer is in adifferent colour.

676 SAXS Structural Studies of Dps from D. radiodurans

scattering functions ofDrDps2 having a similar denseprotein core shape (of an average diameter of 9 nm).The DrDps2 crystal structure was docked within theSAXS molecular envelopes as determined for theapo-form, 24 and 48 iron equivalent samples. In thethree samples analysed, four extra areas (followingthe P23 symmetry) with a globular structural shape(Fig. 6) were identified protruding from the dodecamerprotein core. The N-terminal regions that were notvisible in the crystal structures were attributed to thefour extra regions protruding from the sphere core,coming from the 3-fold N-terminal axes subunits.Comparing these models to the one presented inFig. 3B, it is visible that in this case, only 4 extra areaswere observed instead of the 12 areas, correspondingto the average scattering from the N-terminal regions.This difference is probably related with the differentsample conditions used.The globularly shaped protuberances that could

accommodate the N-terminal extensions are ob-served in both models calculated from the iron-loaded SAXS data; however, the size of the regionsof these globular structures linking to the protein coreis modified with the presence of iron, as observed in

the models generated for 24 and 48 Fe/dodecamer(Fig. 6B and C). This variation could be due to thefact that the fit to the experimental data shows somedeviations, as it can be seen for fringe 1 (Fig. 6C).Indeed, the central profile of the models presentedhas changed from a circular for apo-form to a moretriangular cross-section for 48 iron atoms models(Fig. 6). However, it is important to consider thatthese protruding structures from the dodecamersphere correspond to the average scattering of theN-terminal extensions.As mentioned above, the influence of the iron addi-

tion to the protein suggests an apparent increase ofpolydispersity in iron content as indicated by theprogressive smearing of scatteringminimawith higheriron loading (Fig. 6). The presence of an iron corecannot be simulated by GASBOR, so the molecularenvelopes were determined only for the conditionswith 24 Fe and 48 Fe/dodecamer. The samecalculations were attempted for the other conditions(100, 200, and 400 Fe/dodecamer), but due to thehigh polydispersity or inhomogeneous partitioning ofiron prevalent in these samples, it was not possible togenerate an interpretable envelope. In addition, for the

Table 3. Comparison of the radius of gyration from nativeand iron-loaded samples

DrDps2 Guinier

q.Rg range Rg (nm) +/−

Native 1.15–1.99 4.40 0.0424 Fe/dodecamer 1.33–1.97 4.30 0.0548 Fe/dodecamer 1.35–1.99 4.16 0.05100 Fe/dodecamer 1.35–1.99 3.77 0.05200 Fe/dodecamer 1.50–1.97 3.39 0.09400 Fe/dodecamer 1.50–1.96 2.72 0.07

X-ray structures (dodecamer) CRYSOL

Rg (nm)

Native (PDB 2C2J) 3.6952 Fe (PDB 2C6R) 3.65Native (2C2J) + N‐terminus 4.65

GASBOR

Model Rg (nm) Rf

Apo-form 4.67 0.04924 Fe/dodecamer 4.60 0.05248 Fe/dodecamer 4.66 0.052

Values in italics are measured with q.Rg range N 1.5.CRYSOL Rg takes into account: atoms - excluded volume + shell.The PDB codes 2C2J and 2C6R are respectively the depositedcrystal structure coordinates for the DrDps2 apo-form andiron-loaded form.“Native (2C2J) + N-terminus” is a test model of the native crystalstructure to which a linear N-terminal protrusion of 41 polyalanineresidues (protruding from the spherical assembly) has beenmanually connected to the first observed residue in the crystalstructure.Per each spherical assembly, this forms four 3-fold symmetry-related polyalanine bundles at the N-terminal 3-fold axis and isused for comparative modelling purposes.Rf is a measure of the discrepancy between the model and theexperimental data, given by Rf = Σi [(Scale*Imodel(i) - Iexp(i))q(i)

2]2/Σi [Iexp(i)q(i)

2]2, where “Scale” corresponds to the scale factorproviding the best least squares.

677SAXS Structural Studies of Dps from D. radiodurans

100, 200, and 400 Fe/dodecamer, the scattering athigh q ranges is dominated by the iron core and freeiron clusters present in the solution.The Dmax determined by GNOM from the experi-

mental data of apo-form and iron-loaded samplesremains constant (Table 4). The comparison of nativeand iron-loaded envelopes (Fig. 6) shows an apparentdecrease in the mean radius of the protein sphere,specifically at the level of the 3-fold C-terminal axisfrom native to iron-loaded envelopes. Indeed, fringe 1in the SAXS spectra shifts towards a slightly higherscattering vector value, corresponding to a broaden-ing of size distribution upon iron addition.The iron loading homogeneity within DrDps2 sam-

ples was evaluated by electron microscopy (Fig. 7).DrDps2 in the apo-form shows some microcrystallineareas of DrDps2 spheres aggregated and organised

on a square lattice. Heterogeneities of iron loading areobserved at the ratio of 100 Fe/dodecamer, which isbelow the maximum load that the protein shell canbioencapsulate (Fig. 7). Nevertheless, the proteinsample containing 500 Fe/dodecamer presents ho-mogeneous iron loading indicating the saturation ofthe protein spheres with iron. However, residual ironthat could not be bioencapsulated is visible in theelectronmicrograph (dark spots not surrounded by the~10 nm circular shapes; Fig. 7).

Iron kinetics ofDrDps2measured by stopped-flowtime resolved SAXS

The mechanism for iron incorporation by Dpsoccurs in a multistep process, with the initial phaseoccurring on a time scale of less than 1 min [31,32].Although the details of the iron uptake mechanismfor DrDps2 have not been extensively addressed,previous studies indicate that this protein is able tooxidise and store iron in its cavity, in the presence ofeither oxygen or hydrogen peroxide [18].As part of the current study, iron incorporation by

DrDps2 was investigated using stopped-flow time-resolved SAXS. The addition of 50 Fe/dodecamer tothe DrDps2 showed an evolution of the scatteredintensity as a function of time. Indeed, an increase ofthe scattered intensity at q = 0.95 nm−1 from 0.0010to 0.0021 mm−1 was observedwithin 1 s together witha smearing of the curve around this position as seen inthe steady-state SAXS data. Analysis of the evolutionof Rg with time indicates that the modification of thestructure due to the addition of iron to the proteinsolution occurs within the first second, correspondingto a swelling of the protein sphere as observed by anincrease of Rg of around 0.5 nm (Fig. 8). After the firstsecond,Rg tends to stabilise over the time scale of themeasurement.

The function of the N-termini

In order to address the role of the DrDpsN-terminal regions with respect to DNA, the truncat-ed forms of DrDps, DrDps1tΔ1–50, and DrDps2tΔ1–39were incubated with DNA in the same conditions asdescribed previously for the full-length proteins [18].The result shows that there was no observable DNAgel migration shift in the presence of DrDps1tΔ1–50up to a concentration of 25 μM (Fig. 9A), comparedto only 4 μM of full-length protein that has beenpreviously shown to induce a shift [18]. This in-dicates that the absence of the N-terminal protru-sions compromises the interaction with the DNA,which is in accordance with previous results [11].In contrast, in the case ofDrDps2tΔ1–39, the addition

of 8 μM of protein induced a DNA gel migration shift.DrDps2 and DNA have been previously shown tointeract and form Dps–DNA complexes only with thesupercoiled form of the plasmid DNA [18]. Here, we

Table 4. Data collection and scattering-derived parameters for DrDps2 in as-isolated form and in the presence of iron(Fe/dodecamer)

Apo-form 24 Fe 48 Fe

Data collection parametersBeamline ID02, ESRFBeam geometry 0.3 mm × 0.3 mmWavelength (Å) 0.775q range (nm−1) 0.04–2.2 and 0.15–6Exposure time (s) 0.1Concentration (mg ml−1) 4 4 4

Structural parametersI(0) (from Guinier) 0.0400 0.0473 0.0690Rg (nm) (from Guinier) 4.40 4.30 4.16I(0) (mm−1) [from P(r)] 0.04803 0.04635 0.05526Rg (nm) [from P(r)] 5.27 5.09 5.07Dmax (nm) 17 17 17I(0) (mm−1; from Porod) 0.04277 0.03993 0.04566Rg (nm; from Porod) 4.67 4.40 4.25Porod volume estimate (nm3; +/− 10%) 428.40 415.50 (349.00)Dry volume calculated from PDB 2C2J (truncated assembly; nm3) 270,900 n/a n/aRg (Å; truncated assembly; from CRYSOL) 34.47 n/a n/aDry volume calculated from PDB 2C2J (native (2C2J) + N-terminus) (nm3) 314.30 n/a n/aRg (nm) [native (2C2J) + N-terminus] (from CRYSOL) 4.65 n/a n/a

Molecular mass estimationMolecular mass Mr. (from Porod volume; +/− 10%; kDa) 257 249.3 (209.4)Calculated dodecamer Mr. from sequence (kDa) 279.9 n/a n/a

SoftwaresPrimary data reduction (circular averaging) SAXS utilities (ESRF)Data processing PRIMUS (ATSAS), Origin,

and MS ExcelAb initio analysis GASBOR (ATSAS)Validation and averaging DAMMAVER (ATSAS)Rigid-body modelling n/aComputation of PDB model intensities CRYSOLThree-dimensional graphics representation Pymol

678 SAXS Structural Studies of Dps from D. radiodurans

show that DrDps2tΔ1–39 is able to shift all the forms ofplasmid DNA, being no longer selective for the type ofDNA and suggesting that the N-terminal region playsa key role in this selection event (Fig. 9B).The genomic sequence encoding for DrDps2

contains a signal peptide [15,16], which suggeststhat this protein could interact with the membrane.Therefore, the N-terminal function in association withthe cell membrane was investigated. D. radiodurans

Fig. 7. DrDps2 electron microscopy images. (A) DrDps2dodecamer, and (C) in the presence of 500 Fe/dodecamer.

bacteriawere grown inM53medium, andDrDps2wasdetected byWestern blotting, as previously described[18]. The protein was detected in both membrane andsoluble fractions (Fig. 10) but with different molecularmasses, suggesting that the form present in themembrane corresponds to the full-length DrDps2(without the signal peptide) while the one in thesoluble fraction contains the N-terminus cleaved.These data suggest that the N-terminal extension,

sample in the apo-form, (B) in the presence of 100 Fe/

Fig. 8. Evolution of the radius of gyration of DrDps2 upon iron loading (50 Fe/dodecamer) obtained by stopped-flowtime-resolved SAXS measurements.

679SAXS Structural Studies of Dps from D. radiodurans

with most of the residues containing polar butuncharged side chains, would have a further role inan interaction with the membrane.

Discussion

DrDps SAXS data

Dps from D. radiodurans have long N-terminalextensions before the first helix from the four-helixbundle, which were not modelled in the crystalstructures determined to date (Fig. 1). These regionsare predicted to have a coiled structure, being veryflexible, appear disordered in the crystal, andtherefore not modelled [13,14,16]. The structure ofthese regions was therefore addressed using Dps insolution in order to understand if they are fullydisordered or form some level of stable organisationthat could be visualised. In the present work, the

Fig. 9. Plasmid DNA binding (A) DrDps1tΔ1–50 protein or (B)Plasmid DNA alone (lane1) or increasing the concentrationincubated with plasmid DNA (lanes 2–9).

low-resolution molecular envelopes of the N-terminalextensions of DrDps and their relative positionwith respect to the protein shell were elucidated bySAXS.The ab initio DrDps SAXS envelopes generated by

the programGASBOR/EOMmatch the dimensions ofthe protein dodecameric spheres previously deter-mined by protein crystallography. Besides the centraldense core, the reported model envelopes harbourextra electron density areas (related by enforced P23symmetry) distributed around the sphere surface. Inorder to confirm this interpretation, the SAXS enve-lope model of a DrDps truncated construct lacking50 N-terminal residues (DrDps1tΔ1–50) was analysedand only showed a spherical particle without anyprotruding regions (Fig. 3). Furthermore, the scatter-ing parameters determined from this shorter protein,Rg and Dmax, are in agreement with those obtainedfrom the crystal structures.

DrDps2tΔ1–39 at pH 6.5 using agarose gel electrophoresis.of DrDps1tΔ1–50 (2–25 μM) or DrDps2tΔ1–39 (0.5–8 μM)

Fig. 10. Western blotting detection of DrDps2 in solubleandmembrane fractions fromD. radiodurans cells. Lane 1 –soluble fraction; lane 2 – membrane fraction.

680 SAXS Structural Studies of Dps from D. radiodurans

Two types of models were generated for bothDrDps, a static molecular envelope produced fromGASBOR/DAMMIN representing the average scat-tering, and a flexible model obtained from EOM. Thedifferent models are consistent on the central core,which show a hollow dodecameric sphere.The major difference is observed on the N-terminaltails arrangement; in the static model, 12 symmetricregions protruded from the dodecamer sphere,whereas in the flexible model, the N-terminal regionsthat protruded from central sphere correspond toindividual asymmetric tails with different conforma-tions. Although the low-resolution static modelcontains important structural information, ensemblemethod such as EOM represents a good approachto determine the different conformations that exist insolution.DrDps2 scattering data were measured from two

beamlines; although the models and scatteringparameters are similar, there are a few differences.The buffers used for the two experiments are not thesame, butDrDps2 retains its dodecameric form inbothbuffers. It has been previously shown that this proteinis a stable dodecamer when exposed to variousconditions [18]. The use of the online chromatographysetup has its major advantage in producing a betterquality scattering data from the pure oligomericmoieties and is therefore free of the contribution ofaggregates, whichwould otherwise bepresent. This isreflected in the parameters determined experimental-ly, namelyRg andDmax, which are in better agreement

with those extracted from the crystal structure.However, the advantage of using scattering datafrom beamline ID02 is the accessibility of wider qangles to obtain higher resolution information on thesystem, which, in this case, is important to understandthe structural changes in the presence of iron(discussed later).

DrDps and its N-terminal extension

Both the N- and C-terminal extensions have beenreported to be involved in the interactions betweenDps and DNA. This interaction has been proposed tooccur through a flexible tail rich in positively chargedresidues such as described in the cases ofEcDps andMsDps1 [6,8]. Flexibility of the N-terminal regions hasbeen one of the factors that contribute to theinteraction with DNA; however, in solution andwithoutthe presence of DNA, these regions are not disor-dered but instead present an organised structure asobserved by the average scattering shapes obtainedfrom the GASBOR molecular envelopes. To comple-ment our studies, we applied EOM. Different confor-mations were obtained for the different N-terminaltails, indicating that these regions may adopt differentconformations, which could be important for promot-ing DNA binding.The first 30 aa residues of DrDps1 contain a total

of seven positively charged residues: six lysines(Lys3, Lys4, Lys7, Lys13, Lys15, and Lys16) and onearginine (Arg28; Fig. 1). DNA interaction studies of aDrDps1 truncated form without the first 55 aaresidues [11] and our comparative data withDrDps1 show that this region is involved in DNAbinding, since the truncated forms have less affinityfor DNA (Fig. 9A) compared to the full-length protein[18,33]. This truncated form, which retains thedodecamer form, could still interact with the DNAby the positively charged residue Arg132 located atthe surface of the protein as previously proposed[12]. Nevertheless, the affinity for the DNA in thiscase is much lower than for the full-length protein, asobserved in Fig. 9A.The N-terminal extension of DrDps2 comprises 41

aa residues before the first helix from the four-helixbundle [16]. The SAXSmodel envelopes indicate thatin solution, the N-terminal extensions can form asmuch as 12 individual extra protrusions outside of thehollow sphere (Figs. 3B and 6A). Four protrudingglobular structures (each composed of 3 N-terminalDrDps2 with 41 aa residues each) are attached at theouter part of the 3-foldN-terminal axis, while themodelwith 12 individual extra protrusions corresponds to theindividual N-terminal protrusions. The secondarystructure prediction using the PSIPRED [34] serverbased on the amino acid sequence for this regiongives almost wholly coil, suggesting conformationalflexibility. Since each region contains six prolineresidues, the structural flexibility of polypeptide

681SAXS Structural Studies of Dps from D. radiodurans

regions would be reduced, which may be important toaccommodate the N-terminus into the structuresobserved in theSAXSenvelope, forming the structuralarrangements observed in the GASBOR molecularenvelope corresponding to the average scatteringshapes of these regions obtained in solution. DNAinteraction studies using the DrDps2tΔ1–39 indicatethat this truncated form binds to all the plasmid DNAforms (Fig. 9B). This result differs from the full-lengthDrDps2, which only binds to the supercoiled form fromthe plasmid DNA [18]. The first 41 N-terminal aminoacids only contain one positively charged residue(Arg33), most of which (49% out of 41 residues) arepolar with uncharged side chains (3 × Ser, 8 × Thr,7 × Gln, and 2 × Asn), and finally, 46% are non-polarwith aliphatic side chains (6 × Pro, 6 × Ala, 3 × Gly,2 × Val, and 2 × Leu). Based on this, it can besuggested that although this N-terminal region is notdirectly involved in DNA interactions, it could presum-ably guide the DNA to bind either to the positivelycharged residues located either at the protein surface(Lys46, Lys47, Arg122, and Arg149) or to the positivelycharged residues (Lys204, Arg209, and Arg211) locatedat the C-terminal region (Fig. 1). The absence of theN-terminal extension would expose these positivelycharged regions located at the protein surface,thereby favouring Dps' propensity to interact withDNA (Fig. 9).Analysis of the cell extracts of D. radiodurans

shows that DrDps2 is present in membrane andsoluble fractions (Fig. 10), although under the testedconditions, most of the protein is present in themembrane fraction. This is in agreement withprevious results, which showed the localisation ofthis protein as being close to the membrane region[15]. When comparing the two forms present in thesoluble and membrane fractions, we suggest thatthe form present in the soluble fraction wouldcorrespond to the N-terminal extension-free protein(form observed in the crystal structure), while theform observed in the membrane fraction would bethe full-length. Therefore, the N-terminus wouldallow an interaction with the membrane, indicatingthat the two forms of DrDps2 (full-length andtruncated forms) are present in D. radiodurans butwith distinct physiological roles.

Changes upon iron addition

The solution scattering envelopes were modelledfor apo-form and iron-loaded DrDps2. This character-isation was obtained using the ID02 beamline, whichallowed further information to be obtained from thehigh q region, and the effect of iron incorporation to theprotein structure was exploited.DrDps2 SAXS curveswere determined under apo conditions and also atincreasing iron equivalents per dodecamer from 24 to400. Analysis of the curves indicates that thepresence of iron inducesmodification to the scattering

curves in the q region 0.1–6 nm−1, in which theq-position minima of the first and second fringechange in the presence of iron.The shifts of the minima q-position of the SAXS

scattering functions from apo- to iron-loaded sam-ples are consistent with a change in the contrast ofthe core-shell structure with iron loading (Fig. 5). Thechange on the q-position of the first fringe wasobserved even at the low iron condition of 24 Fe/dodecamer. However, the low amount of ironpresent was expected to load only in the ferroxidasesites plus one iron per monomer and not to initiatethe iron core formation. In the present work, electronmicroscopy was performed on DrDps2 samples withvarious iron loading stages, which showed binarypopulations ofDrDps2 dodecamers with either no (orundetected iron loading) or very high iron loading(Fig. 7). This is indicated by an apparent polydisper-sity or shape variation by SAXS. The heterogeneityis probably a direct consequence of the iron incor-poration kinetics inside the Dps proteins: once theiron nucleus is formed, the biomineral will growaround it as an autocatalytic process [31,32], limitedonly by the size of the protein cavity that willeventually become saturated.It is interesting to note that the shape of the

DrDps2 protein SAXS molecular envelope changesupon iron addition. The Rg determined by Guinier/GNOM shows a decrease from the apo-form to the48 iron-loaded samples, suggesting that the centreof mass becomes more compact due to the iron coreformation. The central sphere structure appears lessrounded for the model with 48 Fe/dodecamercompared with the apo-form, suggesting somestructural modification in the global structure due toiron loading. In fact, the four globular structures onthe 3-fold N-terminal axis seem not to change acrossthe three different models presented, from apo-formto 24 and 48 Fe/dodecamer. This is subject tocaution because the curve fit for the model with 48Fe/dodecamer obtained with GASBOR is lesssatisfactory (Fig. 6C). Although iron core formationwas not expected in the case of 48 iron equivalents,strong heterogeneity in iron bioencapsulation in theDrDps2 spheres already leads to iron core formationin this “low” iron loading condition. The proteinmolecular envelope determined for this conditionbetween the protein dense core and the protrudingregion shows an area that may represent the initiallocations for iron core formation (Fig. 6C). It isinteresting to mention that two irons per monomer,Fe3 and Fe5, observed in the DrDps2 crystalstructure determined in the presence of iron (a totalof 52 Fe/dodecamer) are positioned in the 3-foldC-terminal channels, which were proposed to be anion path with a valve system to control the flow ofcations into the core [16].Although it was possible to locate different iron

sites in the DrDps2 crystal structure (PDB code

682 SAXS Structural Studies of Dps from D. radiodurans

2C6R), none of these correspond to biomineralisediron. The experimental procedure used for the crystalstructure loaded with iron was to soak an existingnative crystal in an iron solution. The packing andprotein rigidity in the crystal lattice therefore limit theformation of an iron core. No crystal cracks wereobserved following the iron soak of the native crystal.Nevertheless, locating the atomic iron core insidethe protein sphere using crystallography is ambigu-ous due to the random positioning of the differentiron atoms inside the protein shell. Comparing theRgdetermined with CRYSOL from both crystal struc-tures, we observed a difference of only 0.04 nmbetween the “as-isolated” crystal structure (Rg =3.69 nm) and iron-loaded crystal structure (Rg =3.65 nm), confirming that there is no significantcrystallographic difference in terms of overall struc-tural changes between the two states. However, theSAXS results under a similar amount of iron loading(sample with 48 Fe/dodecamer) show biominera-lised iron. This effect is probably due to sample ironheterogeneities as observed by electron microscopy(Fig. 7).Iron uptake by the DrDps2 protein has been

investigated using stopped-flow time-resolvedSAXS. The results indicate that the modification ofthe scattering patterns due to the addition of iron tothe protein solution occurs within the first second,corresponding to a swelling of the protein sphere asobserved by an increase ofRg of around 0.5 nm (Fig.8), then tending to stabilise during the time mea-sured. The process of iron oxidation by oxygen andthe subsequent growth of an iron core in Dps areknown to occur on the minute time scale [32]. In fact,previous studies on iron oxidation under aerobicconditions with DrDps2 demonstrated that the coreformation occurs slowly in a time scale of minutes[18]. Consequently, the increase in the radius ofgyration upon iron addition is unlikely to be related tochemical modifications such as iron oxidation or coreformation inside the protein shell. The origin of thisincrease on the radius of gyration is likely to be acombination of iron-triggered conformational chang-es in the entry channels at the level of the N-terminalextensions and a morphological transition from adodecamer to a more sphere-like shape.These results show that the iron binding to DrDps2

occurs in seconds, inducing an increase of theradius of gyration, while the iron core formationoccurs within minutes in which the radius of gyrationdecreases, as observed in the static SAXS mea-surements of DrDps2.

Conclusion

We present results from the SAXS experimentsperformed on both DrDps that show for the first timethat both proteins exist in solution with a definable

particle size corresponding to dodecameric sphereswith protruding N-terminal regions. The SAXS enve-lope determined for DrDps1 shows 12 N-terminalprotrusions pointing outside of the sphere, and themodel generated for the DrDps1 truncated form(DrDps1tΔ1–50) shows that the protein maintains itsspherical dodecameric shape but without the protrud-ing extensions, demonstrating that these protrusionsobserved for the full-length protein are from theN-terminal regions. The truncation of the first 50 aaresidues of the N-terminal region of DrDps1 almostabolishes the interaction with DNA, indicating thatthese tails protruding from the dodecameric sphereare essential for stabilising the protein–DNA complex.DrDps2 SAXS data suggest that the N-terminal

extensions can be organised around the 3-fold axis,grouped with three N-termini from different subunitsor as individual tails from each subunit, as observedwith the two models calculated from scattering dataobtained from two different beamlines. In both cases,these extensions are crucial for maintaining thedodecamer protein structure. The N-terminal protru-sions in DrDps2 are probably involved in theinteraction with DNA by hooking DNA close to thepositively charged residues located at the surface orthe C-terminal part of the protein. However, theseprotrusions can also be important for interacting withthe membrane: the presence of a signal peptidedirectsDrDps2 to themembrane region, and then, theinteraction with the membrane would be maintainedthrough the N-terminal extensions. Nevertheless, theinteraction between DrDps2 and the membraneneeds to be further investigated. The effect of ironon DrDps2 structure was addressed using static andstopped-flow time-resolved SAXS measurements,showing that the addition of iron induces the formationof iron nano-clusters, leading to dynamic conforma-tional changes in the overall protein dodecamer shape(models generated with GASBOR), which are alsoreflected by the variation of the radius of gyration.This study takes advantage of the complementarity

of both crystal X-ray diffraction and protein structureanalysis in solution by SAXS to provide a completecharacterisation of Dps biomolecular structures undernative conditions and in the presence of iron.

Experimental Procedures

Protein production and purification

DrDps1 was purified as previously described [14].DrDps1 truncated protein lacking the 50 N-terminalamino acids residues (DrDps1tΔ1–50) was obtainedfrom the in situ cleavage of E. coli extract withrecombinantDrDps1. The protocol for protein expres-sion and purification was performed at room temper-ature using the same procedures describedfor DrDps1. Sample purity was judged by

683SAXS Structural Studies of Dps from D. radiodurans

SDS-PAGE and Western blotting analysis as de-scribed in Ref. [18], which confirmed that the proteinwas 100% cleaved to the DrDps1tΔ1–50. In order todetermine the starting amino acid sequence, theN-terminal sequencing was made for DrDps1tΔ1–50.This was done first using SDS-PAGE followed byWestern blotting (semi-dry system, BioRad) usingpolyvinylidene fluoride membrane (PVDF). Beforeloading the samples into 15% SDS-PAGE, thegel was submit to a pre-run with a running buffer[1.5% (wt/vol) Tris, 7.2% (wt/vol) glycine, and 0.5%(wt/vol) SDS at pH 8.3] with 40 μM of glutathione for2 h at 100 V. Subsequently, the running buffer wasremoved, and a new running buffer was added,containing an extra 3% (vol/vol) of thioglycolic acid.The sample was applied to the gel and the run wasperformed at 150 V at room temperature. Then, thegel was washed in the blotting running buffer [10%(vol/vol) methanol and 10 mMCAPSat pH 11]. PVDFmembrane was activated in 100% (vol/vol) of meth-anol for 30 s and washed with water and then with theblotting running buffer for 15 min. The sample transferwas done for 30 min at 15 V at room temperature.After the transfer, the PVDF membrane was stainedwithCoomassie brilliant blueR-250 (BioRad) followedby washing with water and 50% (vol/vol) methanol.The N-terminal amino acid residue sequence wasdetermined by Edman reaction using Procise 491 HTProtein Sequencer (Applied Biosystems). These datawere provided by the Analytical Laboratory, AnalyticalServices Unit from Instituto de Tecnologia Química eBiológica António Xavier, Universidade Nova deLisboa.DrDps2 was purified as previously described [16].

A truncated construct lacking the 39 N-terminalamino acid residues (DrDps2tΔ1–39) was clonedinto a plasmid pET151/D-TOPO (Gateway system,Invitrogen) with His-Tag and Tobacco Etch Virus(TEV) cleavage site at the N-terminus. The resultingplasmid was transformed into E. coli BL21 (DE3)Star, and overexpression was obtained by growingcells at 37 °C in LB with 100 μg/ml ampicillin until anoptical density at 600 nm of 0.6. The cells wereinduced with 0.5 mM of IPTG, which were grown for4 h at 37 °C. The cells were collected by centrifu-gation at 16,000g for 20 min at 4 °C, resuspended inlysis buffer [20% (wt/vol) sucrose, 50 mM Tris–HCl(pH 8.0), 100 mM NaCl, 1 mM MgCl2, 0.1 mg/mllysozyme, and 20 μg/ml DNase], and broken in aFrench pressure cell at 35,000 psi.The protein expression was found in inclusion

bodies, which were collected by centrifugation at11,000g for 15 min at 4 °C. The inclusion bodieswere refolded using the previously reported proce-dure [14]. The renaturated sample was loaded in5-ml HisTrap HP column [1.5 ml/min, 20 mM Tris–HCl (pH 7.5), 250 mM NaCl, 2.5% glycerol, 1 mMPMSF, and 10–500 mM Imidazole]. The DrDp2tΔ1–39 sample was eluted at 100–300 mM imidazole, and

it was immediately dialysed against 20 mM Tris–HCl(pH 7.5), 150 mM NaCl, 2.5% glycerol, and 1 mMPMSF. His-Tag was cleaved by an overnightincubation at 4 °C with TEV protease. Finally, thesample was loaded on the 5-ml HisTrap HP column(GE) using the same buffers as described above.The protein was eluted in flow-through, and its puritywas judged by SDS-PAGE analysis.The oligomerisation state for all proteins, after

purification, was resolved by SEC using a Superdex200 10/300 GL (GE Healthcare) column equilibratedwith 20 mM Tris–HCl (pH 7.5) and 150 mM NaClbuffer. The column was previously calibrated usingthe standard's proteins ranging 14–660 kDa (GEHealthcare).

Sample preparation

All samples, DrDps1, DrDps1tΔ1–50, DrDps2, andDrDps2t Δ1–39 used in experiments realised onbeamline BM29 were dialyzed in 20 mM Tris–HCl(pH 7.5) plus 150 mM NaCl buffer and concentratedto 10 mg/ml.The DrDps2 protein used in experiments realised

on beamline ID02 was concentrated to 4.0 mg/ml in50 mM Hepes (pH 8.0) buffer, 150 mM NaCl, andglycerol [50% (wt/vol)]. Fresh 100 mM iron (II)ammonium sulphate solutions were prepared withwater previously degassed under vacuum and thenargon-flushed to avoid iron autoxidation prior tomixing with the protein solution. The final buffersolutions were not degassed in order to allow fasteriron uptake and oxidation within the Dps. The sameprotein concentration was used for a series ofexperiments with 24, 48, 100, 200, and 400 ironequivalents per dodecamer. Each sample wasincubated for 10 min at room temperature beforeSAXS measurements. Salt, buffer, and glycerol[50% (wt/vol)] concentrations were kept constant inall samples analysed. The presence of 50% (wt/vol)glycerol helps avoid radiation damage with the X-raybeam but at the same time reduces the contrast ofthe protein and makes the iron core more visible [26].

ESRF beamline BM29: SAXS setup and datacollection for DrDps1, DrDps1tΔ1–50, and DrDps2

SAXS data were collected at the ESRF BioSAXSbeamline BM29 [19]. An online HPLC system(Viscotek GPCmax, Malvern Instruments) wasused [35] coupled directly to the BM29 samplechanger [20] exposure unit inlet valve. Independent-ly, three DrDps1, two DrDps1tΔ1–50, and two DrDps2samples were loaded into vials and automaticallyinjected onto the SEC, Superdex 200 10/300 GL (GEHealthcare), equilibrated with 20 mM Tris–HCl(pH 7.5) and 150 mM NaCl buffer at room temper-ature via an integrated syringe system.

684 SAXS Structural Studies of Dps from D. radiodurans

SAXS data were collected using X-rays of wave-length of 0.9919 Å and a sample-to-detector dis-tance of 2.81 m corresponding to q ranges of 0.08–4.5 nm−1. About 1500 (1 frame s−1) frames werecollected for each sample, where q is the magnitudeof the scattering vector given by q = 4π/λ sin (θ),with 2θ the scattering angle. All data processing ofeach sample was performed automatically using theEDNA online data analysis [36] pipeline using toolsfrom the EMBL-HH ATSAS 2.5.1 [37], generatingradially integrated, calibrated, and normalisedone-dimensional profiles for each frame. All frameswere compared with the initial frame, and matchingframes were merged to create the reference buffer.Any subsequent frames, which differed from thereference buffer, were subtracted and then proc-essed within the EDNA pipeline. The invariantscalculated by the ATSAS tool (AUTORG) were usedto select a subset of frames from the peak scatteringintensity. Frames with a consistent radius of gyration(Rg) from the peak scattering intensity were auto-matically merged to yield a single averaged framecorresponding to the scattering of an individualSEC-purified species. The peaks of interest werereprocessed manua l l y to max im ise thesignal-to-noise ratio. Individually, the frames (ap-proximately 60 for DrDps1, 40 for DrDps1tΔ1–50, and45 for DrDps2) corresponding to the highest proteinconcentration were merged and used for all furtherdata processing and model fitting. A Kratky plot wasused to confirm the fold of the all proteins samples.

ESRF beamline ID02: SAXS setup and datacollection for DrDps2

Small angle X-ray data were collected using X-raysof wavelength of λ = 0.775 Å onESRFbeamline ID02[25]. SAXS data were collected using two sample-to-detector distances of 0.85 m and 2.5 m corre-sponding to q ranges of 0.15–6 nm−1 and 0.04–2.2 nm−1, respectively. The sample was injected intoa flow-through quartz capillary cell equipped with amotorised syringe. This setup allowed the precisemeasurement of protein solution and the correspond-ing buffer solution at the same spot. Radiationdamage is avoided by displacing the solution betweentwo successive exposures to the X-ray beam. Anexposure time of 0.1 s was chosen for data acquisi-tion, after tests of up to 0.5 s with an X-ray intensity of2.1 × 1012 photons/s were performed on DrDps2 tocheck the resistance of protein against radiationdamage, in both native and iron-loaded samples. Atotal of 10 2D SAXS patterns were collected for eachsample with the samples refreshed after eachexposure. SAXS patterns were normalised to abso-lute scale using the geometrical factors, the incidentflux, and sample transmission and then azimuthallyaveraged to obtain the one-dimensional scatteredintensity as a function of q [25]. Normalised scattered

intensity of the buffer solution was subtracted fromeach sample scattering curve, and the resultingquantity after division by sample thickness is repre-sented as I(q) in mm−1. Due to possible differences ofsalt or glycerol concentration between the buffer andthe sample, a small offset correction (of a few percent)was required for the buffer scattering of somesamplesin order to obtain the expectedbehaviour of I(q) at highq values. This correction has no influence at low qranges fromwhich themain structural information wasdeduced. The emergence of a scattering minimum athigh scattering vector (q) values is a signature ofparticles with core-shell morphology. For data analy-sis, the interparticle interactions are assumed to benegligible for the concentrations used.

Data processing for DrDps1, DrDps1tΔ1–50, andDrDps2

The radius of gyration was computed from the slopeof the Guinier plot of the profile [38] using PRIMUS(ATSAS) [39] represented by this equation: I (q) = I(0)exp [ − (q2 Rg

2)/3], and this equation is valid when themomentum transfer (q) is near zero or whenq.Rg b 1.3 [40] for the data collected at BM29, whilefor some instances of the analysis of the datacollected at ID02, due to possible aggregation andrelated data exclusion from analysis at low angle,q.Rg b 2.0 has been used (which is an approximationin the specific cases of sphere-like particles). Curvefitting was carried out to determine the value of themaximal particle size (Dmax) by calculating the pairdistance distribution function [P(r)] for an arbitrarymonodispersive system model using GNOM [21].Also, Rg and Porod volume were obtained by P(r)function using the GNOM program that is included inATSAS 2.5.1 suite [21]. The P(r) function wasobtained only for the q values b 2 nm−1.Theoretical scattering curves were calculated from

structures of DrDps1 or DrDps2 (PDB codes 2C2Uand 2C2J) and compared with the experimental dataSAXS using CRYSOL [41]. The ab initio shapeswere obtained from the SAXS data for each sampleusing GASBOR program [42] and EOM [23], firstwithout structural information from the crystal struc-ture, applying no symmetry (point group 1), thenimposing the crystal structure point group symmetry(P23) of the DrDps dodecameric particles and thetotal protein residues expected per monomer (207for DrDps1, 158 for DrDps1tΔ1–50, and 211 DrDps2).

SAXS model shape representation

For eachDps (DrDps1,DrDps1tΔ1–50, andDrDps2),a total of 10 ab initio reconstructed models obtainedwith GASBOR were matched by DAMAVER [44]program package (based on the program SUPCOMB[45]).

685SAXS Structural Studies of Dps from D. radiodurans

The models were obtained only from the qrange b 2 nm−1. Low-resolution electron densitymaps of the molecular envelopes were computedwith PDB2VOL and converted to CCP4 format withMAP2MAP {both programs are part of the SITUS(v2.5) program package [46–51]}. Docking of theX-ray crystal structures with the SAXS maps wasperformed using Colacor (SITUS v2.5) and bymanual translation and rotation. The shape recon-struction is contoured at 1 sigma and represented bydensity mesh using PYMOL [52,53].

Stopped-flow time-resolved SAXSmeasurementsof DrDps2

Stopped-flow experiments were carried out at ESRFbeamline ID02 using an apparatus (SFM-3,Bio-Logic,Pont de Claix, France) described in Ref. [25]. Forthesemeasurements, the ratio of protein to ironwas1:1(vol/vol), which corresponds to a final protein concen-tration of 4 mg/ml and is loadedwith 50 Fe/dodecamer.The q range of themeasurementwas 0.20 to 4.65 nm−

1 during the first 3.7 s.

Electron microscopy of DrDps2

DrDps2 electron microscopy images were per-formed in 50 mM Hepes buffer (pH 7.0). The proteinwas diluted to a final concentration of 25 μg/ml. Ironloadings of 100 and 500 iron equivalents per proteindodecameric assembly were performed by adding asolution of iron (II) ammonium sulphate before dilution.Protein samples were applied to the clean side ofcarbon on mica (carbon/mica interface) and negativelystained with 1% sodium silicotungstate (pH 7.0).Micrographs were taken under low-dose conditionswith a JEOL 1200 EX II microscope at 100 kV and acalibrated magnification of 39,750× (using the tobaccomosaic virus particle). Selected negatives were thendigitisedonanEpsonPerfection4990PHOTOscannerat 2400 dpi (2.6 Å at the sample scale).

Electrophoretic mobility shift assay

DrDps1tΔ1–50 (2–25 μM) or DrDps2tΔ1–39 (0.5–8 μM) was incubated with 9 nM pUC19 supercoiledplasmid DNA for 15 min at room temperature in40 mM Bis-Tris (pH 6.5) plus 150 mM NaCl buffer.DNA-binding interaction studies were performed aspreviously described [18].

Detection of DrDps2 in D. radiodurans

D. radiodurans cells were grown in M53 medium(500 mL) at 30 °C, as previously described [18]. Cellswere collected at mid-exponential phase (OD600nm =0.60) and were then harvested at 11000 g andresuspended in 20 mM Tris–HCl (pH 7.5) plus150 mM NaCl. Cells were disrupted in a Frenchpressure cell at 15000 psi. Membrane and solublefractions were obtained after ultracentrifugation for 1 hat 186000 g and at 4 °C. After protein concentration

using 3-kDa protein concentrator (Amicon), the proteinwas quantified using the modification Biuret method[54] for the membrane fraction, and the soluble fractionwas quantified by Bradford method (Bio-Rad) [55]. Atotal protein of 60 μg was loaded in 12% PAGE, andDrDps2 was detected by Western blot analysis aspreviously described [18]. The data presented are fromthree independent samples.

Accession numbers

The scattering data collected in BM29 beamlineare deposited in the Small Angle Scattering Biolog-ical Data Bank (SASBDB) [43], with the followingaccession codes (Table 1):SASDBG7 (DrDps1), SASDBF7 (DrDps2), and

SASDBH7 (DrDps1tΔ1–50).

Acknowledgements

This work was supported by the EuropeanSynchrotron Radiation Facility (ESRF, Grenoble,France), which provided Ph.D. studies' funding toM.G.C. and data collection material and equipment,and by the Fundação para a Ciência e Tecnologiagrants PTDC/BIA-PRO/100365/2008 and PTDC/QUI-BIQ/100007/2008. This work was furthers u p p o r t e d b y M O S T M I C R O(LISBOA-01-0145-FEDER-007660) Research Unitco-funded by FCT, through national funds, and byFEDER under the PT2020 Partnership Agreement.S.P.S. and C.V.R. acknowledge the FCT grantsSFRH/BD/78870/2011 and SFRH/BPD/94050/2013, respectively. We thank G. Schoehn for theelectron microscopy analysis and the IBS (Grenoble)for mass spectrometry assistance.

Received 2 August 2016;Received in revised form 6 January 2017;

Accepted 6 January 2017Available online 11 January 2017

Keywords:DNA;metal;

scattering;signal peptide;

membrane

Present address: M.G. Cuypers, St. Jude Children'sResearch Hospital, 262 Danny Thomas Place, Memphis,

TN 38105, USA.

Abbreviations used:Dps, DNA-binding proteins from starved cells; EcDps, Dpsfrom Escherichia coli; MsDps1, Dps from Mycobacterium

686 SAXS Structural Studies of Dps from D. radiodurans

smegmatis; DrDps, Dps from Deinococcus radiodurans;SAXS, small angle X-ray scattering; EOM, ensemble

optimisation method; SEC, size-exclusion chromatogra-phy; PVDF, polyvinylidene fluoride membrane; TEV,

Tobacco Etch Virus.

References

[1] E. Chiancone, P. Ceci, The multifaceted capacity of Dpsproteins to combat bacterial stress conditions: detoxificationof iron and hydrogen peroxide and DNA binding, Biochim.Biophys. Acta 1800 (2010) 798–805.

[2] T. Haikarainen, A.C. Papageorgiou, Dps-like proteins:structural and functional insights into a versatile proteinfamily, Cell. Mol. Life Sci. 67 (2010) 341–351.

[3] K. Zeth, Dps biomineralizing proteins: multifunctional archi-tects of nature, Biochem. J. 445 (2012) 297–311.

[4] L.N. Calhoun, Y.M. Kwon, Structure, function and regulationof the DNA-binding protein Dps and its role in acid andoxidative stress resistance in Escherichia coli: a review,J. Appl. Microbiol. 110 (2011) 375–386.

[5] R.R. Crichton, J.P. Declercq, X-ray structures of ferritins andrelated proteins, Biochim. Biophys. Acta 1800 (2010) 706–718.

[6] P. Ceci, S. Cellai, E. Falvo, C. Rivetti, G.L. Rossi, E.Chiancone, DNA condensation and self-aggregation ofEscherichia coli Dps are coupled phenomena related to theproperties of the N terminus, Nucleic Acids Res. 32 (2004)5935–5944.

[7] R.A. Grant, D.J. Filman, S.E. Finkel, R. Kolter, J.M. Hogle,The crystal structure of Dps, a ferritin homolog that binds andprotects DNA, Nat. Struct. Biol. 5 (1998) 294–303.

[8] S. Roy, R. Saraswathi, S. Gupta, K. Sekar, D. Chatterji, M.Vijayan, Role of N- and C-terminal tails in DNA binding andassembly in Dps: structural studies of Mycobacterium smeg-matis Dps deletion mutants, J. Mol. Biol. 370 (2007) 752–767.

[9] S. Roy, S. Gupta, S. Das, K. Sekar, D. Chatterji, M. Vijayan,X-ray analysis of Mycobacterium smegmatis Dps and acomparative study involving other Dps and Dps-like mole-cules, J. Mol. Biol. 339 (2004) 1103–1113.

[10] T.J. Stillman, M. Upadhyay, V.A. Norte, S.E. Sedelnikova, M.Carradus, S. Tzokov, P.A. Bullough, C.A. Shearman, M.J.Gasson, C.H. Williams, P.J. Artymiuk, J. Green, The crystalstructures of Lactococcus lactisMG1363 Dps proteins revealthe presence of an N-terminal helix that is required for DNAbinding, Mol. Microbiol. 57 (2005) 1101–1112.

[11] G. Bhattacharyya, A. Grove, The N-terminal extensions ofDeinococcus radiodurans Dps-1 mediate DNA major grooveinteractions as well as assembly of the dodecamer, J. Biol.Chem. 282 (2007) 11,921–11,930.

[12] K.H. Nguyen, A. Grove, Metal binding at the Deinococcusradiodurans Dps-1 N-terminal metal site controls dodeca-meric assembly and DNA binding, Biochemistry 51 (2012)6679–6689.

[13] S.G. Kim, G. Bhattacharyya, A. Grove, Y.H. Lee, Crystalstructure of Dps-1, a functionally distinct Dps protein fromDeinococcus radiodurans, J. Mol. Biol. 361 (2006) 105–114.

[14] C.V. Romao, E.P. Mitchell, S. McSweeney, The crystalstructure of Deinococcus radiodurans Dps protein (DR2263)reveals the presence of a novel metal centre in the Nterminus, J. Biol. Inorg. Chem. 11 (2006) 891–902.

[15] B.J. Reon, K.H. Nguyen, G. Bhattacharyya, A. Grove,Functional comparison of Deinococcus radiodurans Dps

proteins suggests distinct in vivo roles, Biochem. J. 447(2012) 381–391.

[16] M.G. Cuypers, E.P. Mitchell, C.V. Romão, S.M. McSweeney,The crystal structure of the Dps2 from Deinococcus radio-durans reveals an unusual pore profile with a non-specificmetal binding site, J. Mol. Biol. 371 (2007) 787–799.

[17] J.L. Gao, Y. Lu, G. Browne, B.C. Yap, J. Trewhella, N.Hunter, K.A. Nguyen, The role of heme binding by DNA-protective protein from starved cells (Dps) in the tolerance ofPorphyromonas gingivalis to heme toxicity, J. Biol. Chem.287 (2012) 42,243–42,258.

[18] S.P. Santos, E.P. Mitchell, H.G. Franquelim, M.A. Castanho,I.A. Abreu, C.V. Romao, Dps from Deinococcus radiodurans:oligomeric forms of Dps1 with distinct cellular functions andDps2 involved in metal storage, FEBS J. 282 (2015)4307–4327.

[19] P. Pernot, A. Round, R. Barrett, A. De Maria Antolinos, A.Gobbo, E. Gordon, J. Huet, J. Kieffer, M. Lentini, M. Mattenet,C. Morawe, C. Mueller-Dieckmann, S. Ohlsson, W. Schmid,J. Surr, P. Theveneau, L. Zerrad, S. McSweeney, UpgradedESRF BM29 beamline for SAXS on macromolecules insolution, J. Synchrotron Radiat. 20 (2013) 660–664.

[20] A. Round, F. Felisaz, L. Fodinger, A. Gobbo, J. Huet, C.Villard, C.E. Blanchet, P. Pernot, S. McSweeney, M.Roessle, D.I. Svergun, F. Cipriani, BioSAXS sample chang-er: a robotic sample changer for rapid and reliable high-throughput X-ray solution scattering experiments, ActaCrystallogr. D Biol. Crystallogr. 71 (2015) 67–75.

[21] D.I. Svergun, Determination of the regularization parameterin indirect-transform methods using perceptual criteria,J. Appl. Crystallogr. 25 (1992) 495–503.

[22] M.H.J. Koch, P. Vachette, D.I. Svergun, Small-angle scat-tering: a view on the properties, structures and structuralchanges of biological macromolecules in solution, Q. Rev.Biophys. 36 (2003) 147–227.

[23] G. Tria, H.D. Mertens, M. Kachala, D.I. Svergun, Advancedensemble modelling of flexible macromolecules using X-raysolution scattering, IUCrJ 2 (2015) 207–217.

[24] D. Schneidman-Duhovny, M. Hammel, J.A. Tainer, A. Sali,Accurate SAXS profile computation and its assessment bycontrast variation experiments, Biophys. J. 105 (2013)962–974.

[25] P. Panine, S. Finet, T.M. Weiss, T. Narayanan, Probing fastkinetics in complex fluids by combined rapid mixing and small-angleX-ray scattering, Adv. Colloid Interf. Sci. 127 (2006) 9–18.

[26] K. Ibel, H.B. Stuhrmann, Comparison of neutron and X-rayscattering of dilute myoglobin solutions, J. Mol. Biol. 93 (1975)255–265.

[27] T. Narayanan, Small-Angle Scattering, in: D.W. Bruce, D.O'Hare, R.I. Walton (Eds.), Structure from Diffraction Methods,John Wiley & Sons, Ltd, Chichester, UK 2014, pp. 259–324.

[28] X. Li, C.-Y. Shew, L. He, F. Meilleur, D.A. Myles, E. Liu, Y.Zhang, G.S. Smith, K.W. Herwig, R. Pynn, Scattering functionsof platonic solids, J. Appl. Crystallogr. 44 (2011) 545–557.

[29] M. Ballauff, J. Bolze, N. Dingenouts, P. Hickl, D. Potschke,Small-angle X-ray scattering on latexes, Macromol. Chem.Phys. 197 (1996) 3043–3066.

[30] P. Bartlett, R.H. Ottewill, A neutron scattering study of thestructure of a bimodal colloidal crystal, J. Chem. Phys. 96(1992) 3306–3318.

[31] M. Su, S. Cavallo, S. Stefanini, E. Chiancone, N.D. Chasteen,The so-called Listeria innocua ferritin is a Dps protein. Ironincorporation, detoxification, and DNA protection properties,Biochemistry 44 (2005) 5572–5578.

687SAXS Structural Studies of Dps from D. radiodurans

[32] G. Zhao, P. Ceci, A. Ilari, L. Giangiacomo, T.M. Laue, E.Chiancone, N.D. Chasteen, Iron and hydrogen peroxidedetoxification properties of DNA-binding protein from starvedcells. A ferritin-like DNA-binding protein of Escherichia coli,J. Biol. Chem. 277 (2002) 27,689–27,696.

[33] A. Grove, S.P. Wilkinson, Differential DNA binding andprotection by dimeric and dodecameric forms of the ferritinhomolog Dps from Deinococcus radiodurans, J. Mol. Biol. 347(2005) 495–508.

[34] D.T. Jones, Protein secondary structure prediction based onposition-specific scoring matrices, J. Mol. Biol. 292 (1999)195–202.

[35] A. Round, E. Brown, R. Marcellin, U. Kapp, C.S. Westfall, J.M.Jez, C. Zubieta, Determination of the GH3.12 protein confor-mation through HPLC-integrated SAXS measurements com-bined with X-ray crystallography, Acta Crystallogr. D Biol.Crystallogr. 69 (2013) 2072–2080.

[36] M.E.Brennich, J.Kieffer,G.Bonamis, A.DeMariaAntolinos,S.Hutin, P. Pernot, A. Round, Online data analysis at the ESRFbioSAXS beamline, BM29, J. Appl. Crystallogr. 49 (2016).

[37] M.F. Incardona, G.P. Bourenkov, K. Levik, R.A. Pieritz, A.N.Popov, O. Svensson, EDNA: a framework for plugin-basedapplications applied to X-ray experiment online data analysis,J. Synchrotron Radiat. 16 (2009) 872–879.

[38] A. Guinier, G. Fournet, Small-Angle Scattering of X-Rays,John Wiley, New York, 1955.

[39] P.V. Konarev, V.V. Volkov, A.V. Sokolova, M.H.J. Koch, D.I.Svergun, PRIMUS: a Windows PC-based system for small-angle scattering data analysis, J. Appl. Crystallogr. 36 (2003)1277–1282.

[40] C.D. Putnam, M. Hammel, G.L. Hura, J.A. Tainer, X-raysolution scattering (SAXS) combined with crystallography andcomputation: defining accurate macromolecular structures,conformations and assemblies in solution, Q. Rev. Biophys. 40(2007) 191–285.

[41] D. Svergun, C. Barberato, M.H.J. Koch, CRYSOL—a programto evaluate x-ray solution scattering of biological macromole-cules from atomic coordinates, J. Appl. Crystallogr. 28 (1995)768–773.

[42] D.I. Svergun, M.V. Petoukhov, M.H. Koch, Determination ofdomain structure of proteins from X-ray solution scattering,Biophys. J. 80 (2001) 2946–2953.

[43] E. Valentini, A.G. Kikhney, G. Previtali, C.M. Jeffries, D.I.Svergun, SASBDB, a repository for biological small-anglescattering data, Nucleic Acids Res. 43 (2015) D357–D363.

[44] V.V. Volkov, D.I. Svergun, Uniqueness of ab initio shapedetermination in small-angle scattering, J. Appl. Crystallogr.36 (2003) 860–864.

[45] M.B. Kozin, D.I. Svergun, Automated matching of high- andlow-resolution structural models, J. Appl. Crystallogr. 34 (2001)33–41.

[46] W. Wriggers, P. Chacon, Using Situs for the registration ofprotein structures with low-resolution bead models from X-raysolution scattering, J. Appl. Crystallogr. 34 (2001) 773–776.

[47] W.Wriggers, P. Chacon, Modeling tricks and fitting techniquesfor multiresolution structures, Structure 9 (2001) 779–788.

[48] W. Wriggers, S. Birmanns, Using situs for flexible and rigid-body fitting of multiresolution single-molecule data, J. Struct.Biol. 133 (2001) 193–202.

[49] W.Wriggers, R.A. Milligan, J.A. McCammon, Situs: a packagefor docking crystal structures into low-resolution maps fromelectron microscopy, J. Struct. Biol. 125 (1999) 185–195.

[50] W. Wriggers, R.A. Milligan, K. Schulten, J.A. McCammon,Self-organizing neural networks bridge the biomolecularresolution gap, J. Mol. Biol. 284 (1998) 1247–1254.

[51] J.S. Pedersen, Modelling of Small-Angle Scattering datafrom colloids and polymer systems, in: P. Lindner, T. Zemb(Eds.), Neutrons, X-Rays and Light, 1st Ed, Elsevier 2002,pp. 391–420, North Holland.

[52] W.L. DeLano, The PyMOL Molecular Graphics System, 2002.[53] L.L.C. Schrodinger, The PyMOLMolecular Graphics System,

Version 1.3r1, 2010.[54] C. Watters, A one-step biuret assay for protein in the

presence of detergent, Anal. Biochem. 88 (1978) 695–698.[55] M.M. Bradford, A rapid and sensitive method for the

quantitation of microgram quantities of protein utilizing theprinciple of protein–dye binding, Anal. Biochem. 72 (1976)248–254.