Mechanism for accurate, protein-assisted DNAannealing by Deinococcus radiodurans DdrBSeiji N. Sugiman-Marangosa, Yoni M. Weissb, and Murray S. Junopa,c,1

aDepartment of Biochemistry and Biomedical Sciences and M. G. DeGroote Institute for Infectious Disease Research, McMaster University, Hamilton, ON,L8N 3Z5, Canada; bMichael G. DeGroote School of Medicine, McMaster University, Hamilton, ON, L8S 4L8, Canada; and cSchulich School of Medicine andDentistry, Department of Biochemistry, University of Western Ontario, London, ON, N6A 5C1, Canada

Edited by Wei Yang, National Institutes of Health, Bethesda, MD, and approved February 17, 2016 (received for review October 21, 2015)

Accurate pairing of DNA strands is essential for repair of DNAdouble-strand breaks (DSBs). How cells achieve accurate annealingwhen large regions of single-strand DNA are unpaired has remainedunclear despite many efforts focused on understanding proteins,which mediate this process. Here we report the crystal structure ofa single-strand annealing protein [DdrB (DNA damage response B)]in complex with a partially annealed DNA intermediate to 2.2 Å.This structure and supporting biochemical data reveal a mechanismfor accurate annealing involving DdrB-mediated proofreading ofstrand complementarity. DdrB promotes high-fidelity annealingby constraining specific bases from unauthorized association andonly releases annealed duplex when bound strands are fully com-plementary. To our knowledge, this mechanism provides the firstunderstanding for how cells achieve accurate, protein-assistedstrand annealing under biological conditions that would otherwisefavor misannealing.

single-strand DNA annealing | DNA repair | Deinococcus radiodurans |crystal structure | DdrB

How cells overcome the need to protect ssDNA from formingdeleterious secondary structures and at the same time pro-

mote accurate strand annealing represents a longstanding andintriguing question in DNA repair. Single-strand DNA-bindingproteins (SSBs) safeguard DNA fragments harboring single-strand regions from misannealing at biological temperatures bybinding and occluding bases from potential interaction withother strands (1). Overcoming the thermodynamic barrier ofaccurate annealing under biological conditions requires special-ized annealing proteins able to regulate access of unpaired basesfrom different strands. Although single-strand annealing (SSA)proteins have been identified in organisms ranging from bacte-riophage to humans, how they promote faithful annealing of DNAends containing large single-strand regions has remained unclear.One of the best examples of single-strand annealing comes

from members of the Deinococcus family of bacteria, whichare renowned for their ability to withstand and accuratelyrepair genome fragmentation, resulting in hundreds of double-strand breaks (DSBs) (2). Up to one-third of these breaks arerepaired by a RecA-independent pathway that is dependenton single-strand annealing mediated by DdrB (DNA damageresponse B) (3–6).DdrB was first identified from two independent analyses of the

IR-induced transcriptional response of Deinococcus radiodurans(7, 8). DdrB binds ssDNA (9); however, unlike SSB, DdrBpromotes accurate DNA-strand annealing (5, 10), providing ev-idence for a role in repair by single-strand annealing. This bi-ological role is further supported by fluorescence microscopydata demonstrating recruitment of DdrB to the nucleoid in theearly stages of repair (6) and attenuation of RecA-independentsingle-strand annealing repair in a ΔddrB strain (5). Althoughcomputational sequence analysis has delineated three distinctsuperfamilies of single-strand annealing proteins (11), similari-ties in function and quaternary structure suggest a sharedmechanism despite divergent evolutionary origins (12–14).

Here, we report the structure of a single-strand annealing pro-tein in complex with a partially annealed DNA intermediate to2.2 Å. In this structure, DNA is bound to a continuous flatsurface on one face of a pentameric DdrB ring assembly. Tworing structures come together in a face-to-face arrangement,sandwiching DNA strands at the interface, thereby stabilizinga partially annealed ss/dsDNA intermediate. Importantly, DdrBrestricts free access of one-third of bases within the 30 base pairintermediate, thereby preventing misannealing and ensuring suf-ficient time to further sample full complementarity of boundstrands before release of duplex DNA. This model is further val-idated by biochemical analysis showing that disruption ofDdrB–ssDNA interactions, which restrict access of unpaired bases,significantly decreases accuracy of strand annealing. On the basis ofthese results we present a general mechanism for how single-strandannealing proteins ensure accurate annealing of ssDNA.

DdrB Promotes Annealing by Sandwiching DNA Strands atthe Interface of Two Pentameric RingsPrior studies of annealing proteins such as DdrB, ICP8, andRad52 have led to a general model in which these proteins bindssDNA along a continuous surface lining one face of their olig-omeric ring structures (14–16). To further investigate this surfaceand its potential role in DNA annealing we determined thecrystal structure of DdrB in complex with single-strand DNA.An initial structure with noncomplementary 14b ssDNA strands(Fig. S1) confirmed ssDNA binding along one extended face of a

Significance

During repair of DNA double-strand breaks, cells must accuratelyanneal broken strands under temperatures that would normallypromote mispairing of even small stretches of ssDNA. How sin-gle-strand annealing (SSA) proteins such as Rad52 and DdrB (DNAdamage response B) overcome this thermodynamic barrier andachieve accurate strand pairing has remained unclear. Our struc-tural studies of DdrB in complex with partially annealed DNA andsupporting biochemical data reveal a mechanism for accurateannealing involving DdrB-mediated proofreading of strand com-plementarity. DdrB promotes high-fidelity annealing by con-straining specific bases from unauthorized association and onlyreleases annealed duplex when bound strands are fully comple-mentary. To our knowledge, this work provides the first mecha-nistic understanding for accurate strand pairing during SSA-dependent DNA double-strand break repair.

Author contributions: S.N.S.-M. and M.S.J. designed research; S.N.S.-M. and Y.M.W. per-formed research; S.N.S.-M. analyzed data; and S.N.S.-M. and M.S.J. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: Crystallography, atomic coordinates, and structure factors have beendeposited in the Protein Data Bank, www.pdb.org (PDB ID code 4NOE).1To whom correspondence should be addressed. Email: mjunop@uwo.ca.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1520847113/-/DCSupplemental.

4308–4313 | PNAS | April 19, 2016 | vol. 113 | no. 16 www.pnas.org/cgi/doi/10.1073/pnas.1520847113

Dow

nloa

ded

by g

uest

on

June

1, 2

020

DdrB pentameric ring. Unexpectedly, two DdrB ring structuresfurther assembled in a face-to-face complex, effectively sand-wiching DNA strands at the interface. In this “encounter complex”each DdrB subunit bound six bases (30 bases per pentamer), withtwo bases buried and the other four exposed (Fig. S1C). BecausessDNA was not designed for self-complementarity, only partialelectron density was observed for DNA at the “sandwich” in-terface, and most bases were not directly paired to the opposingstrand (Fig. S2). Nevertheless, the structure clearly showed whereDNA is bound and what an initial encounter complex looks likewhen noncomplementary strands are present. To further investi-gate how DdrB promotes accurate annealing we crystallized DdrBbound to ssDNA capable of forming a partially annealed ss/dsDNA intermediate. A substrate consisting of five tandem re-peats of 5′-TTGCGC was used, which enables the GCGCmotif toself-anneal, generating the potential to form a ss/dsDNA hybridwith four paired and two unpaired bases in each of the repeatsequences. The structure (Fig. 1) was solved by molecular re-placement and refined at 2.20 Å to Rwork = 18.8% and Rfree =23.3% (Table S1). All main chain residues were observed, exceptfor a few in loop regions joining β4–β5 and β6–β7 (Fig. 2A and

Fig. S3). Because of added stability available through basepairing, DNA from both strands was able to be entirely modeled.Similar to the encounter complex formed with noncomplemen-tary ssDNA, each subunit of the pentameric ring bound six basesin a repeating pattern of two buried through protein interaction(dT1, dT2) and four exposed (dG, dC, dG, dC) (Fig. 2C and Fig.S3). dT1 is held between L87 and L97 (Fig. 3A), whereas dT2forms a π–π interaction with the phenol group of Y125 (Fig. 3B).DNA binding to one face of the pentameric ring is consistent

with what has been proposed for human Rad52 on the basis ofmutational studies (17). In the case of Rad52, a heptameric ringassociates with a similar length of ssDNA (28 bases comparedwith 30 bases in DdrB) (16, 18), with each monomer interactingwith four bases (19). Interestingly, Rad52 generates a repetitivepattern of hydroxyl radical protection on bound ssDNA, withthree bases protected and the fourth sensitive to hydroxyl attack(19). In light of the current structures it seems likely that Rad52also binds ssDNA in a repetitive manner.

Protein–Protein Interaction Between DdrB Rings MediatesDNA Strand AssociationDdrB pentameric rings within the crystal form a face-to-faceencounter complex sandwiching DNA strands at the interface(Fig. 2B). Protein–protein interactions between ring structuresfunction to stabilize the encounter complex and also positionDNA strands in the correct polarity for proper base pairing.Exposed bases (GCGC motifs) within the encounter complex areheld in the correct distance and in the proper orientation to formthree hydrogen bonds with each of the corresponding bases fromthe opposing DdrB pentamer (Fig. 2C). Encounter complex sta-bility is therefore the net result of contributions from weak pro-tein–protein interactions, burying a surface area of 340 Å2, andbase pairing, which can account for up to an additional 1,150 Å2

of surface contact when full-strand complementarity is present.Given the large proportion of stability dependent on base pairing,the degree of strand complementarity appears to function as themain driver of complex stability and longevity. When strandcomplementarity is absent, an encounter complex could only existtransiently because of the limited amount of protein–proteininteraction available. In addition to these contributions to en-counter complex stability, DdrB also acts to inhibit nonproductiveencounter complexes by limiting the total amount of binding

Fig. 1. Surface electrostatic map of DdrB bound to ssDNA. (A) Top view of a30-base ssDNA bound to the positively charged surface of a DdrB pentamer.(B) Side view of the decameric complex formed by interaction of two DdrBpentamers. ssDNA strands bound by each pentamer are partially annealed atthe pentamer–pentamer interface.

Fig. 2. Structure of the DdrB–ss/dsDNA hybrid complex. (A) Top view of the asymmetric unit containing five monomers of DdrB (individually colored) and the30-base ssDNA substrate (yellow). Residues in loops joining β4–β5 and β6–β7 that were not traceable are represented by dotted lines. (B) Side view of the face-to-face assembly of DdrB nucleoprotein complexes. (C) Magnified view of the base-pairing interactions taking place at the complex interface. G-C hydrogen-bonding interactions are represented by black dashes with an average donor–acceptor distance of 2.9 Å for all 60 base pairs. All structural figures wereproduced with PyMol (www.pymol.org).

Sugiman-Marangos et al. PNAS | April 19, 2016 | vol. 113 | no. 16 | 4309

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

1, 2

020

energy available through base pairing. By restricting initial basepairing from 30 to 20 nucleotides, DdrB is able to reduce theupper limit of encounter complex stability significantly andtherefore ensure that the complex will only remain intact whenfully complementary strands are bound. Having more than thenecessary number of bases exposed would only serve to makediscriminating between full and partial complementarity morechallenging. Thus, encounter complex stability mediated byprotein–protein, base–base, and protein–base interaction is per-fectly balanced to promote strand contact and proper orientationwithout trapping strands in an unproductive state when non-complementarity is encountered.Because ssDNA was not preannealed before crystallization,

initial interaction of ssDNA with DdrB was not biased towardany particular residues being buried or exposed. Nevertheless,the crystal structure adopted the lowest possible energy state(maximum amount of annealing), implying that ssDNA is able tobe repositioned along the DdrB pentamer following initialbinding. It will be interesting to determine whether this involvesrapid equilibrium between free and bound states and/or slidingof ssDNA strands relative to the encounter complex until opti-mal base pairing is achieved.

DdrB Promotes Accurate ssDNA Annealing by RestrictingAccess of Unpaired BasesPartially restricting access of unpaired bases represents an ele-gant means of increasing annealing fidelity. If all bases were si-multaneously exposed within an encounter complex one wouldexpect there to be limited benefit to annealing accuracy overstrands free of secondary structure in solution. Discriminationbetween accurate and inaccurate annealing would still be con-founded by the small differences in binding energies betweenhomo- and hetero-duplex products. Systematically restrictingbase access would overcome this difficulty by allowing annealingto occur in a controlled two-stage process involving initial sam-pling for general complementarity with exposed bases (step one)and a second step checking for additional complementarity be-tween buried bases. Staged annealing would also make it possibleto sample long stretches of ssDNA (30 base) for complemen-tarity without committing to the release of partially annealedDNA until all bases (including those buried) are verified in acontrolled second step.To determine if DdrB uses a two-stage annealing process that

is dependent on initially restricting access to bound bases wecompared annealing of DNA strands containing varying degreesof mismatches for WT and mutant DdrB. In this assay, annealingwas negligible in the absence of DdrB (approximately 12% ± 3%compared with 58% ± 3% at 30 min) (Fig. 3A). As expected, DdrBsignificantly stimulated annealing of complementary strands (5)

(Fig. 3B); however, at higher protein concentrations (>1 μM) theefficiency of annealing decreased. A second mode of ssDNAbinding involving higher-order assembly of DdrB pentamers hasbeen observed (15); however, the functional significance remainsunclear. This effect is similar to what has been observed for humanRad52 (20) where reduced annealing at higher protein concen-tration was attributed to two opposing concentration-dependentmodes of DNA binding (wrapped vs. extended) (18). Neverthe-less, the fact that DdrB dramatically increases the efficiency ofannealing compared with the same strands free in solution un-derscores the important contribution of protein-mediated inter-actions within Deinococcus.Guided by the crystal structure, two mutants altering different

aspects of DdrB–DNA interaction were created and tested fortheir effect on strand annealing. K102A, which disrupts contactwith the sugar-phosphate backbone (Fig. 3) and reduces DNA-binding affinity, was included as a negative control (15). Con-sistent with its DNA-binding defect, K102A displayed no stim-ulation of DNA annealing (Fig. 4B). A second mutant, Y125A,was designed to permit unrestricted access of one of the twoburied bases (dT2). Y125 accounts for most of the direct stabi-lizing base interaction with dT2 (Fig. 3B). When substituted toan alanine, this loss of base interaction resulted in a significantenhancement in annealing activity relative to WT protein (Fig.4B), suggesting the release of buried bases within the encountercomplex increased binding energy of the initial encounter com-plex. To further test effects of Y125A on annealing fidelity, twoadditional ssDNA oligos with decreasing complementarity wereanalyzed. Substrates 12x (one mismatch site every 12 bases) and6x (one mismatch site every 6 bases) were assessed by the sameannealing assay with both WT DdrB and Y125A. AlthoughY125A was as efficient in annealing mismatched substrates asperfectly complementary strands (Fig. 4C), WT DdrB displayeddecreasing annealing efficiency with increasing numbers of mis-matches and surprisingly did not significantly improve annealingfidelity relative to the unassisted control. Under more biologicalconditions where other factors such as SSB are present, DdrBmay have increased annealing fidelity. Nevertheless, the re-markable effect of Y125A clearly demonstrates the critical im-portance of restricting bases at the encounter complex stage toensure optimal protein-assisted annealing fidelity.

A Mechanism for DdrB-Mediated AnnealingResults presented here provide structural evidence for a mech-anism of protein-mediated single-strand DNA annealing, whichwe term “Restricted Access Two-Step” ssDNA annealing(RATS). The structure of DdrB–ssDNA confirms earlier pre-dictions that single-strand annealing proteins such as DdrB andRad52 are able to bind ssDNA in an extended state along one

Fig. 3. DNA-binding residues. (A–C) Magnified views of protein–DNA interactions with nucleic acids dT1, dT2, and dG3–dC6, respectively. Hydrogen bondingand electrostatic interactions are represented by black dashes, and π–π interactions are represented by yellow dashes. Distances are measured in angstroms.

4310 | www.pnas.org/cgi/doi/10.1073/pnas.1520847113 Sugiman-Marangos et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

020

continuous surface of a protein oligomeric ring that ensuresstrands are presented free of secondary structure in a high-en-ergy state. Unexpectedly, the interaction does not permit equalaccess of all bases to solvent, but rather systematically buries twoout of six bases within each subunit (Fig. 2). This mode of ssDNAbinding is not dependent on DNA sequence or degree of strandcomplementarity, as a structure with entirely altered sequenceand less than half the annealing capacity also buried one-third ofbound bases (Fig. S1).In restricted access two-step ssDNA annealing, accurate

pairing of strands is achieved using two sequential searches forcomplementarity. In the first step, DdrB uses an elegant mech-anism to limit the search for complementarity to a subset ofbound bases. DNA strands initially bound to independent pen-tamers are brought together to form a transient encountercomplex primarily stabilized through weak protein–protein in-teractions at a stacked pentamer interface. In this complex, only20 out of 30 bound bases from each strand are permitted accessto the opposing strand. When an encounter complex harboringcomplementary strands is formed, the gain of 20 base pair con-tacts (Fig. 5A) would be expected to further stabilize pentamer–pentamer association but destabilize pentamer–ssDNA interac-tions due to increased torsional strain imposed on the sugarphosphate backbone during base pair formation. This is sup-ported by differences in the structure of DNA from encountercomplexes with complementary versus noncomplementary strands.B-factors for paired bases are significantly reduced, indicating in-creased stability. Further studies will be required to determineif the ability to flip out buried bases in the second stage ofannealing is the result of increased encounter complex stabilityand half-life and/or reduced affinity between individual pen-tamers and single-strand DNA. When full-strand complemen-tarity is achieved between exposed bases within the first step, asecond proofreading step is enabled and required to ensure fullcomplementarity. If only the first step was required, misanneal-ing would occur because stable duplex is readily formed with30-nt strands harboring 1–10 mismatches at biological tempera-tures. In further support for a second step, it would not havebeen possible to crystallize an encounter complex in an arrestedstate with partially annealed DNA if the encounter complexreleased duplex when only 20 out of 30 bases were comple-mentary. In the second stage of annealing, buried bases haveincreased opportunity to flip out and sample additional com-plementarity in the opposing strand. Whether this is the result ofincreased complex longevity or the result of destabilizing pro-tein–ssDNA interaction remains to be determined. Successfulpairing of buried bases, as modeled (Fig. 5B), would result in full

complementarity over all 30 bases and the release of highlystable B-form DNA that is unable to rebind DdrB (21). Thesecond proofreading step is made possible not only byrestricting access of bound bases, but also by exploiting thelimited base-stacking energy available in the extended con-figuration of bound ssDNA. In the extended conformation,exposed bases within the encounter complex do not stackoptimally and therefore have significantly reduced base-stacking interactions relative to B-form DNA. Thus, flipping anoncomplementary base out of a buried position (i.e., where it isbetter stabilized by interactions with residues such as Y125) (Fig.5D) would result in an unfavorable, high-energy transition. Un-der these conditions, noncomplementary bases would remainburied, and the search for complementarity would continue (Fig.5E). The need for a second proofreading step is best demon-strated by the effect Y125A substitution has on annealingfidelity. Premature release of even one of the two buried basesfrom DdrB permits significantly higher levels of annealing withnoncomplementary strands.

A Mechanism for Single-Strand AnnealingThe structure of the DdrB–ssDNA complex shares several keyfeatures with other DNA and RNA single-strand annealingproteins despite these proteins not being true homologs. Of note,Rad52 (16, 17), Redβ (12), Erf (22), RecT (23), Sak (24), ICP8(14), and Hfq (25) assemble into multisubunit ring structuressimilar to DdrB. Conservation of ring structures among single-strand annealing proteins, despite a lack of sequence or struc-tural similarity, is likely the result of convergent evolution.Multisubunit ring structures represent a simple way of formingdiscretely sized (20–30 bases) nucleic acid-binding interfaces thatpermit association of nucleic acid strands in a high-energy statefree of secondary structure. Another ssDNA-binding protein,ICP8 (14), is thought to form face-to-face arrangements ofstacked ring structures in which DNA strands are presented atthe interface. In the case of ICP8, an electron microscopy re-construction clearly illustrates a face-to-face orientation ofoligomeric rings that is ssDNA-dependent; however, because oflimited resolution, the EM structure could not inform de-finitively on the position of DNA or the detailed mechanism ofstrand annealing (14). The structure of Rad52 is not compatiblewith a stacked face-to-face complex and may involve a distinctmechanism from DdrB for strand annealing. However, withoutan ssDNA-bound Rad52 structure, it is difficult to say whethersome rearrangement of the protein would not occur, promoting aface-to-face complex. Similar questions remain for how smallnoncoding regulatory RNAs are annealed with transencoded

Fig. 4. Single-strand annealing assay. Annealing of a 60-base 5′-6FAM labeled oligonucleotide (5 nM) to M13mp18 (1 nM) was assessed by 1% agarose gelelectrophoresis. All reactions were performed at 32 °C, initiated by addition of the 60-base substrate, and quenchedwith 2.5 μMof unlabeled oligomer in 5% (wt/vol)SDS (final concentration of 250 mM and 0.5% SDS). (A) Time-course assay in the absence (◇) and presence (●) of 750 nM DdrB. (B) Concentration-dependentannealing by WT (●), K102A (□), and Y125A (△) DdrB. Reactions were quenched after 20 min. (C) Annealing with complementary (c.) and mismatched (12xand 6x) 5′-6FAM oligonucleotides in the absence and presence of 750 nM WT or mutant (Y125A) DdrB. Reactions were quenched after 20 min.

Sugiman-Marangos et al. PNAS | April 19, 2016 | vol. 113 | no. 16 | 4311

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

1, 2

020

target mRNAs by the ring structure of Hfq. Hfq, a well-charac-terized member of the Hfq–Sm–LSm protein family (26), hasbeen crystallized in complex with ssRNA (25). In this structure,ssRNA is bound along a continuous surface of a hexameric Hfq

ring, with each monomer positioning two bases inward and oneoutward. Although the precise mechanisms are likely to differ onthe basis of biological function, it will be interesting to seewhether it is possible to trap an Hfq encounter complex with

Fig. 5. DdrB-mediated ssDNA annealing. A face-to-face arrangement of DdrB nucleoprotein complexes facilitates the search for homology between twostrands of ssDNA. (A) Base-pairing interactions between exposed bases transiently stabilize the complex, and the buried bases (associated with L87, L97, andY125) sample the opposing strand for complementarity. (B) Accurate pairing of buried bases, as modeled, results in release of the bound DNA (C).(D) Mispairing of buried bases, as modeled, results in a continuation of the search for homology or dissociation of the decameric complex with DNA stillbound to DdrB (E).

4312 | www.pnas.org/cgi/doi/10.1073/pnas.1520847113 Sugiman-Marangos et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

020

partially annealed RNA duplex and whether annealing proceedsthrough a staged annealing mechanism similar to DNA annealing.The structure of encounter complexes bound to noncomple-

mentary and cDNA strands presented here provides high-resolutioninsight into a mechanism of ssDNA annealing. Findings areconsistent with previous observations from other SSA proteins,such as ICP8, and reveal how cells can use a multistep annealingmechanism involving restricted access of bound bases to over-come the difficult challenge of protecting single-strand DNAfrom degradation and secondary structure formation and at thesame time promote single-strand annealing.

Experimental ProceduresCrystallization. DdrB1–144 from D. radiodurans was expressed and purified asreported previously (21). Crystals were grown by the hanging-drop vapordiffusion method. Briefly, 1 μL of DdrB1–144/ssDNA (480 μM protein, 143 μMDNA) was mixed with 1 μL of crystallization condition [6% (wt/vol) PEG 2000,0.1 M Hepes pH 6.5, 0.05 M CaCl2] and 0.2 μL of Hampton Additive Screencondition 48 (0.01 M L-Glutathione reduced, 0.01 M L-Glutathione oxidized)and equilibrated over 1.2 M ammonium sulfate at 20 °C. No additional cryo-protectant was required for data collection. Diffraction data were collected ata wavelength of 1.1 Å on beamline ×25 at the National Synchrotron LightSource at Brookhaven National Laboratory. Decameric DdrB–ssDNA complexformed before crystallization.

Structure Solution. The diffraction data were integrated withmosflm (27) andthen scaled, merged, and converted to structure factors with CCP4 (28, 29).Structure solution was performed with AutoMR from the Phenix softwarepackage (30) using the DdrB1–144 monomer from D. radiodurans (PDBID4HQB) as a search model. The best molecular replacement solution wasgenerated using a search model with manually truncated loops regions,which was then rebuilt-in-place into a simulated annealing OMIT map usingPhenix-AutoBuild. Missing loop regions (87–101 and 105–125) and DNAbases were placed manually and refined in iterative cycles using Coot (31)and Phenix-Refine until R and Rfree values converged and geometry statisticsreached suitable ranges (Table S1). Validation of the final structure using

Ramachandran analysis indicated that 99.1% of amino acids fell within fa-vored positions (0% outliers).

DNA Substrates. DNA substrates used in crystallography and annealing assayswere purchased from Integrated DNA Technologies. The substrate used togenerate cocrystals containing a ss/dsDNA hybrid consisted of a single oligo-nucleotide (5′-TTGCGCTTGCGCTTGCGCTTGCGCTTGCGC). The 6FAM labeled60-base probe (5′- ACACTGAGTTTCGTCACCAGTACAAACTACAACGCCTGTAG-CATTCCACAGACAGCCCTC) is complementary to a region of M13mp18,whereas substrates 12x (5′-ACACTGAGTTTTGTCACCAGTACGAACTACAACGC-TTGTAGCATTCCGCAGACAGCCCTT) and 6x (5′-ACACTAAGTTTTGTCACTAGTA-CGAACTATAACGCTTGTAGTATTCCGCAGACGGCCCTT) are complementary tothe same region with one mismatch every 12 or 6 bases, respectively.

Annealing Assay. DdrB mutants (K102A and Y125A) were generated by site-directed mutagenesis of the WT expression vector (pPROEX-HT_DR0070) andwere purified as previously described (15). Single-strand DNA annealingactivity of DdrB was assessed using a 60-base 5′-6FAM labeled probe (5 nM)complementary to a region of M13mp18 (1 nM). Annealing reactions (25 μL)were carried out at 32 °C in annealing buffer [50 mM Tris pH 7.5, 100 mMNaCl, 25 mM Mg(CH3COO)2, 1 mM DTT, 5% (vol/vol) glycerol] and wereinitiated by the addition of the 60-base probe (1.25 μL of 100 nM). Reactionswere quenched with 2.5 μL of 2.5 μM unlabeled oligomer in 5% (wt/vol) SDSand resolved using a 1% agarose gel run in TBE buffer (90 mM Tris-boratepH 8.3, 2 mM ethylenediaminetetraacetic acid) at 50 V. Annealing reactionproducts were quantified with a GE Amersham Typhoon Trio+ variable-mode imager and ImageJ (32) and were assessed relative to a controlledannealing reaction carried out in a thermocycler in the absence of protein.

ACKNOWLEDGMENTS. We thank Annie Héroux for her technical assistancein the collection of structural data on beamline X25 at the National Synchro-tron Light Source, Brookhaven National Laboratories. We also thank ChrisBrown for editorial comments during the preparation of the manuscript.This work was supported by the Natural Sciences and Engineering ResearchCouncil of Canada through Grant 2008R00075 (to M.S.J.) and a studentship(to Y.M.W.).

1. Pestryakov PE, Lavrik OI (2008) Mechanisms of single-stranded DNA-binding proteinfunctioning in cellular DNA metabolism. Biochemistry (Mosc) 73(13):1388–1404.

2. Battista JR (1997) Against all odds: The survival strategies of Deinococcus radiodurans.Annu Rev Microbiol 51:203–224.

3. Daly MJ, Minton KW (1996) An alternative pathway of recombination of chromo-somal fragments precedes recA-dependent recombination in the radioresistant bac-terium Deinococcus radiodurans. J Bacteriol 178(15):4461–4471.

4. Slade D, Lindner AB, Paul G, Radman M (2009) Recombination and replication in DNArepair of heavily irradiated Deinococcus radiodurans. Cell 136(6):1044–1055.

5. Xu G, et al. (2010) DdrB stimulates single-stranded DNA annealing and facilitatesRecA-independent DNA repair in Deinococcus radiodurans. DNA Repair (Amst) 9(7):805–812.

6. Bouthier de la Tour C, et al. (2011) The deinococcal DdrB protein is involved in an earlystep of DNA double strand break repair and in plasmid transformation through itssingle-strand annealing activity. DNA Repair (Amst) 10(12):1223–1231.

7. Liu Y, et al. (2003) Transcriptome dynamics of Deinococcus radiodurans recoveringfrom ionizing radiation. Proc Natl Acad Sci USA 100(7):4191–4196.

8. Tanaka M, et al. (2004) Analysis of Deinococcus radiodurans’s transcriptional responseto ionizing radiation and desiccation reveals novel proteins that contribute to ex-treme radioresistance. Genetics 168(1):21–33.

9. Norais CA, Chitteni-Pattu S, Wood EA, Inman RB, Cox MM (2009) DdrB protein, analternative Deinococcus radiodurans SSB induced by ionizing radiation. J Biol Chem284(32):21402–21411.

10. Lockhart JS, DeVeaux LC (2013) The essential role of the Deinococcus radiodurans ssbgene in cell survival and radiation tolerance. PLoS One 8(8):e71651.

11. Iyer LM, Koonin EV, Aravind L (2002) Classification and evolutionary history of thesingle-strand annealing proteins, RecT, Redbeta, ERF and RAD52. BMC Genomics3:8–18.

12. Passy SI, Yu X, Li Z, Radding CM, Egelman EH (1999) Rings and filaments of betaprotein from bacteriophage lambda suggest a superfamily of recombination pro-teins. Proc Natl Acad Sci USA 96(8):4279–4284.

13. Erler A, et al. (2009) Conformational adaptability of Redbeta during DNA annealingand implications for its structural relationship with Rad52. J Mol Biol 391(3):586–598.

14. Tolun G, Makhov AM, Ludtke SJ, Griffith JD (2013) Details of ssDNA annealing re-vealed by an HSV-1 ICP8-ssDNA binary complex. Nucleic Acids Res 41(11):5927–5937.

15. Sugiman-Marangos SN, Peel JK, Weiss YM, Ghirlando R, Junop MS (2013) Crystalstructure of the DdrB/ssDNA complex from Deinococcus radiodurans reveals a DNAbinding surface involving higher-order oligomeric states. Nucleic Acids Res 41(21):9934–9944.

16. Singleton MR, Wentzell LM, Liu Y, West SC, Wigley DB (2002) Structure of the single-

strand annealing domain of human RAD52 protein. Proc Natl Acad Sci USA 99(21):

13492–13497.17. KagawaW, et al. (2002) Crystal structure of the homologous-pairing domain from the

human Rad52 recombinase in the undecameric form. Mol Cell 10(2):359–371.18. Grimme JM, et al. (2010) Human Rad52 binds and wraps single-stranded DNA and

mediates annealing via two hRad52-ssDNA complexes. Nucleic Acids Res 38(9):

2917–2930.19. Parsons CA, Baumann P, Van Dyck E, West SC (2000) Precise binding of single-stranded

DNA termini by human RAD52 protein. EMBO J 19(15):4175–4181.20. Lloyd JA, Forget AL, Knight KL (2002) Correlation of biochemical properties with the

oligomeric state of human rad52 protein. J Biol Chem 277(48):46172–46178.21. Sugiman-Marangos S, Junop MS (2010) The structure of DdrB from Deinococcus: A

new fold for single-stranded DNA binding proteins. Nucleic Acids Res 38(10):

3432–3440.22. Poteete AR, Sauer RT, Hendrix RW (1983) Domain structure and quaternary organi-

zation of the bacteriophage P22 Erf protein. J Mol Biol 171(4):401–418.23. Thresher RJ, Makhov AM, Hall SD, Kolodner R, Griffith JD (1995) Electron microscopic

visualization of RecT protein and its complexes with DNA. J Mol Biol 254(3):364–371.24. Ploquin M, et al. (2008) Functional and structural basis for a bacteriophage homolog

of human RAD52. Curr Biol 18(15):1142–1146.25. Link TM, Valentin-Hansen P, Brennan RG (2009) Structure of Escherichia coli Hfq

bound to polyriboadenylate RNA. Proc Natl Acad Sci USA 106(46):19292–19297.26. Vogel J, Luisi BF (2011) Hfq and its constellation of RNA. Nat Rev Microbiol 9(8):

578–589.27. Leslie GWA, Powell HR (2007) Processing diffraction data with mosflm. Evolving

Methods for Macromolecular Crystallography 245:41–51.28. Winn MD, et al. (2011) Overview of the CCP4 suite and current developments. Acta

Crystallogr D Biol Crystallogr 67(Pt 4):235–242.29. French S, Wilson K (1978) On the treatment of negative intensity observations. Acta

Crystallogr A 34(4):517–525.30. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macro-

molecular structure solution. Acta Crystallogr D Biol Crystallogr 66(Pt 2):213–221.31. Emsley P, Cowtan K (2004) Coot: Model-building tools for molecular graphics. Acta

Crystallogr D Biol Crystallogr 60(12):2126–2132.32. Schneider CA, Rasband WS, Eliceiri KW (2012) NIH Image to ImageJ: 25 years of image

analysis. Nat Methods 9(7):671–675.

Sugiman-Marangos et al. PNAS | April 19, 2016 | vol. 113 | no. 16 | 4313

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

1, 2

020