Mechanism of staphylococcal multiresistance plasmidreplication origin assembly by the RepA proteinMaria A. Schumachera,1, Nam K. Tonthata, Stephen M. Kwongb, Naga babu Chinnama, Michael A. Liub,Ronald A. Skurrayb, and Neville Firthb

aDepartment of Biochemistry, Duke University Medical Center, Durham, NC 27710; and bSchool of Biological Sciences, University of Sydney, Sydney, NSW2006, Australia

Edited by James M. Berger, The Johns Hopkins University School of Medicine, Baltimore, MD, and approved May 15, 2014 (received for review April 1, 2014)

The staphylococcal multiresistance plasmids are key contributorsto the alarming rise in bacterial multidrug resistance. A conservedreplication initiator, RepA, encoded on these plasmids is essentialfor their propagation. RepA proteins consist of flexibly linkedN-terminal (NTD) and C-terminal (CTD) domains. Despite their essen-tial role in replication, the molecular basis for RepA function isunknown. Here we describe a complete structural and functionaldissection of RepA proteins. Unexpectedly, both the RepA NTD andCTD show similarity to the corresponding domains of the bacterialprimosome protein, DnaD. Although the RepA and DnaD NTD bothcontain winged helix-turn-helices, the DnaD NTD self-assemblesinto large scaffolds whereas the tetrameric RepA NTD binds DNAiterons using a newly described DNA binding mode. Strikingly,structural and atomic force microscopy data reveal that the NTDtetramer mediates DNA bridging, suggesting a molecular mecha-nism for origin handcuffing. Finally, data show that the RepA CTDinteracts with the host DnaG primase, which binds the replicativehelicase. Thus, these combined data reveal themolecular mechanismby which RepA mediates the specific replicon assembly of staphy-lococcal multiresistant plasmids.

RepA_N | replication initiation | S. aureus | two hybrid

The emergence of multidrug-resistant bacteria is a mountingglobal health crisis. In particular, multidrug-resistant Staph-

ylococcus aureus is a major cause of nosocomial and community-acquired infections and is resistant to most antibiotics commonlyused for patient treatment (1). Hospital intensive care units inmany countries, including the United States, now report methi-cillin-resistant S. aureus infection rates exceeding 50% (2, 3).Antibiotic resistance in contemporary infectious S. aureus strains,such as in hospitals, is often encoded by plasmids that can betransmitted between strains via horizontal DNA transfer mech-anisms. These plasmids are typically classified as small (<5 kb)multicopy plasmids, which usually encode only a single resistancegene; medium-sized (8–40 kb) multirestance plasmids that con-fer resistance to multiple antibiotics, disinfectants, and/or heavymetals; and large (>40 kb) conjugative multiresistance plasmidsthat additionally encode a conjugative DNA transfer mechanism(4–6). Importantly, sequence analyses have shown that moststaphylococcal conjugative and nonconjugative multiresistanceplasmids encode a highly conserved replication initiation protein,denoted RepA_N (5–15). RepA_N proteins are also encoded byplasmids from other Gram-positive bacteria as well as by somephage, underscoring their ubiquitous nature (10). These RepAproteins are essential for replication of multiresistance plasmids,and hence plasmid carriage and dissemination, yet the mech-anisms by which these proteins function in replication are cur-rently unknown.The DNA replication cycle can be divided into three stages:

initiation, elongation, and termination. Replication initiationproteins (RepA) mediate the crucial first step of initiation. Bac-terial chromosome replication is initiated by the chromosomalreplication initiator protein, DnaA, which binds the origin andrecruits the replication components known as the primosome

(16). In Gram-negative bacteria the primosome includes DnaGprimase, the replicative helicase (DnaB), and DnaC (17). Rep-lication initiation in Gram-positive bacteria involves DnaG pri-mase and helicase (DnaC) and the proteins DnaD, DnaI, andDnaB (18–22). DnaD binds first to DnaA at the origin. This isfollowed by binding of DnaI/DnaB and DnaG, which togetherrecruit the replicative helicase (23, 24). Instead of DnaA, plas-mids encode and use their own specific replication initiatorbinding protein. Structures are only available for RepA proteins(F, R6K, and pPS10 Rep) harbored in Gram-negative bacteria.These proteins contain winged helix-turn-helix (winged HTH)domains and bind iteron DNA as monomers to, in some stillunclear manner, drive replicon assembly (25–27).Replication mechanisms used by plasmids harbored in Gram-

positive bacteria are less well understood and are distinct fromtheir Gram-negative counterparts. Indeed, most plasmid RepAproteins in Gram-negative and Gram-positive bacteria showno sequence homology and seem to be unrelated. The multi-resistance RepA proteins are arguably among the most abundantof plasmid Rep proteins, yet how they function is not known.Data suggest that these proteins are composed of three mainregions: an N-terminal domain (NTD) consisting of ∼120 aa, a longand variable linker region (∼30–50 residues), and a C-terminaldomain (CTD) of ∼120 residues (28–31). The NTD and CTDare both essential for replication. The NTD exhibits the highestlevel of sequence conservation, which has resulted in the desig-nation of plasmids that encode these proteins as the RepA_Nreplicon family (10). Although not as well conserved as the NTD,RepA CTD regions show homology between plasmids found in

Significance

The large staphylococcal multiresistance plasmids harbored inGram-positive pathogens contribute significantly to the spreadof multidrug-resistant bacteria and are typified by the presenceof a highly conserved replication initiator protein, RepA, whichis required for plasmid retention. RepA proteins containN-terminal (NTD) and C-terminal (CTD) domains, which are bothrequired for replication. We show that the RepA NTD and CTDshow striking homology to the host primosome protein DnaDyet perform distinct functions; the NTD binds origin DNA ina novel manner and the CTD recruits the replicative helicase.Moreover, NTD–DNA structures reveal the first mechanism oforigin handcuffing. Combined, the data unveil the minimalmechanism by which multiresistance plasmids mediate originassembly via the highly conserved RepA protein.

Author contributions: M.A.S., S.M.K., R.A.S., and N.F. designed research; M.A.S., N.K.T.,S.M.K., N.b.C., and M.A.L. performed research; M.A.S., N.K.T., and N.F. analyzed data; andM.A.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: Coordinates and structure factor amplitudes have been deposited in theProtein Data Bank (PDB), www.pdb.org (PDB ID codes 4PQK, 4PQL, 4PT7, and 4PTA).1To whom correspondence should be addressed. E-mail: maria.schumacher@duke.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1406065111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1406065111 PNAS Early Edition | 1 of 6

BIOCH

EMISTR

YCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

May

27,

202

0

genus-specific clusters, suggesting that this domain may performa host-specific role (28–32). Although the function of the RepACTD remains enigmatic, recent studies have indicated that theNTD mediates DNA binding and interacts with iterons that re-side within the plasmid origin (30). The essential roles played byRepA proteins in multiresistance plasmid retention marks themas attractive targets for the development of specific chemo-therapeutics. However, the successful design of such com-pounds necessitates structural and mechanistic insight. Here,we describe a detailed dissection of the RepA proteins from themultiresistance plasmids pSK41 and pTZ2126. The combineddata reveal the molecular underpinnings of a minimalist repli-cation assembly mediated by multiresistance RepA proteins.

Results and DiscussionStructure of the RepA NTD Contains a Winged HTH with a UniqueDimer-of-Dimers Arrangement. To gain insight into the functionof the RepA NTD, which is conserved in more than 100 proteins,structures were determined of the pSK41 and pTZ2162 RepANTDs. The pSK41 NTD structure was determined first to 2.60 Å(Table S1). The structure revealed that the RepA NTD has anoverall fold composed of a central winged HTH (residues 33–112) flanked by an N-terminal helix-strand-helix (residues 1–27)and C-terminal helix-loop motif (residues 116–132), with overalltopology [α1(residues 8–14)-β1(16–20)-α2(21–27)-α3(33–53)-α4(67–76)-α5(79–94)-β2(94–100)-β3(106–112)-α6(116–129)] (Fig.1A). The helix-strand-helix and C-terminal helix function asoligomerization elements to create an extensive dimer-of-dimersor tetramer (Fig. 1 B and C). To form dimers, the β1 strandsfrom two NTDs interdigitate via antiparallel interactions. Thedimer is further stabilized by multiple contacts between residuesin helices α1 and α6 (Fig. 1B). The resultant dimer interface isextensive and buries 2,005 Å2 of protein surface from solvent.The RepA tetramer is created by the orthogonal packing of twoNTD dimers. Key to the formation of this compact tetramer isthe deep insertion of the N-terminal tails of each subunit intocomplementary hydrophobic cavities located at the junction ofβ2–β3 and α3 (Fig. 1C). The residues mediating oligomerizationare highly conserved (Fig. S1). Moreover, tetramer formationburies 6,865 Å2 of surface area, supporting the notion that RepAis tetrameric.A second pSK41 RepA NTD structure was solved to 3.20 Å

and two structures of the pTZ2162 RepA NTD (70% identicalto pSK41 NTD) were determined to 2.35 Å and 2.45 Å. Notably,all of the structures revealed the same tetramer, which can besuperimposed with rmsds of 0.56–0.78 Å (Fig. 1D). The expan-sive surface area buried by tetramer formation as well as the factthat four different crystal structures revealed the same tetramerprovided compelling support that the RepA NTD is tetrameric.However, to assess the NTD oligomeric state in solution, size-exclusion chromatography (SEC) and small-angle X-ray scat-tering (SAXS) studies were carried out. SEC experimentsrevealed that both the pTZ2162 and pSK41 RepA NTDs elutedas tetramers (Fig. 1E). SAXS data analysis produced radius-of-gyration (Rg) values for the pTZ2162 RepA NTD and pSK41RepA NTD of ∼29 Å and 30 Å, which compares well with thecalculated tetrameric Rgs of 27 and 26 Å, but not with the Rgvalues for monomers and dimers of 18 Å and 20 Å, respectively(Fig. 1F). Thus, the collective data all support that the RepANTD is tetrameric. The large number of conserved RepA pro-teins, and specifically the RepA NTD, has led the National Centerfor Biotechnology Information (NCBI) Conserved DomainDatabase to classify RepA_N as its own NCBI category, pfam-6970(10). The unique tetrameric fold revealed by the RepA NTDstructures can now be assigned to this ubiquitous and importantprotein superfamily.It is notable that the multiresistance RepA proteins contain

a winged HTH, because structures of the F, R6K, and pPS10RepA proteins encoded on plasmids harbored in Gram-negativebacteria also contain this motif (26, 27). The winged HTH ele-ments, however, are the only shared structural feature of these

two Rep families; in contrast to pSK41 and pTZ2162 RepA, theF, R6K, and pPS10 RepA proteins consist of two tandem wingedHTH domains and function as monomers (26, 27). Thus, theseRepA proteins belong to distinct structural families. To identifypossible structural homologs of the RepA NTD, databasesearches were performed. The only protein that showed ho-mology with RepA was the NTD of the Geobacillus kaustophilusDnaD protein (PDB ID code 2VN2), which although onlysharing 13% sequence identity with RepA forms similar oligo-meric arrangements (33). The DnaD and RepA subunits anddimers can be superimposed with rmsds of 3.0 Å and 3.5 Å,respectively (Fig. 1D). The C-terminal regions of the protein,however, differ, leading to the formation of distinct tetramers(rmsd of 5.0 Å) (Fig. 1D, Center). Although the DnaD and RepANTDs display structural similarities they perform diverse func-tions. Specifically, the DnaD NTD does not bind DNA but self-assembles into large scaffold structures, aiding primosome as-sembly, and the RepA NTD carries out the key role of DNAiteron binding at the plasmid origin (30).

Fig. 1. Structure of the multiresistance RepA NTD. (A) Ribbon diagram ofthe pTZ2162 RepA NTD. The winged HTH is colored gray and secondarystructural elements and the N and C termini are labeled. Ribbon diagramswere made using PyMOL (39). (B) RepA NTD dimer. The pink subunit isshown in the same orientation as in A. (C) RepA NTD tetramer. The pink andyellow subunits are in the same position as in B. (D) Superimposition of RepANTD structures determined in this study. The pSK41 tetramers are coloredshades of blue and the pTZ2162 tetramers, orange. (Right) Overlay of theRepA NTD tetramer and dimer (cyan) onto the DnaD NTD tetramer and di-mer (red). The dimers overlay well with the exception of the C-terminal helix.The differences in these C-terminal helices leads to alternate tetramer for-mation of the proteins. (E) SEC analyses. The blue cross represents thestandard curve and pTZ2162 and pSK41 RepA NTD samples are representedas a green circle and yellow square, respectively. (F) SAXS curves of pTZ2162and pSK41 RepA NTD samples plotted against calculated curves based on thecrystal structure tetramers.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1406065111 Schumacher et al.

Dow

nloa

ded

by g

uest

on

May

27,

202

0

RepA NTD–DNA Structure Reveals Basis for Iteron Recognition. Iter-ons boxes bound by the RepA_N family of proteins consist ofrepeats that are ∼20–25 bp (10, 29, 30). Plasmid origins typicallycontain three and sometimes four such boxes. Although boxesbound by different RepA proteins are divergent, they all containextended AT tracts, usually located in the center of each repeat(10). DNase I protection studies demonstrated that pSK41 RepAbinding led to a periodic and nearly complete protection of theiteron region, and electrophoretic mobility shift experiments,which revealed a single supershift upon RepA addition, in-dicated that RepA binding is cooperative (30). To gain insightinto how multiresistance RepA proteins recognize iteron DNA,the structure of the pTZ2162 RepA NTD bound to a 32merbased on pTZ2162 iteron 2 was determined to 3.09 Å (Fig. 2A)(SI Methods and Table S2). To obtain crystals, excess proteinover DNA was necessary and in the structure only one dimer ofthe tetramer is bound to the 32mer DNA. The recognition he-lices, α5, of each HTH (α4–α5) of the RepA dimer dock onto theDNA major grooves, and the wings insert into the adjacent mi-nor grooves (Fig. 2 A and B). Except for folding of the wings,which are largely disordered in the apo structures, there are nostructural changes in the RepA NTD tetramer upon DNAbinding (rmsd of 0.8 Å for Cα superimposition of DNA-boundand apo forms). However, to accommodate binding of consec-utive RepA winged HTH elements, the DNA is significantly bentat the central AT-rich region by 30°. Because AT tracts areknown to be intrinsically bent, the finding that RepA-bounditerons must be distorted to bind RepA_N proteins suggests anexplanation for the conservation of AT tracts in their origins(10, 34). Consistent with this supposition, fluorescence po-larization (FP) DNA binding assays revealed that substitution ofthe AT-rich region to a GC tract reduced the binding affinity ofRepA by ∼10-fold (Fig. 2C). In addition to the compressedcentral minor groove, the major grooves where the recogni-tion helices dock are widened (13 Å compared with 11 Å forB-DNA) and the wings that bind near the ends of the DNAcause distortion, including DNA untwisting. These DNA defor-mations suggest that RepA may participate in or facilitate DNAmelting. Indeed, FP studies show that the NTD can bind ssDNA,which could aid and stabilize origin strand opening (Fig. S2).

The distorted DNA adopts an optimal shape for interactingwith the electropositive RepA NTD, as well as for permittingdirect protein–DNA contacts by the wings and recognition he-lices of the HTHs (Fig. 2 B and D). The deep insertion of thewing is made possible by the presence of Gly101 at the tip of thewing. As expected given its key role in DNA binding, this glycineis one of the most conserved residues among these RepA pro-teins (Fig. S1). The long side chain of wing residue Arg99 rec-ognizes O2 moieties on adjacent pyrimidine nucleotides (Fig.2D). Wing residues Leu102 and Asn103 make additional baseinteractions. Resides from the recognition helices provide fur-ther specificity in RepA DNA recognition by reading nucleo-bases in the major grooves. Lys79 hydrogen bonds to the O6atom of conserved guanines and Glu80 reads the N6 atoms ofadenosine bases and also the O2 atom of a thymine. Finally,Thr83 is positioned to make van der Waals contacts to thymines.In sum, the pTZ2162 RepA NTD-iteron structure indicates thatbase-specific contacts and DNA deformability, in particular thatmediated by the central AT-rich region, are key for RepA DNAbinding specificity. The pTZ2162 RepA NTD-iteron structurewas obtained using a 32-bp element that contained the 23-bppredicted iteron box as well as regions 5′ and 3′ of the repeatbecause the minimal motif required for specific binding by asingle RepA unit had been unclear (29–31). Unexpectedly, thestructure revealed that the minimal site required to permit op-timal contacts to the DNA extended beyond the 23mer motifsuch that the wings of adjacently bound dimers would insert intothe same minor groove (Fig. 2E, Top). Modeling indicates thatthese consecutively bound wings could occupy the same minorgroove without overlap or clash. Interestingly, this modelingexercise also suggests the possibility that, in addition to possiblewing–wing contacts, adjacently bound RepA proteins could in-teract (Fig. 2E). Thus, DNA binding cooperativity could arisefrom minor groove widening by binding the first wing (allowingthe second to bind more readily), wing–wing contacts, and/orinteractions between RepA molecules outside the wing region.The cooperative interaction of multiple RepA proteins to suc-cessive Rep boxes could amplify local untwisting of the DNA.Because excess protein was required to obtain the pTZ2162

RepA NTD-iteron crystals, only one NTD dimer of the tetramer

Fig. 2. Structure of the pTZ2162 RepA–iteron DNAcomplex. (A) pTZ2162 RepA NTD–iteron DNA com-plex. (B) Electrostatic surface representation of thepTZ2162 RepA NTD–iteron complex. (C) Semi-logplot of FP RepA NTD binding to the WT iteron (blue)and one in which the AT tract was replaced by aGC tract (red). (D, Upper) RepA–DNA contacts. (D,Lower) Schematic showing the RepA NTD residuesthat contact bases and the phosphate backbone. (E)Sequence of the pTZ2162 origin site showing fourRep boxes (Top). Repeats originally identified arecolored and the regions bound by each RepA com-plex are under- and overlined. pTZ2162 RepA NTDiteron binding model constructed by docking suc-cessive RepA NTDs onto consecutive Rep boxes aswould be located within the pTZ2162 origin. Shownas close up view is the modeled wing–wing in-teraction. (Middle) RepA NTD binding to two dis-tinct Rep boxes that would be located on differentDNA substrates. (Bottom) A model of RepA tetra-mer binding between Rep boxes on one duplexorigin site. This is only possible if the tetramerbridges between boxes 1 and 3 or 1 and 4 (or 2 and4). However, this model is not supported by DNase Iprotection studies (indicated by the cross throughthe model) (30).

Schumacher et al. PNAS Early Edition | 3 of 6

BIOCH

EMISTR

YCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

May

27,

202

0

was bound to DNA. However, modeling shows that each dimerof the tetramer could bind Rep boxes located on distinct DNAduplexes simultaneously to mediate bridging (Fig. 2E, Middle).By contrast, one RepA tetramer could not wrap two consecutiveRep boxes on the same strand about itself because, as noted,RepA molecules bind close in space to directly abutted iterons.Hence, looping within a single origin would require that RepAproteins bind to nonconsecutive iterons (e.g., iterons 1–3, 1–4, or2–4). However, DNase I protection studies on RepA–origininteractions clearly revealed that when RepA binds, every box isprotected, which would not be consistent with nonconsecutivelooping (28–30). Moreover, the pattern of protection in thefootprints is strikingly periodic, which indicates essentiallyidentical modes of RepA binding to each iteron, in particular forboxes 2–4 in the pSK41 origin (Fig. S3) (30). Box 1 is also pro-tected but compared with boxes 2–4 shows a slightly alteredpattern, likely owing to its departure from the iteron consensus.Nonetheless, the combined data indicate that the origin iteronsare bound by adjacent RepA proteins. Interestingly, however,the RepA proteins bound to all of the iterons within an originwould each have one of its dimer faces free for potential in-teraction with DNA at a distance or on a separate duplex. Hence,to probe the possible higher-order structures that may be formedby RepA on DNA, atomic force microscopy (AFM) was used.

Visualization of RepA-Mediated DNA Bridging Between DNA Duplexesand Atomic Model for Plasmid Copy Number Control by Handcuffing.For AFM studies, DNA sites containing the pSK41 and pTZ2162iteron boxes embedded at the center of 700-bp DNA fragmentswere used (SI Methods). Samples containing pTZ2162 or pSK41RepA NTD formed similar complexes. Consistent with model-ing, the images reveal no looping within one DNA duplex andthe entire central, iteron-containing regions are occupied byRepA proteins (Fig. S4). Strikingly, however, when complexeswere imaged with high DNA concentrations (SI Methods) allimages revealed that two separate DNA strands are linked to-gether between RepA molecules, indicating that the RepA-bound tetramers can mediate DNA bridging (Fig. 3 A–C, i–iii).The DNA bridging visualized by AFM suggests a molecularmechanism for an important means of plasmid copy numberregulation termed “handcuffing.” Regulation of replication isessential for stable plasmid retention in host cells. Plasmids mustbe kept to a minimum to avert metabolic overburdening of thehost cell and out-competition by cells with fewer plasmids.RepA_N replicons are known to be regulated via antisense RNAmolecules, which prevent excess RepA expression and as a resultthe overproduction of plasmids (10). However, studies indicatethat multiple mechanisms of replicon regulation provide optimalcontrol of replication and copy number production (35). Hand-cuffing sterically hinders the initiation process and preventsoverreplication. Although Chattoraj and coworkers (35) pro-posed this mechanism decades ago, how the monomeric F, R6K,and psS10 RepA proteins could mediate this process has beenunclear. RepA tetramer-mediated bridging suggests an idealmolecular mechanism to explain handcuffing. Further evidencefor RepA DNA bridging was provided by a pTZ2162 RepANTD-15mer crystal structure (3.6 Å resolution), which showedtwo DNA duplexes sandwiched between a RepA tetramer(Fig. S5).Isothermal titration calorimetry (ITC), which can provide

binding stoichiometries, was next used to probe DNA bindingand possible DNA bridging by the pTZ2162 and pSK41 RepANTDs. The resulting isotherms of the NTDs binding to singleiteron DNA sites revealed stoichiometries of one RepA NTDdimer:one DNA duplex, or one RepA NTD tetramer binding totwo DNA molecules, for both pTZ2162 and pSK41 RepA, con-sistent with DNA bridging (Fig. 3D). To provide a final assessmentof RepA NTD-mediated DNA bridging, SAXS experiments werecarried out on the pTZ2162 RepA NTD in complex with a 100-bpiteron-containing DNA site. A model constructed of a RepANTD tetramer bridged between two 100mer DNA sites provided

the best fit to the SAXS data (Rg of the model was 87 Å and thedata were 87 ± 5 Å) (SI Methods and Fig. S6). Thus, AFM, X-raycrystallography, ITC, and SAXS analyses all provide support fora RepA_N-mediated DNA handcuffing model (Fig. 3E).

Structure and Function of the RepA CTD: Homology to the DnaD CTDand DnaG Recruitment. The combined data reveal the basis for spe-cific and cooperative origin binding by the RepA NTD. However,

Fig. 3. RepA NTD bridges two replication origins; insight into replicon as-sembly and direct visualization of handcuffing. (A) Top shows the design ofthe DNA used in AFM imaging for pTZ2162 and pSK41 studies. AFM images ofpTZ2162 RepA NTD in complex with DNA700. (i) A RepA NTD–DNA complexcomposed of two DNA strands and RepA NTD tetramers. Also shown isa strand of DNA with no protein bound. (B) AFM images of pSK41 RepANTD in complex with DNA700. (C) AFM images of pTZ2162 RepA NTD andRepA CTD incubated with DNA700. (D) Representative ITC binding isothermof pTZ2162 and pSK41 RepA NTD titrated into 23-bp DNA encoding a singleRep box. (E) Molecular model for RepA-mediated plasmid handcuffing.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1406065111 Schumacher et al.

Dow

nloa

ded

by g

uest

on

May

27,

202

0

the RepA CTD is also required for replication, yet its role(s)in this process are unknown. To assess whether the CTD inter-acts with the NTD and whether it may play a role in DNAbinding or organization of the Rep boxes, AFM was performedon samples containing separated domains of pTZ2162 RepA (theNTD and CTD) with DNA700. The resulting images were in-distinguishable from those obtained from the RepA NTD–DNA700samples (Fig. 3C, i–iii). These data are consistent with previousstudies demonstrating that the CTD binds neither DNA nor theNTD (30). Our FP studies also indicated that the CTD does notbind either DNA or the RepA NTD–DNA complex (Fig. S7). TheCTD sequence revealed no insight into its function. However,although not as generally conserved as the NTD, the RepA CTDshows strong sequence homology between RepA_N plasmidsin genus-specific clusters, suggesting that it may perform host-specific functions necessary for replication. To gain insight intopossible role(s) of the RepA CTD, the structure of the pTZ2162CTD was solved to 3.4 Å resolution. To obtain well-diffractingcrystals, a fusion protein was produced in which the pTZ2162CTD was attached to the C terminus of the maltose bindingprotein (MBP) (Table S2). The structure shows that the RepACTD folds into a compact structure composed of five helices.Helices α1 and α2 pack against a three-helix bundle formed byα3, α4, and α5 (Fig. 4A and Fig. S8). No CTD oligomers wereobserved in the structure, suggesting that this domain may bemonomeric. Indeed, SAXS studies, revealing an experimentallyderived Rg value of 16 Å, were consistent with the CTD monomerRg (17 Å) (Fig. 4B). Finally, the SAXS data were of sufficientquality to use in de novo envelope calculations and docking ofthe CTD into the envelope produced an excellent fit (Fig. 4B).Thus, the combined data support that the RepA CTD exists asa monomeric entity, flexibly tethered to the DNA-bound NTD.Database searches revealed that the RepA CTD shared the

strongest structural homology to the Enterococcus faecalis DnaDCTD (rmsd of 2.9 Å for overlay of 67 Cα atoms) (PDB ID code2I5U) (Fig. 4 C and D). This was a striking finding given that theRepA NTD was found to harbor structural similarity to theDnaD NTD. The CTD RepA and DnaD regions show strongerhomology than the corresponding NTDs, which, as noted, havedifferent functions. Whether the roles of the RepA and DnaDCTDs have also diverged is unclear. In Bacillus subtilis, DnaDplays a central role in replication origin assembly by mediatingrecruitment of essential replication proteins. To determinewhether the RepA CTD interacts with replication factors or anyproteins in its S. aureus host, yeast two-hybrid (Y2H) analyseswere carried out. First, Y2H showed that both full-length (FL)RepA and NTD self-interact, whereas the CTD does not, con-sistent with our structural and biochemical data. An FL pSK41RepA Y2H bait construct was then used to screen a genomic

S. aureus Y2H prey library to identify candidate CTD interactionpartners. Screening of ∼320,000 transformants (approximatelyfivefold library coverage) yielded only two strongly positiveinteracting prey constructs that did not self-activate. Strikingly,sequencing revealed that both contained a DNA fragment en-coding residues 81–313 of DnaG primase (Fig. 4E). However,to directly assay for a RepA CTD–DnaG interaction, FP ex-periments were carried out. These studies examining the abilityof DnaG to bind to a preformed RepA–DNA complex revealeda clear second binding event, indicating that DnaG binds toRepA–DNA (Fig. 4F).DnaG primase is a key component of the chromosomal DNA

replication machinery, synthesizing RNA primers for leading-strand DNA synthesis, and is an essential component of theprimosome in lagging-strand DNA synthesis. Moreover, DnaG isknown to interact with the replicative helicase DnaC; this in-teraction regulates the activities of both proteins and is critical toprimosome formation (23, 36). Y2H baits corresponding topSK41 RepA NTD (residues 1–120) and CTD (201–319) wereconstructed and tested against the DnaG prey. An interactionwas only detected with CTD, thereby corroborating the previousYH studies. Thus, our data indicate that staphylococcal RepA_Nproteins mediate replication initiation by anchoring RepA to theinitiation site via NTD–iteron interactions and recruiting thehost primase and hence helicase proteins via its CTD. Whereasthis is the first example to our knowledge of a plasmid RepAprotein loading the primase, previous work by the Bastia labo-ratory has shown that the monomeric plasmid RepA proteins canrecruit host replication proteins, including DnaA and DnaB,suggesting that plasmid RepA proteins have evolved to co-optthe host replication machinery (37, 38).

Conclusions: A “Minimalist” Replication Initiation Model for RepA_NReplicons. RepA_N initiator proteins are employed by nearly allcharacterized large staphylococcal multiresistance plasmids aswell as plasmids from Gram-positive bacteria including the en-terococcal pheromone-responsive plasmids and plasmids frombacilli, lactococci, and lactobacilli. However, how these ubiquitousRepA proteins function in replication has been unknown. Here wereveal the structures of the RepA protein domains as well as thestructure of a complex of RepA bound to an iteron DNA site.Remarkably, both RepA domains show structural similarity tothe corresponding domains of the S. aureus host primosomeprotein DnaD. This unexpected finding suggests the intriguingpossibility that multiresistance plasmids may have, through hori-zontal gene transfer, acquired the host DnaD gene and mod-ified it for plasmid-specific functions. Indeed, the RepA NTDfunction diverged significantly from DnaD to bind DNA ratherthan mediate self-assembly, as does the corresponding domain in

Fig. 4. Structure of pTZ2162 RepA CTD. (A) The MBP–RepACTD structure. The MBP is green, the maltose molecules areshown as sticks, and the RepA CTD is blue. (B) SAXS scat-tering plot of pTZ2162 CTD compared with the calculatedcurve based on the crystal structure. Also shown is the CTDcrystal structure docked into the de novo SAXS envelope.(C) RepA CTD structure with secondary structure elementslabeled. (D) Superimposition of the RepA CTD (cyan) ontothe DnaD CTD structure (red) revealing their structuralhomology. (E) Identification of the pSK41 RepA–DnaG in-teraction by Y2H analysis. Two prey clones (5A and 6A)strongly activated expression of the selectable marker andreporter genes only in the presence of the RepA bait con-struct. DNA sequencing revealed that they were independentclones expressing DnaG. All other prey clones (3A and 7A) self-activated in the presence of empty bait vector, hence repre-senting false positives. (F ) FP binding of pSK41 RepA to aniteron site until saturation, followed by addition of increasingconcentrations of DnaG (indicated by arrow). A saturablesecond binding event was evident.

Schumacher et al. PNAS Early Edition | 5 of 6

BIOCH

EMISTR

YCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

May

27,

202

0

DnaD. Our data show that the RepA CTD is similar to theDnaD CTD in that it harbors a primosome assembly role.However, distinct from DnaD, the RepA CTD binds DnaG,forming a plasmid-specific scaffold for helicase recruitment andprimosome assembly.In summary, our combined data indicate that RepA_N repli-

cons are minimalist replicons that require only the plasmidRepA protein to bind to the origin and facilitate replicon as-sembly. Binding of the RepA NTD to the origin recruits its at-tached CTD, which binds the host primosome DnaG, and hencethe replicative helicase (23). Thus, RepA_N proteins alone canassemble a functional replicon. A further striking finding fromthe RepA NTD–DNA structure is its bridging capability, whichpresents an optimal conformation for plasmid origin handcuffing.Given that RepA_N proteins are essential and conserved instaphylococcal multiresistance plasmids, this molecular dissectionof RepA_N proteins may enable the development of urgentlyneeded arsenals of new and highly specific chemotherapeuticsto combat the growing threat of multidrug-resistant bacteria.

MethodsThe genes encoding the pSK41 and pTZ2162 RepA proteins were purchasedfromGenscript Corporation and subcloned into pET15b. Proteins were purifiedvia multiple column chromatography steps. Biochemical, structural, and Y2Hstudies were performed as described in SI Methods.

ACKNOWLEDGMENTS. We thank Dr. Charles W. Pemble IV for assisting N.K.T.with data processing and the Nanoimaging Core Facility at the Universityof Nebraska Medical Center (UNMC). The facility is supported by NationalInstitutes of Health (NIH) Grant 1S10 RR023400-01 (Shared Instrument Grantprogram), the UNMC Program of Excellence, and the Nebraska ResearchInitiative. Small-angle X-ray scattering and X-ray crystallographic data werecollected at the Advance Light Source (ALS) and Advanced Photon Source(APS). ALS is a national user facility operated by Lawrence Berkeley NationalLaboratory on behalf of the Department of Energy (DOE), Office of BasicEnergy Sciences, through the Integrated Diffraction Analysis Technologiesprogram, supported by DOE Office of Biological and Environmental Re-search. Additional support comes from project MINOS (MacromolecularInsights on Nucleic Acids Optimized by Scattering) Grant R01GM105404.APS is supported by the US DOE, Office of Science, and the Office of BasicEnergy Sciences under Contract W-31-109-Eng-38. This work was also sup-ported by NIH Grant GM074815 (to M.A.S.) and National Health and MedicalResearch Council (Australia) Grant 571029 (to N.F. and S.M.K.).

1. Goetghebeur M, Landry PA, Han D, Vicente C (2007) Methicillin-resistant Staphylo-coccus aureus: A public health issue with economic consequences. Can J Infect DisMed Microbiol 18(1):27–34.

2. Cardo D, et al.; National Nosocomial Infections Surveillance System (2004) NationalNosocomial Infections Surveillance (NNIS) system report, data summary from January1992 through June 2004, issued October 2004. Am J Infect Control 32(8):470–485.

3. Rosenthal VD, et al.; INICC members (2012) International Nosocomial Infection Con-trol Consortium (INICC) report, data summary of 36 countries, for 2004-2009. Am JInfect Control 40(5):396–407.

4. Lacey RW (1975) Antibiotic resistance plasmids of Staphylococcus aureus and theirclinical importance. Bacteriol Rev 39(1):1–32.

5. Firth N, Skurray R (2006) Genetics: Accessory elements and genetic exchange. Gram-Positive Pathogens, eds Fischetti VA, Novick RP, Ferretti JJ, Portnoy DA, Rood JI (ASM,Washington), pp 413–426.

6. Novick RP (1989) Staphylococcal plasmids and their replication. Annu Rev Microbiol43:537–565.

7. del Solar G, Espinosa M (2000) Plasmid copy number control: An ever-growing story.Mol Microbiol 37(3):492–500.

8. Diep BA, et al. (2006) Complete genome sequence of USA300, an epidemic clone ofcommunity-acquired methicillin–resistant Staphylococcus aureus. Lancet 367(9512):731–739.

9. Weigel LM, et al. (2003) Genetic analysis of a high-level vancomycin-resistant isolateof Staphylococcus aureus. Science 302(5650):1569–1571.

10. Weaver KE, Kwong SM, Firth N, Francia MV (2009) The RepA_N replicons of Gram-positive bacteria: A family of broadly distributed but narrow host range plasmids.Plasmid 61(2):94–109.

11. Kuroda M, et al. (2001) Whole genome sequencing of methicillin-resistant Staphy-lococcus aureus. Lancet 357(9264):1225–1240.

12. Baba T, et al. (2002) Genome and virulence determinants of high virulence commu-nity-acquired MRSA. Lancet 359(9320):1819–1827.

13. Gill SR, et al. (2005) Insights on evolution of virulence and resistance from the com-plete genome analysis of an early methicillin-resistant Staphylococcus aureus strainand a biofilm-producing methicillin-resistant Staphylococcus epidermidis strain.J Bacteriol 187(7):2426–2438.

14. Holden MT, et al. (2004) Complete genomes of two clinical Staphylococcus aureusstrains: Evidence for the rapid evolution of virulence and drug resistance. Proc NatlAcad Sci USA 101(26):9786–9791.

15. Mwangi MM, et al. (2007) Tracking the in vivo evolution of multidrug resistance inStaphylococcus aureus by whole-genome sequencing. Proc Natl Acad Sci USA 104(22):9451–9456.

16. del Solar G, Giraldo R, Ruiz-Echevarría MJ, Espinosa M, Díaz-Orejas R (1998) Replica-tion and control of circular bacterial plasmids. Microbiol Mol Biol Rev 62(2):434–464.

17. Soultanas P (2012) Loading mechanisms of ring helicases at replication origins. MolMicrobiol 84(1):6–16.

18. Konieczny I (2003) Strategies for helicase recruitment and loading in bacteria. EMBORep 4(1):37–41.

19. Bruand C, Ehrlich SD, Jannière L (1995) Primosome assembly site in Bacillus subtilis.EMBO J 14(11):2642–2650.

20. Bruand C, Farache M, McGovern S, Ehrlich SD, Polard P (2001) DnaB, DnaD and DnaIproteins are components of the Bacillus subtilis replication restart primosome. MolMicrobiol 42(1):245–255.

21. Smits WK, Goranov AI, Grossman AD (2010) Ordered association of helicase loaderproteins with the Bacillus subtilis origin of replication in vivo. Mol Microbiol 75(2):452–461.

22. Ishigo-Oka D, Ogasawara N, Moriya S (2001) DnaD protein of Bacillus subtilis interactswith DnaA, the initiator protein of replication. J Bacteriol 183(6):2148–2150.

23. Corn JE, Berger JM (2006) Regulation of bacterial priming and daughter strand syn-thesis through helicase-primase interactions. Nucleic Acids Res 34(15):4082–4088.

24. Rymer RU, et al. (2012) Binding mechanism of metal·NTP substrates and stringent-response alarmones to bacterial DnaG-type primases. Structure 20(9):1478–1489.

25. Abhyankar MM, Reddy JM, Sharma R, Büllesbach E, Bastia D (2004) Biochemical in-vestigations of control of replication initiation of plasmid R6K. J Biol Chem 279(8):6711–6719.

26. Swan MK, Bastia D, Davies C (2006) Crystal structure of pi initiator protein-iteroncomplex of plasmid R6K: Implications for initiation of plasmid DNA replication. ProcNatl Acad Sci USA 103(49):18481–18486.

27. Giraldo R, Fernández-Tresguerres ME (2004) Twenty years of the pPS10 replicon: In-sights on the molecular mechanism for the activation of DNA replication in iteron-containing bacterial plasmids. Plasmid 52(2):69–83.

28. Kwong SM, Lim R, Lebard RJ, Skurray RA, Firth N (2008) Analysis of the pSK1 replicon,a prototype from the staphylococcal multiresistance plasmid family. Microbiology154(Pt 10):3084–3094.

29. Kwong SM, Skurray RA, Firth N (2004) Staphylococcus aureus multiresistance plasmidpSK41: Analysis of the replication region, initiator protein binding and antisense RNAregulation. Mol Microbiol 51(2):497–509.

30. Liu MA, Kwong SM, Pon CK, Skurray RA, Firth N (2012) Genetic requirements forreplication initiation of the staphylococcal multiresistance plasmid pSK41. Microbi-ology 158(Pt 6):1456–1467.

31. Nakaminami H, Noguchi N, Nishijima S, Kurokawa I, Sasatsu M (2008) Characteriza-tion of the pTZ2162 encoding multidrug efflux gene qacB from Staphylococcus au-reus. Plasmid 60(2):108–117.

32. Liu MA, Kwong SM, Jensen SO, Brzoska AJ, Firth N (2013) Biology of the staphylo-coccal conjugative multiresistance plasmid pSK41. Plasmid 70(1):42–51.

33. Schneider S, Zhang W, Soultanas P, Paoli M (2008) Structure of the N-terminal olig-omerization domain of DnaD reveals a unique tetramerization motif and providesinsights into scaffold formation. J Mol Biol 376(5):1237–1250.

34. Carrera P, Azorín F (1994) Structural characterization of intrinsically curved AT-richDNA sequences. Nucleic Acids Res 22(18):3671–3680.

35. Paulsson J, Chattoraj DK (2006) Origin inactivation in bacterial DNA replication con-trol. Mol Microbiol 61(1):9–15.

36. Koepsell SA, Larson MA, Griep MA, Hinrichs SH (2006) Staphylococcus aureus helicasebut not Escherichia coli helicase stimulates S. aureus primase activity and maintainsinitiation specificity. J Bacteriol 188(13):4673–4680.

37. Sharma R, Kachroo A, Bastia D (2001) Mechanistic aspects of DnaA-RepA interactionas revealed by yeast forward and reverse two-hybrid analysis. EMBO J 20(16):4577–4587.

38. Datta HJ, Khatri GS, Bastia D (1999) Mechanism of recruitment of DnaB helicase to thereplication origin of the plasmid pSC101. Proc Natl Acad Sci USA 96(1):73–78.

39. DeLano WL (2002) The PyMOL Molecular Graphics System (DeLano Scientific,San Carlos, CA).

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1406065111 Schumacher et al.

Dow

nloa

ded

by g

uest

on

May

27,

202

0