Structural determinant for switching betweenthe polymerase and exonuclease modes in thePCNA-replicative DNA polymerase complexHirokazu Nishidaa, Kouta Mayanagib, Shinichi Kiyonaric, Yuichi Satoa,1, Takuji Oyamad, Yoshizumi Ishinoc,and Kosuke Morikawae,f,2

aCentral Research Laboratory, Hitachi Ltd., 1-280 Higashi-koigakubo, Kokubunji, Tokyo 185-8601, Japan; bMedical Institute of Bioregulation, KyushuUniversity and Institute for Bioinformatics Research and Development, Japan Science and Technology Agency, 3-1-1 Maidashi, Higashi-ku, Fukuoka-shi,Fukuoka 812-8582, Japan; cDepartment of Genetic Resources Technology, Faculty of Agriculture, Kyushu University and Institute for Bioinformatics Researchand Development, Japan Science and Technology Agency, 6-10-1 Hakozaki, Fukuoka-shi, Fukuoka 812-8581, Japan; dInstitute for Protein Research, OsakaUniversity and Institute for Bioinformatics Research and Development, Japan Science and Technology Agency, Open Laboratories of Advanced Bioscienceand Biotechnology (OLABB), 6-2-3 Furuedai, Suita, Osaka 565-0874, Japan; eInstitute for Protein Research, Osaka University, 6-2-3 Furuedai, Suita,Osaka 565-0874, Japan; and fCore Research for Evolutional Science and Technology, 6-2-3 Furuedai, Suita, Osaka 565-0874, Japan

Edited by John Kuriyan, University of California, Berkeley, CA, and approved October 6, 2009 (received for review July 16, 2009)

Proliferating cell nuclear antigen (PCNA) is responsible for theprocessivity of DNA polymerase. We determined the crystal struc-ture of Pyrococcus furiosus DNA polymerase (PfuPol) complexedwith the cognate monomeric PCNA, which allowed us to constructa convincing model of the polymerase-PCNA ring interaction, withunprecedented configurations of the two molecules. Electron mi-croscopic analyses indicated that this complex structure exists insolution. Our structural study revealed that an interaction occursbetween a stretched loop of PCNA and the PfuPol Thumb domain,in addition to the authentic PCNA-polymerase recognition site (PIPbox). Comparisons of the present structure with the previouslyreported structures of polymerases complexed with DNA, sug-gested that the second interaction plays a crucial role in switchingbetween the polymerase and exonuclease modes, by inducing aPCNA-polymerase complex configuration that favors synthesisover editing. This putative mechanism for fidelity control of rep-licative DNA polymerases is supported by experiments, in whichmutations at the second interaction site caused enhancements inthe exonuclease activity in the presence of PCNA.

DNA clamp � DNA replication � electron microscopy � fidelity control �protein crystallography

DNA replication is a highly coordinated process, in whichDNA polymerase and the DNA sliding clamp play major

roles in ensuring accurate genome duplication. Most replicativeDNA polymerases consist of two independent functional com-ponents: polymerase and exonuclease moieties, which executeDNA synthesis and editing, respectively (1, 2). Since these twoactive sites are apart from each other, the DNA substrate cannotaccess both sites at the same time. These two reactions areperformed alternately, and thus require regulation by temporallydifferent switching, which was previously proposed from acomparison of the crystal structures of RB69 DNA polymerasescomplexed with dsDNA in the polymerase and exonucleasemodes (3, 4).

The DNA clamp, which forms a dimer or trimer, depending onthe domain of life, retains DNA polymerase on the templateDNA strand for processive DNA synthesis. DNA clamps, whichare called PCNA in Archaea and Eukarya, and the � subunit inBacteria, form similar architectures with a central hole. Thus,they encircle the substrate DNA, thereby enabling consecutiveDNA synthesis (5).

PCNA-interacting proteins contain a small conserved motif,the PIP (PCNA-interacting protein) box, which binds to acommon site on PCNA (6, 7). A model building study betweenthe RB69 DNA polymerase-editing complex and PCNA re-vealed the consistent configuration of each component (3). Thestructure of the complex between the ‘‘little finger’’ (LF) domain

of the translesional Y-family DNA polymerase Pol IV and the�-clamp from Escherichia coli showed that the bound polymer-ase domain lies at the outer rim of the clamp ring (8).

We reported previously that the PCNA from the hyperther-mophilic archaeon, Pyrococcus furiosus, bound to the PIP box ofthe family B DNA polymerase (hereafter PfuPol in this report)from the same organism and stimulated DNA strand synthesis invitro (9, 10). Here, we present the crystal structure of PfuPolcomplexed with a single subunit of the cognate PCNA(PfuPCNA). This complex structure exhibited a second inter-action site between the DNA polymerase and the clamp, inaddition to the PIP-binding site. Mutations at the secondinteraction site of PfuPol reduced the physical interaction withPfuPCNA and the PCNA-dependent DNA extension activity,whereas they enhanced the exonuclease activity in the presenceof PCNA. Collectively, our structural study, including an elec-tron microscopic analysis, suggests that the second interaction isinvolved in switching between the polymerase and exonucleasemodes (3, 4).

ResultsOverall Structure of the PfuPol/PCNA Monomer Complex. We deter-mined the 2.7Å resolution crystal structure of PfuPol in complexwith a monomeric subunit of PfuPCNA (Fig. 1A). For thestructural determination, we used a PCNA mutant, in which twoaspartic acid residues essential for ring formation were replacedby alanines (11). The architectures of PfuPol are essentially thesame between the complex with PCNA and the free state (20),although 47 residues in the thumb1 domain and 15 residues in theC terminus, including the PIP box, are structurally disordered inthe polymerase alone.

Remarkably, the superposition of one subunit of the PCNAhomotrimeric ring onto the PCNA monomer, in our complex,allowed us to construct a reasonable model, which exhibits nosteric hindrance between PfuPol and the other two PCNAsubunits (Fig. 1 A). In this complex structure, the PfuPol mole-

Author contributions: H.N. designed research; H.N., K. Mayanagi, S.K., and Y.S. performedresearch; H.N., K. Mayanagi, S.K., and T.O. analyzed data; and H.N., K. Mayanagi, Y.I., andK. Morikawa wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The atomic coordinates and structure factors have been deposited in theProtein Data Bank, www.pdf.org (PDB ID code 3A2F).

1Present address. Department of Biotechnology and Life Science, Tokyo University ofAgriculture and Technology, 2-24-16 Naka-cho, Koganei, Tokyo 184-8588, Japan.

2To whom correspondence should be addressed. E-mail: morikako@protein.osaka-u.ac.jp.

This article contains supporting information online at www.pnas.org/cgi/content/full/0907780106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0907780106 PNAS � December 8, 2009 � vol. 106 � no. 49 � 20693–20698

BIO

CHEM

ISTR

Y

cule is located in front of the ring plane, but does not cover thecentral hole. This configuration is distinct from the previousdocking models of the RB69 DNA polymerase/sliding clampcomplex (so-called active ‘‘tethered’’ states), where the DNApolymerase fully covers the central hole of the clamp in both theediting and polymerizing modes (3, 4). The present complex isalso different from the resting ‘‘locked-down’’ arrangementfound in the E.coli translesion DNA polymerase (pol IV)complexed with the sliding � clamp (10). The superpositionbetween PCNA in our complex and the corresponding domainsin the � clamp of the pol IV/� clamp complex indicated that polIV is rotated by �90° relative to PfuPol, whereas the C-terminalclamp-interacting peptide regions coincide with each other (Fig.1B). The positions of PfuPol and PCNA in our structure appearto correspond to an intermediate state between the activetethered state and the locked-down state (Fig. 1C), thus sug-gesting that our structure represents the ‘‘standby’’ state ofreplicative DNA polymerases. As described later, our electronmicroscopy data are consistent with the notion that this standbyconfiguration exists in solution.

The sequence alignment of the second interaction site ofPfuPol is noteworthy, in that two hydrophilic residues, Tyr-654and Arg-706, interact with Glu-171 at the particular turn pro-

truding from a � sheet in PCNA (Fig. 2A). This longest turn isreferred to as the ‘‘switch-hook’’ hereafter in this report, ac-cording to its functional role discussed later. The switch-hook,containing an acidic residue (Glu or Asp) at the same site, ishighly conserved among the DNA clamps from several speciesof archaea (i.e., PDB IDs: 1RWZ from Archaeoglobus fulgidus,2IX2, and 2IZO from Sulfolobus solfataricus), yeast (2OD8 fromSaccharomices cerevisiae), human (1VYM), and Escherichia coli(2POL).

Surface Plasmon Resonance Analyses. To examine the contributionof this site to the PfuPol/PCNA complex formation, we analyzedthe affinities of PfuPol Arg-706 mutants for PCNA. Accordingto the sequence alignment in the second interaction site, Arg-706is more conserved than Tyr-654. Except for the replacement ofArg-706 by lysine in the Thermococcus and Sulfolobus subdo-mains, the neighboring residues exhibited substantial conserva-tion (Fig. 2 A).

Surface plasmon resonance (SPR) analyses using the immo-bilized PfuPCNA trimeric ring revealed that the substitution ofglutamate for Arg-706 (R706E) reduced the binding ability ofPCNA, whereas the R706A and R706K mutants exhibitedmoderate reductions (wild-type, KD � 69.1 � 5.6 nM; R706E,

Fig. 1. Structure of the PfuPol/PCNA complex in comparison with othercomplex structures reported previously. (A) Superposition of our PfuPol(green)/PCNA monomer (orange) complex structures on the PfuPCNA trimericring (khaki), in which the surface representation (blue: polymerase domain,red: exonuclease domain) is overlaid with the PfuPol (23). (B) Stereo repre-sentation of the superimposed structures of the PfuPol/PCNA complex on theE. coli pol IV/� clamp complex. The little finger (LF) domain of pol IV and the� clamp are colored purple and pink, respectively. The PIP regions of pol IV LFand PfuPol (thick worm representations within the black oval) coincide witheach other, whereas pol IV LF and PfuPol adopt completely different config-urations with respect to the DNA clamp. (C) Schematic diagrams illustratingthe different clamp-binding angles of the polymerases in the three states.Hexagonal clamp rings (yellow) are shown together with dsDNA substrates(gray) running through the clamp rings. The models of the locked-down(purple) and standby (green) polymerases were generated from the crystalstructures (10), and that of the tethered polymerase (blue) was derived fromthe previous docking model (4).

Fig. 2. Second interaction site between PfuPol and PfuPCNA (A), and resultsfrom the surface plasmon resonance analyses of mutants (B and C). (A) Surfacerepresentations overlaid with ribbon diagrams of the PfuPol (green)/PCNA(orange) complex (Left), in which the surface regions corresponding to thetwo contact sites (PIP and the second interaction site) are colored dark green.The inset is a close-up view of the second interaction site in the complex.Alignment of the residues neighboring the two residues (Tyr-654, Arg-706)involved in the second interaction site of PfuPol and the correspondingsequences of seven other thermophilic archaeons from three subdomains(Right). The positions of PfuPol Arg-706 and the corresponding residues fromother species are highlighted in cyan. Abbreviations: Pfu, Pyrococcus furiosus;Pab, Pyrococcus abyssi; Pgl, Pyrococcus glycovorans; Tli, Thermococcus litora-lis; Tko, Thermococcus kodakaraensis; Tgo, Thermococcus gorgonarius; Sto,Sulfolobus tokodaii; Sac, Sulfolobus acidocaldarius. (B) Sensorgrams of thebinding of PfuPol and its mutants to PCNA immobilized on the chip. Associ-ation and dissociation phases were both obtained for 60 s each. The zero timefor all of the sensorgrams was set to the injection point. (C) Overlaid sensor-grams representing the interactions of PfuPol and its mutants with PCNAadded to the biotinylated primed-DNA immobilized on the SA chip. The lefthalf of the sensorgrams represents the association (60 s) and dissociation (90s) phases of PCNA onto the immobilized primed-DNA, and the right halfindicates the association (60 s) and dissociation (120 s) phases of thepolymerases.

20694 � www.pnas.org�cgi�doi�10.1073�pnas.0907780106 Nishida et al.

KD � 141 � 6.7 nM; R706A, KD � 112 � 7.1 nM; R706K, KD �88.9 � 4.2 nM)(Fig. 2B).

When PfuPCNA captured by the immobilized primed-DNAwas used as a ligand (Fig. 2C), the SPR examination of a seriesof Trp-769 mutants (W769A, W769A/R706K, and W769A/R706E), which suppress the PIP binding ability, was even moreintriguing. When the immobilized PCNA alone was used as aligand, W769A exhibited no binding ability (Fig. 2B). On theother hand, when the PCNA was captured by the immobilizedprimed-DNA, W769A significantly recovered binding ability(KD � 561 � 10.0 nM). Interestingly, the W769A/R706K doublemutant showed the same degree of binding as W769A, whereasW769A/R706E completely lost the binding ability (Fig. 2C). Thisimplies that the positive charge of Arg-706 is critical for PfuPol/PCNA complex formation, in the absence of the functionalPIP box.

Stimulation of Primer Extension Reaction by PCNA. To clarify theroles of the second interaction site during the cooperative actionof PfuPol and the clamp, we measured the stimulation effects ofPCNA on the primer extension reaction by the PfuPol wild-typeand R706 mutants, and found a good correlation between theextent of stimulation and the order of physical interaction (Figs.2C and 3A). For instance, the wild-type and R706K PfuPolsimilarly showed the highest stimulation effect by PCNA, al-though R706E and R706A, respectively, exhibited the lowest andmoderate stimulation effects among the mutants at Arg-706. Onthe other hand, in agreement with the data from SPR analyses,W769A exhibited much lower stimulation than the other mutantsat Arg-706. These results indicate that the binding ability at thesecond interaction site affects the DNA extension activity facil-itated by PCNA.

Exonuclease Activities Stimulated by PCNA. To determine whetherthe second interaction site stabilizes only the polymerase mode,we measured how PCNA enhances the exonuclease activity ofwild-type PfuPol and its mutants, and found an obvious differ-ence in the PCNA-enhanced exonuclease activity between thewild-type and the R706E mutant. These two enzymes exhibitedthe same degrees of exonuclease activities in the absence ofPCNA. On the other hand, the R706E mutant complexed withPCNA showed �2-fold higher exonuclease activity than thewild-type with PCNA at both 30 and 60 min (Fig. 3B). It appearsthat the dissociation of the second interaction site was favorablefor the exonuclease reaction in the PfuPol/PCNA complex.

Electron Microscopic Analysis. Our gel filtration and protein com-ponent analyses indicated that three polymerase molecules aresimultaneously loaded onto the single PCNA ring in the absenceof DNA, instead of forming a 1:1 complex, which consists of onetrimeric ring of PCNA and one bound PfuPol molecule. How-ever, the addition of various types of primed DNA allowed us toproduce the 1:1 complex. We analyzed negatively stained imagesof this complex by the electron microscopic single particlemethod.

Until now, the optimal conditions for visualizing the standbystate as a stable complex in the presence of the primed DNA hadnot been defined. Approximately 99% of the images of thecomplex showed the polymerase molecule fully covering thecentral hole of the PCNA ring, implying that the polymerase ismostly in the tethered state with a substrate DNA. However,�1% of the images displayed the polymerase standing on theedge of the PCNA ring, in good agreement with our crystalstructure. These images did not show the clear density of theprimed DNA, presumably because of smearing caused by themultiple positions of the primed DNA within the complex.Alternatively, this could result from the general tendency, whereDNA molecules in complexes with proteins are more poorly

visualized by negative staining. We selected 184 images of thecomplex, which is considered not to be in the tethered state, andthe 2-D class-averaged image (Fig. 4A) was calculated from theimages belonging to the major class (83 images).

To compare this averaged image with the structure of theproposed PfuPol/PCNA trimer complex, we calculated 2-Dprojection maps from this atomic model in various orientations.The projection map (Fig. 4C) calculated from the complex model

Fig. 3. Stimulatory effects of PfuPCNA on the primer extension (A) andexonuclease (B) activities of wild-type and mutant PfuPols. (A) The DNAstrands synthesized by wild-type and mutant PfuPols using substrates con-taining [�32P]dTTP were subjected to alkaline agarose gel electrophoresis, andthe products were visualized by autoradiography (Top). The sizes indicated onthe left were from BstI-digested � phage DNA, labeled with 32P at each 5� end.The bar chart (Bottom) represents the quantitated total count of the[�32P]dTTP incorporated by the wild-type and mutants. (B) Gel image (Top)and the corresponding bar chart (Bottom) in the comparison of the exonu-clease activities of the wild-type and R706E polymerases with and withoutPfuPCNA. Error bars in the chart represent � values for each peak profile. In theabsence of PCNA, the digested products were exactly the same length be-tween the wild-type and R706E polymerases at both 30 and 60 min. On theother hand, in the presence of PCNA, the exonuclease activities were en-hanced for both the wild-type and R706E, and the activity of R706E was abouttwice as high as that of the wild-type.

Nishida et al. PNAS � December 8, 2009 � vol. 106 � no. 49 � 20695

BIO

CHEM

ISTR

Y

with a certain orientation (Fig. 4B) was in good agreement withthe class averaged electron microscopic image (Fig. 4A). Wetherefore conclude that our model of the PfuPol/PCNA ringcomplex proposed in this study is in fact present in solution.

DiscussionThe switching mechanism between the polymerizing and editingmodes was previously discussed, on the basis of the two distinctcrystal structures of the RB69 DNA polymerase-DNA com-plexes (3, 4). The helical axes of the substrate DNA in each modewere differently oriented relative to the DNA polymerase, andhence the polymerase, tethered to PCNA (Fig. 1C, Right), has topivot upon switching from the polymerizing to editing mode (4).It is possible that this ‘‘pivot’’ motion is regulated by unknowninteractions between the polymerase and clamp, besides themain PIP contact.

In combination with mutational and electron microscopicdata, our crystal structure represents the standby state of thereplicative DNA polymerase. Notably, this state retains a strong,but flexible connection with a DNA clamp, so that PfuPol canundergo a rotating motion about the axis connecting the twointeraction sites to convert smoothly to the tethered state. Thetethered state lies on the path of this motion that brings the twoactive sites of PfuPol to the front of the PCNA ring plane.

Recent electron microscopic studies revealed that the orien-tations of DNA relative to the PCNA ring could be diverse,depending upon the proteins bound to the DNA clamps (25, 26).The crystal structure of the bacterial DNA clamp-DNA complexsuggested that the DNA in the PCNA may tilt up to �22°, inagreement with the molecular dynamics simulation study (12,24). We thus examined the reasonable polymerase-PCNA con-figurations deduced from our crystal structure, assuming that thedsDNA is bound to PfuPol in the exonuclease mode.

The extensions of dsDNA (red) bound to PfuPol rotated by 0°(Fig. 5A, Top) clashed with PCNA, and could not pass throughthe clamp hole without a sharp kink. In the case of the 60°rotation of PfuPol (Fig. 5A, Bottom), the dsDNA obliquelycrosses the center of the PCNA ring at a tilt angle of �45°, whichexceeds the limit described above. The substrate dsDNA in theexonuclease mode can never run through the central hole ofPCNA at any rotation angle of PfuPol, which simultaneouslyretains the two interactions.

On the other hand, when superimposing the dsDNA bound tothe RB69 DNA polymerase replicating complex on the PfuPolstructure with the two simultaneous interactions (4), the sub-strate dsDNA in the polymerase mode can pass through theclamp hole without any collision (Fig. 5B).

These relationships between PfuPol, PfuPCNA, and DNAwere supported by the experiments for the in vitro primerextension and exonuclease experiments. We found that theimpairment of the second interaction by the R706E mutation

benefits the exonuclease reaction in cooperation with PCNA(Fig. 3). Therefore, we examined how the second interactionbetween PfuPol and PfuPCNA is related to the previouslyproposed polymerase-exonuclease switching model (4), whichexhibited the tilt angle difference of�40° between the two DNApolymerase molecules in the polymerase and exonucleasemodes. This examination allowed us to propose that the secondinteraction plays a crucial role in the reaction-mode switching ofthe DNA polymerase, as illustrated in Fig. 6. This switching,involving the pivot motion, uses the PIP contact as a pivoting footand is executed in a different direction from the ‘‘standby-tethered’’ rotation, as depicted in Fig. S1. The detachment of thesecond interaction would generate sufficient mobility of thepolymerase on the PCNA ring, thereby allowing the DNA toaccess the exonuclease active site (Fig. 6A). This notion isconsistent with the previous model (4), and provides the struc-tural basis for the smooth switching between the polymerase andexonuclease modes (Fig. 6B).

It is intriguing to imagine that other PCNA binding proteinscontain similar extra sites interacting with PCNA. A notablesecond interaction site in the vicinity of the PIP-box was foundin the PCNA-Fen1 complex crystal structure (27). Such second-ary interactions may function as regulatory switches at varioussteps in DNA transactions.

Materials and MethodsData Collection, Structure Determination, and Refinement. Purification andcrystallization of the PfuPol/PCNA complex were described in refs. 13 and 14.Diffraction datasets, obtained at 100 K using cryo-cooled crystals, for themultiple anomalous dispersion (MAD) method based on selenomethioninederivatives, were collected by a CCD detector (ADSC) on beamlines BL38B1 andBL41XU at SPring-8. Data were processed and scaled using the HKL2000program suite (15). The MAD-phased initial electron density map at 3.3-Åresolution was obtained with SOLVE (16), and was improved with DM (17). Themodel was built with O (18), and was refined with CNS (19) at 2.7-Å resolution,using the best dataset. Statistics are presented in Table S1.

Surface Plasmon Resonance Analyses. The Biacore system was used for SPRanalyses. To monitor the binding of PfuPol and its mutants to immobilizedPCNA, highly purified PfuPCNA was fixed on a Sensor Chip CM5 (Biacore),according to the manufacturer’s recommendations.

To examine the interactions of PfuPol and its mutants with PCNA on theimmobilized primed-DNA, a 5�-biotinylated 70mer ssDNA (5�-TGCCGCCTC-CAAT TCTAATACGACTCACTATAGGGAGAAGGAAACTCCACCAACGATCTGA-CTACTGCCT-3�) was allowed to bind to a Sensor Chip SA (Biacore) coated withstreptavidin, and then the complementary 25mer ssDNA (5�-AGGTGGTTGCT-AGACTGATGA CGGA-3�), designed to hybridize with the 3�- end of the 70mertemplate, was applied to form the substrate primed-DNA. For each measu-rement of the binding of PfuPol to PCNA with primed-DNA, PfuPCNA wasadded to the immobilized primed-DNA, and PfuPol was applied to the chip.

To investigate the kinetic parameters, various concentrations of the puri-fied wt, R706A, R706K, and R706E proteins were applied to the sensor chips.All measurements were performed at a continuous flow rate of 30 �L/min, in

Fig. 4. Comparison of the crystal structure with an electron microscopic image. (A) Class average of electron microscopic images of the PfuPol/PCNA/DNAcomplex, corresponding to the side view of the not-tethered complex. The size of the image is 20.4 � 20.4 nm. (B) Ribbon diagram of the proposed PfuPol/PCNAtrimer ring model, viewed from the direction corresponding to A. PfuPol is colored green and the PfuPCNA trimer ring is colored yellow. (C) Two-dimensionalprojection map calculated from the proposed atomic model of the PfuPol/PCNA trimer complex. The size of the image is equivalent to that in A.

20696 � www.pnas.org�cgi�doi�10.1073�pnas.0907780106 Nishida et al.

a buffer containing 10 mM HEPES (pH 7.4), 150 mM NaCl, and 0.005% Tween20. At the end of each cycle, the bound protein was removed by washing thechip with 4M MgCl2. The association and dissociation phase data were simul-taneously fitted using a data analysis program, BIAevaluation 3.2 (Biacore). Allof the measurements for the wild-type and mutants were performed fivetimes to calculate the standard deviations for the kinetic parameters.

In Vitro Primer Extension Reaction. The in vitro primer extension abilities ofpolymerases in the absence and presence of PfuPCNA were compared by using

M13mp18 single-stranded DNA annealed with an oligonucleotide as a sub-strate. The reaction mixture, containing 20 mM Tris�HCl (pH 8.8), 100 mM NaCl,2 mM MgSO4, 10 mM (NH4)2SO4, 50 mM KCl, 0.1% Triton X-100, 200 �Mdeoxyribonucleoside triphosphates (containing 5 �Ci of [�32P]dTTP), and 5 nMof template-primer DNA (20 �L), was preheated at 70 °C for 1 min, and thereaction was then started by adding DNA polymerases to 10 nM. After an

Fig. 5. Modeling of dsDNA with the PfuPol/PCNA complex in the exonuclease(A) and polymerase (B) modes. (A) Substrate DNA in the exonuclease mode. Thesurfaces of the two interacting sites (encircled by green ovals) are represented indark green. The Top view of the complex corresponds to the present crystalstructure without rotation about the axis (magenta) connecting the two inter-acting sites, while the Bottom exhibits the view after a 60° rotation. The Leftexhibit views along the axis. When the exonuclease domains of the RB69 DNApolymerase and PfuPol are superimposed on each other, the substrate dsDNA(red ribbons), bound to RB69 DNA polymerase in the exonuclease mode, isextended (transparent red cylinder), causing possible collisions (enclosed in redcircles) with the PCNA ring. (B) Substrate DNA in the polymerase mode. Theribbon diagram of dsDNA (blue), docked to PfuPol (green surface) in the poly-merizing mode and the possible extension of the substrate dsDNA are depictedby a cylinder (transparent blue) for clarity. Note that a 55° rotation of PfuPolallows the bound substrate DNA to pass through the central hole of the clamp,while maintaining both interactions.

Fig. 6. Schematic diagrams illustrating that the disconnection of the secondinteraction site is requisite for the exonuclease mode. (A) The disengagementof the second interaction can eliminate the steric hindrance between thedsDNA in the exonuclease mode and the clamp. In the ternary complex wherePfuPol is rotated by 60° about the axis connecting the two interacting sites, theputative dsDNA would seriously clash with the inside of the PCNA hole, asdepicted in Fig. 5A. However, the loss of the second interaction site allowedPfuPol to pivot freely, thereby enabling the bound dsDNA to pass through thecentral hole without any steric violation. (B) Schematic diagrams representingthe switching between the polymerase and exonuclease modes. The substrateDNA is depicted as gray ribbons. PCNA and bound PfuPol are depicted as thickstick models (yellow, PCNA; green, PfuPol), with the polymerase active sitemoiety in PfuPol colored dark blue and the exonuclease active site colored red.The PIP region in PfuPol is also indicated as a thick stick branching from PfuPol,and is labeled ‘‘PIP.’’ The second interaction site is located at the end of thehorseshoe-shaped PfuPol, and the PIP nearly protrudes from this end. The stickmodels of PfuPCNA and PfuPol were constructed from the actual coordinatesin the crystal structure. When the second interaction is maintained, thepolymerase active site (dark blue) of PfuPol interacts with the substrate DNAthrough PCNA (left model). By contrast, the exonuclease active site cannotinteract with DNA, without uncoupling the second site interaction. Thus,together with PIP, the second site is used for stabilizing the polymerase mode,but not the exonuclease mode. The disruption of this site allows the conver-sion to the exonuclease mode (right model).

Nishida et al. PNAS � December 8, 2009 � vol. 106 � no. 49 � 20697

BIO

CHEM

ISTR

Y

incubation at 70 °C for 2 min, the reaction was stopped by adding a solutioncontaining 98% deionized formamide, 1 mM EDTA, 0.1% xylene cyanol, and0.1% bromophenol blue. For the reactions with PfuPCNA and the cognate RFC(replication factor C), these factors were added to the reaction mixture at anequal molar ratio to the DNA polymerases. The reaction products wereanalyzed by electrophoresis on an alkaline agarose gel (1%) containing 50 mMNaOH.

In Vitro Exonuclease Activity. The in vitro exonuclease activities of the wild-type and mutant PfuPols in the absence and presence of PfuPCNA weremeasured by using the pET-21a vector (Novagen, 5443-bp) digested with theBamH-I (Takara) restriction enzyme. The reaction mixture contained 20 mMTris�HCl (pH 8.8), 100 mM NaCl, 2 mM MgSO4, 15 mM (NH4)2SO4, 5%vol/volglycerol, 0.5 nM BamH-I-treated pET-21a (20 �L), and 1.5 nM polymerase.After an incubation at 60 °C for either 30 or 60 min, the reaction was stoppedby adding a solution containing 95% deionized formamide and 20 mM EDTA.For the reactions with PfuPCNA, PfuPCNA was added to the reaction mixtureat an equal molar ratio to the polymerase. The reaction products wereanalyzed by electrophoresis on an agarose gel (0.8%). After staining withSYBR Green I (Molecular Probes), the gel was analyzed with a Fluor Imager 595(Molecular Dynamics).

Electron Microscopic Analyses. PfuPol and PfuPCNA were mixed with theprimed-DNA in a buffer containing 50 mM Tris�HCl (pH 8) and 5 mM MgCl2. Themixture was repurified by chromatography on a Superdex 200 gel filtration

column. The fraction corresponding to the reconstructed complex was dilutedwith a buffer containing 50 mM Tris�HCl (pH 8), 50 mM NaCl, and 5 mM MgCl2.A 3-�L aliquot of the sample solution was applied to a carbon-coated grid andstained with 2% uranyl acetate. The specimens were examined with a JEM1010 electron microscope (JEOL) operated at an accelerating voltage of 100kV. Images were obtained with a Bioscan CCD camera (GATAN). The step sizeof a pixel of the image was calibrated to be 5.1 Å, using TMV as a referencesample. A minimum dose system (MDS) was used to reduce the electronradiation damage. Image processing was performed using the software pack-ages EMAN (21) and IMAGIC (22). Particle images corresponding to thenot-tethered complex were isolated, using the program BOXER in EMAN.

ACKNOWLEDGMENTS. We are grateful to Drs. K. Hasegawa and H. Sakai atbeamline BL38B1, and Dr. M. Kawamoto at beamline BL41XU for their help inX-ray data collection at SPring-8, and Drs. D. Tsuchiya, M. Tanabe, and K. Torifor helpful discussions. This work was partly supported by a research grantendorsed by the New Energy and Industrial Technology Development Orga-nization. Y. I. was supported by a grant from the Human Frontier ScienceProgram and a grant-in-aid from the Ministry of Education, Culture, Sports,Science, and Technology of Japan, H. N. was partly supported by a grant fromthe Genome Network Project of the Ministry of Education, Culture, Sports,Science, and Technology of Japan. K. Mayanagi, S. K., T. O., and Y. I. weresupported by the Institute for Bioinformatics Research and Development,Japan Science and Technology Agency. K. Morikawa and T. O. were supportedby the Core Research for Evolutional Science and Technology, Japan Scienceand Technology Agency.

1. Kunkel TA (2004) DNA replication fidelity. J Biol Chem 279:16895–16898.2. Patel PH, Loeb LA (2001) Getting a grip on how DNA polymerases function. Nat Struct

Biol 8:656–659.3. Shamoo Y, Steitz TA (1999) Building a replisome from interacting pieces: Sliding clamp

complexed to a peptide from DNA polymerase and a polymerase editing complex. Cell99:155–166.

4. Franklin MC, Wang J, Steitz TA (2001) Structure of the replicating complex of a polalpha family DNA polymerase. Cell 105:657–667.

5. Indiani C, O’Donnell M (2006) The replication clamp-loading machine at work in thethree domains of life. Nat Rev Mol Cell Biol 7:751–761.

6. Warbrick EM (2000) The puzzle of PCNA’s many partners. BioEssays 22:997–1006.7. Moldovan GL, Pfander B, Jentsch S (2006) PCNA, the maestro of the replication fork.

Cell 129:665–679.8. Bunting KA, Roe SM, Pearl LH (2003) Structural basis for recruitment of translesion DNA

polymerase Pol IV/DinB to the �-clamp. EMBO J 22:5883–5892.9. Matsumiya S, Ishino S, Ishino Y, Morikawa K (2002) Physical interaction between

proliferating cell nuclear antigen and replication factor C from. Pyrococcus furiosusGenes Cells 7:911–922.

10. Tori K, Kimizu M, Ishino S, Ishino Y (2007) DNA polymerases BI and D from thehyperthermophilic archaeon Pyrococcus furiosus both bind to proliferating cell nuclearantigen with their C-terminal PIP-box motifs. J Bacteriol 189:5652–5657.

11. Matsumiya S, Ishino S, Ishino Y, Morikawa K (2003) Intermolecular ion pairs maintainthe toroidal structure of Pyrococcus furiosus. Prot Sci 12:823–831.

12. Georgescu RE, et al. (2008) Structure of a sliding clamp on DNA. Cell 132:43–54.13. Matsumiya S, Ishino Y, Morikawa K (2001) Crystal structure of an archaeal DNA sliding

clamp: Proliferating cell nuclear antigen from Pyrococcus furiosus. Prot Sci 10:17–23.14. Nishida H, Matsumiya S, Tsuchiya D, Ishino Y, Morikawa K (2006) Stoichiometric

complex formation by proliferating cell nuclear antigen (PCNA) and its interactingprotein: Purification and crystallization of the DNA polymerase and PCNA monomermutant complex from Pyrococcus furiosus. Acta Crystallogr F 62:253–256.

15. Otwinowski Z, Minor W (1997) Processing of X-ray diffraction data collected in oscil-lation mode. Methods Enzymol 276:307–319.

16. Terwilliger TC, Berendzen J (1999) Automated MAD and MIR structure solution. ActaCrystallogr D 55:849–861.

17. CCP4 (1994) Collaborative Computational Project Number 4. The CCP4 suite: Programsfor protein crystallography. Acta Crystallogr D 50:760–763.

18. Jones TA, Zou ZY, Cowan SW, Kjeldgaard M (1991) Improved methods for buildingprotein models in electron density maps and the location of errors in these models.Acta Crystallogr A 47:110–119.

19. Brunger AT, et al. (1997) Crystallography and NMR system: A new software suite formacromolecular structure determination. Acta Crystallogr D 54:905–921.

20. Kim SW, Kim DU, Kim JK, Kang LW, Cho HS (2008) Crystal structure of Pfu, the highfidelity DNA polymerase from Pyrococcus furiosus. Int J Biol Macromol 42:356–361.

21. Ludtke SJ, Baldwin PR, Chiu W (1999) EMAN: Semiautomated software for high-resolution single-particle reconstructions. J Struct Biol 128:82–97.

22. van Heel M, Harauz G, Orlova EV, Schmidt R, Schatz M (1996) A new generation of theIMAGIC image processing system. J Struct Biol 116:17–24.

23. Pettersen EF, et al. (2004) UCSF Chimera—A visualization system for exploratoryresearch and analysis. J Comput Chem 25:1605–1612.

24. Ivanov I, Chapados BR, McCammon JA, Tainer JA (2006) Proliferating cell nuclearantigen loaded onto double-stranded DNA: Dynamics, minor groove interactions, andfunctional implications. Nucl Acid Res 34:6023–6033.

25. Mayanagi K, et al. (2009) Mechanism of replication machinery assembly as revealed bythe DNA ligase-PCNA-DNA complex architecture. Proc Natl Acad Sci USA 106:4647–4652.

26. Miyata T, et al. (2005) Open clamp structure in the clamp-loading complex visualizedby electron microscopic image analysis. Proc Natl Acad Sci USA 102:13795–13800.

27. Sakurai S, et al. (2005) Structural basis for recruitment of human flap endonuclease 1to PCNA. EMBO J 24:683–693.

20698 � www.pnas.org�cgi�doi�10.1073�pnas.0907780106 Nishida et al.