ARTICLES

Nucleotide-dependent conformational changes in the DnaA-like core of the origin recognition complex

Megan G Clarey1, Jan P Erzberger1, Patricia Grob1, Andres E Leschziner2, James M Berger1, Eva Nogales1–3 & Michael Botchan1

Structural details of initiator for DNA replication have provided clues to the molecular events in this process. EM reconstructions of the Drosophila melanogaster origin recognition complex (ORC) reveal nucleotide-dependent conformational changes in the core of the complex. All five AAA+ domains in ORC contain a conserved structural element that, in DnaA, http://www.nature.com/nsmb promotes formation of a right-handed helix, indicating that helical AAA+ substructures may be a feature of all initiators. A DnaA helical pentamer can be docked into ORC, and the location of Orc5 uniquely positions this core. The results suggest that ATP- dependent conformational changes observed in ORC derive from reorientation of the AAA+ domains. By analogy to the DNA- wrapping activity of DnaA, we posit that ORC together with prepares origin DNA for loading through mechanisms related to the established pathway of prokaryotes.

ORC is a conserved, multiprotein complex used to initiate DNA DNA replication, melting of the template provides a portal for replication in all eukaryotes characterized to date1–3. Binding of this loading the helicase. The precise molecular architecture of the eukar- six-subunit assembly to DNA origins marks the first step in DNA yotic helicase has not been resolved completely, but recent data replication and is followed by ORC-directed recruitment of the pre- show that a complex containing the six MCM proteins (MCM2– replication complex (pre-RC) and eventual establishment of bidirect- MCM7), Cdc45 and the four GINS proteins performs this 3 16–18

Nature Publishing Group Group 200 6 Nature Publishing ional DNA replication forks . unwinding activity . In this complex, the ATP-dependent

© Despite differences in the recruitment and regulation of replisomal functions of the helicase are provided by the MCM proteins. These factors, numerous structural and functional similarities indicate that MCM proteins associate with the template before helicase there are important mechanistic parallels between prokaryotic and activation and before the acquisition of Cdc45 and GINS, and the eukaryotic DNA replication initiation3–6. For example, all initiator tight binding of the MCMs to the pre-RC requires, at minimum, proteins function as either homomeric or heteromeric complexes of the ORC and Cdc6 proteins19,20. To date, ORC and Cdc6 have AAA+ subunits that are activated through ATP binding7,8.Flankingthe not been found to melt or otherwise deform DNA in preparation AAA+ core, initiators have C-terminal DNA-targeting modules that for MCM loading, as would be anticipated by a strict analogy to mediate origin interactions. In bacteria, DNA targeting precedes ATP- DnaA. Consequently, a widely held speculation has been that the dependent oligomerization of the initiator molecule, DnaA, whereas in AAA+ ensemble of ORC plus Cdc6 functions as the ‘loader’ of a eukaryotes, ATP modulates origin interactions in the context of the ring-like helicase, much as the five-subunit AAA+ replication preformed ORC assembly9–12. Notably, recent studies of human ORC factor C (RFC) brings the PCNA sliding clamp to encircle the have shown that, in addition to stimulating DNA-binding activity, ATP DNA-primer template at the replication fork. The details of binding also has an essential structural role in the formation and how ORC and Cdc6 work to stably engage the MCM proteins stability of the complex12. This finding parallels the assembly mechan- have not yet been revealed. We wished to elucidate the structure ism of DnaA8,13.Additionally,bothDnaAandORChaveanincreased of ORC and its relationships to the functional form of the affinity for negatively supercoiled DNA3,4,14,15, indicating that super- ATP-DnaA filament (described in the accompanying work by coiling may be generally important for the initial engagement of Erzberger et al.21) and to the RFC structure. To that end, we origins. These observations suggest that bacterial initiators may provide have derived an EM reconstruction of the Drosophila melanogaster useful mechanistic insights into eukaryotic initiation, and vice versa. complex and tested the possible docking of these high- In all organisms and viruses, a key step in the pathway to initiation resolution X-ray models into the volume of the eukaryotic of DNA synthesis is the loading of a DNA helicase. In prokaryotic ORC structure.

1Division of Biochemistry & Molecular Biology, Molecular & Cell Biology Department, 1 Barker Hall, University of California, Berkeley, California 94720, USA. 2Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, California 94720, USA. 3Howard Hughes Medical Institute, Molecular and Cell Biology Department, LSA 355 #3200, University of California, Berkeley, California 94720-3200, USA. Correspondence should be addressed to M.B. ([email protected]) or E.N. ([email protected]). Received 14 April; accepted 21 June; published online 9 July 2006; doi:10.1038/nsmb1121

684 VOLUME 13 NUMBER 8 AUGUST 2006 NATURE STRUCTURAL & MOLECULAR BIOLOGY ARTICLES

resolution of 34 A˚ , according to the 0.5 Fourier-shell correlation (FSC) abApo-ORC criterion (Fig. 1 and Supplementary Fig. 2 online) and shows an even coverage of angular space, without any apparent missing cone (Supplementary Fig. 2). From both the two-dimensional (Fig. 1b) and three-dimensional data (Fig. 1c), we observed that apo-ORC forms an elongated structure, with maximal dimensions of 170 A˚ 115 A˚ .Thethree- dimensional structure reveals a large major domain with distinct secondary features (Supplementary Video 1 online). The major domain forms the core of the complex and has a spiral-crescent shape that encompasses a B25-A˚ -wide channel. Density protruding off the top of this toroidal core results in an S-shaped molecule. This c 90° 90° 90° secondary density appears as a right-handed ‘boxing glove,’ where the ‘thumb’ contributes to the top of the channel and the ‘fingers’ curl back toward the major domain. Peripheral to the channel, at the base Channel 170 Å * of the complex, is a that exits through a small circular ‘collar’ Apo resting on the back of the core domain (Fig. 1c). Collar To visualize nucleotide-dependent changes in the ORC structure, we obtained images of the ATPg-S–bound form of ORC (Fig. 2a). Figure 1 EM analysis of Drosophila melanogaster ORC. (a) Electron Reference-free class averages (Fig. 2b) confirmed the expected simi- micrograph of negatively stained apo-ORC (scale bar ¼ 50 nm). (b) Examples of reference-free two-dimensional class averages. (c)34-A˚ larity between complexes with and without added nucleotide. We http://www.nature.com/nsmb resolution reconstruction of Drosophila ORC rendered as a gray isosurface. then produced a three-dimensional reconstruction of ATPg-S–bound 23 Each view represents a 901 counter-clockwise turn about the y-axis. Dashed ORC by iterative reference projection matching ,usingthe circle marks the collar. apocomplex as the initial reference (Fig. 2, Supplementary Fig. 2 and Supplementary Video 1). To avoid reference bias during projection matching, the initial apo-ORC reference volume was RESULTS filtered to 60 A˚ (Supplementary Fig. 3 online). The final ATP-ORC Nucleotide-dependent conformational changes in ORC structure, refined to 32-A˚ resolution, shows that the addition of In preparation for a structural analysis of Drosophila ORC, the nucleotide induces a conformational change that results in a tighten- recombinant was purified to homogeneity (Supplementary ing of the core spiral element (Fig. 2d, see below). In addition, changes Fig. 1 online). Drosophila ORC has emerged as a stable metazoan are observed in the collar, which opens in comparison to the apo ORC complex; unlike human ORC, all six subunits remain assembled structure (Fig. 2c,d). Concomitant with this change, thickening is even in the absence of nucleotide, making it a more ideal complex observed in the mass that connects the uppermost subdomain to the 12 Nature Publishing Group Group 200 6 Nature Publishing for structural studies . Electron micrographs of the negatively core. To verify that the differences between the two reconstructions

© stained complex reveal that it is stable and does not aggregate on are substantial compared to the internal variability within each data the sample grid (Figs. 1 and 2). Tilt-pair data were collected from the set, three-dimensional variance maps were calculated for each recon- apocomplex for an ab initio reconstruction. The untilted particle data struction24,25. For both the apo and the ATP reconstruction, regions set was subjected to two-dimensional reference-free alignment and of high variance are very limited and localize mostly to the classification (Fig. 1b), and the tilted data was used to generate class stain-retaining cavities in the complex (Supplementary Fig. 4 online). volumes by the random conical tilt (RCT) method22. After merging A recent EM reconstruction of ATP-bound Saccharomyces cerevisiae together five class volumes, we used the resulting structure ORC26 also shows an elongated complex of similar dimensions, with a for projection-matching refinement. The final reconstruction has a ring domain resembling the collar in the Drosophila ORC reconstruc- tion. However, our structure is quite different in overall appearance from the yeast ORC a ATP-ORC b c 90° 90° 90° complex. In particular, it is difficult to trace a helical path on the yeast structure, as it appears more planar than the Drosophila 170 Å * ORC reconstruction. Although the overall ATP conservation in the amino acid sequences of the subunits suggests a substantial degree of ° d 180 structural homology, the Drosophila complex may be more stable than the yeast counter- Apo part, which may suffer from substoichio- ATP metricamountsofsomeofthecomponents or have a greater degree of conformational flexibility. Structural differences between the Figure 2 EM analysis of Drosophila melanogaster ATP-ORC, showing nucleotide-dependent confor- two models could, to some extent, be mational changes. (a) Electron micrograph of negatively stained ATPg-S–ORC (scale bar ¼ 50 nm). explained by possible differences in the angu- (b) Examples of reference-free two-dimensional class averages. (c) ATP-ORC reconstruction at 33-A˚ resolution rendered as a red isosurface. Asterisk marks nucleotide-dependent opening in the collar. lar coverage of the structures (presence or (d) Front and back stereo views of superimposed apo-ORC (gray surface) and ATP-ORC (red wire mesh) absence of a missing cone) or, more probably, reconstructions (Supplementary Video 1). by different levels of preservation of the

NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 13 NUMBER 8 AUGUST 2006 685 ARTICLES

Walker A α2 β1 α3 β2 PaCdc5 ScOrc1 DmOrc1 ScOrc4 DmOrc4 ScOrc5 DmOrc5 ScOrc2 DmOrc2 ScOrc3 DmOrc3

Helical insert α3 α4 PaCdc5 ScOrc1 DmOrc1 ScOrc4 DmOrc4 ScOrc5 DmOrc5 ScOrc2 DmOrc2 ScOrc3 DmOrc3

Walker B Sensor I Box VII β3 α5 α6 β4 α7 α8 β5 PaCdc5 ScOrc1 DmOrc1 ScOrc4 DmOrc4 ScOrc5 DmOrc5 ScOrc2 DmOrc2 ScOrc3 http://www.nature.com/nsmb DmOrc3

Figure 3 Secondary structure predictions of Orc1–Orc5 show the presence of a helical insert. Saccharomyces cerevisiae (Sc) and Drosophila melanogaster (Dm) AAA+ ORC components were aligned to one another and to Pyrobaculum aerophilum (Pa) Cdc6/Orc1 using information from the secondary structure prediction program PSIPRED48 and the threading programs 3D-PSSM49 and PHYRE (http://www.sbg.bio.ic.ac.uk/phyre/) and manually adjusted using JALVIEW50. Secondary structure elements of PaCdc6/Orc1 are shown above the alignment. Blue, a-helix; yellow, b-strand; red, helical insert.

samples during the staining procedures (different levels of ‘flattening’). that this feature has functional relevance in the remodeling of However, we do expect to find some bona fide structural differences in replication origins. the two complexes, and biochemical and biological differences between them should not be disregarded. For example, the yeast Pentameric AAA+ filament docks in the ORC core

Nature Publishing Group Group 200 6 Nature Publishing and Drosophila Orc1 counterparts function in silencing by interacting To test the predictions from structural conservation described above,

© with different molecular partners27, and Orc6 in metazoans has a we compared the architecture of the ATP-DnaA filament and Droso- unique role in cytokinesis not found in budding yeast28,29. phila ORC. Docking of a helical ATP-DnaA pentamer into the ORC envelope reveals a good fit, supporting the idea that a right-handed Initiator-specific helix drives filament formation The accompanying 3.5-A˚ X-ray crystal structure of ATP-bound DnaA reveals that the bacterial initiator forms a right-handed helical filament upon oligomerization21. This quaternary packing arrange- abWalker A/B ment is unique among AAA+ proteins and is principally mediated by Sensor I

the presence of an extra a-helix at one edge of the AAA+ domain, α3 which acts as a ‘steric wedge’ between subunits to drive ATP-DnaA α4 α3 into a spiral filament. This a-helical insert marks a defining feature of α4 30 Helical insert the initiator clade of AAA+ ATPases and is absent in other AAA+ 180˚ 90° assemblies, such as the eukaryotic clamp loader RFC31 and the 32 bacterial transcription factor NtrC . Box VII

As DnaA is the closest structural homolog of both the archaeal α3 initiator Cdc6/Orc16,33 and the eukaryotic initiator proteins Orc1– α3 Orc5, sequence alignments were used to investigate whether this helix α4 α4 is present in the AAA+ domains of Orc1–Orc5. Secondary structure predictions and structural threading calculations reveal that this Figure 4 A conserved structural element common to DnaA and Orc1–Orc5 additional helix is found in all five AAA+ ORC components suggests a common oligomerization interface. (a) Secondary structure (Fig. 3). A superposition of the archaeal Cdc6/Orc1 structure on an arrangement and surface conservation profile of the archaeal Cdc6/Orc1 ATP-DnaA dimer shows that the ATP-DnaA packing orientation AAA+ domain (PDB entry 1FNN) reveals areas of high sequence provides excellent steric complementarity for the archaeal initiator, conservation (green). The helical insert (a3-a4; red) is present in the five ORC AAA+ subunits (Fig. 3). (b) Stereo images of a least-squares specifically accommodating interactions of the initiator-specific helix superposition of two Cdc6/Orc1 monomers on an ATP-DnaA dimer and matching the surface conservation pattern of archaeal Cdc6/Orc1 (r.m.s. deviation ¼ 1.3 A˚ among core Ca atoms) show extensive (Fig. 4). These observations support the idea that initiators have a complementarity between 3-A˚ envelopes (blue and gold) around interface Ca unique oligomerization state not found in other AAA+ complexes and atoms. Surface interactions match regions of highest surface conservation.

686 VOLUME 13 NUMBER 8 AUGUST 2006 NATURE STRUCTURAL & MOLECULAR BIOLOGY ARTICLES

Figure 5 ATP-DnaA pentamer as a model for the ORC core. (a) Top and side a bc view of AAA+ domains in the ATP-DnaA helical filament, with five protomers alternately colored red and orange. (b) Manual docking of a pentameric ATP- DnaA AAA+ subassembly into apo-ORC (Supplementary Video 2). (c)Ball- and-stick representation of the helical path in the ORC core (ORC-EM; cyan) and the centers of mass of the DnaA pentamer (red) and RFC superposed on the second DnaA monomer (green). The EM envelope is shown as a surface 90° 90° 90° DnaA ORC-EM for the full volume and as a mesh for core domains, thresholded to 420 kDa RFC and 114 kDa respectively. (d) Front (left) and side (right) views of apo- and ATP-ORC core masses. Arrows denote the trajectory of the ORC core upon ATP binding. (e) Model of ORC with five AAA+ domains in green and red, and four winged-helix domains (WHD; corresponding to the collar) in yellow.

docking model, we performed subunit mapping studies with a demonoclonal antibody to the Orc5 subunit. The Orc5 subunit is an ideal candidate for testing this prediction, as it is composed almost entirely of the AAA+ domain. Our hypothesis demands that this AAA+ subunit be found in a distinct position within the mass of the toroidal Apo core. We incubated ORC with various concentrations of the mono- ATP WHD clonal antibody and defined conditions for isolation of single ORC– antibody complexes. These complexes were separated from unbound ORC and free antibody by glycerol-gradient sedimentation. Images of http://www.nature.com/nsmb helical AAA+ oligomer forms the core of the eukaryotic initiation negatively stained antibody-bound ORC were subjected to two- complex (Fig. 5 and Supplementary Video 2 online). We then dimensional alignment and classification (Fig. 6). The EM images increased the threshold of the EM envelope to highlight the core confirm that ORC binds a single antigen-binding domain of the Orc5 features of apo-ORC. The distinct spiral organization of this envelope antibody. The resulting class averages were then aligned to reprojec- permitted the placement of five evenly spaced markers into the core of tions of the ORC complex alone (Fig. 6a). The additional Y-shaped the envelope. Indeed, the curvature, dimensions and center of mass of density seen in the two-dimensional ORC-plus-antibody class averages the DnaA pentamer globally matches the core density within apo-ORC, can be assigned to the Orc5 monoclonal antibody. The general whereas the pitch and diameter of the split-ring RFC assembly are location of the Orc5 monoclonal antibody maps to the toroidal core drastically different and do not dock into the ORC volumes (Fig. 5c). of the ORC complex where the AAA+ pentamer docks, supporting Furthermore, the DnaA model containing five AAA+ domains corre- our hypothesis that the spiral element of ORC is composed of an sponds to 42% of the molecular weight of apo-ORC and occupies AAA+ backbone (Fig. 6b). B40% of the ORC volume, as determined by protein volume calcula- There is a clear preferential orientation of the antibody-labeled 34 Nature Publishing Group Group 200 6 Nature Publishing tions . This good match between the actual volume occupied by the ORC particles about ORC’s elongated axis on the sample grid. This

© DnaA helix and the calculated protein masses helps to validate the preferred orientation is most probably due to the large additional mass docking model. The small differences in pitch between the ATP-DnaA contributed by the full-length antibody and its binding position on filament and the core of ORC are probably attributable to slight ORC. Although the antibody has a certain degree of conformational structural differences in their initiator-specific helices and modest flexibility around its antigen-binding site, the Orc5 subunit is clearly repositions of AAA+ subunit interfaces (Figs. 3 and 4). On the basis of the ATP-DnaA docking model, we propose that Orc1–Orc5 form the core of the complex, assembling into a right-handed helical oligomer a through head-to-tail interactions between the AAA+ domains. ORC alone

Orc5 is located in the toroidal core of ORC 1 2 3 4 The 34-A˚ resolution of our ORC reconstruction is sufficient to attempt docking of the high-resolution models of RFC and DnaA, ORC + Orc5 to compare the overall architecture of these assemblies. Having antibody observed that DnaA model can be docked with great complementarity 1 2 34 to the prokaryotic DnaA filament, we endeavored to establish that this docking was valid by a direct structural approach. To test the unique Orc5 antibody localization

Figure 6 Orc5 is located in the toroidal core of ORC. (a) Different views of the three-dimensional apo-ORC reconstruction (top row) aligned to two- b dimensional class averages of negatively stained ORC plus a monoclonal antibody raised against the Orc5 subunit (middle row). The general location 180° of the Orc5 antibody mass is represented as a mesh cone in the corresponding three-dimensional structure of each view (bottom row). (b) ‘Front’ and ‘reverse’ views of the ORC reconstruction, showing where the antibody to Orc5 labels the particle. The spiral region of ORC, which closely matches the pitch and dimensions of five AAA+ domains extracted from the DnaA filament, contains Orc5, a known AAA+ subunit.

NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 13 NUMBER 8 AUGUST 2006 687 ARTICLES

situated in the top ‘shoulder’ of the semicircular core of ORC, nucleotide-dependent conformational changes along the AAA+ precisely along the right-handed helical axis of the complex, where helix19. Just as has been suggested for RFC, structural changes the AAA+ pentamer docks. mediated by nucleotide binding at one end of a hetero-oligomeric complex could be propagated to neighboring subunits31,42.Orc1isthe DISCUSSION active ATPase in the AAA+ core, and a position at the base of the Our data support the hypothesis that the five AAA+ domains of ORC helical domain would allow it to bind other factors that need to are organized in a manner structurally homologous to the right- interact with ORC for DNA replication. handed ATP-DnaA filament. Perhaps the most significant implication A functional consequence of the open helical packing arrangement is of such results is that this initiator-specific superstructure may confer that AAA+ subunit surfaces are left exposed at either end of the an intrinsic DNA-remodeling activity that can act to destabilize assembly. This configuration means that terminal AAA+ domains are origins during replication initiation. The data, and the speculation available to interact with and be regulated by other AAA+ proteins that follows from them, challenge prevalent models19,20,26,35–37 for during initiation. The location of the AAA+ modules in ORC proposed ORC based on RFC clamp-loader structures. The ORC AAA+ here would provide an unoccupied AAA+ surface at the base of the domains cannot accommodate the tight closed spiral of an RFC-like complex for interaction with Cdc6. Cdc6 binding at this location may arrangement, probably owing to the presence of the additional close off the central channel, as seen in the yeast Cdc6-ORC structure26. initiator-specific helix in Orc1–Orc5. It is possible that a distinct The studies presented here show that a conserved AAA+ helical and perhaps unique loader mechanism for ORC plus Cdc6 is to be substructure underlies initiator architecture in the three domains of uncovered, despite the structural homology to DnaA. However, for life. This finding in turn suggests that there are likely to be strong reasons of parsimony, we favor the hypothesis of conservation of mechanistic commonalities in the ways that initiators engage and function with the prokaryotic initiator. remodel replication origins, as well as in how they facilitate Previous studies have linked ORC to RFC by establishing the assembly. In bacteria, ATP promotes the cooperative assembly of http://www.nature.com/nsmb analogy that ORC loads MCMs as RFC loads PCNA. However, the DnaA monomers into a right-handed helical filament that wraps available biochemistry on ORC equally accommodates aspects of DNA; with negatively supercoiled DNA, this wrapping is predicted to DnaA-type activity. As is observed for ATP-DnaA during replication promote the melting of DNA unwinding elements in the origin to initiation, ORC would be bound to replication origins in an ATP- maintain a constant topological linking number. Although archaeal bound state. In a mechanism consistent with ATP-dependent loading and eukaryotic initiators have not yet been observed to carry out this of the prokaryotic replicative helicase DnaB by the prokaryotic heli- precise function, modulation of local DNA architecture is likely to be a case loader DnaC, Cdc6 bound to ATP would allow for MCM crucial property of pre-RC formation. Our structural data show that interaction with the pre-RC8,19,38. Just as ATP hydrolysis by DnaC is although ORC exists as a preformed assembly, its AAA+ modules and required for loading of the DnaB helicase onto the DNA template, WHDs can flex in response to nucleotide status, a conformational ATP hydrolysis by Cdc6 is required for MCM loading. If these switch that probably underlies changes in DNA affinity known to arise prereplication parallels between eukaryotes and prokaryotes are mean- from ATP binding3,4. This structural transition may also regulate the ingful, one might ask why DNA-remodeling activity of the pre-RC accessibility of exposed AAA+ surfaces, allowing ORC, for example, to

Nature Publishing Group Group 200 6 Nature Publishing factors ORC and Cdc6 has not been observed. Perhaps other factors or interact with Cdc6 through direct AAA+ interactions, much as the

© in vitro conditions need to be investigated. To date, DNA-binding re-replication inhibitor Hda binds DnaA in E. coli43. Hda-DnaA studies with ORC and Cdc6 have been performed with linear DNA. In binding inhibits the progression of the replication process in E. coli; the context of our discussion, it is germane to point out that negative in contrast, Cdc6 activates the pathway in eukaryotes in an unknown supercoiling is required for substrate melting and that HU and other way. Further structural studies on how these ancillary factors affect the minor groove–binding proteins that stabilize DNA bends are impor- path and structure of the DNA with the initiators should provide a key tant cofactors for DnaA-induced melting39–41. to the mechanisms of activation and repression. The DnaA pentamer fits best into the apo-ORC structure, suggest- ing that there can be conformational flexion of this region, owing METHODS either to structural differences hardwired into the different ORC Protein purification. The recombinant Drosophila ORC was purified as subunits or to allosteric effects arising from ATP binding. Consistent described previously7. An additional glycerol-gradient purification step was with this idea, inspection of the high-density regions of the ORC core performed to obtain homogenous complex (Supplementary Fig. 1). reveals that nucleotide-dependent conformational changes are propa- Electron microscopy and image reconstruction. Purified ORC was used to gated along the predicted AAA+ skeleton, causing tightening and prepare negatively stained EM grids. Approximately 150 ng of ORC was applied elongation of the superhelical axis (Fig. 5d). The changes along the to the surface of glow-discharged holey carbon grids covered with continuous AAA+ helical axis also modulate the opening of the collar, which we carbon. The samples were then stained with 3% (w/v) uranyl acetate. RCT was propose contains the winged-helix domains (WHDs) located on the used to generate an initial structure22. A tilted-pair data set was taken at a C-terminal side of the AAA+ core in at least four of the ORC AAA+ magnification of 49,000 on a Tecnai electron microscope at 120 keV with a subunits (Fig. 5e). These assignments are based upon the volume of tilt angle of 501 and a defocus range of –1.5 to –2 mm. Reference-free two- these domains in the structure and their connectivity with the core dimensional alignment of the untilted particles was performed in IMAGIC44, helical domain. A subset of the WHDs is known to mediate DNA then the alignment parameters were transferred to SPIDER, where the tilted contacts, but in the context of ORC, it is likely that some interactions data was used to generate class volumes by RCT. After progressively merging together five class volumes, we used the resulting structure for projection also serve to stabilize the complex. The collar may therefore serve not matching refinement45 of the full untilted data set. The final reconstruction only as a DNA-binding region, but also as an oligomerization element includes B4,500 particles and has a resolution of 34 A˚ as estimated by the FSC (Figs. 2 and 5). cutoff at 0.5 (ref. 46). The three-dimensional volume was thresholded to the We suggest that Orc1, the largest ORC subunit at 103 kDa, is calculated value for a molecular weight of 420 kDa. For the ATP-ORC situated at the base of the complex, so that nucleotide binding between structure, ATPg-S was added to glycerol gradient–purified ORC to a final the active site of Orc1 and the arginine finger of Orc4 effects the concentration of 5 mM and only untilted images were collected. Reference-free

688 VOLUME 13 NUMBER 8 AUGUST 2006 NATURE STRUCTURAL & MOLECULAR BIOLOGY ARTICLES

two-dimensional alignment of the untilted particles was performed in 2. Gossen, M., Pak, D.T., Hansen, S.K., Acharya, J.K. & Botchan, M.R. A Drosophila IMAGIC44. A three-dimensional reconstruction was generated from B8,000 homolog of the yeast origin recognition complex. Science 270, 1674–1677 particles by iterative reference projection matching in SPIDER using the apo- (1995). ˚ 3. Bell, S.P. The origin recognition complex: from simple origins to complex functions. ORC structure filtered to 60 A as the initial reference. The FSC criterion Dev. 16, 659–672 (2002). estimate of the final resolution was 33 A˚ . This three-dimensional volume was 4. Remus, D., Beall, E.L. & Botchan, M.R. DNA topology, not DNA sequence, is a critical also thresholded to a molecular weight of 420 kDa. Figures were generated determinant for Drosophila ORC-DNA binding. EMBO J. 23, 897–907 (2004). using PyMOL (http://pymol.sourceforge.net). Docking of the X-ray crystal 5. Vashee, S. et al. Sequence-independent DNA binding and replication initiation by the human origin recognition complex. Genes Dev. 17, 1894–1908 (2003). structure of the ATP-DnaA model was done manually in PyMOL. DnaA 6. Erzberger, J.P., Pirruccello, M.M. & Berger, J.M. The structure of bacterial DnaA: protein volume calculations were done in VOIDOO34. To highlight the core implications for general mechanisms underlying DNA replication initiation. EMBO J. mass of the ORC complex, the density threshold was calculated to include an 21, 4763–4773 (2002). approximate molecular mass of 114 kDa. 7. Chesnokov, I., Gossen, M., Remus, D. & Botchan, M. Assembly of functionally active Drosophila origin recognition complex from recombinant proteins. Genes Dev. 13, 1289–1296 (1999). Three-dimensional variance calculation. Three-dimensional variance maps 8. Bramhill, D. & Kornberg, A. Duplex opening by dnaA protein at novel sequences in were calculated as described24,25. Briefly, 500 bootstrap versions of both datasets initiation of replication at the origin of the E. coli . Cell 52, 743–755 were picked randomly and used to calculate as many three-dimensional (1988). 9. Chesnokov, I., Remus, D. & Botchan, M. Functional analysis of mutant and wild-type reconstructions by backprojection using the Euler angles previously determined Drosophila origin recognition complex. Proc. Natl. Acad. Sci. USA 98, 11997–12002 2 by projection matching. The three-dimensional variance estimate sB between (2001). these B ¼ 500 partial reconstructions was computed. The same procedure is 10. Lee, D.G. & Bell, S.P. ATPase switches controlling DNA replication initiation. Curr. used to calculate the bootstrap estimate of the background noise using ‘noise Opin. Cell Biol. 12, 280–285 (2000). 11. Klemm, R.D. & Bell, S.P. ATP bound to the origin recognition complex is important for particles’ extracted from the area surrounding the data particles. The real-space preRC formation. Proc. Natl. Acad. Sci. USA 98, 8361–8367 (2001). three-dimensional variance can then be calculated from the bootstrap estimates 12. Ranjan, A. & Gossen, M. A structural role for ATP in the formation and stability of the as follows: human origin recognition complex. Proc. Natl. Acad. Sci. USA 103, 4864–4869 (2006). 2 2 2 13. Bramhill, D. & Kornberg, A. A model for initiation at origins of DNA replication. Cell http://www.nature.com/nsmb s ¼ Kðs s Þ struct B back 54, 915–918 (1988). where K is the total number of particles in each dataset and s2 is the average 14. Lee, J.K., Moon, K.Y., Jiang, Y. & Hurwitz, J. The Schizosaccharomyces pombe origin back recognition complex interacts with multiple AT-rich regions of the replication origin of the background noise variance. The resulting three-dimensional variance DNA by means of the AT-hook domains of the spOrc4 protein. Proc. Natl. Acad. Sci. map was superposed to the corresponding three-dimensional reconstruction USA 98, 13589–13594 (2001). for both states of the ORC complex in Chimera47 (supported by US National 15. Fuller, R.S. & Kornberg, A. Purified dnaA protein in initiation of replication at the Institutes of Health grant P41 RR-01081). Escherichia coli chromosomal . Proc. Natl. Acad. Sci. USA 80, 5817–5821 (1983). 16. Gambus, A. et al. GINS maintains association of Cdc45 with MCM in replisome Subunit mapping. Protein A–purified Orc5 monoclonal antibody was added to progression complexes at eukaryotic DNA replication forks. Nat. Cell Biol. 8, 358–366 Mono S–purified ORC and subjected to glycerol-gradient sedimentation. The (2006). peak fraction containing the mono-bound ORC–antibody complex was used to 17. Pacek, M., Tutter, A.V., Kubota, Y., Takisawa, H. & Walter, J.C. Localization of prepare negatively stained grids as described above. Untilted images were MCM2–7, Cdc45, and GINS to the site of DNA unwinding during eukaryotic DNA replication. Mol. Cell 21, 581–587 (2006). collected as described above. Two-dimensional alignment and classification of 18. Moyer, S.E., Lewis, P.W. & Botchan, M.R. Isolation of the Cdc45/Mcm2-7/GINS (CMG) B3,000 antibody-labeled particles was performed in IMAGIC. complex, a candidate for the eukaryotic DNA replication fork helicase. Proc. Natl. Acad. Sci. USA, published online 23 June 2006 (doi:10.1073/pnas.0602400103). Nature Publishing Group Group 200 6 Nature Publishing Accession codes. EM Data Bank: Three-dimensional EM maps of apo-ORC 19. Bowers, J.L., Randell, J.C., Chen, S. & Bell, S.P. ATP hydrolysis by ORC catalyzes © g reiterative Mcm2–7 assembly at a defined origin of replication. Mol. Cell 16, 967–978 and ATP -S–ORC have been deposited with accession codes EMD-4751 and (2004). EMD-4820, respectively. 20. Randell, J.C., Bowers, J.L., Rodriguez, H.K. & Bell, S.P. Sequential ATP hydrolysis by Cdc6 and ORC directs loading of the Mcm2–7 helicase. Mol. Cell 21, 29–39 (2006). Note: Supplementary information is available on the Nature Structure & Molecular 21. Erzberger, J.P., Mott, M.L. & Berger, J.M. Structural basis for ATP-dependent DnaA Biology website. assembly and replication-origin remodeling. Nat. Struct. Mol. Biol. advance online publication 9 July 2006 (doi:10.1038/nsmb1115). ACKNOWLEDGMENTS 22. Radermacher, M. Three-dimensional reconstruction of single particles from random and nonrandom tilt series. J. Electron Microsc. Tech. 9, 359–394 (1988). We thank M. Gossen at the Max Delbruck Institute for numerous gifts of the 23. Penczek, P.A., Grassucci, R.A. & Frank, J. The ribosome at improved resolution: new Drosophila Orc5 monoclonal antibody, D. Remus for helpful discussions and the techniques for merging and orientation refinement in 3D cryo-electron microscopy of Office of Biological and Environmental Research of the US Department of Energy biological particles. Ultramicroscopy 53, 251–270 (1994). for support of E.N. E.N. is a Howard Hughes Medical Institute Investigator. 24. Grob, P. et al. Cryo-electron microscopy studies of human TFIID: conformational This work was supported by US National Institutes of Health grant R39 breathing in the integration of regulatory cues. Structure 14, 511–520 CA 30490 to M.B. (2006). 25. Penczek, P.A., Yang, C., Frank, J. & Spahn, C.M. Estimation of variance in single- particle reconstruction using the bootstrap technique. J. Struct. Biol. 154, 168–183 AUTHOR CONTRIBUTIONS (2006). M.G.C. contributed ORC sample preparation and EM analysis; J.P.E. contributed 26. Speck, C., Chen, Z., Li, H. & Stillman, B. ATPase-dependent cooperative binding of DnaA docking and sequence alignments; P.G. contributed three-dimensional ORC and Cdc6 to origin DNA. Nat. Struct. Mol. Biol. 12, 965–971 (2005). variance analysis; A.E.L. contributed computational assistance; J.M.B. contributed 27. Pak, D.T. et al. Association of the origin recognition complex with heterochromatin and ORC cartoon model, docking supervision and DnaA mass/volume calculations; HP1 in higher eukaryotes. Cell 91, 311–323 (1997). E.N. supervised EM analysis and contributed to the manuscript; M.B. supervised 28. Chesnokov, I.N., Chesnokova, O.N. & Botchan, M. A cytokinetic function of Drosophila ORC6 protein resides in a domain distinct from its replication activity. Proc. Natl. biochemistry and contributed to the manuscript. Acad. Sci. USA 100, 9150–9155 (2003). 29. Prasanth, S.G., Prasanth, K.V. & Stillman, B. Orc6 involved in DNA replication, COMPETING INTERESTS STATEMENT chromosome segregation, and cytokinesis. Science 297, 1026–1031 (2002). The authors declare that they have no competing financial interests. 30. Iyer, L.M., Leipe, D.D., Koonin, E.V. & Aravind, L. Evolutionary history and higher order classification of AAA+ ATPases. J. Struct. Biol. 146, 11–31 (2004). Published online at http://nature.com/nsmb/ 31. Bowman, G.D., O’Donnell, M. & Kuriyan, J. Structural analysis of a eukaryotic sliding Reprints and permissions information is available online at http://npg.nature.com/ DNA clamp-clamp loader complex. Nature 429, 724–730 (2004). reprintsandpermissions/ 32. Lee, S.Y. et al. Regulation of the transcriptional activator NtrC1: structural studies of the regulatory and AAA+ ATPase domains. Genes Dev. 17, 2552–2563 (2003). 33. Capaldi, S.A. & Berger, J.M. Biochemical characterization of Cdc6/Orc1 binding to the 1. Bell, S.P. & Stillman, B. ATP-dependent recognition of eukaryotic origins of DNA replication origin of the euryarchaeon Methanothermobacter thermoautotrophicus. replication by a multiprotein complex. Nature 357, 128–134 (1992). Nucleic Acids Res. 32, 4821–4832 (2004).

NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 13 NUMBER 8 AUGUST 2006 689 ARTICLES

34. Kleywegt, G.J. & Jones, T.A. Detection, delineation, measurement and display of 42. Neuwald, A.F. Evolutionary clues to eukaryotic DNA clamp-loading mechanisms: cavities in macromolecular structures. Acta Crystallogr. D Biol. Crystallogr. 50,178– analysis of the functional constraints imposed on AAA+ ATPases. 185 (1994). Nucleic Acids Res. 33, 3614–3628 (2005). 35. Davey, M.J., Jeruzalmi, D., Kuriyan, J. & O’Donnell, M. Motors and switches: 43. Su’etsugu, M., Shimuta, T.R., Ishida, T., Kawakami, H. & Katayama, T. Protein AAA+ machines within the replisome. Nat. Rev. Mol. Cell Biol. 3, 826–835 (2002). associations in DnaA-ATP hydrolysis mediated by the Hda-replicase clamp complex. 36. Mendez, J. & Stillman, B. Perpetuating the double helix: molecular machines at J. Biol. Chem. 280, 6528–6536 (2005). eukaryotic DNA replication origins. Bioessays 25, 1158–1167 (2003). 44. van Heel, M., Harauz, G., Orlova, E.V., Schmidt, R. & Schatz, M. A new generation of 37. Perkins, G. & Diffley, J.F. Nucleotide-dependent prereplicative complex assembly by the IMAGIC image processing system. J. Struct. Biol. 116, 17–24 (1996). Cdc6p, a homolog of eukaryotic and prokaryotic clamp-loaders. Mol. Cell 2,23–32 45. Frank, J. et al. SPIDER and WEB: processing and visualization of images in 3D (1998). electron microscopy and related fields. J. Struct. Biol. 116, 190–199 (1996). 38. Davey, M.J., Fang, L., McInerney, P., Georgescu, R.E. & O’Donnell, M. The DnaC 46. Saxton, W.O. & Baumeister, W. The correlation averaging of a regularly arranged helicase loader is a dual ATP/ADP switch protein. EMBO J. 21, 3148–3159 (2002). bacterial cell envelope protein. J. Microsc. 127, 127–138 (1982). 39. Leonard, A.C. & Grimwade, J.E. Building a bacterial orisome: emergence of new 47. Pettersen, E.F. et al. UCSF Chimera–a visualization system for exploratory research and regulatory features for replication origin unwinding. Mol. Microbiol. 55, 978–985 analysis. J. Comput. Chem. 25, 1605–1612 (2004). (2005). 48. McGuffin, L.J., Bryson, K. & Jones, D.T. The PSIPRED protein structure prediction 40. Baker, T.A. & Kornberg, A. Transcriptional activation of initiation of replication from the server. Bioinformatics 16, 404–405 (2000). E. coli chromosomal origin: an RNA-DNA hybrid near oriC. Cell 55, 113–123 (1988). 49. Kelley, L.A., MacCallum, R.M. & Sternberg, M.J. Enhanced genome annotation using 41. Skarstad, K., Baker, T.A. & Kornberg, A. Strand separation required for initiation of structural profiles in the program 3D-PSSM. J. Mol. Biol. 299, 499–520 (2000). replication at the chromosomal origin of E. coli is facilitated by a distant RNA–DNA 50. Clamp, M., Cuff, J., Searle, S.M. & Barton, G.J. The Jalview Java alignment editor. hybrid. EMBO J. 9, 2341–2348 (1990). Bioinformatics 20, 426–427 (2004). http://www.nature.com/nsmb Nature Publishing Group Group 200 6 Nature Publishing ©

690 VOLUME 13 NUMBER 8 AUGUST 2006 NATURE STRUCTURAL & MOLECULAR BIOLOGY