Structural mechanism of helicase loading onto replication origin DNA by ORC-Cdc6

Significance The loading of the core Mcm2-7 helicase onto origin DNA is essential for the formation of replication forks and genomic stability. Here, we report two cryo-electron microscopy (cryo-EM) structures that capture helicase loader–helicase complexes just prior to DNA insertion. These pre-loading structures, combined with a computational simulation of the dynamic transition from the pre-loading state to the loaded state, provide crucial insights into the mechanism required for topologically linking the helicase to DNA. The helicase loading system is highly conserved from yeast to human, which means that the molecular principles described here for the yeast system are likely applicable to the human system.

D NA replication is central for cellular life and genomic stability. Most macromolecular processes are controlled at the level of initiation (1). In eukaryotes, DNA replication is separated into two distinctive steps occurring in different cell cycle phases (2). During G1 phase of the cell cycle, the helicase core consisting of the minichromosome-maintenance 2-7 (Mcm2-7) proteins becomes loaded onto DNA, a process termed prereplicative complex (pre-RC) formation. Then, in S phase, the Mcm2-7 hexamer becomes activated, leading to origin firing and DNA synthesis. Helicase loading at DNA replication origins depends on the Origin Recognition Complex (ORC), Cdc6, and Cdt1 (3)(4)(5)(6)(7). Yeast ORC binds the replication origin DNA and is associated with DNA through the cell cycle (8,9). During late M-phase, Cdc6 binds to ORC, transforming it into an active complex, which is now competent to load, with the help of Cdt1, two Mcm2-7 hexamers into Mcm2-7 double hexamer that encircles DNA (6,7). Prior to helicase loading, the Mcm2-7 single hexamer adopts an open spiral configuration (10)(11)(12). However, within the ORC-Cdc6-Cdt1-Mcm2-7 (OCCM) pre-RC intermediate, which forms in the absence of ATP hydrolysis, DNA is already inserted into Mcm2-7 and the helicase adopts a ringshaped conformation, with a small gap at the Mcm2-Mcm5 interface (13)(14)(15)(16)(17). Based on a recent reanalysis of the wild-type (WT) cryo-EM dataset (18), we were able to separate a minor population of OCCM in which the DNA is fully inserted and the Mcm2-Mcm5 gate is fully closed at both N-and C-tiers. To facilitate the description of the structural transition in this report, we refer to these two WT OCCM conformers as "OCCM" (gate partially closed) and "gate-closed OCCM" (gate fully closed), respectively.
Orc1-5 and Cdc6 belong to the AAA+ family of ATPases (25,26), while Orc6 shares homology with the transcription factor II B (TFIIB) (27,28). Orc1, Orc4, Orc5, and Cdc6 participate in ATP binding and are composed of an AAA+ domain followed by a winged-helix domain (WHD), while Orc2 and Orc3 do not bind to ATP and are structurally more diverged (29). In particular,

Significance
The loading of the core Mcm2-7 helicase onto origin DNA is essential for the formation of replication forks and genomic stability. Here, we report two cryo-electron microscopy (cryo-EM) structures that capture helicase loader-helicase complexes just prior to DNA insertion. These pre-loading structures, combined with a computational simulation of the dynamic transition from the pre-loading state to the loaded state, provide crucial insights into the mechanism required for topologically linking the helicase to DNA. The helicase loading system is highly conserved from yeast to human, which means that the molecular principles described here for the yeast system are likely applicable to the human system.
Orc3 is characterized by a large α-helical insertion located between its AAA+ domain and the WHD, giving rise to the typical asymmetric shape of the ORC complex (30,31). More recent structural work revealed that ORC adopts a two-tiered hexameric structure, with one narrow tier made up by five Orc1-5 WHDs and one wider tier composed of the five AAA+ domains (18,(32)(33)(34). Orc6 is attached to the side of the ORC complex via Orc3, but also has major interactions with Orc2 and Orc5 (18,33,35). When ORC is DNA-bound, DNA passes through its central channel.
On the narrow C-terminal side of the complex, Orc3, Orc5, and Orc6 bend the DNA by 60°utilizing patches of positively charged amino acids (33).
Biochemical and genetic work has shown that the Mcm3 C-terminal WHD is essential to promote initial Cdt1-Mcm2-7 interactions with ORC-Cdc6 (15). Indeed, point mutants in the Mcm3 WHD have been shown to abrogate complex formation. In addition, the Mcm3 WHD is sufficient to interact with ORC-Cdc6, and so it was proposed that Mcm3 could initiate Mcm2-7 loading on DNA (15). Moreover, it was shown that a Cdt1-Mcm6 interaction remodels the Mcm2-7 complex, relieving an autoinhibitory activity of the C-terminal WHD of Mcm6. In budding yeast, this remodeling is critical for the initial ORC-Cdc6 interaction with Cdt1-Mcm2-7 (14). In metazoans, where Cdt1 and Cdc6 bind directly to ORC (36,37), an ORC-Cdc6-Cdt1 complex may regulate the recruitment of Mcm2-7 via a Cdt1-Mcm6 interaction.
To study the initial phase in ORC-Cdc6-Cdt1-Mcm2-7 complex formation, it is necessary to slow down the loading process. Here we employed a previously reported Mcm6 mutant, which is missing the C-terminal WHD (amino acids 839 to 1017) and is referred to as McmΔC6 or "mutant Mcm2-7" (14). This mutant Mcm2-7 supports the first step in helicase recruitment, the interaction between Cdt1-Mcm2-7 and ORC-Cdc6, but fails to induce pre-RC ATP hydrolysis for unknown reasons and does not allow double hexamer formation (14). Thus, pre-RC formation is blocked at a stage prior to full OCCM complex formation. Using cryo-EM, we have now captured two OCCM intermediates, "semi-attached OCCM" and "pre-insertion OCCM," which reveal the step-wise recruitment of budding yeast Cdt1-Mcm2-7 to ORC-Cdc6. Furthermore, to understand the dynamic DNA insertion process, we performed molecular dynamics simulations, revealing the energy landscape and the large conformational changes accompanying origin DNA insertion into Mcm2-7. Our work explains why ORC alone in the absence of Cdc6 is incapable of recruiting Cdt1-bound Mcm2-7, thereby demonstrating how ORC-Cdc6 guides and positions the DNA near the DNA entry gate in Mcm2-7 prior to helicase loading and how DNA enters into the Mcm2-7, providing mechanistic insights into a key stage of the highly conserved eukaryotic helicase-loading process.

Results and Discussion
Cryo-EM Captures Two OCCM Intermediates Using the Mcm6-ΔC6 Mutant. To examine the structural rearrangements occurring during helicase loading in budding yeast, we employed an Mcm2-7 complex containing McmΔC6 and the ATP analog ATPγS, which both slow down the pre-RC reaction cascade ( Fig. 1 A) (14). Analysis of 2D class averages showed one set of classes where the density of the helicase loader ORC-Cdc6 is well resolved while Cdt1-Mcm2-7 is clearly present but is blurred ( Fig. 1 B, Upper). This suggests the capture of a very early intermediate in which Mcm2-7 is connected to ORC-Cdc6 yet the two complexes have not docked onto each other. We termed this intermediate semi-attached OCCM. Further analysis of the 2D class averages revealed a second intermediate in which both ORC-Cdc6 and Cdt1-Mcm2-7 densities were well resolved, indicating that ORC-Cdc6 and Cdt1-Mcm2-7 have docked onto each other ( Fig. 1 B, Center) (18). Interestingly, the origin DNA exiting from the ORC-Cdc6 complex is found in the cleft between ORC-Cdc6 and the top of the Mcm2-7 helicase core, clearly outside the Mcm2-7 central channel. Since this intermediate is prior to loading of Mcm2-7 on DNA, but more advanced than semi-attached OCCM, we have termed it "pre-insertion OCCM." We also observed 2D class averages in which origin DNA resides in the Mcm2-7 central channel, suggesting the presence of a loaded conformation (termed "mutant OCCM") and that the removal of Mcm6 WHD did not completely block the DNA insertion ( Fig. 1 B, Lower). During 3D classification, we further obtained a 3D map of the mutant Mcm2-7 hexamer in the presence of ATPγS at 7.7-Å resolution (SI Appendix, Table S1 and Figs. S1 and S2), which was similar to the reported 7 to 8-Å resolution maps of the WT yeast Mcm2-7 bound to ADP or AMPPNP (10), albeit missing the Mcm6 C terminus. Therefore  Table S1 and Figs. S1 and S3). In this complex, ORC and Cdc6 form a six-membered elongated ring with Cdc6 bridging the gap between Orc1 and Orc2. The ring is characterized by interdigitated domain-swapping interactions between the WHDs and the AAA+ domains of adjacent subunits to form a two-tiered ring. The density of ORC-Cdc6-DNA is nearly complete, but most of the Mcm2-7 density is missing, except for the Mcm3 and Mcm7 WHDs. The large AAA+ C-tier collar of ORC-Cdc6 is not obstructed and available for interaction with the C-terminal interface of the incoming Cdt1-Mcm2-7 ( Fig. 2 Orc6 Orc6 Origin DNA Orc6

ORC Cdc6
Orc6 ORC-Cdc6 and B). From 2D classes, we can deduce a clear gap between the two main components, though the cloud-like appearance indicates Cdt1-Mcm2-7 is flexibly attached and displays a large degree of conformational flexibility relative to the helicase loader (Fig. 1B, dashed lines).
In the semi-attached OCCM, we observed the entire C-terminal domain of Orc6, which was making contacts with the origin DNA ( Fig. 2A). This is in contrast to the previously published OCCM and the Drosophila ORC structures, in which only a single Orc6 αhelix was resolved and the Orc6-DNA interaction was absent (18,38). Thus, the data strongly suggest that the C-terminal domain of Orc6 is stabilized by DNA, similar as in the ORC-DNA cryo-EM structure (33).
The ATPase subcomplex Orc1-Orc4-Orc5 in the semiattached OCCM is largely unchanged when compared to the yeast ORC-DNA structure, except for the Orc2 WHD that is rotated and moved out to accommodate Cdc6 (Fig. 2 B and C). Therefore, the Orc2 WHD appears to serve as a flexible gate that allows DNA and Cdc6 insertion. This observation is consistent with the reported ORC-DNA structures in which the Orc2 WHD occupies different positions (33)  the center of the ring, generating a topological link, which explains ORC's strong affinity for DNA (39).
As mentioned above, in the semi-attached OCCM complex, DNA near the narrow C-terminal tier of the helicase loader is in a bent conformation, and is held in place by Orc6. The bent DNA corresponds to the B1-element of the ARS1 origin and makes extensive contacts with Orc2, 5, and 6 via many positively charged and well-conserved residues similar to those in the ORC-DNA structure (33) (Fig. 2A). Moreover, in the semiattached OCCM, the Orc4 insertion helix (IH) is seen binding the DNA major groove (18,33), but not the basic patch of Orc1 that interacts with the ARS consensus sequence (ACS) (33, 40) ( Fig. 2A). The Orc4 insertion loop, the sole connection between ORC-Cdc6 and Cdt1 as observed in the WT OCCM (18), was invisible in the semi-attached OCCM. This indicates that the Orc4 loop is flexible before it binds in an Mcm6 WHDdependent manner to Cdt1 and that Orc4, Cdt1, and Mcm6 WHD form an intricate structural network. WHDs trap DNA in ORC-Cdc6, thereby stabilizing the complex, and in this way advancing pre-RC formation. These findings are consistent with a previous report, which showed that Mcm3 WHD regulates ORC-Cdc6 ATP hydrolysis and is essential for recruitment of Cdt1-Mcm2-7 to ORC-Cdc6 (15). Overall, the semi-attached OCCM structure suggests that, during the initial binding, Mcm2-7 first projects a double-anchor-the Mcm3 and Mcm7 WHDs-onto the ORC-Cdc6 platform (Fig. 2D). Thus, this intermediate may represent the earliest encounter between ORC-Cdc6 and Cdt1-Mcm2-7.
Structure of the "Pre-Insertion OCCM". Another helicase loading intermediate could be resolved, yielding an 8.1-Å resolution 3D map of the "pre-insertion OCCM" (Fig. 3 A-D and SI Appendix,  Fig. 3A). It is noteworthy that, in this complex, the Mcm7 WHD is in direct contact with its AAA+ domain (Fig. 3 A and B). Importantly, the Orc1 WHD establishes close contact with the AAA+ core of  Table S1 and Figs. S1 and S4). The previously reported 3.9-Å resolution structure of the WT OCCM aligns very well with the mutant 3D map (Fig. 4A) (18,42). This implies that deletion of the Mcm6 WHD slowed down OCCM formation, but did not completely abolish the ORC-Cdc6-dependent insertion of DNA into Mcm2-7. During the transition from preinsertion OCCM to WT OCCM, the main body of Mcm2-7 rotates and moves >100 Å to latch onto ORC-Cdc6 (Fig. 4B). In addition, Mcm2-7 switches from the spiral ring to a flat and largely closed ring conformation (18) One key event in this transition is the establishment of the interaction between Cdt1 and Mcm6, which is essential for pre-RC formation in yeast, mice, and humans (14,(43)(44)(45). Since the Mcm2-Mcm5 gate is not fully closed in the OCCM, there will also be significant changes in these two gate subunits as the structure transitions to the gate-closed OCCM, where the Mcm2-Mcm5 interface is fully engaged. To gain insight into the dynamic conformational changes accompanying these transitions, we employed state-of-the-art chain-of-replicas molecular simulation approaches. Specifically, we used the string method with swarms of trajectories (46) to compute an optimal path connecting the experimentally observed start state (pre-insertion OCCM) and the end state (gate-closed OCCM), with the OCCM as an experimental intermediate. The semi-attached OCCM was not used as the start point because the main body of Mcm2-7 was missing in that structure. The path was represented by 49 replicas of the simulation system, initially obtained by taking evenly spaced snapshots from a preliminary targeted molecular dynamics (MD) run. The path was then optimized by the string method on the 49 replicas in the space of two rmsd collective variables (CVs) as described in Methods.
The resultant minimum energy path (MEP) from the preinsertion OCCM to the gate-closed OCCM shows a series of distinct motions (Fig. 5 A-C and Movie S1). ORC-Cdc6 first undergoes a tilt and a concomitant twist toward the Mcm2-Mcm5 gate (Fig. 5 B and C). Initially, ORC-Cdc6 and Mcm2-7 interact through the Mcm4-Orc1 and Mcm7-Orc1 interfaces with additional stabilizing contacts between the Mcm3 and Mcm6 WHDs with Cdc6 and Orc4-Orc5 subunits (Fig. 5A). The global tilt of the ORC-Cdc6 results in the DNA outside the ORC central cavity becoming wedged between Orc5-6 and Mcm2. Importantly, the transition of DNA from the bent to the straight conformation occurs gradually through several discrete intermediates (Fig. 5A) (28). The switch in WHD positioning occurs in the latter stages of the conformational transition and is facilitated by the global tilt of ORC-Cdc6, which creates an opening between Orc1 and Mcm7 that allows the passage of the WHD through the newly formed gap (Fig. 5E).

Mcm5
Mcm2 Cdt1  PCA Analysis Captures 10 On-Path DNA Insertion Intermediates. To more extensively sample the conformational ensemble along the optimal energy path, we further performed free MD simulations of all on-path intermediates. Releasing all string replicas from the imposition of spatial restraints allowed the simulation trajectories to sample in minima of the underlying free energy landscape. We then carried out principal component analysis (PCA) on the combined trajectories, which is well suited to separate the observed conformational states and identify the predominant motions that the protein complex adopt (47). Fig. 6A shows the histogram of our trajectory data projected onto the first two principal components (PC1 and PC2), representing the two largest motions involved in the transition: PC1 is capturing the global tilt and twist motion of ORC-Cdc6 with respect to the Mcm2-7 core (Movie S2), and PC2 is capturing the gate opening and closing motion of the Mcm2 and Mcm5, along with an out-of-plane motion of the Mcm2-7 ring (Movie S3). Our sampling covered the entire path from the pre-insertion OCCM to the gate-closed OCCM. Minima on the PC1 vs. PC2 histogram corresponded to discerned intermediates whose structures can be extracted for further analysis. Thus, we performed an agglomerative clustering analysis to subdivide our conformational ensemble into 10 consecutive on-path intermediary states (S1 to S10) and extracted centroid conformations from each cluster to analyze residue-level contacts (Fig. 6 B-D and Movie S4).

pre-insertion OCCM
Preinsertion intermediates (S1 to S4  capturing ORC-Cdc6. Moreover, these interactions guarantee that Cdt1-Mcm2-7 recognizes the ORC-Cdc6 complex but not ORC alone, since both WHDs contact Cdc6, thus providing complex specificity. This is consistent with the observation that a Mcm3-Mcm5 subcomplex can interact with ORC-Cdc6. However, the extreme Mcm3 C terminus, which is essential for Orc-Cdc6 interaction (15), was not visible in our structure, probably due to flexibility. S. cerevisiae ORC-Cdc6 bends DNA, similarly as seen in 2D class averages of the Drosophila ORC-Cdc6-DNA complex (34) and the high-resolution structure of the budding yeast ORC-DNA complex (33). The pre-insertion OCCM is characterized by additional ORC-Cdc6-Cdt1-Mcm2-7 interactions.  Step 1 Initial anchoring Step 2 Mcm2-7 docking Step 3 DNA inserts into Mcm2-7 Step   Cryo-Grid Preparation. To prepare cryo-EM grids, we applied 3 μL of the loading reaction sample at a final concentration of 0.7 mg/mL to glowdischarged C-flat 1.2/1/3 holey carbon grids, incubated for 10 s at 6°C and 95% humidity, blotted for 3 s, then plunged into liquid ethane using an FEI Vitrobot IV. We loaded the grids into an FEI Titan Krios electron microscope operated at 300-kV high tension and collected images semiautomatically with SerialEM under low-dose mode at a magnification of ×22,500 and a pixel size of 1.31 Å per pixel. A Gatan K2 Summit direct electron detector was used under superresolution mode for image recording with an underfocus range from 1.5 to 3.5 μm. The dose rate was 10 electrons per Å 2 per second, and the total exposure time was 9 s. The total dose was divided into a 30-frame movie, and each frame was exposed for 0.3 s.
Image Processing and 3D Reconstruction. Approximately 10,000 raw movie micrographs were collected. The movie frames were first aligned and superimposed by the program Motioncorr (52). Contrast transfer function parameters of each aligned micrograph were calculated using the program CTFFIND4 (53). All of the remaining steps, including particle auto selection, 2D classification, 3D classification, 3D refinement, and density map postprocessing were performed using Relion-2.0 (54). We manually picked ∼10,000 particles from different views to generate 2D class averages, which were used as templates for subsequent automatic particle selection. Automatic particle selection was then performed for the entire data set. A total of 838,544 particles were initially selected. Particles were then sorted by similarity to the 2D references; about 10% of particles with the lowest z-scores were deleted from the particle pool. Two-dimensional classification of all remaining particles was performed, and particles in unrecognizable classes were removed. The remaining "good" particles were divided into four subsets according to  (55). The DNA structure was also manually adjusted and extended using COOT. The obtained model was then refined in real space against the cryo-EM 3D map using the phenix.real_space_refine module in PHENIX (56). Finally, the quality of the refined atomic models was examined using MolProbity (57). For modeling of the pre-insertion OCCM map at 8.1 Å, the atomic models of Mcm2-7 and ORC-Cdc6-DNA were extracted separately from the WT OCCM and were directly docked as rigid bodies into the EM map in Chimera (58). The initial rigid-body docking was followed by manual adjustment using COOT (55). Due to the low resolution and per the community custom, the manually adjusted atomic model was not subject to further refinement. We did not build atomic models for the 3D maps of the "mutant Mcm2-7" hexamer and the "mutant OCCM" because these mutant structures were similar to the corresponding WT structures that had been reported (18,33). Structural figures were prepared in Chimera and PyMOL (https://pymol.org/2/).
Computational Modeling. We constructed models of pre-insertion OCCM and the gate-closed OCCM complex. We then used molecular dynamics flexible fitting (MDFF) (59) to refine the models into the respective EM density maps. Prior to MDFF, each model was solvated with equilibrated TIP3P solvent in a simulation box with 15 Å spacing from the protein complex to the edge of the box. Counterions, Na + and Cl − , were added to neutralize the overall charge of the complex and adjust salt concentration to 150 mM. The solvated systems were then minimized for 10,000 steps, heated in the NVT ensemble, and then equilibrated in the NPT ensemble (1 atm and 300 K). During equilibration, positional restraints on all heavy atoms were gradually reduced from 5 to 0 kcal mol −1 Å −2 while simultaneously employing MDFF grid forces with a scaling factor of 0.1. We then guided the equilibrated structures of the pre-insertion OCCM into the gate-closed OCCM configuration using targeted molecular dynamics (TMD). The TMD production run was completed in 20 ns and employed 1,000 kcal/mol force constant on the backbone atoms of the protein and DNA. Detailed procedures are described in SI Appendix, Supporting Information Methods. The initial MCM-loading path was represented by 49 evenly spaced snapshots taken from the targeted MD trajectory. This path was then optimized using the finitetemperature string method with swarms of trajectories. String method protocol and definition of collective variables are provided in the SI Appendix. The optimized configurations were then released for 50 ns of free unbiased MD per replica. This resulted in ∼2.5 μs of aggregate simulation time. The combined trajectories were then subject to principal component analysis and agglomerative clustering with CPPTRAJ (60). VMD and UCSF Chimera packages were used for analysis and visualization (58,61).