Changing protein–DNA interactions promote ORC binding-site exchange during replication origin licensing

Significance Bidirectional DNA replication, in which two replication forks travel in opposite directions from each origin of replication, is required for complete genome duplication. To prepare for this event, two copies of the Mcm2-7 replicative helicase are loaded in opposite orientations at each origin. Using single-molecule assays, we identified a sequence of changing protein–DNA interactions involved in this process. These stepwise changes gradually reduce the DNA-binding strength of ORC (origin recognition complex), the primary DNA-binding protein involved in this event. This reduced affinity promotes ORC dissociation and rebinding in the opposite orientation on the DNA, facilitating the sequential assembly of two Mcm2-7 molecules in opposite orientations. Our findings identify a coordinated series of events that drive proper DNA replication initiation.

Eukaryotic chromosomal replication initiates by building two oppositely oriented replication forks at origins of replication. During G1, the core component of the replicative helicase, the Mcm2-7 complex, is assembled around origins in a process termed origin licensing or helicase loading. This process marks all potential origins of replication, and a subset of the loaded helicases are activated in S phase to form the core of the replication machinery (1). To ensure bidirectional replication, each pair of Mcm2-7 helicases must be assembled around origin DNA in a head-to-head orientation (2)(3)(4)(5)(6).
In addition to protein-protein interactions, ORC forms extensive but changing interactions with DNA (Fig. 1A, red dashed boxes). In budding yeast, the primary ORC binding site is defined by the ARS consensus sequence (ACS) and B1 element (11,13,14). ORC encircles the ACS as an open ring, forming additional interactions with DNA including and beyond the B1 element to create a strong ~80° bend in DNA (15). Structural studies have captured a "preinsertion OCCM" intermediate with bent DNA that contains all helicase-loading proteins (Fig. 1A) (16). However, the bent DNA straightens and is inserted in the Mcm2-7 open ring in the later OCCM intermediate (17). These intermediates

Significance
Bidirectional DNA replication, in which two replication forks travel in opposite directions from each origin of replication, is required for complete genome duplication. To prepare for this event, two copies of the Mcm2-7 replicative helicase are loaded in opposite orientations at each origin. Using single-molecule assays, we identified a sequence of changing protein-DNA interactions involved in this process. These stepwise changes gradually reduce the DNA-binding strength of ORC (origin recognition complex), the primary DNA-binding protein involved in this event. This reduced affinity promotes ORC dissociation and rebinding in the opposite orientation on the DNA, facilitating the sequential assembly of two Mcm2-7 molecules in opposite orientations. Our findings identify a coordinated series of events that drive proper DNA replication initiation.
together suggest that a segment of DNA flanking the bend, which we refer to as the bend-proximal region (BPR) (Fig. 1A), is transferred from ORC to Mcm2-7, allowing the DNA bend to straighten as the BPR enters the Mcm2-7 central channel. Although molecular dynamics models have simulated the DNA insertion process (16), DNA unbending and deposition into Mcm2-7 have yet to be monitored in real time. In addition, the trigger(s) for this change in ORC-DNA interactions has not been defined.
For one ORC to load both Mcm2-7 helicases at an origin, ORC has to switch between two oppositely oriented DNA-binding sites. Natural yeast origins include at least one B2 element, each of which includes a partial match to the ACS but in the inverted orientation (Fig. 1A, purple arrow) (18)(19)(20)(21). As a result, the B2 element is proposed to serve as a weaker secondary binding site for ORC. Studies with artificial origins showed that a second inverted binding site is essential for robust helicase loading (22).  Interestingly, B2 elements are found at variable distances from the ACS (18,20,23). A key question raised by the "one-loader" hypothesis for helicase loading is how ORC releases from its strong initial binding site to bind a weaker, oppositely oriented secondary binding site a variable distance away.
Combining colocalization single-molecule spectroscopy (CoSMoS) (24,25) and single-molecule Förster resonance energy transfer (sm-FRET) (26), we monitored a variety of interactions between ORC and Mcm2-7 with the origin DNA and identified key triggers for changes in the observed protein-DNA interactions. We demonstrated that the ORC-DNA complex predominantly contains bent DNA but infrequently transitions to a less stable conformation with straightened DNA. Cdc6 binding does not change the bent state of the ORC-bound DNA but does decrease the dissociation of ORC from DNA. Although the OCCM is static on the DNA, upon loss of Cdc6 (Fig. 1A, step 4) we observe changes in sm-FRET supporting sliding of the ORC-Cdt1-Mcm2-7 (OC 1 M) complex. Based on our data, we propose that the sequential events of DNA unbending, Cdc6 release, and DNA sliding progressively reduce ORC stability on DNA to facilitate ORC dissociation from DNA and flipping to bind the inverted B2 element.

The ORC-DNA Complex Predominantly Exists in a Bent DNA
State. To understand how and when origin DNA changes its conformation during helicase loading, we developed a sm-FRET assay to monitor the dynamics of ORC-induced DNA bending. Previous cryo-EM structures showed that the C-terminal face of ORC, particularly the C-terminal domain of Orc6, is in close proximity to the BPR of origin DNA (15). To position a donor fluorophore close to the BPR, we generated an ORC fluorescently labeled on its C-terminal face (ORC C ). To this end, we deleted residues 1 to 266 of Orc6 and attached a fluorophore to the truncated N terminus. Although defective in recruiting a second , truncation of this region in Orc6 does not compromise OCCM formation in ensemble helicase-loading assays (SI Appendix, Fig. S1A). An acceptor fluorophore was coupled to DNA at an internal position 51 bp from the ARS1 origin ACS (ARS1 +51-Cy5 ). This position was chosen to maximize proximity to the Orc6 label when ORC is bound to the BPR and minimize their proximity when the DNA is straight (Fig. 1B). We refer to the FRET between this donor-acceptor pair as "ORC C •+51" FRET ( Fig. 1B), and the corresponding interaction between ORC and the BPR as the "ORC-BPR" interaction ( Fig. 1A, first panel, dashed box). Using total internal reflection fluorescence microscopy, we monitored the colocalization of ORC C in solution with surface-tethered origin DNA (24). Throughout each ORC-DNA colocalization event, we measured effective ORC C •+51 FRET efficiency (E FRET ) to examine ORC-induced DNA bending.
When ORC arrived on DNA, we consistently observed a high ORC C •+51 E FRET state, indicating that ORC was bound to the ACS and BPR and that the bound DNA was bent (blue background in Fig. 1 C and D). The majority of ORC C •+51 E FRET values were centered at 0.62 (Fig. 1D). Although they represented less than 0.5% of the time of ORC-DNA colocalization, we also observed short periods of lower E FRET values (<0.28, orange background in Fig. 1 C and D and SI Appendix, Fig. S1B). These low E FRET values were not caused by photobleaching, as we examined the DNA-coupled fluorophore by acceptor excitation after the experiment and restricted analysis to DNA molecules that maintained fluorescence to the end of the experiment. In addition, control experiments showed that these low E FRET values were not caused by ORC sliding along the DNA to move away from the ACS (SI Appendix, Fig. S2). We excluded a small fraction (approximately 2%) of ORC molecules that did not show a high E FRET signal at any time during the DNA colocalization, as these likely represent nonspecific ORC binding. The poor fit of a single Gaussian model at low E FRET values (Fig. 1D, red curve in Inset) suggest the presence of distinct populations at the bent or unbent conformational states. We conclude that the low E FRET values reflect an unbent state in which the ORC remains bound at the ACS but the ORC-BPR interaction is lost (Fig. 1B). Addition of Cdc6 increases ORC stability on DNA, leading to a near twofold decrease in its apparent dissociation rate from 0.0131 s −1 to 0.0073 s −1 (SI Appendix, Fig. S1C and Table S1). However, Cdc6 did not shift the ORC C •+51 E FRET distribution between the bent and unbent states, consistent with Cdc6 not significantly altering the ORC-BPR interaction (SI Appendix, Fig. S1C) (16,28,29). Thus, ORC-bound DNA is predominantly in the bent state with infrequent transitions to the unbent state.
We frequently observed the low E FRET unbent state immediately prior to ORC dissociation from DNA ( Fig. 1C and SI Appendix, Fig. S1B). Thus, we asked whether DNA unbending increases the rate of ORC dissociation from DNA. Indeed, ORC in the low E FRET state (<0.28) dissociates more than 40 times faster from DNA than ORC in the high E FRET state (≥0.28) (Fig. 1E). These data indicate that the ORC-BPR interaction increases ORC stability on DNA.
As a further test of the ORC-BPR contribution to the kinetic stability of the ORC-DNA complex, we mutated 9 amino acids in the ORC region that interacts with the phosphate backbone of the BPR (15) (SI Appendix, Fig. S1D). When these mutations are incorporated into labeled ORC (ORC C-BPR ), we observed a limited decrease in E FRET (centered at 0.52, SI Appendix, Fig. S1E), suggesting that these interactions contribute to but are not solely responsible for ORC-induced DNA bending. Nevertheless, ORC C-BPR stability on DNA is notably reduced (SI Appendix, Fig. S1E), reflected by a significant increase in its apparent dissociation rate (0.0849 s −1 ) compared to WT (0.0131 s −1 , SI Appendix, Table S1). We conclude that the kinetic stability of ORC on DNA is strongly dependent on the ORC-BPR interaction, as a partial loss of this interaction results in a strong reduction in ORC-DNA stability.

Mcm4 and Mcm6 Winged-Helix Domains (WHDs) Trigger DNA
Unbending. After Mcm2-7 recruitment, the BPR disengages from ORC and is inserted into the helicase central channel (17). To understand the transition of the BPR DNA from being bound to ORC-Cdc6 to being deposited into the helicase, we investigated the kinetics of DNA unbending during helicase loading. We included Cdc6 and labeled Mcm2-7 4N-650 -Cdt1 in our experiment to simultaneously monitor ORC-BPR interactions and Mcm2-7 recruitment. To avoid using red-excited fluorophores on both DNA and Mcm2-7, we replaced the DNA-coupled acceptor with the fluorescence quencher Black Hole Quencher-2 (BHQ-2) at the same DNA site (ARS1 +51-BHQ2 ). In this experiment, DNA bending induced by ORC-BPR interaction results in quenching of green-excited ORC C fluorescence, and DNA unbending due to loss of the ORC-BPR interaction leads to unquenching ( Fig. 2A). Using increase in green-excited fluorescence as a marker for DNA unbending, we consistently detected a temporal delay between Mcm2-7 arrival (Fig. 1A, step 2) and DNA unbending ( Fig. 2B and SI Appendix, Fig. S3). Unbending occurs 3.9 ± 0.3 s (mean ± SEM) after Mcm2-7 arrival, which is significantly earlier (Fig. 2C) than Cdc6 release (Fig. 1A, step 4) (7). Given the temporal delay between Mcm2-7 recruit.ment and DNA unbending, we explored the possibility of additional regulatory steps between these two events.
To ensure that this E FRET decrease is associated with successful helicase loading, we limited our analysis to Mcm2-7-DNA colocalization events that resulted in highsalt-resistant, loaded Mcm2-7 hexamers (7,32). Using data from the experimental setup described in Fig. 3A, we plotted a twodimensional heat map for these productive loading events to visualize MCM•+51 E FRET values 0 to 100 s after the first Mcm2-7 arrival (Fig. 4A). We divided the heat map into four time windows (TW, Fig. 4A) and fit a two-component Gaussian model to the E FRET distribution in each TW (Fig. 4B, red curves, SI Appendix, Table S1). The centers of the low and high Gaussian components in TW1 and TW2 were not significantly different ( Fig. 4B and SI Appendix, Table S1). Consistent with Fig. 3B, the mixture of two Gaussian components in TW1 and TW2 suggests that most helicase-loading intermediates transitioned from a low-E FRET DNA-bent state to a high-E FRET DNA-deposited state between TW1 and TW2.
After DNA deposition, E FRET values decreased in TW3 to a lower level maintained in TW4 (Fig. 4A). Although TW3 shows a gradual decrease in the aggregate heat map (Fig. 4A), single-molecule records show examples of both gradual and sharp E FRET decline (SI Appendix, Fig. S5). Notably, the peak centers of TW3 and TW4 did not overlap with those of TW1 and TW2 (Fig. 4B), suggesting the presence of new conformational states distinct from the DNA-bent and the DNA-deposited states. Mcm2-7 sliding on DNA (SI Appendix, Fig. S6A). To confirm Mcm2-7 sliding, we created two additional DNA substrates, each with a single DNA-coupled acceptor at a position located more than 30 bp (>10 nm) away from the +51 position used to detect DNA deposition (Fig. 4C). The −4 dye (ARS1 -4-Cy5 ) is intended to detect leftward Mcm2-7 3N-550 sliding, and the +82 dye (ARS1 +82-Cy5 ) rightward sliding ( Fig. 4C and SI Appendix, Fig. S6B).
In experiments with DNA-coupled acceptors at either the −4 or +82 positions, more than half of the DNA-bound Mcm2-7 molecules exhibited one or more E FRET peaks (Fig. 4 D and E), implying that the Mcm2-7 molecules can slide leftward or rightward along the DNA. Because some molecules displayed more than one E FRET peak (Fig. 4D, orange arrow), we infer that the sliding movement can change direction. For Mcm2-7 colocalization events that displayed high E FRET , the time to sliding detected by the −4 and +82 dyes is significantly longer than time to deposition after Mcm2-7 arrival (Fig. 4F), which is consistent with sliding occurring only after initial DNA deposition. Together, these findings suggest that after DNA deposition, Mcm2-7 single hexamers can slide back and forth on DNA with no preferential direction.  (Fig. 5A). Although we observed a spectrum of different E FRET patterns in the mutants (SI Appendix, Fig. S7), we consistently observed a prolonged duration of the high E FRET state (Fig. 5A) compared to WT Mcm2-7 (Fig. 4A). Thus, efficient Mcm2-7 sliding is initiated by at least one round of ATP hydrolysis at each Mcm2-7 subunit interface.
We next investigated the temporal relationship between the start of Mcm2-7 sliding and the sequential releases of Cdc6 and Cdt1 from the OCCM (Fig. 5 B-F). We performed MCM•+51 FRET experiments in the presence of Cdc6 or Cdt1 labeled with a red-excited dye to monitor Cdc6 or Cdt1 release simultaneously with MCM•+51 E FRET (Fig. 5 B and E). We labeled the N terminus of Cdc6 (Cdc6 N-649 ), which is distant from the Mcm3 N-terminal FRET donor based on structural studies (17) and   Fig. S8). We used alternating red and green laser excitation to monitor protein binding to DNA and FRET: Green excitation enabled monitoring of FRET and DNA association of green-excited Mcm2-7, while red excitation monitored DNA association of red-excited Cdc6 or Cdt1.
In these MCM•+51 FRET experiments, Cdc6 or Cdt1 release is represented by a decrease in red-excited fluorescence (Fig. 5 C  and F, red arrows). The initiation of Mcm2-7 sliding is indicated by E FRET decrease below the defined threshold (Fig. 5 C and F, red dotted line, SI Appendix, Fig. S8 B and E). More than 90% of DNA-bound Mcm2-7 molecules exhibited E FRET decrease only after Cdc6 release (Fig. 5D, positive values, SI Appendix, Fig. S8C). In contrast, we found that 80% of Mcm2-7 exhibited the E FRET decrease before Cdt1 release (Fig. 5G, negative values, SI Appendix,    S8F). Taken together, these findings show that Mcm2-7 sliding primarily occurs after Cdc6 release but before Cdt1 release.

Mcm2-7, and Cdt1. Because the initial interactions between ORC
and Mcm2-7 remain stable until Cdt1 release (9), our observation that Mcm2-7 sliding typically begins before Cdt1 release strongly suggests that the initial sliding complex also includes Cdt1 and ORC. To confirm this hypothesis and characterize the timing of ORC sliding relative to other steps in helicase loading, we developed a FRET assay that monitors ORC-ACS interaction.
We placed a donor fluorophore on the N-terminal face of ORC (ORC N , see Materials and Methods) and an acceptor fluorophore adjacent to the ACS (ARS1 -4-Cy5 ) (Fig. 6A) , Fig. S9B). By determining the time interval between Mcm2-7 arrival and ORC N •-4 E FRET decrease, we found that ORC leaving the ACS occurs significantly after DNA unbending or deposition into the Mcm2-7 central channel (Fig. 6C).
The decrease in ORC N •-4 E FRET could be due to ORC sliding away from ACS or ORC flipping associated with MO formation. If ORC begins sliding simultaneously with Mcm2-7, ORC N •-4 E FRET decrease would fall between Cdc6 and Cdt1 release (Fig. 5  B-G). However, if the E FRET decrease corresponds to ORC flipping, which occurs after Cdt1 release (9), we would expect ORC N •-4 E FRET decrease to follow Cdt1 release. To distinguish between these two possibilities, we determined when ORC leaves the ACS relative to Cdc6 and to Cdt1 release. To this end, we performed ORC N •-4 E FRET experiments in the presence of red-excited Cdc6 N-649 or Cdt1 N-649 . Consistent with the predicted distances from structural studies (17), neither Cdc6 N-649 nor Cdt1 N-649 exhibits significant E FRET to ORC N (SI Appendix, Fig. S9A).
ORC N •-4 E FRET consistently decreased after Cdc6 release (80/84 positive values in Fig. 6D and SI Appendix, Fig. S10A). In contrast, ORC N •-4 E FRET almost always decreased before Cdt1 release (89/91 negative values in Fig. 6E and SI Appendix, Fig. S10B). These distributions show that ORC movement away from ACS predominantly begins between Cdc6 release and Cdt1 release, matching the time interval during which Mcm2-7 starts sliding. Importantly, we detected negligible background sliding of ORC in the absence of other helicase-loading factors (SI Appendix, Fig. S2), consistent with ORC sliding being temporally controlled during helicase loading.
To detect ORC sliding, we examined E FRET between ORC and a DNA probe distant from the ACS. By placing the ARS1 DNA acceptor at the +51 position, a high ORC N •+51 E FRET is only expected if ORC moves to the proximity of the +51 dye (SI Appendix, Fig. S6C). Importantly, ORC bound at B2 would place the donor fluorophore too far (>8.2 nm) from the +51 DNA fluorophore to exhibit high E FRET (SI Appendix, Fig. S6C). In the presence of labeled Cdt1 N-649 , we observed high ORC N •+51 E FRET peaks that occurred before Cdt1 release (Fig. 6F and SI Appendix,  Fig. S10C). These E FRET peaks indicate that ORC slides away from the ACS while still bound to the C-terminal face of Mcm2-7 and before flipping to form the MO complex.
Given our previous observation that OC 1 M sliding is bidirectional (Fig. 4E), we would expect ORC to revisit the ACS during sliding. If true, then in the ORC N •-4 E FRET experimental setup, after an initial decrease in E FRET, we would sometimes observe increases that represent ORC reengagement with the ACS. Indeed, 42.9% of ORC molecules exhibited a return to high ORC N •-4 E FRET signal after the initiation of sliding (SI Appendix, Fig. S11A). Consistent with ORC having low affinity for the ACS under these conditions, these rebinding events were short-lived relative to ORC independently binding the ACS (SI Appendix, Fig. S11B). Interestingly, only 3.3% of ORC molecules showed a high ORC N •-4 E FRET signal after Cdt1 release, suggesting that Cdt1 release inhibits ORC rebinding to the ACS (SI Appendix, Fig. S11A).
Together, our studies show that sliding of ORC and Mcm2-7 on the DNA is temporally controlled. This process is triggered by the release of Cdc6 from the OCCM but does not require Cdt1 release. Because the initial ORC-Mcm2-7 interactions remain stable until Cdt1 release (9), we conclude that the OC 1 M complex, consisting of ORC, Cdt1, and Mcm2-7, exhibits bidirectional sliding on dsDNA. ORC sliding away from the ACS represents a third mechanism to reduce ORC affinity for DNA, preparing it to rapidly release from the DNA upon Cdt1 release and disruption of the initial ORC-Mcm2-7 interface. Together, these events drive ORC release from the ACS and facilitate its subsequent flipping, B2 element binding, and MO complex formation.

Discussion
Our previous studies showed that a single ORC can mediate loading of both helicases found at licensed origins (7,9). This process requires the same ORC to sequentially bind two distinct and oppositely oriented sites on the DNA. This model raises the question of how ORC is released from its initial high-affinity binding site ACS.
Using sm-FRET assays to examine ORC and Mcm2-7 interactions with DNA during recruitment of the first Mcm2-7, our studies revealed a coordinated series of changes that promote ORC site exchange. We propose an updated helicase loading model that incorporates the findings from our study (Fig. 7). DNA-bound ORC consistently induces DNA bending by simultaneously interacting with the ACS and the BPR (Fig. 1D). Shortly after recruitment of the first Mcm2-7 onto ORC-Cdc6 bound to bent DNA (Fig. 7, 1st Mcm2-7 recruitment), DNA unbending is triggered by interactions formed during OCCM assembly, such as between ORC and the WHDs of Mcm4 and Mcm6 (Fig. 2D). This leads to the rapid DNA deposition into the Mcm2-7 central channel (Figs. 3D and 7, circled 1). Loss of the ORC-BPR interaction destabilizes ORC on DNA (Fig. 1E), and subsequent Cdc6 release (Fig. 7, circled 2) further weakens ORC-DNA interactions (SI Appendix, Fig. S1C). Mcm2-7 ATP hydrolysis activity initiates ORC-Cdt1-Mcm2-7 (OC 1 M) sliding on DNA (Figs. 5A and 7, circled 3), which both fully releases ORC from the ACS and enables access to B2 elements located at variable distances from the ACS. The sequential events of DNA unbending, Cdc6 release, and OC 1 M sliding progressively reduce ORC stability on DNA (Fig. 7, gray wedge) to promote ORC dissociation from its strong site, allowing for its subsequent rebinding to the inverted B2 element (Fig. 7, ORC Flip). After ORC flipping, B2-bound ORC in the context of the MO complex recruits a second Mcm2-7 in the inverted orientation, completing helicase loading (Fig. 7, 2nd Mcm2-7 recruitment and DH).
Stepwise Reduction in ORC Stability on DNA. Our finding that ORC is progressively destabilized on DNA is consistent with previous structural studies. ORC makes numerous contacts with the BPR when the DNA is in a bent state (15), and we show that mutation of the amino acids responsible for a subset of these interactions significantly reduces ORC-DNA stability (SI Appendix, Fig. S1D). Similar ORC-BPR interactions observed in metazoan ORC suggest that DNA bending is a common mechanism for regulating ORC-DNA complex stability (37). A role for the Mcm4 and Mcm6 WHDs in triggering DNA unbending is consistent with structural studies showing that these domains orient the Mcm2/5 gate next to the BPR for subsequent deposition (38). Notably, the Orc1 basic patch region that interacts with the ACS in the ORC-ACS structure becomes disordered in the OCCM structure, which may further contribute to the lowered ACS affinity of ORC (15,17).
Following DNA unbending, Cdc6 release serves a dual purpose. Structural studies of the ORC-Cdc6-DNA complex show that Cdc6 binding to ORC completes an ORC-Cdc6 protein ring around the DNA (17,28). Cdc6 dissociation is thus required to reopen the ORC ring for DNA release from its central channel during ORC flipping. Moreover, the WHD and initiator-specific motif of Cdc6 form extensive specific contacts with the ACS (28). Both of these functions are consistent with previous data showing that Cdc6 enhances ORC binding to the ACS (39) and our data that Cdc6 stabilizes DNA-bound ORC (SI Appendix, Fig. S1C). The decreased affinity of ORC for the ACS after Cdc6 release would lower the energy barrier for OC 1 M sliding on DNA, which would further reduce specific ORC-DNA interactions. Even upon reencountering the ACS site during sliding, ORC in this destabilized state fails to maintain stable interactions with the ACS, resulting in its rapid disengagement from the site (SI Appendix, Fig. S11). The combined effects of DNA unbending, Cdc6 release, and OC 1 M sliding promote ORC dissociation from DNA necessary for ORC flipping.
Our studies also provide insights into the importance of the ordered release of Cdc6 and Cdt1 from the OCCM. Cdt1 release is associated with the disruption of the interaction between ORC and the C-terminal domain of Mcm2-7 involved in initial Mcm2-7 recruitment (9). Thus, the retention of Cdt1 as sliding commences means that both ORC and Mcm2-7 remain as a complex during initial sliding. Once ORC associates with nonspecific DNA, we predict that the stability of ORC on DNA is highly dependent on its interaction with the C-terminal domain of Mcm2-7. When this ORC-Mcm2-7 interaction is disrupted during Cdt1 release, ORC is poised to rapidly dissociate from DNA and seek a new binding site.
Two key questions remain to be addressed by future studies.  , Fig. S11A). Future experiments will be necessary to determine the precise mechanism of ORC binding to the inverted B2 site.
Energy Requirements that Lead to ORC Flipping. Because the DNA binding strength of ORC is sequentially decreased to facilitate ORC flip and MO formation (Fig. 7, gray wedge), we propose that the resulting increase in the free energy of ORC-DNA interactions is compensated by additional stabilizing interactions or by energy derived from ATP hydrolysis. First, destabilization caused by the loss of ORC-BPR interaction during DNA unbending is offset by stabilizing interactions formed between DNA and the central channel of Mcm2-7 during DNA deposition (MCM-BPR in Figs. 1 and 3), as well as interactions between ORC and Mcm2-7 that involve Mcm4 and Mcm6 WHDs (Fig. 2D). Second, ATP hydrolysis by Cdc6 has been reported to drive Cdc6 release (41). Third, consistent with previous studies showing that Mcm2-7 ATP hydrolysis is required for helicase loading (36,42), our study demonstrated that Mcm2-7 ATP hydrolysis is required for the onset of OC 1 M sliding (Fig. 5A). A likely explanation for this requirement is the ATP control of a transition of the Mcm2-7 βhairpin loops from a DNA-engaged (as seen in the OCCM) (3,17) to a weakly DNA-bound open conformation (as seen in the DH) that favors DNA sliding (2,43).
Although ATP hydrolysis is important for initiating sliding, our data suggest that OC 1 M movement involves passive diffusion instead of ATP-powered translocation. We observed evidence for movement in both directions from the initial ORC-Mcm2-7 binding sites (Fig. 4 D and E), which differs from the biased random walk exhibited by the CMG helicase during active translocation along ssDNA (43). In addition, ORC, single or double Mcm2-7 hexamers, and potential loading intermediates including both ORC and Mcm2-7 have been shown to passively slide on dsDNA (5,6,8,35,44,45). Analogous sliding has been observed in many different protein complexes including MutS (46,47) and Type III restriction enzymes (48).  Fig. 7. A helicase loading model. We propose that DNA unbending and deposition, Cdc6 release, and OC 1 M sliding (red circled numbers) sequentially lower the DNA binding strength of ORC (gray wedge). These events facilitate ORC to release DNA from its central channel and rebind to an inverted binding site (ORC Flip), which enables loading of the second Mcm2-7 in the correct orientation and completes formation of the Mcm2-7 DH. steric obstruction of B2 by the first Mcm2-7. At other origins, such sliding would similarly provide ORC and the first Mcm2-7 access to one or more potential B2 elements. Our data also agrees with previous studies that single roadblocks on either side of the origin do not impact helicase loading (49) because the OC 1 M complex is able to slide away from the roadblocks to access distant B2 elements. The lack of sliding directionality is also consistent with observations that artificial origins with B2 positioned on either side of ACS can support helicase loading (22). A strong requirement for a specific B2 element is observed when flanking roadblocks limit OC 1 M sliding such that only a single B2 element is accessible (49), or in artificial origins where only a single B2-like sequence is present (22). As such, sliding enables ORC to search for another near-ACS sequence in the inverted orientation. Though any ACS-related sequence on either side can serve as a secondary ORC landing site, the distance between the two binding sites is likely confined in vivo by originflanking nucleosomes that limit sliding (50,51). As a result, the B2 elements that are found to be important in vivo (18,20) are likely the most accessible secondary sites relative to the ACS.

Role of
Our study identifies the OC 1 M as the first sliding intermediate and establishes that sliding is prevented before this intermediate is formed, but it is likely that later helicase-loading intermediates also slide on DNA to search for a stable secondary binding site. Because previous studies proposed that ORC remains tethered to the helicase to prevent its dissociation while exchanging binding sites (8,9), it is possible that the DNA-bound Mcm2-7 single hexamers travel along DNA while the tethered ORC actively searches for a suitable landing site. Moreover, given that ORC can search DNA via diffusion (34,35), and the MO complex has been found to occupy DNA sites other than the initial recruitment site (8), the MO complex can likely diffuse on DNA prior to ORC identifying a stable binding site in the inverted orientation. Future experiments will be required to demonstrate and characterize sliding of other helicase-loading intermediates.

Materials and Methods
Nomenclature for Fluorescently Modified Proteins and DNAs. We use a superscript notation to describe proteins and DNA with fluorescent or dark quencher dye modifications. The superscript notation consists of two parts separated by a dash: the site of modification and the type of modification. If the type of modification is denoted by a number, it refers to the Dylight™ fluorophore conjugated. For example, ARS1 +51-BHQ2 refers to ARS1 origin labeled at the +51 nucleotide position relative to ACS with a BHQ-2; Mcm2-7 4N-550 refers to Mcm2-7 labeled with Dylight 550 at the Mcm4 N terminus. We used abbreviations for two specific ORC constructs: ORC C for ORC labeled at the C-terminal face with Dylight 550 and ORC N for ORC labeled at the N-terminal face with Dylight 550 (see subsections below).
SI Appendix, Table S2 summarizes all the fluorescently-labeled proteins and DNA used in this study.

Preparation of Internally Modified
+82-Cy5 ). Internally modified ARS1 DNAs were assembled by ligating two overlapping PCR fragments. A list of oligos (Integrated DNA Technologies) used in this study is provided in SI Appendix , Table S3. The internal Cy5 (iCy5) and internal BHQ2 (iBHQ2) fluorophores were directly attached through the DNA phosphate backbone. The biotinylated PCR fragments were generated with 5′ biotin-labeled oligos, and the dye-coupled PCR fragments were generated using internally labeled oligos that anneal to the overlap region. The two PCR fragments were ligated as described previously (49) to yield the final 1.4-kb DNA construct containing a biotin end modification and an internal dye modification near the point of ligation.
Sortase-Mediated Protein Labeling and Purification. Fluorescence labeling at the N or C terminus of proteins was performed via sortase-mediated coupling and purified as described (9).

Preparation of ORC
C and ORC C-BPR . S. cerevisiae strains overexpressing the codon-optimized ORC subunits with the indicated mutations were grown, arrested, induced, harvested, and lysed as described above. Following elution from the FLAG resin, peak fractions were coupled to Dylight 550 via sortase as above. . Cdc6 N-649 and Cdt1 N-649 were purified as described (7). To separate the Cdt1-conjugated fluorophore from the donor fluorophore on Mcm2-7 3N-550 , a 3× FLAG tag and a rigid alpha-helical (EAAAK) 10 linker (53) was placed at the N terminus of Cdt1 after the sortaserecognition motif GGG. Cdt1 was purified via FLAG resin and labeled at the N terminus via sortase-mediated conjugation as described above.
Single-Molecule Assays. A micromirror total internal reflection microscope was used to perform multiwavelength single-molecule imaging (25). Glass slides were functionalized with PEG and Biotin-PEG; fiducial markers and biotinylated DNA were coupled to the slide as described (9). All reactions were performed in buffer containing 25 mM HEPES-KOH pH 7.6, 0.3 M potassium glutamate, 5 mM Mg(OAc) 2 , 3 mM ATP, 1 mM dithiothreitol, 1 mg/mL bovine serum albumin, with an oxygen scavenging system (glucose oxidase/catalase), and 2 mM Trolox (54). ORC C •+51 and ORC N •-4 experiments contained 1 nM ORC, 1 µM 60-bp nonspecific DNA generated by annealing two oligonucleotides (SI Appendix, Table S3), and 3 nM Cdc6 if specified. For helicase loading assays, 5 to 10 nM Mcm2-7/Cdt1 was added. DNA molecules labeled with Alexa488 or Cy5 were identified before the experiments using 488-nm or 633-nm excitation, respectively. Three different protocols were used for experimental acquisition: 1) reactions containing only green-excited proteins: continuous 532-nm excitation of specified exposure time; 2) reactions containing green-excited ORC, red-excited Mcm2-7, and quencher-labeled DNA: simultaneous 532-nm and 633-nm excitation; 3) reactions containing green-excited ORC, redexcited Mcm2-7, and red-excited DNA: alternating 532-nm and 633-nm excitation. Cy5 internally labeled DNA that did not photobleach was identified by 633-nm excitation after the experiments. To identify salt-resistant helicases, a high-salt buffer containing 25 mM HEPES-KOH pH 7.6, 0.5 M KCl, and 5 mM Mg(OAc) 2 was applied to wash away loading intermediates at the end of the experiments.
FRET Data Analysis. Data were analyzed as described previously (7) and fluorescence intensity values were corrected for background fluorescence as described (55). Apparent FRET efficiency (E FRET ) calculations were performed as described (9).
E FRET values were maximum likelihood fit using the models and yielding the fit parameters given in SI Appendix , Table S1. The E FRET threshold used to differentiate the two E FRET states was defined as the position of the trough in the fit. An E FRET transition was considered to have taken place once the E FRET value crossed the threshold for at least two consecutive frames.

ORC Apparent Dissociation Rate in High and Low E FRET States.
In the ORC C •+51 FRET experiment, we used the threshold 0.28 to distinguish between the high and low E FRET states. The frequency of ORC release (Fig. 1E) in its high and low E FRET states was determined by first identifying the E FRET state of ORC in