An Orc6 tether mediates ORC binding-site switching during replication origin licensing

Orc6N Interacts with Mcm2-7N Rapidly Upon Mcm2-7 Recruitment.

A single ORC molecule transitioning between the OCCM and MO complexes could be explained by the Orc6N domain binding to Mcm2-7N, and the Orc6 flexible linker tethering the rest of ORC to Mcm2-7. If Orc6 forms such a tether before the ORC binding-site transition, then Orc6N should bind to Mcm2-7N before MO complex formation. To determine when Orc6N associates with the Mcm2-7N, we developed a smFRET assay to detect such interactions using S. cerevisiae proteins (Fig. 1 A and B). Orc6 was labeled at a peptide inserted adjacent to the N-terminal folded domain (position 107) using the SFP enzyme (ORC^6N-550, SI Appendix, Fig. S1B; 33). The Mcm2-7 helicase was labeled at its N-tier by deleting the Mcm6 N-terminal extension (amino acids 1 to 103) and attaching a fluorescently-labeled peptide adjacent to this site using Sortase (Mcm2-7^6N-650) (SI Appendix, Fig. S1B). Previous studies showed the Mcm6 extension is dispensable for helicase loading (27). Importantly, bulk helicase-loading remained robust in reactions using one or both of these modified proteins (SI Appendix, Fig. S1C). Using total internal reflection fluorescence microscopy, we monitored the colocalization of ORC^6N-550 and Mcm2-7^6N-650 with individual origin DNA molecules (Fig. 1A; 32). Alternate excitation of the donor and acceptor fluorophores allowed observation of the association of either ORC or Mcm2-7 with DNA. Interactions between Orc6N and Mcm2-7N were monitored by determining the apparent FRET efficiency (E_FRET) during donor excitation (Fig. 1B). We restricted our analysis to events in which ORC and Mcm2-7 were both simultaneously present on DNA for at least five frames (>~12 s) to focus on productive ORC–Mcm2-7 interactions. We will refer to association events that meet these criteria as “stably recruited” Mcm2-7 helicases.

When both ORC^6N-550 and Mcm2-7^6N-650 colocalized with DNA, E_FRET values frequently and rapidly increased after Mcm2-7^6N-650 recruitment (Fig. 1C and SI Appendix, Fig. S1D). We found that 91.5% (184/201) of stably recruited Mcm2-7^6N-650 led to establishment of a high E_FRET state (defined as two consecutive E_FRET measurements above a defined threshold, SI Appendix, Fig. S2A). A heat map of the E_FRET distribution vs. the time after Mcm2-7 recruitment revealed the timing of the E_FRET transition (Fig. 1D). Upon first Mcm2-7 recruitment, most molecules are in a low E_FRET state (peak at 0.33), but transition to a high E_FRET state (peak at 0.71) shortly afterward (Fig. 1D and SI Appendix, Fig. S2A). Kinetic analysis showed a 7.2 ± 1.2 s (median ± SE) time to form the high E_FRET state after Mcm2-7–Cdt1 recruitment (SI Appendix, Fig. S1E), substantially faster than the median time of MO formation after Mcm2-7–Cdt1 recruitment (38.3 ± 2.8 s; 25). Once formed, the median duration of the Orc6N–Mcm2-7N interaction was 26.4 ± 2.4 s, with a subset of the events showing much longer interactions (SI Appendix, Fig. S1F). The high E_FRET state, once formed, was almost always detected until ORC release. Indeed, in cases of successful recruitment of a second Mcm2-7 (Fig. 1C and SI Appendix, Fig. S1D), the end of the Orc6N–Mcm2-7N interaction typically occurred when ORC was released from the DNA. Because the MO complex dissociates before ORC release (25), we conclude that the Orc6N–Mcm2-7N interaction forms before and is lost after the MO complex ends.

Because Orc6N is tethered to the rest of ORC via an unstructured linker, we considered the possibility that the high E_FRET we observed reflected the general proximity of Orc6N to Mcm2-7 rather than a specific interaction with the Mcm2-7N-tier. To distinguish between these possibilities, we performed the same analysis except using Mcm2-7 labeled at its C-terminal tier via Mcm2 (Mcm2-7^2C-650, SI Appendix, Fig. S3 A and B). Using Mcm2-7^2C-650 and ORC labeled at the same site on Orc6 as above, we did not observe stable formation of a high E_FRET state (SI Appendix, Fig. S3C). Instead, a small drop in E_FRET values occurred at a similar time (~8 s) as the transition to the high E_FRET state seen when Mcm2-7 was labeled at its N-terminal domain (SI Appendix, Fig. S3D). This finding is consistent with binding of Orc6N to the opposite end of the helicase reducing an already weak E_FRET signal. We conclude that the elevated E_FRET observed in the FRET assay (Fig. 1B) reflects a stable interaction between the Orc6N and Mcm2-7N tier. For reasons that will become clear, we subsequently refer to this high-FRET state (Fig. 1D) as the Orc6 tether interaction and the assay presented in Fig. 1 as the Orc6-tether assay.

Mcm2N and Two Regions of Orc6 Contribute to the Tether Interaction.

Given the known interaction between the Orc6N and Mcm2N domains in the MO complex (25, 26), we asked whether Mcm2N region is involved in the interaction observed in our FRET assay. Although the MO structure is not resolved sufficiently to identify specific amino acid interactions (26), we identified a region of Mcm2 proximal to Orc6N in the MO complex (amino acids 348 to 365). We incorporated a mutant Mcm2 that scrambled these amino acids into Mcm2-7 labeled as above (Mcm2-7^2scr-6N-650) and tested its impact on the Orc6 tether interaction. Consistent with similar Orc6N and Mcm2N interactions observed in the MO complex mediating the tether, we found that this mutant eliminated the transition to the high E_FRET state (Fig. 2B, left panel).

We also investigated the involvement of regions of Orc6 in the tether interaction. Initially we tested ORC that lacked Orc6N (ORC^6∆N-550 in SI Appendix, Fig. S4A and Fig. 2 A, Middle left panel). Interestingly, upon stable recruitment of Mcm2-7^6N-650 by this ORC, we still observe a clear transition to a higher E_FRET state (Fig. 2 B, Middle left panel, and SI Appendix, Fig. S5A). The timing and stability of the high E_FRET state measured when using ORC^6∆N-550 were largely unaltered (Fig. 2 C and D), however, one notable kinetic difference was observed. ORC^6∆N-550 showed a clear reduction in longer-lived tether interactions compared to ORC^6N-550 (Fig. 2D, >40 s), consistent with the known defect in MO complex formation for this mutant (25, 26). In addition, there was a modest reduction in the percentage of molecules that reached the high E_FRET state [82.3% (102/124) for ORC^6∆N-550 vs. 91.5% for ORC^6N-550 (184/201), Fig. 2E], and the average high E_FRET value was reduced [high E_FRET state centered around 0.62 for ORC^6∆N-550 vs. 0.71 for ORC^6N-550 (SI Appendix, Fig. S2 A and B, Left)]. Together, these findings suggest that, although the Orc6N domain is involved in the tether interaction, additional regions of Orc6 are likely to contribute to the interaction.

We next investigated the role of the unstructured Orc6 linker using the Orc6-tether assay. We focused on a linker region (aa 120 to 185) previously identified as important for viability (Orc6 lnk1; 10). To eliminate effects due to altered linker length, we scrambled the Orc6 lnk1 region by altering the order but not the identity of its residues (SI Appendix, Fig. S4A, ORC^{6-lnk1scr-550}). Like the WT Orc6 lnk1, the scrambled sequence is predicted to be unstructured (SI Appendix, Fig. S4B).

Scrambling the Orc6 lnk1 region had a significant impact on the characteristics of Orc6 tether interactions (Fig. 2 B, Middle right panel, and SI Appendix, Fig. S5B). The median time for this mutant to form the interaction (i.e., the time to enter the high E_FRET state) is twofold longer than WT ORC (Fig. 2C, compare black to blue). Additionally, the median lifetime of the tether interactions with the ORC^{6-lnk1scr-550} mutant is more than twofold shorter (Fig. 2D). The latter finding is consistent with the frequent oscillations between high and low E_FRET observed in individual traces (SI Appendix, Fig. S5B) and the bimodal E_FRET values observed after ~20 s (Fig. 2 B, Middle right panel). Consistent with a weaker interaction, ORC^{6-lnk1scr-550} showed a reduction in the percentage of molecules attaining the high E_FRET state (Fig. 2E). We also tested two additional scramble mutations of the lnk1 regions and found that these had similar effects on the tether interaction (SI Appendix, Fig. S6), indicating that the original scramble mutant is not a gain of function mutant. These data suggest that the lnk1 region in Orc6 is important, but not solely responsible for the observed Orc6 tether interaction.

Because neither elimination of the Orc6N domain nor mutation of Orc6-lnk1 prevented the tether interaction, we tested the impact of combining these mutations (Fig. 2A, ORC^{6∆N-lnk1scr-550}). This mutant results in a dramatic loss of the high E_FRET signal (Fig. 2 B, Right panel, and SI Appendix, Fig. S5C). Of the 115 instances where Mcm2-7^6N-650 complexes are stably recruited by ORC^{6∆N-lnk1scr-550}, only 4.4% (5/115) transitioned to a high E_FRET state (compared to 91.5% of WT ORC^6N-550 molecules, Fig. 2E). In the rare instances that FRET is observed with ORC^{6∆N-lnk1scr-550}, the duration is much shorter than the WT or either single Orc6 mutant (Fig. 2 D, Inset). Together our data indicate that the Orc6N and Orc6-lnk1 regions work together to form the Orc6 tether interaction.

We asked whether any of the mutations that disrupt the tether interaction have a phenotype in vivo (SI Appendix, Fig. S4C). Previous studies showed that deletion of Orc6N is lethal to cells (10). Using a plasmid shuffle assay, we tested whether the mcm2-348-365scr, orc6-lnk1scr, or orc6-∆N,lnk1scr are similarly defective. When any of these alleles is the only copy of MCM2 or ORC6 present in vivo the associated cells are inviable (SI Appendix, Fig. S4C).

The Orc6 Tether Interaction Retains ORC After Cdt1 Release.

For one ORC to load both Mcm2-7 helicases without leaving the site of helicase loading, ORC must remain linked to the site of loading throughout its binding-site transition. After release of Cdt1, OM interactions are broken and the MO complex is formed rapidly (median 8.3 ± 2.0 s; (25)). The interaction between Orc6 and Mcm2-7N represents a possible mechanism to tether ORC to Mcm2-7 as it flips and changes binding sites. For such a mechanism to operate, the interaction would need to form before the OM interaction and the ORC–DNA interaction are disrupted and be retained during the transition to MO complex formation (Fig. 3A).

Because the OM interaction ends when Cdt1 is released from Mcm2-7, we first investigated when this potential tethering interaction occurred relative to Cdt1 release. Although comparison with previously measured Cdt1 release data (23) indicated that, on average, the Orc6N interacts with Mcm2-7N before Cdt1 release (SI Appendix, Fig. S7A), it was possible that this order only occurred in a fraction of loading events. To determine whether this order of events was consistently the case, we measured the times of these events at individual DNA molecules. To this end, we performed the Orc6-tether assay as in Figs. 1 and 2 but included Cdt1 labeled with the same, red-excited fluorophore (Cdt1^C-650) as Mcm6N (Fig. 3B). When Mcm2-7^6N-650-Cdt1^C-650 is recruited to DNA, we see an initial increase in red-excited fluorescence corresponding to two red dyes (Fig. 3B). Successful progression of the helicase-loading process results in the release of Cdt1^C-650 while Mcm2-7^6N-650 remains bound to DNA (23, 25), which can be observed as a halving of red-excited fluorescence (Fig. 3B). Importantly, the Cdt1^C-650 used in this experiment does not exhibit strong E_FRET with the ORC^6N-550 dye during helicase loading (SI Appendix, Fig. S7B).

Comparing the time of the first Cdt1 release to the time of the Orc6 interaction with Mcm2-7N revealed a clear order of events. When ORC^6N-550 recruited Mcm2-7^6N-650-Cdt1^C-650 a high E_FRET state consistently formed (152/162, Fig. 3C and SI Appendix, Fig. S7C). Of the 152 recruitment events that exhibited high E_FRET, all but one (151/152) showed that the Orc6 tether interaction formed before Cdt1 departure (median time 16.9 ± 1.7 s prior to Cdt1 release, Fig. 3D). Thus, Orc6 consistently interacts with Mcm2-7N before Cdt1 release and MO complex formation.

We expect that lack of the Orc6 interaction with Mcm2-7N should decrease ORC retention on the DNA after the loss of the OM interaction. To test this prediction, we compared the times of ORC retention after Cdt1 release for events that did or did not exhibit the tether interaction at the time of Cdt1 release (Fig. 3E). When the Orc6 tether interaction was present at the time of Cdt1 release, ORC was subsequently retained for a median time of 23.8 ± 8.8 s. In contrast, when the Orc6 tether interaction was absent at the time of Cdt1 release, ORC retention was much shorter (median time of 3.8 ± 3.4 s). Of the molecules that lacked the tether interaction at the time of Cdt1 release but exhibited longer ORC retention times (>10 s), more than half (5/9) formed the tether interaction within 10 s of Cdt1 release and this was true for all but one event that lasted more than 25 s.

We also asked how Orc6 mutants we tested in the tether assay impacted ORC DNA retention after Cdt1 release using the same assay. Both mutants that removed the Orc6 N-terminal domain showed strong defects in ORC retention (Fig. 3F). In each case, ~80% of molecules lacking Orc6N were released within 7.8 s (three frames) after Cdt1 release [ORC^6∆N-550, 80 ± 4% (85/106); ORC^{6ΔNlnk1scr550}, 83 ± 4% (88/106)] compared to only 39 ± 4% (58/147) for WT ORC^6N-550. Similarly, although 24 ± 4% (35/147) of WT ORC^6N-550 were retained for more than 100 s on DNA, either very few [ORC^6∆N-550, (1/106)] or none [ORC^{6ΔNlnk1scr550}, (0/106)] of the ORCs lacking Orc6N were retained for this period. These longer lasting events would include tethered ORC as well as ORC involved the MO complex and subsequent events. Mutating the linker region resulted in a hybrid set of defects. ORC^{6-lnk1scr-550} showed a significant increase in ORC molecules rapidly released after Cdt1 release relative to WT ORC^6N-550 (63 ± 5% (66/104) released after 7.8 s). Unlike the mutants lacking Orc6N, mutation of the Orc6 linker alone showed an intermediate level of long-lasting ORC retention events, with 13 ± 3% (13/104) of these molecules being retained for longer than 100 s. These data suggest that the Orc6N domain is critical for retaining ORC for longer periods and that the Orc6 lnk1 region facilitates the establishment of these interactions (Discussion). Together, our studies of ORC retention after Cdt1 release strongly support a model in which Orc6 establishes a tether interaction with Mcm2-7 before Cdt1 release that retains ORC at the site of helicase loading during its binding-site transition.

ORC Phosphorylation Interferes with the Orc6 Tether.

CDK phosphorylation of ORC inhibits helicase loading in vivo and in vitro (6, 10, 11, 14). Orc6 is one of two ORC subunits modified by CDK, and recent studies demonstrated that Orc6 phosphorylation fully inhibits MO-complex formation (11). Interestingly, the CDK modification sites on Orc6 are either in or proximal to the Orc6-lnk1 sequence (6), raising the possibility that ORC phosphorylation interferes with formation of the Orc6 tether interaction. To address this hypothesis, we performed the Orc6-tether assay using CDK-phosphorylated ORC^6N-550.

ORC phosphorylation (both Orc2 and Orc6) altered several characteristics of the tether interaction (Fig. 4 A and B). First, ORC phosphorylation delayed the median time of tether formation more than twofold relative to unphosphorylated ORC (Fig. 4C). More importantly, the lifetime of the phosphorylated ORC interaction is two and half times shorter than with unphosphorylated ORC (Fig. 4D). We also observed more frequent transitions between high and low E_FRET states compared to unphosphorylated ORC^6N-550 (Fig. 4 A, Right). E_FRET values for phosphorylated ORC^6N-550 split into a distinct high and low E_FRET states after Mcm2-7 recruitment with more data in the low E_FRET region (Fig. 4B). Consistent with a defect in tether formation, only 53.9% (77/143) of the Mcm2-7^6N-650 molecules recruited by phosphorylated ORC^6N-550 ever exhibited a stable high E_FRET state, compared to 91.5% (184/201) for unphosphorylated ORC (Fig. 4E).

To address the contribution of Orc6 vs. Orc2 phosphorylation, we repeated these experiments with ORC mutations that prevented phosphorylation of Orc6 (ORC^2phos-6N-550) or Orc2 (ORC^6phos-6N-550; 6). We focused on the impact of each subunit’s phosphorylation on the time to formation and duration of the Orc6 tether interaction. Only Orc6 phosphorylation significantly delayed tether formation (SI Appendix, Fig. S8 A–C). In contrast, phosphorylation of either Orc2 or Orc6 reduced the interaction lifetimes (SI Appendix, Fig. S8 D–F).

CDK phosphorylation of Orc6 occurs in or proximal to the Orc6 lnk1 region (6, 34). Given the overlapping targets, we compared the effects of Orc6 phosphorylation to the ORC^{6-lnk1scr-550} mutant. Interestingly, the distributions of initial times to tether formation are remarkably similar for the two ORCs (SI Appendix, Fig. S8G). The median lifetimes of the tether interactions obtained were also similar (12.0 ± 1.3 s vs. 9.7 ± 1.9 s, SI Appendix, Fig. S7F). However, when looking at the full distribution of interaction durations, we noticed a subpopulation of ORC^{6-lnk1scr-550} mutant events that exhibited long-lived tether interactions that were absent when Orc6 was phosphorylated (SI Appendix, Fig. S8H). This observation raises the possibility that a subset of ORC^{6-lnk1scr-550} molecules can form one or more downstream complexes (e.g., MO complex) in helicase loading.

Orc6 lnk1 Is Important for Mcm2-7 Double-Hexamer Formation.

To investigate the role of the Orc6 linker in the formation of complexes downstream of the OCCM, we performed a previously described single-molecule MO-complex-formation assay (Fig. 5A; 25). Briefly, ORC was labeled with a donor fluorophore on the Orc6 C-terminus, and Mcm2-7 is labeled with an acceptor fluorophore on the Mcm3 N terminus. When Mcm2-7 was initially recruited by ORC and forms the OCCM complex, these dyes are far apart and exhibit low E_FRET. When ORC flips over Mcm2-7 to form the MO complex, the donor fluorophore on the Orc6 C-terminus is brought into close proximity with the acceptor fluorophore on the Mcm3 N terminus resulting in a transition to a high E_FRET state (Fig. 5A).

Interestingly, although both ORC^6C-550 and ORC^{6C-lnk1scr-550} showed interactions between Orc6C and Mcm2-7N, the mutant complex showed strong defects in downstream events. Performing the MO formation assay using ORC^{6C-lnk1scr-550} revealed a modest defect in the percentage of Mcm2-7 recruitment events that result in MO complex formation (Fig. 5B). However, a striking phenotype was observed when later steps in the reaction were examined. For WT ORC^6C-550, the MO complex dissociates shortly after 2nd Mcm2-7 recruitment (Fig. 5C, second vertical blue-dashed line). In contrast, the ORC^{6C-lnk1scr-550} mutant is trapped in the MO complex for extended periods of time as indicated by the persistent high E_FRET signal (Fig. 5D). Although the MO complexes containing ORC^{6C-lnk1scr-550} repeatedly recruit 2nd Mcm2-7 complexes, the recruited helicases are rapidly released (Fig. 5D, red arrowheads). This results in a severe defect in DH formation compared to WT ORC (Fig. 5E). Consistent with this defect, ensemble helicase loading assays using ORC^{6C-lnk1scr-550} show a strong defect in salt-stable Mcm2-7 helicase loading (SI Appendix, Fig. S9A). These data indicate an important role for the Orc6 linker to stably recruit the second Mcm2-7 complex.

A Tether Function for Orc6 During Helicase Loading.

Based on our findings, we propose a model in which Orc6 performs a tether function during helicase loading (Fig. 6). Upon recruitment of Mcm2-7 to DNA-bound ORC-Cdc6, ORC embraces the first Mcm2-7 complex by first interacting with the Mcm2-7 C-terminal domains via the Orc1-5 C-tier along with Cdc6 (OM interactions). Shortly afterward, the Orc6-N terminal and lnk1 regions bind to the Mcm2-7N-terminal region, forming a “tether” between ORC and the first Mcm2-7 (Fig. 1). The interaction forms efficiently and rapidly once Mcm2-7 is recruited and is present well before Cdt1 release (Fig. 3). Because Cdt1 release occurs simultaneously with the breaking of OM interactions and anticipates the formation of the MO complex (25), this timing means the tether interaction is present prior to the ORC binding-site switch. In successful double-hexamer formation events, the Orc6 tether-associated FRET interactions last until after recruitment of a second Mcm2-7 complex (and therefore throughout MO complex formation) (Figs. 1 and 3). This interaction keeps ORC bound to the site of helicase loading as it flips over the 1st Mcm2-7 helicase to form new interactions with the N-terminal domains of Mcm2-7 and the inverted B2 DNA binding site.

The rapid and nearly quantitative formation of this tethering interaction upon first Mcm2-7 recruitment has interesting implications for the mechanism of helicase loading. These findings strongly suggest that in most cases (>90% in our experiments) the first approach taken to form a double hexamer is a single-ORC mechanism. Because the tether interaction forms between ORC and Mcm2-7 that are already bound to one another, an intramolecular event, even high concentrations of ORC would be unlikely to change this percentage. Multiple aspects of origin structure and helicase loading would prevent the involvement of a second ORC at this point. First, very few origins are arranged such that the ACS and B2 would be spaced appropriately to allow a second ORC to interact with the B2 element and the N terminus of the first Mcm2-7 (14). Second, because the tethering interaction involves the same regions of the Orc6 and Mcm2 involved in the MO interaction (Fig. 2), the Orc6 tether interaction would interfere with a second ORC forming the MO complex. The rapidity of formation and intramolecular nature of the tether interactions makes it unlikely that a second ORC would form an MO interaction before the Orc6 tether interaction.

Despite the likelihood that a one-ORC mechanism is the first approach used for origin licensing, this does not mean that it is the only mechanism used. Our single-molecule studies have also observed two-ORC reactions, albeit at a lower rate than the one-ORC mechanism (25). A likely explanation for two-ORC events is the frequent release of ORC as it attempts to make its binding-site transition. Even when a tether interaction was present as Cdt1 was released, ~25% of ORC molecules released within one video frame (~2.6 s) after Cdt1 release (28/114, Fig. 3 E, Inset). When the tether is not present as Cdt1 is released, approximately half of ORC molecules are released in the same time frame (16/33, Fig. 3 E, Inset). Therefore, we see ORC release at this stage as a common failure point in the one-ORC mechanism. The release of the first ORC provides an opportunity for a second ORC to form the MO complex and complete helicase loading (Fig. 6, two-ORC mechanism). We note, however, that the time for a second ORC to form the MO complex is limited. If the first ORC fails to form the MO complex, it will also fail to hold the first Mcm2-7 in a stable closed-ring state (11, 26). A single Mcm2-7 alone on DNA is an unstable state, having a median lifetime of 11.2 ± 2.0 s (25). Release of the first Mcm2-7 prior to second ORC binding would reset the licensing process (Fig. 6, reinitiation pathway). Thus, after first ORC release, a race between a second ORC binding and Mcm2-7 release would determine the extent of use of the two-ORC pathway. Thus, as ORC concentration increases so would the likelihood of a two-ORC pathway being used and it is very likely that both pathways are used in cells.

Why have electron microscopy studies of the OCCM complex structure (24) not revealed an interaction between Orc6N and Mcm2-7N? The OCCM structure was trapped using a slowly hydrolyzable ATP analog, ATPγS, to prevent Cdc6 departure and downstream steps (e.g., Mcm2-7 ATP hydrolysis which is required for subsequent translocation/sliding; 17, 35). In contrast, all reactions in the present study were performed with ATP. More recent cryo-EM studies using ATP but incorporating ATPase mutations in Mcm2 or Mcm5 also did not detect the tether (35). The discrepancy between the electron microscopy and FRET results suggests that an Mcm2-7 ATP-dependent step is required to form the Orc6 tether interaction. We attempted to perform the tether assay with ATPγS, however, we were unable to obtain results due to poor ORC–DNA binding under these conditions. Although not detected by electron microscopy, crosslinking-mass spectrometry conducted in the presence of ATPγS (24), identified a crosslink between the Orc6 lnk1 sequence and the N-terminal A domain of Mcm2 close to the site mutated in Mcm2-7^2scr-6N-650 (Mcm2 338 vs. Mcm2 348-365 in the mutant). Thus, the beginning of the interaction between Orc6N and Mcm2-7N may be occurring in the OCCM, although the longer-lasting interaction that we observe is not.

Two Regions of Orc6 Contribute to Tether Interaction.

The data presented in this study reveal a role for both the Orc6N domain and the Orc6 lnk1 sequence in forming the Orc6 tether interaction. The Orc6N domain binds to Mcm2N in the MO structure (26) and deletion of the Orc6N domain prevents MO complex formation (25, 26). It is, therefore, most likely that the Orc6N domain interacts with the Mcm2-7N during tethering using the same or a similar interface. Strongly supporting this conclusion, a mutant in Mcm2 predicted to disrupt the interaction of Orc6N and Mcm2N in the MO complex completely disrupts Orc6 tether formation (Fig. 2B). How the Orc6 linker contributes to the interaction observed between Orc6N and the first Mcm2-7N is less clear. This region of the Orc6 protein has not been resolved in any ORC-containing structures to date. Previous coimmunoprecipitation studies showed that the Orc6N + lnk1 (aa 1 to 185) region binds Cdt1 (10), suggesting that Orc6 lnk1 contributes to tether interactions by binding Cdt1. Such an interaction raises the interesting possibility that Cdt1 release stimulates ORC reorientation during the flipping process by changing its interactions. Consistent with this idea, we found that once released from the ACS, ORC exhibits short ACS rebinding events only before Cdt1 release but never after (17).

Although mutations in Orc6 lnk1 led to a longer time to form Orc6 tether interactions as well as shorter duration of these interactions (Fig. 2 C and D), these mutations did not eliminate long periods of ORC retention after Cdt1 release (Fig. 3F) or MO FRET formation (Fig. 5B). In contrast, deletion of Orc6N dramatically reduces ORC retention at the DNA after Cdt1 release (Fig. 3F) and prevents stable MO-complex formation (25, 26). These findings suggest that Orc6 lnk1 interaction helps to position Orc6 N terminus for tether formation but is not essential for tethering ORC during its binding-site transition. This model is supported by the observation that, although mutation of Orc6 lnk1 leads to shorter interactions for the majority of ORC molecules, a subpopulation of Orc6 lnk1-scr molecules form long-lived tether interactions (see bimodal curves in Figs. 2D and 3F).

It has been proposed that the Orc2 IDR could have a role in tethering ORC to the site of helicase loading (36). This conclusion is based on the finding that the first 200 amino acids of Orc2 IDR interact with purified OCCM complexes (36). We show that deletion of Orc6N alone leads to rapid release of ORC after Cdt1 release (Fig. 3F), suggesting that Orc2 IDR is not sufficient to retain ORC during the binding-site switch. We propose that a role of the Orc2 IDR in tethering could support but not substitute for the Orc6 tether interaction and likely plays a more prominent role in stabilizing the MO complex (14, 36).

Our findings also reveal that the Orc6 linker functions after MO complex formation. ORC can recruit the first Mcm2-7 and form an OCCM complex without Orc6 (37). Although current models based on structural studies suggest that recruitment of the second Mcm2-7 helicase uses the same interactions as the first (4, 5), a role for Orc6 in stable recruitment of the second Mcm2-7 suggests otherwise. How could Orc6 be involved in recruitment of the second but not the first Mcm2-7? Although we did not detect any difference in the high E_FRET states adopted by WT ORC^6N-550 and ORC^{6C-lnk1scr-550} in our MO assay (SI Appendix, Fig. S9B), one possibility is that the ORC^6-lnk1scr mutant does not form a proper MO complex. Alternatively, the Orc6 lnk1 region may form an uncharacterized interaction with the second Mcm2-7–Cdt1 complex that facilitates its stable loading. Such an interaction could be related to the interaction we observe between the Orc6 lnk1 region and the first Mcm2-7–Cdt1 complex (ORC^{6-lnk1scr-550}, Fig. 2).

CDK Phosphorylation of ORC Inhibits Orc6 Tether Interactions.

Previous studies have shown that phosphorylation of ORC inhibits MO complex formation (11, 14). Our data suggest that this effect is at least partly due to inhibition of tether formation (Fig. 4). Phosphorylation of Orc6 alone, but not Orc2 alone, decreases the rate of formation of this interaction (SI Appendix, Fig. S8). Based on these data, as well as the fact that Orc6 phosphorylation sites are either in or proximal to the Orc6 lnk1 sequence (6), we propose that Orc6 phosphorylation directly inhibits tether formation. Phosphorylation of either Orc6 or Orc2 reduces the duration of Orc6N–Mcm2-7N interactions (SI Appendix, Fig. S8). Again, the proximity of the Orc6 phosphorylation sites to Orc6N suggests that modification of Orc6 directly interferes with binding of the Orc6N to Mcm2-7N, reducing its duration. The mechanism of Orc2 phosphorylation reducing Orc6 tether duration is less clear but is likely related to interactions between the Orc2 IDR and Mcm2-7 in either the OCCM (36) or the MO complex (14, 36) helping to stabilize ORC:Mcm2-7 interactions.

Implications for Origin Licensing in Other Organisms.

A major difference between replication initiation in budding yeast and most other eukaryotic organisms is the lack of specific sequences driving helicase loading in most organisms. The one-ORC mechanism seems particularly useful in a situation in which there are no high affinity ORC binding sites on DNA since this would reduce the need for recruitment of two ORC molecules to low-affinity binding sites. Consistent with a similar mechanism being used, recent studies of human origin licensing showed that a similar MO complex can be detected (38, 39). Nevertheless, a key element of the one-ORC mechanism is changed. Orc6 in human cells (HsOrc6) is not robustly bound to the remaining ORC subunits and is not required for all helicase loading (although it improves efficiency of loading; 38–40). In addition, the location and length of the linker region is changed. Nevertheless, mutations in either the N-terminal region or C-terminal helix of HsOrc6 that lead to Meier-Gorlin syndrome have a similar defect in helicase loading to reactions lacking ORC6 (38), suggesting that Orc6 is important for proper DNA replication and cellular function, potentially through a one-ORC, MO-dependent mechanism. Regardless, it will require more direct experiments to resolve whether a one-ORC mechanism can operate in human cells and it is certain that it will not be the only mechanism that contributes to origin licensing, as reactions without Orc6 are possible.

Growth of Cells for Protein Purification.

Growth of cells for protein purification was performed as described (11, 41). In brief, S. cerevisiae strains were grown to OD₆₀₀ = 1.0 in 6L of YEP supplemented with 2% glycerol (w/v) at 30 °C. Cells were arrested in G1 using α-factor (100 ng/mL) and expression of proteins was induced using 2% galactose (w/v). After induction, cells were harvested and sequentially washed with 100 mL wash buffer and 50 mL lysis buffer. The washed pellet was resuspended in approximately 1/3 of packed cell volume with lysis buffer containing cOmplete Protease Inhibitor Cocktail Tablet (1 tablet per 25 mL total volume; Roche) and frozen dropwise in liquid nitrogen. Frozen cells were lysed in a SamplePrep freezermill (SPEX) and the powder was saved at −80 °C. Upon thawing of the cell powder, the lysate was clarified by ultracentrifugation in a Type 45Ti rotor at 36 krpm (150,000×g) for 1 h at 4 °C. This lysate was then used for further steps of protein purification as described below.

Preparation of Unlabeled Proteins.

Wild-type Mcm2-7, Cdc6, and Cdt1 were purified as described previously (11, 41). Clb5-Cdk1 was purified from ySK119 and Sic1 was purified from BL21-DE3-Rosetta bacteria transformed with the plasmid pGEX-Sic1 as described previously (42).

Preparation of Labeled ORC^6N-550, ORC^{6-lnk1scr-550}.

For ORC^6N-550 and ORC^{6-lnk1scr-550}, an S6 tag (GDSLSWLLRLLN) along with 3× GGS flexible linker on either side of the tag was inserted after amino acid 107 in Orc6. Dy550-CoA dyes were synthesized as described in ref. 43 using maleimide-Dylight 550 (Thermo Scientific) and Coenzyme A trilithium salt (Research Products International). ORC was purified from lysate using FLAG affinity resin (Sigma-Aldrich) as previously described (25). ORC, SFP synthase (NEB), and Dy550-CoA were incubated in a 1:4:10 ratio for 45 min at room temperature, then at 4 °C overnight. Labeled ORC was then purified on a Superdex 200 Increase 10/300 gel filtration column. Peak fractions were aliquoted and stored at −80 °C.

Preparation of Labeled ORC^6∆N-550, ORC^{6∆N-lnk1scr-550}, ORC^6C-550.

For sortase-mediated labeling, peptides (NH2-CHHHHHHHHHLPETGG-COOH for N-terminal labeling, NH2-GGGHHHHHHHHHHC-COOH for C-terminal labeling) were labeled with maleimide-derivatized Dylight 550 or Dylight 650 and HPLC purified.

ORC was purified from lysate using FLAG affinity resin (Sigma-Aldrich) as previously described (25). Sortase-mediated labeling was performed by incubating purified protein with equimolar amount of Sortase along with 100 nmol of either the labeled N- or C-peptide in buffer supplemented with 5 mM CaCl₂ for 15 min at room temperature. After incubation, the reaction was quenched with 15 mM EDTA and applied to a Superdex 200 Increase 10/300 gel filtration column to remove Sortase and unreacted peptide-dye. Peak protein fractions were incubated with 0.3 mL of Ni-NTA Agarose Resin (Qiagen) in buffer supplemented with 5 mM imidazole overnight at 4 °C. The resin was collected and then washed first with 5 mL buffer supplemented with 15 mM imidazole, then washed with 5 mL buffer supplemented with 25 mM imidazole. Labeled protein was then eluted in buffer supplemented with 300 mM imidazole. Peak fractions were aliquoted and stored at −80 °C.

Preparation of Labeled Mcm2-7^6N-650,Mcm2-7^2scr-6N-650, and Mcm2-7^2C-650.

Mcm2-7 protein was purified using FLAG resin and labeled via Sortase-mediated conjugation as described above. For Mcm2-7^6N-650 and Mcm2-7^2scr-6N-650, the N-terminal extension of Mcm6 (2 to 103) (which is not important for helicase loading, 27) was deleted and replaced with an N-terminal Sortase recognition sequence Ubiquitin-GGG. During expression in cells, the ubiquitin is cleaved from the protein, and after Sortase labeling the N-terminal Sortase peptide coupled to Dylight-650 is attached to the N terminus of Mcm6∆103. For Mcm2-7^2C-650, a C-terminal Sortase recognition sequence (LPETGG) was placed at the C-terminus of Mcm2. During Sortase labeling, the C-terminal Sortase peptide coupled to Dylight-650 is attached to the C-terminus of Mcm2.

Preparation of Labeled Mcm2-7^3N-650 and Cdt1^C-650.

Mcm2-7^3N-650 and Cdt1^C-650 were purified as described (23, 25).

Single-Molecule Assay for Helicase Loading.

A micromirror total internal reflection microscope was used to perform multiwavelength single-molecule imaging (31). Glass slides were functionalized with Biotin-PEG and PEG, and biotinylated DNA along with streptavidin-coated fiducial markers (0.04 micron, ThermoFisher TransFluoSpheres T10711) were coupled to the slide as described (25). All reactions were performed as described previously (25) in buffer containing 25 mM HEPES-KOH pH 7.6, 300 mM 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) as well as 2 mM Trolox (44). Reactions used 0.5 to 1 nM ORC, 3 to 10 nM Cdc6, and 10 to 15 nM Mcm2-7-Cdt1. All reactions also included 0.5 µM of a 60 bp nonspecific DNA generated by annealing the following two oligonucleotides:

(5′-CTTGTTATTT TACAGATTTT CTCCATTCTT CTTTTATGCT TGCAAAACAA AAGGCCTGCA-3′) and (5′-TGCAGGCCTT TTGTTTTGCA AGCATAAAAG AAGAATGGAG AAAATCTGTA AAATAACAAG-3′).

Positions of DNA molecules labeled with Alexa Fluor 488 were located before experiments using 488 nM excitation. Experimental acquisition was performed using 1 s alternating 532 nm and 633 nm excitation, with approximately 0.2 s of dead time between acquisitions.

Single-Molecule Data analysis.

CoSMoS datasets were analyzed as described previously (23) and fluorescence intensity values were background corrected as described (45). Details of FRET data analysis can be found in SI Appendix.

Ensemble Helicase Loading Assays.

Ensemble helicase loading assays (SI Appendix, Figs. S1B and S9) were performed as described previously (41).

Plasmid Shuffle Assays.

Plasmid shuffle assays to test viability of orc6 and mcm2 mutants (SI Appendix, Fig. S4C) were performed essentially as described previously (46). The yeast strains used for testing were ySC136 (orc6 mutants) and ASY1055.1 (mcm2 mutants).