To test the idea that PP2A^Cdc55 targets Cdc6, we examined Cdc6 protein levels in Δclb2 and Δcdc55 cells by western blot analysis. The Cdc6 protein level was lower in Δclb2 compared to wild type in log-phase cells, which supports previous results of Cdc6 destabilization in the Cdc6-lxf mutant that does not bind to Clb2 (Figure 1B, C, Örd et al., 2019). There was low Cdc6 protein expression in G1 arrested cells, which was not impacted by deletion of CLB2 or CDC55 (Figure 1B, middle). During mitosis, we observed less Cdc6 in Δclb2 cells and more in Δcdc55 cells compared to wild type; this difference in the Cdc6 protein level was more prominent in mitosis than in log-phase cells (Figure 1B, C). These results indicate that PP2A^Cdc55 and Clb2 regulate Cdc6 protein levels in a contradictory manner. An elevated Clb2 protein level and Cdc6 stabilization in wild-type cells confirmed that cells were in mitosis (Figure 1B, right). The S-phase cyclin Clb5, as well as another regulatory subunit of PP2A, Rts1, did not alter Cdc6 protein levels (Figure 1C). We conclude that Cdc6 is regulated by mitotic cyclin Clb2 and PP2A^Cdc55 in a specific manner.

Next, we analyzed Cdc6 protein levels in Δclb2 and cdc55-101 cells during the cell cycle. The cdc55-101 mutant excludes Cdc55 from the nucleus, allowing us to study Cdc55 nuclear function. Unlike Δcdc55, cdc55-101 mutant cells do not exhibit morphological defects, which allows us to monitor G1 arrest induced by α-factor and cell cycle progression (Sasaki et al., 2000). In wild-type cells, Cdc6 protein levels were suppressed throughout S phase, increased in mitosis at 80 min and then diminished (Figure 1D, E; Drury et al., 2000; Drury et al., 1997; Piatti et al., 1995). In Δclb2 cells, Cdc6 protein levels were suppressed throughout S phase and particularly in mitosis (Figure 1D–F). This result suggests a role for Clb2 in Cdc6 stabilization during mitosis possibly by masking the Cdc6 phospho-degron. In contrast, Cdc6 was more stabilized in cdc55-101 cells during mitosis, with a peak accumulation between 60 and 100 min (Figure 1D, E). Flow cytometry analysis did not show a significant difference in the cell cycle profile between wild-type and Δclb2 cells (Figure 1F). In cdc55-101 cells, we observed a 20-min S-phase delay, indicating that PP2A^Cdc55 positively regulates S-phase progression (Figure 1F). We conclude that the early Cdc6 accumulation in cdc55-101 cells is not due to cell cycle delay.

To examine how PP2A^Cdc55 impacts Cdc6 localization and protein level, we analyzed Cdc6-Citrine by time-lapse microscopy in wild-type and Δcdc55 cells. Whi5 is a Start transcriptional repressor imported into the nucleus in M/G1 and exported in late G1. Whi5 nuclear import was used as a cell cycle indicator that was set as time 0 for a temporal framework (Figure 1G; Doncic et al., 2011). Cdc6-Citrine began accumulating in the nucleus after Whi5 nuclear import and reached maximum accumulation at 9 min after Whi5 nuclear import in wild-type cells (Figure 1H). In Δcdc55 and cdc55-101 cells, we observed more Cdc6-Citrine accumulation (Figure 1H). The timing of Cdc6-Citrine expression and degradation remained unchanged, as it started dropping around 15 min after Whi5 nuclear import in all strains (Figure 1H). These data suggest that PP2A^Cdc55 negatively regulates Cdc6 stability in the nucleus during mitosis. We also observed Cdc6-GFP nuclear localization in early mitosis in asynchronous cdc55-101 cells (Figure 1I, arrow). The similar Cdc6 protein accumulation patterns observed in Δcdc55 and cdc55-101 support the idea that Cdc6 is solely regulated by nuclear PP2A^Cdc55 function.

Clb2 binds to the Cdc6 N-terminus through a conserved LxF motif (Örd et al., 2019). We tested if low Cdc6 protein levels in Δclb2 cells are mediated through the Cdc6 N-terminus. The CDC6-T368A mutant contains an alanine substitution in the C-terminal phospho-degron, stabilizing Cdc6 in mitosis (Figure 2A, lanes 4 + 5, Figure 2B; Perkins et al., 2001). This mutant retains the N-terminal Cdc6 phospho-degron at T39-S43. Cdc6-T368A was unstable in Δclb2 log-phase cells (Figure 2A, lanes 2 + 3) and was more prominent in mitotic-arrested cells, supporting the idea that Clb2 shields the Cdc6 N-terminal phospho-degron (Figure 2A, lanes 5 + 6, Figure 2B; Örd et al., 2019). The CDC6-T39A-T368A double mutant, containing mutations at both the N- and C-terminal phospho-degron, was synthetically lethal in Δclb2 cells (Figure 2C). When Cdc6 is overexpressed and controlled under the GAL promoter, we observed stabilized Cdc6-T39A-T368A in Δclb2 cells during mitosis, suggesting that stabilized Cdc6 is lethal when Clb2’s inhibitory role in origin licensing is abolished (Figure 2—figure supplement 1). In contrast, GAL-Cdc6 and GAL-Cdc6-T368A were degraded in Δclb2 cells, which suggest the importance of the N-terminal phospho-degron in Cdc6 degradation when Clb2 is abolished (Figure 2—figure supplement 1).

Cdc6 stabilization in Δcdc55 cells indicates that PP2A^Cdc55 inhibits Cdc6 protein accumulation. Therefore, we excluded the possibility that PP2A^Cdc55 targets neither Cdc6 phospho-degron. We hypothesized that PP2A^Cdc55 releases Clb2 from Cdc6, exposing the Cdc6 phospho-degrons, which subsequently leads to Cdc6 degradation. To test this possibility, we compared the Clb2:Cdc6 protein-binding ratio between wild-type and Δcdc55 cells during mitosis by co-immunoprecipitation (co-IP). Cdc6 was stabilized in Δcdc55 cells, confirming the results in Figure 1B (Figure 2D, lanes 3 + 6, Figure 2E). There was more Clb2 bound to Cdc6 in Δcdc55 cells compared to wild-type cells (Figure 2D, lanes 3 + 6). The relative ratio between Clb2 to Cdc6 was reproducibly higher in Δcdc55 cells, suggesting that PP2A^Cdc55 disrupts Cdc6–Clb2 binding (Figure 2E). From this result, we conclude that Clb2 binds tighter to a fraction of Cdc6 in Δcdc55 cells, which may mask the Cdc6 N-terminal phospho-degron, and thereby contribute to Cdc6 stability in Δcdc55 cells.

Our finding supports a scenario in which PP2A^Cdc55 dephosphorylates Cdc6 at the N-terminus and disrupts Cdc6–Clb2 binding. We hypothesized that PP2A^Cdc55 dephosphorylates Cdc6 at residue T7, which removes the Cks1 docking site, thus inhibiting Clb2 binding. We generated two CDC6-9MYC constructs that contain alanine mutations at the Cdk1 phosphorylation sites to test if Cdc6-T7 site is dephosphorylated by PP2A^Cdc55. The CDC6-6A mutant strain contains six alanine mutations at T7A, T23A, T39A, S43A, T368A, and S372A, serving as a negative control. The CDC6-T7 mutant strain only retains the T7 phosphorylation site while all other Cdk1 phosphorylation sites are mutated to alanine (T23A, T39A, S43A, T368A, and S372A) (Figure 3A). Thus, the CDC6 constructs eliminate the function of both phospho-degrons. We raised a custom Cdc6-T7 phospho-specific antibody to detect T7 phosphorylation (Cdc6-T7p). Cdc6-T7 phosphorylation was not detected in the CDC6-6A mutant, confirming the specificity of the custom-made Cdc6-T7 phospho-specific antibody (Figure 3B). The Cdc6-6A protein level did not differ between CDC55 and Δcdc55 cells (Figure 3B). Cdc6-T7 phosphorylation was detected in the Cdc6-T7 construct and was slightly enhanced in Δcdc55 cells, supporting the idea that PP2A^Cdc55 dephosphorylates the Cdc6-T7 site (Figure 3B, C). Cdc6-6A did not interact with Clb2, supporting a role for Cdc6 phosphorylation in protein binding (Figure 3B). Cdc6-T7 associated with Clb2, supporting our previous results that Cks1 docking to T7 is crucial for the Cdc6–Clb2 interaction (Figure 3B). The Cdc6-T7 and Clb2 interaction was enhanced in Δcdc55 cells (Figure 3B, C).

The CDC6-T7 mutant contains mutations on phospho-degrons. Why is the Cdc6-T7 protein level slightly stabilized in the absence of Cdc55? (Figure 3B). There is a possibility that the T7 site serves as a phospho-degron, because Clb2 may bind and mask phosphorylated Cdc6-T7 in Δcdc55. Interestingly, the Cdc6-6A protein level was noticeably less stable than Cdc6-T7 (Figure 3B). In order to understand the mechanism of these observations, we examined endogenous Cdc6, and the mutants Cdc6-6A and Cdc6-T7 protein levels in cdc4-1 cells (Figure 3—figure supplement 1). Endogenous Cdc6 was stabilized after 1–2 hr at 37°C as reported previously (Drury et al., 1997; Figure 3—figure supplement 1). In contrast, Cdc6-6A was still degraded in cdc4-1 cells, indicating that Cdc6 has another degradation mechanism rather than SCF-mediated degradation. The Cdc6-T7 protein level stayed the same in cdc4-1 cells, which indicates the SCF^Cdc4 does not target T7 for degradation (Figure 3—figure supplement 1).

Next, we examined Cdc6-T7 phosphorylation during the cell cycle by α-factor block and release. Cdc6 was expressed during mitosis at 80 minutes in wild type and was more stabilized in cdc55-101 cells (Figures 3D and 1D). Cdc6-T7 phosphorylation was suppressed in wild-type cells (Figure 3D) throughout the cell cycle. In contrast, there was a spike of T7 phosphorylation in cdc55-101 mutant cells at 80 minutes during mitosis (Figure 3D). The Cdc6-T7 phosphorylation peak coincides with Cdc6 stabilization (Figure 3D, E). Flow cytometry analysis showed that both wild-type and cdc55-101 mutant cells are in mitosis at 80 min (Figure 3—figure supplement 2). From these results, we conclude that Cdc6-T7 phosphorylation is reversed by PP2A^Cdc55, which releases Clb2.

It has been reported that Cdc14 targets Cdc6 and other pre-RC proteins (Zhai et al., 2010). Consequently, we examined Cdc6 protein levels in cdc14-3 mutant cells. Cdc14 was inactivated at the nonpermissive temperature upon α-factor release. We show that Cdc6 accumulation was consistently lower in cdc14-3 compared to cdc15-1, an alternative mitotic mutant (Figure 4—figure supplement 1A). This result was especially prominent during mitosis at 120–180 minutes. The cell cycle profiles between cdc14-3 and cdc15-1 were the same, confirming that the observed difference in Cdc6 protein levels was not caused by cell cycle progression defects (Figure 4—figure supplement 1B). From these data, we conclude that Cdc14 stabilizes Cdc6 during late mitosis.

PP2A^Cdc55 and Cdc14 exhibited opposite effects on Cdc6 stability (Figure 1 and Figure 4—figure supplement 1A). We hypothesized that Cdc14 may target a Cdc6 phospho-degron, leading to Cdc6 stabilization. PP2A^Cdc55 and Cdc14 specificity toward Cdc6 was examined by in vitro phosphatase assays. Recombinant Cdc6 was preincubated with Clb5–Cdk1 in the presence of [γ-³²P]-ATP. The reaction mixture was further incubated with purified Cdc14 or PP2A^Cdc55 for dephosphorylation. Purified Sic1 was also added to the mixture in order to inhibit further phosphorylation of Cdc6 by Clb5–Cdk1 after addition of the phosphatases. The intensity of [³²P]-label on Cdc6 was measured by autoradiography. Full-length wild-type Cdc6 was moderately dephosphorylated by both PP2A^Cdc55 and Cdc14 (Figure 4A), indicating that Cdc6 is directly targeted by these phosphatases. The phosphate on Cdc6-T7 was efficiently removed by PP2A^Cdc55 when T7 was the only phosphorylation site available, but not by Cdc14 (Figure 4B). The same trend was observed with Cdc6-T23, but to a lesser extent (Figure 4C). When the Cdc6 C-terminal phospho-degron was the only phosphorylation site available, it was dephosphorylated by Cdc14, but not by PP2A^Cdc55 (Figure 4D). N-Swe1 and Csa1 were used as positive controls to show PP2A^Cdc55 Harvey et al., 2011 and Cdc14 specificity (Örd et al., 2020), respectively (Figure 4E, F). Neither Cdc6-T39 nor -S43 was dephosphorylated by PP2A^Cdc55 or Cdc14, indicating that the Cdc6 N-terminal phospho-degron is not controlled by these phosphatases (Figure 4—figure supplement 2). These results suggest that PP2A^Cdc55 directly dephosphorylates Cdc6-T7 and -T23, while Cdc14 dephosphorylates the Cdc6 C-terminal phospho-degron at T368-S372. Thus, PP2A^Cdc55 and Cdc14 showed sequence specificity toward Cdc6.

Deletion of CDC55 slows down S-phase progression. However, cells were still capable of completing S phase (Figure 1F). The key regulator of Clb2–Cdk1 inhibition, Sic1, is a promoter of origin licensing in G1 in conjunction with APC^Cdh1 (Lengronne and Schwob, 2002; Wäsch and Cross, 2002). It is possible that Sic1 also drives origin licensing through Cdc6. Both Cdh1 and Sic1 act in overlapping mechanisms that promote mitotic exit, which is potentially coupled to origin licensing (Visintin et al., 1998). We hypothesized that this is due to high level of Clb2 binding to Cdc6. A CDC6-T39A-T368A Δcdh1 strain was used to obtain stable Cdc6 expression. Transient Sic1 expression was induced in a GAL-SIC1 strain using galactose. Overexpression of Sic1 released Clb2 from Cdc6 in mitotic cells arrested by nocodazole (Figure 5A, B).

Next, we tested if Sic1 promotes Mcm2–7 DNA loading using a biochemical approach with purified proteins (Ayuda-Durán et al., 2014; Remus et al., 2009; Seki and Diffley, 2000). The stringent regulation of intracellular Cdc6 protein levels by proteolytic degradation in both G1 and S phase severely hampers the purification of endogenous Cdc6 from these cell cycle stages even after galactose-induced overexpression. However, we observed that extended overexpression of TAP^TCP-tagged Cdc6 upon release from α-factor arrest induces a mitotic arrest phenotype in yeast cells, characterized by a 2C DNA content and the presence of a large elongated bud (data not shown). Intriguingly, Cdc6 isolated from whole-cell extracts of such mitotically arrested cells exists in a stable stoichiometric complex with three additional proteins that we identified as Clb2, Cdk1, and Cks1 by mass spectrometry (Figure 5C). The formation of this complex after Cdc6 overexpression is consistent with the previous observation of a Cdc6·Clb2·Cdk1 complex in G2/M cells (Mimura et al., 2004). Importantly, using gel-filtration analysis, we find that Cdc6 is released from the Clb2–Cdk1 complex in the presence of the Cdk1 inhibitor Sic1, which itself forms a stable complex with Clb2–Cdk1 (Figure 5D).

Clb2–Cdk1 binding to Cdc6 has been proposed to inhibit the origin licensing activity of Cdc6 as part of cellular re-replication control mechanisms (Mimura et al., 2004). We tested the ability of Cdc6·Clb2·Cdk1·Cks1 to promote Mcm2–7 loading onto DNA in conjunction with purified ORC and Cdt1·Mcm2–7, analogous to experiments described previously with purified Cdc6⁵². Using this approach, we find that Cdc6·Clb2·Cdk1·Cks1 does not support the formation of salt-stable MCM complexes on DNA, indicating that Cdc6 in complex with Clb2·Cdk1·Cks1 is inactive for Mcm2–7 loading (Figure 5E, lanes 1 + 2). Importantly, normal Mcm2–7 loading was observed in reactions containing both purified Sic1 and Cdc6·Clb2·Cdk1·Cks1 (Figure 5E, lanes 3 + 4), indicating that Clb2·Cdk1·Cks1 sequesters Cdc6 and that this form of Cdc6 inhibition can be reversed by release of Cdc6 from the Cdc6·Clb2·Cdk1·Cks1 complex by Sic1. However, we noted that Orc6 is efficiently phosphorylated in the presence of Cdc6·Clb2·Cdk1·Cks1, as indicated by the pronounced gel-mobility shift of Orc6 (Figure 5E, compare lanes 1 + 3). This Orc6 phosphorylation is inhibited in the presence of Sic1, indicating that Clb2–Cdk1 retains partial kinase activity in the Cdc6·Clb2·Cdk1·Cks1 complex. It is known that Cdk1 phosphorylation inhibits the Mcm2–7 loading function of ORC (Frigola et al., 2013). It was possible that the inhibition of Mcm2–7 loading in the presence of Cdc6·Clb2·Cdk1·Cks1 was due to ORC phosphorylation and not Cdc6 sequestration. However, we noted that not only Mcm2–7 loading, but even Mcm2–7 recruitment to ORC/Cdc6, which can be monitored in the presence of ATPγS, is inhibited in the presence of Cdc6·Clb2·Cdk1·Cks1 (Figure 5E, lanes 5 + 6), and this inhibition is again reversed in the presence of Sic1 (Figure 5E, lanes 7 + 8). Since ORC phosphorylation specifically inhibits Mcm2–7 loading but not Mcm2–7 recruitment (Frigola et al., 2013), these data demonstrate that the mechanism of licensing inhibition by Cdc6·Clb2·Cdk1·Cks1 complex formation is distinct from that by Cdk1 phosphorylation of ORC. In summary, we conclude that Clb2–Cdk1 inhibits Cdc6 by physical sequestration and this inhibition is reversible by the Sic1-mediated release of Cdc6 from the Cdc6·Clb2·Cdk1·Cks1 complex.

The necessary structural elements in Cdc6 responsible for protein binding are not well understood. Protein-binding motifs often fall in IDRs, which are not well characterized structurally due to the limitations of crystal structure analysis (Wright and Dyson, 2015; Fuxreiter et al., 2007). IDRs in Cdc6 were identified by several programs including IUPred (Mészáros et al., 2018) and Anchor (Mészáros et al., 2009), which include: (1) Cdc6-N terminus residues 1–59 and (2) Cdc6 C-terminus residues 350–378 (Figure 6A and Figure 6—figure supplement 1A). Next, we constructed a full-length Cdc6 structure model by extracting the known structure from Protein Data Bank (PDB ID: 5V8F) (Yuan et al., 2017), which was lacking some of the key functional residues and regions, for example, 1–59, 68–80, 129–163, and 349–387. Indeed, these regions fall into the IDRs identified in Figure 6A. We generated a complete structural representation of Cdc6 through predictions of both IDRs and secondary structure combined with template-based modeling and ab initio approaches (see Methods). The missing regions modeled in the Cdc6 N-terminus span residues 1–59 (Figure 6B, left) as well as the centrally located residues 350–378 (Figure 6B, right), which correspond to the predicted IDRs (Figure 6A). The Cdc6 N-terminal IDR comprises four Cdk1 consensus sites (T7, T23, T39, and S43) and two cyclin-binding motifs RxL (30–32) and LxF (47–49), all of which are critical for Cdc6 phospho-regulation (Figure 6B, left). The Cdc6 C-terminal IDR (residues 350–378) contains phospho-degron 368–372 (Figure 6B, right). We also compared our Cdc6 structural model with the recently released AlphaFold and evaluated the model quality using Verify 3D software (Figure 6—figure supplement 1B). The 3D-1D profiles from Verify 3D measure the compatibility of the primary structure of a protein with its three-dimensional structural environment in the model. An average 3D-1D score of 0.2 (shown by the green line in the plots) denotes the threshold of acceptable model quality, that is, scores around this threshold or above are inferred as correctly modeled regions (Figure 6—figure supplement 1C). The phospho-degrons (red boxes) scored close to or at the 0.2 threshold in our model, while both regions score well below the threshold in the AlphaFold model (Figure 6—figure supplement 1C). This is also consistent with low confidence scores (less than 70) generated by AlphaFold’s internal confidence rating (Figure 6—figure supplement 1B, yellow and orange). The scores in the structured regions of both models are comparable. Thus, our model outperforms the AlphaFold model in the region of the modeled IDRs (Figure 6—figure supplement 1C).

Finally, we predicted a potential scenario of Cdc6–Clb2 interaction, where Cdc6 E45 interacts with Clb2 K270 in hp via a salt bridge (Figure 6C, D) that highlights the importance of the IDR in this interaction. We speculate that a transiently structured helix adopted by the PEKLQF motif in our Cdc6 model may be important for its interaction with the hp in Clb2. The N-terminus of Clb2 located from residues 1–210 (data not shown) contained an IDR which includes the KEN box (residues 85 and 100) and the destruction box (D-box) (residues 25–33), which are both necessary for Clb2 degradation via APC (Pfleger and Kirschner, 2000; Schwab et al., 1997; Glotzer et al., 1991; Figure 6C). The ordered regions of Clb2 include the hp (N260, L264, Y267, and K270) that facilitates protein binding to its substrate such as Cdc6 (Örd et al., 2019; Hunt, 1991; Figure 6C). Importantly, our Cdc6–Clb2 protein interaction model also captured the Cdc6 N-terminal phospho-degron protected by Clb2, thereby inhibiting Cdc6 degradation.

In this study, we show that PP2A^Cdc55 dephosphorylates Cdc6 at T7 and T23 to disrupt the Cdc6–Clb2 complex in mitosis. As cells progress into late mitosis, Sic1 releases Clb2·Cdk1·Cks1 from Cdc6 to promote origin licensing. Following this, Cdc14 dephosphorylates Cdc6 at T368-S372 to stabilize Cdc6 before origin licensing. Our bioinformatical analyses demonstrate that highly adaptable IDRs in the Cdc6-N terminus are critical for Cdc6-binding dynamics. The data provide a temporal model of origin licensing control by Clb2·Cdk1·Cks1 release and sequential dephosphorylation of Cdc6 (Figure 7A).

Neither PP2A^Cdc55 nor Cdc14 dephosphorylated the Cdc6 N-terminal phospho-degron, suggesting that there are additional mechanisms involved (Figure 4—figure supplement 2). The Cdc6-T7 site is located in an IDR, where PP2A^Cdc55 and Cks1 might gain easier access for docking (Figure 6A, B), which may trigger a cascade of Cdc6 dephosphorylation in early mitosis (Touati et al., 2019). Our study also supports a model in which PP2A^Cdc55 prefers phospho-threonines over phospho-serines (Godfrey et al., 2017).

Previous work in Xenopus egg extracts indicated that immunodepletion of the trimeric PP2A complex inhibits replication initiation without regulating pre-RC assembly (Lin et al., 1998). A follow-up study suggested that PP2A targets proteins required for DNA replication initiation but not elongation proteins, for example, Cdc45 (Chou et al., 2002). Sld2 and Sld3 are potential substrates for PP2A (Bloom and Cross, 2007). Therefore, PP2A might regulate multiple replication proteins at different steps. Mammalian PP2A is regulated by a diverse group of regulatory subunits, termed, B, B’, B’’, and B’’’ (Janssens and Goris, 2001). B55 is a mammalian homolog of Cdc55 which plays a role in mitosis (Schmitz et al., 2010). PP2A^B55 also selectively targets phospho-threonines over phospho-serines in vitro (Cundell et al., 2016; Agostinis et al., 1990). PR48 (B’’ family) has been identified as a Cdc6-binding partner through yeast two-hybrid screening (Yan et al., 2000). PR48 mediates protein interaction with the Cdk1 phosphorylation site-containing Cdc6 N-terminus (Yan et al., 2000). PR70, another B’’ family subunit, has also been implicated in Cdc6 regulation by enhancing phosphatase activity toward Cdc6 (Yan et al., 2000; Wlodarchak et al., 2013; Davis et al., 2008). Thus, our data define Cdc6 dephosphorylation by PP2A in a simple system, which might be a conserved mechanism in eukaryotes.

The Cdc6 C-terminal phospho-degron is targeted by SCF^Cdc4 for protein degradation in mitosis (Drury et al., 2000). Here, we show that the Cdc6 C-terminal phospho-degron is dephosphorylated by Cdc14 to stabilize Cdc6 (Figure 4). Cdc14 is inhibited by PP2A^Cdc55 through Net1 dephosphorylation (Queralt et al., 2006), which indicates that Cdc14-dependent Cdc6 dephosphorylation is not triggered until Clb2–Cdk1 activity overrides PP2A^Cdc55 activity. Elevated Clb2–Cdk1 activity phosphorylates Net1 to release Cdc14 from nucleolus (Figure 7B). PP2A^Cdc55-dependent Cdc6-T7 dephosphorylation releases Clb2, which may enhance Net1 phosphorylation because Clb2 is free from its inhibitor Cdc6 (Figure 7B, left). Once Cdc14 is released from Net1, Cdc14 dephosphorylates Cdc6, Cdh1, and Sic1 to license the origin (Figure 7B, right). This temporal regulation of Cdc14 ensure that Cdk1 substrates are sequentially dephosphorylated (Kataria et al., 2018). Such a temporal regulation is also achieved by cell cycle-dependent Cdc55 localization; nuclear Cdc55 inhibits the metaphase–anaphase transition through APC inhibition (Rossio et al., 2013; Queralt and Uhlmann, 2008).

There is evidence that Cdc14 is a key regulator for promoting DNA replication. When Cdc14 is not properly sequestered in CDC14-TAB6 mutant, cells show DNA replication defects when CLB5 is deleted (Bloom and Cross, 2007). Cdc14 dephosphorylates replication initiation proteins Orc2, Orc6, Cdc6, and Mcm3 (Zhai et al., 2010). Ectopic Cdc14 expression induces DNA re-replication, supporting the idea that Cdc14-dependent dephosphorylation of pre-RC components is a requirement for origin licensing (Zhai et al., 2010). Thus, Cdc14 is a key component to assemble pre-RCs in late mitosis.

Our results show that Clb2·Cdk1·Cks1 binding to Cdc6 may hinder SCF^Cdc4 recognition of the N-terminal phospho-degron and consequently leads to Cdc6 stabilization (Figures 2 and 6C). This indicates that the Cdc6 C-terminal phospho-degron is the key element for Cdc6 degradation until Clb2 is released. We have previously shown that the Cdc6 C-terminal phospho-degron is targeted by Mck1, a yeast homolog of GSK3, for protein degradation (Al-Zain et al., 2015; Ikui et al., 2012). Despite high Clb2–Cdk1 activity and an exposed phospho-degron, Cdc6 can be partially stabilized in mitosis until Mck1 phosphorylates Cdc6-T368. It is of interest to study how Mck1 is activated during the cell cycle. Mck1 is involved in DNA damage signaling (Al-Zain et al., 2015; Li et al., 2019), therefore the Cdc6 C-terminus, which is available throughout the cell cycle, plays a role in controlling the timing of licensing under stress conditions. We also showed evidence that Cdc6 is degraded in SCF^Cdc4-independent manner; for example, Cdc6-6A mutant is unstable despite the mutations at phospho-degrons (Figure 3—figure supplement 1).

APC ^Cdh1 targets Clb2 for degradation via the KEN- and D-boxes (Pfleger and Kirschner, 2000; Schwab et al., 1997; Glotzer et al., 1991). Cdh1 is inhibited until it is dephosphorylated by Cdc14. The KEN- and D-boxes in Clb2 are located in IDRs, which may also enhance interaction with the APC to facilitate Clb2 degradation upon mitotic exit (Figure 6C; Radivojac et al., 2010).

We showed that Sic1 releases Clb2 from Cdc6 which supports origin licensing (Figure 5). Sic1 contains multiple Cdk1 phosphorylation sites, which includes a phospho-threonine Cks1 docking site in the N-terminus (Kõivomägi et al., 2011; Venta et al., 2020). Cks1 docking triggers an N- to C-directed multisite phosphorylation cascade both in Sic1 and Cdc6, which results in the creation of phospho-degrons targeted by SCF^{Cdc4 14,18}. Thus, Cdc6 and Sic1 share a functional similarity in their phosphorylation patterns. Since PP2A^Cdc55 dephosphorylates Cks1 docking site T7 in Cdc6, it is worth examining if PP2A^Cdc55 dephosphorylates Sic1 at the N-terminal Cks1 docking site. Sic1, Cdh1, and Cdc6 are Clb2–Cdk1 inhibitors as well as Cdc14 substrates and SCF^Cdc4 targets (Örd et al., 2019; Calzada et al., 2001; Visintin et al., 1998; Verma et al., 1997). Cdc14-dependent Swi5 dephosphorylation also activates Sic1 nuclear import and transcription at the end of mitosis (Visintin et al., 1998). Thus, Cdc14 release from the nucleolus targets at least four components to achieve a complete inhibition of Clb2–Cdk1 activity during mitotic exit, leading to origin licensing (Figure 7B).

Recent studies have expanded the role of cyclin-specific short linear motifs (SLiMs), which are short functional sequence motifs often found in IDRs. SLiMs such as RxL and LxF motifs dictate Cdk1 substrate specificity and control cell cycle progression (Örd et al., 2019; Örd et al., 2020; Faustova et al., 2021; Bhaduri and Pryciak, 2011). This is also conserved in mammalian cells (Takeda et al., 2001; Allan et al., 2020; Topacio et al., 2019). Phosphatases also utilize SLiMs for substrate docking. In yeast, Cdc14 scaffolds to a PxL motif in substrates to mediate dephosphorylation events (Kataria et al., 2018). Currently, there is no identified docking motif for PP2A^Cdc55 in yeast. Mammalian homolog PP2A^B55 recognizes a polybasic docking sequence that flanks CDK phosphorylation sites (Cundell et al., 2016). PP2A^B56-specific SLiMs are defined as LSPIxE in humans (Wang et al., 2016). Future research should focus on docking motifs for phosphatases and their regulatory subunits to understand phosphatase specificity in yeast.

Our data illustrate multiple IDRs in Cdc6 (Figure 6A). Notably, the N-terminal Cdc6 IDR includes SLiMs such as RxL and LxF motifs, which supports a model that depicts the Cdc6 N-terminus as a hub for multiple protein–protein interactions (Figure 6B). Cdc6 IDRs are readily available for protein binding in G1/S, during which Cdc6 disengages from ORC and is subjected to SCF-dependent degradation (Drury et al., 1997; Piatti et al., 1995). Cdc6 is then shifted toward an origin licensing preparation period where Cdc6 is stabilized but not licensed via Clb2 binding. This temporal mode is released when Cdc14 outcompetes with Clb2–Cdk1 activity at the end of mitosis.

In humans, cyclin A–CDK2 was shown to utilize SLiMs to promote interaction with Cdc6 and ORC (Hossain et al., 2021). Our studies on Cdc6 regulation shed light on the dynamics of phosphorylation and dephosphorylation that drive DNA replication initiation upon mitotic exit. In this study, we showed first evidence to link PP2A with DNA replication. This study also provided a stepwise origin licensing process mediated by PP2A, Cdc14, and Sic1.

Yeast extract peptone medium with glucose (YPD) was used to grow cells for western blot and flow cytometry analysis. Low fluorescence medium with glucose supplemented with adenine was used to visualize Cdc6-GFP under fluorescence microscope (Sheff and Thorn, 2004). G1 arrest was achieved by α-factor at 50 nM for 2 hr at 30°C. Cells were washed three times and resuspended in YPD to release the G1 arrest. Mitotic arrest was achieved by nocodazole at 15 μg/ml for 2 hr at 30°C.

Standard methods were used for mating, tetrad analysis, and transformation. All yeast strains are haploid congenic to W303 background. Strain list in this study is in Supplementary file 1.

Cells were lysed in TBT buffer containing protease inhibitors and Phos STOP (Roche) with acid-washed glass bead agitation using Fast-Prep as previously described (Bloom et al., 2011). Proteins were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) with Novex 4–20% Tris–glycine polyacrylamide gel (Life Technologies). Cdc6-9MYC was detected using anti-cMYC antibody 9E10 (Sigma-Aldrich Cat# M4439, RRID:AB_439694) at 1:5000 dilution and anti-Pgk1 (Thermo Fisher Scientific Cat# 459250, RRID:AB_2532235) at 1:5000 as a loading control. Cdc6-protein A was detected by HRP-conjugated rabbit IgG (Sigma-Aldrich Cat# P1291, RRID:AB_1079562). Anti-Clb2 antibody is a gift from Frederick Cross. Images were developed using a Fuji LAS 4000 Imager (GE Healthcare Life Sciences, Pittsburgh, PA) or Odyssey CLx imager (LI-COR Lincoln, NE). Co-immunoprecipitation was performed with rabbit anti-MYC-conjugated agarose beads (Sigma-Aldrich, St Louis, MO) for 1 hr at 4°C. Cdc6-T7 phospho antibody is custom antibody raised in rabbit (Covance, Princeton, NJ). Sic1-HA and Cdc6-HA were detected by HRP-conjugated rat HA (Roche Cat# 12013819001, RRID:AB_390917).

Cells were fixed and stained with propidium iodide staining as previously described (Epstein and Cross, 1992). Flow cytometry analysis was performed using a BD Accuri C6 flow cytometer (BD Biosciences). A total of 20,000 cells were analyzed per sample. FL2 channel was used to detect the fluorescent signal.

Cdc6, Sic1, Swe1(1–450), and Cdc14 were expressed in E. coli BL21RP cells from pET28a-based vectors as fusions with 6xHis tag. Csa1 was expressed as GST fusion in BL21RP cells from pGEX-4T1-based vector. The expression of Cdc6 and Swe1(1–450) was induced at 23°C using 0.3 mM IPTG, Sic1 at 37°C using 1 mM IPTG, Cdc14 at 23°C using 0.125 mM IPTG, and Csa1 at 16°C using 0.3 mM IPTG. 6xHis-Cdc6, 6xHis-Sic1, and 6xHis-Swe1(1–450) were purified by standard cobalt affinity chromatography with 200 mM imidazole used for elution. 6xHis-Cdc14 was purified using nickel affinity chromatography with 250 mM imidazole used for elution. GST-Csa1 was purified using Glutathione Sepharose (GE Healthcare). Clb5–Cdk1 complex was purified from S. cerevisiae culture where TAP-tagged Clb5 was overexpressed from GAL1 promoter as described previously (Puig et al., 2001; Ubersax et al., 2003). The yeast lysate was prepared using Mixer Mill MM 400 (Retch). PP2A^Cdc55 complex was purified from yeast cells containing 3HA-tagged Cdc55 (Anastasia et al., 2012). The yeast culture was grown in YPD to OD₆₀₀ = 1.4, when the cells were collected and snap frozen. The cells were lysed using Mixer Mill MM 400. The lysate was cleared by centrifugation and the supernatant was incubated with anti-HA agarose beads for 3 hr. The beads were washed thoroughly and PP2A^Cdc55 was eluted with buffer containing HA dipeptide.

Substrate proteins Cdc6, Swe1(1–450), and Csa1 at 500 nM concentration were phosphorylated for 60 min at room temperature using 2 nM Clb5–Cdk1 in a buffer containing 50 mM HEPES–KOH, pH 7.4, 150 mM NaCl, 5 mM MgCl₂, 20 mM imidazole, 2% glycerol, 0.2 mg/ml BSA, and 500 μM ATP [(with added [γ-³²P]-ATP (Hartmann Analytic))]. Prior to addition of phosphatase mix, an aliquot of the reaction was pipetted to SDS sample buffer. Then, a mixture containing Sic1 and phosphatase Cdc14 or PP2A^Cdc55 was mixed with the phosphorylation reactions to inhibit Clb5–Cdk1 and to measure the dephosphorylation. The final concentration of Sic1 was 1 µM. At 4, 8, and 16 min, an aliquot of the reaction was mixed with SDS sample buffer to stop the reaction. The samples were loaded on SDS–PAGE and following electrophoresis, the gels were stained using Coomassie Brilliant Blue R-250 dye and dried.

Γ-³²P phosphorylation signals were detected using an Amersham Typhoon 5 Biomolecular Imager (GE Healthcare Life Sciences). Signals were quantified using ImageQuant TL (Image Quant TL, RRID:SCR_018374). The dephosphorylation assays were performed in three independent replicate experiments.

Images were acquired with Nikon Eclipse Ti2 microscope with ×60 Oil CFI Plan APO Lambda lens. Fluorescence illumination is provided by Sola light engine and images were captured by Nikon DS-Qi2. Images were acquired by Nikon NIS element software.

Yeast cultures were grown at 30°C in synthetic complete media with 2% glucose (SC) to OD600 0.2–0.6, and placed on 0.08 mm cover glass and covered with 1 mm thick 1.5% agarose (NuSieveTM GTGTM Agarose, Lonza) pad made with SC. Images were acquired using Zeiss Observer Z1 microscope with Axiocam 506 mono camera (Zeiss), ×63/1.4NA oil immersion objective and Tempcontrol 37-2 digital (PeCon) to keep the sample at 30°C. Time-lapse imaging was performed using ZEN software, an automated stage and Definite Focus. Phase-contrast, Citrine, and mCherry images were taken every 3 min for 8 hr from up to 12 positions. Colibri LED modules at 25% power were used to excite the fluorescent proteins. Cdc6-Citrine was imaged using Zeiss Filter Set 46 and Colibri 505 LED module for 500 ms. Whi5-mCherry was imaged using Zeiss Filter Set 61 HE and Colibri 540–580 LED module for 750 ms. The analysis was carried out using MATLAB (MATLAB, RRID:SCR_001622). The cells were segmented and tracked based on the phase-contrast image, followed by quantification of the nuclear fluorescence signals as described in Doncic et al., 2013. The nuclear signal was obtained by applying a 2D Gaussian fit to the brightest point in the cell and assigning the 25% with highest signal as the nucleus. The cytoplasmic fluorescence signal was considered as background and was subtracted from the nuclear signal. For every strain, data are from at least two repeats with different transformants.

Sic1 was purified as described before (Gros et al., 2014). ORC and Cdt1·Mcm2–7 were purified as described previously (Remus et al., 2009). For Cdc6·Clb2·Cdk1·Cks1, cells (strain YDR12) were grown in 10 l of YPD at 25°C to a density of 2 × 10⁷ ml⁻¹, arrested in G1 phase by addition of 50 μg ml⁻¹ α-factor, harvested by centrifugation, resuspended in 10 L YP/2% galactose lacking α-factor, and incubated at 25°C overnight to induce expression of Cdc6-TAP^TCP. Next day, cells were harvested by centrifugation, washed twice with cold 25 mM HEPES, pH 7.6/1 M sorbitol, once with buffer H (25 mM HEPES–KOH pH 7.6/0.02% NP-40/10% glycerol)/300 mM KCl, resuspended in 0.5× volume of buffer H/300 mM KCl/2 mM DTT/protease inhibitor cocktail, and frozen as droplets in liquid nitrogen. The resulting ‘popcorn’ was stored at −80°C until further processing. Cell lysate was prepared by crushing the frozen popcorn in a freezer mill (SPEX CertiPrep 6,850 Freezer/Mill) for 6 cycles of 2 min at a rate of 15 impacts per second. Crushed cell powder was thawed on ice, resuspended with 1 volume of buffer H/0.3 M KCl/1 mM DTT. Insoluble material was pelleted by centrifugation of the lysate in a type 45 Ti rotor (Beckman) for 60 min at 40,000 rpm. The clarified extract was supplemented with 2 mM CaCl₂ and incubated with 1 ml calmodulin affinity beads for 2 hr at 4°C. The calmodulin resin was washed with 10 CV of buffer H/300 mM KCl/2 mM CaCl₂/1 mM DTT, and bound protein eluted with 10 CV of buffer H/300 mM KCl/1 mM EDTA/2 mM EGTA/1 mM DTT. Peak fractions were pooled, concentrated by centrifugation through an Amicon spin concentrator, digested for 4 hr at 4°C with TEV protease to remove the TAP^TCP tag, and subsequently fractionated by gel filtration using a 24 ml Superdex 200 column equilibrated in buffer H/300 mM KCl/1 mM EDTA/1 mM EGTA/1 mM DTT. Peak fractions were pooled, diluted with 2× volumes of buffer H/1 mM EDTA/1 mM EGTA/1 mM DTT to a final salt concentration of 100 mM KCl, and fractionated on a 1 ml Mono S ion exchange column using an elution gradient of 0.1–1 M KCl over 10 CV. Peak fractions were pooled and stored in aliquots at −80°C after snap freezing in liquid nitrogen. For the original identification of the Cdc6-associated proteins by mass spectrometry, purified complex was fractionated by SDS–PAGE, individual Coomassie-stained bands excised from the gel, and proteins subjected to mass-spectrometric analysis after in-gel trypsin digestion using standard protocols.

6 μg of Sic1, 4 μg Cdc6·Clb2·Cdk1·Cks1, or a mixture of 6 μg Sic1 and 4 μg Cdc6·Clb2·Cdk1·Cks1 were incubated in a 50 μl reaction volume containing buffer H/300 mM KCl/1 mM EDTA/1 mM EGTA/1 mM DTT for 10 min at room temperature, fractionated on a 2.4 ml Superdex 200 PC 3.2/30 column, and fractions analyzed by SDS–PAGE and silver stain.

Purified Mcm2–7 loading onto linear 1 kbp ARS-containing DNA immobilized on paramagnetic beads using was performed as described previously (Remus et al., 2009), except that 50 nM Cdc6·Clb2·Cdk1·Cks1 was used in place of Cdc6. Where indicated, Sic1 was included at 500 nM.

The IDRs and disordered binding regions of Cdc6 were predicted using DISOPRED (UCL Bioinformatics Group, RRID:SCR_010248; Ward et al., 2004), IUPRED2A (IUPRED, RRID:SCR_014632; Mészáros et al., 2018), PrDOS (PrDOS, RRID:SCR_021886), MFDp2 (MFDp2, RRID:SCR_021885; Mizianty et al., 2013), and CSpritz (CSpritz, RRID:SCR_021884; Walsh et al., 2011). A consensus was derived by combining the high confidence predictions from all these programs.

The atomic coordinates for Cdc6 isolated from the cryo-EM structure of ORC-Cdc6-Cdt1-Mcm2–7 intermediate (OCCM; PBD ID: 5V8F; Yuan et al., 2017) were extracted and used to build a full-length template of Cdc6. The 140 missing residues from this structure for Cdc6 was modeled separately via an ab initio modeling program QUARK (QUARK, RRID:SCR_018777; Zhang et al., 2016). The full-length Cdc6 model was constructed by combining abovementioned fragment and extracted coordinates of Cdc6 using MODELLER (MODELLER, RRID:SCR_008395; Eswar et al., 2006) and further refined using 3DRefine (Bhattacharya et al., 2016). Clb2 was modeled using a hybrid threading approach, ab initio, and template-based approach based on the algorithm, I-TASSER (Iterative Threading ASSEmbly Refinement) (I-TASSER, RRID:SCR_014627; Roy et al., 2010). I-TASSER identifies structural templates from the PDB by a multithreading server, LOMETS (LOMETS, RRID:SCR_021882; Wu and Zhang, 2007), and then constructs full-length atomic models by iterative template-based fragment assembly simulations. The IDR in Clb2 was modeled independently using QUARK (Zhang et al., 2016) and incorporated to create a full-length model in similar manas described above for Cdc6. Both the full-length modeled structures were meticulously evaluated by various structure verification programs, including VERIFY3D (University of California at Los Angeles – Department of Energy Institute for Genomics and Proteomics, RRID:SCR_001921; Eisenberg et al., 1997), Voromqa (VoroMQA, RRID:SCR_021881; Olechnovič and Venclovas, 2017), Proq3 (ProQ3, RRID:SCR_021880; Uziela et al., 2016), and ProsaWeb (ProSA-web, RRID:SCR_021879; Wiederstein and Sippl, 2007). These programs employ a variety of methods inclusive of deep learning approaches to evaluate the quality of the models. The final models were chosen by selecting the best evaluation profiles and their correlation with known functional features. The models were visualized and analyzed for their biophysical properties using the visualization programs Pymol (PyMOL, RRID:SCR_000305; Schrodinger, 2010) and ChimeraX (UCSF ChimeraX, RRID:SCR_015872; Pettersen et al., 2021).