Antagonistic control of DDK binding to licensed replication origins by Mcm2 and Rad53

Eukaryotic replication origins are licensed by the loading of the replicative DNA helicase, Mcm2-7, in inactive double hexameric form around DNA. Subsequent origin activation is under control of multiple protein kinases that either promote or inhibit origin activation, which is important for genome maintenance. Using the reconstituted budding yeast DNA replication system, we find that the flexible N-terminal extension (NTE) of Mcm2 promotes the stable recruitment of Dbf4-dependent kinase (DDK) to Mcm2-7 double hexamers, which in turn promotes DDK phosphorylation of Mcm4 and −6 and subsequent origin activation. Conversely, we demonstrate that the checkpoint kinase, Rad53, inhibits DDK binding to Mcm2-7 double hexamers. Unexpectedly, this function is not dependent on Rad53 kinase activity, suggesting steric inhibition of DDK by activated Rad53. These findings identify critical determinants of the origin activation reaction and uncover a novel mechanism for checkpoint-dependent origin inhibition.


Introduction
To ensure the timely, accurate, and complete duplication of their genomes prior to cell division, eukaryotic cells initiate DNA replication at many replication origins distributed along the length of each chromosome. From each origin two replication forks emanate in opposite direction to form a replication bubble.
Although bidirectional origin firing has long been recognized as a universal feature of chromosomal DNA replication in all domains of life (Huberman and Riggs, 1968;Prescott and Kuempel, 1972), how pairs of oppositely oriented replication forks are established at chromosomal origins remains poorly understood. In eukaryotes, two copies of the replicative DNA helicase, Mcm2-7, are loaded as a stable double-hexameric complex around double-stranded DNA (dsDNA) at the origin (Evrin et al., 2009;Miller et al., 2019;Remus et al., 2009). Mcm2-7 comprise six related proteins of the AAA+ family of ATPases that assemble into a hexameric ring with defined subunit order. Intriguingly, the two hexamers in a Mcm2-7 double hexamer (DH) associate in a head-to-head configuration, thus providing a platform for the establishment of oppositely oriented sister replisomes. (Abid Ali et al., 2017;Evrin et al., 2009;Li et al., 2015;Miller et al., 2019;Noguchi et al., 2017;Remus et al., 2009). However, Mcm2-7 DHs are catalytically inactive and require the regulated association of the essential helicase co-factors Cdc45 and GINS to form two active replicative DNA helicase complexes, termed CMG (Cdc45-MCM-GINS), which encircle single-stranded DNA (ssDNA) during unwinding (Bell and Labib, 2016). Therefore, to ensure bidirectional origin firing, mechanisms must exist that ensure simultaneous progression of both helicase complexes from the origin. As the head-to-head orientation of CMG helicases assembled at the origin requires them to pass each other during origin activation, it has been proposed that an active CMG encircling ssDNA is blocked by a dsDNA-encircling inactive CMG potentially formed around the opposite Mcm2-7 hexamer, thereby imposing origin bidirectionality (Douglas et al., 2018;Georgescu et al., 2017).
The replication of chromosomes from multiple replication origins necessitates strict control mechanisms that prevent origins from re-firing within one cell cycle in order to maintain genome stability. Such re-replication control is achieved by a two-step mechanism that temporally separates helicase loading in late M and G1 phase from helicase activation in S phase (Bell and Labib, 2016). Helicase activation is controlled by two cell cycle-regulated protein kinases, Dbf4-dependent kinase (DDK) and cyclin-dependent kinase (CDK), which act in conjunction with a defined set of co-factors, comprising Sld3·7, Sld2, Dpb11, Pol e, and Mcm10, in addition to GINS and Cdc45, to mediate CMG assembly (Douglas et al., 2018). The essential targets of CDK during CMG assembly are Sld2 and Sld3, which physically interact with distinct Dpb11 BRCT domains when phosphorylated to recruit GINS and Pol e to the origin (Bell and Labib, 2016). The mechanism by which DDK promotes CMG assembly is somewhat obscure. Genetic and biochemical studies demonstrate that  are the essential targets for DDK during DNA replication (Deegan et al., 2016;Hardy et al., 1997;Randell et al., 2010;Sheu and Stillman, 2010;Yeeles et al., 2015). Accordingly, DDK phosphorylation of the flexible Nterminal tails of Mcm4 and -6 promotes the recruitment of Sld3·7 to Mcm2-7 DHs by generating phosphorylation-dependent binding sites for Sld3, which in turn recruits Cdc45 (Deegan et al., 2016;Gros et al., 2014;Heller et al., 2011;Tanaka et al., 2011;Yeeles et al., 2015). However, the nature and stoichiometry of the Sld3 interaction with Mcm2-7 DHs is unclear, as Sld3 exhibits limited sequence specificity for phosphodependent binding sites, and phospho-mimicking mutations in either Mcm4 or -6 are sufficient to bypass the requirement for DDK (Deegan et al., 2016;Randell et al., 2010). Moreover, other mutations in Mcm5 and Mcm4 that do not involve phospho-mimetic amino acid substitutions can also bypass the requirement for DDK (Hardy et al., 1997;Sheu and Stillman, 2010). It is not known, if these mutations allow Sld3 recruitment in the absence of Mcm2-7 phosphorylation, or if they bypass the requirement for Sld3 altogether.
Replication origins do not fire simultaneously upon S phase entry, but in a staggered programmatic manner throughout S phase (Rhind and Gilbert, 2013). The replication timing program is in part imposed by limiting concentrations of the initiation factors Sld2, Dpb11, Sld3, Cdc45, and Dbf4, which restrict CMG assembly to subsets of origins throughout S phase (Mantiero et al., 2011;Tanaka et al., 2011). While limiting initiation factors that interact transiently with origins are thought to be recycled from early to late origins in a normal S phase, such recycling is prevented by the S phase checkpoint to inhibit late origin firing. The S phase checkpoint is a vital kinase signaling cascade that induces a range of cellular responses in addition to the inhibition of origin firing, including increasing dNTP levels, stabilization of stalled replication forks, and DNA repair to promote genome maintenance during replication stress (Pardo et al., 2017).
In budding yeast, the minimal targets for checkpoint-dependent origin inhibition are Sld3 and Dbf4, which are substrates for the checkpoint effector kinase, Rad53 (Lopez-Mosqueda et al., 2010;Zegerman and Diffley, 2010). Rad53 phosphorylation of Sld3 inhibits its physical interactions with Mcm2-7, Cdc45, and Dpb11 (Deegan et al., 2016;Lopez-Mosqueda et al., 2010;Zegerman and Diffley, 2010). How Rad53 inhibits Dbf4 is not clear. Rad53 was found to inhibit DDK kinase activity in vitro, but the mechanism of inhibition is unknown (Kihara et al., 2000;Weinreich and Stillman, 1999). Moreover, DDK activity in these studies was assessed using isolated Mcm2 or Mcm7 subunits as substrates or by measuring DDK autophosphorylation (Kihara et al., 2000;Weinreich and Stillman, 1999). However, since DDK exhibits high specificity for Mcm4 and -6 in the context of Mcm2-7 DHs, the effect of Rad53 on DDK activity at licensed origins remains to be determined (Francis et al., 2009;Randell et al., 2010;Sheu and Stillman, 2006;Sun et al., 2014). Rad53 has also been reported to disrupt DDK-chromatin association in cells treated with hydroxyurea (HU) (Pasero et al., 1999). DDK binds to chromatin via an interaction with Mcm2-7 at licensed replication origins (Dowell et al., 1994;Francis et al., 2009;Jares and Blow, 2000;Jares et al., 2004;Sato et al., 2003;Sheu and Stillman, 2006;Takahashi and Walter, 2005;Weinreich and Stillman, 1999;Yanow et al., 2003). Several potential DDK docking sites have been identified in Mcm2-7 based on pairwise interaction studies with individual Mcm2-7 subunits (Ramer et al., 2013;Sheu and Stillman, 2006). However, it has not been tested to what extent these contribute to DDK binding in the context of the Mcm2-7 DH, which is known to greatly stimulate DDK substrate specificity and kinase efficacy (Francis et al., 2009;Sun et al., 2014). How Rad53 may affect the DDKchromatin association has not been addressed.
Here we employ the reconstituted budding yeast DNA replication system to investigate the mechanism of DDK docking to Mcm2-7 DHs and its regulation by Rad53 (Devbhandari et al., 2017;Devbhandari and Remus, 2020). We find that the flexible Mcm2 N-terminal tail is necessary for DDK docking onto Mcm2-7 DHs, Mcm4 and -6 phosphorylation, and origin firing. The data suggests that the N-terminal tails of both Mcm2 protomers in the Mcm2-7 DH are required for efficient origin activation. Intriguingly, Rad53 disrupts DDK binding to Mcm2-7 DHs in a manner that is independent of Rad53 kinase activity, but dependent on prior activation of Rad53. These observations have important implications for the mechanism of bidirectional origin firing and its regulation by the replication checkpoint.
First, we tested if proteolytic truncation of the Mcm2 N-terminal tail affects Mcm2-7 DH stability. For this we performed reconstituted Mcm2-7 loading reactions on origin-containing DNA using purified ORC, Cdc6, and either wildtype Cdt1×MCM or Cdt1×MCM 2-TEV (Remus et al., 2009). Following Mcm2-7 loading, DNA-bound complexes were washed to remove free proteins, treated with TEV protease, and analyzed by SDS-PAGE.
The DNA was immobilized on paramagnetic streptavidin-coated beads via a single photo-cleavable 5'-terminal biotin moiety to allow elution of the DNA from the beads with UV light for analysis. To differentiate  DHs loaded around DNA from potential other more loosely associated complexes, such as the OCCM (ORC-Cdc6-Cdt1-MCM), DNA-bound complexes were washed with a high-salt buffer prior to DNA elution (Remus et al., 2009;Yuan et al., 2017). While Mcm2-7 DHs are resistant to salt-elution from the DNA, loading factors and loosely associated Mcm2-7 complexes are efficiently disrupted by stringent salt washes. In addition, control Mcm2-7 loading reactions were carried out in the presence of non-hydrolyzable ATPgS as Mcm2-7 DH formation is strictly dependent on ATP hydrolysis (Bell and Labib, 2016;Remus et al., 2009). Using this approach, we demonstrate comparable DNA loading efficiencies for wildtype Cdt1×MCM and Cdt1×MCM 2-TEV .
Moreover, truncation of the Mcm2 N-terminal tail from Mcm2-TEV-containing DHs, despite being efficient, did not negatively affect Mcm2-7 DH retention on DNA ( Figure 1C). Thus, maintenance of Mcm2-7 DHs is not dependent on residues 1-127 of the Mcm2 N-terminal tail.

Residues 1-127 of Mcm2 are important for DNA replication
Next we tested if residues 1-127 of the Mcm2 N-terminal tail are required for DNA replication in vitro.
For this we performed DNA replication reactions both on naked DNA templates and reconstituted chromatin as described previously (Devbhandari et al., 2017). As FACT and Nhp6 have been demonstrated to promote replisome progression through chromatin in vitro, purified FACT and Nhp6 were also included in chromatin replication reactions here (Supplementary Figure 2) (Kurat et al., 2017). TEV protease cleavage of the Mcm2 N-terminal tail was induced following Mcm2-7 loading (Figure 2A).
Intriguingly, truncation of Mcm2 residues 1-127 largely attenuated DNA replication ( Figure 2B). Figure 3). The DNA replication defect was not due to a loss of HBD function, as mutation of two conserved tyrosine residues, Y82 and Y91, in the Mcm2 HBD (Cdt1×MCM 2-2A , Figure 2C) that have been previously shown to disrupt histone H3/H4 and FACT binding to Mcm2 has little effect on DNA replication using either DNA or chromatin as a template (Foltman et al., 2013). This is consistent with previous reports demonstrating that yeast cells harboring the mcm2-2A allele are viable and exhibit only mild replication defects (Foltman et al., 2013). This data reveals that the Mcm2 N-terminal tail performs a fundamental function during normal DNA replication that is distinct from its histone H3/H4 chaperone activity.

The Mcm2 N-terminal tail promotes DDK function during the initiation of DNA replication
In order to dissect which step in the DNA replication reaction is defective in the absence of residues 1-127 of Mcm2 we performed order-of-addition experiments by adding TEV protease at various steps of the origin firing pathway ( Figure 3A). In control experiments, TEV protease did not disrupt the DNA replication proficiency of wild-type Cdt1×MCM when added prior to DDK immediately after Mcm2-7 loading ( Figure 3B).
Conversely, purified Cdt1×MCM 2D127 , obtained by prior TEV protease digestion of Cdt1×MCM 2-TEV , was deficient for DNA replication, as expected (we show below that the replication defect is not due to a MCM loading defect). As before, in the presence of Cdt1×MCM 2-TEV , addition of TEV protease to the reaction immediately after Mcm2-7 loading inhibited origin activation. In striking contrast, addition of TEV protease after DDK or Sld3×7 allowed DNA replication to proceed normally. This data indicates that Mcm2 residues 1-127 promote DDK function during the initiation of DNA replication but are dispensable at later steps of the DNA replication reaction.
As the essential function of DDK is the phosphorylation of the N-terminal tails of Mcm4  Phosphorylation of the Mcm4 and -6 N-terminal tails by DDK promotes the assembly of the CMG helicase. We, therefore, tested if truncation of the Mcm2 N-terminal tail impairs CMG assembly using an origindependent CMG helicase assay that detects CMG helicase activity by the generation of highly unwound circular plasmid DNA, termed U-form DNA (Douglas et al., 2018). As expected, generation of U-form DNA in the presence of either Mcm2-WT or Mcm2-TEV is dependent on both CDK and DDK, demonstrating that plasmid unwinding is dependent on CMG assembly ( Figure 3D). Importantly, generation of U-form DNA was suppressed specifically in the presence of Mcm2-TEV when TEV protease was added to the reaction after the Mcm2-7 loading step, prior to the addition of DDK and other initiation factors. This data is consistent with residues 1-127 of the Mcm2 N-terminal tail promoting CMG assembly. In summary, we conclude that the Mcm2 N-terminal tail is important for DNA replication by promoting the phosphorylation of Mcm4 and -6 by DDK and subsequent CMG assembly, whereas it is dispensable for DNA replication after CMG assembly.

The Mcm2 N-terminal tail promotes DDK docking onto Mcm2-7 DHs
Next we addressed how Mcm2 may promote the phosphorylation of Mcm4 and -6 by DDK. Previous studies had proposed a docking mechanism by which a stable association of DDK with Mcm2-7 DHs promotes processive multi-site phosphorylation of the Mcm4 and -6 N-terminal tails (Francis et al., 2009;Sheu and Stillman, 2006). These studies, however, either utilized complex cell extracts to load Mcm2-7 onto DNA, which may include unknown proteins that bridge the DDK-Mcm2-7 interaction, or examined the binding of DDK to the To generate Cdt1·MCM 2D127 , we digested Cdt1·MCM 2-TEV with TEV protease and re-purified Cdt1·MCM 2D127 from the digestion reaction by gel-filtration chromatography. Truncation of the Mcm2 Nterminus did not have a detectable effect on Cdt1·MCM stability, as Cdt1·MCM 2D127 eluted in a mono-disperse peak during gel-filtration ( Figure 5A). Cdt1·MCM 2D127 also did not exhibit any discernible Mcm2-7 loading defects, demonstrating that the Mcm2 N-terminus is dispensable for Mcm2-7 recruitment and loading onto DNA ( Figure 5B). Next we assessed the contribution of the Mcm2 N-terminus to DNA replication. For this, wildtype or mutant Cdt1·MCM complexes were included at 80 nM during the Mcm2-7 loading step, a concentration that supports near maximal DNA replication levels. Importantly, at Cdt1·MCM concentrations below 80 nM origin firing efficiency strongly correlates with Cdt1·MCM concentration, thus allowing sensitive detection of loss of Cdt1·MCM activity (Supplementary Figure 4A). Similar to our previous approach (  Figure 4B). Importantly, we did not detect any evidence for asymmetric, or uni-directional, origin firing arising from the activation of a single hexamer within heterologous Mcm2-7 DHs assembled from both Cdt1×MCM 2-WT or Cdt1×MCM 2D127 , which may either generate greater than half-unit length leading strands, or cause a disproportional increase in lagging strand products as leading strand synthesis is initiated by the lagging strand of the sister replisome (Aria and Yeeles, 2018). In summary, we conclude that the Mcm2 N-termini of both hexamers are required for efficient origin firing, while the presence of a single Mcm2 N-terminal tail at a Mcm2-7 DH does not result in the establishment of single replication forks at an origin.

Rad53 sterically inhibits DDK binding to Mcm2-7 DHs
We have shown that DDK binding to Mcm2-7 DHs is required for origin activation. Interestingly, previous studies have indicated that a physical interaction between Rad53 and DDK contributes to the inhibition of origin activation by the checkpoint (Chen et al., 2013;Duncker et al., 2002;Matthews et al., 2014;Varrin et al., 2005). We, therefore, asked whether Rad53 might control origin activity by inhibiting DDK binding to Mcm2-7 DHs. For this we purified recombinant wildtype Rad53, which undergoes autoactivation during overexpression in E. coli, or the catalytically dead Rad53-D339A mutant, designated Rad53-kd below (Gilbert et al., 2001). Indeed, pre-incubation of DDK with Rad53 in the presence of ATP prior to addition of  DHs prevented both the binding of DDK to Mcm2-7 DHs and Mcm4 and -6 phosphorylation by DDK ( Figure   6A, lanes 4+5). Moreover, Rad53 was able to displace DDK from Mcm2-7 DHs when DDK binding to  DHs preceded addition of Rad53 (lanes 4+7). However, this displacement action of Rad53 was slightly less efficient at disrupting the DDK-Mcm2-7 DH interaction than the action of preventing DDK recruitment (lanes 5+7). Intriguingly, Rad53-kd also largely inhibited stable binding of DDK to Mcm2-7 DHs when co-incubated with DDK prior to addition to Mcm2-7 DHs, demonstrating that DDK phosphorylation by Rad53 is not essential to prevent stable DDK recruitment to Mcm2-7 (lanes 4+6). However, some residual DDK binding and significant Mcm4 and -6 phosphorylation occurred in the presence of Rad53-kd, indicating that Rad53-kdmediated inhibition of DDK is inefficient. Moreover, unlike Rad53-WT, Rad53-kd was unable to displace DDK from Mcm2-7 DHs when DDK was bound to Mcm2-7 DHs prior to Rad53-kd addition (lanes 4+8). These observations are consistent with genetic data demonstrating that kinase-dead alleles of Rad53 are deficient in origin inhibition (Lopes et al., 2001;Pellicioli et al., 1999). We conclude that activated Rad53 inhibits DDK binding to Mcm2-7 DHs, which in turn inhibits Mcm4 and -6 phosphorylation.
To further address if Rad53 phosphorylation of DDK is required for the inhibition of DDK binding to Mcm2-7 DHs we monitored DDK-Mcm2-7 DH complex formation in the presence of AMP-PNP. As we have shown above, AMP-PNP promotes DDK binding to Mcm2-7 DHs to the same extent as ATP ( Figure 4E).
Strikingly, the ability of Rad53 to prevent DDK recruitment to Mcm2-7 DHs or to displace DDK from  DHs was undiminished in the presence of AMP-PNP ( Figure 6B). Thus, DDK phosphorylation by Rad53 is not required to control DDK-Mcm2-7 DH complex formation. This raises the question why Rad53-kd is deficient at inhibiting DDK. Intriguingly, Rad53-WT and Rad53-kd exhibit very different elution profiles during gel-filtration chromatography ( Figure 6C). While Rad53-kd elutes at a volume that is consistent with a monomeric structure, Rad53-WT elutes as oligomeric complex. Rad53 and its human homolog, Chk2, are known to associate into homo-dimeric complexes during activation by trans-autophosphorylation, suggesting the oligomeric form of Rad53 observed here is a dimer (Ahn and Prives, 2002;Cai et al., 2009;Oliver et al., 2006;Wybenga-Groot et al., 2014;Xu et al., 2002). However, contrary to our observation, both kinase-dead Rad53 and Chk2 have been shown previously to also form dimers (Cai et al., 2009;Wybenga-Groot et al., 2014). As these previous studies were carried out with truncated kinase versions it is possible that the additional domains present in our full-length Rad53-kd construct affect its dimerization. For example, the isolated Chk2 kinase domain adopts a highly distinct dimer configuration from that of a Chk2 construct spanning the FHA and kinase domain, supporting the notion that domain composition can affect the oligomeric structure of Rad53/Chk2 (Cai et al., 2009;Oliver et al., 2006). We conclude that Rad53 activation results in the formation of Rad53 dimers that can sterically inhibit DDK binding to Mcm2-7 DHs, suggesting a novel non-catalytic mechanism for Rad53dependent origin control.

Discussion
The head-to-head orientation of the hexamers in the Mcm2-7 DH requires CMG helicases to pass each other during origin firing (Douglas et al., 2018;Georgescu et al., 2017). Based on this configuration it was proposed that an inactive CMG encircling dsDNA blocks the progression of a CMG formed around the opposite hexamer to impose bidirectional origin firing (Douglas et al., 2018;Georgescu et al., 2017). However, a stalled CMG helicase encircling ssDNA would pose a threat to DNA integrity due to the exposure of the displaced strand at the active CMG. As an alternative fail-safe mechanism we propose that simultaneous activation of both Mcm2-7 hexamers at an origin is controlled by an interdependent mechanism. Our data identify the Mcm2 N-terminal tail as a critical component of such a mechanism, as loss of a single Mcm2 N-terminus at a Mcm2-7 DH inhibits origin activity without inducing unidirectional firing. Mechanistically, we propose that both Mcm2 Nterminal tails in a Mcm2-7 DH are required for tethering of a pair of DDK molecules to promote symmetric phosphorylation of both hexamers ( Figure 7A). Supporting an interdependent hexamer activation mechanism, recent findings have demonstrated that both Mcm2-7 hexamers loaded at an origin have to be physically associated to support CMG activation (Champasa et al., 2019). Interdependent Mcm2-7 activation may be required at the origin melting step, which may involve two CMGs working in opposite direction against each other (Froelich et al., 2014;Langston and O'Donnell, 2019;Noguchi et al., 2017).
Our proposed role of the Mcm2 N-terminal tail as a DDK docking site is supported by a previous yeast two-hybrid study that detected a specific pairwise interaction between Dbf4 and the Mcm2 N-terminus (Ramer et al., 2013). Although a docking interaction between DDK and the structured NTD of Mcm4 was suggested by in vitro studies with purified DDK and Mcm4 (Sheu and Stillman, 2006), we find that in the context of the  (Francis et al., 2009;Sun et al., 2014). In addition, prior phosphorylation of Mcm2-7 by other kinases has also been shown to promote DDK binding to Mcm2-7 at origins (Francis et al., 2009). Structural characterization of the DDK·Mcm2-7 DH complexes isolated here will help define the details of the DDK-Mcm2-7 interface. Such an analysis may also help resolve the mechanism of Mcm4/6 phosphorylation by DDK, which may occur across the hexamer-hexamer interface or within the hexamer bound by a DDK molecule (Sun et al., 2014).  (Mantiero et al., 2011;Tanaka et al., 2011). Similarly, retention of DDK at unfired origins has been proposed to restrict DDK targeting of Eco1 until late S phase to control cohesion in the cell cycle (Seoane and Morgan, 2017). These mechanisms imply that DDK release from Mcm2-7 DHs is regulated by origin activation. Since DDK targets Mcm2-7 DHs specifically, hexamer separation may be sufficient to induce DDK release (Francis et al., 2009;Sun et al., 2014). Alternatively, tethering of DDK to the long and flexible Mcm2 Nterminal tail may allow DDK to remain bound to replisomes. Indeed, the association of DDK with replisomes was shown to link DNA replication with meiotic recombination (Murakami and Keeney, 2014). Analogously, DDK bound to replisomes in a mitotic S phase may promote spatio-temporal coordination of DNA replication with other chromosomal processes such as chromatin assembly or DNA repair (Furuya et al., 2010;Gerard et al., 2006). It will, therefore, be interesting in future experiments to determine the dynamics of the DDK-Mcm2-7 interaction during DNA replication.
We find that activated Rad53 inhibits DDK binding to Mcm2-7 DHs by a steric mechanism that is independent of Dbf4 phosphorylation by Rad53. Dbf4 contains three evolutionarily conserved sequence motifs, termed N, M, and C that constitute parts of distinct domains separated by flexible linkers (Hughes et al., 2012;Masai and Arai, 2000). The M and C domains embrace the Cdc7 kinase to mediate Cdc7 activation, while the N motif forms part of an N-terminal BRCT domain that is dispensable for Cdc7 activation (Almawi et al., 2016;Hughes et al., 2012). Intriguingly, an N-terminal fragment encompassing the BRCT and M domains is required and sufficient to target Dbf4 to Mcm2-7 complexes loaded at replication origins (Dowell et al., 1994;Francis et al., 2009), while a physical interaction between Rad53 and the Dbf4 BRCT domain has been proposed to contribute to checkpoint-dependent origin control (Chen et al., 2013;Duncker et al., 2002;Matthews et al., 2012;Matthews et al., 2014). Thus, competitive binding of Rad53 to the Dbf4 BRCT domain is likely in part responsible for the inhibition of DDK binding to Mcm2-7 DHs observed here. The N-terminal region of Dbf4 has also been shown to interact with ORC (Duncker et al., 2002). However, consistent with a previous study in whole-cell yeast extracts (Francis et al., 2009) (Figure 7B, left). On the other hand, as DDK is monomeric in solution, both Rad53-WT and Rad53-kd can sequester DDK prior to Mcm2-7 DH binding, but the interaction of Rad53-kd with DDK is unstable (Figure 7B, right). Thus, Rad53 would block DDK binding to Mcm2-7 DHs in vivo only after checkpoint-induced activation and dimerization, as expected.
A Rad53 kinase-independent mechanism for DDK inhibition was unexpected as previous studies have demonstrated that mutation of Rad53 phosphorylation sites in Dbf4 and Sld3 allows late origin firing in the presence of HU or MMS (Lopez-Mosqueda et al., 2010;Zegerman and Diffley, 2010). It is possible that origin efficiency under this condition will further increase upon disruption of the Rad53-DDK interface. Alternatively, phosphoacceptor-site mutations in Dbf4 may also affect the physical interaction between Dbf4 and Rad53.
These possibilities remain to be tested in the future.

DDK
Two DDK variants, harboring either a removable C-terminal TAP tcp tag (Gros et al., 2014) or a ybbR-FLAG tag on Dbf4 were used interchangeably. Both variants behave identically and are fully proficient for DNA replication in vitro.
The ybbR-FLAG-tagged DDK was purified from strain YSA35. Cells were grown in 48L YP / 2 % glycerol / 2 % lactic acid pH 5.5 (YPLG) at 30ºC to a density of 2 x 10 7 cells / mL. Protein expression was induced with 2 % galactose for 4 hours. Cells were collected by centrifugation, washed with 25 mM HEPES-KOH pH 7.6 / 1 M sorbitol and resuspended in 0.5 volumes of buffer A (45 mM HEPES-KOH pH 7.6 / 0.02 % NP-40 substitute / 10 % glycerol) / 100 mM NaCl / 1mM DTT / 1x protease inhibitor cocktail (Pierce). The cell suspension was pipetted dropwise into liquid nitrogen to generate frozen popcorn and stored in -80ºC. Cells were lysed by crushing the popcorn in a Spex freezer mill, using 10 cycles of 2 minute run + 1 minute cooldown at 15 CPS.
Resulting whole cell lysate was thawed and supplemented with 1 volume of 45 mM buffer A / 100mM NaCl / 1mM DTT / 1x protease inhibitor cocktail. 5 M NaCl was added to the lysate to a final concentration to 300 mM.
After 20 minutes of gentle agitation at 4ºC, cell lysate was centrifuged in a T-647.5 rotor (Thermo Fisher) at 40,000 rpm for 1 hour at 4ºC. The clear soluble phase was recovered and DDK pulled down with 1 mL packed FLAG affinity agarose beads (Sigma) for 4 hours at 4ºC with gentle rocking. Beads were collected by centrifugation and washed with 10 volumes of buffer A / 300 mM NaCl/1 mM DTT. Beads were resuspended in 1 volume of buffer A / 300 mM NaCl / 2 mM MnCl2 / 1 mM DTT and incubated with λ protein phosphatase (NEB) at 50 U / mL for 1 hour at 23ºC with agitation. Beads were collected and protein eluted in 5 volumes of buffer A / 300 mM NaCl / 1 mM DTT supplemented with 0.25 mg / mL 3xFLAG peptide. Eluates were analyzed by SDS-PAGE. The FLAG pulldown was repeated until DDK was depleted from the extract. Fractions containing DDK were pooled, and the volume reduced to 0.5 ml using an Amicon spin concentrator (Millipore).
The pooled, concentrated eluate was fractionated on a 24 ml Superdex 200 Increase 10/300 GL (GE Healthcare) gel filtration column in buffer A / 300 mM NaCl / 1 mM DTT. Fractions were analyzed by SDS-PAGE and peak fractions containing DDK were pooled and concentrated using Amicon spin concentrator before dialysis against buffer A / 100 mM KOAc / 2 mM β-mercaptoethanol. The concentration of the purified DDK was determined by SDS-PAGE and Coomassie stain using BSA standards. Purified DDK was stored in aliquots a -80ºC.

Nhp6
Nhp6 was expressed as a N-terminal 6x His-tag fusion protein in E.coli BL21-CodonPlus (DE3)-RIL cells (Agilent). A colony of cells freshly transformed with plasmid p1035 was grown in 3 L of LB supplemented with 50 μg/mL ampicillin and 34 μg/mL chloramphenicol at 37ºC. At OD600 ~ 0.6, 1 mM IPTG was added and the temperature reduced to 4ºC. After 1 hour, the temperature was raised to 20ºC and the cells were incubated for an additional 16 hours. Cells were collected by centrifugation, rinsed twice with dH2O, once with buffer B (50 mM Tris-HCl pH 7.5 / 1 mM EDTA / 10 % glycerol / 10 mM benzamidine / 150 mM NaCl), and resuspended in buffer B supplemented with 1x protease inhibitor cocktail (Pierce) and 1 mM DTT. Cells were lysed by addition of 10 mg lysozyme (Thermo Scientific) and incubation for 30 mins at 4ºC followed by sonication. The clear, soluble phase was isolated after centrifugation of the whole-cell lysate in a T-647.5 rotor (Thermo Fisher) at 40,000 rpm for 30 minutes at 4ºC. Nhp6 was pulled down with 0.5 ml packed Ni-NTA agarose beads (Qiagen) for 3 hours at 4ºC with gentle agitation. Beads were collected and washed with 20 volumes of Buffer B / 1 mM DTT. Protein was eluted with 5 volumes of buffer B / 1 mM DTT / 100 mM imidazole, and eluates were analyzed by SDS-PAGE. Eluate fractions containing Nhp6 were pooled and the volume reduced to 0.5 ml using an Amicon spin concentrator (Millipore). The pooled concentrate was fractionated by gel filtration chromatography using a Superdex 200 10/300 GL (GE Healthcare) column in buffer B / 1 mM DTT. Elution fractions were analyzed by SDS-PAGE and Nhp6 peak fractions were pooled and concentrated using a spin concentrator. Purified Nhp6 was aliquoted, snap-frozen, and stored at -80ºC.

FACT FACT complex was purified after overexpression in yeast cells harboring codon-optimized copies of SPT16
and N-terminally CBP-tagged POB3 under control of the GAL1,10 promoter (strain YDR 125). Cells were grown in 12 L YPLG at 30ºC up to a density of 2 x 10 7 cells / mL. Protein expression was induced with 2 % galactose for 4 hours. Cells were collected by centrifugation, washed with 25 mM HEPES-KOH pH 7.6 / 1 M sorbitol, and resuspended in 0.5 volumes of Buffer C (25 mM Tris-HCl pH 7.5 / 0.02 % NP-40 substitute / 10 % glycerol) / 100 mM NaCl / 1 mM DTT / 1x protease inhibitor cocktail (Pierce). The cell suspension was pipetted dropwise into liquid nitrogen to generate frozen popcorn and stored at -80ºC. Cells were lysed by crushing in a Spex freezer mill with 10 cycles of 2 minute run and 1 minute cooldown at 15 CPS. Thawed whole cell lysate was supplemented with 1 volume of Buffer C / 100 mM NaCl / 1 mM DTT / 1x protease inhibitor cocktail and the final concentration of NaCl adjusted to 300 mM using a 5 M NaCl stock solution. After 20 minutes of gentle agitation at 4ºC, the cell lysate was centrifuged in a T-647.5 rotor (Thermo Fisher) at 40,000 rpm for 1 hour at 4ºC. The clear soluble phase was recovered and supplemented with 2mM CaCl2. FACT was pulled down from the extract with 0.5 ml packed calmodulin affinity resin (Agilent) for 4 hours at 4ºC with gentle rocking. Resin was collected by centrifugation and washed with 10 volumes of Buffer C / 300 mM NaCl / 2 mM CaCl2 / 1mM DTT. Protein was eluted from the resin with 7 volumes of Buffer C / 300 mM NaCl / 1 mM EDTA / 2 mM EGTA / 1 mM DTT, and eluates were analyzed by SDS-PAGE and Coomassie stain. The calmodulin pulldown was repeated until FACT was depleted from the extract. Fractions containing FACT were pooled and incubated for 16 hours at 4ºC with 400μg TEV protease to remove the CBP tag. The digest was diluted 3-fold with buffer D (25 mM Tris-HCl pH 7.5 / 1 mM EDTA / 10 % glycerol) / 1mM DTT to reduce the final NaCl concentration to 100 mM, and fractionated on a MonoQ 5/50 GL (GE Healthcare) column in buffer D / 1mM DTT using a gradient of 0.1 -1 M NaCl over 20 column volumes. Fractions containing FACT were pooled and concentrated to a volume of 0.5 ml using an Amicon spin concentrator (Millipore). The concentrate was gel-filtered on a 24 ml Superdex 200 10/300 GL (GE Healthcare) column equilibrated in 25 mM HEPES-KOH pH 7.5 / 1 mM EDTA / 10 % glycerol / 300 mM KOAc / 1 mM DTT. Elution fractions were analyzed by SDS-PAGE and Coomassie stain. FACT-containing peak fractions were pooled, spin-concentrated, aliquoted, snap-frozen and stored at -80ºC.

DNA beads
MCM loading, MCM phosphorylation, and MCM-DDK binding assays were performed on a linear 3 kb ARS305-containing DNA covalently linked on one end to HpaII methyltransferase and immobilized on paramagnetic beads via a 5' photocleavable biotin on the other end. The template was PCR-amplified from p470 using oligo DR772, which contains a photocleavable 5' biotin moiety, and oligo DR2417, which contains a M.HpaII-binding sequence modified with 5-fluoro-2′-deoxycytidine (BioSynthesis). The purified PCR product was coupled to Dynabeads M280 streptavidin magnetic beads (Invitrogen). M.HpaII (NEB) was conjugated to bead-bound DNA in 50 mM Tris-HCl pH 7.5, 10 mM EDTA, 100 μM SAM at a ratio of 4 units M.HpaII per 90 fmol of DNA for 16 hours at 37ºC with agitation. M.HpaII-conjugated bead-bound DNA was washed and stored in 10 mM HEPES-KOH pH 7.6 / 50 mM KOAc / 1 mM DTT at 4ºC.

Plasmids
The plasmid unwinding assay was performed on circular 3 kb ARS1-containing p79, while the in vitro DNA replication assays were performed on ARS1-containing p1017 (4.8 kbp) or ARS305-containing p470 (10 kbp) DNA. Plasmid DNAs were initially isolated using a maxiprep kit (Qiagen). To remove nicked plasmid species, purified plasmid DNA was fractionated on a 10 -40 % sucrose gradient in 20 mM Tris-HCl pH 7.5 / 1mM EDTA / 1M NaCl using an AH-629 swinging bucket rotor (Thermo Scientific) at 27,000 rpm for 20 hours at 20ºC. 0.5 ml fractions were collected and analyzed by agarose gel-electrophoresis in the absence of ethidium bromide. The gel was stained post-run with ethidium bromide. Supercoiled DNA-containing fractions were pooled, dialyzed against 10 mM Tris pH 7.5 / 2mM EDTA, concentrated using an Amicon spin concentrator (Millipore) to 1 to 2 mg/ml, and stored in aliquots at -20ºC.
Beads were magnetically separated from the supernatant and washed once with Wash Buffer / 300 mM KOAc, once with Wash Buffer / 500 mM NaCl, and once with Binding Buffer. Beads were resuspended in 20 μL Wash Buffer / 50 mM KOAc / 1 mM DTT, and the DNA eluted from the beads by exposure to UV312 nm for 10 minutes using a hand-held UV lamp. The supernatant, containing the DNA and DNA-bound proteins, was analyzed by SDS-PAGE followed by silver staining.

MCM phosphorylation assay
MCM loading was carried out as described above. Following the TEV protease or mock cleavage and wash steps, beads were resuspended in Binding Buffer / 5mM ATP / 1mM DTT and supplemented with purified DDK at 150 nM or indicated concentrations in a total volume of 40 μL. The reaction was incubated for 20 minutes at 30ºC with agitation. Beads were magnetically separated from the supernatant, and washed once with Wash Buffer / 300 mM KOAc, once with Wash Buffer / 500 mM NaCl, and once with Binding Buffer. Beads were resuspended in 20 μL Wash Buffer / 50 mM KOAc / 1 mM DTT, and the DNA was eluted from the beads by exposure to UV312 nm for 10 minutes. The supernatant, containing the DNA and DNA-bound proteins, was analyzed by SDS-PAGE and silver staining.

DDK-MCM DH binding assay
MCM loading was carried out as described above. Following the TEV protease or mock cleavage and wash steps, beads were resuspended in Binding Buffer / 5 mM ATPγS / 1mM DTT and supplemented with purified DDK at 150 nM or indicated DDK concentrations in a total volume of 40 μL. For Figures 4E, 6A and 6B, ATPγS was substituted with the indicated ATP analog. Binding reactions were incubated for 20 minutes at 30ºC with agitation. For Figures 6A & 6B, the indicated Rad53 variant was either pre-mixed with DDK or added sequentially. To pre-mix, 250 nM Rad53 was incubated with 150 nM DDK in 10 mM Mg(OAc)2 and 5 mM ATP or AMP-PNP for 20 minutes at 30ºC before adding to the resuspended beads. For sequential addition, the binding reaction was carried out as above for 10 minutes instead of 20 minutes, at which point 250 nM Rad53 was added and the reaction carried out for another 10 minutes. Beads were magnetically separated from the supernatant and washed once with Wash Buffer /100 mM KOAc. For Figure 4C, another round of TEV or mock cleavage was performed in Binding Buffer / 5mM ATPγS / 1mM DTT for 1 hour at 30ºC followed by a wash with Wash Buffer / 100mM KOAc. For Figure 4D, beads were washed in Wash Buffer with the indicated salt concentrations. Beads were resuspended in 20 μL Wash Buffer / 50 mM KOAc /1 mM DTT, and the DNA was eluted from the beads by exposure to UV312 nm for 10 minutes. The supernatant, containing the DNA and DNAbound proteins, was analyzed by SDS-PAGE and silver staining.
Right: Plot of total normalized DNA synthesis. (B) DDK titration experiment using standard DNA replication conditions, but Cdt1·MCM 2-D127 in place of Cdt1·MCM 2-WT . Template: p1017 (4.8 kbp). Left: Reaction products were analyzed by denaturing agarose gel-electrophoresis and autoradiography. Right: Plot of total normalized DNA synthesis.