Sld3-MCM Interaction Facilitated by Dbf4-Dependent Kinase Defines an Essential Step in Eukaryotic DNA Replication Initiation

Sld3/Treslin is an evolutionarily conserved protein essential for activation of DNA helicase Mcm2-7 and replication initiation in all eukaryotes. Nevertheless, it remains elusive how Sld3 is recruited to origins. Here, we have identified the direct physical association of Sld3 with Mcm2 and Mcm6 subunits in vitro, which is significantly enhanced by DDK in vivo. The Sld3-binding domain (SBD) is mapped to the N-termini of Mcm2 and Mcm6, both of them are essential for cell viability and enriched with the DDK phosphorylation sites. Glutamic acid substitution of four conserved positively charged residues of Sld3 (sld3-4E), near the Cdc45-binding region, interrupts its interaction with Mcm2/6 and causes cell death. By using a temperature-inducible degron (td), we show that deletion of Mcm6 SBD (mcm6ΔN122) abolishes not only Sld3 enrichment at early origins in G1 phase, but also subsequent recruitment of GINS and RPA during S phase. These findings elucidate the in vivo molecular details of the DDK-dependent Sld3-MCM association, which plays a crucial role in MCM helicase activation and origin unwinding.

In this study, we report that Sld3 is recruited through binding directly to the N-termini of Mcm2 and Mcm6 subunits in vitro, which are enriched with the phosphorylation sites of DDK. Indeed, their interactions are pronouncedly enhanced by DDK in vivo. According to the crystal structure of Sld3, we are able to identify that four highly conserved positively charged residues of Sld3, near the Cdc45-binding region, are important for associating with Mcm2/6 and cell viability. By utilizing a temperature-inducible degron (td) strategy, we show that the interaction defective mutations in Mcm6 (mcm6 122) interrupt the origin association of Sld3, but not Mcm2-7 per se. The subsequent recruitment of GINS to Mcm2-7 becomes abolished, which in turn compromises the Mcm2-7 helicase-driven origin unwinding as evidenced by the RPA-ChIP analysis. These data provide the detailed mechanism by which Sld3 is recruited to the Mcm2-7 complex in a DDK-dependent manner, which defines an upstream step in helicase activation and replication initiation in eukaryotes.

Sld3 Interacts with Mcm2-7 in a DDK-Dependent Manner
To investigate the molecular mechanism of Sld3 recruitment and CMS assembly, we set out to characterize the protein-protein interactions mediated by Sld3. In the yeast two hybrid assay, Sld3 interacted with Mcm2 as previously reported ( Figure 1A; Herrera et al., 2015). Meanwhile, we found that Sld3 also displayed robust interaction with Mcm6 subunit.
Since DDK has been shown to be involved in Sld3 recruitment (Heller et al., 2011;Tanaka et al., 2011), we next tested whether DDK promotes Sld3-MCM interaction. To this end, we adopted an auxin-inducible degron (aid) method to deplete cellular Dbf4 proteins (Nishimura et al., 2009). When 500 µM indole-3-acetic acid (IAA) was supplemented in the medium, yeast cells carrying the carboxy-terminal aid tagged DBF4 (dbf4-aid) failed to grow, suggesting the efficient degradation of Dbf4 proteins which are essential for cell growth ( Figure 1B). Meanwhile, SLD3 gene was tagged with 13MYC at the genomic locus, which did not alter the normal cell growth. Sld3-13MYC was immunoprecipitated with an anti-MYC antibody and detected by western blotting following SDS-PAGE. In the absence of IAA, Mcm6 was detected in the immunoprecipitate of Sld3-13MYC from dbf4-aid cells as well as from wild-type (WT) cells ( Figure 1C, lanes 4 and 6). These results corroborate the physical association between Sld3 and Mcm6 in vivo. However, after IAA treatment for 2 h, the levels of Mcm6 co-precipitated with Sld3 were significantly decreased ( Figure 1C, compare lane 5 to 6). In an independent experiment, co-precipitation of Mcm2 with Sld3 was also abrogated by the auxin-induced Dbf4 depletion (Figure 1D, compare lane 6 to 5). These data indicate that Sld3-MCM association is facilitated by DDK, through catalyzing the phosphorylation of Mcm2-7 subunits as demonstrated in a very recent study (Deegan et al., 2016).

Sld3-MCM Interaction is Essential for Cell Growth
Sld3 does not bear any apparent known phosphopeptide-binding motifs. Interestingly, it comprises two conserved positively charged regions close to each other as revealed by the crystal structure (Itou et al., 2014). One basic patch (Sld3-BP1, a.a. 301-330) mediates the interaction with an acidic region of Cdc45. Interestingly, the second basic patch (Sld3-BP2, K181, R186, K188, R192, K404, K405) is not involved in Cdc45 binding, but is also important for cell growth albeit with unknown function (Itou et al., 2014). We hypothesized that Sld3-BP2 might be involved in binding to phosphorylated MCM subunits. To test this, we first constructed the Glu-substitution mutants of the positively charged residues within Sld3-BP2. In the yeast two hybrid assay, the interaction with either Cdc45 or Mcm2/6 was not dramatically affected in the Sld3-2E(K188E, R192E) mutants ( Figure 1E). Sld3-4E (K188E, R192E, K404E, K405E) also retained positive interaction with Cdc45 ( Figure 1E, lower panel), which is consistent with the previous co-purification results (Itou et al., 2014). In contrast, Sld3-4E completely lost the interactions with both Mcm2 and Mcm6 ( Figure 1E). These results indicate that the DDK-dependent Sld3-MCM interaction is mainly mediated by the positively charged Sld3-BP2 close to the Cdc45 binding interface, which might be an evolutionarily conserved event given that these basic residues are highly conserved between Sld3 and Treslin.
We next examined the physiological role of Sld3-MCM interaction in yeast. Since SLD3 is essential for cell growth, we constructed the sld3 mutants via plasmid shuffling. Briefly, WT SLD3 was introduced by a plasmid with URA3 selective marker. The genomic copy of SLD3 was then knocked out. The sld3 mutants were expressed from a HIS3 plasmid. The URA3 plasmid can be counter-selected on a 5-fluoro-orotic acid (5-FOA) plate. Therefore, cell growth on the 5-FOA plates reflects the physiological function of the remaining sld3 copy in the HIS3 plasmid. Strikingly, sld3-4E was not able to support cell growth, whereas sld3-2E showed moderate sick growth ( Figure 1F), correlating with their ability to interact with MCM. These data suggest that Sld3-MCM interaction is essential for cell viability.

Sld3 Binds Directly with the N-Termini of Mcm2 and Mcm6
Then, we mapped the Sld3-binding domain (SBD) in Mcm2. Through construction of a series of Mcm2 truncations, we identified that a small region (a.a. 300-390) close to its N-terminus was required for interaction with Sld3 (Figure 2A). Moreover, the very N-terminal 299 amino acids were sufficient to bind Dbf4, which is consistent with a previous report that devoid of the N-terminal 63 amino acids in Mcm2 abolishes the interaction with Dbf4 (Ramer et al., 2013). These results indicate that Sld3 and Dbf4 interact with two adjacent regions within the Mcm2 amino terminus (Figure 2B) (Figures 2C,D), indicating that the SBD of Mcm6 is separable from the interface of the Mcm2-7 hexameric complex. When pull-down experiments were conducted with purified recombinant proteins, Sld3 was successfully detected together with both GST -Mcm2N (1-390) and GST -Mcm6N (1-439), indicating a direct physical association between Sld3 and Mcm2/6 N-termini ( Figure 2E). Taken together, these data suggest that both N-termini of Mcm2 and Mcm6 mediate interaction with Sld3, which is enriched with the DDK phosphorylation sites (Randell et al., 2010;Sheu and Stillman, 2010). These results are in agreement with the notion that Sld3-MCM interaction can be facilitated by DDK as shown in Figure 1 and other studies (Heller et al., 2011;Tanaka et al., 2011;Deegan et al., 2016).

Mcm6 SBD is Indispensable for Replication Initiation
Next, we asked whether the Sld3 interaction defective mutations in Mcm2/6 affect normal cell growth. To this end, we adopted a temperature-inducible degron (td) to deplete the endogenous Mcm2 proteins (Kanemaki et al., 2003). The expression of ubiquitin ligase E3 Ubr1 for td-labeled protein degradation is under control of a galactose-inducible promoter. The separationof-function mutations were introduced in a plasmid copy of MCM2. When cells were switched to the galactose plates incubated at 37 • C, the endogenous td-Mcm2 proteins were degraded and resulted in cell death ( Figure 3A). Strikingly, the lethality could be rescued by an extra copy of WT MCM2, but not an mcm2 (300-390) allele. Similarly, a Mcm6 mutant devoid

FIGURE 4 | Continued
Sld3 recruitment and GINS-MCM interaction are compromised in mcm6 N122. (A) Sld3 recruitment at early origin ARS607 is abrogated in mcm6 N122. Sld3-13MYC ChIP-qPCR was conducted as described in Experimental Procedures. A non-origin region, 17 kb downstream of ARS607, is detected as a negative control. (B) Enrichment of Mcm6 at ARS607 is not affected in mcm6 N122. Anti-Mcm6 ChIP was conducted basically as described above. (C) Interaction between Psf2 (a GINS subunit) and Mcm2 is abolished in the absence of the Sld3 binding module of Mcm6. After cultured to OD600 0.3 at 23 • C, yeast cells were arrested by α-factor in G1 at 23 • C for 3 h and released into galactose media with α-factor at 37 • C for 1 h, then released into fresh galactose media for the indicated times (S phase). The endogenous td-Mcm6 proteins were depleted when cells were shifted to 37 • C during synchronization, but not in the asynchronized samples (Asy in the box). Cell lysates were subjected to Psf2 immunoprecipitation and analyzed by western blot. (D) The origin association of Rfa1 becomes compromised in the interaction defective mutant mcm6 N122. Cells were arrested in G1 and the endogenous td-Mcm6 proteins were depleted as above. After released into fresh galactose media for another 20 min (early S phase), cells expressing WT MCM6 or mcm6 N122 were withdrawn from the culture and subjected to Rfa1-13MYC ChIP-qPCR as described in (A). of SBD, the N-terminal 122 amino acids (mcm6 N122), was not able to support yeast growth either ( Figure 3B). These results are consistent with the phenotype of the interaction defective Sld3-4E mutants described in Figure 1F. Recently, Itou et al reported a hetero-tetrameric structure of Sld3-Sld7 (Itou et al., 2015), which provides one possible scenario that two Sld3 molecules bind to Mcm2 and Mcm6, respectively. Putting together, these data suggest that both Mcm2 and Mcm6 N-termini mediated interactions with Sld3 are essential requirements for cell viability.
Since both MCM and Sld3 play essential roles in DNA replication initiation. We then examined whether the lethality of the Sld3-MCM interaction defective mutant is due to the failure in DNA replication by flow cytometry. Notably, both mcm2 (300-390) and mcm6 N122 mutant cells were significantly compromised in S phase progression, thus implying that both Mcm2-and Mcm6-mediated interactions with Sld3 are crucial for replication initiation (Figure 3C).

Mcm6 SBD Participates in Sld3 Recruitment and Origin Unwinding
To investigate the exact role of Mcm6 N-terminus medicated interaction with Sld3, we carried out three sets of experiments. First, a chromatin immunoprecipitation (ChIP) assay was performed to evaluate the enrichment levels of Sld3 at early origins in G1-arrested cells. The endogenous Sld3 proteins carrying a 13MYC tag were immunoprecipitated by an anti-MYC antibody from the formaldehyde cross-linked cell lysates. The co-immunoprecipitated DNA was analyzed by quantitative PCR (ChIP-qPCR). A significant amount of ARS607 DNA was detected in WT cells, but not in mcm6 N122 mutant ( Figure 4A). Meanwhile, the origin localization of Mcm6 itself was not changed after deletion of Mcm6 SBD (Figure 4B). These results indicate that the origin recruitment of Sld3 is inhibited when Mcm6 loses Sld3 binding capacity. The recruitment of Sld3 to pre-RC has been shown to be required for subsequent GINS recruitment (Kanemaki and Labib, 2006;Bruck and Kaplan, 2015a,b). Therefore, Psf2, a subunit of GINS complex, was examined for its association with MCM during the cell cycle. G1-arrested cells were harvested every 20 min after release into the S phase. Cell lysates were subjected to anti-Psf2 immunoprecipitation. In WT, the amounts of Mcm2 coprecipitated with Psf2 were gradually increased with the S phase progression ( Figure 4C). Strikingly, there was barely detectable Mcm2 in the precipitate of mcm6 N122 (Figure 4C). On the contrary, if the endogenous Mcm6 was not depleted like in the asynchronized cells, Mcm2 was detected normally (lane 7). These results suggest a severe defect in the assembly of active CMG.
Third, we asked whether origin unwinding by MCM helicase depends on Mcm6 SBD. When duplex origin DNA is unwound by activated MCM helicase, the single-stranded DNA (ssDNA) is immediately protected by RPA. Therefore, RPA localization at origins can reflect the MCM-created ssDNA levels in vivo. The lysates were prepared from cells synchronized at G1 or early S phase (20 min after release from G1). Rfa1-3HA was precipitated from the lysates of G1 or S phase cells by an anti-HA antibody. Coprecipitated DNA was measured by qPCR. The endogenous Mcm6 was depleted by the td degron as described above. In the presence of WT MCM6, the relative enrichment of Rpa1 at early origin ARS607 substantially increased after release from G1 phase for 20 min ( Figure 4D).
Nevertheless, in mcm6 N122, the signals of Rpa1 at this origin were about 50% lower compared to the ones in WT, indicating a severe defect in the origin recruitment of RPA ( Figure 4D). These results suggest that origin unwinding by MCM helicase activity is dependent on the Sld3-MCM interaction.
In this study, we provide the molecular details of the DDKdependent recruitment of Sld3 to the MCM complex. These findings, together with the previous notion of Sld3-dependent Cdc45 and GINS loading (Kanemaki and Labib, 2006;Bruck and Kaplan, 2015b;Herrera et al., 2015;Deegan et al., 2016), shed new light on the assembly of the CMS complex, which defines a crucial transient step upstream of the recruitment of GINS and assembly of the active form of the replicative helicase, the CMG complex, for duplex DNA unwinding and replication initiation.

EXPERIMENTAL PROCEDURES Yeast Strains and Basic Methods
The yeast strains used in this study are derived from W303-1a (Table 1) and constructed basically as previously described (Lou et al., 2008). Plasmids constructed for this study are listed in Table 2 and validated by sequencing. Yeast two-hybrid, plasmid shuffling, cell synchronization and flow cytometry analysis were performed as described in (Quan et al., 2015).

Immunoprecipitation (IP)
Immunoprecipitation (IP) was carried out basically as described previously with some modifications (Lou et al., 2008). Briefly, yeast cultures were arrest at G1 phase with α-factor for 3 h. Cells (4 × 10 8 ) were crosslinked and collected and lysed at 4 • C with glass beads (BeadBeater) in IP buffer [45 mM HEPES, pH 7.2,  (Roche)]. Whole cell extract was mixed with 2 µl indicated antibodies and rotated for 3 h at 4 • C, and then 20 µl of protein G beads were added and incubated for another hour. Beads were then washed three times with 1 ml IP buffer and boiled in 50 µl SDS-sample buffer. Western analysis was performed to detect specific proteins. Blots were probed with the indicated antibodies in phosphate-buffered saline containing 0.1% Tween and 2% dried milk. 9E10 (1:1000) was used to detect an MYC tag, M2 (1:1000) to detect Flag-tagged proteins, 12CA5 (1:1000) to detect an HA epitope. Polyclonal sera against Mcm2 (1:10000) and Mcm6 (1:10000) was used to detect corresponding proteins.

Chromatin Immunoprecipitation (ChIP-qPCR)
ChIP experiments were performed with extracts of formaldehyde cross-linked cells, using 1.0 µg of purified anti-Sld3 or anti-Mcm6 antibodies (Tanaka et al., 2011). After The DNA fragments in precipitates were quantified by real-time PCR for each genomic locus. The ratio of immunoprecipitated DNA to total DNA in input was normalized and shown as fold enrichment (Natsume et al., 2013).