ATP-dependent Pre-replicative Complex Assembly Is Facilitated by Adk1p in Budding Yeast*

Pre-replicative complex (pre-RC) assembly is a critical part of the mechanism that controls the initiation of DNA replication, and ATP binding and hydrolysis by multiple pre-RC proteins are essential for pre-RC assembly and activation. Here, we demonstrate that Adk1p (adenylate kinase 1 protein) plays an important role in pre-RC assembly in Saccharomyces cerevisiae. Isolated from a genetic screen, adk1G20S cells with a mutation within the nucleotide-binding site were defective in replication initiation. adk1Δ cells were viable at 25 °C but not at 37°C. Flow cytometry indicated that both the adk1-td (temperature-inducible degron) and adk1G20S mutants were defective in S phase entry. Furthermore, Adk1p bound to chromatin throughout the cell cycle and physically interacted with Orc3p, whereas the Adk1G20S protein had a reduced ability to bind chromatin and Orc3p without affecting the cellular ATP level. In addition, Adk1p associated with replication origins by ChIP assay. Finally, Adk1-td protein depletion prevented pre-RC assembly during the M-to-G1 transition. We suggest that Adk1p regulates ATP metabolism on pre-RC proteins to promote pre-RC assembly and activation.

Mutation analysis showed that all conserved ATP-binding motifs of these proteins are essential for cell viability (12,17,18).
There have been numerous reports on the roles of binding and hydrolysis of ATP in pre-RC assembly. In S. cerevisiae, ATP binding of Orc1p activates ORC assembly and ORC binding to replication origins (1). Mutation within the Orc1p ATP-binding site prevents ORC binding to DNA in vitro and is lethal in vivo (19). Other studies indicated that mutations in the Walker A motif eliminate the ATP-binding and hydrolysis activities of Orc1p (20,21). When ATP binds to Orc1p, an initial round of chromatin loading of MCM proteins is permitted, whereas ATP hydrolysis is required for other rounds of MCM loading in vitro (13,22). Human ORC assembly in vitro is dependent on ATP binding and impaired by mutations in Orc4p or Orc5p ATP-binding sites (23,24). In S. cerevisiae, Cdc6p binding to ORC, which is dependent on the Cdc6p ATPase activity, changes the ORC structure and contributes to pre-RC assembly (25). Cdc6p mutated in the Walker B motif cannot interact with Orc1p in vitro (26) and shows decreased chromatin loading and lethality in vivo (18). A similar mutation in human CDC6 also eliminates its ATP-binding and hydrolysis activities (27). However, the enzyme(s) that may regulate ATP metabolism during pre-RC assembly has not been reported.
Adenylate kinases are phosphotransferases that catalyze the interconversion reaction of ATP ϩ AMP 7 2ADP and control nucleotide metabolic processes and thus the cell growth rate in eukaryotes (28). Adk1p is important for cell proliferation but not essential for cell viability by gene disruption analysis in S. cerevisiae (29), and two isozymes of Adk1p, termed Adk2p and Ura6p, have been found (30 -32). In Schizosaccharomyces pombe, the Adk1p homolog is required for cell viability (33). In mammalian tissues, six isozymes, designated AK1-AK6, have been identified and characterized (34 -36). These adenylate kinases show functional similarities in ATP metabolism.
From a random ethane methyl sulfonate mutagenesis followed by a phenotypic screen (which we called the initiation of DNA replication, or IDR, screen) to identify genes involved in or regulating replication initiation in S. cerevisiae (37,38), 4 we isolated an adk1 G20S mutant that loses a single-ARS plasmid at a high rate and a multiple-ARS plasmid at a reduced rate. We show that both the adk1-td and adk1 G20S mutants have replication initiation defects, suggesting that Adk1p plays an important role in DNA replication initiation. Furthermore, we dem-onstrate that Adk1p binds to pre-RC components and replication origins and becomes essential for pre-RC assembly and cell viability at 37°C.
The anti-HA antibody (12CA5) was from Roche Applied Science. Anti-ORC and anti-MCM antibodies were gifts from Bruce Stillman (Cold Spring Harbor Laboratory).
Protein Extraction, Co-immunoprecipitation, Immunoblotting, ChIP Assay, and ATP Assay-Protein extraction, co-immunoprecipitation, and ChIP assays were performed as described (4,42). ATP assays were performed as described (43). After the yeast cell wall was digested (4), the ATPlite TM assay system (PerkinElmer Life Sciences) and 10 7 cells/sample were used for each measurement.
Cell Synchronization, FACS Analysis, and Chromatin Binding-Cell cycle block and release with ␣-factor, hydroxyurea, or nocodazole were carried out as described (4). adk1-td cells were cultured to early log phase and then arrested with the cell cycle inhibitors in yeast/extract/peptone/dextrose (YPD)-or synthetic complete medium (SCM)-based selective medium containing 0.1 mM CuSO 4 at 25°C. YPRG medium (10 g/liter yeast extract, 20 g/liter peptone, 20 g/liter raffinose, and 5 g/liter galactose) without CuSO 4 was then used to induce GAL-UBR1 expression at 25°C for 1 h and to degrade the Adk1-td protein at 37°C for 1 h. FACS analysis and chromatin binding assays were performed as described (4,42,44).

adk1 G20S Mutant Cells Have Defects in DNA Replication
Initiation-We carried out a sensitive yeast phenotypic screen with randomly mutagenized yeast cells to identify proteins related to replication initiation using a pair of tester plasmids, p1ARS and p8ARSs (4). It is known that mutants in genes that function in or regulate DNA replication initiation exhibit high plasmid loss rates in p1ARS transformants and lower plasmid loss rates in p8ARSs transformants (4,6,(45)(46)(47). Therefore, we employed these plasmids to identify mutants defective in DNA replication initiation. Among many mutants in known and unknown replication initiation proteins (37,38), 4 an adk1 G20S mutant was identified to be a replication initiation mutant.
The original adk1 mutant isolated from our screen contains two point mutations: G20S within the putative nucleotidebinding site ( 13 GPPGAGKST 21 ) (48) and P138L. We separated the two mutations, integrated them separately into an adk1⌬ strain, and then examined the plasmid loss rates of the integrants using a colony color sectoring assay (49) and a colony size assay. Cells in the ade2-1 ade3-1 background turn red on nonselective YPD plates when they contain p1ARS or p8ARSs bearing the reporter gene ADE3-2. In the colony color sectoring assay, the white/red sectoring morphology of the colony indi-

. adk1 G20S mutant cells have replication initiation defects, and
Adk1p is essential for cell viability at 37°C. A, colony color sectoring assays of different adk1 mutants. The wild-type ADK1 strain (YL36) and different adk1 mutants transformed with p1ARS and p8ARSs separately were streaked onto YPD plates and incubated at 25°C for 3-5 days to form colonies. B, colony size assays of the adk1 G20S mutants. adk1⌬ cells containing pRS416-adk1 G20S were transformed with p1ARS and p8ARSs separately, spread onto the plasmid loss testing plate (SCMϪUraϪLeu) and the control SCMϪUra plate (pRS416 contains the URA3 marker), and incubated at 25°C for 5 days to allow colony formation. C, quantitative plasmid loss assay for wild-type and adk1 G20S mutant cells. D, 10-fold serial dilutions of log phase adk1⌬ cells were spotted onto YPD plates and incubated at 25 and 37°C to test cell viabilities. E, adk1-td cells transformed with pRS416 or pRS416-ADK1 were streaked onto a SCM-Ura/dextrose (SCM-Ura/D) plate at 25°C and a SCMϪUra/raffinose/galactose (SCMϪUra/RG) plate at 37°C to test cell viabilities.
cates the plasmid loss rate of the cells in the colony (red cells contain the plasmid, and white cells have lost the plasmid). The adk1 G20S mutant lost p1ARS at a high rate and p8ARSs at a lower rate, whereas the adk1 P138L mutant lost both plasmids at low rates (Fig. 1A). Therefore, the G20S (but not P138L) mutation was responsible for the replication initiation phenotypes of the original adk1 mutant isolated from the screen.
In the colony size assay, the background of the mutants is leu2-1; thus, adk1⌬/pRS416-adk1 G20S cells (the pRS416 vector contains the URA3 marker) need to carry p1ARS or p8ARSs bearing the LEU2 marker to grow on SCMϪUraϪLeu (synthetic complete medium lacking uracil and leucine) plates. When mutant cells have replication initiation defects, p1ARS transformants form small colonies because of the high rate of p1ARS loss, whereas p8ARSs transformants form bigger ones. Consistent with results from the colony color assays, the adk1 G20S mutant exhibited smaller colony size with p1ARS than p8ARSs on SCMϪUraϪLeu plates, whereas the difference in colony size disappeared on SCMϪUra plates, which were nonselective for LEU2 (Fig. 1B). Furthermore, quantitative plasmid assays (Fig. 1C) confirmed the plasmid loss phenotypes of the adk1 G20S mutant. These results suggest that the adk1 G20S mutant is defective in DNA replication initiation. ADK1 Becomes Essential for Viability at 37°C-Consistent with a previous report that ADK1 is important for growth but not essential for viability at normal growth temperatures (29), the adk1⌬ cells that we generated could grow at 25°C, but they grew more slowly than the wild-type cells (Fig. 1D). However, the adk1⌬ cells were not viable at 37°C (Fig. 1D), indicating that ADK1 is essential for cell viability at 37°C. To further confirm this conclusion, we generated a temperature-inducible degron adk1-td strain, transformed the adk1-td cells with pRS416-ADK1 or with the pRS416 vector alone, and then tested the transformants at 25 and 37°C. The adk1-td cells transformed with pRS416 were viable at 25°C but not at 37°C, whereas the adk1-td cells transformed with pRS416-ADK1 grew normally at both temperatures (Fig. 1E). These results confirm that ADK1 is essential for cell viability at 37°C. adk1 G20S Mutant and Wild-type ADK1 Cells Have Similar in Vivo ATP Levels and Growth Rates-To rule out the possibility that the plasmid loss phenotypes of the adk1 G20S mutant resulted from an altered cellular ATP level, we measured the in vivo ATP levels in wild-type, adk1 G20S , and adk1⌬ log phase cells. Wild-type and adk1 G20S mutant cells had similar ATP levels in vivo, whereas the adk1⌬ cells had a much lower ATP level ( Fig. 2A). These results indicate that the replication initiation defects of the adk1 G20S cells did not result from ATP homeostasis imbalance.
We also measured the cell growth rates of wild-type ADK1 and the adk1 mutants at 25°C. Compared with the wild-type cells, the growth rates of the adk1 G20S and adk1 P138L mutants were only slightly reduced (Fig. 2B). Together with the data on the cellular ATP levels, these results support the notion that the adk1 G20S mutant cells exhibit allele-specific defects in replication initiation not being associated with ATP homeostasis or growth rate.
Adk1p Is Required for the G 1 -to-S Transition at 37°C-To examine the role of Adk1p in the cell cycle, we examined the DNA contents of wild-type, adk1-td, and adk1 G20S cells by flow cytometry. The cells were first arrested in G 1 phase by ␣-factor. After Adk1-td protein degradation in galactose-containing medium (galactose induces the overexpression of Ubr1p to facilitate td protein degradation (40)) at 37°C, the adk1-td cells were released from the G 1 block into fresh medium at 37°C. Control experiments were performed with glucose-containing medium at 25°C. The adk1-td cells at 25°C (Fig. 2C) and the , the cell growth rates of the adk1⌬ cells transformed with pRS416-ADK1, pRS416-adk1 G20S , or pRS416-adk1 P138L in SCMϪUra medium at 25°C were measured at A 600 . C and D, wild-type, adk1-td, and adk1 G20S cells were arrested in G 1 phase by ␣-factor at 25°C. After Adk1-td protein degradation in YPRG medium at 37°C, cells were released from the ␣-factor block into fresh YPRG medium at 37°C (D). Cells in the control experiments were cultured in YPD medium at 25°C (C). The adk1 G20S mutant cells were cultured in YPD medium at 25°C (C) or 37°C (D). Aliquots of cells were analyzed by FACS and bud counting. Asy., asynchronous.
wild-type cells at both temperatures (Fig. 2, C and D) entered and completed S phase after release from the ␣-factor block. In contrast, the adk1-td cells at 37°C were mostly defective in S phase entry, with only a small fraction of the cells starting to enter S phase at 150 min after ␣-factor removal (Fig.  2D). These results suggest that Adk1p is essential for the G 1 -to-S transition at 37°C. The adk1 G20S mutant cells showed a notable delay in the G 1 -to-S transition compared with the wild-type cells at both 25 and 37°C (Fig. 2, C and D). On the other hand, Adk1p was not required for S phase progression after the cells were released from the early S phase block by hydroxyurea (supplemental Fig. S1, A and B) or for mitosis after the cells were released from the G 2 /M block by nocodazole (supplemental Fig. S1, C and D) at 25 or 37°C.

Adk1p Binds to Chromatin throughout the Cell Cycle, and the adk1 G20S Mutant Protein Has Reduced Chromatin Binding-
To determine whether Adk1p binds chromatin for its role in replication initiation and to understand the basis for the adk1 G20S mutant phenotypes, we carried out chromatin binding assays with ADK1-HA and adk1 G20S -HA cells. In the asynchronous cells, the Adk1 G20S protein had much reduced chromatin binding compared with wild-type Adk1p using Orc3p as the loading control (Fig. 3A). To examine the cell cycle pattern of chromatin association of the Adk1p and Adk1 G20S proteins, ADK1-HA and adk1 G20S -HA cells were arrested in G 1 phase by ␣-factor and then released into fresh medium at 25°C. Samples were collected for chromatin binding assays at different time points after release. Adk1p-HA was more or less constant in both the supernatant and chromatin fractions throughout the cell cycle, whereas the constant chromatin binding of Orc3p and the cell cycle-regulated chromatin association of Mcm2p were as expected ( Fig. 3B; see the accompanying FACS data in Fig. 3C). In the adk1 G20S -HA cells, Orc3p was constant, whereas the release of Mcm2p from chromatin was delayed (Fig. 3D), in accordance with the slower cell cycle progression of the adk1 G20S -HA cells (Fig.  3E) compared with the wild-type cells (Fig. 3C). Consistent with the results from the asynchronous cells, the Adk1 G20S -HA protein was reduced in the chromatin fractions and increased in the supernatant fractions (Fig. 3D) compared with the wild-type Adk1p protein (Fig. 3B).

Adk1p Physically Interacts with Orc3p, and the Adk1 G20S Mutant Protein Has Mostly Lost the Interaction with Orc3p-
To test if Adk1p physically interacts with ORC, co-immunoprecipitation assay was carried out with asynchronous ADK1-HA and adk1 G20S -HA cells. Orc3p could be co-immunoprecipitated with Adk1p-HA by the anti-HA antibody (Fig. 3F, upper  panel, lane 6) but not by the control mouse IgG (lane 5) from the ADK1-HA cell extracts. An untagged strain gave a negative co-immunoprecipitation signal as expected (Fig. 3F, lane 3). For the adk1 G20S -HA cells, no obvious co-immunoprecipitation of Orc3p could be detected (Fig. 3F, lower panel, lane 6). These results suggest that Adk1p physically interacts with ORC in vivo and that the Adk1 G20S mutant protein has mostly lost this interaction.
Adk1p Binds to Replication Origins in Vivo-After knowing that Adk1p binds to chromatin and interacts with Orc3p, we examined if the Adk1-HA protein associates with replication origins in vivo by ChIP assay with the anti-HA antibody and with the anti-Orc3 antibody and mouse IgG as the positive and negative controls, respectively. Specific PCR primer pairs were  SEPTEMBER 24, 2010 • VOLUME 285 • NUMBER 39 used to amplify two replication origins, ARS1 and ARS501, and their corresponding non-ARS control sequences, R2.5 and (501ϩ11 kb), respectively. When the anti-Orc3 (Fig. 3G, lane 4) or anti-HA (lane 5) antibody was used, ARS1 and ARS501, but not the non-ARS control regions, were detected in the immunoprecipitated chromatin DNA. The negative control for PCR without DNA template (Fig. 3G, lane 2), the untagged strain control (lane 3), and the IgG control (lane 6) produced no ChIP signal. Together with the results of interaction of Adk1p with chromatin and ORC, these ChIP data suggest that Adk1p binds to replication origins probably through its interaction with ORC and perhaps other pre-RC proteins.

Adk1p Facilitates Pre-RC Assembly
Adk1p Is Required for Pre-RC Assembly at 37°C-During the Mto-G 1 transition, MCM proteins, known as the last markers for pre-RC assembly, are loaded onto chromatin at replication origins. We examined the in vivo pre-RC assembly efficiency by detecting Mcm2p and Cdc6p on chromatin when the Adk1p-td protein was depleted. The adk1-td cells were arrested at the G 2 /M boundary by nocodazole. After Adk1-td protein degradation, the cells were released from the G 2 /M block into G 1 phase in ␣-factor-containing medium at 37°C and then harvested at various time points for chromatin binding assays (Fig. 4A) and for FACS analysis to monitor cell cycle progression (Fig. 4B). In the control experiment, in which the Adk1-td protein was not degraded at 25°C, Mcm2p and Cdc6p were successfully loaded onto chromatin as expected (Fig.  4A, lanes 1-5). In contrast, when the Adk1p-td protein was degraded at 37°C, Mcm2p and Cdc6p were mostly absent on chromatin (Fig.  4A, lanes 6 -10), suggesting that the chromatin association of MCM proteins and Cdc6p is significantly impaired. Orc3p was constant in the chromatin fractions at different time points at both temperatures. In the supernatants, Orc3p and Mcm2p were constant at both temperatures, whereas Cdc6p was cell cycle-regulated as expected (Fig. 4A,  lanes 1-10). These results reveal that Adk1p is critical for pre-RC assembly at 37°C without affecting the levels of MCM proteins and Cdc6p in vivo.
We also examined the efficiency of pre-RC assembly during the M-to-G 1 transition in the adk1 G20S -HA cells. Compared with the wild-type cells at 25 and 37°C, Mcm2p chromatin loading was delayed and reduced in the adk1 G20S -HA cells ( Fig. 4C; see FACS data in D and E), suggesting that the Adk1 G20S mutant protein is partially defective in promoting pre-RC formation.
Adk1p Is Dispensable for Pre-RC Maintenance in G 1 Phase at 37°C-In G 1 phase, MCM proteins are maintained on chromatin. To determine whether Adk1p is required for pre-RC maintenance in G 1 phase, we examined Mcm2p in the chromatin FIGURE 4. Adk1p is essential for pre-RC assembly in vivo at 37°C, and the adk1 G20S mutant cells are partially defective in pre-RC assembly. A-E, adk1-td (A and B), wild-type ADK1 (C and D), and adk1 G20S -HA (C and E) cells were arrested at the G 2 /M boundary in YPD medium containing nocodazole (Noc.) at 25°C. After the Adk1-td protein was depleted in YPRG medium at 37°C for 1 h, adk1-td cells were released into ␣-factorcontaining YPRG medium at 37°C. Wild-type ADK1 and adk1 G20S -HA cells were shifted to 37°C for 1 h before being released from the nocodazole block into ␣-factor-containing YPD medium at 37°C. Control cells were kept in YPD medium at 25°C. Cells at various time points after release from the nocodazole block were harvested for chromatin binding assays (A and C) and FACS analysis (B-E). F and G, adk1-td cells were arrested in G 1 phase with ␣-factor-containing YPD medium at 25°C. After Adk1-td protein degradation in YPRG at 37°C, the cells were kept in G 1 phase at 37°C for chromatin (Chr.) binding assays (F) and FACS analysis (G). Control cells were kept in ␣-factor-containing YPD medium at 25°C. Pel., pellet; Sup., supernatant.

Adk1p Facilitates Pre-RC Assembly
fractions when Adk1p-td was depleted in ␣-factor-containing medium at 37°C. Mcm2p was stable in both the chromatin and supernatant fractions when the Adk1p-td protein was stable at 25°C or degraded at 37°C (Fig. 4F; see FACS data in Fig. 4G), suggesting that the chromatin association of MCM proteins in G 1 phase does not require Adk1p. Because Adk1p is required for the G 1 -to-S transition at 37°C (Fig. 2D), we suggest that ATP metabolism is important for pre-RC activation and/or the transition from pre-RC to the preinitiation complex in addition to its function in pre-RC assembly.

DISCUSSION
Our study reveals that Adk1p facilitates the ATP-dependent pre-RC assembly and that Adk1p becomes essential for pre-RC assembly and cell viability at 37°C. The data from the plasmid loss assays show that the adk1 G20S mutant is specifically defective in DNA replication initiation. On the other hand, the adk1 G20S mutant and wild-type cells have similar in vivo ATP levels and growth rates, indicating that the replication initiation defects of the adk1 G20S mutant do not result from an ATP homeostasis imbalance. We also show that Adk1p physically interacts with ORC, chromatin, and replication origins and that Adk1p is essential for pre-RC assembly and activation at 37°C. These physical interaction and functional data support a direct role of Adk1p in promoting pre-RC assembly and activation.
Our data also indicate that the Adk1 G20S mutant protein has weakened association with chromatin and ORC compared with wild-type Adk1p. These results provide the probable molecular basis for the defects of the adk1 G20S mutant cells in pre-RC assembly and replication initiation. In addition, in a study unrelated to DNA replication, the Adk1 G20S mutant protein has been reported to have a lower adenylate kinase activity than wild-type Adk1p (48). Therefore, we suggest that the Adk1 G20S mutant protein has a reduced ability to regulate ATP metabolism at pre-RCs due to the impairment of both its enzymatic activity and its association with ORC and perhaps other pre-RC proteins, resulting in decreased efficiencies of pre-RC assembly and activation.
Adk1p has two isozymes, Ura6p and Adk2p, and Ura6p can partly compensate for the loss of function of Adk1p in S. cerevisiae (31,32). Because adk1⌬ cells are viable with growth defects under normal growth conditions, it is possible that the physiological role of Adk1p in ATP-dependent pre-RC assembly can be partially compensated by Ura6p and/or Adk2p at normal growth temperatures when the Adk1p activity is absent. This is consistent with the report that Adk1p contributes ϳ90% to the total adenylate kinase activity in yeast cells (29). We have shown that Adk1p becomes essential for pre-RC assembly and for cell viability at 37°C, whereas it is not required for S phase progression or mitosis, consistent with Adk1p playing a critical role in pre-RC assembly. It is possible that a higher cellular adenylate kinase activity is needed for faster ATP metabolism at 37°C; thus, the 10% remaining adenylate kinase activity provided by Ura6p and Adk2p is not sufficient for cell viability.
In conclusion, we propose that Adk1p may catalyze the interconversion reaction of ATP ϩ AMP 7 2ADP and regulate ATP metabolism directly on pre-RC proteins so as to facilitate the ATP binding-and hydrolysis-dependent pre-RC assembly and activation. Adk1p has also been reported to interact with Cdc14p (50), which dephosphorylates multiple pre-RC components and is required for pre-RC assembly (38); hence, it is also possible that Adk1p may regulate phosphorylation and dephosphorylation processes of pre-RC components during pre-RC assembly. This study reveals a novel function of Adk1p in the ATP-dependent pre-RC assembly and opens a new avenue to investigate the mechanism and cell cycle control of DNA replication initiation.