A Second Human Dbf4/ASK-related Protein, Drf1/ASKL1, Is Required for Efficient Progression of S and M Phases*

Cdc7-Dbf4 kinase is conserved through evolution and regulates initiation and progression of DNA replication. In human, ASK/hsDbf4 binds and activates huCdc7 during S phase and this kinase complex is essential for DNA replication and cell proliferation. Drf1/ASKL1, a second human Dbf4/ASK-related protein, shares three conserved Dbf4 motifs previously identified on all of the Dbf4-related molecules. Drf1/ASKL1 can bind and activate huCdc7, and Cdc7-ASKL1 complex phosphorylates MCM2. ASKL1 transcription and protein levels oscillate during cell cycle and increase at late S to G2/M phases. The protein is detected predominantly in the nuclear-soluble fraction but not in the chromatin-bound fraction. Inhibition of Drf1/ASKL1 expression by siRNA results in attenuation of cell growth and in the increase of late S and G2/M phase population. siRNA treatment on synchronized cell population revealed that S phase progression is delayed when ASKL1 protein level is decreased. S phase delay may be linked to replication fork block, because increased levels of γH2AX and activated form of Chk2 are detected with ASKL1 siRNA in the absence of any additional DNA damages. Furthermore, mitotic progression is retarded in ASKL1 or Cdc7 siRNA-treated cells. Our results suggest that ASKL1 in a complex with Cdc7 may play a role in normal progression of both S and M phases.

Cell cycle progression of eukaryotic cells is strictly regulated by a series of phosphorylation events. Among them, multiple cyclin-dependent kinases play crucial roles to facilitate the progression of cell cycle at various stages. Multiple cyclins and cyclin-dependent kinases have been identified in various eukaryotes, each acting at specific stages of the cell cycles (1)(2)(3)(4)(5)(6)(7). Conserved cyclin-box sequences have been identified on the cyclin molecules, and significant homology is present in the kinase-conserved domains of various cyclin-dependent kinases (8,9).
Cdc7-Dbf4 is another family of serine-threonine kinases originally identified in Saccharomyces cerevisiae (10,11). Recent studies have shown conservation of Cdc7-Dbf4 kinases throughout evolution and their essential roles in DNA replication in higher eukaryotes (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23). In fission yeast, the presence of a second set of Cdc7-Dbf4 kinase, Spo4-Spo6, was reported (24,25), whereas only one set of Cdc7-Dbf4 has been identified on the genome of budding yeast. Spo4 and Spo6 were shown to play essential roles during meiosis but not during the mitotic cell cycle, in contrast to Hsk1-Dfp1/Him1, which is essential for cell viability. The presence of more than one Cdc7-Dbf4-related kinase in a single eukaryotic species suggests a possibility that they also constitute a novel kinase family, each member of which may play distinct roles in cell cycle progression.
We have identified a novel human cDNA, designated ASKL1, 1 which shares significant homology with Dbf4 protein family. ASKL1 is identical to Drf1 reported recently by Montagnoli et al. (26). It was reported that Drf1 binds to huCdc7 and activates its kinase activity and that its transcription and protein levels increase during late S to G 2 /M. The protein exists in nuclei, and its role in S phase progression is speculated (26). However, the functional analyses of Drf1 have not been conducted and its precise roles during cell cycle remained elusive. Therefore, we have undertaken studies aimed at clarifying the roles of the Cdc7-ASKL1 complex during cell cycle regulation of human cells.
We have found that Drf1/ASKL1 is primarily recovered in nuclear-soluble fractions but is not associated with Tritoninsoluble chromatin fractions. To elucidate biological functions of the Cdc7-ASKL1 complex, we examined the effects of downregulation of Drf1/ASKL1 expression by siRNA on cell cycle progression. Inhibition of Drf1/ASKL1 expression resulted in retardation of both S phase and M phase progression. We discuss possible mechanisms of how Drf1/ASKL1 may regulate the cell cycle through association with huCdc7 kinase subunit.

EXPERIMENTAL PROCEDURES
Development of Antibody against ASKL1 Protein-DNA encoding a polypeptide of ASKL1 (residues 211-355; minimum ASKL1) was amplified by PCR (5Ј-CGGGGATCCCCGGAACATGTCCAGCAGCA-3Ј and 5Ј-CGGAATTCTCACCTGGGGAGGCCAGCCTG-3Ј) and cloned into the BamHI-EcoRI site of pGEX-5X-3 to generate a GST-ASKL1[minimum] fusion protein. The protein was insoluble after overexpression and purified from inclusion bodies by extraction into SDS-PAGE running buffer followed by dialysis against PBS containing 0.1% SDS. The antibody was developed in rabbit, and the serum was run through a GST-affinity column to remove GST-reactive antibody and was further affinity-purified by adsorption to polyvinylidene difluoride membrane to which purified His-tagged ASKL1c protein (448 amino acids, also see Supplemental Fig. S2B) was transferred. The ASKL1c cDNA was amplified by 5Ј-GAGGATCCCATATGAGCGAACCGG-GAAAGGGAGAC-3Ј and 5Ј-CGAATTCTAGTGATGATGGTGATGAT-GACAGGAATGTCCGCCGGGGCA-3Ј to be subcloned into a T7 promoter vector, pT7-7. The antibody was eluted from the membrane by 0.2 M glycine, pH 2.8, and 1 mM EGTA, and the pH of the eluate was adjusted by addition of 0.1 volume of 1 M Tris base.
Establishment of Stable Cell Lines Expressing GST-ASKL1-A NotIfragment containing GST-ASKL1[minimum] or GST coding region was cloned at the NotI site of pKU3 vector, and the resulting plasmid DNAs were transfected into HeLa cells with Lipofectamine Plus reagent (Invitrogen). After 2 days of incubation, cells were grown in a medium containing 800 g/ml G418 for 10 days and the surviving colonies were selected. They were further cloned and amplified to establish HeLa cells constitutively expressing GST-ASKL1 or GST protein.
Synchronization of Cell Cycle and Cell Fractionation-Cell cycle of HeLa cells was synchronized either by release from G 1 /S arrest or that from mitotic arrest. For G 1 /S arrest, the cells were treated with 2.5 mM thymidine for 14 -16 h twice with a 9-h interval of growth without the drug. For mitotic arrest, the cells were first treated with 2.5 mM thymidine for 16 h and then treated with 50 ng/ml nocodazole for 8 or 10 h. The cells then were incubated in the absence of the drug for indicated times and harvested. Half of the cells were resuspended in 70% ethanol and were processed for FACS analyses of cell cycle distribution as described previously (21), and the remaining half was used for cell fractionation. To quantify DNA content histogram, FACS data were analyzed by ModFit LT software (Verity Software House). To fractionate cells into Triton-soluble and Triton-insoluble fractions, the cells were resuspended in CSK buffer (20 mM HEPES-KOH, pH 7.6, 40 mM potassium glutamate, 1 mM MgCl 2 , 1 mM EGTA, 300 mM sucrose, 1 mM dithiothreitol, 1 mM Na 3 VO 4 , and protease inhibitors) containing 0.1% Triton X-100, incubated on ice for 20 min, and centrifuged at 3,000 rpm for 3 min to obtain Triton-soluble extracts. The Triton-insoluble pellet enriched for chromatin-bound proteins after being washed once with the CSK buffer was resuspended in the same buffer.
To fractionate cells into cytoplasm, nuclear-soluble and nuclear pellet fractions, cells washed with hypotonic buffer (20 mM HEPES-KOH, pH 7.6, 5 mM potassium acetate, 0.5 mM MgCl 2 , and 0.5 mM dithiothreitol), were homogenized in a minimal volume of the same buffer followed by centrifugation at 5,000 rpm for 3 min to separate the supernatant (cytoplasm) and pellet (nuclei). The pellet was resuspended in CSK buffer containing 0.5% Triton X-100 and incubated on ice for 10 min followed by centrifugation at 3,000 rpm for 3 min to separate the supernatant (nuclear-soluble fractions) and pellet (chromatin-enriched nuclear-insoluble fractions). The pellet was washed once with CSK-Triton X-100 buffer and resuspended in the same buffer.
Detection of BrdUrd Incorporation-After treatment with siRNA and synchronization of HeLa cells, BrdUrd was added at 10 M and incubation was continued for 30 min before cell harvest at each time point. For immunostaining of BrdUrd-incorporated DNA, recovered cells were fixed with 50 mM glycine, 70% ethanol, pH 2.0, at Ϫ30°C for 30 min and washed with PBS. Immunostaining was conducted as recommended by the manufacturer (5-bromo-2Ј-deoxy-uridine labeling kit, Roche Applied Science). DNA was costained with 4,6-diamidino-2-phenylindole (DAPI). The coverslips then were washed with PBS and mounted with PBS containing 90% glycerol, 2.5% 1,4-diazabicyclo[2.2.2] octane (DABCO, Sigma). Samples were observed under a fluorescence microscope (Carl Zeiss Axiophot) equipped with a Hamamatsu ORCA-ER CCD camera. Images were processed with software Aquacosmos (Hamamatsu Photonics). Classification of "strong" or "weak" positive cells was determined by pixel intensities of the BrdUrd-incorporating cells in the digital fluorescent images.

RESULTS
Identification of Human ASKL1-We previously reported the presence of three short stretches of amino acids conserved in Dbf4-related molecules and named them Dbf4 motif-N, -M, and -C (27). Further characterization of these motifs revealed that the motif-M and -C polypeptides are capable of binding to the catalytic subunit and that the bipartite binding of these two motifs is essential and sufficient for activation of Cdc7 kinase (28,29).
Data base search revealed the presence of a genomic DNA segment derived from the human chromosome 17, which has a potential of encoding an amino acid segment sharing a homology with Dbf4 motif-C. We then further searched the genomic region for the presence of sequences resembling Dbf4 motif-M and -N. We found genomic segments potentially encoding motif-N and motif-M, enabling us to amplify the cDNA, which we named ASKL1. We also found the potential C-terminal segment of ASKL1 on the genomic data base. Consequently, we were able to obtain a putative full-length ASKL1 cDNA encoding a 615 amino acid coding region, which contained the three conserved Dbf4 motifs and shared 25% overall identity with human ASK. ASKL1 turned out to be identical to Drf1 as reported by Montagnoli et al. (26). Drf1/ASKL1 is composed of a N-terminal segment (35% identity with huASK) carrying conserved Dbf4 motif-N, -M, and -C and a long C-terminal tail that is much less conserved (Supplemental Fig. S1).
Complex Formation with huCdc7 and Kinase Activation-As reported previously, Drf1/ASKL1 binds to huCdc7 and forms a complex. We have confirmed this by expressing them both in insect cells (Supplemental Fig. S2A) and mammalian cells.
We have identified four distinct transcripts for Drf1/ASKL1 (ASKL1a, ASKL1b, ASKL1c, and ASKL1d, see Supplemental Fig. S2, B and C). We then expressed these four forms of the FLAG-tagged ASKL1 derivatives in combination with GSTtagged huCdc7 in insect cells and attempted to purify the huCdc7-ASKL1 complex through glutathione-Sepharose beads (Fig. S2D). Affinity-purified huCdc7-ASKL1a or huCdc7-ASKL1c complex can phosphorylate MCM2 in the MCM2⅐ MCM4⅐MCM6⅐MCM7 complex causing the downward mobility shift on SDS-PAGE (Supplemental Fig. S2E, lanes 3 and 1), as observed with the huCdc7-ASK complex. However, ASKL1b or ASKL1d lacking motif-M or both motif-M and motif-C, respectively, did not generate a complex with Cdc7 (Supplemental Fig. S2D, lanes 5-8). We have also coexpressed GST-fused ASKL1 polypeptide (residues 211-355; GST-ASKL1[minimum]) and hemagglutinin-tagged huCdc7 in bacterial cells and purified the complex. The wild-type complex was active in phosphorylating MCM2 in the MCM2⅐MCM4⅐MCM6⅐MCM7 (data not shown). These results indicate that Drf1/ASKL1 can form active kinase complexes with huCdc7 and that the amino acid segment containing motif-M and motif-C is required and sufficient for the generation of an active kinase.
We also examined whether endogenous huCdc7 associates with Drf1/ASKL1. We have established a stable HeLa cell line expressing GST-ASKL1 fusion protein containing the "minimum" ASKL1 region (Dbf4 motif-M and -C) under a constitutively active promoter. In control, a stable cell line expressing GST alone was also established. The GST-ASKL1[minimum] stable cell line was synchronized by hydroxyurea arrest, and GST-ASKL1 was pulled down from the Triton-soluble extracts at 0 h (early S) and 9 h (G 2 /M) after release. Western blotting confirmed the presence of GST-ASKL1 protein. More GST-ASKL1 was pulled down from the G 2 /M extracts (Fig. 1A, lanes 3 and 4). In the same fraction, huCdc7 protein was detected and its amount was roughly proportional to that of GST-ASKL1. In control, the pull down from HeLa cells expressing GST alone indicated the presence of an equal amount of GST protein in early S and G 2 /M extracts but no huCdc7 was detected (Fig. 1A,  lanes 1 and 2). Endogenous ASKL1 protein was immunoprecipitated next with affinity-purified anti-ASKL1 antibody from Triton-soluble extracts of asynchronous culture (Fig. 1B, left panel, lane 1), and Cdc7 was detected in the immunoprecipitates (Fig. 1B, right panel, lane 4). The interaction was detected at any stages of cell cycles, but ASKL1-bound fractions of huCdc7 appeared to increase during S phase (Fig. 1C, lanes 2  and 3). These results indicate that endogenous huCdc7 forms a complex with ASKL1 throughout the cell cycle.
Expression of Drf1/ASKL1 during Cell Cycle-Transcription of human ASK is known to fluctuate during the cell cycle, increasing at late G 1 and staying at a high level during the S phase (16). HeLa cells were synchronized, and the expression of Drf1/ASKL1 was examined by Northern analyses during cell cycle progression. After release from hydroxyurea arrest, the Drf1/ASKL1 transcript increased at 8 -10 h after release when the cells were in late S to G 2 and started to increase again at the next late S phase (24 h after release) (Supplemental Fig.  S3A). In the release from arrest at mitosis by TN-16 treatment, transcription of Drf1/ASKL1 was high at the time of release and decreased as the cell cycle progressed into G 1 /S. It increased again in the next late S phase (Supplemental Fig. S3B). This result indicates that transcription of Drf1/ASKL1 is regulated during cell cycle and increases at the late stage of DNA synthesis through G 2 and mitosis. Transcription of Drf1/ ASKL1 was detected in various other cell lines including human kidney-derived 293 cells and human amnion-derived FL cells, and its level in these cell lines also increased after arrest with TN-16 (data not shown).
Affinity-purified anti-ASKL1 antibody was used to identify the endogenous Drf1/ASKL1 protein. Triton-soluble and -insoluble fractions were prepared from asynchronously growing HeLa cells and were examined by Western blotting. The bands migrating slightly faster than FLAG-tagged ASKL1 expressed in COS7 cells were identified in the Triton-soluble fraction. To confirm that this band indeed represents the endogenous Drf1/ ASKL1 protein, we conducted siRNA experiments. We have designed four different siRNA derived from different portions of Drf1/ASKL1 as well as control siRNA derived from the luciferase gene. The intensities of the putative Drf1/ASKL1 protein bands decreased with ASKL1-1 and ASKL1-2 siRNA but not with other siRNAs (Fig. 2A). This result strongly indicates that they represent the endogenous ASKL1 protein. We were not able to determine the cause for the presence of two bands. Drf1/ASKL1 protein level was examined in cells in which cell cycle was synchronized by release from double thymidine block-induced G 1 /S arrest. The Drf1/ASKL1 protein was detected almost exclusively in the nuclear-soluble fraction (Fig.  2B). The protein level increased as the cells entered late S through G 2 /M (9 -12 h), slightly decreased during G 1 -S (18 h), and increased again at next late S to G 2 (24 h) (Supplemental Fig. S3C). Drf1/ASKL1 protein was detected in the cytoplasmic fraction at a low level but was not detected in the Triton-soluble chromatin-enriched fractions throughout the cell cycle (data not shown). These results indicate that the Drf1/ASKL1 protein level oscillates during cell cycle and increases at late S to G 2 /M and that the protein is almost exclusively present in the nuclear-soluble fraction.
To examine whether the protein level oscillates in a manner independent from transcriptional regulation, stably expressed GST-ASKL1[minimum] protein under the control of a constitutive promoter was examined. The cells were released from hydroxyurea arrest and were fractionated into cytoplasm and nuclear-soluble and nuclear pellet fractions. The levels of the stably expressed proteins were detected by anti-GST antibody (Supplemental Fig. S3D). The majority of GST protein was detected in cytoplasmic fractions, and a small amount was detected in nuclear-soluble fractions at a constant level throughout the cell cycle. In contrast, GST-ASKL1 protein was detected in both cytoplasmic and nuclear-soluble fractions but not in nuclear pellet fractions. The total amount of GST-ASKL1 protein peaked at late S to G 2 /M phase. The protein level in nuclear-soluble fractions reached maximum at 6 -8 h after release (late S to G 2 ), decreased at 10 -12 h (M to G 1 ), and increased again at 24 h (late S), whereas the protein level in cytoplasm reached maximum at 10 h after release. In controls, RPA p34 subunits were detected in nuclear pellet (enriched in chromatin-bound proteins) only during S phase and I␤ protein was recovered only in cytoplasm fractions, verifying the procedure of cell fractionation. Because the transcription of the exogenous GST-ASKL1 is under a constitutive promoter (30), the result indicates that the level of Drf1/ASKL1 protein is regulated at post-transcriptional levels as well.
Potential Roles of Drf1/ASKL1 during Cell Cycle Progression-To obtain insight into the roles of Drf1/ASKL1 in human cell cycle progression, we have examined the effect of reducing the level of Drf1/ASKL1, ASK, and Cdc7 proteins in HeLa cells. The knockdown of target proteins or transcript was confirmed by immunoblotting or RT-PCR in HeLa cells treated with siRNA specific to ASKL1, ASK, or Cdc7. Drf1/ASKL1 and Cdc7 protein levels were reduced in Drf1/ASKL1 and Cdc7 siRNAtreated cells, respectively (Fig. 2C, lanes 3 and 4 and 7 and 8). Drf1/ASKL1 protein level was reduced also in Cdc7 siRNAtreated cells (Fig. 2C, lane 7). Reduction of ASK expression was confirmed by RT-PCR with ASK siRNA-treated cells (Fig. 2D,  3 and 4), ASK (lanes 5 and 6), or Cdc7 (lanes 7 and 8) siRNA were collected at 48 h after transfection and were fractionated into Triton-soluble (S) and -insoluble (P) fractions. Proteins were detected by immunoblotting with the antibodies indicated. D, RT-PCR was conducted with total RNA prepared from the HeLa cells transfected with control siRNA (lanes 1, 3, and 5) or ASK siRNA (lanes 2, 4, and 6). The transcription of ASK (upper panel) or actin (lower panel) was analyzed by specific primers. E, the numbers of viable HeLa cells in 60-mm culture dishes (upper panel) or those of viable U2OS cells in a 24-well plate (lower panel) were counted at the times indicated after transfection. As controls, sense-strand RNA for ASKL1 (HeLa) or for luciferase siRNA (U2OS) were used. This experiment was conducted at least three times with similar results, and representative data are shown. Open column, 2 days after transfection; filled column, 3 days after transfection. F, FACS analyses of DNA content of control or ASKL1 siRNA-transfected HeLa cells at 48 h after transfection. G, phase-contrast images of ASKL1, ASK, or Cdc7-depleted cells at 48 h after transfection. Magnification: ϫ10 (upper) or ϫ20 (lower). Images were observed by IX50 (Olympus) and recorded by digital camera C-4100 (Olympus).  4, and 6). Interestingly, in ASK-depleted cells, the Cdc7 protein level also significantly decreased (Fig. 2C, lanes 5 and  6) but not in ASKL1-depleted cells.
Drf1/ASKL1 siRNA slowed down the rate of cell growth. In HeLa cells, the knockdown of ASKL1 resulted in the reduction of cell counts by 36% compared with control siRNA at 3 days after treatment (Fig. 2E, upper panel). Significant growth retardation was observed also in U2OS cells treated with ASKL1 siRNA (Fig. 2E, lower panel). The majority of ASK-depleted cells appeared to stop growth and undergo cell death at as early as 24 h after siRNA treatment. The phenotypes of Cdc7-depleted cells were similar to those of ASKdepleted cells, although cell death was not triggered until later time point (data not shown). In contrast to ASK and Cdc7, ASKL1 siRNA did not affect cell viability but the rate of cell proliferation was reduced.
FACS analyses of DNA content of asynchronously growing HeLa cells, transfected with ASKL1 siRNA for 48 h, revealed a decrease of G 1 cells and increase of late S to G 2 /M phase cells (Fig. 2F). Under the microscope, the population of mitotic cells (in a globular shape with condensed chromatin) apparently increased in both Cdc7 and ASKL1 siRNA-treated cells ( Fig.  2G; see below).
Depletion of ASKL1 Slows S Phase Progression-To evaluate whether the slow cell growth caused by siRNA was due to blockage at a specific stage of cell cycle or not, we first examined the effects of knockdown on S phase progression and DNA synthesis. We conducted double thymidine block to synchronize HeLa cells at the G 1 /S boundary after transfection of siRNA and then released the cells into cell cycle (Fig. 3A). In both untreated or control (sense primer for ASKL1) cells, DNA contents changed from 2N (time 0) to 4N (time 9 h) within 9 h after thymidine release. However, this transition occurred more slowly in ASKL1-depleted cells (Fig. 3B).
With the same set of samples, BrdUrd was incorporated for 30 min before cell harvest and BrdUrd incorporation was analyzed by FACS. Dot plots of control or ASKL1 siRNA-treated cells showed that the significant level of BrdUrd incorporation was observed in cells at 3-9 h after release (Fig. 3C). At 3 h after release, over 60% cells incorporated BrdUrd in control cells, whereas only 35% incorporated BrdUrd in ASKL1 siRNAtreated cells. In contrast, at 9 h after release, 61% ASKL1 siRNA cells were still in S phase, whereas only 31% were in S phase in control cells (Fig. 3D). Thus, the onset of BrdUrd uptake is delayed and S phase continued longer in ASKL1 siRNA-treated cells.
To determine the precise timing of initiation and elongation of S phase in ASKL1-depleted cells, DNA synthesis was analyzed by immunostaining in the cells progressing from mitosis (see Fig. 3E for time course). The HeLa cells transfected with siRNA for luciferase, Cdc7, or ASKL1 were released from nocodazole block and collected for FACS analyses after BrdUrd incorporation for 30 min. In this experiment, depletion of Drf1/ ASKL1 was rather mild and delay in S phase progression was not as striking as the one shown in Fig. 3B (Fig. 3F). The timing for the start of BrdUrd incorporation was around 6 h after release for all of the cells (Fig. 3G). Slight delay in initiation was observed for Cdc7 or ASKL1 siRNA-treated cells (Fig. 3H, see 6 h after release). By 15 h after the release, the level of DNA synthesis substantially decreased in control cells and Ͻ40% cells were labeled only at a low level, whereas over 70% Cdc7depleted cells were still labeled with BrdUrd, and nearly 40% was labeled at a high intensity. In ASKL1 siRNA-treated cells, nearly 70% cells incorporated BrdUrd at 15 h after release, albeit at a low level. These data indicate that decreased expression of ASKL1 in HeLa cells causes delay in progression of DNA synthesis at later stages of S phase in addition to slight delay in the initiation of S phase.
Depletion of ASKL1 Slows Down Mitotic Progression and Inhibits Release from Mitosis-We previously noted that mitotic cells seemed to be increased when siRNA for ASKL1 was transfected into HeLa cells. Quantification of each cell cycle phase of the FACS data revealed that G 2 /M phase cells in ASKL1 siRNA-and control siRNA-treated cells were 33.1 and 11.6%, whereas G 1 cell populations were 29.0 and 41.7%, respectively (Fig. 2, F and G), suggesting the increase of mitotic cells after ASKL1 siRNA treatment in HeLa cells. To strictly distinguish the cells at interphase including G 2 and those at mitosis, siRNA-treated cells were subjected to detailed microscopic observation. The precise stage of mitosis at which the mitotic cells were arrested was determined by visualizing chromatin structures with Hoechst 33342 staining. The results indicated that the increase in mitotic cells in ASKL1 or Cdc7 siRNA-treated cells was statistically valid (Fig. 4A).
We next examined the mitotic progression in synchronized cells. Even under the mild depletion of ASKL1 where the delay of S phase progression was not obvious, the population of M phase cells was higher in ASKL1-depleted cells than that in luciferase siRNA-treated cells at 12 and 15 h after release from double-thymidine block (Fig. 4B). The delay of M phase was observed also with Cdc7 depletion under the similar condition (data not shown).
In the next set of experiments, siRNA treatment was combined with mitotic cell arrest. HeLa cells transfected with siRNA were blocked by nocodazole and then released into cell cycle for 4 h. The cells were observed by phase-contrast microscopy (Fig. 4C). In comparison with the control untransfected cells in which the majority of the population exited the M phase and entered G 1 phase, ASKL1-and Cdc7-depleted cells at 4 h after nocodazole release displayed increased fractions of mitotic population: 89.5, 26.2, and 13.8% in Cdc7 siRNA-treated, ASKL1 siRNA-treated, or untreated cells, respectively (Fig.  4D). These results indicate that mitotic progression is delayed or inhibited in Cdc7 or ASKL1 siRNA-treated HeLa cells. The magnified images of aberrant mitotic cells at 4 h after nocodazole release included those arrested at various stages of mitosis (data not shown), indicating that the arrest was not at a specific stage of mitosis. The population analyses of mitotic stages in ASKL1 or Cdc7 siRNA-treated asynchronous cells also supported the conclusion that mitotic cells at a particular stage were not accumulated (Supplemental Fig. S4A). In addition, the multinucleated population also increased in ASKL1 and Cdc7 siRNA-treated cells (Supplemental Fig. S4B), suggesting a defect in nuclear division or cytokinesis. Double-stranded DNA Breaks Are Generated in ASKL1 siRNA-treated Cells-The above results indicate that S phase progression is retarded by depletion of ASKL1 protein in HeLa cells, suggesting a possibility that replication forks are stalled. Therefore, we have examined the induction of DNA damages by using ␥H2AX antibody, which detects double-stranded DNA breaks. ␥H2AX signal was induced by hydroxyurea or UV treatment of control cells (Fig. 5, lanes 4 and 6). In contrast, in ASKL1 siRNA-treated cells, ␥H2AX signals were detected without the damaging agents (Fig. 5, lane 8). Similarly, damage-independent ␥H2AX signal was detected in ASK siRNAtreated cells (Fig. 5, lane 14). We next examined whether Chk2 is activated in these cells. The antibody that recognizes the phosphorylated Thr 68 residues detected Chk2 only after hydroxyurea or UV treatment in control cells (Fig. 5, lanes 3 and  5), whereas it reacted even without DNA damages in ASKL1or ASK-treated cells (Fig. 5, lanes 7 and 13). On the other hand, Chk1 was not activated by ASKL1 siRNA treatment without hydroxyurea or UV treatment (Fig. 5, lane 7). Cdc2 tyrosine 15 phosphorylation did not increase nor was the CyclinB-Cdc2 kinase activity inhibited in response to ASKL1 siRNA (data not shown), suggesting that G 2 /M checkpoint signal is not generated. These results indicate that down-regulation of ASKL1 in HeLa cells indeed arrests the replication fork progression, which leads to the generation of double-stranded DNA breaks or unusual DNA structures and activation of Chk2 kinase, although mitotic inhibition signals are not induced.

DISCUSSION
Cdc7-Dbf4 kinase complex plays crucial roles in regulating initiation and progression of DNA replication, and their functions appear to be conserved throughout evolution (15,19). Dbf4 resembles cyclin molecules in that it is an essential regulatory subunit for the Cdc7 catalytic subunit (11) and that its expression is cell cycle-regulated, although there is no homology on the level of their primary structures. It has not been known whether Dbf4 also forms a protein family, each member of which plays distinct functions. Recently, a second set of the Cdc7-Dbf4-related kinase complex, Spo4-Spo6, was identified in fission yeast (24,25). Hsk1-Dfp1/Him1 kinase (fission yeast homologue of Cdc7-Dbf4 complex) plays essential mitotic functions in fission yeast by phosphorylating Cdc19 (MCM2) protein (13,14,25), whereas Spo4-Spo6, dispensable for viability, plays essential roles in the sporulation stage during meiosis (24,25). This finding has suggested a possibility that Cdc7-Dbf4 is also a part of a novel kinase-regulator family, although the presence of additional Cdc7-Dbf4-related molecules in higher eukaryotes has not been reported. Recently, Drf1, a second Dbf4/ASK-related molecule in human and its possible Xenopus homologue were reported (26,31).
We independently isolated ASKL1 and its isoforms present on the human chromosome 17. One isoform of ASKL1 is identical to Drf1. Drf1/ASKL1 contains the three conserved Dbf4 motifs and shares 35% identity with human ASK in the Nterminal region. The C-terminal amino acid sequences are diverged as is found between human and mouse ASK (32). Transcription of Drf1/ASKL1 oscillates during cell cycle, increasing at late S phase and peaks at G 2 /M phase in mammalian cells (Supplemental Fig. S3, A and B), consistent with a previous report (26). The level of the endogenous Drf1/ASKL1 protein also increases at late S to G 2 /M phase (Supplemental Fig. S3C). The level of GST-ASKL1[minimum] containing only motif-M and -C of Drf1/ASKL1, stably expressed in HeLa cells under a constitutively active promoter (SR␣), also increased from late S to G 2 /M phase, indicating the presence of regulation at a posttranscriptional level (Supplemental Fig. S3D).
We found that Drf1/ASKL1 protein is present almost exclusively in nuclear-soluble fractions all through the cell cycle. Although a putative Xenopus homologue of huDrf1/ASKL1 was reported to be loaded onto chromatin during DNA replication (31), human Drf1/ASKL1 was not detected in chromatin-enriched Triton-insoluble fractions, even after DNA replication block or DNA damages (data not shown).
Drf1/ASKL1 expressed in insect cells bound to huCdc7 (Supplemental Fig. S2A), and motif-M is essential for its binding (Supplemental Fig. S2D). The purified Cdc7-ASKL1 complex could phosphorylate MCM2 in the MCM2⅐MCM4⅐ MCM6⅐MCM7 complex in vitro (Supplemental Fig. S2E). We have shown that Cdc7 indeed forms a complex with Drf1/ ASKL1 in vivo throughout the cell cycle (Fig. 1C). Therefore, it can be regarded as a second activation subunit for huCdc7 kinase.
To more precisely understand the functions of Drf1/ASKL1, we have conducted siRNA experiments. Decreased expression of Drf1/ASKL1 by siRNA in HeLa cells lead to retarded growth. By the combination of siRNA treatment with cell cycle synchronization, this retarded proliferation was shown to be caused by slowed progression of both S phase and mitosis. The cell viability was not noticeably affected in ASKL1-depleted HeLa cells, in contrast to Cdc7 and ASK siRNA-treated cells, which undergo apoptosis.
Although the precise functions of Drf1/ASKL1 during cell cycle are not clear at this moment, we could make some speculation on the basis of the phenotypes of siRNA-treated cells. ASKL1 siRNA slows down the cell cycle (Figs. 3B and 4B) but does not kill the cell unlike ASK siRNA (Fig. 2G). Drf1/ASKL1 protein is not on the chromatin (Fig. 2B) and therefore is not expected to constitute a part of the replication complex essential for initiation. This would be conducted by ASK, which binds to chromatin during G 1 and whose abundance on the chromatin increases during the S phase (33). Cdc7 kinase is a known target of replication checkpoint in Xenopus egg extracts. Dfp1/Him1, a fission yeast Dbf4 homologue, is hyperphosphorylated in response to replication fork arrest signals (34). This hyperphosphorylation may be associated with temporal downregulation of Cdc7-ASK kinase activity or its dissociation from chromatin (35,36). To restart replication forks, Cdc7 kinase may need to be reactivated to phosphorylate components of replication machinery. It is possible that ASKL1, which is normally present in nuclei, associates with Cdc7 and facilitates the reactivation of Cdc7 kinase after replication fork arrest. It has become realized that replication forks are blocked by various causes, even during the normal course of DNA replication (37). Therefore, the role of Drf1/ASKL1 may be to facilitate the fork progression by assisting the replication fork restart during the unexpected fork stalling that may occur during the S phase. In the absence of Drf1/ASKL1, the fork restart process would be less efficient and S phase may take more time to complete. Induction of phosphorylation of histone H2AX and Chk2 kinase in ASKL1 siRNA-treated cells (Fig. 5) may be related to the impaired fork reconstruction.
Down-regulation of Drf1/ASKL1 also affects the G 2 /M phase progression. This delay of G 2 /M phase progression could be the results of checkpoint-mediated G 2 /M arrest induced by replica-  7-12), or ASK siRNA (lanes [13][14][15][16][17][18]. Lanes 1,2,7,8,13, and 14 (NT), untreated cells. Lanes 3,4,9,10,15, and 16 (hydroxyurea HU), cells incubated with hydroxyurea (5 mM) for last 6 h before harvest. Lanes 5,6,11,12,17, and 18 (UV), cells irradiated with UV (50 J/m 2 ) at 2 h before harvest. Untreated or ASKL1 siRNA-treated cells were recovered at 48 h after transfection, whereas with ASK siRNA transfection (lanes 13-18), cells were harvested at 24 h after transfection to avoid cell death caused by this siRNA. Triton-soluble (S) and -insoluble (P) extracts were prepared from each sample, and proteins were detected by immunoblotting with antibodies indicated. NT, transfection without siRNA. tion fork block. However, this is not likely because we did not detect an increase of Cdc2 tyrosine 15 phosphorylation or Cdc2 kinase inhibition by ASKL1 siRNA (data not shown). Morphological analyses of mitotic cells in ASKL1 siRNA-treated population indicated they are not arrested at a particular stage of mitosis, although cells at prometaphase are slightly more abundant (Supplemental Fig. S4A). We also detected an increase of multinucleated cells (Supplemental Fig. S4B), which may result from aberrant nuclear division or the failure of cytokinesis. Similar mitotic delay and increase of multinucleated cells were observed with Cdc7 siRNA.
ASKL1 forms a complex with Cdc7 in nuclei, and the amounts of this complex increase at late S through G 2 /M. Cdc7-ASKL1 may phosphorylate the mitotic apparatus to facilitate the progression of M phase. Alternatively, the loss of ASKL1 may affect the proper processing of stalled replication forks, which may lead to altered chromatin structures of replicated molecules, including impairment of sister chromatid cohesion or of chromatin condensation. These defects in chromatin structures may directly or indirectly affect the processes of mitosis. In fact, we have shown previously (21) that sister chromatid cohesion is partially impaired in a fission yeast hsk1ts mutant. Furthermore, we recently obtained evidence showing that Hsk1 is required also for mitosis of fission yeast. 2 These results indicate that the roles of Cdc7 during M phase may be conserved.
In summary, our results show that huCdc7 is regulated by two distinct activation subunits, ASK and ASKL1. ASKL1 may not be essential for DNA replication but may play supplemental roles in S phase progression and/or emergency roles in restoration of stalled replication forks as well as roles in M phase progression in conjunction with Cdc7 kinase.