Proliferating Human Cells Hypomorphic for Origin Recognition Complex 2 and Pre-replicative Complex Formation Have a Defect in p53 Activation and Cdk2 Kinase Activation*

The Origin Recognition Complex (ORC) is a critical component of replication initiation. We have previously reported generation of an Orc2 hypomorph cell line (Δ/–) that expresses very low levels of Orc2 but is viable. We have shown here that Chk2 is phosphorylated, suggesting that DNA damage checkpoint pathways are activated. p53 was inactivated during the derivation of the Orc2 hypomorphic cell lines, accounting for their survival despite active Chk2. These cells also show a defect in the G1 to S-phase transition. Cdk2 kinase activation in G1 is decreased due to decreased Cyclin E levels, preventing progression into S-phase. Molecular combing of bromodeoxyuridine-labeled DNA revealed that once the Orc2 hypomorphic cells enter S-phase, fork density and fork progression are approximately comparable with wild type cells. Therefore, the low level of Orc2 hinders normal cell cycle progression by delaying the activation of G1 cyclin-dependent kinases. The results suggest that hypomorphic mutations in initiation factor genes may be particularly deleterious in cancers with mutant p53 or increased activity of Cyclin E/Cdk2.

DNA replication requires the action of numerous protein complexes to accurately and completely duplicate the genome in a timely fashion. In most organisms, a series of steps must occur to load the replication machinery onto the chromatin. This sequence of events has been examined in several model systems and is fairly well conserved (for review, see Ref. 1). Briefly, the Origin Recognition Complex (ORC), 2 a complex of six subunits, is thought to bind DNA, thus marking a replication origin (2). ORC recruits Cdc6 and Cdt1 to the chromatin, both of which are required for loading the Mcm2-7 complex (3)(4)(5)(6)(7). The Mcm2-7 complex is thought to be the helicase responsible for unwinding the doublestranded DNA (8 -12), allowing polymerases access to the DNA. The loading of Mcm2-7 completes formation of the pre-Replicative Complex (pre-RC), which must occur before the end of G 1 . At this stage, origins are "primed" and are awaiting the activity of cyclin-dependent kinases and Dbf4/Cdc7 kinase. These kinase activities are required to load downstream factors that make up the Pre-initiation Complex, Cdc45/Sld3, Mcm10, Dpb11/Sld2, and GINS. These proteins are thought to be important for polymerase loading by mechanisms that are not fully understood (reviewed in Refs. 1 and 13).
The ORC itself is of great interest, as it defines the region in the genome where replication will begin. This function is vital, as evenly spaced initiations are important to complete replication of the entire genome as quickly as possible. In fact, ORC was initially identified in budding yeast based on its ability to interact with the yeast origin sequence (ARS) (2). ORC is formed of six subunits, Orc1-6, that range in size from 97 to 28 kDa. It has been shown that the ORC subunits interact with each other, albeit with different affinities (14,15). These studies show that Orc2 and Orc3 seem to form a core subcomplex with which other ORC members interact. Early studies on the ORC proteins showed that Orc1 (16), Orc2 (17,18), Orc3 (19), Orc4 (19), Orc5 (19), and Orc6 (16) are all essential in yeasts. Although ORC is expected to be essential for human chromosomal replication, a detailed study of chromosomal replication with low levels of human ORC has not yet been published. Recent reports using small interfering RNA indicate that acute depletion of an ORC subunit leads to cell cycle arrest, making it impossible to study chromosomal replication under such conditions (20,21).
To study ORC in human cells, we have previously generated an Orc2 hypomorph cell line, in which Orc2 is only being expressed from one allele at 10% of wild type levels (22). This Orc2 ⌬/Ϫ cell line (containing one hypomorph Orc2 allele and one null Orc2 allele) was found to be viable, presumably because of compensatory changes in other genes, and was able to proliferate for many generations. The low levels of Orc2 did, however, prevent replication of an exogenous plasmid containing a single Epstein-Barr virus origin. We believe that this difference in replication ability stems from a threshold effect; the levels of Orc2 were high enough to support chromosome replication, but not episome replication. In this study, we have examined the effects of low Orc2 on the replication of the cellular chromosomes, as well as the cell as a whole.
Pulse-Chase-Logarithmic cells were starved of methionine for 1 h in Dulbecco's modified Eagle's medium, supplemented with dialyzed fetal bovine serum. Cells were then labeled with 300 Ci of [ 35 S]methionine for 4 h. Cells were washed to remove unincorporated methionine and were chased with McCoy's 5A medium supplemented with fetal bovine serum for the indicated lengths of time. Cells were lysed in the same buffer as described above, but with 300 mM NaCl. Orc3 was immunoprecipitated with the above antibody and separated by SDS-PAGE. The gel was then dried and exposed to a phosphorimaging plate for visualization and analysis.
FACS Analysis-Cells were prepared as previously described (27). The analysis was carried out on a BD Biosciences FACS Calibur using Cellquest and FloJo software.
Chromatin Immunoprecipitation-In vivo cross-linking was performed as described in Ref. 28 with some modifications. In brief, 80% confluent HeLa, HCT116, or HCT116 ⌬/Ϫ cells were grown as described above and then treated with formaldehyde (1%). Cross-linked cell nuclei were sonicated 10 times for 30 s each time, and the chromatin size was monitored by electrophoresis (29). This treatment generated fragments of ϳ20 kb. To further reduce the chromatin size to smaller fragments of 1.5-3.5 kb, DNA was digested with SphI, HindIII, PstI, and EcoRI restriction endonucleases in NEB2 buffer (100 units of each; New England Biolabs, Beverly, MA) at 37°C for 6 h. Sheared chromatin-lysed extracts were incubated with 50 l of protein G-agarose (Roche Applied Science) to reduce background caused by nonspecific adsorption of irrelevant cellular proteins/DNA to proteins. These cleared chromatin lysates were incubated at 4°C for 6 h on a rocker platform with either 50 l of preimmune rabbit serum (Santa Cruz Biotechnology) or 5 g of anti-Orc2, anti-Orc3, anti-Orc4, or anti-Orc6 antibodies. Protein G-agarose (50 l) was added, and the incubation was continued for 12 h. The precipitates were successively washed two times for 5 min with 1 ml of each buffer, lysis buffer, WB1 (50mM Tris-HCl, pH 7.5, 500 mM NaCl, 0.1% Nonidet P-40, 0.05% sodium deoxycholate), WB2 (as WB1 with no NaCl), and 1 ml of TE (20 mM Tris-HCl, pH 8.0, 1 mM EDTA). The precipitates were finally resuspended in 200 l of extraction buffer (1% SDS/TE). The samples were incubated at 65°C overnight to reverse the protein/DNA cross-links, followed by a 2-h incubation at 37°C with 100 g of proteinase K (Roche Applied Science). Finally, the samples were processed for DNA purification by passing them through QIAquick PCR purification columns (Qiagen, Valencia, CA). PCR reactions were carried out in 20 l with 1/200th of the immunoprecipitated material with the use of LightCycler capillaries (Roche Applied Science) and the LightCycler-FastStart DNA Master SYBR Green I (Roche Applied Science). The real-time PCR quantification was performed as described in Ref. 30, using primer sets LB2 and LB2 C1.
Cell Cycle Analysis-Cells were arrested in 40 ng/ml of nocodazole for 16 h. To release the cells from mitosis, they were washed three times in sterile phosphate-buffered saline and replated in fresh warmed medium. After the desired incubation time, cells were labeled with 10 M BrdUrd for 1 h and then prepared for FACS or Western blot analysis as above. Alternately, after the desired incubation time, cells were treated with 1.25 Ci of [ 3 H]thymidine for 1 h. Cells were then washed and incubated with cold stop solution (10% trichloroacetic acid, 200 mM sodium pyrophosphate). Cells were washed with 95% ethanol and solubilized (1% SDS, 10 mM NaOH.) The resulting solution was spotted on Whatman paper, dried, and counted on a Beckman LS 6000 scintillation counter. To examine cells in S-phase, cells were treated with 2 mM thymidine for 12 h, released into untreated medium for 12 h, and further treated with 1 g/ml of aphidicolin for 12 h.
Chromatin Fractionation-Chromatin fractions were isolated as described previously (31). Briefly, cells (2 ϫ 10 6 ) were lysed in 100 l of CSK buffer (10 mM Pipes, pH 7.0, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 ) containing 0.5% Triton X-100, 1 mM ATP, 1 mM Na 3 VO 4 . Lysates were incubated on ice for 20 min and then centrifuged at 1500 rpm for 5 min at 4°C. Supernatant (S1) was removed, and pellets were washed with 1 ml of lysis buffer and centrifuged again. Pellets were incubated in 100 l of lysis buffer containing 1 mM CaCl 2 and 120 units of micrococcus nuclease (Worthington) for 10 min at 37°C and centrifuged. Supernatant (S2, chromatin-bound fraction) was removed, and pellets were washed with 1 ml of lysis buffer and centrifuged again. Pellets were boiled in 100 l of 1ϫ sample buffer (P2).
DNA Combing and Detection by Fluorescent Antibodies-Cells were synchronized at very early S-phase by treatment with 2 mM thymidine for 12 h, release into untreated medium for 12 h, and further treatment with 1 g/ml of aphidicolin for 12 h. Cells were released into medium containing 50 M BrdUrd for the indicated time points. Cells were then washed in medium containing 100 M thymidine and incubated in 10 M thymidine until 8 h post release. Cells were then harvested with trypsin/EDTA and suspended in 1% agarose. Agarose blocks were incubated with ESPK (0.5 M EDTA, 1% sarkosyl, 2 mg/ml proteinase K (Fisher)) for 24 h twice. Blocks were then washed with TE/phenylmethylsulfonyl fluoride and stored in 0.5 M EDTA.
DNA from whole cells was extracted and combed on silanised coverslips as described (32). Combed DNA was dehydrated in a series of ethanol (70, 90, and 100%), denatured with 1N NaOH for 30 min, again dehydrated, and blocked in a blocking solution (1ϫ phosphate-buffered saline, 0.1% Tween, 1% bovine serum albumin) for 1 h. BrdUrd was detected with an anti-BrdUrd antibody (Abcys), followed by an anti-rat antibody conjugated with AlexaFluor 488 and an anti-goat AlexaFluor 488 antibody (33). Total DNA was visualized afterward by an antiguanosine antibody (Argene), followed by an anti-mouse AlexaFluor 594 (Molecular Probes). Antibody incubations were in general for 30 min and were separated by three-four washes with 1ϫ phosphate-buffered saline, 0.1% Tween. Coverslips were mounted in Vectashield solution. To analyze data, images of the combed DNA molecules were acquired by a Leitz DC300F camera associated with the LeicaFW4000 software and measured by ImageGauge 4.2 software. Fields of view were chosen at random in the AlexaFluor 594 channel and then photographed under the AlexaFluor 594 and AlexaFluor 488 filters. The replication extent of each sample was defined as the sum of all eye lengths divided by the total length of the molecules. Fork density is the total number of forks divided by total DNA length (kb) in each sample, as detailed in Ref. 33. Total DNA length in each case was normalized using the number of cells entering S-phase as shown in Fig. 5D).

RESULTS
Absence of Orc2 Affects Core ORC Complex Stability-A hypomorph mutation of Orc2 in HCT116 cells (which we designate the ⌬/Ϫ cell line) decreases Orc2 protein level by 90% (22). We also observed decreased levels of Orc3 and Orc5 protein even though their genomic loci remained unaltered (Ref. 22 and data not shown.) Orc1, Orc4, Orc6, and Cdt1 protein levels are unaffected in ⌬/Ϫ, indicating that the Orc2,3,5 subcomplex alone is affected by the lack of Orc2. mRNA levels of Orc3 and Orc5 were unchanged in the ⌬/Ϫ cells (supplemental Fig.  S1), suggesting that the protein decrease is due to another mechanism after transcription.
Orc3 and Orc5 proteins might be destabilized due to low levels of the Orc3 binding partner. To examine this, Orc3 stability was determined by a [ 35 S]methionine-labeling pulse-chase experiment. Orc3 was less stable in the ⌬/Ϫ cells (Fig. 1A), and quantification of the fluorogram revealed a marked decrease in half-life: 11 h in HCT116 cells, compared with 6 h in the ⌬/Ϫ cells (Fig. 1B). It is possible that the destabilization results from proteasomal degradation. However, Orc3 was not stabilized in either cell line after MG132 treatment, in contrast to p27, a known target of the proteasome (34) (Fig. 1C). Therefore, Orc3 is not targeted for proteasomal degradation. Similar results were observed for Orc5 (data not shown). These results suggest that the decrease in Orc3 protein level is due to destabilization in the absence of the partner protein Orc2 by mechanisms that do not depend on the proteasome. This destabilization of the Orc2,3,5 subcomplex may adversely affect pre-RC loading in the ⌬/Ϫ cells.
ORC and Pre-RC Loading Is Decreased in Orc2 Hypomorph Cells-In yeast, Orc2 is required for DNA replication initiation (35). To investigate ORC loading at a specific origin, chromatin immunoprecipitation was performed using several ORC subunits. The associated chromatin was assayed by quantitative PCR using primers in the Lamin B2 region, a previously reported human origin (36). Orc2, 3, 4, and 6 associated with the Lamin B2 origin in HeLa cells, but not a nearby control region, whereas normal rabbit serum did not associate with either region ( Fig.   2A). The experiment was then performed on HCT116 ϩ/ϩ and ⌬/Ϫ cells, which revealed a decrease of ORC loading on Lamin B2 in ⌬/Ϫ cells (Fig. 2B). The amount of cross-linked DNA molecules pulled down in HCT116 cells was 3-to 7.7-fold higher than in the ⌬/Ϫ cells. This indicates that origin association of the entire ORC (including Orc4 and Orc6, whose levels are not changed) is dependent on the presence of Orc2. This extends our earlier observation that Orc4 loading decreases in a chromatin fractionation experiment (22). Interestingly, origin specificity (the ratio of ORC signal at Lamin B2 versus the control region) was retained in ⌬/Ϫ cells despite the low levels of ORC binding.
ORC association with origins recruits Cdt1 and Cdc6, which in turn recruit the MCM2-7 complex to complete a functional pre-RC (5, 6). To investigate pre-RC loading, we fractionated cells and examined MCM7 levels in the chromatin fractions. Chromatin-bound MCM7 was drastically decreased in the ⌬/Ϫ cells (Fig. 2c), which indicates that MCM7 loading, and thus pre-RC formation, is dependent on ORC loading. (The use of p53Ϫ/Ϫ cells as a further control is explained later.) The decrease in pre-RC loading suggests that replication may be affected in the ⌬/Ϫ cells, as chromatin-loaded MCM2-7 is required not only to recruit additional factors for replication but also to unwind the DNA for fork firing.
Chk2 Is Phosphorylated in ⌬/Ϫ Cells-The low levels of pre-RC are expected to lead to replication stress. This could cause fork stalling (and checkpoint activation) as forks extend from more sparsely firing origins. To test for checkpoint activation, we looked at phosphorylation of Chk1 at serine 317, a target of the DNA damage modulator ATR (37). We also looked at phosphorylation of Chk2 at threonine 68, a target of the other major modulator, ATM (38). Chk1 phosphorylation, although induced by hydroxyurea, is not increased in ⌬/Ϫ cells (data not shown.) However, we found that Chk2 phosphorylation at Thr-68 is increased in the ⌬/Ϫ cells (Fig. 3A). It therefore appears that a DNA damage checkpoint is activated in these cells.
To determine the downstream effects of Chk2 phosphorylation, we examined several known targets. Cdc25C is phosphorylated by Chk2 at serine 216 (39,40), which serves to inhibit its normal role in activating  A, chromatin immunoprecipitation was performed on HeLa cells using the indicated antibodies. Precipitated DNA was analyzed by quantitative PCR using primers against the Lamin B2 locus (origin) or an outside region (control). The y-axis indicates the amount of associated DNA. B, chromatin immunoprecipitation was performed using HCT116 and ⌬/Ϫ cells as in panel A. Note that even though associated DNA amounts are less in ⌬/Ϫ cells, origin binding is still specific (more Lamin B2 DNA than control DNA associates with the ORC proteins). C, chromatin fractionation shows a decrease in MCM7 loading. Cells were fractionated and immunoblotted for MCM7. Chromatin-bound proteins are found in the S2 fraction (nuclease soluble). A Ponceau stain of the membrane is shown as a loading control.
Cyclin B/Cdc2 (41), thus preventing mitotic entry. Phosphorylation of Cdc25C was slightly decreased in ⌬/Ϫ cells, indicating it is not a target of the active Chk2 (supplemental Fig. S2a). Cdc25A is also phosphorylated by Chk2, causing its degradation (42), which, in turn, would lead to increased Tyr-15 phosphorylation on Cdk2 (43,44). We did not, however, observe any decrease in the amount of Cdc25A in the ⌬/Ϫ cells (it was actually increased) (supplemental Fig. S2a). Likewise, the total amount of Tyr-15 phosphorylation on Cdk2 was less in ⌬/Ϫ cells, the opposite of what is expected if Cdc25A were inactive (supplemental Fig.  S2b). Therefore, Cdc25A and the inhibitory phosphorylation of Cdk2 do not seem to be affected by activated Chk2 in the ⌬/Ϫ cells.
Perhaps the most well studied target of Chk2 is p53. Chk2 stabilizes p53 by phosphorylating it at serine 20 (45,46). We were surprised to see that p53 was completely absent in these cells at both the protein and message level (Fig. 3, B and C). Induction with ␥ irradiation failed to induce p53 or p21 in ⌬/Ϫ cells, indicating this DNA damage response pathway is completely disrupted in these cells at a step after Chk2 activation.
To determine at which point p53 loss occurred compared with Orc2 loss, we looked at p53 levels in each of the ⌬/Ϫ precursor cell lines. Surprisingly, p53 was lost in the ⌬/ϩ cells created after Cre-mediated recombination (Fig. 3, D and E). These cells have a wild type allele of Orc2, and we have observed that Orc2 protein expression is similar to wild type cells. The decrease of p53 in these cells, therefore, predated the decrease in Orc2.
To ask whether p53 decrease could account for any observed phenotypes, we repeated several previous experiments with p53Ϫ/Ϫ HCT116 cells (47). The p53Ϫ/Ϫ cells actually expressed slightly higher levels of ORC proteins (data not shown) and did not show any pre-RC loading defect (Fig. 2C) when compared with wild type HCT116 cells. Likewise, the p53 Ϫ/Ϫ cells do not have increased Chk2 Thr-68 phosphorylation (Fig. 3A). These effects are unique to the ⌬/Ϫ cells and thus can be attributed to low levels of Orc2. We suspect that this prior loss of p53 allows the ⌬/Ϫ cells to survive with Chk2 activation by avoiding p53mediated cell cycle arrest and apoptosis.
Replication and Growth Rates-We have previously observed a growth defect in the ⌬/Ϫ cells (22). To determine the cause of this slower growth in ⌬/Ϫ cells, we examined the length of G 1 and S-phases during a cell cycle. Cells were arrested in nocodazole and released into regular medium. At different time points following release, cells were pulsed with 3 H-labeled thymidine. The time between nocodazole release and half maximum DNA incorporation approximately indicates G 1 length. Additionally, the length of S-phase can be determined by the width of the DNA incorporation peak. [ 3 H]Thymidine incorporation reaches half maximum slightly later in ⌬/Ϫ cells (7.4 h) compared with HCT116 (6 h) (Fig. 4A). Even though equal numbers of cells were plated, the ⌬/Ϫ cells incorporated much less radionucleotide. The major peak of replication seems to be delayed in the ⌬/Ϫ cells, most likely because of the delay in entry of cells into S-phase.
Low levels of DNA synthesis may indicate that many ⌬/Ϫ cells either never enter S-phase or that they enter S-phase but cannot continue DNA synthesis because of some defect. To distinguish these possibilities, we analyzed the percentage of cells entering S-phase by BrdUrd labeling at different time points following release from nocodazole arrest. These cells were fixed, labeled with fluorescein isothiocyanateconjugated anti-BrdUrd antibody, stained with propidium iodide, and analyzed using two-color FACS (Fig. 4B). HCT116 and p53Ϫ/Ϫ cells incorporate BrdUrd similarly at each time point, indicating there is no S-phase entry defect in the p53Ϫ/Ϫ cells. In contrast, a much smaller percentage of ⌬/Ϫ cells incorporate BrdUrd at every time point. Cells enter S-phase, as shown by an increase in this percentage as time passes, but they do so with a significant delay compared with controls. Recent work has described a possible role for mammalian Orc2 in mitosis (20). The failure of S-phase entry we observed may stem from a  mitotic deficiency due to low levels of Orc2. However, we did not observe any defect in mitotic exit as determined by chromatin condensation. Nocodazole-arrested HCT116 and ⌬/Ϫ cells showed equal DNA condensation in nocodazole and equal levels of decondensation 3 h after release (Fig. 4C). This indicates that there is no delay in exit from mitosis and that the S-phase entry delay is caused by a G 1 -S transition defect.
Orc2-deficient Cells Have Cell Cycle Progression Defects-⌬/Ϫ cells released from a nocodazole arrest appear to have difficulties transitioning from G 1 to S-phase. We therefore examined several cell cycle factors important for G 1 progression. Rb protein is a known regulator of the transition from G 1 to S-phase. Its hyperphosphorylation before S-phase by cyclin-dependent kinases (CDKs) causes its dissociation from the transcription factor E2F, which allows transcriptional activation of many genes required for cell cycle progression (for review, see Ref. 48). After release from nocodazole, ⌬/Ϫ cells have more hypophosphorylated Rb compared with wild type cells. (Fig. 5A) We also observed less phosphorylation of two specific sites on Rb, serine 780 and serine 807/ 11. This persistence of the inhibitory form of Rb indicates that cell cycle progression through G 1 is delayed in the ⌬/Ϫ cells.
We next examined Cdk2-associated kinase activity after release from a nocodazole arrest. Cyclin E/Cdk2 activity is required for S-phase entry, as well as for pre-initiation complex loading. Cdk2 kinase activity is significantly decreased in the ⌬/Ϫ cells (Fig. 5, B and C). Although the kinase activity does increase over time, it does so much more slowly than in the wild type cells, confirming that ⌬/Ϫ cells have difficulty progressing through G 1 into S-phase. Some of the most important regulators of Cdk2 kinase activity are its cyclin binding partners. In G 1 phase, Cyclin E/Cdk2 forms an active complex (49), so we investigated Cyclin E protein levels after release from nocodazole arrest. Although Cyclin E levels in HCT116 cells increase rapidly, Cyclin E increase is very much delayed in ⌬/Ϫ cells (Fig. 5B). The levels of Cyclin E seem to correlate with Cdk2-associated kinase activity in both cell lines, suggesting that the low Cyclin E levels are rate-limiting in ⌬/Ϫ cells. Cdk2 itself is not thought to cycle, and its levels are unchanged in both cell lines (Fig. 5B).
Cyclin E transcription is tightly regulated during the cell cycle. Northern blotting total RNA shows that Cyclin E message is drastically reduced in the ⌬/Ϫ cells at time points after nocodazole release (Fig.  5D). This transcript decrease could account for the low levels of Cyclin E protein, as well as the low Cdk2 kinase activity.
Rb is phosphorylated by both Cyclin D/Cdk4 and Cyclin E/Cdk2 (50 -54). As Cyclin E expression is mediated by E2F (55,56), low Cyclin E levels may be due to E2F inhibition by hypophosphorylated Rb. As E2F is autoregulatory (for review, see Ref. 57), E2F levels reflect E2F transactivation capacity. We find that levels of E2F are low after nocodazole release (Fig. 5E), suggesting that E2F activity is inhibited.
It has been previously reported that different sites on Rb are preferentially phosphorylated by different cyclin/CDK pairs. Ser-780 is thought to be phosphorylated by Cyclin D/Cdk4 (53,58). Decreased Cyclin D/Cdk4 activity might explain the Rb hypophosphorylation on Ser-780. However, Cdk4-associated kinase activity after nocodazole release was similar between the two cell lines at all time points (Fig. 5, E  and F), suggesting other factors may be required for Rb phosphorylation at Ser-780. Therefore, the primary defect seems to be a failure of E2F activation, perhaps due to failure to phosphorylate Rb.
These observations implicate a defect in progression through G 1 in the ⌬/Ϫ cells. Although we have observed an increase in Chk2 phosphorylation in the ⌬/Ϫ cells, it is unclear whether this is responsible for the G 1 delay. In fact, we have observed that Cdc25A (a positive factor for Cdk2 activation) is not inhibited in asynchronous cells and is most likely not involved. To determine whether Chk2 is relevant for the G 1 delay, we looked at Chk2 phosphorylation levels after nocodazole release. Chk2 phosphorylation increases in both cell lines 3 h after release (Fig.  5A). Although this phosphorylation subsequently decreases in HCT116, it persists longer in the ⌬/Ϫ cells. Therefore, it is possible Chk2 activation plays an as yet unknown function in G 1 regulation. However, small interfering RNA-mediated knockdown of Chk2 did not increase BrdUrd incorporation in the ⌬/Ϫ cells (data not shown.) Cells That Enter S-phase Can Initiate Replication-Despite the S-phase entry difficulty, a population of the ⌬/Ϫ cells does enter S-phase in each cell cycle, allowing the line to survive. We can therefore examine S-phase dynamics with low levels of Orc2 without the influence of the Cyclin E activation defect. To avoid this influence, we used thymidine/aphidicolin to arrest the cells in very early S-phase.
We first examined the ability of the cells to progress through S-phase after thymidine/aphidicolin release. The percentage of HCT116 and ⌬/Ϫ cells incorporating BrdUrd is very similar at time points after release (Fig. 6A). This indicates that once cells have entered S-phase, they are able to proceed normally through the replication phase. Rb phosphorylation was equal between the two cell lines after thymidine/ aphidicolin release (Fig. 6B), suggesting Cyclin E/Cdk2 kinase activity is normal in the ⌬/Ϫ cells once they have entered S-phase. Therefore, cells that enter S-phase have enough Cyclin E/Cdk2 kinase activity to phosphorylate Rb and promote the G 1 -S transition.
Once ⌬/Ϫ cells enter S-phase, they seem to be able to proceed normally with replication. We have shown that initiation at ␤-globin and c-Myc origins was normal in ⌬/Ϫ cells (22) but could not rule out that these origins were exceptional. To look at origin firing globally, we used DNA combing on cells that have entered S-phase (32). Briefly, cells were synchronized with thymidine/aphidicolin and released in the presence of BrdUrd. Genomic DNA was combed and prepared for immunofluorescent microscopy. DNA lengths labeled with BrdUrd were compared with total lengths to determine the rate at which new forks appeared during the first 160 min of S-phase. Finally, total DNA amounts were normalized by the percentage of cells actually entering S-phase as determined in Fig. 6A. The average fork density (all three time points together) is similar among the three cell lines (Fig. 6C). These results indicate that similar numbers of forks are firing in the cell lines. Furthermore, the replication progression (percentage of DNA replicated at a given time) is also very similar between ⌬/Ϫ and wild type cells (Fig. 6C). These results suggest that once ⌬/Ϫ cells enter S-phase, they are able to fire forks relatively normally despite very low levels of pre-RC loading.

DISCUSSION
One of the most immediate ramifications of low Orc2 levels is the equal decrease of Orc3 protein and, to a lesser extent, Orc5 protein. These decreases are not transcriptional, as Orc3 and Orc5 mRNA levels are unchanged between HCT116 and ⌬/Ϫ cells. The half-life of Orc3 is significantly reduced in the ⌬/Ϫ cells, but this instability is not caused by proteasomal degradation. Therefore, Orc3 seems to be more prone to non-proteasomal forms of degradation in the absence of Orc2. Biochemical studies of the human ORC proteins show a tight binding between Orc2 and Orc3 (14,15); this interaction may serve to stabilize Orc3. When Orc3 cannot bind to Orc2, residues normally buried in the Orc2/3 interface would be exposed. This could result either in degradation by an unknown active mechanism or loss of structural integrity, allowing the protein to be degraded by housekeeping mechanisms. Dhar et al. (14) also reported that Orc5 is able to bind Orc2/3, forming a stable subcomplex. The proteasome-independent decrease in Orc5 levels may also be due to a decrease in stability resulting from the absence of binding partners. The other ORC subunits are unaffected by loss of Orc2/3 (or Orc5), supporting the notion that Orc1, Orc4, and Orc6 form relatively labile interactions with Orc2, 3,5. As several members of the ORC are decreased in ⌬/Ϫ cells, it is not surprising that the origin loading of those subunits is decreased. However, Orc4 and Orc6 loading is drastically reduced despite unchanged protein levels, which indicates that their chromatin loading is dependent on the presence of the Orc2,3,5 subcomplex. The lack of ORC origin binding also results in decreased MCM2-7 loading on chromatin in human cells, which confirms results seen in other model systems.
The overall decrease in Orc2 seems to cause several unique problems for replication. ⌬/Ϫ cells released from mitosis begin incorporating labeled nucleotides later than controls, and the total level of labeling is decreased despite equal starting numbers of cells (Fig. 4A). Fewer ⌬/Ϫ cells enter S-phase, and those cells that do enter seem to do so later than control cells. This difficulty in entering S-phase is not due to any mitotic defect resulting from low ORC but is a defect in the G 1 /S transition.
Cyclin E protein and message levels are decreased in G 1 in the ⌬/Ϫ cells, and Cdk2 kinase activation is significantly delayed. This kinase activation is one of the critical events in the G 1 to S transition. In addition to phosphorylating various factors required for replication fork loading, Cyclin E/Cdk2 also phosphorylates Rb, which allows release of the transcription factor E2F. E2F is then free to activate transcription of a variety of important cell cycle genes, including Cyclin E, and many different replication factors (59). We found that the ⌬/Ϫ cells have a large population of hypophosphorylated Rb, further preventing entry into S-phase. Additionally, E2F levels are low, suggesting that the G 1 -S signal amplification between Rb, E2F, and the CDKs is disrupted in the ⌬/Ϫ cells. Because these proteins form a complex, cross-and autoregulatory amplification signal cascade, it is difficult to identify the initiating event. However, despite previous reports of Cyclin D/Cdk4 involvement in Ser-780 phosphorylation on Rb (an initial event required before further phosphorylation), we find the Cdk4-associated kinase activity is unchanged in the two cell lines. This suggests that although Cyclin D/Cdk4 may be essential for Ser-780 phosphorylation, it is not sufficient. An additional factor seems to be missing, preventing phosphorylation of Ser-780 in cells with low Orc2 levels. The resulting G 1 delay explains why fewer ⌬/Ϫ cells enter S-phase and could also account for the overall slower apparent proliferation rate of these cells.
Despite this G 1 delay, it appears that once cells exit G 1 , they can proceed through S-phase relatively normally. ⌬/Ϫ cells released from a thymidine/aphidicolin arrest have no cyclin/CDK defect, have equivalent fork firing, and show normal Rb phosphorylation. We imagine that some cells are able to amplify the G 1 signal cascade, eventually increasing Cyclin E protein levels and thus Cyclin E-associated kinase activity. Once the kinase activity reaches a threshold, a given cell is able to enter S-phase.
It is interesting to speculate which ORC-dependent events are important for cell cycle progression through G 1 . One possibility is that pre-RC loading onto chromatin is monitored so that the cells only proceed into S-phase when a minimum threshold is reached. Allowing the ⌬/Ϫ cells extra time in early S-phase may give them the opportunity to reach such a threshold, at which point the origins fire and replicate normally.
Even if a pre-RC loading threshold exists, the proliferation of the ⌬/Ϫ cells is still striking when one considers the huge decrease in pre-RC loading observed in these cells (Fig. 2). Additionally, we do not see a significant increase in pre-RC loading in thymidine/aphidicolin-arrested cells (data not shown). These results suggest that pre-RCs are loaded in vast excess of a threshold required for a functional replication event, supporting an earlier model termed the "Jesuit Model," in which many pre-RCs are "called" but few are chosen (60). Although the ⌬/Ϫ cells have fewer potential origins available for firing, the amount remaining is sufficient for normal replication in those cells that can enter S-phase. However, this also suggests that any potential selection of the best origins may be lost in the ⌬/Ϫ cells, perhaps leading to subtle detrimental effects during replication. Future work may be able to determine whether such changes in origin selection occurs and the ramifications of losing such selectivity. Although the cyclin/CDK machinery is known to regulate replication initiation factors, evidence is emerging that the converse may also be true, at least in mammalian cells. Overexpression of a stable form of geminin in G 1 cells to inhibit the loading of MCM proteins activated a checkpoint in primary cells that led to G 1 arrest and inhibition of Cdk2 (61). In that study, though, the "licensing checkpoint" was not apparent in cancer cells. Acute depletion of Orc2 by transient transfection of small interfering RNA duplexes in a non-cancerous breast epithelial cell and in primary fibroblasts was shown recently to increase Cdk2 inhibitors p27 and p21 and thus inhibit Cyclin E/Cdk2 activity (21). We are impressed, therefore, by the inhibition of Cyclin E/Cdk2 that we see in the Orc2 ⌬/Ϫ colon cancer cells where Orc2 levels have been chronically depressed and the cells had a chance to select for compensatory changes in the cell cycle machinery. The exact mechanism of cyclin/ Cdk2 inhibition is different in the two studies from our laboratory, because the primary defect in the cancer cells chronically decreased for Orc2 appears to be in the up-regulation of Cyclin E without any induction of Cdk2 inhibitors (e.g. p27 in Fig. 1A). The end result, though, is the same. The cell is prevented from prematurely activating Cyclin E/Cdk2, which would have inhibited further loading of pre-RCs and thus condemned the cell to DNA replication with suboptimal amounts of origin licensing and attendant problems in genomic stability. Clearly, much more needs to be known to establish the mechanism by which decrease of ORC leads to inhibition of Cyclin E/Cdk2 activity. Given the pleiotropic effectors of the checkpoint, increase of p21 and p27 or decrease of Cyclin E, it is tempting to speculate that changes in chromatin structure due to decrease in ORC affects the transcription of key genes leading to cell cycle inhibition. It is also quite interesting that if the Cyclin E/Cdk2 activation defect can be overridden, the cells are able to proceed normally, even with low levels of Orc2. This suggests that in this system Orc2 plays a more important role in Cyclin E up-regulation than in its previously described functions for replication initiation.
The link between replication initiators and the CDK cycle also suggests that hypomorphic mutations in genes encoding replication initiators might be lurking in cancers with mutations in the checkpoint pathways implicated here. Loss of p53 is a common event in human cancers, and the discovery that the ⌬/Ϫ cells are also p53Ϫ suggests that p53Ϫ cancers may be more tolerant of hypomorphic mutations in initiation factors. Likewise, decrease of p27 or increase of Cyclin E have both been described in human cancers, and our results suggest that hypomorphic mutations in initiation factors could be more dangerous in these backgrounds. First, the loss of the licensing checkpoint will allow the cancer cells to proliferate despite the low levels of initiation factors. Second, DNA synthesis from few licensed origins may lead to fork stalling and DNA breaks that would promote extensive genomic instability.