Diminished Origin-Licensing Capacity Specifically Sensitizes Tumor Cells to Replication Stress Free

This article is featured in Highlights of This Issue, p. 313

To replicate the human genome in a timely manner, replication is initiated bidirectionally from multiple origins. However, this necessitates that replication origins only fire once during each cell cycle to avoid rereplication. Origin licensing occurs from late mitosis to G₁ phase and involves assembly of the origin recognition complex (ORC), encompassing ORC1 to ORC6, onto origin sequences (1, 2). Together with CDC6 and CDT1, ORC loads the heterohexameric MCM2-7 complex which provides helicase activity, generating the prereplication complex (pre-RC; refs. 3–6). Cells exploit several mechanisms to prevent origin refiring, including the inhibition of MCM2-7 loading onto origins during S–G₂ and the tight regulation of other pre-RC components via proteasome-mediated degradation (1, 2).

Only a fraction of licensed origins are used for replication, with nonfired origins being considered dormant (7–9). Following replication stress, the activation of checkpoint kinases stabilizes stalled replication forks (10). In addition, dormant origins can be exploited to promote recovery from replication stress (11–15). However, replication fork stalling also activates an intra-S-phase checkpoint response that inhibits late-firing origins (16–18). Although apparently conflicting with the notion that replication fork stalling exploits dormant origins, recent studies have shown that origins are organized in clusters, with activation occurring stochastically and inactivating further origins within the cluster (14, 19). Although damage response signaling inhibits the firing of origins in new clusters, a distinct process promotes dormant origin firing within a cluster in which double fork stalling has occurred (14). This model is intrinsically appealing as it implies that dormant origins are only activated near a stalled fork where they are needed, whereas new replication is diminished elsewhere to preclude further replication in the presence of DNA damage. Support for this model has come from studies involving siRNA-mediated depletion of MCM2-7 in human cells, which suppresses dormant origin usage, inhibits the rate of DNA synthesis, and reduces cell survival in response to replication-inhibiting agents (11–13).

Until recently, studies on dormant origin usage were predominantly undertaken in tumor cells and it was unclear whether the same process occurs in primary cells. Interestingly, an increase in the number of stalled replication forks in unchallenged S-phase cells was observed in mouse embryonic fibroblasts (MEF) expressing a hypomorphic Mcm4^Chaos3 allele, which impairs MCM2-7 complex stability and reduces the number of dormant origins (15). Significantly, Mcm4^Chaos3 mice are cancer prone (20). Importantly, Mcm4^Chaos3 cells have a normal rate of replication and helicase activity. Thus, reducing the number of dormant origins need not affect replication but can impede recovery from either endogenous or damage-induced replication stress. Although other pathways of replication fork recovery exist, a failure to use dormant origins is proposed to cause genomic instability.

Two recent studies identified mutations in origin-licensing components (ORC1, ORC4, ORC6, CDT1, and CDC6) in Meier–Gorlin syndrome (MGS), a disorder characterized by microcephaly, proportionate dwarfism, and bone abnormalities including small or absent patellae (21–23). Cells from patients with MGS, despite having substantially reduced origin licensing capacity, grow well in culture consistent with the notion that only a fraction of licensed origins are required to sustain replication (22).

Carcinogenesis necessitates multiple genetic changes to support often rapid and uncontrolled proliferation. Most tumor cells suffer high replication stress, due to uncontrolled proliferation and/or enhanced genomic instability. Interestingly, several studies have reported that origin-licensing proteins are overexpressed in tumor-derived cell lines (24–27). Given this, we reasoned that tumor cells might have a greater demand for origin licensing than nontransformed cells, either to sustain rapid replication and/or to enhance recovery from the increased level of replication stalling/collapse. Given the finding that nontransformed cells can grow efficiently with substantially reduced licensing capacity, we considered that ORC proteins might represent targets to specifically sensitize tumor cells. Here, we examine this possibility by investigating the impact of diminished origin licensing capacity in tumor versus nontransformed cells. Strikingly, our results suggest that tumor cells more frequently rely on dormant origin usage following exposure to agents that cause replication stress compared with nontumor cells.

Cell lines were purchased from the American Type Culture Collection (ATCC) or established and authenticated in-house or by scientific collaborators indicated in references. All cell lines were tested for mycoplasma contamination before use and assessed for ORC1 expression by immunoblot. Control primary skin fibroblasts (48BR), control hTERT-immortalized fibroblasts 1BR3hTERT or BJhTERT (ATCC), and ORC1-P4hTERT, derived from an ORC1-deficient patient with MGS, were cultured in Dulbecco's modified Eagle's medium supplemented with 15% fetal calf serum (FCS, Invitrogen; ref. 22, 28–30). Medium for BJ-MYC-ER, a derivative of BJhTERT expressing a tamoxifen-inducible c-Myc gene, was supplemented with 2 μg/mL puromycin (Invitrogen). MRC-5 is a primary fetal lung fibroblast cell line. MRC5, U2OS, and HeLa cells (ATCC) were cultured in minimal essential medium containing 10% FCS. Cells were transfected with siRNA oligonucleotide pools (Thermo Scientific Dharmacon; ORC1, p53, ORC6, or CDC6) or Stealth siRNA targeting ORC1 (Invitrogen; ref. 22) using HiPerFect (Qiagen) or DharmaFECT (Thermo Scientific Dharmacon). siControl represents scrambled oligonucleotides (Thermo Scientific Dharmacon).

siRNA-transfected cells were seeded in 96-well dishes, treated as described and viability was assessed using the CellTiter-Blue assay (Promega). Viability was normalized to the siRNA-transfected but untreated control. The half maximal inhibitory concentration (IC₅₀) values from viability curves were calculated with SigmaPlot (Systat Software, Inc.) using the five-parameter logistic nonlinear regression model. IC₅₀ values represent the mean ± SD of 3 independent experiments.

For whole-cell extracts, cells were lysed in buffer A (22) containing 500 mmol/L NaCl and supplemented with protease (Sigma-Aldrich) and phosphatase inhibitors (Thermo Fisher Scientific) for one hour on ice and sonicated at 4°C. Fractionation of chromatin-bound and unbound proteins was conducted as previously described (22). Lysates were resolved by electrophoresis, transferred onto polyvinylidene difluoride (GE Healthcare) and immunoblotted using α-ORC1, ORC6, CDC6, p53 (Santa Cruz Biotechnology, Inc.), or β-actin (Abcam) antibodies.

siRNA-transfected cells were seeded into dishes and grown for 48 hours. Hydroxyurea (Sigma-Aldrich) or H₂O₂ (Fisher Scientific/Acros Organics) was added for the indicated times. Cells were washed 3 times in PBS and incubated in fresh medium as indicated.

Clonogenic survival was assessed as previously described (31). Briefly, cells were transfected with siRNA, treated as described, and incubated for 10 days (U2OS) or for 21 days using irradiated feeder cells to enhance plating efficiency (1BR3hTERT). Survival was normalized to the siRNA-transfected but untreated control. Plotted values represent the mean ± SD of 3 independent experiments.

Cells were labeled with 25 μmol/L chlorodeoxyuridine (CldU) for 20 minutes, washed 3 times with medium, incubated in 2 mmol/L hydroxyurea for 24 hours, washed 3 times again, and pulse-labeled with 250 μmol/L iododeoxyuridine (IdU) for 1 hour. Labeled cells were harvested and DNA fibre spreads were prepared as previously described (32). CldU and IdU were detected by incubating acid-treated fibre spreads with rat α-bromodeoxyuridine (BrdUrd) (1:1,000, AbD Serotec) and mouse α-BrdUrd (1:750, Becton Dickinson) monoclonal antibodies for 1 hour. Slides were fixed with 4% paraformaldehyde and incubated with AlexaFluor 555-conjugated goat α-rat IgG (1:500, Molecular Probes) and AlexaFluor 488-conjugated goat α-mouse IgG (1:500, Molecular Probes) for 1.5 hours. Images of DNA fibres were acquired on a Nikon E600 microscope using a × 60 (1.3NA) lens, a Hamamatsu digital camera, and the Volocity package (Perkin Elmer). For quantification, at least 130 structures were counted per experiment using ImageJ software (NI H, http://rsbweb.nih.gov/ij/).

siRNA-transfected cells were treated with hydroxyurea and labeled with 50 μmol/L BrdUrd; Becton Dickinson 30 minutes before harvesting. Cells were fixed, BrdUrd-labeled, propidium iodide-stained and analyzed by fluorescence-activated cell sorting (FACS) as previously described (31).

Cells were processed for γ-H2AX analysis as previously described (33) using α-γ-H2AX and α-CENPF (Abcam) and 4′, 6–diamidino–2–phenylindole (DAPI) labeling of DNA. Cells harboring S-phase–associated DNA damage, referred to as γ-H2AX⁺, were detected by bright, pan-nuclear γ-H2AX and minimal CENPF signal (34, 35). γ-H2AX⁺ cells were manually scored in more than 500 cells per condition. Images of cells were acquired with identical exposure settings on a Zeiss Axioplan2 microscope using a × 40 (0.75 NA) lens, a Hamamatsu digital camera, and SimplePCI software (Hamamatsu).

We aimed to compare how diminished origin licensing capacity affects recovery from replication stress in tumor-derived versus nontumor cell lines. We used 1BR3hTERT and U20S osteosarcoma cells as nontumor and tumor cell lines, respectively. 1BR3hTERT, an hTERT-immortalized fibroblast line derived from a normal individual, has a stable karyotype, shows genomic stability, and has an intact G₁–S checkpoint (Unpublished observations). Because ORC1 is essential, we sought knockdown conditions that reduce ORC1 protein levels without impeding proliferation. Viability, monitored using the CellTiter-Blue assay, was assessed following siRNA-mediated knockdown of ORC1 in 1BR3hTERT and U20S using a range of siRNA oligonucleotide (scrambled or ORC1-specific) concentrations. We observed diminished proliferation with increasing siORC1 concentrations, consistent with the notion that oligonucleotide concentration correlates with knockdown efficiency (Fig. 1A and B). Since tumor and nontumor cell lines differ in the efficiency of siRNA-mediated knockdown and requirement for ORC1, the impact was distinct for each line. The highest siORC1 concentration that did not significantly impede viability was 5 or 0.6 nmol/L for 1BR3hTERT and U20S, respectively (Supplementary Fig. S1A). Immunoblotting revealed that U20S has higher ORC1 protein levels compared with 1BR3hTERT, consistent with previous findings that tumor-derived cells overexpress ORC proteins (Fig. 1C; ref. 24–27). Although α-ORC1 antibodies are inefficient for immunoblotting, a marked reduction in ORC1 protein could be observed in both lines following siORC1 (Fig. 1C). Routinely, low residual ORC1 was detectable in siORC1-treated U20S cells, whereas residual ORC1 was not detectable in siORC1-transfected 1BR3hTERT. As ORC1 is essential, it is likely that 1BR3hTERT retain residual, although undetectable, ORC1. This suggests that U20S cells require a higher level of ORC1 to maintain viability compared with 1BR3hTERT, consistent with the notion that tumor cells have a greater need for origin-licensing proteins compared with nontumor cells. Having identified knockdown conditions that substantially deplete ORC1 without impeding viability, which we anticipated would substantially reduce the level of dormant origins, we proceeded to examine the impact on recovery from damage-induced replication arrest.

We examined sensitivity to hydroxyurea, which depletes ribonucleotide reductase and enhances fork stalling/collapse, in 1BR3hTERT and U20S following siControl or siORC1. siRNA transfection was conducted as described above and cells were exposed to differing concentrations of hydroxyurea for 24 hours. Hydroxyurea was removed and viability monitored 4 days later (Fig. 1D and E). To compare the effect of siORC1 between the cell lines, we estimated the hydroxyurea concentration that reduced viability by 50% (the IC₅₀ value). The relative IC₅₀ value was calculated by comparison with the IC₅₀ of siControl-transfected cells (Fig. 1H). Strikingly, while siORC1 did not affect hydroxyurea sensitivity in 1BR3hTERT, it significantly enhanced sensitivity in U20S. Similar effects were observed following transfection of cells with a distinct pool of ORC1 siRNA oligonucleotides (Supplementary Fig. S1B–S1D). To verify that the resistance of 1BR3hTERT cells is not simply a consequence of their slower cell-cycle progression, resulting in a lower fraction of cells progressing into S-phase, we monitored the population doubling time (Supplementary Fig. S1E-S1F) and rate of hydroxyurea -induced γ-H2AX formation in the 2 cell lines (Fig. 1F and G). We estimated that by 48 hours, all 1BRhTERT cells had undergone replication fork arrest after hydroxyurea. However, examination of viability following 48-hour hydroxyurea treatment yielded similar results (Fig. 1H). Thus, the resistance of 1BR3hTERT cells to siORC1-induced hypersensitivity is not explained by their slower cell-cycle progression. To verify that the viability assay reflects clonogenicity, we also examined clonogenic survival of U20S following 24-hour hydroxyurea treatment (Fig. 1I) and of 1BR3hTERT following 48-hour hydroxyurea treatment (Fig. 1J). Although these 2 assays monitor different endpoints, a similar impact was observed, validating use of the viability assay.

To substantiate these findings without relying on siRNA-mediated depletion, we also examined an hTERT-immortalized fibroblast line derived from an ORC1-deficient MGS patient (ORC1-P4hTERT; ref. 22). ORC1 is expressed at normal levels in ORC1-P4hTERT but chromatin binding of ORC1 is impaired (ref. 22; Fig. 1K).We observed resistance rather than marked sensitivity of ORC1-P4hTERT cells to hydroxyurea at higher concentrations potentially due to a lower number of replication origins (Fig. 1L). We also adjusted hydroxyurea treatment time to achieve complete hydroxyurea-induced S-phase arrest in 1BR3hTERT and ORC-P4hTERT (representing 48 or 72-hour exposures, respectively). Under these conditions, resistance, but not sensitivity, to 0.5 mmol/L hydroxyurea was also observed (Fig. 1M).

To examine whether siORC1 affects replication recovery, 1BR3hTERT or U20S were treated with siRNA as described above and exposed to 2 mmol/L hydroxyurea for 24 hours. Following hydroxyurea removal, cells were incubated for 2, 4, or 24 hours, and BrdUrd added for the final 30 minutes. BrdUrd incorporation, representing recovery of DNA synthesis, was assessed by FACS (Fig. 2A and B). Hydroxyurea treatment abolished BrdUrd incorporation in both cell lines, consistent with replication inhibition. Strikingly, in 1BR3hTERT, BrdUrd incorporation was substantially recovered at 2 hours after hydroxyurea removal and was similar in siControl or siORC1-treated cells. In marked contrast, although siControl-treated U20S cells also recovered DNA synthesis at 2 hours following hydroxyurea removal, DNA synthesis was dramatically reduced at this time in siORC1-treated U20S cells. Because this difference is observed at early times (2 hours) after hydroxyurea removal, this suggests that siORC1 does not impair rapid recovery of replication in hydroxyurea -treated 1BR3hTERT but does so in U20S.

We also examined whether the inability of siORC1-treated U2OS to recover replication causes accumulated DNA damage. Either immediately (0) or 24 hours following treatment with 2 mmol/L hydroxyurea, cells were examined for γ-H2AX, a marker of DNA damage, and CENPF, a G₂–M-phase marker. Immediately following hydroxyurea treatment, most cells were CENPF⁻ (consistent with S-phase arrest) and showed pan-nuclear γ-H2AX staining, showing the presence of collapsed/stalled replication forks; untreated cells had a lower fraction of γ-H2AX⁺ cells (Fig. 2C and D). 24 hours after hydroxyurea removal, few siControl or siORC1-transfected 1BR3hTERT cells were γ-H2AX⁺ (Fig. 2D), consistent with the observed recovery of replication. Similarly, the number of γ-H2AX⁺ siControl-treated U20S cells was dramatically reduced 24 hours after hydroxyurea removal (Fig. 2C and D). In stark contrast, approximately 50% of siORC1-treated U20S cells retained γ-H2AX staining 24 hours after hydroxyurea removal. γ-H2AX⁺ cells were negative for the G₂–M marker, CENPF, consistent with the notion that they represent damaged S-phase cells. This analysis shows that 1BR3hTERT undergo replication fork arrest and activate the DNA damage response after hydroxyurea treatment but efficiently recover despite substantial depletion of ORC1.

The DNA fibre assay allows replication at new versus preexisting origins to be monitored (36).We exploited the technique to assess new origin firing in U2OS cells after release from hydroxyurea exposure. siRNA-transfected cells were exposed to CldU for 20 minutes, then either exposed to IdU for 20 minutes (control) or CldU was washed out, and cells were exposed to hydroxyurea for 24 hours. Following hydroxyurea removal, IdU was added for 1 hour. CldU⁺/IdU⁻ replication tracks are considered to represent stalled forks that have not reinitiated replication; CldU⁻/IdU⁺ tracks represent ones with newly fired origins (Supplementary Fig. S2C–S2D). To monitor new origin firing, the fraction of CldU⁻/IdU⁺ tracks was assessed. In untreated U20S cells, ORC1 siRNA did not significantly impact upon new origin firing (Fig. 2E). Following hydroxyurea, although siORC1 did not impact upon the level of stalled forks, new origin firing was substantially diminished. These findings strongly suggest that siORC1 diminishes replication restart by new origin firing after hydroxyurea while not affecting new origin firing in unperturbed cells.

We next examined whether the sensitivity of U2OS cells to hydroxyurea is impacted following depletion of additional origin-licensing factors. We examined ORC6 and CDC6 as they are also causal defects for MGS (22). A total of 0.6 nmol/L siORC6 or CDC6 substantially depleted ORC6 or CDC6 but did not impede cellular proliferation (Fig. 3A; data not shown). Viability assessment revealed a similar level of hydroxyurea sensitivity following siORC6 or CDC6 to that observed following siORC1 (Fig. 3B and C). However, neither siORC6 nor siCDC6 affected hydroxyurea sensitivity in 1BR3hTERT (Supplementary Fig. S3).

To extend our findings, we examined additional nontumor fibroblasts (BJhTERT, 48BR, and MRC-5) and tumor-derived lines (HeLa and MDA-MB-231). 48BR and MRC5 represent primary fibroblast lines to complement the analysis of the hTERT-immortalized line. For all lines, we examined the optimum siORC1 oligonucleotide concentration that failed to impact upon viability (Supplementary Fig. S4). BJhTERT cells showed slightly diminished viability above 5 nmol/L siORC1 similar to 1BR3hTERT cells; 5 nmol/L was chosen for analysis (Supplementary Fig. S4A). HeLa cells were resistant to high oligonucleotide concentrations; 0.6 nmol/L was chosen to allow comparison to U20S (Fig. 4A and B, Supplementary Fig. S4D). 48BR, MDA-MB-231, and MRC-5 displayed diminished viability above 1 nmol/L siORC1; 1 nmol/L was chosen for analysis (Supplementary Fig. S4B, S4C, and S4E). The time of hydroxyurea treatment required to achieve complete hydroxyurea -induced S-phase damage was also assessed in each cell line (Supplementary Fig. S5). The time indicated was used for subsequent viability experiments. Similar to 1BR3hTERT and U20S cells, although ORC1 protein levels were greater in siORC1-depleted HeLa and MDA-MB-231 cells compared with BJhTERT, 48BR, or MRC-5 cells, siORC1 enhanced hydroxyurea sensitivity of HeLa and MDA-MB-231 without substantially impacting upon sensitivity of BJhTERT, 48BR, or MRC-5 cells (Fig. 4A, C–G). For MRC-5, slightly enhanced sensitivity was observed at high hydroxyurea doses but there was no impact on the IC₅₀ value.

Having shown that siORC1 hypersensitizes U20S and HeLa but not 1BR3hTERT or BJhTERT cells to hydroxyurea, we used a similar approach to evaluate whether recovery from oxidative damage, which can indirectly induce replication stress, might also involve differential dormant origin usage. Strikingly, while siORC1 did not affect sensitivity of 1BR3hTERT or BJhTERT to H₂0₂, U20S and HeLa cells showed marked hypersensitivity (Fig. 5A–E). The slightly higher resistance of siORC1-treated 1BR3hTERT cells to H₂0₂ compared with siControl cells likely reflects their slightly slower replication. Nonetheless, the distinction between 1BR3hTERT/BJhTERT and U20S/HeLa cells to combined siORC1 and H₂0₂ was marked.

p53 loss abrogates the damage-induced G₁–S checkpoint, enhancing S-phase progression and replication stalling (37). In addition, p53 is required for a licensing checkpoint which prevents S-phase entry until sufficient origins have been licensed (38, 39). We examined whether sip53 in 1BR3hTERT affects viability and hydroxyurea sensitivity following siORC1. 1BR3hTERT cells were transfected with siControl, sip53, siORC1 or combined siORC1+sip53 for 48 hours, then viability assessed in untreated or hydroxyurea -treated cells as described above. p53 was efficiently depleted in 1BR3hTERT (Fig. 6A). As above, siORC1 did not affect the viability of 1BR3hTERT cells (Fig. 6B). sip53 alone slightly inhibited viability but combined sip53+siORC1 diminished viability by approximately 1.7 (Fig. 6B). These findings suggest that in undamaged cells, siORC1 more markedly affects viability in the absence of p53. siORC1 did not significantly affect hydroxyurea sensitivity similar to the findings in Fig. 1D; there was a modest but not statistically significant impact of sip53 on hydroxyurea sensitivity but a marked decrease following sip53+siORC1 (Fig. 6C and D).

Myc overexpression, which enhances proliferation and replication stress, is frequently observed during carcinogenesis (40–42). We examined whether Myc expression influences the requirement for origin licensing capacity using a BJhTERT derivative that expresses c-Myc fused to a tamoxifen-inducible estrogen receptor (40–42). First, anticipating that tamoxifen concentration affects the level of c-Myc expression, we estimated the tamoxifen concentration promoting endogenous DNA damage (assessed by γ-H2AX). 24 hours after 2 μmol/L tamoxifen, 30% of siControl-transfected cells were γ-H2AX⁺, suggesting that Myc overexpression induces replication stress (Fig. 7A). Following siORC1, 60% of BJhTERT cells were γ-H2AX⁺, raising the possibility that siORC1 causes enhanced or persistent c-Myc–induced replication arrest. Next, assessment of viability 5 days following exposure to different tamoxifen concentrations revealed substantial sensitivity following siORC1 (Fig. 7B and C) suggesting that depletion of origin-licensing capacity diminishes the ability to cope with c-Myc–induced replication stress.

We previously observed that MGS patient-derived cell lines grow efficiently in culture despite 10-fold lower levels of origin-licensing components (21, 22). This ability to sustain substantial proliferation is consistent with findings that only 10% of licensed origins are used during unchallenged replication (7–9). Recent studies have provided evidence that dormant origins can be used to promote recovery from replication fork stalling or collapse (11–15). Because tumor cells show elevated oxidative and replicative stress, we predicted that they might have an enhanced reliance on origin-licensing capacity compared with normal cells, raising the possibility that targeting origin-licensing components could specifically inhibit cancer cell growth. Here, we evaluate this possibility.

Consistent with previous findings, we observed that 3 tumor-derived cell lines showed high ORC1 expression compared with nontumor lines (24–27). In general, higher ORC1 levels were required to maintain viability in the tumor lines (although MRC-5 cells also had a requirement for a higher ORC1 level). Nonetheless, the ability to detect higher residual ORC1 in the tumor than nontumor cell lines shows that our findings cannot simply be explained by more efficient ORC1 knockdown in the tumor versus nontumor cells.

To enhance the applicability to exploit inhibition of origin licensing for tumor therapy, we examined whether partial ORC1 depletion affected the response to hydroxyurea, a chemotherapeutic agent. Significantly, we observed marked sensitization of the tumor -derived lines to hydroxyurea (or H₂0₂) compared with the nontumor cells. In addition, we observed that ORC1 depletion enhanced hydroxyurea sensitivity of p53-depleted nontumor cells and also conferred sensitivity to Myc overexpression. This important result suggests that enhancing the level of replication stress and/or rate of proliferation, both of which arise following c-Myc expression, increases the demand for origin-licensing capacity.

Collectively, using our panel of 3 tumor and 4 primary or hTERT-immortalized cell lines, our findings suggest that tumor cells have a greater demand for origin-licensing capacity following replication fork arrest compared with nontumor lines and that loss of p53 or c-Myc expression enhances this demand in nontumor cells. Our findings could have several explanations. One possibility is that following replication stress, stalled forks more readily collapse in tumor compared with nontumor cells and that dormant origin firing enhances recovery from replication fork collapse. However, we are not aware of studies supporting this suggestion. Alternatively, it is possible that fork collapse occurs similarly in tumor and nontumor cells, but that tumor cells more frequently exploit dormant origins for recovery and are hence hypersensitive when this route is unavailable. Although the original studies describing dormant origin usage after replication stress used tumor cells, a recent study involving primary MEFs showed that they also exploit dormant origins for recovery from replicative stress (15). However, it is difficult to evaluate the comparative usage of dormant origins in tumor versus nontumor cells from these studies. An alternative and appealing possibility is that tumor cells override the origin-licensing checkpoint to enhance proliferation and, therefore, enter S-phase with diminished dormant origins. Indeed, the upregulation of origin licensing proteins in tumor cells may reflect their need to effect rapid origin licensing during their short G₁ phase. Thus, further reducing licensing capacity may provide a situation where there are insufficient dormant origins to exploit following replication fork arrest. It should be noted that our knockdown conditions were designed to prevent any impact on unperturbed cell growth. This model is consistent with the known function of p53 in enhancing G₁-phase progression and/or abolishing the G₁–S checkpoint. Importantly, in the present context, p53 is also required for the origin-licensing checkpoint as codepletion of p53 and CDC6, another licensing component, in normal fibroblasts permits S-phase entry with insufficient origin capacity (43). c-Myc also enhances G₁-phase progression and disrupts p53 activity (44, 45). However, both p53 loss and c-Myc expression have multiple additional impacts including an influence on replication (44–46). Thus, although the latter explanation is appealing, further work is required to define the basis underlying our observations. It is likely, moreover, that there could be multiple impacts. Our findings to date are based on a restricted number of tumor or nontumor cell lines. Nonetheless, the relationship seems marked and further work will be required to examine the extent to which this represents a phenotype of many tumor cell lines. It should be noted that around 50% of tumor cell lines show an upregulation of origin-licensing components (24–27).

Our goal was to examine whether the origin-licensing complex represents a suitable target to specifically sensitize tumor cells. Importantly, we report that 3 tumor cells require greater origin-licensing capacity following exposure to DNA damaging agents than 4 nontumor cells. In addition, we show that p53 loss (in the presence of hydroxyurea) or c-Myc expression in nontumor cells enhances the reliance on ORC1. Interestingly, the BAH domain of ORC1 was recently reported to bind H4K20me2 with high specificity and affinity via an aromatic cage, which could provide a route for drug targeting (47). In summary, we provide evidence that the downregulation of ORC1 and other origin-licensing proteins enhances the sensitivity of tumor but not nontumor cell lines to replicative stress, providing a potential route for specific sensitization of tumor cells.

No potential conflicts of interest were disclosed.

Conception and design: E. Petermann, P.A. Jeggo

Development of methodology: P.A. Jeggo, K.M. Zimmerman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.M. Jones, K.M. Zimmerman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.M. Zimmerman, R.M. Jones, P.A. Jeggo

Writing, review, and/or revision of the manuscript: K.M. Zimmerman, E. Petermann, P.A. Jeggo

Study supervision: P.A. Jeggo

The authors thank Drs. M. O'Driscoll for discussions and O. Fernandez-Capetillo for kindly providing BJ-MYC-ER.

The P.A. Jeggo laboratory is supported by an MRC programme grant, the Association for International Cancer Research, the Wellcome Research Trust, and the EMF Biological Research Trust. The E. Petermann laboratory is supported by an MRC project grant, Cancer Research UK, and the University of Birmingham.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.