Forced Mitotic Entry of S-Phase Cells as a Therapeutic Strategy Induced by Inhibition of WEE1

Cell-cycle progression is regulated by multiple overlapping checkpoints that stall the cell cycle in response to DNA damage or other cellular insults. Functionally distinct checkpoints have been described that block the initiation of S-phase (G₁–S checkpoint), the progression through S-phase (intra–S-phase checkpoint), the initiation of mitosis (G₂–M checkpoint), and the initiation of anaphase (spindle checkpoint; ref. 1). The disruption of these checkpoints provides a potential approach to cancer treatment, synergizing with chemotherapy to drive cells inappropriately onward through the cell cycle without first repairing DNA damage (2).

Entry into mitosis is a tightly regulated process ultimately driven by the cyclin B-CDK1 complex (3). In normal cells, cyclin B expression is low at the start of S-phase and peaks at the end of G₂ phase, leading to the build-up of cyclin B-CDK1 complexes. However, until the end of G₂ phase, cyclin B-CDK1 is kept inactive through inhibitory phosphorylation of cyclin-dependent kinase (CDK)1 at Thr14 and Tyr15 by WEE1 (predominantly Tyr15) and MYT1 (Thr14 and Tyr15). Mitotic entry is ultimately initiated by dephosphorylation of these residues by the CDC25 family of phosphatases, initiating a positive feedback loop, in which cyclin B-CDK1 inhibits its inhibitors (WEE1 and MYT1) and activates its activators (CDC25 phosphatases), ensuring rapid amplification of CDK1 activity (reviewed in ref. 3). In response to DNA damage, checkpoint kinases such as CHK1 (4) mediate the inhibition and/or degradation of the CDC25 phosphatases to inhibit CDK1 activation until the damage is repaired, a safeguard mechanism termed the G₂–M checkpoint (1).

Inhibitors of WEE1 have been developed as potential cancer therapeutics, originally conceived to synergize with chemotherapy by abrogating the G₂–M checkpoint and drive cells into mitosis with unrepaired DNA damage (reviewed in ref. 5). Previous work with chemical inhibitors of WEE1 and siRNA-mediated depletion (6) suggested that WEE1 inhibition abrogated the G₂ checkpoint and selectively sensitized p53-deficient cells to various DNA-damaging agents, such as gemcitabine, carboplatin, and cisplatin (7, 8), and inhibited tumor growth in in vivo models (7, 9). In addition, WEE1 inhibition or depletion sensitized tumor cells to ionizing radiation (10–13).

However, the proposed mechanism of sensitization via abrogation of the G₂–M checkpoint, discounts the importance of the intra–S-phase checkpoint that is engaged when replication fork progression is hampered in response to stresses such as depletion of deoxyribonucleotides [deoxynucleotide triphosphate (dNTP)] or DNA damage. The checkpoint machinery inhibits the initiation of DNA replication from unfired replication origins and protects the integrity of stalled replication forks (14). Here, we show that WEE1 inhibitors have a distinct and novel mechanism of action, forcing cells arrested in S-phase directly into mitosis. We go on to show that this mechanism is also shared, although to a lesser extent, by inhibitors of CHK1. We explore the pathophysiologic changes that underlie the reliance on WEE1 and CHK1 to prevent inappropriate engagement of mitosis while arrested in S-phase, and provide evidence that forced mitosis targets the oncogenic phenotype of poor prognosis cancers characterized by TP53 mutation and high expression of mitotic cyclins and polycomb protein EZH2.

We investigated whether synergy occurred between gemcitabine and the WEE1 inhibitor MK-1775 in a panel of 25 breast cell lines (23 breast cancer cell lines and 2 breast epithelial cell lines; Fig. 1A; Supplementary Table S1), using the combination index (CI) method of Chou and Talalay (15). Gemcitabine inhibits DNA synthesis by targeting ribonucleotide reductase, resulting in depletion of the dNTP pool and by competing with endogenous dCTPs for incorporation into DNA. A synergistic interaction between gemcitabine and MK-1775 was found in 44% (11 of 25) of the cell lines. As shown previously, TP53 wild-type cell lines displayed no evidence of substantial synergy (7), nor did the untransformed breast epithelial cell lines. There was a substantial variation in the degree of synergy in the TP53-mutant cell lines (Fig. 1A).

To examine the mechanisms underlying synergy, we initially chose to study 2 model cell lines: MCF7 (nonsynergistic, CI = 1.04; TP53 wild-type) and CAL120 (synergistic, CI = 0.61; TP53-mutant; Fig. 1B). MK-1775 treatment alone increased the proportion of cells in mitosis with a full 4N complement of DNA (CAL120 4N mitotic cells: 3.9%–27.8%, MCF7 4N mitotic cells: 1.9%–4.9%; Fig. 1C), most likely reflecting a subpopulation of G₂ phase cells undergoing appropriately scheduled mitosis. In response to gemcitabine, both cell lines arrested in early S-phase (Fig. 1C). Unexpectedly, the addition of MK-1775 to gemcitabine-arrested cells triggered a large proportion of CAL120 cells to engage mitosis without completing DNA synthesis (46.9% mitotic cells with <4N DNA content), an effect that was not seen in MCF7 cells (1.4% mitotic cells with <4N DNA content). Addition of MK-1775 triggered mitosis in CAL120 cells, but not in MCF7 cells, that were arrested in early S-phase by hydroxyurea treatment or by a double thymidine block (Fig. 1D), blocking DNA synthesis by depleting the dNTP pool through inhibition of ribonucleotide reductase, suggesting that mitotic entry was independent of the mechanism of inducing S-phase arrest. Mitosis without completing S-phase was also observed with other chemical inhibitors of WEE1 (PD0166285, WEE1 inhibitor II, and PHCD; Supplementary Fig. S1A) and siRNA-mediated silencing of WEE1 (Supplementary Fig. S1B), confirming the specificity of the results obtained with MK-1775.

We assessed mitotic entry in an expanded set of breast cancer cell lines (Fig. 1E; Supplementary Fig. S2A and S2B). Unscheduled mitosis, before completion of DNA synthesis, was seen in multiple TP53-mutant cell lines upon gemcitabine and MK-1775 treatment, but not in TP53 wild-type cell lines (Fig. 1E). Time course analysis revealed that unscheduled mitosis occurred earlier (after 6 hours; Supplementary Fig. S3A) and at lower doses of MK-1775 (300 nmol/L; Supplementary Fig. S3B) in the synergistic cell lines, whereas nonsynergistic cell lines responded only to 1 μmol/L MK-1775 at later time points (after 12 hours). This suggests that unscheduled entry into mitosis is the predominant mechanism underlying synergy between MK-1775 and gemcitabine, with the variation in synergy explained by the susceptibility to forced mitotic entry. There was no correlation between cell doubling time and synergy as assessed by CI (r = 0.15, P = 0.63; Pearson correlation coefficient).

Taken together, these results show that after stalling in S-phase, p53-deficient cell lines rely, in part, on WEE1 activity to delay mitosis until DNA replication is complete. Inhibition of WEE1 by MK-1775 in combination with DNA-damaging agents overrides the induced early S-phase arrest and forces cells directly into mitosis without completing DNA synthesis.

We examined nuclear morphology in response to gemcitabine, MK-1775, and the combination of both drugs. After gemcitabine treatment, only interphase nuclei with uncondensed chromatin were observed, whereas cells with condensed chromosomes typically aligned at the metaphase plate were observed after MK-1775 treatment (Fig. 2A). In contrast, gemcitabine and MK-1775 treatment resulted in highly abnormal mitoses with unaligned and partially condensed chromosomes (Fig. 2A and B), showing characteristics of S-phase premature chromosome condensation (PCC; ref. 16). To assess this further, we examined microtubule organization by labeling cells treated with hydroxyurea and MK-1775 for α-tubulin and phospho-histone H3 (pH3; Fig. 2C). After hydroxyurea and MK-1775 combination treatment, 65% of the pH3-positive cells showed highly disorganized bipolar spindles with dispersed chromosomes and little evidence of alignment to the metaphase plate, a phenotype that was only rarely seen in control cells or with MK-1175 alone (Fig. 2C and D). The occurrence of mono- or multipolar spindles was not affected by WEE1 inhibition. A small number of interphase nuclei in control cells stained positive for focal pH3, indicative of pericentromeric H3 phosphorylation in cells with incipient mitotic entry (17).

To assess the consequences of unscheduled mitosis, CAL120 cells were arrested in early S-phase by hydroxyurea, followed by addition of MK-1775 for 8 hours or continued hydroxyurea treatment alone. Treatments were washed out and cell-cycle progression was monitored by flow cytometry. Hydroxyurea-arrested control cells readily reentered the cell cycle and reached late S–G₂ phase after 8 hours, went through mitosis at 16 hours after release, and subsequently progressed into the next cell cycle (Fig. 3A; Supplementary Table S2). In contrast, cells treated with hydroxyurea and MK-1775 progressively exited mitosis, 8 hours after release the percentage of pH3-positive cells had dropped from 60.1% to 8.5%, but did not progress into S- or G₂ phase for up to 48 hours after release (Fig. 3A and data not shown).

Normally, entry into mitosis is triggered by progressive activation of cyclin B1-CDK1 at the end of G₂ phase (3). Cyclin B1-CDK1 activity remains high until anaphase onset, when cyclin B1 is targeted for degradation by anaphase-promoting complex or cyclosome (APC/C) after the spindle assembly checkpoint (SAC) is satisfied by proper attachment of all chromosomes to the mitotic spindle. Cyclin B1 degradation results in inactivation of CDK1, which is required for anaphase progression and exit from mitosis. To gain more insight into the kinetics of unscheduled entry and exit from mitosis, we conducted a detailed time course analysis (Fig. 3B and C). After MK-1775 addition, cells gradually entered mitosis during the first 6 hours and subsequently exited mitosis many hours later, regardless of whether WEE1 inhibition was maintained (Fig. 3B). As the majority of cells exited mitosis, 10 to 14 hours after MK-1775 addition (Fig. 3B), cyclin B1 levels decreased and CDK1-Y15 phosphorylation partially increased (Fig. 3C), suggesting that after a prolonged mitotic arrest, loss of cyclin B1-CDK1 activity allowed cells to exit mitosis without undergoing complete chromosome segregation and cell division, a process termed “mitotic slippage” (18).

The nuclear morphology of cells exiting unscheduled mitosis was highly abnormal with gross micronuclei formation (Fig. 3D and E). Accompanying mitotic exit, PARP-1 was cleaved, indicating engagement of apoptosis (Fig. 3C). Substantial induction of apoptosis was confirmed with both Annexin V staining (Supplementary Fig. S4A and S4B) and caspase-3/7 activation (Supplementary Fig. S4C). This suggested that on mitotic exit, the dispersed chromosomes were unable to assemble into a single nucleus, resulting in apoptosis or long-term cell-cycle arrest.

The intra–S-phase checkpoint is predominantly regulated by the checkpoint kinases ATR and CHK1 (4). In response to replicative stress, CHK1 is activated by ATR through phosphorylation at Ser317 and Ser345 (19). CHK1 inhibits the CDC25 phosphatases, keeping CDK1 in its inactive Y15-phosphorylated state (20). Inhibition of WEE1 per se is insufficient to result in CDK1-Y15 dephosphorylation, if CDC25 phosphatases are also inactivated by CHK1. Consistent with this, we observed only partial loss of CDK1-Y15 phosphorylation with WEE1 kinase inhibition alone (Fig. 3C).

To further investigate the role of CHK1 in preventing unscheduled mitotic entry induced by WEE1 inhibition, we treated CAL120 and MCF7 cells with gemcitabine for 24 hours followed by CHK1 inhibitor AZD7762 (21) alone or in combination with MK-1775. Inhibition of CHK1 in gemcitabine-arrested cells induced premature mitosis in 33% of the CAL120 cells and only 2% in the MCF7 cells (Fig. 4A). Time course analysis revealed that within 1 hour after addition of both inhibitors, 52% of the cells had entered mitosis, compared with 4% and 14% treated with AZD7762 or MK-1775 alone, respectively (Fig. 4B).

We extended these observations by showing that the combination of AZD7762 and MK-1775, in the absence of DNA-damaging agents, was sufficient to drive late S-phase cells into mitosis (12% mitotic CAL120 cells with <4N DNA content; Fig. 4C and D). No induction of mitosis was seen with the combination in MCF7 cells (Fig. 4C). AZD7762 increased the sensitivity of CAL120 cells to MK-1775 alone, shifting the MK-1775 SF₅₀ value from 329 to 75 nmol/L (Supplementary Fig. S5A). We explored the molecular basis for the synergistic interaction between CHK1 and WEE1 inhibition in CAL120 cells (Fig. 4E; Supplementary Fig. S5B). Combined inhibition of CHK1 and WEE1 led to a greater decrease in CDK1-Y15 phosphorylation than WEE1 inhibitor alone (Fig. 4E), consistent with the rapid engagement of mitosis with dual inhibition (Fig. 4B).

To examine the possibility that WEE1 inhibition interfered with induction and maintenance of the intra–S-phase checkpoint, CAL120 cells were pulse-labeled with bromodeoxyuridine (BrdU) followed by treatment with the topoisomerase I poison camptothecin (Fig. 4F). The recovery of DNA replication in S-phase cells was determined by monitoring cell-cycle progression of the BrdU-labeled cells. Camptothecin induced a substantial S-phase delay due to engagement of the intra–S-phase checkpoint [Fig. 4F (iii)]. AZD7762 abrogated the camptothecin-induced S-phase delay, causing BrdU-labeled cells to progress into G₂ phase at 8 hours after camptothecin [Fig. 4F (v)]. In contrast, MK-1775–treated BrdU-labeled cells showed a similar delay in S-phase progression as controls [Fig. 4F, (iv) vs. (iii), respectively], suggesting that WEE1 is not required for maintenance of the intra–S-phase checkpoint.

We investigated the molecular changes that predispose to unscheduled mitotic entry. Unscheduled mitotic entry after gemcitabine and MK-1775 combination treatment was blocked by cotreatment with the selective CDK1 inhibitor RO-3306 (Fig. 5A). Silencing of CDK1 with siRNA, but not CDK2, blocked unscheduled mitosis (Fig. 5B; Supplementary Fig. S6A), confirming that CDK1, not traditional S-phase CDKs, triggered entry into mitosis.

In TP53 wild-type MCF7 cells, silencing of both p53 and p21^Waf1/Cip1 predisposed to forced mitotic entry by WEE1 inhibition (Fig. 5C; Supplementary Fig. S6B). This suggested that in TP53 wild-type cells, both p53 and p21^Waf1/Cip1 prevent activation of CDK1 in S-phase. We examined changes in expression of mitotic proteins in MCF7 and CAL120 cells in response to gemcitabine, MK-1775, or the combination (Fig. 5D). The expression of mitotic cyclins in normal cells starts at the beginning of S-phase and peaks during G₂ phase in preparation for mitotic entry. In response to DNA damage, expression of the mitotic genes CCNB1 and CDC2 is suppressed in a process mediated by p53 and p21^Waf1/Cip1 (22). After gemcitabine induced early S-phase arrest (Fig. 1C), TP53 wild-type MCF7 cells expressed relatively low levels of mitotic proteins, whereas TP53-mutant CAL120 cells expressed high levels of mitotic proteins and CDK1-Y15 phosphorylation (Fig. 5D). In CAL120 cells, the expression of CDK1 varied with different cell-cycle phases, in a similar manner to cyclin B1 expression (Supplementary Fig. S7A).

We investigated the factors that promote substantial unscheduled mitosis in some TP53-mutant cell lines, but not in others (Fig. 1E). We have previously published whole-genome gene expression profiles for 20 TP53-mutant breast cancer cell lines used in this study (23). The gene expression levels for each individual probe were correlated with gemcitabine/MK-1775 CI using Pearson correlation coefficient (Fig. 5E). We hypothesized that genes correlating positively with the CI (correlating with lack of synergy) may be involved in resistance to unscheduled mitotic entry, and conversely genes with a negative correlation (correlating with synergy) may promote unscheduled mitotic entry. Because of the small number of cell lines examined relative to the large number of gene expression probes, this analysis would have a high false discovery rate but could also highlight genes and processes that would be validated by subsequent functional experiments. High expression of cyclin D1 was correlated with lack of synergy (r = 0.46,P = 0.041; Pearson correlation coefficient), whereas high expression of cyclin E1 (r = −0.46, P = 0.040) and B-type cyclins (r = −0.38, P = 0.09) was correlated with synergy.

To confirm the observation that high level of mitotic genes may predispose to mitotic entry, we assessed cyclin B1 and CDK1 expression levels in a larger panel of cell lines (Fig. 5F; Supplementary Fig. S7B). Cyclin B1 protein level correlated with CCNB1 mRNA gene expression data (r = 0.7), supporting the gene expression analysis and suggesting that cyclin B1 protein levels were regulated by transcription. Although gemcitabine induced an early S-phase arrest (Supplementary Fig. S2B), a robust induction of cyclin B1 and CDK1 expression was observed, particularly in the cell lines that showed high levels of premature mitosis. In contrast, TP53 wild-type cell lines MCF7 and ZR75.1 did not show a strong induction of cyclin B1 and CDK1, confirming that p53 is involved in suppressing mitotic gene expression upon DNA damage (22). CHK1 was activated (phospho-Ser317) in response to gemcitabine in all TP53-mutant cell lines but neither of the TP53 wild-type lines.

Taken together, these data suggest that the switch from expression of G₁–S cyclins to late S–G₂–M cyclins, which is known to occur in poor prognosis breast cancers (24), is associated with synergy, and following S-phase arrest, TP53-mutant cells become reliant on CHK1–WEE1 function to prevent mitotic entry.

We noted that in TP53-mutant cell lines, high levels of EZH2 expression correlated with synergy (r = −0.58, P = 0.009; Fig. 6A), and conversely high levels of CDKN1A (p21^Waf1/Cip1) expression correlated with lack of synergy (r = 0.81, P < 0.0001; Fig. 6B). We silenced these genes to examine whether they contributed to the forced mitotic phenotype in TP53-mutant cells. Silencing of EZH2 with 2 independent siRNAs reduced unscheduled mitosis in CAL120 cells (Fig. 6C; Supplementary Fig. S8A). In contrast, silencing of p21^Waf1/Cip1 did not induce unscheduled mitotic entry in TP53-mutant SUM44 and JIMT-1 cells (Fig. 6D; Supplementary Fig. S8B), suggesting that residual p21^Waf1/Cip1 in TP53-mutant cell lines was not responsible for preventing premature mitotic entry.

It has previously been shown that EZH2 negatively regulates p21^Waf1/Cip1 expression (25), and EZH2 gene expression was negatively correlated with CDKN1A expression in a publicly available data set of TP53-mutant breast cancers (Fig. 6E). This suggested that the observed correlation between CDKN1A and CI (Fig. 6B) may be reflective of the negative regulation of CDKN1A expression by EZH2 in TP53-mutant cell lines. High expression of the polycomb protein EZH2 is associated with poor prognosis in breast cancer (26) and has been shown to promote mitotic entry (27). Expression of CCNB1 and EZH2 is positively correlated in breast cancers (r = 0.51, P < 0.0001; Fig. 6F). Combined assessment of CCNB1 and EZH2 expression was particularly robust in identifying a poor prognostic group (Fig. 6G), suggesting that high expression of both EZH2 and mitotic cyclins may promote an aggressive breast cancer phenotype while providing a permissive environment for unscheduled mitosis that could potentially be targeted therapeutically by WEE1 inhibitor combinations. Interestingly, basal breast cancer cell lines showed a greater degree of synergy between MK-1775 and gemcitabine than luminal cell lines (Fig. 6H).

We examined for potential efficacy of MK-1775 and gemcitabine combinations in vivo and examined for evidence of forced mitosis. Because we were unable to robustly xenograft the synergistic breast cancer cell line CAL120, we used the WiDr colon cancer model (7, 8), which showed a substantial degree of forced mitosis in vitro (Fig. 7A; Supplementary Fig. S9A).

We treated WiDr xenografts with gemcitabine followed 24 hours later by MK-1775. Within 8 hours of MK-1775 treatment, a substantial increase in pH3-positive cells was seen, with H3 phosphorylation in interphase nuclei indicative of incipient mitotic entry (Fig. 7B and C). Treatment of xenografts with gemcitabine induced an early S-phase arrest (Supplementary Fig. S9B), suggesting that the pH3 induction occurred in early S-phase cells. For efficacy experiments, we used the WiDr derivative HT-29 that grows robustly in nude mice. In this model, prolonged MK-1775 exposure induced mitoses characteristic of forced mitosis following combination treatment (Fig. 7D and E), and the combination was substantially more effective at reducing tumor growth than either treatment given alone (Fig. 7F).

In this study, we show that forced mitotic entry of S-phase arrested cancer cells provides a potential novel therapeutic approach for aggressive breast cancers. In cell lines that showed synergy between WEE1 kinase inhibitor MK-1775 and chemotherapy, MK-1775 triggered mitotic entry in S-phase–arrested cells. Forced mitotic entry before completion of DNA synthesis triggered a highly abnormal mitosis characterized by dispersed chromosomes and highly disorganized bipolar spindles. Subsequently, cells “slipped” through the forced mitosis, with mitotic exit resulting in gross micronuclei formation and apoptosis.

Prior investigations of the combination of MK-1775 and chemotherapy have focused on disruption of the classical G₂–M checkpoint as a mechanism by which WEE1 inhibition sensitizes to DNA-damaging agents (6–13, 28). However, we show here that p53-deficient breast cancer cells engage the intra–S-phase checkpoint and arrest in early S-phase in response to gemcitabine (Fig. 1C; Supplementary Fig. S2B). Therefore, understanding the role of WEE1 during S-phase rather than the classical G₂–M checkpoint is key to clarifying the mechanism of sensitivity to WEE1 inhibitors. Multiple p53 and p21^Waf1/Cip1 functions protect against unscheduled mitosis (Fig. 7G). p53 and p21^Waf1/Cip1 inhibit mitotic gene transcription in response to DNA damage switching the LINC complex from active to repressive states (22, 29) and p21^Waf1/Cip1 directly inhibits CDK1 activity (30–32). In contrast, in TP53-mutant cancer cells, loss of p53 and p21^Waf1/Cip1 function makes cells more reliant on the CHK1-dependent intra–S-phase checkpoint in response to DNA damage (CHK1 pS317; Fig. 5F; Supplementary Fig. S7B), while at the same time, mitotic gene expression is deregulated and WEE1 becomes critical to restrain CDK1 activity.

This model suggests that the primary role of chemotherapy in combination with WEE1 inhibition is to arrest cells in S-phase to allow accumulation of mitotic gene expression to levels that are sufficient to trigger mitosis. WEE1 inhibition alone appears insufficient to trigger mitosis in the absence of S-phase perturbation, potentially as a critical level of CDK1 activity is not achieved. The combination of WEE1 and CHK1 inhibition was sufficient to force mitosis during unperturbed late S-phase (Fig. 4C and D), potentially as CDK1 activity reaches the critical level required to trigger mitosis. However, the combination of both inhibitors may have additional effects other than inducing premature mitosis (33), and potential toxicity of the combination would need careful assessment before clinical development as the combination modestly sensitizes the breast epithelial cell line MCF10A (Supplementary Fig. S5A). Our data emphasize the potential benefit of strategies to enhance forced mitotic entry. Only a restricted set of TP53-mutant cell lines showed evidence of synergy between gemcitabine and WEE1 inhibition in vitro (Fig. 1A). Yet, forced mitotic entry could be triggered in the nonsynergistic TP53-mutant cell lines SUM44 and MFM223 (Supplementary Fig. S3) but only at levels insufficient to result in synergy, which suggests that there may be potential to exploit this mechanism of cell death in all TP53-mutant cancer cell lines.

During unperturbed cell-cycle progression, CHK1 is localized at the centrosomes where it inhibits cyclin B-CDK1 activity during S-phase via phosphorylation of the CDC25 family phosphatases (34, 35). Chemical inhibition of CHK1 induces premature centrosome separation and activation of centrosome-associated CDK1 (34). Because centrosome separation and positioning are crucial for correct formation of a bipolar spindle (reviewed in ref. 36), premature activation of centrosomal CDK1 could be the underlying mechanism for the disorganized mitotic spindle phenotype observed after WEE1 inhibition (Fig. 2C).

Recent evidence has indicated that WEE1 silencing with siRNA results in stalled and collapsed replication forks in S-phase cells (37, 38), potentially resulting from disruption of the interaction between WEE1 and the Mus81-Eme1 endonuclease (39). However, S-phase delay was not observed with WEE1 inhibition by MK-1775 (Figs. 1C and 4F; Supplementary Fig. S2B). Potentially, these differences simply reflect analysis at different time points and conditions or the genetic background of the cell lines examined, although it is possible that inhibition of the kinase with a small molecule and silencing with siRNA may have different phenotypes.

The oncogenic transformation of aggressive breast cancers results in a permissive environment for unscheduled mitotic entry, characterized by TP53 mutation and high levels of cyclin B1 and EZH2 (Fig. 7G; refs. 24, 40–43). A current major focus of breast cancer research is triple-negative and basal-like breast cancers because of the lack of effective therapies. This subtype has the highest frequency of TP53 mutation, as well as generally high expression of mitotic B-type cyclins and EZH2 (24, 40–43). In our study, basal breast cancer cell lines show a greater degree of synergy between MK-1775 and gemcitabine than luminal cell lines (Fig. 6H). Although it has been reported that WEE1 inhibitors may target triple-negative breast cancer cell lines as single agents (44), basal cancer cell lines were not more sensitive to single-agent MK-1775 than luminal cell lines (median MK-1775 SF₅₀ basal 310 nmol/L vs. luminal 320 nmol/L, P = 0.5; Mann–Whitney U test, data not shown). Therefore, we believe that WEE1 inhibition is a potential therapeutic approach for basal-like breast cancers but likely only in combination with chemotherapy or CHK1 inhibition.

Our data emphasize the importance of biomarkers in the clinical development of WEE1 inhibitors, as only a restricted proportion of cancer cell lines show synergy (Fig. 1A). WEE1 inhibitors are in early clinical development and we look forward to assessing the expression of CCNB1 and EZH2 in these studies, although further research will be required to ascertain whether similar predictors apply to cancer types other than breast cancer. High expression of the mitotic cyclins, and potentially EZH2, may be a downstream consequence of multiple different oncogenic drivers in aggressive breast cancer. Therefore, forced mitotic entry with WEE1 inhibitors represents a novel therapeutic approach that potentially targets the consequences of oncogenic transformation without requiring a direct understanding of the individual oncogenic mutations present.

Cell lines were obtained from American Type Culture Collection or Asterand (Supplementary Table S1) and maintained in phenol red–free Dulbecco's Modified Eagles’ Media (DMEM) or RPMI with 10% FBS (FBS Gold, PAA Laboratories) and 2 mmol/L l-glutamine (Sigma-Aldrich). All cell lines were banked in multiple aliquots on receipt to reduce risk of phenotypic drift and identity confirmed by short tandem repeat profiling with the StemElite ID System (Promega) and array comparative genetic hybridization profiling. Cell lines were assigned to luminal and basal subtypes as defined by Chin and colleagues (45).

For synergy analysis between gemcitabine and MK-1775, cells were plated in 96-well plates at 2,000 to 5,000 cells per well. The following day, cells were exposed to different fixed-ratio combinations of gemcitabine (dose range, 1.56–100 nmol/L) and MK-1775 (dose range, 31.3 nmol/L–2 μmol/L), and survival was assessed after 72-hour exposure with the Cell Titer-Glo Luminescent Cell Viability Assay (Promega). The CI was determined according to the median-effect model of Chou–Talalay (15). Although a CI < 1.0 can be inferred as synergy, we chose a cutoff point of <0.75 to more stringently define synergy.

For siRNA transfections, cells were reverse-transfected in 6-well plates with 50 nmol/L siRNA using Lipofectamine RNAiMAX (Invitrogen) or DharmaFECT3 (Dharmacon/Thermo Scientific) according to the manufacturers’ instructions. Drug treatments were started 48 hours after transfection and protein extracts prepared 72 hours after transfection. siRNAs were obtained from Dharmacon: WEE1 siGENOME SMARTpool (siWEE1, M-005050-02), CDC2 siGENOME SMARTpool (siCDK1, M-003224-03) and individual siRNA duplexes (D-003224-09/10/11/12), CDK2 siGENOME SMARTpool (siCDK2, M-003236-04) and individual siRNA duplexes (D-003236-06/08/09/10), CDKN1A siGENOME SMARTpool (sip21, M-003471-00) and individual siRNA duplexes (D-003471-01/03), TP53 siGENOME SMARTpool (siTP53, M-003329-03) and individual siRNA duplexes (D-003329-05/26), and EZH2 siGENOME SMARTpool (siEZH2, M-004218-03) and individual siRNA duplexes (D-004218-01/02), and siGENOME Non-Targeting siRNA #1 (siCON, D-001210-01).

The following chemical inhibitors were used at the indicated concentrations, unless stated otherwise, WEE1 inhibitor MK-1775 (1 μmol/L; Merck), CHK1 inhibitor AZD7762 (100 nmol/L; Selleck Chem), CDK1 inhibitor RO-3306 (10 μmol/L; Tocris), gemcitabine (100 nmol/L), camptothecin (1 μmol/L), hydroxyurea (3 mmol/L), thymidine (2.5 mmol/L), and nocodazole (200 ng/mL; all from Sigma-Aldrich).

Antibodies used were WEE1 (sc-5285), PARP-1 (sc-8007), CDK2 (sc-163), and actin (sc-1616; all from Santa Cruz Biotechnology); CHK1-Ser317 (A300-163A; Bethyl Laboratories); phospho-Histone H3-Ser10 (pH3, 06-570), phospho-Histone-H2AX-Ser139 (05-636), and EZH2 (05-1319; all from Upstate); α-tubulin (T4026; Sigma-Aldrich); β-Tubulin (T4026; Sigma-Aldrich); cyclin B1 (4135), phospho-CDK1-Tyr15 (9111), CDK1 (9112), phospho-CDC25C-Ser216 (4901), CDC25C (4688), and p21^Waf1/Cip1 (2947; all from Cell Signalling Technology).

Cells were seeded in 60-mm dishes (2 × 10⁵ cells per dish), treated for 24 hours after plating with 100 nmol/L gemcitabine for 24 hours, after which 1 μmol/L MK-1775 or 100 nmol/L AZD7762 was added to gemcitabine for another 8 hours. Cells were fixed in ice-cold 70% ethanol, permeabilized with 0.25% Triton X-100 in PBS, incubated with 1 μg/mL anti-pH3 antibody for 2 hours at 4°C, followed by a secondary antibody conjugated to AlexaFluor 488 for 1 hour at 4°C. DNA was stained with 20 μg/mL propidium iodide (PI) in the presence of 100 μg/mL RNase A.

For intra–S-phase checkpoint assessment, CAL120 cells were incubated with 10 μmol/L BrdU (Sigma-Aldrich) for 30 minutes, washed, and then treated with 1 μmol/L camptothecin for 30 minutes, washed, and grown in drug-free medium for 2 hours, after which MK-1775 or AZD7762 were added for another 22 hours. Ethanol-fixed cells were incubated in pepsin (0.4 mg/mL in 0.1N HCl) for 30 minutes at room temperature. DNA was denatured using 2N HCl and neutralized with 0.1 mol/L sodium borate. Cells were labeled with fluorescein isothiocyanate–conjugated anti-BrdU antibody (347583; Becton Dickinson) and DNA was stained with PI. Samples were analyzed on a BD LSR II flow cytometer using BD FACSDiva software (BD Biosciences).

Cells were plated on 100-mm dishes, treated as indicated, and whole-cell extracts were prepared using NP-40 lysis buffer [10 mmol/L Tris-HCl, pH 8.0, 150 mmol/L NaCl, 1 mmol/L EDTA, pH 8.0, 1% NP-40, 5 mmol/L sodium pyrophosphate, 50 mmol/L NaF, 1 mmol/L Na₃VO₄, 5 mmol/L dithiothreitol (DTT), complete protease inhibitor cocktail (Roche)]. Lysates were run on precast 4% to 12% Bis–Tris gels (Invitrogen) and transferred onto nitrocellulose membranes (Bio-Rad). Signals were visualized using enhanced chemiluminescence (Amersham ECL Western Blotting Detection Reagents, GE Healthcare).

CAL120 cells were plated on coverslips and treated with chemical inhibitors the following day as described above. After drug treatment, cells were fixed in 4% paraformaldehyde for 1 hour, washed, permeabilized in 0.2% Triton X-100 in PBS for 20 minutes, washed, and blocked in IFF [1% bovine serum albumin, 2% FBS in PBS] followed by incubation with primary and secondary AlexaFluor-conjugated antibodies for 1 to 2 hours at room temperature each. DNA was stained with 4′,6-diamidino-2-phenylindole (DAPI). Coverslips were mounted on glass slides using Vectashield. Images were acquired on a Leica confocal microscope (×63 oil immersion objective). Mitotic and nuclear phenotypes of at least 200 cells per condition were scored for 2 independent replicate experiments.

For preparation of chromosome spreads, CAL120 cells were treated with KaryoMAX Colcemid solution (20 ng/mL; Gibco-Invitrogen) for 2 hours before harvesting. Cells were treated with 30 mmol/L NaCitrate for 15 minutes at 37°C, fixed in methanol:acetic acid (3:1), and dropped onto glass slides. Chromosomes were stained with DAPI.

All animal experiments were conducted in accordance with local good animal practice. For the WiDr xenograft experiments, nonobese diabetic/severe combined immunodeficient female mice ages 5 to 6 weeks were obtained from Charles River Laboratories. Mice were inoculated with 5 × 10⁶ WiDr cells in 100 μL (1:1 PBS:growth factor–reduced Matrigel; BD Biosciences) subcutaneously (s.c.) into both flanks. Gemcitabine was administered intraperitoneally (i.p.) at 100 mg/kg (in PBS), and 24 hours later, MK-1775 was administered at a dose of 30 mg/kg (in 0.5% methylcellulose) by oral gavage (p.o.). Tumors were taken 8 hours after MK-1775 administration for fluorescence-activated cell-sorting (FACS) analysis and pH3 immunofluorescence. For efficacy experiments, we used the WiDr derivative HT-29 that grows robustly in nude mice. CD-1 nu/nu female mice ages 4 to 6 weeks were obtained from Charles River Laboratories. Mice were inoculated with 2.5 × 10⁶ to 5 × 10⁶ WiDr (HT-29) cells in 100 μL (1:1 Matrigel: PBS) s.c. into the right flank. When tumor volume reached 325 mm³ (±70), mice were randomized to treatment groups. For efficacy analysis, gemcitabine was administered i.v. at 100 mg/kg once per week, and 24 hours later, MK-1775 was administered at 15 or 30 mg/kg p.o. twice daily (3 hours). Tumor measurements were recorded biweekly and tumor volume was calculated as (width × length ×length)/2. For pH3 immunofluorescent analysis, gemcitabine treatment was followed by MK-1775 for 2.5 days (5 doses), with tumors collected 2 hours after the final dose, fixed in 4% paraformaldehyde, paraffin-embedded, and sectioned.

All statistical tests were conducted with GraphPad Prism version 5.0. Unless stated otherwise, P values were 2-tailed and considered significant if P < 0.05. Error bars represent SEM of 3 experiments unless stated otherwise.

Cell line whole-genome gene expression data were assessed on Illumina human-ref6 v2 BeadChips as previously described (23). Gene expression, from probes detectable in more than 90% of cell lines, was correlated with gemcitabine/MK-1775 CI using Pearson correlation coefficient. Gene expression data on a publicly available data set of TP53 wild-type and mutant cell lines was analyzed from Miller and colleagues (46). Kaplan–Meier survival curves were plotted for EZH2, CCNB1, and mean of EZH2 and CCNB1, in 2 groups, gene expression < median versus ≥ median.

M.S. Hurd and S.D. Shumway are employees of Merck Sharp & Dohme Corp. C. Toniatti was previously an employee of Merck Sharp & Dohme Corp. No potential conflicts of interests were disclosed by other authors.

The authors thank Asha Kigozi and Antoinette van Weverwijk for their assistance with the mouse xenograft experiments and the Breakthrough Centre Histopathology Core Facility for processing the xenograft tumors.

This work was supported by Cancer Research UK grant A10038 and Susan G. Komen for the Cure (to M. Aarts, R. Sharpe, I. Garcia-Murillas, A. Ashworth, N.C. Turner) and the Netherlands Organisation for Scientific Research (NWO; M. Aarts). The authors acknowledge NHS funding to the NIHR Biomedical Research Centre.