Background
DNA damage activates cell cycle checkpoints that arrest cell cycle progression and thereby provide time for repair and recovery. This has led to the development of checkpoint inhibitors as adjuvants to DNA damaging agents with the anticipation that they will enhance therapeutic activity. Chk1 is the primary checkpoint protein against which many small molecule inhibitors have been developed [
1‐
3]. Chk1 is activated when the kinases ATM and/or ATR detect double-strand breaks or large single-strand regions of DNA, respectively. Once activated, Chk1 phosphorylates and inactivates CDC25 phosphatases that are required for CDK activation and cell cycle progression. Inhibition of Chk1 results in premature activation of CDC25 phosphatases and CDK1/2, and progression through the cell cycle before adequate repair has occurred. Increased DNA damage occurs as cells progress through S phase with a damaged template, followed by lethal mitosis once they have reached the G2 phase [
4].
Antimetabolites such as gemcitabine and hydroxyurea inhibit ribonucleotide reductase, thereby rapidly depleting deoxyribonucleotide pools and stalling replication fork progression. These agents do not directly induce DNA breaks, and arrest occurs without the need for Chk1 activation. However, Chk1 stabilizes the stalled replication forks and, when inhibited, the replication forks collapse thus producing DNA double-strand breaks [
5]. Hence, there is a significant difference in the outcome of Chk1 inhibition depending on the type of DNA damage that occurs; in the latter case, new lethal events occur where no DNA damage existed previously. Consequently, we have found that Chk1 inhibition can induce a far more dramatic sensitization to antimetabolites that induce this replication arrest compared to other DNA damaging agents that activate Chk1 through the DNA damage-induced checkpoint [
6].
Gemcitabine is a deoxynucleoside analogue that is metabolized to a deoxynucleotide triphosphate, a precursor for incorporation into DNA, and to a deoxynucleotide diphosphate that irreversibly inhibits ribonucleotide reductase. As a consequence, low concentrations of gemcitabine rapidly deplete deoxyribonucleotide pools, inhibit DNA synthesis and induce a long S phase arrest. Here we focus on the combination of gemcitabine with the Chk1 inhibitor MK-8776 [
7]. We report the efficacy of this combination in cell lines from many different cancers. We also report that the time of addition of MK-8776 can significantly impact the response of tumor cells to gemcitabine both
in vitro and in xenograft tumor models. The schedule dependence is critical because of the relatively short half-life of MK-8776 in patients’ plasma [
8]. These results have important implications for the design of clinical trials of this combination.
Discussion
Chk1 participates in multiple functions in a cell [
3]. It was originally recognized as a mediator of the DNA damage response, preventing cell cycle progression so that cells could repair DNA damage. The underlying mechanism involves Chk1-mediated inhibition of CDC25, thereby preventing activation of CDK1 and 2. Inhibition of Chk1 leads to activation of CDK1/2, cell cycle progression and aberrant mitosis. Recently, it has been recognized that some cell lines are hypersensitive to brief inhibition of Chk1 alone, with γH2AX foci and/or DNA double-strand breaks appearing within 6 h [
6,
14,
20]. This damage occurs only in S phase cells and is also mediated by activation of CDK2. In addition, Chk1 is now recognized as having additional roles in replication fork stability, replication origin firing and homologous recombination, and it is the latter of these roles that appears important for the efficacy of the combination of gemcitabine with MK-8776. Mechanistically, homologous recombination results when Chk1 phosphorylates the C-terminal domain of BRCA2 which then interacts with and recruits RAD51 to single-stranded DNA. In addition Chk1 can directly phosphorylate RAD51 and this is also required for recruitment of RAD51 to single-stranded DNA [
17,
21]. Our results demonstrate that inhibition of Chk1 can also result in dissociation of RAD51 from DNA which we suggest is due to the dynamic status of regressed replication forks which likely shorten or grow in length continuously and thereby displace RAD51.
These different functions of Chk1 can explain why Chk1 inhibitors exhibit variable efficacy in sensitizing cells to DNA damaging agents. Our previous experiments involved incubation of cells with the topoisomerase I inhibitor SN38 [
4,
6]. Replication forks collide with the inhibited topoisomerase complex creating DNA breaks that rapidly activate Chk1 and prevent cell cycle progression. Yet while inhibition of Chk1 induced cell cycle progression, it had little impact on overall cytotoxicity because lethal breaks were already induced by SN38 alone. In contrast, when gemcitabine or hydroxyurea inhibit ribonucleotide reductase, replication stalls rapidly and independently of Chk1. Indeed, we previously demonstrated that hydroxyurea can arrest DNA replication without activating Chk1, and this observation is reiterated here at low concentrations of gemcitabine [
6]. Upon removal of gemcitabine, these arrested cells are able to recover. However, inhibition of Chk1 rapidly induces collapse of replication forks, and this is new DNA damage that dramatically enhances cell killing. Other investigators have observed activation of Chk1 upon incubation with either hydroxyurea or gemcitabine, but in general those experiments involved higher concentrations of each drug that exceed those needed to arrest the cells [
22‐
24]. We have observed slight activation of Chk1 when western blots are over-exposed, but this level of phosphorylation is far lower than observed after replication forks have collapsed as a consequence of Chk1 inhibition. Similar observations were made in a study of gemcitabine alone which showed phosphorylation of Chk1, but a subsequent paper also showed this to be negligible compared to that induced by concurrent inhibition of Chk1 [
25,
26]. In the case of cells incubated with gemcitabine alone, we question whether the low level activation of Chk1 is due to incorporation of gemcitabine into DNA and the chain termination that then occurs rather than to the inhibition of ribonucleotide reductase.
Here, we show that MK-8776 markedly sensitizes multiple cell lines to gemcitabine. In further dissecting the mechanism, we noted that γH2AX did not appear until about 16 h of co-treatment. We therefore delayed the addition of MK-8776 and demonstrated that, when added for the final 4 h of a 24-h incubation of gemcitabine, it induced as much γH2AX signal as it did when incubated concurrently with gemcitabine for the entire 24 h. Our results demonstrate that stalled replication forks evolve with time to become more Chk1 dependent, and this correlates with a delay in loading of Rad51 onto DNA. When Chk1 was inhibited, these Rad51 foci disappeared and very strong γH2AX signal was observed. Evolution of stalled replication forks and delayed appearance of RAD51 foci have previously been observed during incubation with hydroxyurea, but it was concluded that RAD51-dependent recombination occurred in response to collapsed replication forks [
27]. Here we observed very few γH2AX-positive foci prior to recombination, but a dramatic increase once RAD51 loading was prevented by inhibiting Chk1. This implies that the appearance of γH2AX is a consequence of inhibiting recombination and not the stimulus for recombination. That inhibition of recombination is important for the observed sensitization is also suggested by the TK10 cells which were sensitive to gemcitabine alone, and were not further sensitized by MK-8776. This cell line has been reported to have a defect in recombination which would explain this observation [
28].
The requirement for only a brief incubation with MK-8776 to enhance cytotoxicity is an important observation given that, in clinical trials, the plasma concentration of MK-8776 was shown to exceed 1 μmol/L for only about 6 h [
8]. MK-8776 dissociates rapidly from Chk1 when the drug is removed (data not shown), so it is unlikely that Chk1 will remain inhibited significantly beyond 6 h.
We extended these experiments to more closely reflect the clinical situation by incubating cells briefly with gemcitabine, and then permitting the cells to recover. Because ribonucleotide reductase remains inhibited for a long time, it took several days for the cells to recover; the rate of recovery depended on the concentration of gemcitabine. Cells in G
1 also progressed into S phase during this time, so the number of cells potentially susceptible to Chk1 inhibition continued to increase. Hence there are two reasons why delayed addition of MK-8776 can enhance sensitivity to gemcitabine: first, there is an increased number of cells arrested in S phase, and second, the arrested cells have been given adequate time to become Chk1 dependent (i.e., to initiate recombination). The current experiments indicated that addition of MK-8776 at 18 h provided the greatest decrease in IC50 for gemcitabine in four cell lines (Figure
4). However, these experiments only reflect growth inhibition, and the S phase arrest at these low concentrations was very transient. Higher concentrations of gemcitabine induce a longer arrest with more cells accumulating in S phase. Consequently, it is possible that later addition of MK-8776 may have improved cell killing as the cells newly arrested in S phase at 18 h may not yet have become Chk1 dependent.
To more directly assess the relevance of these in vitro observations, we assessed the S/G2 phase arrest that occurred in two different tumor models in vivo. This was quantified as the ratio of geminin-positive to Ki67-positive cells. Eighteen hours after administration of 150 mg/kg gemcitabine, there was a marked increase in geminin-positive cells suggesting that up to 83-95% of the Ki67-positive cells were in S or G2 phase. By 42 h, this percentage had partially reverted to the starting value reflecting recovery of the cells. This dose of gemcitabine is considered equivalent to a dose of 450 mg/m2 in patients, which is about half the standard dose administered (1000 mg/m2). We are currently performing a clinical trial to assess the S/G2 phase arrest that occurs in patients receiving gemcitabine as a guide for subsequent administration of a Chk1 inhibitor.
Finally, we assessed the impact of schedule on the response of human tumor xenografts to the combination of gemcitabine and MK-8776. The results clearly demonstrated that administration of MK-8776 18 h after gemcitabine, but not 30 min after, caused significant decrease in tumor growth compared to gemcitabine alone, consistent with the observations made in vitro. This conclusion held in two different tumor models. The pharmacokinetics of MK-8776 in mice is currently being assessed, and we believe it may be possible to increase the length of exposure of tumors to drug and thereby further enhance the therapeutic response.
The clinical development of Chk1 inhibitors has taken many years. The first candidate, UCN-01, was a broad kinase inhibitor but had unfavorable pharmacokinetic properties [
29,
30]. Three subsequent Chk1 inhibitors that entered clinical trial, AZD7762, XL9844 and PF-00477736, have been discontinued; whether this is due to mechanism-based toxicity or off-target effects remains to be determined (the latter drug was reportedly terminated for business reasons rather than concerns for safety or efficacy;
http://www.clinicaltrials.gov). Clinical trials are currently ongoing with LY2606318, LY2606368 and GDC-0425. In most cases, these inhibitors are being studied in combination with gemcitabine or, in one case, pemetrexed [
31]. One issue with all these drugs is that they inhibit several other targets, and in most cases this includes Chk2, although the published information is limited. Indeed, there are currently no publications reflecting the preclinical development of these other agents with which we can compare our current results.
MK-8776 may have an advantage over other Chk1 inhibitors in being much more selective for Chk1 and additionally, it does not inhibit Chk2 [
7]. MK-8776 has completed Phase I clinical trials in combination with gemcitabine although the schedule was based on a 30 min interval between the two drugs. The results of a second Phase I clinical trial in combination with cytarabine has just been reported [
8]. In this case a different schedule was used: cytarabine was administered as a 72 h infusion with MK-8776 given on day 2 and 3 [
8]. The schedule with other Chk1 inhibitors could vary depending upon the time frame over which it can inhibit Chk1, and the DNA damaging agent with which it is combined. For example, LY2603618 has recently been shown to have a plasma half-life of 5 – 25 h, though whether this drug remains bioavailable throughout this time frame is unknown [
31]. Our results provide justification for a schedule of administration whereby gemcitabine is administered 18 h prior to MK-8776, and this justification should apply to clinical trials of gemcitabine with any other Chk1 inhibitor.
Competing interests
The authors declare that they have no competing of interests.
Authors’ contributions
AE designed the overall study. NK designed the in vivo experiments. RM performed the majority of the in vitro experiments with help from RT, IC and AE. HH and NK performed the in vivo experiments. RM and AE wrote the manuscript which was then reviewed and approved by all other authors.