Checkpoint-mediated dependency of tumor cells can be deployed to selectively kill them without substantial toxicity to normal cells. Specifically, loss of CHK1, a serine threonine kinase involved in the surveillance of the G2–M checkpoint in the presence of replication stress inflicted by DNA-damaging drugs, has been reported to dramatically influence the viability of tumor cells. CHK1′s pivotal role in maintaining genomic stability offers attractive opportunity for increasing the selectivity, effectivity, and reduced toxicity of chemotherapy. Some recently identified CHK1 inhibitors entered clinical trials in combination with DNA antimetabolites. Herein, we report synthesis and profiling of MU380, a nontrivial analogue of clinically profiled compound SCH900776 possessing the highly unusual N-trifluoromethylpyrazole motif, which was envisioned not to undergo metabolic oxidative dealkylation and thereby provide greater robustness to the compound. MU380 is a selective and potent inhibitor of CHK1 which sensitizes a variety of tumor cell lines to hydroxyurea or gemcitabine up to 10 times. MU380 shows extended inhibitory effects in cells, and unlike SCH900776, does not undergo in vivo N-dealkylation to the significantly less selective metabolite. Compared with SCH900776, MU380 in combination with GEM causes higher accumulation of DNA damage in tumor cells and subsequent enhanced cell death, and is more efficacious in the A2780 xenograft mouse model. Overall, MU380 represents a novel state-of-the-art CHK1 inhibitor with high potency, selectivity, and improved metabolic robustness to oxidative N-dealkylation. Mol Cancer Ther; 16(9); 1831–42. ©2017 AACR.
Genomic instability, one of the hallmarks of cancer, is attributed to genetic perturbations caused by both endogenous and exogenous sources. Sophisticated and evolutionarily conserved DNA damage response (DDR) machinery in coordination with cell-cycle checkpoints ensure tight surveillance of the functionally redundant repair pathways and act as protective barrier against cancer (1). Checkpoint kinase 1 (CHK1) is a gatekeeper cell-cycle checkpoint kinase with an integral role in DDR (2–4). CHK1 links the upstream sensors ATM and ATR to effector proteins (i.e., cdk/cyclins) to allow proper response to DNA damage. Its role in elimination of the accumulation of damages, maintaining genomic stability and thereby preventing malignancy, can be of considerable therapeutic value (5). In highly proliferating tumor cells, replication stress is caused by inefficient replication, which leads to slowed or stalled replication forks and further increases the reliance on checkpoints (4). Hence, such checkpoint-mediated dependency can be deployed to kill tumor cells selectively without substantial toxicity to normal cells (6). Current line of chemotherapy uses DNA-damaging agents and antimetabolites to produce harmful DNA lesions and thereby activate DDR pathways that can trigger checkpoints. Indeed, loss of CHK1 in the presence of replication stress created by DNA-damaging drugs has been reported to be synthetic lethal in tumor cells (6–8). This is accompanied by S and G2 arrest caused by combination of CHK1 inhibition with a variety of genotoxic agents (9). Several CHK1 inhibitors have been developed over the past years with some entering clinical trials (10). In spite of their clinical potential, none of them have reached the bedside, mainly because of associated off-target effects, toxicity, and poor pharmacologic properties (11). SCH900776 (MK8776), which is currently in phase II clinical trials, represents a state-of-the-art CHK1 inhibitor with minimal intrinsic antagonistic properties (12). SCH900776 has been found to potentiate the action of several DNA antimetabolites including gemcitabine (7), cytarabine (8, 13), and hydroxyurea (12, 14). SCH900776 contains the N-methylpyrazole motif, which is one of the most frequent pharmacophores used in medicinal chemistry (15). However, this moiety has been shown to undergo oxidative demethylation (16) and the resulting metabolite can be the source of undesirable off-target effects (as shown in this study) and/or its formation can cause (at least partly) decreased bioavailability of the parent compound. To overcome these limitations, we have synthesized compound MU380, a fluorinated analogue of SCH900776 possessing highly unusual N-trifluoromethylpyrazole motif, which was envisioned not to undergo metabolic oxidative dealkylation and thereby provide greater robustness to the compound.
In this study, we have tested the efficacy of MU380 and compared its selectivity, potency, and metabolic stability to the clinical candidate SCH900776. Our results show that MU380 is a comparably selective inhibitor of CHK1 and significantly more strongly potentiates sensitization to different DNA-damaging drugs across a panel of tumor cell lines. At the molecular level, the heightened efficacy is reflected through accumulation of lethal amounts of DNA damage, which drives the tumor cells to death in contrast to nontumorigenic cells. This enhanced potency of the compound is accompanied with more persistent MU380-mediated phenotypes in cells and a better pharmacokinetic profile in the mouse. The potential of the compound to be used as a drug and/or a chemical biology probe is further supported by the human xenograft studies. Finally, to our best knowledge, MU380 represents the first example of a biologically active compound with the N-trifluoromethylpyrazole pharmacophore with documented in vivo robustness to the oxidative N-dealkylation.
Materials and Methods
Detailed description of all material and methods is given in the Supplementary Material and Methods.
The synthetic procedures and spectral characteristics of all newly prepared compounds are described in detail in the Supplementary Data (Synthesis).
All cell lines were authenticated using AmpFLSTR Identifiler PCR Amplification Kit. Briefly, cells were grown in appropriate medium supplemented with l-glutamine, penicillin (100 U/mL), streptomycin (0.1 mg/mL), and 10% FBS. SW480, SW620, Caco-2, HT-29, PC-3, DU 145, SKOV-3, MCF10A, A549, and H441 cell lines were obtained from ATCC (LGC Standards in 2009, 2009, 2009, 2000, 2011, 2014, 2001, 2001, 2009, 2008, and 2011). U2OS, A2780, and A2780cis were obtained from European Collection of Authenticated Cell Cultures (ECACC, in 2010, 2001, and 2001, respectively). MDA-MB-231 cells were obtained from Dr. Petr Beneš (Masaryk University) in 2013. MiaPaCa-2, PANC-1, and CAKI- 2 cells were obtained from Dr. Roman Hrstka (Masaryk Memorial Cancer Institute) in 2013. Sk-Br-3 cells were obtained from prof. György Vereb (University of Debrecen) in 2008. MDCK cells were obtained from prof. Jan Vondráček (Institute of Biophyics CAS) in 2013. HCT 116 p53−/−, and HCT 116 p53+/+ were obtained from Prof. Dr. Bert Vogelstein (The Johns Hopkins University) in 2006. HCT 116 PTEN−/−, HCT 116 PTEN+/+ were obtained from Dr. Todd Waldman (Georgetown University) in 2007. BPH-1 and BPH-1 CAFTD04 cells were obtained from Prof. Simon W. Hayward (Vanderbilt-Ingram Cancer Center) in 2009. MiaPaCa-2 luc cells were obtained from Dr. Dawn Quelle (The University of Iowa Carver College of Medicine) in 2015. HeLa.S3-Fucci cells were obtained from Dr. Miyawaki, Atsushi (Riken Brain Science Institute) in 2010.
Cells were seeded in appropriate density and allowed to attach overnight. Attached cells were treated with hydroxyurea or gemcitabine for 24 hours followed by addition of CHK1 inhibitors (either SCH900776 or MU380) for 2 hours. Thereafter, the cells were refurbished with fresh medium and harvested for appropriate assay at indicated time points.
Cell proliferation assays
Cells were treated as described above, followed by proliferation assays 48 hours and 72 hours posttreatments using CyQuant or WST-1, respectively.
Cells were harvested and lysed in radioimmunoprecipitation assay (RIPA) buffer with protease and phosphatase inhibitors. Equivalent protein quantities (30–50 μg) were separated by SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes. The following primary antibodies were used: CHK1, pS296 CHK1, pS345 CHK1, and β-actin. Appropriate secondary antibodies, anti-mouse IgG, or anti-rabbit IgG were used followed by detection of antibody reactivity.
U2OS cells were harvested at 48 hours after treatment, washed and fixed in ice-cold 1% methanol-free formaldehyde solution, and incubated in primary anti-phospho-Histone H2A.X (Ser139) antibody overnight followed by FITC-conjugated secondary antibody. Cells were simultaneously stained with propidium iodide and then analyzed by flow cytometer. For apoptotic assay, Annexin V/PI staining was done according to the manufacturer's protocol.
The Eurofins Kinase Profiler service was used to obtain general selectivity data for SCH900776, MU378, MU379, and MU380 and the IC50 values for inhibition of CHK1.
High content microscopy
DU 145 cells were treated with 12.5 nmol/L gemcitabine alone or 4 μmol/L of CHK1 inhibitors alone or in combination. Thereafter, the cells were allowed to recover in drug-free medium. Experiment was terminated by cell fixation, permeabilization, blocking, and staining with primary γH2AX antibody, followed by secondary antibody and simultaneous DAPI staining. The plates were scanned for FITC and DAPI by ImageXxpress Micro System.
NMOL/L RI/CD1 mice were injected (i.p.) with a single dose of SCH900776 and MU380 in sterile Kolliphor ELP and blood was collected in time intervals. Plasma samples were analyzed by high-performance liquid chromatography (HPLC) under isocratic elution on a Gemini-NX C18 chromatographic column with mass spectrometry detection. The technique of atmospheric pressure chemical ionization in positive mode was chosen for analyte's ion formation.
Mouse xenograft experiments
NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ, JAX) and SHO (Crl:SHO-PrkdcscidHrhr, Charles River Laboratories) mice were subcutaneously injected with A2780 cells or MiaPaCa2 luc cells. After one week, mice were randomly divided into 6 cohorts and treated with either vehicle, gemcitabine, or SCH900776 or MU380, or combinations. For survival study, the treatment was repeated after two weeks. The survival analysis was performed for the period of 66 days and represented as the Kaplan–Meier survival plot. The study was performed in two independent repetitions (n = 9 for each cohort). MiaPaCa2 luc tumor-bearing mice were treated and the tumor size was analyzed under isoflurane anesthesia using optical imaging system IVIS Lumina XR (Perkin Elmer) after intraperitoneal luciferin injection.
For cell line screening, data were standardized as percent of control. A nonlinear regression was applied to generate curves that best fitted the data. The level of statistical significance in all analyses was α = 0.05. All alternative hypotheses were two-sided. Analyses were performed using IBM SPSS Statistics 23, Statistica for Windows 12, and GraphPrism 5.
Caco-2 permeability drug assay
Caco-2 cell monolayers were cultured for 3 weeks. Then the tested compounds (4 μmol/L) were applied either in apical or basolateral compartments. Samples were withdrawn and analyzed by LC/MS. Integrity of the monolayers was checked and the apparent permeabilities, efflux ratio, and mass balance were calculated.
Real-time cell analysis
Acea 96-well E-plates and an xCELLigence real-time cell analysis (RTCA) SP system were used to monitor dynamics of cytotoxic effects of drugs and combinations. After treatments with hydroxyurea and CHK1 inhibitors, the cell index was monitored continually to obtain maximal value (plateau) of the control. HeLa.S-Fucci cells (17) were used for live fluorescence monitoring of the cell-cycle dynamics. Data were analyzed using TrackMate Fiji plugin (18).
Fixed skin samples were processed and stained with CHK1 pS345 (clone 133D3; Cell Signaling Technology; ref. 12). Fixed tumor samples were sliced and stained with γH2A.X (Ser139, clone 20E3; Cell Signaling Technology) Slides were scanned using TissueFAX system (TissueGnostics) and data were analyzed using CellProfiler software (ver. 2.2.0.; ref. 19).
Newly synthesized compound MU380 is a potent and selective inhibitor of CHK1
To identify more selective and robust analogues of SCH900776 with less pronounced off-target effects, we used two strategies: (i) we prepared selected known racemic pyrazolo1,5-a]pyrimidine–based inhibitors with comparable in vitro potency (20, 21) and separated their enantiomers (i.e., MU377 and MU378, Fig. 1A); (ii) we prepared the compound MU380 as a potentially more robust analogue of SCH900776, plus other derivatives possessing the N-trifluoromethylpyrazole motif.
In the synthesis of MU380, we utilized intermediate 6, which is a close analogue of the previously reported compound with ethoxymethyl-protecting groups instead of 2-(trimethylsilyl)ethoxymethyl. The key installation of N-trifluoromethylpyrazole at position 3 of the pyrazolo[1,5-a]pyrimidine scaffold was carried out via the Suzuki coupling of 6 with boronate 9, which was prepared from iodide 8 (Fig. 1). Of note, both iodide 8 and boronate 9 were prepared in house (see the Supplementary Information), as they were not available (and known) at the beginning of the project. To our best knowledge, there is no synthesis of iodide 8 reported in the literature to date. The resulting Suzuki product was fully deprotected and the enantiomers of the corresponding intermediate were separated by HPLC on chiral stationary phase. Finally, the sequence of deprotection and reprotection steps followed by regioselective bromination afforded enantiomerically pure target compound MU380 (experimental details are given in the Supplementary Information). The absolute configuration of MU380 was assigned indirectly, based on the similarity of its CD spectrum with that of SCH900776.
While among the tested compounds, MU377 showed poor in vitro activity, MU378 and MU380 were identified as the active enantiomers with IC50 values of 28 nmol/L and 2 nmol/L, respectively (Supplementary Fig. S1A). Therefore, they were additionally profiled against a panel of 207 human kinases and compared with the clinical candidate SCH900776. As demonstrated in the kinome tree, MU378 and MU380 showed very good selectivity, comparable with the SCH900776 profile (Fig. 1B). Specifically, both MU378 and MU380 were approximately 80 times more active towards CHK1 than CHK2.
MU380 is a more potent sensitizer of U2OS cells to hydroxyurea than SCH900776 and other analogues
In vitro potency of MU380 and MU378 prompted us to test the cellular activity of these compounds. As CHK1 is activated by DNA-damaging agents, such as hydroxyurea, we compared relative efficacy of individual CHK1 inhibitors alone or in combination with hydroxyurea by determination of growth curves and the corresponding EC50 values in U2OS cells (Fig. 2A; Table 1A). Cells were treated for 24 hours with 1 mmol/L hydroxyurea, a dose which was minimally toxic, but sufficient to activate CHK1 (Supplementary Fig. S1C), in the presence or absence of compounds SCH900776, MU378, and MU380. The combination treatment with MU380 was the most effective with approximately 4 times lower EC50 (0.33 μmol/L) compared with SCH900776 (Table 1B). On the other hand, MU378 was relatively inactive, with EC50 value of 2.73 μmol/L.
|Compound .||IC50 (nmol/L) .|
|(B) EC50 values of the tested compounds alone and in combination with HU|
|Compound||EC50 (μmol/L) mean ± SD (n = 3)|
|MU380||0.80 ± 0.08|
|MU380 + HU||0.33 ± 0.06|
|SCH900776||3.57 ± 0.20|
|SCH900776 + HU||1.39 ± 0.17|
|MU378||4.92 ± 0.36|
|MU378 + HU||2.73 ± 0.35|
|Compound .||IC50 (nmol/L) .|
|(B) EC50 values of the tested compounds alone and in combination with HU|
|Compound||EC50 (μmol/L) mean ± SD (n = 3)|
|MU380||0.80 ± 0.08|
|MU380 + HU||0.33 ± 0.06|
|SCH900776||3.57 ± 0.20|
|SCH900776 + HU||1.39 ± 0.17|
|MU378||4.92 ± 0.36|
|MU378 + HU||2.73 ± 0.35|
Abbreviation: HU, hydroxyurea.
Additional in vitro profiling of MU380 along with SCH900776 demonstrated that the chemical modification did not cause significant detrimental changes in the pharmacologic properties. Specifically, both compounds were observed to be only moderately protein bound (Supplementary Table S1) and showed no significant difference in inhibition of human cytochrome P450 isoforms 1A2, 2C9, 2C19, 2D6, and 3A4 (Supplementary Table S2). The human microsomal stabilities (Supplementary Table S3A) were also comparable, with both compounds exhibiting a relatively low clearance (Supplementary Table S3B). The only significant difference was observed in the thermodynamic solubility values (90.6 μmol/L for MU380 and 5974 μmol/L for SCH900776); however, in both cases they were sufficient, that is, significantly higher than the pharmacologically active concentrations used in the cell-based assays and the concentrations observed in the pharmacokinetic studies (see below).
Because of the observed activity of MU380, we used the developed synthetic methodology and prepared additional new target compounds MU381, MU382, MU383, and MU385 (Supplementary Fig. S2A) whose motifs at position 5 of the pyrazolo[1,5-a]pyrimidine scaffold could mimic the piperidine moiety, but do not contain a stereogenic center and their synthesis therefore does not have to be enantioselective. However, all compounds were significantly less potent in the cell (Supplementary Fig. S2B).
Greater potentiation effects of MU380 correlate with biomarkers of CHK1 inhibition in the cell
Next, we assessed the specific inhibition of CHK1 by analyzing ATR-mediated phosphorylation of CHK1 at Ser345 and autophosphorylation of CHK1 at Ser296 upon hydroxyurea treatment (Fig. 2B). A pronounced reduction of Ser296 phosphorylation upon treatment by MU380 alone or in combination with hydroxyurea was observed in comparison with SCH900776, which suggests the higher efficiency of CHK1 inhibition by MU380 in cells. On the other hand, phosphorylation at Ser345 increased, indicating ATR-mediated phosphorylation of CHK1 due to higher level of DNA damage (Fig. 2B). Importantly, these treatments did not alter the total CHK1 level, indicating no change in the stability of the protein. In agreement with its poor cellular activity, MU378 did not effectively reduce Ser296 autophosphorylation (Supplementary Fig. S1C). On the basis of its selectivity and efficient CHK1 inhibition, MU380 was chosen for further cell-based characterization in comparison with SCH900776.
Thereafter, accumulation of DNA damage which contributes to enhanced cell death was monitored. Flow cytometry using γH2AX and Annexin-V/PI staining to assess DNA damage and determine the mechanism of cell death were used, respectively. Combination of MU380 with hydroxyurea induced almost 3-fold increase of γH2AX (Figs. 2C; Supplementary Fig. S3A) and enhanced cell death, as indicated by higher population of apoptotic cells compared with SCH900776 and hydroxyurea alone or in combination (Fig. 2D; Supplementary Fig. S3B). These results were comparable with the response of shRNA-mediated CHK1 knockdown cells treated with hydroxyurea (Supplementary Fig. S3C), suggesting that cells accumulating DNA damage eventually undergo cell death by apoptosis, which is attributed to the specific inhibition of CHK1 by MU380.
MU380 is more potent than SCH900776 in a variety of tumor cell lines
To determine whether sensitizing effects of treatments with various DNA-damaging agents followed by CHK1 inhibition are sufficiently general, we used a panel of 23 cell lines (Fig. 3A). First, in vitro screen covering a wide range of hydroxyurea concentrations alone or in combination with the CHK1 inhibitors was performed. According to the observed proliferation and obtained dose–response curves (Supplementary Fig. S4; Supplementary Table S4), the measure of sensitivity was set as the ratio between the EC50 value of hydroxyurea alone over the EC50 value of the most effective hydroxyurea plus CHK1 inhibitor treatment 0. Most cell lines showed enhanced sensitivity to DNA damage upon CHK1 inhibition. Importantly, in nearly all cases, MU380 was significantly more potent than SCH900776. The most sensitive cell lines have the ratio between 7.0 and 8.8 (Fig. 3A), which corresponds to the combinational treatment with MU380 causing almost 9-times higher sensitization over the hydroxyurea treatment alone. This group includes colon carcinoma cell line HT29 and kidney and colon carcinoma Caki-2 and SW620 cell lines, respectively. However, we did not observe any difference in sensitization of p53+/+ versus p53−/− HCT116 colon carcinoma cells. Importantly, the cell lines derived from nontumorigenic tissues used as controls, including MCF10A, MDCK, and BPH-1, did not reveal marked additional sensitivity with the ratio being near to 1.0 (Fig. 3A). The only cancer cell line that did not show any increase of sensitivity upon CHK1 inhibition was the lung adenocarcinoma cell line H441. To confirm the above described differences between nontumorigenic and cancer cells and to analyze kinetics of the response, we employed xCelligence system for real-time label-free analysis of impedance (22). Kinetic analysis of the response in nontumorigenic MDCK and cancer DU 145 cells confirmed the increased sensitivity of cancer cells to hydroxyurea and CHK1 inhibitor combinations and the higher effectivity of MU380 to potentiate hydroxyurea-induced cytotoxicity (Supplementary Fig. S5).
Next, we tested whether the sensitization described above would be observed also for another DNA-damaging agent (gemcitabine). In the same manner, we measured dose–response curves for eight selected cell lines including the most sensitive, most resistant, and nontumorigenic ones for effect of gemcitabine in combination with the CHK1 inhibitors (Supplementary Fig. S6; Supplementary Table S5). The sensitivity ratios were quite comparable with those obtained for hydroxyurea, except for the H441 line, where we observed higher sensitization (Fig. 3B). Moreover, MU380 was again more potent compared with SCH900776 in nearly all cases. These results indicate the significant effect of CHK1 inhibition upon DNA damage and suggest that the effect is not specific for a particular antimetabolite. However, no significant sensitization was observed when CHK1 inhibition was combined with irradiation (Supplementary Fig. S7).
Next, the Fucci technology (17) was used for direct correlation of the cell-cycle dynamics with the treatment condition in HeLa.S-Fucci cells. Representative Fucci traces of cells treated with gemcitabine confirmed arrest of cells in S/G2–M phase [mAG-hGem (1/100) sensor does not resolve these phases] after they progressed normally into S-phase (Supplementary Fig. S5E). On the other hand, cells treated with gemcitabine in combination with both CHK1 inhibitors showed tracks characteristic for induction of mitotic death (23)—arrested cells reacquired red signal. Although MU380 was significantly more potent than SCH900776 in combination with gemcitabine in these cells, the duration of S/G2–M and G1 did not differ (Supplementary Fig. S5G and S5H). However, we observed increased time interval between first and second cell divisions (Supplementary Fig. S5F).
To exclude the possibility that the observed differences stemmed from various levels of CHK1 or varying intrinsic amounts of DNA damage causing activation of CHK1 to different extent, we analyzed selected cell lines for endogenous CHK1 levels and its basal autophosphorylation using immunoblotting. The levels of CHK1 were approximately the same in all cases. The basal levels of autophosphorylation between cell lines differed (Fig. 3C); however, they did not correlate with the observed sensitivity to the combinational treatment with gemcitabine and MU380/SCH900776.
To further validate molecular effects of MU380, we analyzed the cell lines for autophosphorylation of CHK1 after the combinational treatment with gemcitabine at EC50 concentrations. In all cell lines, we observed activation of CHK1 upon the DNA damage (Supplementary Fig. S8). Treatment of the cells with CHK1 inhibitors in combination with gemcitabine resulted in reduction of CHK1 autophosphorylation (Supplementary Fig. S8), with MU380 again showing more pronounced inhibition compared with SCH900776.
Taken together, we observed that upon combinational treatments with DNA damage-inducing drugs, various cancer cell lines responded better to MU380 than SCH900776, whereas the nontumor cells were not sensitized. This was supported by both cell proliferation assessment and inhibition of CHK1 activity.
The N-trifluoromethylpyrazole motif prevents in vivo demethylation of MU380 to the less selective metabolite
To understand the underlying cause of the enhanced potency of MU380 and to test its stability in cells, we compared the persistence of CHK1 inhibition in U2OS cells. The cells were treated with both MU380 and SCH900776 inhibitors (1 μmol/L) alone or in combination with HU (5 mmol/L), and the time course of their recovery was followed. At the initial time point, the western blot analysis revealed suppression of pS296 autophosphorylation to basal levels upon hydroxyurea treatment with both inhibitors (Fig. 4A, right). In case of SCH900776, the inhibitory effect disappeared as early as 1 hour after the drug removal. However, MU380-mediated inhibition persisted up to 12 hours while the level of total CHK1 remained constant (Fig. 4A).
Next, we elucidated the differences in the kinetics of accumulation of DNA damage caused by CHK1 inhibition. As DU 145 cells were sensitized to both hydroxyurea and gemcitabine upon CHK1 inhibition, we treated them with gemcitabine followed by either CHK1 inhibitors, allowed them to recover in drug-free medium for 0, 6, 24, and 48 hours, and monitored the intensity of γH2AX staining by immunofluorescence microscopy. Relatively fewer γH2AX foci, similar to the basal levels, were observed in cells treated with both CHK1 inhibitors alone (Fig. 4B), indicating that the inhibitors themselves did not increase the DNA damage in these cells. Treatment of the cells with gemcitabine alone or in combination with the CHK1 inhibitors induced at least 4 and 6 times higher occurrence of pan-nuclear γH2AX staining over the control (DMSO) and CHK1 inhibitors alone, respectively. In addition, the extent of damage accumulated with MU380 and gemcitabine was significantly higher than in the experiment with SCH900776 after 48 hours of the compound removal (Fig. 4B). Interestingly, while a substantial amount of damage induced by gemcitabine alone was repaired within 48 hours to the basal levels, the damage induced by the combinational treatment persisted.
To investigate the bioavailability and compare distribution properties of MU380, we employed the widely used Caco-2 cell line model of permeability and efflux transport (24). Correlations with human intestinal absorption data show that compounds with high Apparent Permeability (Papp) in a Caco-2 assay (Papp value of >10 × 10−6 cm/sec) are likely to be well absorbed. An efflux ratio (Papp basal>apical/Papp apical>basal) of >2 indicates active efflux of the compound from the apical surface of the cells. Our data indicate that MU380 has good ability to penetrate through biological barriers as well as bioavailability, comparable with those of SCH900776 (Supplementary Table S6).
Next, we wanted to investigate whether there is a relationship between drug accumulation defects in the tested cell lines and the sensitivity indexes for MU380 and SCH900776 (Fig. 3A and B). Therefore, we performed clustering and correlation analysis of sensitivity indexes to drug retention indexes (DRI). DRIs were measured by fluorescent assay using JC-1 and/or mitoxanthron fluorescent substrates. Our results showed that there was no statistically significant correlation between MU380 and SCH900776 sensitivity indexes and drug retention activity in the tested cell lines (Supplementary Fig. S9A and S9B).
To assess and compare the metabolic stability of MU380 to SCH900776, we employed pharmacokinetic profiling of the compounds in the mouse. The pharmacokinetic profile of MU380 was found to be significantly better than that of SCH900776 (Fig. 5A; Supplementary Table S7). Specifically, while the concentration of MU380 in plasma was still sufficiently high (0.4 μmol/L) after 16 hours, the concentration of SCH900776 was almost undetectable already after 8 hours. The observed concentrations of MU380 correlated well with the number of CHK1 pS345-positive cells (12, 25) detected in mice hair follicles (Supplementary Fig. S10).
To address the hypothesis that SCH900776 is demethylated in vivo, we prepared the putative metabolite MU379 in enantiomerically pure form (Fig. 1A). The synthetic route was analogous to that used for MU380, but we had to optimize the set of protecting groups to avoid undesired bromination on the pyrazole part, which generally produced unseparable mixtures of the mono- and dibrominated isomers (see Supplementary File Synthesis). Interestingly, MU379 is significantly less selective than the parent compound SCH900776: when profiled in a panel of 207 human kinases at 1 μmol/L concentration, it inhibited additional 42 kinases whose residual activity was <50% as compared with SCH900776 (Supplementary Table S8). Moreover, MU379 impaired in a dose-dependent manner cytotoxic effects of both SCH900776 and MU380 in combination with hydroxyurea (Supplementary Fig. S2C). Next, we used compound MU379 as a standard for LC/MS analysis of mouse plasma samples obtained in the pharmacokinetic study. Indeed, MU379 was detected in the experiment with SCH900776 at 1 and 2 hours at concentrations 0.28 to 0.57 μmol/L respectively, while in the experiment with MU380, the metabolite was not detected (limit of detection 0.04 μmol/L).
MU380 is more potent than SCH900776 in A2780 in vivo xenograft model
Finally, we performed in vivo experiments to investigate the antitumor activity of the combination of gemcitabine (150 mpk) and MU380 (25 mpk) compared with gemcitabine (150 mpk) and SCH900776 (25 mpk) combination and gemcitabine (150 mpk) alone. On the basis of the results of previous study of Guzi and colleagues (12), we employed the A2780 xenograft model and survival analysis. A2780 tumor-bearing animals treated with two boluses of gemcitabine and the CHK1 inhibitors (concurrent treatment) showed statistically significant longer median survival only for the combination with MU380. The median survival observed for the combination of gemcitabine with SCH900776 was similar to that of the gemcitabine monotherapy (Fig. 5B; Supplementary Table S9). This difference was confirmed by statistical analysis using log-rank (Mantel–Cox) test (Supplementary Table S10). Next, we used MiaPaCa2 luc pancreatic in vivo model which is less sensitive to gemcitabine-induced cytotoxicity (26), but showed significant sensitization when gemcitabine was combined with MU380 or SCH900776 in our screen in vitro (Fig. 3B). On the basis of the previously published results and our pilot experiments, the animals were scheduled to receive three boluses on every fifth or seventh day with 4-hour or 24-hour delays between applications of gemcitabine and the CHK1 inhibitors. Nude SHO mice were monitored by bioluminescence imaging (BLI). Extra cohorts of animals were sacrificed 24 hours after the first cycle of therapy for determination of γH2AX in MiaPaCa2 tumors. Combinations of gemcitabine and SCH900776 or gemcitabine and MU380 were both tolerable, as indicated by monitoring of animal weight during the experiment (Supplementary Fig. S11B). In the 4-hour schedule, both compounds enhanced gemcitabine efficacy to a similar extent (Supplementary Fig. S11A). In the 24-hour schedule, the effect of SCH900776 was relatively marginal while it was statistically significant for MU380 (Fig. 5C). SCH900776 and MU380 dosed as monotherapy induced minimal γH2AX positivity, comparable with vehicle in MiaPaCa tumors. However, administration of the CHK1 inhibitors to animals previously injected with gemcitabine augmented number of γH2AX-positive nuclei (Fig. 5D; Supplementary Fig. S11C and S11D). Statistically significant difference between gemcitabine monotherapy and gemcitabine/CHK1 inhibitor combination was observed only in the case of combination with MU380.
Despite rapid development in the area of biologicals, small molecules still represent a substantial part of modern drugs. Targeting certain protein kinases has been particularly attractive in modern oncology (27, 28). Among these, kinase signaling pathways within the DDR mechanism have been identified as suitable therapeutic targets, especially for combinational treatment (29–31). Specifically, the attractiveness of targeting CHK1 is illustrated by the number of preclinically and clinically profiled inhibitors (32). However, it remains to be seen whether any of the substances possess optimal combination of potency, selectivity, and good pharmacologic properties to reach the ultimate clinical use. Therefore, identification of new compounds fulfilling these requirements is of significant importance.
Herein, we describe synthesis and characterization of a novel CHK1 inhibitor MU380, which is a nontrivial analogue of clinically profiled compound SCH900776, with enhanced inhibitory potency toward CHK1, heightened response to chemosensitization in tumor cells and, importantly, improved stability/bioavailability in the cell and in vivo.
While SCH900776 has reached phase II clinical trials (NCT01870596), here we show that it has a relatively short half-life. It also undergoes rapid in vivo demethylation and the resulting metabolite (MU379) is significantly less selective than the parent compound. The metabolite can not only cause unspecific off-target effects, but it can attenuate the desired phenotype, for example, by inhibition of CDKs (12).
To prevent this undesirable demethylation, we have prepared MU380, with a very unusual N-trifluoromethylpyrazole motif. Compared with SCH900776, MU380 shows a significantly better pharmacokinetic profile. MU380 is not highly protein bound in the mouse, has good aqueous solubility to reach active concentrations in vivo, does not significantly inhibit major human P450 enzymes, and has a relatively low clearance in human microsomes. Importantly, MU380 does not undergo the undesired N-dealkylation, which is supported by the absence of the significantly less selective metabolite MU379 in mouse blood plasma. Introduction of fluorine into the molecule to improve metabolic stability, physicochemical properties, and binding to the biological target is a standard strategy in medicinal chemistry. Correspondingly, a number of small-molecule drugs contain fluorine (33), and the development of selective and mild fluorinations represents a very active area of organic synthesis (34). However, N-trifluoromethylated azoles are rare and reliable and sufficiently mild methods of their preparation have come into the focus only recently (35). The fact that the in vivo profile of MU380 is superior to SCH900776 is further supported by A2780 xenograft model, where we observed increased median survival of the animals treated with gemcitabine in combination with the CHK1 inhibitor. When we used a different model, gemcitabine-refractory MiaPaCa2 luc, we obtained similar results.
Mechanistically, MU380 showed prolonged CHK1 blockage in the cell, evidenced by extended inhibition of autophosphorylation and inability to repair DNA damage. The overall kinase selectivity profile of MU380 is similar to that SCH900776, with marked selectivity toward CHK1 over CHK2. This is important, as targeting CHK1 over CHK2 is more desirable, as it has been shown that dual depletion or inhibition of CHK2 and CHK1 do not further sensitize cancer cells to DNA antimetabolites (12). Likewise, similar ability to pass through cell membranes and efflux from cells via passive or active mechanisms was observed for both compounds in the Caco-2 permeability assay. Drug transporter ABCG2 has been suggested to play role in sensitivity/resistance to nucleoside analogues through drug efflux modulation (36, 37). Therefore, we performed correlation analysis between MU380 and SCH900776 sensitivity indexes and drug retention indexes determined using fumitremorgin C, a selective ABCG2 inhibitor. The absence of significant correlation indicates that ABCG2-dependent drug efflux is not a mechanism responsible for the observed differences in sensitivity of screened cell lines to the tested drug combinations.
Current clinical trials of CHK1 inhibitors principally consist of their combination with antimetabolites. Along with this scenario, MU380 was able to sensitize a wide range of cancer cell lines to hydroxyurea and gemcitabine, ranging from 4- to 12-fold sensitization. Importantly, we did not observe significant sensitization of nontumor cells. However, no significant sensitization was observed when either compound was combined with irradiation, which is in accordance with previous reports (38).
Preclinical data suggest that SCH900776 can sensitize tumor cells to antimetabolites in a p53-dependent manner, which is logical as the dual loss of CHK1 and p53 would result in abrogation of all checkpoints and further potentiate sensitivity to chemotherapy (39). However, in our study, no difference was observed for p53-deleted HCT116 colon cancer cells compared with their wild-type counterpart, which indicates possible activation of compensatory pathway(s). This is in agreement with the previous report where CHK1 inhibitor AZD7762 or ATR inhibitor VE-821 did not sensitize p53−/− cells (40). Moreover, we did not observe correlation between the sensitization of the cell lines and the status of 69 known or putative tumor suppressors identified in Cancer Cell Line Encyclopedia (Supplementary Table S11; ref. 41). The differences in sensitivity of individual cell lines could be explained neither by different endogenous CHK1 levels nor heightened CHK1 autophosphorylation activity in response to DNA-damaging agents.
While we and others were unable to identify other potential predictive biomarkers apart from general pS296 and γH2AX for selective sensitization of tumor cells by CHK1 inhibition (42, 43), in specific cases, a mechanistic link might be more evident. Utilization of a specific genetic background has been successfully applied in synthetic lethal (SL) treatments with PARP inhibitors sensitizing breast or ovarian cancers carrying BRCA mutations (44) and in a more general context in targeted therapy and personalized medicine (32, 45). Similarly, SL studies based on siRNA screening suggest that CHK1 ablation might be beneficial for cancers with deficiency of FANC genes (46) and Wee1 (29, 47). In addition, replication stress induced by specific genetic background could also represent a sensitivity determinant as reported for treatment of Burkitt lymphomas with ATR inhibitors (48).
In summary, MU380 represents a novel state-of-the-art CHK1 inhibitor with excellent potency, good selectivity, bioavailability, and metabolic stability, suitable for further preclinical and potentially clinical progression. In addition, to our best knowledge, this study provides the first experimental evidence of the in vivo metabolic robustness of the N-trifluoromethylpyrazole motif, suggesting that this moiety can be used more generally in modern medicinal chemistry.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: F. Nikulenkov, L. Krejčí, K. Paruch, K. Souček
Development of methodology: P. Samadder, T. Suchanková, O. Hylse, P. Khirsariya, F. Nikulenkov, A. Hampl
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Samadder, T. Suchanková, O. Hylse, P. Khirsariya, F. Nikulenkov, S. Drápela, N. Straková, P. Vaňhara, K. Vašíčková, L. Binó, P. Kollár, J. Jaroš, A. Hampl, K. Souček
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Samadder, T. Suchanková, O. Hylse, F. Nikulenkov, S. Drápela, P. Vaňhara, P. Ovesná, R. Fedr, M. Ešner, J. Jaroš, K. Paruch
Writing, review, and/or revision of the manuscript:P. Samadder, T. Suchanková, F. Nikulenkov, K. Vašíčková, A. Hampl, L. Krejčí, K. Paruch, K. Souček
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Khirsariya, H. Kolářová, L. Binó, M. Bittová
Study supervision: L. Krejčí, K. Paruch, K. Souček
The authors thank Prof. Dr. Bert Vogelstein for his kind gift of HCT 116 p53+/+ and p53−/− cells, Prof. Simon W. Hayward for BPH-1 and BPH-1 CAFTD04 cells, Dr. Todd Waldman for HCT 116 PTEN+/+ and PTEN−/− cells, Dr. Dawn Quelle MiaPaCa-2 luc cells, Dr. Roman Hrstka for MiaPaCa-2, PANC-1, and CAKI- 2 cells, Dr. Miyawaki Atsushi for HeLa.S-Fucci cells, and Prof. Raděk Vrtěl for cell line authentication. The authors also thank Zuzana Kahounová (née Pernicová) for help with pilot experiments, Iva Lišková, Martina Urbánková, Katarína Marečková, and Kateřina Svobodová for superb technical assistance, Pavla Řezníčková, Ráchel Víchová, and Nina Charvátová for maintenance of animal facility.
This work was supported by Ministry of Health of the Czech Republic (grant no. 15-33999A), all rights reserved (to L. Krejčí, K. Paruch, K. Souček), project no. LQ1605 from the National Program of Sustainability II (MEYS CR) and by the European Union - project ICRC-ERA-HumanBridge (no. 316345; to L. Krejčí, K. Paruch, A. Hampl, K. Souček), HistoPARK (CZ.1.07/2.3.00/20.0185, to A. Hampl and K. Souček), CELLBIOL (CZ.1.07/2.3.00/30.0030, to T. Suchánková), and project CZ-OPENSCREEN: National Infrastructure for Chemical Biology (LM2015063, to K. Paruch).
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.