BOS172722 (CCT289346) is a highly potent, selective, and orally bioavailable inhibitor of spindle assembly checkpoint kinase MPS1. BOS172722 treatment alone induces significant sensitization to death, particularly in highly proliferative triple-negative breast cancer (TNBC) cell lines with compromised spindle assembly checkpoint activity. BOS172722 synergizes with paclitaxel to induce gross chromosomal segregation defects caused by MPS1 inhibitor–mediated abrogation of the mitotic delay induced by paclitaxel treatment. In in vivo pharmacodynamic experiments, BOS172722 potently inhibits the spindle assembly checkpoint induced by paclitaxel in human tumor xenograft models of TNBC, as measured by inhibition of the phosphorylation of histone H3 and the phosphorylation of the MPS1 substrate, KNL1. This mechanistic synergy results in significant in vivo efficacy, with robust tumor regressions observed for the combination of BOS172722 and paclitaxel versus either agent alone in long-term efficacy studies in multiple human tumor xenograft TNBC models, including a patient-derived xenograft and a systemic metastasis model. The current target indication for BOS172722 is TNBC, based on their high sensitivity to MPS1 inhibition, the well-defined clinical patient population with high unmet need, and the synergy observed with paclitaxel.
The key mechanism ensuring proper chromosome segregation during mitosis is the spindle assembly checkpoint (SAC), which monitors the correct bipolar attachment and tension of microtubules (MT). When all MTs have been properly attached to the kinetochores, the cells enter anaphase by releasing activators of the anaphase promoting complex/cyclosome (APC/C; refs. 1, 2). One of the pivotal proteins of the SAC is MPS1 kinase (also known as TTK). MPS1 is vital for the recruitment of kinetochore components, namely, a complex of MAD2 and MAD1, to unattached kinetochores (3, 4) which in turn bind and lock the APC/C coactivator cdc20, keeping the APC/C inhibitory complex inactive. MPS1 is further essential for sustaining this inhibitory complex throughout mitosis (5–7) and for correcting improperly attached chromosomes (8). Consequently, if MPS1 is inhibited, the time cells spend in mitosis is drastically reduced, resulting in elevated chromosome segregation errors and overall aneuploidy reaches detrimental levels (3, 6, 9, 10). Although most cancers have a high frequency of aneuploidy (11, 12) and chromosome instability (CIN) is a common feature (13–15), even these cells cannot tolerate aneuploidy beyond a certain threshold and increasing CIN has been shown to have a negative impact on their overall viability (16–18). Thus, instant generation of unsustainable aneuploidy induced by MPS1 inhibition poses an attractive area for therapeutic intervention in cancer, and several inhibitors have been previously reported (19–26).
Breast cancer comprises distinct subtypes (27–29) with both human epidermal growth receptor 2 (HER2) overexpressing and basal-like breast cancer, including triple-negative breast cancers (TNBC), having a significantly worse prognosis than luminal and normal-like cancers (28, 30–32). Within the distinct subgroups, TNBCs are associated with the highest proliferation rate, and the expression of gene signatures is associated with the cell cycle (29, 33, 34). For these reasons, antimitotic chemotherapy seems to be a rational option in TNBC, and we sought to identify synergistic combinations with established therapeutics in order to maximize benefit while minimizing the potential for the emergence of secondary resistant tumors.
Materials and Methods
Reagents and antibodies
Eribulin was purchased from Eisai Pharmaceuticals, and HOECHST33342 and DAPI from Life Technologies. All other reagents were obtained from Sigma.
Cell lines were purchased from ATCC and DSMZ. In-house authentication of cell lines by SNP profiling was carried out, and cultured cells were passaged for less than 6 months before replacement from early passage frozen stocks. Cells were regularly screened for Mycoplasma, using a PCR-based assay (VenorGem; Minerva Biolabs).
We performed a high-throughput screen at Horizon Discovery Inc. as described below. The endpoint readout of this assay is based upon quantitation of ATP as an indicator of viable cells (except when noted in Analyzer). Once cells reached expected doubling times, screening begins. Cells are equilibrated in assay plates via centrifugation and placed in CO2 incubators (attached to the Dosing Modules) at 37°C for 24 hours before treatment. At the time of treatment, a set of assay plates (which do not receive treatment) are collected, and ATP levels are measured by adding ATPLite (PerkinElmer). These T-zero (T0) plates are read using ultrasensitive luminescence on Envision plate readers. Assay plates are incubated with compound (10-point treatment) for 120 hours and are then analyzed using ATPLite. All data points are collected via automated processes and are subject to quality control. GI50 and synergy was determined using Horizon's proprietary software Chalice. Assay plates are accepted if they pass the following quality control standards: relative raw values are consistent throughout the entire experiment, Z-factor scores are greater than 0.6, and untreated/vehicle controls behave consistently on the plate. GI50 determination in breast cancer cell lines was carried out in-house in 384 (Greiner Bio-One, #781091), as described below. Cells were seeded at individual optimal cell densities, and drugs were added using an Echo liquid handler (Labcyte). After 5 days, cells were incubated with HOECHST 33342 stain (10 μg/mL) and propidium iodide (1 μg/mL) and assays read on a Celigo Imaging Cytometer (Nexcelom) using the Dead&Total application. GI50 values were assessed using GraphPad prism and a sigmoidal fit. For synergy screening in breast cancer cell lines, the respective drug concentrations were 0.001 and 0.002 μmol/L paclitaxel, 0.0001 and 0.0002 μmol/L eribulin, 0.002 and 0.004 μmol/L doxorubicin, and analysis was performed using the “MacsynergyII” spreadsheet. In brief, cells were seeded at optimal densities in 96-well plates. BOS172722 and paclitaxel were added at a fixed ratio determined by the respective GI50 values for each compound and cell line. Cell viability was assayed after 5 days using the MTT reagent and calculations were done using the Compusyn program.
Meso scale discovery (MSD) assay
In-house electrochemiluminescence (MSD) assays were developed to measure tubulin acetylation and histone H3 phosphorylation. After treatment, cells were washed with PBS and lysed with RIPA buffer [150 mmol/L NaCl, 50 mmol/L Tris pH 7.5, 1 mmol/L EDTA pH 8.0, 1% (v/v) NP40, 1% sodium deoxycholate, 0.1% SDS, 10 mmol/L NaF, protease inhibitor tablet, and phosphatase inhibitor cocktails] and sonicate briefly (3–4 pulses at mid power). Protein lysates were then diluted 1:10 in lysis buffer [50 mmol/L NaCl, 20 mmol/L Tris pH 7.5, 1 mmol/L EDTA, 1 mmol/L EGTA, 1% (v/v) Triton X-100, 10 mmol/L NaF, protease inhibitor tablet and phosphatase inhibitor cocktails] to be compatible with the MSD buffer content required. For the acetylated tubulin MSD assay, 25 μL of lysate (0.2–0.4 μg/μL) was loaded onto MSD plates that were precoated with anti-tubulin antibody (1:100 in PBS, mouse monoclonal, Sigma, cat. no. 9026) and blocked with 3% (w/v) BSA, and protein lysates were incubated on the plate for 1 hour at room temperature on a shaker. Plates were with MSD wash buffer, and 25 μL of anti-acetylated tubulin antibody (rabbit polyclonal, Cell Signaling Technology, cat. no. 5335) 1:100 diluted in 1% (w/v) BSA was added followed by incubation for a further 1 hour at RT. Plates were washed with MSD wash buffer and incubated with 25 μL of anti-rabbit sulfo-TAG antibody (Meso Scale Discovery, cat. no. R32AB) diluted in 1% (w/v) BSA) for 1 hour. After the final incubation, plates were washed with MSD wash buffer and read in the presence of 1× MSD read buffer. IC50 values were determined using GraphPad prism. For phosphorylated histone H3 MSD assay, the same preparation was used except that the plate was precoated with anti-pan histone antibody (2 μg/mL diluted in PBS, mouse monoclonal, Millipore, cat. no. MAB3422) and anti–phospho-histone H3 antibody (rabbit polyclonal, Millipore, cat. no. 06-570) 1:100 diluted in 1% (w/v) BSA was used to detect the phosphorylated Histone H3.
Flow cytometry to determine the cell-cycle profile in HeLa cells after treatment with 100 and 200 nmol/L BOS172722 for 24 hours was performed as previously described (23).
Cells were seeded in 10-cm dishes. The next day, paclitaxel or 0.25% DMSO was added. After 36 hours, cells were arrested using nocodazole (100 ng/mL) and further incubated for 4 hours. Mitotic cells were collected via mitotic shake off, pelleted, resuspended in 0.75 mmol/L KCl and incubated at 37°C for 8 minutes. After centrifugation, cells were fixed using a −20°C solution of 4:1 methanol:acetic acid. Cells were pelleted and fixative removed. Cell suspension (15 μL) was dropped onto a glass slide and stained with 10 μg/mL DAPI. Pictures were taken on a Zeiss Imager.D1 microscope equipped with an AxioCam MRm using Axiovision software (Zeiss).
Live-cell imaging, immunofluorescence, and immunoprecipitation
Analysis by immunofluorescence and time-lapse microscopy as well as immunoprecipitations was performed as previously described (7).
IHC and immunofluorescence in human tumor xenografts
Tumors were fixed in 10% neutral buffered formalin and embedded in paraffin. Phosphorylation of T875 on KNL1 was determined by IHC. Heat-based antigen retrieval was performed by boiling the 4-mm-thick tissue sections in pH = 6 citrate buffer (TCS Biosciences Ltd., HDS05, 1:100 dilution) for 5 minutes in a pressure cooker. The sections were incubated with a rabbit polyclonal T875-KNL1 antibody for 2 hours and detected using a Vectastain Elite ABC kit (Vector Laboratories) and DAB reagent (Dako). Nuclei of the cells were located by counterstaining the sections with Harris' hematoxylin. The T875-KNL1 antibody was generated by immunizing rabbits with phosphorylated peptides CNDMDI(pT)KSYTI (Eurogentec). T875-KNL1–positive cells were quantified by manually counting positively stained cells in 8 random fields of the tumor section. Phosphorylation of Histone H3 was determined by immunofluorescence. Heat-based antigen retrieval was performed as described above, except that the slides were heated in the microwave for 10 minutes. The sections were incubated with rabbit polyclonal p-histone H3 (S10) antibody (Millipore, 06-570) for 1 hour, then incubated with secondary goat anti-rabbit IgG (H+L) Alexa Fluor 488 (Invitrogen, A-11034) and counterstained with DAPI. For each section, 9 fluorescent images were captured, and quantification of p-Histone H3-positive cells was done in CellProfiler software (www.cellprofiler.org).
Real-time quantitative PCR
RNA from cells was extracted using the Quick-RNA kit (Zymo Research) according to the manufacturer's instructions. Real-time quantitative PCR reactions of the MPS1 gene were carried out using the TaqMan Universal PCR Master Mix (Applied Biosystems) in the Applied Biosystems StepOne-plus Real-time PCR System, following the manufacturer's instructions. We used commercially available primers and probes for PCR analyses (TaqMan Gene-Expression Assays, Assay ID: Hs01009870_m1 for MPS1, and Hs03003631_g1 for 18S and Hs04195421_s1 for PP1A as endogenous controls; Applied Biosystems). PCR conditions were as follows: 95°C for 10 minutes, followed by 40 cycles at 95°C for 15 seconds and 60°C for 1 minute. Each sample was assayed in triplicate with RNase-free water as negative control. Relative gene-expression quantifications were calculated according to the comparative Ct method using 18S or PP1A as an endogenous control. Final results were determined by the formula 2−ΔΔCt.
In vivo efficacy studies
The MDA-MB-468 and OD-BRE-503 patient-derived xenograft studies were carried out at Oncodesign S.A., France. Statistical analyses of mean tumor volumes at randomization were performed using ANOVA, and pairwise tests were performed using the Bonferroni/Dunn correction in case of significant ANOVA results. A P value < 0.05 was considered as significant. For the metastatic model, MDA-MB-231 luciferase-expressing cells were injected i.v. in the tails of NOD SCID mice. Mice were dosed starting 6 days after tumor cell implantation, with paclitaxel at 15 mg/kg on days 0 (day 0 is 6th day after tumor cell implantation), 7, 14, and 21 i.v. and BOS172722 at 30 and 40 mg/kg p.o. on days 0 + 1, 7 + 8, 14 + 15, and 21 + 22, and tumor burden was assessed by whole body bioluminescent imaging. Animals were culled when they showed signs of deterioration due to tumor burden (body weight loss and rapid breathing). All animal studies were carried out in accordance with the UK Animals (Scientific Procedures) Act 1986 and national guidelines (35).
BOS172722 is a novel, potent, and orally bioavailable MPS1 inhibitor in biochemical and cellular assays
BOS172722 is a novel, orally bioavailable, potent, and highly selective small-molecule inhibitor of MPS1 kinase discovered from our in-house lead optimization studies on a pyridopyrimidine series of compounds (Fig. 1A; ref. 36). In vitro kinase assays using purified recombinant MPS1 protein showed that BOS172722 inhibited MPS1 activity with IC50 values of 0.004 μmol/L at low ATP (10 μmol/L) and 0.01 μmol/L at high (1 mmol/L) ATP concentrations, respectively (Fig. 1A). The in vitro profiling of BOS172722 was tested in a wide panel of more than 400 kinases (36). Only a small number of other kinases were inhibited by BOS172722, in particular JNK1, JNK2, JNK3, and LRRK2 at >80% at 1 μmol/L. Follow-up IC50 values were obtained (JNK1 IC50 = 0.092 μmol/L, JNK2 IC50 = 0.076 μmol/L, JNK3 IC50 = 0.242 μmol/L, and LRRK2 IC50 = 0.048 μmol/L), showing that BOS172722 is highly selective for MPS1 over these other kinases (36). We have previously described the use of an in-house electrochemiluminescence (MSD) assay to quantitatively measure MPS1 auto-phosphorylation in cells at T33/S37 sites (22, 23). BOS172722 potently inhibited MPS1 T33/S37 auto-phosphorylation in MSD assay with IC50 value of 0.06 ± 0.03 μmol/L (Fig. 1A). To prove that the observed inhibition of MPS1 phosphorylation it is not a consequence of a mitotic exit, nocodazole-arrested HCT116 cells, a colon cancer cell line sensitive to MPS1 inhibition widely used for MPS1 activity investigation (20), were cotreated with BOS172722 and MG132 proteasome inhibitor to block exit from mitosis, for 2 hours. Nocodazole-induced MPS1 auto-phosphorylation on T33/S37 was completely inhibited by treatment with BOS172722, indicating that BOS172722 specifically inhibits the activity of MPS1 (Fig. 1B). Next, HCT116 cells treated with different concentrations of BOS172722 for 24 and 48 hours were analyzed for inhibition of histone H3 phosphorylation and PARP cleavage by immunoblotting. Histone H3 phosphorylation at Ser10 was inhibited in a time-dependent manner (Supplementary Fig. S1A). Induction of apoptotic cell death upon drug treatment also increased as determined by levels of cleaved PARP (Supplementary Fig. S1A). To identify the time that is required for BOS172722 to induce maximum effect on tumor cell growth inhibition, we performed wash-off experiments. HCT116 cells were treated with BOS172722 for 2 to 96 hours followed by the measurement of growth inhibition (GI50) at 96 hours. We observed that 24 hours of treatment showed a comparable GI50 to 96 hours treatment (Supplementary Fig. S1B), indicating that 24 hours of MPS1 inhibition is sufficient to achieve maximum growth inhibition in tumor cells.
Effects of BOS172722 on SAC activity and cell cycle
MPS1 activity is required for activation of SAC, and inhibition of its activity results in SAC override and mis-segregation of chromosomes. In order to investigate the mechanism of action of BOS172722, we performed live-cell imaging of H2B-mCherry–transfected HeLa cells, a model widely used to study mitosis. We measured time in mitosis and chromosome segregation defects after treatment with BOS172722. We found that treatment with 200 nmol/L of BOS172722 resulted in early mitotic exit of 11 minutes compared with 52 minutes for the untreated HeLa cells (Fig. 1C). This sharp decrease in time in mitosis resulted in gross chromosomal abnormalities, including unaligned chromosomes (∼83%) and decondensation of chromosomes without division (∼17%; Fig. 1D). In addition, induction of aneuploidy and loss of normal cell-cycle profile was observed in cells treated with BOS172722 for 24 hours with the indicated concentrations (Fig. 1E). MPS1 activity is required for the recruitment of SAC components to the unattached kinetochores. To test this, HeLa cells were pretreated with BOS172722 for 1 hour, then treated with nocodazole, MG132, and BOS172722 for an additional 1 hour to arrest the majority of cells in mitosis, followed by fixation and staining with the indicated antibodies (Supplementary Fig. S1C). Treatment of cells with 200 nmol/L BOS172722 resulted in loss of recruitment of MAD1, MAD2, BUBR1, and KNL1 to the unattached kinetochores. A reduction in MPS1 T33/S37 auto-phosphorylation was also observed with BOS172722 treatment, whereas MPS1 levels at kinetochores remained unchanged (Supplementary Fig. S1C).
Cell proliferation rate and SAC activity are important indicators for BOS172722 potency
Prompted by the observation that PTEN-deficient cell lines are more sensitive to MPS1 inhibition (37) and in order to identify additional populations that may benefit from an MPS1 inhibitor, we tested 50 cancer cell lines, including 25 PTEN-proficient and 25 PTEN-deficient, with BOS172722 in a 5-day growth inhibition assay (Supplementary Table S1A). The results showed that, although there is a clear trend in sensitivity to MPS1 inhibition with BOS172722 between PTEN-deficient and PTEN-proficient cell lines, the difference was not statistically significant (Fig. 2A). Treatment of 50 cancer cell lines from 8 different tissues of origin, in addition to the data we have generated using TNBC cell lines, showed that lung cancer cell lines overall show similarly sensitivity to MPS1 inhibition (Supplementary Fig. S2A). In addition, similar to TNBC cell lines, the average doubling time of the cell lines above the median GI50 was higher than the average doubling time of the cell lines below the median GI50. Moreover, there was a statistically significant difference between the doubling time of the most sensitive cell lines (top tercile, n = 17, 38 hours) relative to the rest of the cell lines (n = 33, 53 hours; Fig. 2B), suggesting that more rapidly dividing cells show greater susceptibility to MPS1 inhibition. Toward the identification of mutations that could predict sensitivity to BOS172722, we extracted nonsilent mutations in the coding regions of genes for 16 sensitive cell lines (GI50 < 50 nmol/L) and 16 resistant cell lines (GI50 > 200 nmol/L) from the Cancer Cell Line Encyclopaedia database and looked for enrichment of mutated genes in the sensitive group. We selected genes that were mutated in at least 3 cell lines in the sensitive group over the resistant group and in no more than two cell lines in the resistant group and calculated the fold enrichment in the two sets (Supplementary Table S1B). The most enriched mutated gene in the sensitive group was PI4KB, a protein that has been shown to be active in mitosis (38) and to prevent formation of polylobed nuclei (39), which is indicative of mitotic exit with aberrant chromosome segregation and aneuploidy (Fig. 2C). Similarly, mutations in ARID1A and SMARCA4 have been shown to induce genomic instability and aneuploidy (40) that is further increased to intolerable levels by MPS1 inhibition, resulting in cell death. In addition, mutations in NUMA1 and TPR, that have been shown to be involved in mitotic spindle assembly and in the activation of the SAC (41–43), may directly sensitize cells to MPS1 inhibition.
We have previously shown that basal breast cancer cell lines (including TNBC) were more sensitive to our tool compound MPS1 inhibitor CCT271850 in comparison with luminal breast cancer cell lines (23). To investigate further, we used a panel of TNBC and non-TNBC cell lines to investigate the association between the potency of BOS172722 to (i) TNBC versus non-TNBC cell lines; (ii) MPS1 expression; (iii) proliferation rate; and (iv) SAC activity. We confirmed that TNBC cell lines are more sensitive to BOS172722 in comparison with non-TNBC cell lines (Supplementary Fig. S2B and Supplementary Table S2). Our data from breast cancer cell lines are in agreement to the published data from primary breast tumors (44) that MPS1 expression is higher in TNBC versus non-TNBC cell lines (Supplementary Fig. S2C and S2D). Importantly, we found that the cellular potency of BOS172722 at Emax (representing maximum effect) significantly associated with the proliferation rate of the cell lines as measured by their doubling times (Fig. 2D), similar to the panel of the 50 cell lines from different types of human cancers described above (Fig. 2B); TNBC cell lines, being more sensitive, have shorter doubling times (Supplementary Fig. S2E). We then investigated the association between BOS172722 potency and levels of SAC activity. To our surprise, cell lines with reduced SAC activity, as measured by MPS1 phosphorylation and BUB1 localization upon mitotic arrest, were more sensitive to BOS172722 treatment (Fig. 2E and F; Supplementary Table S2), indicating that cell lines with a compromised SAC may require lower doses of an MPS1 inhibitor to abrogate the mitotic checkpoint, thereby inducing gross chromosomal abnormalities and cell death. TNBC cell lines overall showed an overall weaker SAC. A t test or a one-way ANOVA, shows P < 0.009. However, we used the u test as these data failed to pass the Shapiro–Wilk normality test. Based on the u test, TNBC cells have on average lower pMPS1/mitotic index ratios (mean 4.75, standard deviation 2.31) relative to non-TNBC cells (mean 32.5, standard deviation 21.6), including two extreme non-TNBC outliers with ratios of 1.1 and 2.2. However, although low ratios may be found in other populations, TNBC cells represent a homogeneous group. These data together suggest that the proliferation rate together with the SAC activity may potentially be used as stratification markers for TNBC patient selection to achieve maximum efficacy.
BOS172722 shows synergistic effect with paclitaxel in TNBC cell lines
Based on its promising in vitro profile, we progressed BOS172722, to human tumor xenograft models of TNBC to initially evaluate single-agent efficacy. In established orthotopic MDA-MB-231 xenografts, BOS172722 given at 50 mg/kg orally, twice a week for 47 days showed significant but moderate reduction of tumor growth compared with vehicle-treated mice (tumor growth inhibition: TGI = 66%, P = 0.0001; Supplementary Fig. S3). However, to support clinical trials as a single agent would require evidence of tumor stasis or regression in preclinical models. We therefore screened a panel of TNBC cell lines for synergism between BOS172722 with paclitaxel (Tax) at 1 and 2 nmol/L, doxorubicin (Dox) at 2 and 4 nmol/L and eribulin (Eri) at 0.1 and 0.2 nmol/L, to a wide range of BOS172722 concentrations (0–500 nmol/L). The data were then analyzed using MacSynergyII (45). Of the combinations of compounds tested, only paclitaxel with BOS172722 showed consistent synergy across our panel of TNBC cell lines (Fig. 3A). By contrast, treatment of BOS172722 together with eribulin or doxorubicin did not show any synergy in the majority of the cell lines tested, with few exceptions (Fig. 3A). We therefore focused on further characterizing the mechanism of the synergism between BOS172722 and paclitaxel. In our screen, we used concentrations of 1 and 2 nmol/L of paclitaxel and the maximum synergism was observed at 1 nmol/L or <2 nmol/L for all cell lines (Supplementary Table S3). Importantly, the observed synergistic concentrations were relevant to the clinical concentrations of paclitaxel (46). In addition, the paclitaxel concentrations where maximum synergy is observed are either at or below the respective GI50 values of paclitaxel alone and are related to clinical dosing schedules (ref. 46; Supplementary Table S3).
In order to investigate the mechanism of action of the synergism of paclitaxel with BOS172722, we performed flow cytometry analysis of the cell cycle of MDA-MB-231 after treatment with each drug individually or in combination. Treatment of MDA-MB-231 cells with 1 nmol/L paclitaxel or 100 nmol/L BOS172722 for 24 hours did not show any significant cell-cycle effect on the cells. However, the combination of both agents led to a pronounced decrease in height and broadening of the G1 cell-cycle peak, consistent with the induction of aneuploidy (Fig. 3B). At these drug concentrations, the cell-cycle profile was completely abolished (an expected minor effect on the cell-cycle histogram was observed at 2 nmol/L paclitaxel alone; Fig. 3B).
MPS1 inhibition reduces paclitaxel-induced mitotic delay and potentiates gross chromosome mis-segregation errors
The therapeutic effect of paclitaxel had long been attributed to the induction of a mitotic arrest (activating the SAC), resulting in cell death. Recent work, however, demonstrated that paclitaxel exerts its effect mainly by the induction of aneuploidy via a multipolar mitosis (46). In contrast, MPS1 inhibition leads to premature abrogation of the SAC and, as a consequence, detrimental aneuploidy (7).
We first assessed the abrogation of mitotic checkpoint by each drug, both individually and in combination using live-cell microscopy of HeLa cells. Incubation of cells with BOS172722 resulted in a dose-dependent decrease in the median length of time cells spent in mitosis (Fig. 3C). In order to measure the type and the magnitude of the chromosomal damage induced by combining BOS172722 and paclitaxel, we performed live-cell imaging of H2B-mCherry–transfected HeLa cells. We could thereby quantify chromosome segregation errors induced by the single agents and their combinations. As seen in Fig. 3D, paclitaxel or BOS172722 at low concentrations alone induce minor mitotic abnormalities. At higher concentrations of the individual drugs, the drug-inherent phenotypes were evident: paclitaxel induces mainly multipolar mitotic figures in contrast to BOS172722, which predominantly induced division with unaligned chromosomes. In the case of the combination of both drugs, the number of abnormal mitosis which mainly exhibit unaligned chromosomes is synergistically increased. Importantly, even at 1 nmol/L concentration of paclitaxel, where the highest synergy scores were observed in our screen of growth inhibition, it induces only low levels of aneuploidy. We confirmed this result in MDA-MB-231 cells using metaphase spreads (Supplementary Fig. S4A). Because both MPS1 inhibition and paclitaxel treatment cause chromosome alignment errors, we reasoned that the synergy between paclitaxel and BOS172722 may arise through increasing the amount of erroneous MT-kinetochore attachments. Indeed, using immunofluorescence-based assays, the amount of chromosome alignment errors increased following combination of the drugs, only additively, in all tested combinations, when mitotic exit was prevented with MG132 (Supplementary Fig. S4B and S4C).
Paclitaxel induces a weak mitotic checkpoint delay
Because we did not observe consistent synergy with eribulin in our initial screen, we then investigated whether a SAC which has been activated to the same extent by MT-depolymerizing or whether MT-stabilizing drugs shows a differential response to MPS1 inhibition. In order to achieve comparable checkpoint activation with MT depolymerizing and stabilizing agents, we treated HeLa cells with nocodazole or paclitaxel and determined the concentration at which maximal activation of the checkpoint was achieved as observed by a plateau in mitotic timing. Cells were then arrested for 16 hours with the respective MT poisons, different doses of BOS172722 added and the fraction of cells in mitosis measured. Of note, treatment with the respective concentrations of nocodazole or paclitaxel alone led to a very similar duration of the time cells remained arrested in mitosis (Fig. 4A). In both cases, the mitotic block was overcome by MPS1 inhibition but, markedly, in paclitaxel-arrested cells, this override was achieved with much lower concentrations of BOS172722 compared with nocodazole. A comparable rate of mitotic exit was achieved with 25 nmol/L BOS172722 in paclitaxel-arrested cells and 100 nmol/L BOS172722 (4 times the concentration) in nocodazole-arrested cells. We therefore considered that the maximal SAC induced by MT depolymerizing agents is comparably stronger than the analogous SAC induced by paclitaxel. In line with this hypothesis, when we analyzed the recruitment of BUB1 to the kinetochore by immunofluorescence, BUB1 levels were, on average, markedly lower in the cells arrested with paclitaxel, being on average 50% of the levels seen in nocodazole and with a much larger range (Fig. 4B). Taken together, these data suggest that the override of the weak paclitaxel-induced SAC with low concentrations of BOS172722 may explain the synergistic increase in cell death.
Simultaneous combination of BOS172722 with paclitaxel for 24 hours induces maximum synergy
In order to better delineate the time that is required for the combination of paclitaxel and BOS172722 to exert a synergistic effect, we incubated MDA-MB-231 cells with BOS172722 for distinct periods of time and analyzed the induced synergism with paclitaxel. We observed that incubation of both drugs for 12 hours had only a limited synergistic effect in MDA-MB-231 cells (Supplementary Fig. S4D). We then explored the synergy of paclitaxel and BOS172722 in association with the incubation time in wash-off long-term clonogenic assays in MDA-MB-231 cells. We found that the combination of 1 nmol/L paclitaxel with 10 nmol/L BOS172722 was as efficacious as higher concentrations of each individual drug (Fig. 4C). We were also interested in investigating whether sequential addition of the drugs had any benefit over simultaneous treatment on the degree of synergism seen. The addition of BOS172722 following paclitaxel treatment or the opposite had no superior effect over the simultaneous addition (Supplementary Fig. S4E). Therefore, simultaneous administration of drugs is potentially the most beneficial in clinical studies. Based on our data, we propose the following model (Fig. 4D): Treatment with paclitaxel induces mitotic arrest with unaligned chromosomes due to impaired MT dynamics and partial inactivation of the SAC due to the presence of kinetochore-MT attachments in some chromosomes. Cancer cells can escape mitotic arrest by mitotic slippage and/or following metabolism/excretion of the drug. Not all cells will have lethal levels of chromosomal abnormalities. Treatment with paclitaxel in combination with BOS172722 completely prevents chromosome alignment in cancer cells due to impaired MT dynamics and dramatically reduced time in mitosis. All cells exit mitosis with gross chromosomal abnormalities and are not viable.
In vivo pharmacodynamic activity of BOS172722
In pharmacodynamic (PD) experiments in vivo BOS172722 potently inhibits the SAC induced by paclitaxel in human tumor xenograft models of TNBC (MDA-MB-231), as measured by inhibition of the mitotic marker phosphorylated histone H3 (p-HH3) by immunofluorescence microscopy (Fig. 5A, right graph; 5B, bottom). We confirmed that this effect is mediated by MPS1 inhibition by demonstrating reduction of the mechanism-related proximal biomarker phosphorylated-KNL1 (p-KNL1) by IHC (Fig. 5A, left graph; 5B, top). KNL1 is a natural substrate of MPS1 and is phosphorylated upon initiation of SAC activation by MPS1 (47).
Having shown that BOS172722 abrogates SAC as measured by paclitaxel-induced p-HH3 and p-KNL1 inhibition, we investigated whether we could identify target-engagement biomarkers to measure the activity of both paclitaxel and BOS172722 simultaneously in the same samples. It is known that taxanes induce acetylation of tubulin due to tubulin polymerization (48), when at the same time MPS1 inhibition should not affect tubulin modification. Paclitaxel and BOS172722 were administrated simultaneously into HCT116 tumor bearing mice. Figure 5C shows paclitaxel-induced acetylation of tubulin at all time points whereas addition of BOS172722 had no effect on tubulin acetylation. In contrast, paclitaxel-induced histone H3 phosphorylation was significantly inhibited by BOS172722 at 2 and 6 hours. To optimize these assays in a high-throughput format, we developed quantitative electrochemiluminescence assays (MSD). The results confirmed the immunoblotting data, indicating a potential use of these biomarkers in the clinic (Fig. 5D).
Therapeutic activity of BOS172722 in in vivo TNBC models
Based on in vitro activity and PD data, human TNBC xenograft experiments in athymic mice were undertaken to evaluate the therapeutic activity of paclitaxel alone or in combination with BOS172722. We initially used an MDA-MB-468 orthotopic (mouse mammary fat pad) xenograft model. Combination of BOS172722 with paclitaxel gave significant tumor regressions and a clear benefit in comparison with paclitaxel alone (Fig. 6A). A study using a TNBC patient-derived xenograft (PDX) model also showed tumor regression and a significant benefit of combination treatment in comparison with paclitaxel alone (Fig. 6B).
We then performed in vivo studies using a TNBC model to simulate breast cancer metastases. Tail-vein–injected MDA-MB-231-luciferase-expressing TNBC cells in SCID mice give rise predominantly to lung metastases. On day 28, the flux expressed as a percentage of vehicle control was as follows: paclitaxel alone 18.8%; combination with BOS172722 at 30 mg/kg 4.8%, and 40 mg/kg 5.7%, confirming a significant benefit of combination treatment in both tumor growth and survival (up to day 63) at BOS172722 doses ≥ 30 mg/kg (Fig. 6C and D). Taken together, the data described above demonstrate that our selective MPS1 inhibitor BOS172722 in combination with paclitaxel synergistically induces increased cell death in TNBC cell lines in vitro and regression and/or reduced growth rate of human tumor xenografts in vivo compared with treatment with either agent alone.
We initially discovered MPS1 as a potential therapeutic target during a siRNA screening campaign where we showed that a subgroup of breast cancer cell lines with a deregulated PTEN tumor suppressor gene were susceptible to cell death upon MPS1 depletion (37). We further investigated these findings using BOS172722. Fifty cell lines, 25 PTEN-proficient and 25 PTEN-deficient from a variety of human cancer types, were tested upon treatment with BOS172722. We found a clear trend of sensitivity to BOS172722 of cell lines with PTEN deficiency, irrespective of the type of cancer. Although not statistically significant, patients with PTEN-deficient tumors may represent a target population for treatment with an MPS1 inhibitor. However, the strongest, statistically significant corollary of sensitivity to MPS1 inhibition was cell proliferation rate. Cells with shorter doubling times were more sensitive to death upon MPS1 inhibition, TNBC cell lines being the most sensitive. A novel indicator for sensitivity to MPS1 inhibition is the SAC activity. We found that cell lines with reduced SAC activity were more sensitive to BOS172722, suggesting that lower SAC activity requires reduced concentrations of the MPS1 inhibitor to abrogate mitosis, thus inducing detrimental aneuploidy in cancer cells.
Due to the moderate levels of tumor growth inhibition by BOS172722 in xenograft studies, we focused on combination studies with the standard-of-care agents in TNBC. We identified paclitaxel as a favorable combination agent for use with MPS1 inhibition as it exerts robust synergistic effects throughout our panel of TNBC cell lines. This combination has also been identified by others and clinical trials have been initialized (refs. 19, 49; NCT02366949). Importantly, we discovered that a reduced SAC checkpoint is easier to override with an MPS1 inhibitor.
The same MPS1 mechanism of action is observed in paclitaxel-treated human TNBC xenografts in vivo. Athymic mice carrying TNBC human tumor xenografts were treated with vehicle, paclitaxel alone at clinically relevant dose, or in combination with single-dose BOS172722. Immunofluorescence microscopy and IHC of tumor sections showed a significant reduction of phospho-histone H3 and reduction of phospho-KNL1. These data confirm the mechanistic contribution of MPS1 inhibition in vivo. In addition, we have suggested a novel target-engagement biomarker strategy to be able to measure simultaneously the effect of paclitaxel and BOS172722 in tumor biopsies. The therapeutic benefit of BOS172722 in combination with paclitaxel was demonstrated in three TNBC in vivo models: MDA-MB-468 orthotopically transplanted in mouse mammary fat pads, systemic metastatic MDA-MB-231, and in a TNBC PDX.
In summary, BOS172722 is a highly potent and selective, orally bioavailable MPS1 inhibitor with favorable PK. Robust efficacy was demonstrated at well-tolerated doses in combination with paclitaxel in multiple xenograft models of TNBC, including PDX. BOS172722 is now in phase I dose escalation clinical trials in combination with standard-of-care paclitaxel treatment (NCT03328494).
Disclosure of Potential Conflicts of Interest
S.J. Anderhub has ownership interest (including stock, patents, etc.) in Rewards to Inventors Scheme. G.W.-Y. Mak has ownership interest (including stock, patents, etc.) in Rewards to Inventors Scheme. M.D. Gurden has ownership interest (including stock, patents, etc.) in Patent and reward to inventors scheme for MPS1 inhibitor. A. Faisal is associate professor at Lahore University of Management Sciences and has ownership interest (including stock, patents, etc.) in Rewards to Inventors Scheme. H.L. Woodward, P. Innocenti, I.M. Westwood, S. Naud, H. Saville, R. Burke, and R.L.M. Van Montfort have ownership interest (including stock, patents, etc.) in Rewards for Inventors Scheme. J. Blagg is VP Drug Discovery for NeoPhore Ltd and Azeria Therapeutics and has ownership interest (including stock, patents, etc.) in NeoPhore Ltd and Azeria Therapeutics. S. Hoelder reports receiving a commercial research grant from the sixth element pioneer fund who also funded the research leading up to this manuscript and is a is a consultant/advisory board member for ICR and is part of the ICR reward for inventors scheme. S.A. Eccles has ownership interest (including stock, patents, etc.) in ICR Rewards to inventors. S. Linardopoulos has ownership interest (including stock, patents, etc.) in Rewards to Inventors Scheme. No potential conflicts of interest were disclosed by the other authors.
Conception and design: S.J. Anderhub, S.A. Eccles, S. Linardopoulos
Development of methodology: S.J. Anderhub, G.W.-Y. Mak, M.D. Gurden, A. Faisal, K. Drosopoulos, S.A. Eccles, S. Linardopoulos
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.J. Anderhub, G.W.-Y. Mak, M.D. Gurden, A. Faisal, K. Drosopoulos, K. Walsh, E. Theofani, S. Filosto, S. Linardopoulos
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.J. Anderhub, G.W.-Y. Mak, M.D. Gurden, A. Faisal, K. Drosopoulos, S.A. Eccles, S. Linardopoulos
Writing, review, and/or revision of the manuscript: S.J. Anderhub, G.W.-Y. Mak, M.D. Gurden, A. Faisal, K. Drosopoulos, S.A. Eccles, S. Linardopoulos
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H.L. Woodward, P. Innocenti, I.M. Westwood, S. Naud, A Hayes, H. Saville, R. Burke, R.L.M. van Montfort, F.I. Raynaud, J. Blagg, S. Hoelder, S.A. Eccles, S. Linardopoulos
Study supervision: S. Linardopoulos
Other (synthesis of key compound BOS172722): H.L. Woodward, P. Innocenti
This work was supported by Cancer Research UK (grant number C309/A11566). We also acknowledge the Cancer Research Technology Pioneer Fund and Sixth Element Capital for funding and NHS funding to the NIHR Biomedical Research Centre. S. Linardopoulos, M.D. Gurden, and K. Drosopoulos are also supported by Breast Cancer Now (grant ref: CTR-Q3).
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.