Purpose: PARP1/2 inhibitors are a class of anticancer agents that target tumor-specific defects in DNA repair. Here, we describe BMN 673, a novel, highly potent PARP1/2 inhibitor with favorable metabolic stability, oral bioavailability, and pharmacokinetic properties.
Experimental Design: Potency and selectivity of BMN 673 was determined by biochemical assays. Anticancer activity either as a single-agent or in combination with other antitumor agents was evaluated both in vitro and in xenograft cancer models.
Results: BMN 673 is a potent PARP1/2 inhibitor (PARP1 IC50 = 0.57 nmol/L), but it does not inhibit other enzymes that we have tested. BMN 673 exhibits selective antitumor cytotoxicity and elicits DNA repair biomarkers at much lower concentrations than earlier generation PARP1/2 inhibitors (such as olaparib, rucaparib, and veliparib). In vitro, BMN 673 selectively targeted tumor cells with BRCA1, BRCA2, or PTEN gene defects with 20- to more than 200-fold greater potency than existing PARP1/2 inhibitors. BMN 673 is readily orally bioavailable, with more than 40% absolute oral bioavailability in rats when dosed in carboxylmethyl cellulose. Oral administration of BMN 673 elicited remarkable antitumor activity in vivo; xenografted tumors that carry defects in DNA repair due to BRCA mutations or PTEN deficiency were profoundly sensitive to oral BMN 673 treatment at well-tolerated doses in mice. Synergistic or additive antitumor effects were also found when BMN 673 was combined with temozolomide, SN38, or platinum drugs.
Conclusion: BMN 673 is currently in early-phase clinical development and represents a promising PARP1/2 inhibitor with potentially advantageous features in its drug class. Clin Cancer Res; 19(18); 5003–15. ©2013 AACR.
PARP1/2 inhibitors are a class of anticancer drugs that target tumor-specific defects in DNA repair. Here, we describe a novel PARP1/2 inhibitor, BMN 673, which shares features such as tumor selectivity with existing inhibitors but has strikingly increased antitumor potency and markedly improved pharmacokinetic attributes. The anticancer activity and selectivity of existing PARP1/2 inhibitors has been proven in early proof-of-concept clinical trials where patient benefit has been seen with limited toxicity. However, research to define patient selection, scheduling, and whether these agents should be used in combination with other anticancer drugs is still ongoing. BMN 673 is already being assessed in clinical trials and represents an exciting new PARP1/2 inhibitor at a time when optimal clinical use of these agents is being established.
DNA is constantly exposed to a range of environmental and endogenous factors that result in DNA damage (1). The repair of the resultant DNA lesions is mediated by a variety of mechanisms, including base excision repair, mismatch repair, nucleotide excision repair, homologous recombination, nonhomologous end joining, and direct reversal (1). Cancer cells often display deficiencies in one or more of these DNA repair pathways and these DNA repair defects can render tumor cells more susceptible to pharmaceutical intervention of the remaining DNA repair pathways than normal cells (2).
PARP1 and the similar enzyme PARP2 play important roles in DNA repair (3). DNA strand breaks result in the recruitment and binding of PARP1/2 to DNA at the site of damage. DNA-bound PARP1/2 catalyzes the synthesis of poly(ADP-ribose) (PAR) onto a range of DNA-associated proteins that mediate DNA repair. PARP1 also undergoes auto-PARsylation, a molecular change that ultimately leads to its release from DNA. Small-molecule inhibitors of PARP1/2 represent a class of anticancer agents that exert their cytotoxic effect by modulating the PARsylation activity of PARP1/2 (4). Inhibition of PARP1/2 is synthetically lethal with loss-of-function of either the BRCA1 or BRCA2 tumor suppressor genes, an effect that enables tumor cells with BRCA gene defects to be selectively targeted with PARP1/2 inhibitors (5). It is believed that loss of BRCA1/2 gene function causes a deficiency in homologous recombination–mediated double-strand DNA break repair (DSBR), making these cells highly susceptible to DNA lesions caused by PARP inhibition (4, 6, 7). In the clinic, phase I and II studies have shown that the PARP1/2 inhibitor olaparib (AstraZeneca/KuDOS) can elicit significant and sustained antitumor responses as a single agent in patients with cancer with BRCA1- or BRCA2-mutant tumors, while still achieving a favorable toxicity profile (4, 8–10). Furthermore, in a phase II study in high-grade serous ovarian cancer, olaparib reduced the risk of recurrence when used as a maintenance therapy after chemotherapy (11).
PARP1/2 inhibitors have also been shown to sensitize tumor cells to cytotoxic drugs such as the alkylating agents, temozolomide and cyclophosphamide, and the topoisomerase I inhibitors, irinotecan and topotecan (12, 13). This characteristic forms the basis of potential combination therapies where PARP1/2 inhibitors could be used together with DNA-damaging anticancer agents to enhance the antitumoral response.
There are currently at least seven PARP inhibitors at various stages of clinical development (4). Here, we report the characteristics of a new, potent, and selective PARP1/2 inhibitor, BMN 673. BMN 673 exhibits many of the biochemical and cytotoxic profiles found with earlier generation PARP1/2 inhibitors such as olaparib (AZD2281, KU0059436; AstraZeneca/KuDOS), rucaparib (AG-014699, PF-01367338; Clovis/Pfizer), and veliparib (ABT-888; Abbott Laboratories). However, BMN 673 is able to achieve antitumor cell responses and elicit DNA repair biomarkers at much lower concentrations than these other PARP1/2 inhibitors, an effect commensurate with its enhanced biochemical potency. Moreover, the favorable metabolic stability, oral bioavailability, and pharmacokinetic properties of BMN 673 suggest that it is a useful addition to existing targeted agents in oncology.
BMN 673 potently and selectively inhibits PARP1/2
Through a medicinal chemistry approach (Wang and colleagues, manuscript in preparation), we designed a series of drug-like small molecules that were able to inhibit the catalytic activity of PARP1. One compound, LT-00628 (Fig. 1A), showed a PARP1 IC50 of 1.82 nmol/L. LT-00628 contains two chiral centers and comprises a racemate that in theory consists of four isomers: L/R, R/L, L/L, and R/R. Chiral separation of LT-00628 indicated that LT-00628 is primarily made of trans isomers (L/R and R/L) with negligible amount of cis isomers. The trans isomers were obtained with chiral purity of greater than 97%. The absolute stereochemistry of BMN 673 was confirmed by single-crystal X-ray diffraction analysis (unpublished data). We found one of the trans isomers, LT-00673, to be highly potent with average PARP1 IC50 of 0.57 nmol/L (Fig. 1A). The other trans isomer, LT-00674, was relatively inactive (IC50 against PARP1 >100 nmol/L). As a residual amount of LT-00673 remained in LT-00674 (up to 0.8%), it was possible that the weak activity shown by LT-00674 may have been caused by contamination with LT-00673. LT-00673 was later renamed as BMN 673 and its structure, (8S,9R)-5-fluoro-8-(4-fluorophenyl)-9-(1-methyl-1H-1,2,4-triazol-5-yl)-8,9-dihydro-2H-pyrido[4,3,2-de]phthalazin-3(7H)-one), is shown in Fig. 1A. In a side-by-side comparison, we found BMN 673 to be more potent than veliparib, rucaparib, and olaparib with IC50s of 4.7, 2.0, and 1.9 nmol/L, respectively (Table 1). Even though the cis isomers were undetectable in the LT-00628 racemate, we designed a synthetic route to make the cis isomers alone; analysis of either cis-isomer showed these to be rather inactive as PARP inhibitors (Supplementary Table S1).
|.||PARP1 enzyme inhibitionIC50, nmol/L .||Cellular PAR synthesisEC50, nmol/L .||Capan-1 cytotoxicityIC50, nmol/L .||Temozolomide potentiationGI50, nmol/L .|
|.||PARP1 enzyme inhibitionIC50, nmol/L .||Cellular PAR synthesisEC50, nmol/L .||Capan-1 cytotoxicityIC50, nmol/L .||Temozolomide potentiationGI50, nmol/L .|
NOTE: Activities of BMN 673, LT-00628, and three clinical PARP1/2 inhibitors veliparib, rucaparib, and olaparib were compared in four in vitro assays: (i) concentration for 50% inhibition in PARP1 enzyme assay (IC50); (ii) concentration for 50% inhibition in cellular PAR synthesis assay in LoVo cells (EC50); (iii) concentration for 50% Capan-1 cell survival reduction in single-agent cytotoxicity assay (IC50); and (iv) concentration that, when combined with 200 μmol/L of temozolomide, resulted in 50% growth inhibition of LoVo cells in temozolomide potentiation assay (GI50). Assay conditions are described in the Materials and Methods. Values are average data from three to four independent experiments.
The kinetic characteristics of BMN 673 binding to PARP1 were assessed using Biacore T200. BMN 673 bound to PARP1 with an on-rate of 3.68 × 105 (1/ms), an off-rate of 1.05 × 10−4 (1/s), and a dissociation constant (KD) of 2.90 × 10−10 mol/L (Fig. 1B, top). In contrast, veliparib under the same conditions displayed an on-rate of 1.74 × 106 (1/ms), an off-rate of 4.10 × 10−3 (1/s), and a KD of 2.39 × 10−9 mol/L (Fig. 1B, bottom), suggesting BMN 673 to have a dissociation rate that is nearly 40 times slower than that of veliparib.
Most PARP1 inhibitors are known to also inhibit the homologous enzyme PARP2 due to the sequence similarity of PARP1 and -2 catalytic domains (14). We found that BMN 673 inhibited PARP1 and -2 to a similar extent, with Ki of 1.20 and 0.85 nmol/L, respectively. PARP1 and -2 are nuclear enzymes that synthesize PAR chains on target proteins as a form of posttranslational modification. To assess the ability of BMN 673 to inhibit intracellular PARP activity, we exposed LoVo cells to hydrogen peroxide (H2O2) to induce PAR synthesis and examined the ability of BMN 673 to inhibit PAR formation. Under these conditions, BMN 673 inhibited intracellular PAR formation with an IC50 of 2.5 nmol/L and was modestly more potent than veliparib, rucaparib, and olaparib, which had cellular PAR formation IC50s of 5.9, 4.7, and 3.6 nmol/L, respectively (Table 1).
Several assays were used to examine the inhibition specificity of BMN 673. We first assessed the effect of BMN 673 on poly (ADP-ribose) glycohydrolase (PARG), a protein that is structurally related to PARP1/2. PARG degrades PAR modifications on proteins and in doing so can counter the effects of PARP1/2 signaling (15). Although BMN 673 profoundly inhibited PARP1/2, it had no effect on PARG activity at concentrations up to 1 μmol/L (data not shown). To identify other potential off-target activities, we screened BMN 673 against two commercially available protein panels, both from MDS Pharma: the Hit Profiling Screen Panel and the Adverse Reaction Enzyme Panel. At a 10 μmol/L concentration, BMN 673 did not have significant interaction, either inhibitory or stimulatory, with any of the receptors, ion channels, or enzymes assayed (data not shown). The effect of BMN 673 on the potassium ion channel hERG (the human Ether-à-go-go-Related Gene protein) was determined in vitro, with no significant inhibition observed at BMN 673 concentrations as high as 100 μmol/L, suggesting that BMN 673 is unlikely to cause clinical cardiac QTc elongation.
Identification of genetic determinants of BMN 673 sensitivity
Previous work has shown that PARP1/2 inhibitors selectively inhibit tumor cells with genetic defects that abrogate homologous recombination–mediated DSBR (6, 7, 16). To rapidly identify genetic determinants of BMN 673 sensitivity in an unbiased fashion, we conducted a series of parallel RNA interference drug sensitization screens (17) and compared the genetic sensitization profile of BMN 673 (i.e., the list of genes that modulated BMN 673 response) with those for three earlier generation clinical PARP1/2 inhibitors–olaparib, rucaparib, and veliparib. To do this, we used a siRNA library targeting 960 genes, encompassing kinases and kinase-related genes (17) as well as a series of tumor suppressors and DNA repair proteins (Supplementary Table S2). We used a moderately PARP inhibitor-resistant breast tumor cell line (CAL51) previously used in similar studies (17) and screened each drug at a concentration required to elicit a 20% reduction in cell survival (surviving fraction 80, SF80) so as to maximize the potential for identifying PARP1/2 inhibitor sensitization effects (17, 18). We did note that for BMN 673, SF80 in CAL51 was achieved at 12.5 nmol/L, compared with the micromolar concentrations of veliparib, rucaparib, or olaparib required to reach this level of cell inhibition. The effect of each siRNA on drug sensitization was quantified by the calculation of a drug effect Z score, with siRNAs returning drug effect Z scores of <−2 being considered significant sensitization effects (17).
The siRNAs that significantly sensitized tumor cells to BMN 673 are shown in Supplementary Table S3. This analysis revealed that the most profound effects on BMN 673 sensitivity were caused by siRNAs targeting genes involved in homologous recombination/DSBR including BRCA2, BRCA1, SHFM1 (aka DSS1), PNKP, PALB2, ATM, ATR, CHEK1, FANCM, FANCA, etc. (Fig. 2A and Supplementary Table S3 genes highlighted in bold text). This observation suggested that homologous recombination/DSBR defects caused by any one of a number of genes caused cellular sensitivity to BMN 673, as is the case for other PARP1/2 inhibitors. To assess whether the overall genetic sensitization profile for BMN 673 was different from that of earlier generation PARP1/2 inhibitors, we compared the drug effect Z scores for all 960 genes from the BMN 673 screen to those derived from the other PARP1/2 inhibitor screens. The genetic sensitization profile for BMN 673 was not significantly different than the profiles generated by olaparib, rucaparib, or veliparib (Supplementary Table S4). These findings, collected in a relatively unbiased fashion, suggested that BMN 673 had the potential to target cells with defects in any one of a number of homologous recombination/DSBR genes.
BMN 673 targets tumor cells with defects in homologous recombination
Although siRNA screens are a rapid and useful means of identifying multiple genetic determinants of drug sensitivity in an unbiased fashion, and in this case confirmed that a number of different homologous recombination/DSBR genes modulated the response to BMN 673, the variable extent and stability of gene silencing achieved by siRNA often limits their ability to accurately define the scale of sensitization caused by a particular gene–drug combination. To formally assess the scale of selectivity of BMN 673 for tumor cells with homologous recombination gene defects, we measured the ability of BMN 673 to inhibit cells with either BRCA1 or BRCA2 gene mutations. We first tested the inhibitory effects of BMN 673 and other clinical PARP1/2 inhibitors in a panel of human tumor cell lines (Table 2). Models such as SW620 and MDA-MB-231 that do not have BRCA gene mutations or homologous recombination/DSBR defects were relatively resistant to BMN 673 with SF50s of 0.13 and 1.85 μmol/L, respectively (Table 2). Likewise, nontransformed MRC-5 cells were also resistant to BMN 673. In contrast, tumor models that were either BRCA1-deficient (MX-1 and SUM149) or BRCA2-deficient (Capan-1) were profoundly sensitive to BMN 673 (Table 2). PTEN deficiency (for example caused by N-terminal PTEN truncating mutations) has previously been shown to cause a defect in homologous recombination and sensitivity to other PARP1/2 inhibitors (19). We found BMN 673 to be highly potent in inhibiting human tumor cells with PTEN deficiency (Table 2). For example, the SF50 values for BMN 673 in the PTEN-null models MDA-MB-468, LNCap, and PC-3 were 6, 3, and 4 nmol/L, respectively, values comparable with SF50s in BRCA deficient models.
|.||SF50, μmol/L .|
|.||MX-1 (BRCA1 deficient) .||SUM149 (BRCA1 deficient) .||Capan-1 (BRCA2 deficient) .||MB-468 (PTEN deficient) .||LNCap (PTEN deficient) .||PC-3 (PTEN deficient) .||SW620 .||MDA-MB-231 .||MRC-5 (normal) .|
|.||SF50, μmol/L .|
|.||MX-1 (BRCA1 deficient) .||SUM149 (BRCA1 deficient) .||Capan-1 (BRCA2 deficient) .||MB-468 (PTEN deficient) .||LNCap (PTEN deficient) .||PC-3 (PTEN deficient) .||SW620 .||MDA-MB-231 .||MRC-5 (normal) .|
NOTE: Cultured human tumor cell lines were treated with veliparib, rucaparib, olaparib, or BMN 673, and tumor-killing effects were assessed either by colony formation assays (SUM149) or by two-dimensional cytotoxicity assays (all other cell lines). Survival curves were plotted and the IC50 was calculated as the concentration required to kill 50% of the cells. Known deficiency of homologous recombination DNA repair genes is indicated in parenthesis. Cells with BRCA1/BRCA2 mutations or PTEN deficiency were more sensitive to all PARP inhibitors than cells lacking these mutations.
Abbreviation: ND, not determined.
It was notable that compared with other clinical PARP1/2 inhibitors, BMN 673 was at least 18-fold more potent in BRCA-deficient tumor cells. In the BRCA1-mutant, triple-negative [estrogen receptor (ER), progesterone receptor (PR), and HER2 negative) breast tumor cell line model SUM149, BMN 673 delivered a SF50 of 8 × 10−12 mol/L, 1 × 105–fold lower than that of veliparib (SF50 = 0.8 μmol/L) and 922 and 231 times lower than that of rucaparib and olaparib, respectively (Table 2). In contrast, the differences in PARP1/2 inhibitor SF50 between different PARP1/2 inhibitors in cells without homologous recombination/DSBR defects was significantly less (Table 2). To confirm the BRCA selectivity of BMN 673, we also assessed cell growth inhibition in isogenic models of BRCA deficiency, namely mouse embryonic stem cells carrying BRCA1 gene defects (20) as well as human DLD1 tumor cells carrying homozygous BRCA2 gene knockout (21). In both model systems, BMN 673 selectivity inhibited BRCA-deficient cells and delivered a therapeutic window between BRCA-proficient and -deficient models at much reduced concentrations when compared with veliparib, rucaparib, or olaparib (Fig. 2B and C).
Compared with other PARP1/2 inhibitors, BMN 673 is about 3- to 8-fold more potent at inhibiting PARP1/2 enzymatic activities, but has a much greater potency advantage in inhibiting BRCA-deficient cells when used as a single agent (Table 2 and Fig. 2). This raised the possibility that the ability of BMN 673 to inhibit these cells might be partially due to activities that are unrelated to PARP1/2 inhibition. To investigate this possibility, we compared the in vitro activities of BMN 673 and its trans isomer LT-00674, which, despite the structural similarity, had a greatly reduced PARP1/2 inhibitory activity when compared with BMN 673. There was a strikingly tight correlation between PARP1 enzyme inhibition and the ability to inhibit the BRCA2-deficient cancer cell line Capan-1 as well as the ability to sensitize tumor cells to temozolomide, another well-established property of PARP1/2 inhibitors (Supplementary Table S5). The correlation of the chiral selectivity strongly suggested that the potent inhibition of cancer cells is a direct effect of BMN 673 PARP inhibition.
BMN 673 induces DNA damage at picomolar concentrations
One working hypothesis that could explain the homologous recombination-selectivity of PARP1/2 inhibitors centers on their ability to cause a DNA lesion or lesions that inhibit the normal function of the replication fork (22). The frequency of stalled and damaged replication forks caused by PARP1/2 inhibitors can be monitored by estimating the formation of nuclear γ-H2AX foci (6, 23). We assessed the ability of BMN 673 to induce nuclear γ-H2AX foci formation, as measured by immunofluorescence and confocal microscopic imaging. We found that BMN 673 induced nuclear γ-H2AX foci at concentrations as low as 100 pmol/L (Supplementary Fig. S1). In contrast, 100 nmol/L of olaparib was required to elicit a similar γ-H2AX response, suggesting that the increased intrinsic potency of BMN 673 also resulted in an increased ability to elicit a DNA response biomarker.
Metabolism and pharmacokinetic properties of BMN 673
One of the objectives of our PARP inhibitor discovery program was to improve metabolic stability, pharmacokinetic properties, and oral bioavailability over existing PARP1/2 inhibitors. In vitro metabolism studies of BMN 673 in liver microsomes from rats, dogs, and humans showed that BMN 673 had excellent liver microsome stability; more than 90% of BMN 673 remained after 2 hours of incubation at 1 μmol/L concentration (Supplementary Table S6). In rat pharmacokinetic studies, BMN 673 showed oral bioavailability of more than 40% when dosed in 0.5% carboxylmethyl cellulose, and pharmacokinetic properties that would predict a human half-life that is sufficient to support a regimen of once daily administration (manuscript in preparation). In vitro studies assessing the potential for inhibition of human cytochrome P450 enzymes (CYP450s) showed that at concentrations up to 10 μmol/L, BMN 673 did not inhibit any of the five major human hepatic CYP450 enzymes (CYP1A2, 2C9, 2C19, 2D6, and 3A4; data not shown). Overall, BMN 673 showed excellent metabolic stability, oral bioavailability, and pharmacokinetic properties.
Antitumor effect of BMN 673 oral administration in xenograft tumor models
To assess the in vivo antitumor effects of BMN 673, when used as a single agent, we treated nude mice bearing established subcutaneous MX-1 tumor xenografts with BMN 673. MX-1 is a human mammary carcinoma cell line that harbors BRCA1 deletion events and is BRCA1 deficient (24). Oral administration of BMN 673 for 28 days (once a day dose of 0.33 mg/kg) significantly inhibited the growth of MX-1 xenografts in mice, with 4 of 6 mice achieving a complete response (CR; tumor impalpable; Fig. 3A). At the lower dose of 0.1 mg/kg, oral BMN 673 only had a small effect on tumor growth after extended treatment (>21 days; Fig. 3A), but was still more effective than olaparib dosed orally at 100 mg/kg once daily. BMN 673 at these doses (0.33 and 0.1 mg/kg) was well tolerated, with no animal lethality or significant weight loss observed after 28 consecutive, once daily, oral doses.
To assess the in vivo pharmacodynamics of BMN 673 (PARP inhibition), we gave MX-1 xenograft-bearing mice a single oral administration of 1 mg/kg BMN 673. Tumors were harvested at 2, 8, and 24 hours postdrug dosing, and intratumor PAR levels were determined using an anti-PAR ELISA. BMN 673 treatment drastically decreased intratumoral PAR levels at 2 and 8 hours following drug administration. Partial recovery of basal PAR levels was observed at 24 hours after dosing (Fig. 3B), an effect probably due to the clearance of BMN 673. In comparison, a single oral administration of olaparib at 100 mg/kg produced a significant decrease of intratumoral PAR level at 2-hour postdosing, with partial recovery at 8 hours and complete recovery at 24 hours.
Subsequent studies that were designed to assess various intermittent dosing schedules of BMN 673 showed little benefit in terms of tumor growth delay (data not shown), suggesting that continuous suppression of PARP might be required for a therapeutic effect. As PAR levels partially recovered in mouse xenograft tumors between 8 and 24 hours postadministration of BMN 673 (Fig. 3B), we reasoned that twice a day dosing might be more effective than a once a day dosing regimen. To test this hypothesis, we compared the antitumor effect of administration of BMN 673 at 0.33 mg/kg/dose once a day for 28 days versus 0.165 mg/kg/dose twice a day for 28 days in MX-1 tumor-bearing mice. The results of these studies indicated that both once a day and twice a day dosing regimens inhibited MX-1 tumor growth with significant tumor regression (Fig. 3C). Once a day treatment generated four CRs out of 6 animals, consistent with the previous experiment. However, while tumors in the once a day–treated cohort did eventually reestablish after BMN 673 treatment ceased, the twice a day for 28 days treatment schedule resulted in 6 of 6 CRs with no reestablishment of tumor until the end of the study, 8 weeks after BMN 673 dosing ceased (Fig. 3C). One of six mice treated with twice a day had significant weight loss (>20%) and was sacrificed on day 53. All other animals in the study tolerated the treatment well. Taken together with the results from the in vivo PAR inhibition study (Fig. 3B), these results suggested that continuous suppression of PARP1/2 is required for maximum antitumor effect in the context of a single-agent application, and that in mice, twice a day dosing of BMN 673 was necessary to achieve optimal therapeutic effect, perhaps by continuously suppressing PARP activity. It should be noted that the half-life of BMN 673 in human is expected to be much longer (due to slower metabolism) than in mice. Therefore, we anticipate that once a day dosing will be sufficient in human patients to continuously suppress PARP and show anticancer effect. The effectiveness of once a day oral dosing in human has been validated in phase I clinical trials of BMN 673 (data to be published in ASCO 2013, Chicago).
The antitumor effect of BMN 673 on growth of PTEN-null tumors was also examined in vivo (Supplementary Fig. S2). Two PTEN-null cell lines were established as subcutaneous xenograft models in nude mice, and treated with 0.33 mg/kg BMN 673 (oral, once a day for 28 days). Both tumor models responded well to BMN 673 treatment, resulting in tumor growth delay of 15.9 days (MDA-MB-468) and 22.8 days (LNCap) when compared with the vehicle-treated control cohort. In a separate study, we showed that treatment with BMN 673 at 0.165 mg/kg/dose twice a day for 28 days was slightly more effective than 0.33 mg/kg/dose once a day for 28 days in the MDA-MB-468 xenograft model, consistent with the MX-1 tumor study results described earlier (Supplementary Fig. S3).
BMN 673 sensitizes tumor cells to DNA-damaging chemotherapies
The chemosensitizing property of PARP1/2 inhibitors has been well documented (25). One of the most robust chemopotentiation effects involves the combination of temozolomide with PARP1/2 inhibitors (26–29). To investigate whether BMN 673 shared this characteristic, we tested its ability to potentiate the cytotoxic effect of temozolomide using an in vitro assay. We found that single-agent temozolomide exposure (200 μmol/L) resulted in approximately 15% cell growth inhibition after a 5-day treatment in LoVo tumor cells (Fig. 4A). Combining BMN 673 with 200 μmol/L temozolomide resulted in a significant potentiation of temozolomide cytotoxicity (Fig. 4A and Table 1). We also examined the effect of BMN 673 on SN-38, active metabolite of the DNA topoisomerase I inhibitor irinotecan. BMN 673 potentiated the cytotoxic effect of SN38 in MX-1 cells in vitro in a dose-dependent manner (Fig. 4B).
We observed a clearly additive effect of combining BMN 673 with platinum drugs in vitro (data not shown). The ability of BMN 673 to potentiate platinum drugs in vivo was also readily shown. We first examined the effect of BMN 673 on cisplatin-induced antitumor effect. Growth of MX-1 tumor xenografts implanted subcutaneously in female athymic mice was significantly inhibited in a dose-dependent manner when animals were treated with BMN 673 in combination with a 6 mg/kg intraperitoneal injection of cisplatin (Fig. 4C). However, we did note that the combination regimens involving cisplatin resulted in moderate body weight loss. Maximum average weight loss of 11%, 6%, 5%, and 3% were observed for groups that contained BMN 673 doses of 1, 0.33, 0.1, and 0.033 mg/kg, respectively. In the same experiment, animals that were treated with cisplatin alone had maximum average body weight loss of 3%. One animal died in the BMN 673 1 mg/kg plus cisplatin group on day 20, whereas most animals recovered their body weight after treatment terminated. Olaparib at 100 mg/kg also showed activity in this treatment regimen, with a maximum body weight loss of 3%. In a separate study, BMN 673 showed significant potentiation of carboplatin antitumor effect in vivo (Fig. 4D). No animal lethality or significant body weight loss was observed.
Here, we describe a novel, potent, and selective PARP1/2 inhibitor BMN 673. Like other PARP1/2 inhibitors, BMN 673 is selective for tumor cells with defects in homologous recombination, as shown by a siRNA drug sensitivity profile that encompasses genes involved in homologous recombination/DSBR and by assessment in isogenic and nonisogenic models of BRCA1, BRCA2, and PTEN deficiency. These effects were also observed in vivo in mouse xenograft models where tumors with either BRCA1 or PTEN defects showed significant tumor growth delay after oral administration of BMN 673. Furthermore, the antitumor effects of temozolomide, cisplatin, and carboplatin were all potentiated by BMN 673. These therapeutic effects were achieved with tolerable toxicity, evidence of PAR inhibition in vivo in animal tumor models, and favorable pharmacokinetic properties that allow once a day oral dosing in human patients.
With Ki of 1.2 and 0.9 nmol/L against PARP1 and -2, respectively, BMN 673 is the most potent PARP inhibitor reported to date. Still, we were initially surprised by the much greater cytotoxicity in homologous recombination–deficient cells compared with other PARP1/2 inhibitors that have apparently comparable potency against PARP catalytic activity. We initially suspected that BMN 673 might have activity other than PARP1 and -2 inhibition, and that this activity may be responsible for or contributing to the increase in tumor cell inhibition. However, chiral selectivity (Supplementary Table S5) shown by the BMN 673/LT-00674 isomer pair strongly suggests that its remarkable cytotoxic properties in homologous recombination/DSBR–deficient cells is very likely a direct result of its ability to inhibit PARP1/2.
BMN 673 showed a notable preference for tumor cells harboring BRCA1, BRCA2, or PTEN dysfunction. Although other PARP inhibitors such as veliparib, rucaparib, and olaparib showed selectivity in homologous recombination–deficient cells versus homologous recombination–proficient cells when used at micromolar concentrations, the selectivity margin of BMN 673 could be achieved with nanomolar or even picomolar concentrations in isogenic models. In some nonisogenic systems (Table 2), the therapeutic window between BRCA-deficient and -proficient models was enhanced compared with other clinical PARP1/2 inhibitors (Fig. 2). The siRNA screen also confirmed that, out of 960 genes tested, the most profound effects on BMN 673 sensitivity were caused by siRNAs targeting genes involved in homologous recombination/DSBR. Although a two-sided t test did not show statistically significant differences between BMN 673 and other PARP inhibitors in their sensitization profiles (Supplementary Table S4), it is intriguing to note that silencing of DSBR genes BRCA1, BRCA2, PNKP, ATR, CHEK1, and PALB2 resulted in sensitization for BMN 673 (Supplementary Table S3) at a low nmol/L dose of the drug. This may generate a large therapeutic window for BMN 673, making it potentially more efficacious while maintaining manageable toxicity.
PARP1/2 inhibitors are an exciting new class of anticancer agents that exploit the biologic concept of synthetic lethality as the basis for their antitumor selectivity (6, 7). However, the clinical use of PARP1/2 inhibitors is still in its infancy. Despite promising phase I and II trial results with olaparib, rucaparib, and niraparib (4, 8–11) further development has been slow with problems such as toxicity when combined with chemotherapeutic agents (30, 31), difficulty in defining suitable patient populations beyond those with BRCA1 or BRCA2 mutations (32), or formulation issues. However, given the effectiveness of PARP1/2 inhibitors in patients with germline BRCA mutations (8–10), this area is still ripe for further study and exploration. Two phase I clinical trials are currently ongoing for BMN 673 that are assessing its safety, pharmacokinetic, and pharmacodynamic properties and preliminary efficacy in human patients. In January 2011, a first in human, single-arm, open-label study in patients with advanced tumors with DNA-repair pathway deficiencies was initiated followed by a two-arm, open-label study in patients with advanced hematologic malignancies, which was started in June 2011. The discovery and characterization of BMN 673, as a potent, selective, orally bioavailable PARP1/2 inhibitor and its advancement into phase I studies thus provides a welcome addition to this field.
Materials and Methods
Drugs and cell lines
Synthesis of LT-00628, BMN 673, LT-00674, and LT-00878 is described elsewhere (33, 34). Olaparib, rucaparib, and veliparib used in this study were either synthesized as previously described (35–37) or obtained from Selleck Chemicals. Drug stock solutions were prepared in dimethyl sulfoxide (DMSO) and stored in aliquots at −20°C. Drugs (alone or in combination) were added to cell cultures so that final DMSO concentrations were constant at 1% (v/v). MX-1 was obtained from National Cancer Institute (Bethesda, MD). All other cell lines were obtained from American Type Culture Collection and maintained as exponentially growing monolayers according to the supplier's instructions.
PARP enzyme assays
The ability of a test compound to inhibit PARP1 enzyme activity was assessed using Trevigen's PARP Assay Kit (Trevigen CAT#4676-096-K) following the manufacturer's instruction. IC50 values were calculated using GraphPad Prism5 software. For PARP inhibitor Ki determination, enzyme assays were conducted in 96-well FlashPlate (PerkinElmer) with 0.5 U PARP1 enzyme (Trevigen; Cat#4667-250-EB), 0.25× activated DNA (Trevigen), 0.2 μCi [3H] NAD (PerkinElmer; Cat#NET443H250UC), and 5 μmol/L cold NAD (Sigma) in a final volume of 50 μL reaction buffer containing 10% glycerol (v/v), 25 mmol/L HEPES, 12.5 mmol/L MgCl2, 50 mmol/L KCl, 1 mmol/L dithiothreitol (DTT), and 0.01% NP-40 (v/v), pH 7.6. Reactions were initiated by adding NAD to the PARP reaction mixture with or without inhibitors and incubated for 1 minute at room temperature. Fifty microliter of ice-cold 20% trichloroacetic acid (TCA) was then added to each well to stop the reaction. The plate was sealed and shaken for a further 120 minutes at room temperature, followed by centrifugation. Radioactive signal bound to the FlashPlate was determined using TopCount. PARP1 Km was determined using Michaelis–Menten equation from various substrate concentrations (1–100 μmol/L NAD). Compound Ki was calculated from enzyme inhibition curve according to the formula: Ki = IC50/[1+(substrate)/Km]. Km for PARP2 enzyme and compound Ki were determined with the same assay protocol except 30 ng PARP2 (BPS Bioscience; Cat#80502), 0.25× activated DNA, 0.2 μCi [3H] NAD, and 20 μmol/L cold NAD were used in the reaction for 30 minutes at room temperature.
Recombinant human PARP1 (rhPARP1) catalytic domain (residues 662–1011) with N-terminal 6× His-tag was generated in house at BioMarin Pharmaceutical Inc. and used in binding assay for PARP inhibitor interaction using Biacore T200 (GE Healthcare). rhPARP1 was immobilized on a CM5 sensor chip (GE Healthcare) by amine coupling method. Briefly, one flow cell of a CM5 chip was first activated by a 7-minute injection at 10 μL/min of freshly prepared 50 mmol/L N-hydroxysuccinimide (NHS): 200 mmol/L 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) (1:1; GE Healthcare) at rate of 10 μL/min. Then rhPARP1 (100 μg/mL, in 10 mmol/L MES pH 6.5) was injected onto the flow cell for 60-second at 10 μL/min. The remaining active coupling sites were blocked with a 7-minute injection of 1 mol/L ethanolamine at 10 μL/min. The immobilization buffer contains 10 mmol/L HEPES pH 7.4, 150 mmol/L NaCl, 0.05% Surfectant P20, 5 mmol/L MgCl2, and 0.5 mmol/L TCEP [tris(2-carboxyethyl)phosphine]. The immobilization level was approximately 7,600 response unit (RU). For binding kinetics measurement, PARP inhibitors at increasing concentrations (12.5, 25, 50, 100, and 200 nmol/L) were injected over the chip surface for 60 seconds per injection. The exposure was followed by a dissociation phase of 3,600 seconds in running buffer (immobilization buffer + 1% DMSO) after the last injection. The flow rate was 50 μL/min. After sensorgrams were corrected for signals from a reference flow, kinetics was calculated with Biacore T200 evaluation software ver.1.0 (Biacore; GE Healthcare).
Intracellular PAR formation assay
Cellular PAR synthesis assay assesses the ability of a test compound to inhibit polymerization of PAR. LoVo human colorectal tumor cells grown in 96-well microtiter plates overnight were pretreated with increasing concentrations of PARP inhibitors for 30 minutes before H2O2 was added at a final concentration of 50 mmol/L. After a 5-minute treatment at room temperature, cells were fixed for 10 minutes with prechilled methanol/acetone (7:3) at −20°C. Fixed cells were incubated with anti-PAR monoclonal antibody (Trevigen) for 60 minutes, followed by incubation with fluorescein isothiocyanate (FITC)-coupled goat anti-mouse immunoglobulin G (IgG; diluted at 1:100) and 1 μg/mL 4′,6-diamidino-2-phenylindole (DAPI) for 60 minutes. FITC signal was normalized with DAPI signal, and EC50 values were calculated using GraphPad Prism.
CAL51 cells plated in 96-well plates were transfected 24 hours later with siRNA (final concentration 100 nmol/L), using Oligofectamine (Invitrogen) as per the manufacturer's instructions. At 48 hours after transfection, three replica plates were treated with 0.01% (v/v) DMSO vehicle in media and three replica plates with each PARP1/2 inhibitor in media. Cell viability was assessed after 5 days of PARP1/2 inhibitor exposure using CellTiter-Glo Luminescent Cell Viability Assay (Promega) as per the manufacturer's instructions. Data from each screen were analyzed as described previously (18).
Colony formation survival assays
Colony formation assays were conducted as described previously (6). In brief, cells were seeded into 6-well plates at a concentration of 500 to 2,000 cells per well. After 24 hours, media was replaced with fresh media containing PARP1/2 inhibitor. This procedure was repeated twice weekly for 14 days, at which point colonies were fixed with TCA and stained with sulforhodamine B. Colonies were counted and surviving fractions calculated by normalizing colony counts to colony numbers in vehicle-treated wells. Survival curves were plotted using a four-parameter logistic regression curve fit.
Single-agent cytotoxicity and chemosensitization assays
In single-agent assays, cells are seeded in a density that allows linear growth for 10 to 12 days in 96-well plates (typically 500–3,000 cells/well). Cells were treated in their recommended growth media containing increasing concentrations of PARP inhibitors for 10 to 12 consecutive days (media changed with fresh compounds every 5 days). In chemosensitization assays, PARP inhibitors at various concentrations were either combined with 200 μmol/L temozolomide to treat LoVo cells or with SN-38 (0–10 nmol/L), an active metabolite of Irinotecan to treat MX-1 cells for 5 days. After treatments, cell survival was determined by CellTiter-Glo assay (Promega), expressed as relative to mock treatment control (0.1%–0.5% DMSO), and IC50 or GI50 values were calculated using GraphPad Prism5 software.
Cells were seeded on coverslips placed in 6-well plates and after 24 hours treated with several concentrations of olaparib or BMN 673. Twenty-four hours after treatment, the cells were fixed in 10% formalin (3.7% PFA) for 1 hour. Cells were permeabilized with 0.2% Triton X-100 in PBS for 20 minutes, treated with 50 μL DNase I (Roche; diluted at 1:10 in PBS) for 1 hour at 37°C and then blocked with IFF (PBS + 1% BSA and 2% FBS followed by filter sterilization) for 1 hour. The coverslips were then incubated with rabbit anti-γ-H2Ax primary (Millipore) and mouse anti-RAD51 primary (Epitomics Lot YI031608C; both 1:1,000 in 50 μL immunofluorescent buffer (IFF)) overnight at 4°C. Next day, the cells were incubated with anti-mouse Alexa Fluor 546 secondary and anti-rabbit Alexa Fluor 488 secondary (both 1:1,000 in 50 μL IFF) for 1 hour. Cells were then washed in PBS containing DAPI 1:10,000 for 10 minutes and attached on glass plates using Vectashield and nail polish. A minimum of four pictures were made of each coverslip using the Leica confocal microscope, and cells were subsequently counted. At least 100 cells were assessed per coverslip, being positive for γ-H2Ax if they had more than 5 foci per nucleus. The percentage of positive cells was plotted.
Female athymic nu/nu mice (8–10-week old) were used for all in vivo xenograft studies. Mice were quarantined for at least 1 week before experimental manipulation. Exponentially growing cells (LNcap and MDA-MB-468) or in vivo passaged tumor fragments (MX-1) were implanted subcutaneously at the right flank of nude mice. When tumors reached an average volume of approximately 150 mm3, mice were randomized into various treatment groups (6–8 mice/group) in each study. Mice were visually observed daily and tumors were measured twice weekly by calliper to determine tumor volume using the formula [length/2] × [width2]. Group median tumor volume (mm3) was graphed over time to monitor tumor growth. In single-agent studies, olaparib (100 mg/kg), BMN 673 (various doses as indicated), or vehicle (10% DMAc, 6% Solutol, and 84% PBS) was administered by oral gavage (per os), once daily or BMN 673 (0.165 mg/kg) twice daily for 28 consecutive days. Mice were continuously monitored for 10 more days after last day of dosing. In cisplatin combination study, BMN 673, olaparib, or vehicle was administered per os once daily for 8 days starting on day 1. Cisplatin at a dosage of 6 mg/kg or its vehicle (saline) was administered intraperitoneally as a single injection on day 3, 30 minutes after PARP inhibitor was administered. Combination with carboplatin was conducted in a similar way in MX-1 model in which BMN 673 was administered per os once daily for either 8 days or 5 days and carboplatin was injected intraperitoneally at single dose of 35 mg/kg, 30 minutes after BMN 673 on day 3.
PAR assay in vivo
MX-1 tumor xenografts were prepared as described in Materials and Methods. When tumors reached an average volume of approximately 150 mm3, olaparib (100 mg/kg), BMN 673 (1 mg/kg), or vehicle was administered in a single per os dosing. Tumors were harvested at 2, 8, and 24 hours after drug dosing, snap-frozen in liquid nitrogen. Tumor tissue was then homogenized in PBS on ice and extracted with lysis buffer (25 mmol/L Tris pH 8.0, 150 mmol/L NaCl, 5 mmol/L EDTA, 2 mmol/L EGTA, 25 mmol/L NaF, 2 mmol/L Na3VO4, 1 mmol/L Pefabloc, 1% Triton X-100, and protease inhibitor cocktail) containing 1% SDS. Levels of PAR in the tumor lysates were determined by ELISA using PARP in vivo PD Assay II kit (Trevigen).
Disclosure of Potential Conflicts of Interest
L.E. Post, Y. Shen, Y. Feng and B. Wang have ownership interest (including patents) in BioMarin Pharmaceutical Inc. A. Ashworth and C.J. Lord may benefit financially from the development of PARP inhibitors through patents held jointly with AstraZeneca through the Institute of Cancer Research's rewards to inventors scheme. No potential conflicts of interest were disclosed by the other authors.
Conception and design: Y. Shen, F.L. Rehman, C.J. Lord, L.E. Post, A. Ashworth
Development of methodology: Y. Shen, Y. Feng, C.J. Lord
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F.L. Rehman, J. Boshuizen, R. Elliott, C.J. Lord, A. Ashworth
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Shen, F.L. Rehman, Y. Feng, C.J. Lord, L.E. Post, A. Ashworth
Writing, review, and/or revision of the manuscript: Y. Shen, F.L. Rehman, Y. Feng, C.J. Lord, L.E. Post, A. Ashworth
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Shen, I. Bajrami
Study supervision: Y. Shen, Y. Feng, C.J. Lord, A. Ashworth
Compound invention: B. Wang
The authors thank Drs. Daniel Chu and Peter Myers for their excellent scientific input and continuous support throughout this work; Shanghai HD Biosciences, Inc. for developing and conducting in vitro and in vivo assays, Shanghai ChemPartner for chemistry work and BiaCore kinetics work; Drs. Mika Aoyagi-Scharber and Paul Fitzpatrick for generating PARP1 protein for the BiaCore study; Dr. Myron Jacobson for excellent scientific discussion and helping with running the PARP2 assay; Dr. Bill Wold (Southern Research Institute, Birmingham, AL) for insightful discussion about animal xenograft models.
Part of this work was funded by Breakthrough Breast Cancer, Cancer Research UK, and the AACR as part of the SU2C Breast Cancer Dream Team initiative. F.L. Rehman is a Breakthrough Breast Cancer Avon Clinical Fellow and is funded as part of the Avon Breast Cancer Crusade. The authors also acknowledge NHS funding to the NIHR Royal Marsden Hospital Biomedical Research Centre.
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.