6368 J. Med. Chem. 2010, 53, 6368–6377 DOI: 10.1021/jm100394y
Discovery of a Potent, Selective, and Orally Bioavailable Pyridinyl-Pyrimidine Phthalazine Aurora Kinase Inhibitor )
)
)
)
)
)
)
Victor J. Cee,*,† Laurie B. Schenkel,† Brian L. Hodous,† Holly L. Deak,† Hanh N. Nguyen,† Philip R. Olivieri,† Karina Romero,† Annette Bak,‡ Xuhai Be,§ Steve Bellon, Tammy L. Bush,^ Alan C. Cheng, Grace Chung,^ Steve Coats,^ Patrick M. Eden,^ Kelly Hanestad,^ Paul L. Gallant,# Yan Gu, Xin Huang, Richard L. Kendall,^ Min-Hwa Jasmine Lin,§ Michael J. Morrison,# Vinod F. Patel,† Robert Radinsky,^ Paul E. Rose, Sandra Ross,# Ji-Rong Sun,^ Jin Tang, Huilin Zhao, Marc Payton,^ and Stephanie D. Geuns-Meyer† )
Departments of †Medicinal Chemistry, ‡Pharmaceutics, §Pharmacokinetics and Drug Metabolism, Molecular Structure, ^Oncology Research, and #Molecular Pharmacology, Amgen Inc., 360 Binney Street, Cambridge, Massachusetts 02142 and Amgen Inc., One Amgen Center Drive, Thousand Oaks, California 91320 Received March 29, 2010
The discovery of aurora kinases as essential regulators of cell division has led to intense interest in identifying small molecule aurora kinase inhibitors for the potential treatment of cancer. A high-throughput screening effort identified pyridinyl-pyrimidine 6a as a moderately potent dual inhibitor of aurora kinases -A and -B. Optimization of this hit resulted in an anthranilamide lead (6j) that possessed improved enzyme and cellular activity and exhibited a high level of kinase selectivity. However, this anthranilamide and subsequent analogues suffered from a lack of oral bioavailability. Converting the internally hydrogen-bonded sixmembered pseudo-ring of the anthranilamide to a phthalazine (8a-b) led to a dramatic improvement in oral bioavailability (38-61%F) while maintaining the potency and selectivity characteristics of the anthranilamide series. In a COLO 205 tumor pharmacodynamic assay measuring phosphorylation of the aurora-B substrate histone H3 at serine 10 (p-histone H3), oral administration of 8b at 50 mg/kg demonstrated significant reduction in tumor p-histone H3 for at least 6 h. Introduction The aurora family of serine/threonine kinases is represented by three paralogous genes (aurora-A, -B, and -C).1 Aurora-A and -B are essential mitotic regulators required for chromosome alignment, segregation, and cytokinesis, while aurora-C function appears to be restricted mainly to spermatogenesis. Despite the high degree of structural similarity, aurora-A and -B have distinct subcellular localizations and functions during mitosis. As cells enter mitosis, aurora-A is localized to the centrosomes and the proximal mitotic spindle, whereas aurora-B is localized to the kinetochore/centromeric region. At the onset of anaphase, as sister-chromatids separate, aurora-B localizes to the spindle midzone where it coordinates exit from mitosis.1,2 Aurora-A can function as an oncogene capable of transforming rodent fibroblast cells in culture.1,3 Expression of aurora-A and -B is elevated in a variety of human cancers, with increased expression correlating with poor prognosis in many cases.1,4 In tumor cells, inhibition of aurora-B or dual inhibition of aurora-A and -B leads to premature exit from mitosis, resulting in undivided 4N DNA-containing cells in the G1-phase of the cell cycle. These cells can progress through further rounds of genome replication without cellular division, a process referred to as endoreduplication, which ultimately results in cell death.1 The described mechanism of action for an auroraA/B kinase inhibitor is distinct from that of traditional antimitotic agents in that death of tumor cells is driven by continued cell-cycle progression rather than by arrest in mitosis (e.g., paclitaxel and docetaxel). *To whom correspondence should be addressed. Phone: 805-3135500. Fax: 805-480-1337. E-mail:
[email protected].
pubs.acs.org/jmc
Published on Web 08/04/2010
A considerable research effort from both the pharmaceutical industry and academic laboratories has led to many reports of small molecule aurora kinase inhibitors.5 Both pan-aurora kinase inhibitors and subtype-selective inhibitors are currently under evaluation in human clinical trials. The earliest example of a pan-aurora kinase inhibitor in clinical development was MK-0457 (VX-680) (1), which inhibits aurora-A and -B as well as a number of other kinases implicated in cancer (Figure 1).5 The earliest examples of isoform-selective aurora kinase inhibitors in clinical development are MLN8054, an aurora-A selective inhibitor (2),6 and AZD-1152 (3a), a phosphate prodrug of the selective auroraB inhibitor 3b.7 Herein we report the discovery of a pyridinylpyrimidine series of pan-aurora kinase inhibitors and describe the optimization of this series, which culminated in phthalazine 8b, a potent and selective inhibitor of aurora-A and -B with properties suitable for oral dosing. Chemistry Pyridinyl-pyrimidine chloride derivative 48 was reacted with 4-aminophenol or a substituted 4-aminophenol in the presence of cesium carbonate at elevated temperature to produce penultimate aniline intermediates 5a-c (Scheme 1). These aniline intermediates (5a-c) were converted to amides by HATU-mediateda reactions with a variety of carboxylic a Abbreviations: ATP, adenosine triphosphate; Cl, clearance; HATU, O-(7-azabenzotriazol-1-yl)-N,N,N0 ,N0 -tetramethyluronium hexafluorophosphate; HTS, high throughput screening; KDR, kinase insert domain receptor; RLM, rat liver microsome; Tie-2, tyrosine kinase with extracellular epidermal growth factor homology domains 2; TPX2, targeting protein for Xenopus plus end-directed kinesin-like protein 2.
r 2010 American Chemical Society
Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 17
6369
Scheme 1. Preparation of Amide Targetsa
a
Reagents and conditions: (a) 4-aminophenol, Cs2CO3, DMSO, 130 C (90-95%); (b) ArCO2H, HATU, DIPEA, CHCl3, rt (33-84%).
Scheme 2. Preparation of Phthalazine Targetsa
Reagents and conditions: (a) 1,4-dichlorophthalazine, s-BuOH, 100 C (88%); (b) boronic acid, Pd(dppf)Cl2, aq Na2CO3, dioxane, 100 C (44-60%). a
Table 1. Potency and Selectivity Summary for HTS Hit 6a
enzyme IC50 (nM)a aurora-A/TPX2 586 aurora-B 66 Lck 385 Abl 503 Tie-2 630 IC50 > 8000 nM: EGFR, EphB4, IGFR-1, JNK2, p70s6, PLK1, Erk1, PKBR, CDK5.
Figure 1. Representative aurora kinase inhibitors and pyridinylpyrimidine phthalazine 8b.
acids to give target amides 6a-6l. Phthalazines 8a and 8b were also derived from intermediate 5a, which was treated with 1,4dichlorophthalazine to provide chloride 7. Suzuki reaction of 7 with phenyl- or 4-fluorophenylboronic acid provided 8a and 8b, respectively (Scheme 2). Discussion High throughput screening (HTS) of a kinase-focused proprietary compound collection identified pyridinyl pyrimidine amide 6a as a moderately potent inhibitor of both aurora-A/ TPX2 (aurora-A in complex with a fragment [1-43] of partner protein TPX29) and aurora-B enzymes, with a reasonable level of selectivity against a small panel of kinases (Table 1). In an aurora-B cellular assay (hereafter termed 24 h DNA ploidy)
c-kit 1350 c-fms 1580 KDR 3270 Src 5750 p38R 8000 Zap70, Jak2, Jak3, BTK,
a All kinase assays were run at the Km for ATP and represent an average of at least two determinations. For statistical analysis of averaged data, see the Supporting Information.
measuring the ability of a test compound to induce an accumulation of cells with g4N DNA content, compound 6a was not able to increase the number of HeLa cells with g4N DNA content relative to control in a 24 h time period at concentrations up to 1200 nM. Therefore, our initial goal was to improve the aurora cellular potency of the screening hit 6a. Table 2 illustrates selected structure-activity relationships for variations of the terminal aryl ring. A positional scan of a chlorine substituent revealed that substitution at the 3- and 4-position (6c and 6d) was superior to 2-substitution (6b), although cellular activity was still not achieved. It was hypothesized that the larger phenoxy substituent might improve inhibitory activity by increased van der Waals interactions between ligand and protein, so phenoxy-substituted analogues were also investigated. In contrast to chlorine substitution, the 2-substituted 6e showed the most potent activity
6370 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 17
Table 3. Potency Data for Pyridinyl-Pyrimidine Anthranilamides
Table 2. Potency Data for Pyridinyl-Pyrimidine Amides
enzyme, IC50 (nM)a 6a 6b 6c 6d 6e 6f 6g 6h
Cee et al.
cell, EC50 (nM)b
R
aurora-A/TPX2
aurora-B
DNA ploidy
3-CF3 2-Cl 3-Cl 4-Cl 2-OPh 3-OPh 4-OPh 2-NHPh
586 >25000 1546 2342 1544 964 ND 62
66 >25000 430 335 56 251 542 17
>1200c ND >1200c >1200c >1200c >1200c ND 89
a All kinase assays were run at the Km for ATP and represent an average of at least two determinations. b Average of at least two determinations, unless otherwise specified. c One determination. For statistical analysis of averaged data, see the Supporting Information.
toward aurora-B enzyme in this series, although cellular activity was still not observed. Replacement of the 2-OPh substituent with 2-NHPh produced anthranilamide (6h), which was an important breakthrough, as this compound exhibited double-digit nanomolar potency against both auroraA/TPX2 and aurora-B enzymes as well as an 89 nM EC50 value in the 24 h DNA ploidy cellular assay. Although 6e and 6h differ by only a single atom, the conformational differences imposed by intramolecular hydrogen bonding between the ether oxygen and amide NH of salicylamide 6e and aniline NH and carbonyl O of anthranilamide 6h are significant, and this may contribute to the difference in potency between 6e and 6h. The structure-activity relationship of the central paraaminophenoxy portion of the molecule is illustrated in Table 3. Moving the methyl group to the position adjacent to the aniline nitrogen (6i) caused a significant loss of aurora-A/ TPX2 and aurora-B enzyme potency as well as poor cellular activity. Removing the methyl group (6j) led to an improvement in both intrinsic potency and cellular activity. These data suggest that an unsubstituted central aminophenoxy ring provides optimal potency, and further efforts focused on this series. The selectivity profile of 6j against a panel of phylogenetically diverse kinases and the pharmacokinetic properties of 6j in Sprague-Dawley rats are presented in Table 4. Aside from aurora-A and -B, only Tie-2 and KDR enzymes were inhibited at submicromolar concentrations (IC50= 440 nM and 794 nM, respectively), resulting in a minimum selectivity factor of 29-fold over the EC50 of the 24 h DNA ploidy cellular assay (440 nM against Tie-2 enzyme vs 15 nM DNA ploidy). Unfortunately, 6j exhibited suboptimal rat pharmacokinetics, with clearance (Cl = 3.6 L/h/kg) approaching rat liver blood flow. The high clearance was consistent with rapid turnover of 6j in rat liver microsomes (RLM Cl = 663 μL/min/mg). When dosed orally, 6j was not detectable in rat plasma, indicating a lack of oral bioavailability. Two fluorinated anthranilamide analogues, 6k and 6l (Table 5), were synthesized in order to assess the impact of
a All kinase assays were run at the Km for ATP and represent an average of at least two determinations. b Average of at least two determinations, unless otherwise specified. c One determination. For statistical analysis of averaged data, see the Supporting Information.
Table 4. Potency, Selectivity, and Pharmacokinetic Properties of 6j
Enzyme IC50 (nM)a aurora-A/TPX2 16 aurora-B 9 Tie2 440 KDR 794 IC50 > 1000 nM: p38R, JNK2, cMET, Jak3, c-kit, Src, Lck, Abl(T315I), BTK, IGFR-1, CDK1, CDK2, CDK5, PKBR, p70s6k, PLK1, Erk1 Rat Pharmacokineticsb Cl (L/h/kg) Vss (L/kg) t1/2 (h) %F
3.6 4.8 2.8 25000 nM: KDR, Lck, Src, BTK, IGFR-1, Jak3, JNK2, CDK5, Erk1, c-Met, p706s6k, PLK1, PKBR, PKBβ, MSK1, Abl(T315I), c-Kit a All kinase assays were run at the Km for ATP and represent an average of at least two determinations.
Figure 3. Compound 8b inhibits aurora-A and aurora-B activity and induces polyploidy in human tumor cells. (A) Western blotting of cell lysates from HeLa cells synchronized in mitosis and treated with DMSO, compound 8b at the indicated concentrations, MLN8054 (aurora-A kinase inhibitor) at 0.5 μM, or AZD1152 (aurora-B kinase inhibitor) at 1 μM for 3 h. (B) HeLa cells were treated with DMSO or 8b at 0.313 μM for 24 h. The levels of p-histone H3 Ser 10 (red) and DNA content (2N, 4N, and 8N) were determined by flow cytometry. (C) Representative merged images of HeLa cells treated with DMSO or compound 8b at 0.313 μM for 48 h. Cells were immunostained with anti-R-tubulin (red) and anti-pericentrin (green) antibodies and DNA counterstained with DAPI (blue).
Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 17
6373
not assessed, but our experience with related pyridyl-pyrimidine phthalazines indicates that g6 h/day suppression of tumor p-histone H3 results in robust inhibition of tumor growth.15 Conclusion
Figure 4. Mice bearing established COLO 205 xenograft tumors were administered a single oral dose of vehicle alone or 8b at 50 mg/ kg. Tumors were collected at 3, 6, and 24 h after treatment (N = 4 per time-point) and the level of p-histone H3 was determined by immunofluorescence-based flow cyotmetry. Pharmacodynamic response is represented as the percentage of p-histone H3 positive cells in G2/M (bars, mean þ SD) plotted against concentration of 8b in mouse plasma (b) (mean ( SD). The controls (black columns) represent vehicle-treated tumors. Statistical significance was determined using ANOVA followed by Scheffe’s post hoc analysis (*P < 0.0001 vs vehicle-treated control).
inhibitor)6 and 1 μM AZD1152 (3a, aurora-B kinase inhibitor)7 selectively blocked the autophosphorylation of either aurora-A or aurora-B, respectively, as expected. Flow cytometry experiments (Figure 3B) established that HeLa cells incubated with 0.313 μM 8b had reduced phosphorylation of histone H3 on Ser 10 (a proximal substrate of aurora-B) and an increased percentage of cells with >4N DNA content, a direct measure of polyploidy. To investigate the effects of 8b on cellular morphology, HeLa cells were treated with compound for 48 h and stained with DAPI (40 ,6-diamidino-2-phenylindole) and antibodies to R-tubulin and to percentrin. As anticipated with inhibition of aurora-B, 8b treatment increased the nuclear size of mitotic and interphase cells and induced irregular spindle geometry (Figure 3C). Although cell viability studies were not conducted with 8b, our general experience with pyridinyl-pyrimidine phthalazines like 8b is that inhibition of cell proliferation correlates well with inhibition of histone H3 phosphorylation.15 On the basis of excellent potency, selectivity, and acceptable rat pharmacokinetics, 8b was selected for further evaluation in a tumor pharmacodynamic assay (Figure 4). To assess whether 8b can inhibit phosphorylation of the aurora-B substrate histone H3 at serine 10 in vivo, mice with established COLO 205 xenograft tumors of ∼300 mm3 were orally administered a single dose of vehicle or 8b at 50 mg/kg (N = 12 per group). At 3, 6, and 24 h post-treatment (N = 4 per time-point), plasma samples were collected and analyzed for 8b, and tumor tissues were harvested and analyzed for p-histone H3 by immunofluorescence-based flow cytometry. After 3 and 6 h, 8b treatment resulted in statistically significant inhibition of p-histone H3 at 50 mg/kg (g96% inhibition, P e 0.0001) with plasma concentrations of 8b g20000 ng/mL (g38 μM). By 24 h, p-histone H3 returned to baseline levels with a plasma concentration of 8b equal to 2460 ng/mL (4.8 μM). A qualitatively similar result was also observed in mouse bone marrow (data not shown). As 8b is highly bound to mouse plasma proteins (>99.9% bound), the unbound concentration at the 6 h time point is
