Aurora Isoform Selectivity: Design and Synthesis of Imidazo[4,5-b]pyridine Derivatives as Highly Selective Inhibitors of Aurora-A Kinase in Cells

Aurora-A differs from Aurora-B/C at three positions in the ATP-binding pocket (L215, T217, and R220). Exploiting these differences, crystal structures of ligand–Aurora protein interactions formed the basis of a design principle for imidazo[4,5-b]pyridine-derived Aurora-A-selective inhibitors. Guided by a computational modeling approach, appropriate C7-imidazo[4,5-b]pyridine derivatization led to the discovery of highly selective inhibitors, such as compound 28c, of Aurora-A over Aurora-B. In HCT116 human colon carcinoma cells, 28c and 40f inhibited the Aurora-A L215R and R220K mutants with IC50 values similar to those seen for the Aurora-A wild type. However, the Aurora-A T217E mutant was significantly less sensitive to inhibition by 28c and 40f compared to the Aurora-A wild type, suggesting that the T217 residue plays a critical role in governing the observed isoform selectivity for Aurora-A inhibition. These compounds are useful small-molecule chemical tools to further explore the function of Aurora-A in cells.


■ INTRODUCTION
Over the past decade there has been extensive interest in Aurora kinases as anticancer targets. 1−3 Aurora proteins, a family of three kinases designated as Aurora-A, -B, and -C, play key and distinct roles in different stages of mitosis 4−8 and are overexpressed in a wide range of human malignancies. 9−15 Research toward the discovery of Aurora kinase modulators as cancer therapeutics led to the identification of a significant number of small-molecule inhibitors of Aurora kinases, the majority of which inhibit all three isoforms A, B, and C. 1 Compounds displaying isoform selectivity have also been reported, including AZD1152, 16 a selective Aurora-B inhibitor, and 1, 17 2, 18 3, 19 4, 20 and 5, 21 which selectively inhibit Aurora-A ( Figure 1). Many inhibitors of Aurora kinases have reached clinical evaluation. 1−3 In relation to isoform selectivity, the ideal inhibitor profile (i.e., selective isoform inhibition versus pan-Aurora inhibition) is still unclear, but the ongoing clinical evaluation of Aurora inhibitors may provide new insights to better answer this question. However, it should be noted that the Aurora kinases have distinct cellular functions and their inhibition has divergent effects on cells undergoing mitosis. 22 Aurora-A is involved in the initiation of mitosis by promoting centrosome maturation and bipolar mitotic spindle formation, and its inhibition leads to increased G2/M accumulation and spindle defects. 17,22,23 Aurora-B is part of the chromosomal passenger complex (CPC), which also includes borealin, survivin, and the inner centromere protein (INCENP), and plays a critical role in the regulation of chromosome alignment and also in cytokinesis. 23 Inhibition of Aurora-B leads to abrogation of the spindle assembly checkpoint, so that cells rapidly undergo chromosome segregation even in the presence of spindle defects. 22 The effects of Aurora-B inhibition mask the effects of Aurora-A inhibition, and therefore, selective inhibitors of both Aurora-A and Aurora-B are needed to carry out detailed cell-based and in vivo studies on this kinase family. 22 The role of Aurora-C during mitosis is not well understood; however, it is established that Aurora-C plays a specific role in spermatogenesis. 24 Other key differences between the Aurora kinases are emerging. Aurora-A alone forms a complex with N-Myc, an oncoprotein that induces neuroblastoma, which protects N-Myc from proteosomal degradation, 25,26 and Aurora-A-selective inhibitors MLN8054 and MLN8237 disrupt the Aurora-A/N-Myc complex. 26 Significantly, the Aurora-A/N-Myc complex was not disrupted by the Aurora-A-selective inhibitor MK-5108, which belongs to a different chemotype. 26 This demonstrates the importance of having access to a range of structurally diverse selective Aurora isoform inhibitors, in addition to the existing pan-Aurora modulators. They can serve as useful small-molecule chemical tools to further explore the cellular function of Aurora proteins toward their eventual therapeutic application in cancer treatment.
We have previously reported the discovery of novel imidazo- [4,5-b]pyridine derivatives as potent, orally bioavailable inhibitors of Aurora kinases (Figure 2), 27,28 with a compound in this series being identified as a dual Aurora/FLT3 preclinical development candidate. 28 In parallel to this work, we also investigated the design and synthesis of Aurora-A-selective inhibitors based on the imidazo [4,5-b]pyridine scaffold. Herein, we report our medicinal chemistry program aimed at the identification of imidazo [4,5-b]pyridine-and 7-azaindole-based inhibitors displaying a high degree of selectivity for Aurora-A over Aurora-B.

■ CHEMISTRY
The synthesis of 7-azaindoles 16a−d is shown in Schemes 2 and 3, with the key step being formation of the azaindole ring via a Pd-catalyzed heteroannulation reaction. 29−31 Starting from 8, access was readily gained to the alkyne 10 using standard Boc protection followed by a Sonogashira crosscoupling reaction (Scheme 1). 5-Chloro-3-iodo-2-aminopyridine (11; Scheme 2) was obtained from 2-amino-5-chloropyridine upon treatment with N-iodosuccinimide and p-toluenesulfonic acid. Azaindole ring formation (compound 12) was effected by reacting 11 with alkyne 10 in the presence of DABCO and Pd(PPh 3 ) 2 Cl 2 (Scheme 2). 29−31 The silyl/iodo exchange proceeded in good yield using N-iodosuccinimide as previously described. 31,32 Subsequent Boc group removal followed by sulfonamide formation afforded the key intermediate 15. The final products 16a,b were then obtained via a Suzuki crosscoupling reaction (Scheme 2).
Starting from key intermediate 14, access to the acetamide derivatives 16c,d (Scheme 3, Table 1) was accomplished by acetylation of the amino functionality in 14 followed by the introduction of 1-methylpyrazol-4-yl and 4-methoxylphenyl via a Suzuki cross-coupling reaction (Scheme 3).

■ RESULTS AND DISCUSSION
Aurora-A, -B, and -C kinases share a high degree of primary sequence homology; there are only three sequence differences in the ATP-binding site. 35 In Aurora-A, these residues are Leu215

Scheme 3 a
(Arg in Aurora-B/C) and Thr217 (Glu in Aurora-B/C) on the C-lobe extension of the hinge region and Arg220 (Lys in Aurora-B/C) on the αD helix. The majority of reported Aurora-Aselective inhibitors exploit the Thr217 ATP-binding site sequence difference to drive selectivity for Aurora-A inhibition over Aurora-B inhibition. [18][19][20]36 Crystallographic analysis of Aurora-A in complex with small-molecule ligands provides insight into the attractiveness of targeting the Thr217 residue to elicit Aurora isoform selectivity. For example, in the case of the imidazo[1,2-a]pyrazine derivative 3 (Figure 1), it was proposed that the observed selectivity for Aurora-A inhibition is driven by the interaction of a sulfonamide with the side chain of Thr217 in Aurora-A and a steric clash with the equivalent Glu residue in Aurora-B/C. 19 In the crystal structure of 3 bound to Aurora-A, the Leu215 side chain points away from the active site, whereas the Arg220 is highly mobile and disordered. 19 Exploitation of these two residues (i.e., Leu215 and Arg220) to drive selectivity for Aurora-A inhibition over Aurora-B/C inhibition is therefore considered to be challenging. We have previously reported the discovery of imidazo [4,5-b]pyridine derivatives as inhibitors of Aurora kinases, and compounds in this series were cocrystallized with Aurora-A, providing us with a clear insight into the interactions of this class of compounds with Aurora kinases. 27,28 The structural knowledge on the ligand−protein interactions across the imidazo[1,2-a]pyrazine 19 and imidazo [4,5-b]pyridine 27,28 series formed the basis of a design principle for imidazo [4,5-b]pyridine-derived Aurora-A-selective inhibition. Overlay of the imidazo [4,5-b]pyridine-based inhibitor 6 27 with the selective inhibitor 3 19 bound to Aurora-A ( Figure 3) indicated that Thr217 could be accessed via an appropriate C7-imidazo [4,5-b]pyridine derivatization (e.g., compounds 21c in Scheme 5 and 21d,f in Scheme 6) or elaboration through the N1-position of the imidazopyridine ring (e.g., 7-azaindole structures 16a−d in Schemes 2 and 3). With regard to the former approach, we recognized that Thr217 targeting for   Figure 1). 18 Docking of structure 28c (Scheme 8) into the active site of Aurora-A (PDB code 3H0Z) 18 and a comparison with 2 bound to Aurora-A indicated a good overlay of the N-phenylbenzamide fragments ( Figure 4).
Our initial efforts focused on the evaluation of the 7-azaindole scaffold (Table 1) in relation to Aurora inhibitory potency and Aurora-A isoform selectivity. Assuming a mode of binding to the hinge region similar to that observed for the imidazo [4,5b]pyridine-based inhibitors, 27,28 p-methoxyphenyl and 1-methylpyrazol-4-yl were chosen as optimal representative R 2 substituents for the 7-azaindole scaffold. 27,28 1-Methylpyrazol-4-yl derivatives 16a and 16c displayed higher Aurora-A and -B inhibitory potencies compared with their p-methoxyphenyl counterparts 16b and 16d (Table 1). On the other hand, both the sulfonamide and the acetamide groups appear to have similar Aurora-A inhibitory effects (16a vs 16c and 16b vs 16d, Table 1). Regarding selectivity for Aurora-A inhibition, although a trend to more potent Aurora-A inhibition was seen with all compounds listed in Table 1, the level of the observed isoform selectivity was not considered significant.
Attempts to cocrystallize compounds of this 7-azaindole class with Aurora-A failed to provide quality crystals; as a result we were unable to rationalize these results on a structural basis. At this point, we revisited the docking of 16a,b into the active site of Aurora-A, which indicated an additional binding mode to that previously observed. The preferred binding mode for 16a was similar to that observed with the imidazo [4,5-b]pyridine-based inhibitors of Aurora with the 2-aryl/heteroaryl substituent, in both imidazo [4,5-b]pyridine and 7-azaindole scaffolds, pointing to the solvent-accessible area 27,28 (see Figure 5A and compound  Figure 3). However, an additional flipped binding mode was observed in docking experiments with compound 16b compared to 16a (compare Figure 5B with Figure 5A). It should also be noted that the p-methoxyphenyl derivatives 16b and 16d exhibited low Aurora-A inhibitory potencies. Taking all these findings into account, we decided to terminate our interest in the 7-azaindole scaffold and focus our efforts on evaluating the imidazo [4,5-b]pyridine-based approaches for Aurora-A-selective inhibition.

in
Targeting Aurora-A Thr217 for selective Aurora-A isoform inhibition was continued by investigation of C7-imidazo [4,5b]pyridine derivatization, i.e., introduction of pyrrolidine-and piperidine-based substituents ( Table 2). This approach afforded highly potent inhibitors of both Aurora-A and -B (21a−f, Table  2), with all compounds displaying similar inhibitory effects against these two Aurora isoforms. On this occasion, structural insight into ligand−protein interactions was gained by cocrystallization of 21c and 21a with Aurora-A. The complexes formed crystals in space group P6 1 22, with one Aurora/ compound complex per asymmetric unit.
Compound 21c occupies the ATP-binding site, with the pyridine N and imidazole NH forming hydrogen bonds to Ala213 in the hinge region of Aurora-A consistent with previous reports ( Figure 6A). 27,28 The amide substituent of the pyrrolidine in 21c points to Thr217 but is neither in the right orientation nor close enough (5.9 Å distance between the −OH of Thr217 and the amide carbonyl of compound 21c) to form a  strong hydrogen bond with the side chain of Thr217. Likewise, compound 21a binds to the hinge region as previously observed with closely related derivatives in this series. However, the substituted pyrrolidine ring adopts an unfavorable conformation for interaction with Thr217, with the acetamido group pointing away from this residue ( Figure 6B). These crystal structures were consistent with the lack of selectivity for compounds 21a and 21c, and we therefore, investigated alternative substitutions at C7.
We continued our attempts to target Thr217 for Aurora-Aselective inhibition via C7-imidazo [4,5-b]pyridine derivatization by the introduction of a benzylamino substituent (compound 28a, Table 3) and 4-amino-N-phenylbenzamide substituent (compound 28b, Table 3). The benzylamino derivative 28a inhibited Aurora-A and -B with similar IC 50 values (Table 3). However, the N-phenylbenzamide derivative 28b was a significantly more potent inhibitor of Aurora-A compared to Aurora-B, the IC 50 values were determined as 0.075 and 4.12 μM, respectively (Table 3). A similar trend was observed in Hela cervical cancer cells; 28b inhibited the autophosphorylation of Aurora-A at T288 (a cell-based biomarker for Aurora-A inhibition) 17,19 with an IC 50 value of 0.32 μM and histone H3 phosphorylation at S10 (a cell-based biomarker for Aurora-B inhibition) 17,19 with a significantly higher IC 50 value of 18.6 μM (58-fold difference). Subsequent replacement of the C7-NH linker in 28b with an oxygen gave compound 28c, which displayed a higher degree of selectivity in inhibiting Aurora-A over Aurora-B in biochemical assays (Table 3) and in cells. In the biochemical assays, the IC 50 values for inhibition of Aurora-A and -B were determined as 0.067 and 12.71 μM, respectively (Table  3). In Hela cervical cancer cells, 28c inhibited the autophosphorylation of Aurora-A at T288 (p-T288 IC 50 = 0.16 μM) 480-fold more potently compared with Aurora-B inhibition (p-HH3 IC 50 = 76.84 μM). A similar trend was also observed in HCT116 human colon carcinoma cells; the IC 50 values for the inhibition of autophosphorylation of Aurora-A at T288 and histone H3 phosphorylation at S10 were determined as 0.065 and 24.65 μM, respectively. The selectivity of 28c against the kinome was also determined by profiling in a 110-kinase panel at a concentration of 1 μM. In this panel, 28c inhibited only three kinases at a level higher than 80%, namely, Aurora-A, VEGFR (VEGFR1), and GSK3b (Table S3, Supporting Information). On this basis, compound 28c is a highly selective kinase inhibitor with a Gini coefficient 37 calculated as 0.719. In an attempt to understand possible reasons for the observed inhibition of VEGFR1 and GSK3b by 28c, we examined conserved residues in the binding site between Aurora-A and VEGFR1 and between Aurora-A and GSK3b. We identified E211, Y212, G216, and L263 (residue numbers for Aurora-A) as conserved binding motifs between Aurora-A and VEGFR1 and residues L210, Y212, P214, and Table 2. C7-Imidazo [4,5-b]pyridine Derivatization: Pyrrolidine and Piperidine Substituents a a Results are mean values of two independent determinations (±SD) unless specified otherwise. A superscript Roman "a" indicates results from a single experiment (each concentration run in duplicate).  T217 (residue numbers for Aurora-A) as conserved binding motifs between Aurora-A and GSK3b. However, determination of whether these conserved regions influence the inhibition of VEGFR1 and GSK3b would require additional crystallography beyond the scope of this study. The antiproliferative activity in human tumor cell lines was also studied, and 28c inhibited the growth of HCT116 human colon carcinoma cells with a GI 50 value of 2.30 μM. We have previously shown that compounds of the imidazo [4,5-b]pyridine class exhibit potent dual Aurora-A/ FLT3 inhibition, 28 and consistent with this finding, we observed FLT3 inhibition in vitro (IC 50 = 0.162 μM) 38 and potent antiproliferative activity in MV4-11 human AML cells (GI 50 = 0.299 μM).
Compound 28c displayed high mouse and liver microsomal stability (after a 30 min incubation with mouse and human liver microsomes, only 22% and 9% of 28c was metabolized, respectively). This promising stability prompted in vivo mouse pharmacokinetic profiling that revealed 28c as a highly orally bioavailable compound (F = 100%) with moderate clearance (0.053 L/h, 44.16 mL/min/kg) and volume of distribution (0.036 L, 1.8 L/kg).
The effects of the 6-Cl and C2-pyrazolyl substituents on Aurora-A inhibition were subsequently studied by preparing compounds 34 and 35, respectively (Scheme 8, Table 4). Both 34 and 35 were significantly less potent inhibitors of Aurora-A compared to 28c (Tables 3 and 4), indicating the requirement for both C6-Cl and C2-aromatic or -heteroaromatic substituents and consistent with previously reported SARs. 27,28,39 Having identified 28c as a highly selective Aurora-A inhibitor, our efforts focused on replacing the aniline moiety in 28c, a potential toxicophore, 40,41 with a range of aliphatic and heteroaryl amines ( Table 5). All replacements were well tolerated in relation to Aurora-A inhibitory potency, and the selectivity for Aurora-A over Aurora-B inhibition was generally maintained (Table 5). Compounds were also tested for the cellular inhibition of both Aurora-A and -B, and 40a inhibited Aurora-A in HCT116 cells significantly more potently compared to Aurora-B (p-T288 IC 50 = 0.095 μM versus p-HH3 IC 50 = 4.93 μM, 52-fold difference). Likewise, 40c was a more potent inhibitor of Aurora-A than Aurora-B in Hela cells (p-T288 IC 50 = 0.28 μM versus p-HH3 IC 50 = 19.72 μM, 70-fold difference). A similar trend was seen with 40b; in Hela cells it inhibited Aurora-A more potently compared to Aurora-B (p-T288 IC 50 = 0.58 μM versus p-HH3 IC 50 = 19.74 μM, 34-fold difference). Compound 40f displayed the highest potency inhibiting Aurora-A in the biochemical assay (IC 50 = 0.015 μM, Table 5), with Aurora-B inhibition being determined as 3.05 μM (Table 5). In Hela cells, 40f inhibited Aurora-A 346 times more potently compared to Aurora-B (p-T288 IC 50 = 0.070 μM versus p-HH3 IC 50 = 24.24 μM). Profiling of 40f in a 50-kinase panel at a concentration of 1 μM revealed a highly selective inhibitor; only one kinase, namely, VEGFR (VEGFR1), was inhibited higher than 80% (Table S4, Supporting Information). Compound 40f exhibited high mouse and liver microsomal stability (after a 30 min incubation with mouse and human liver microsomes, 28% and 22% of 40f was metabolized, respectively). However, an in vivo pharmacokinetic profiling in mouse revealed a lower oral bioavailability (14%) compared to that for 28c (100%).
Many attempts to cocrystallize 28c and 40f with Aurora-A were unsuccessful. However, the docking of 28c into the active site of Aurora-A suggested that the aniline moiety resides in close proximity to Thr217 (Figure 4). On this basis, we probed whether Thr217 (Glu in Aurora-B) is the main residue governing the selectivity for Aurora-A inhibition. Testing of 28c against the Aurora-A wild type and its T217E mutant expressed in Hela cells revealed that the Aurora-A T217E mutant was significantly less sensitive to inhibition (40-fold) compared to the Aurora-A wild type (p-T288 IC 50 = 4.11 and 0.107 μM, respectively). Subsequently, both 28c and 40f were tested against the Aurora-A wild type and its T217E, L215R, and R220K mutants in HCT116 cells (Table 6, Figure 7, and Figure S1 in the Supporting Information). Both 28c and 40f inhibited the Aurora-A L215R and R220K mutants with IC 50 values similar to those seen for the Aurora-A wild type (Table 6, Figure 7, and Figure S1). On the other hand, the Aurora-A T217E mutant was significantly less sensitive to inhibition by 28c and 40f compared to the wild type (33-fold and 64-fold, respectively; Table 6,    Figure S1). This body of evidence suggests that the Thr217 residue (Glu in Aurora-B/C) plays an important role in governing the observed selectivity for Aurora-A inhibition. In the above experiment, the inhibition of Aurora-B by 40f was also investigated by measuring the reduction in the phosphorylation of histone H3 at S10. As shown in Figure S2 (Supporting Information), inhibition of histone H3 phosphorylation at S10 was only achieved at high concentrations of 40f (partial inhibition at 25 μM and complete inhibition at 50 μM). Interestingly, at concentrations where phosphorylation of Aurora-A was completely inhibited (for example, at 1.5 μM), there was an increase in histone H3 phosphorylation ( Figure S2), most likely due to an increase in the percentage of mitotic cells as previously reported for other Aurora-A-selective inhibitors. 17,42 However, at higher concentrations, histone H3 phosphorylation was inhibited, indicating onset of Aurora B inhibition ( Figure  S2).
In conclusion, structural knowledge of imidazo[4,5-b]pyridine and imidazo[1,2-a]pyrazine-based ligand−Aurora-A protein interactions formed the basis of a design principle for imidazo [4,5-b]pyridine-derived Aurora-A-selective inhibition. This led to the discovery of imidazo [4,5-b]pyridine derivatives displaying a high degree of selectivity for inhibition of Aurora-A over Aurora-B in both biochemical and cellular assays. In particular, 28c and 40f were highly selective in inhibiting Aurora-A in both Hela cervical cancer and HCT116 human colon carcinoma cells. Testing of 28c and 40f against the Aurora-A wild type and its T217E, L215R, and R220K mutants suggested that the Aurora-A Thr217 residue (Glu in Aurora-B/C) plays an important role in governing the observed selectivity for Aurora-A inhibition. In addition, 28c and 40f are a different chemotype compared to other known Aurora-A-selective inhibitors, and 28c is highly selective within the tested kinome, with a kinase profiling differing from that of MK-5108 42 and MLN8237. 43 These compounds could therefore serve as useful small-molecule tools 44,45 to further explore the function of Aurora-A in cells; indeed, the need for structurally diverse inhibitors displaying Aurora isoform selectivity has recently been highlighted by Brockmann et al. 26 In their study, the Aurora-A-selective inhibitors MLN8054 and MLN8237 disrupted the Aurora-A/ N-Myc complex and induced apoptosis in neuroblastoma cell lines and tumors, a phenotype different from that seen with the structurally differentiated Aurora-A-selective inhibitor MK-5108 from a different chemotype. 26 These data suggest differences in the mechanism of action and also indicate a specific target patient population for treatment with an Aurora-A-selective inhibitor. For the Aurora-A kinase the enzyme reaction (total volume 10 μL) was carried out with 5 nM N-terminal HIS-tagged AurA, which was expressed and purified as previously described, 46   The reader provides a software package ("Reviewer") which converts the peak heights into percent conversion by measuring both product and substrate peaks. The percent inhibition was calculated relative to the total wells (containing enzyme) and blank wells (containing no enzyme plus DMSO). IC 50 values were calculated from a four-parameter logistics fit of percentage inhibition versus log concentration using GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA) or the Studies package (Dotmatics, Bishops Stortford, U.K.).
Kinase Selectivity Profiling. Compound 28c was profiled against a panel of 110 kinases at the International Centre for Protein Kinase Profiling, Division of Signal Transduction Therapy, University of Dundee. Compound 40f was profiled against a panel of 50 kinases (MRC-PPU Express Screen) also at the International Centre for Protein Kinase Profiling.
Cell Viability Assay. GI 50 values (50% cell growth inhibitory concentration) were determined as previously described. 27,47 Inhibition of Aurora-B and the Aurora-A Wild Type and Mutants in Cells. Determination of cellular inhibition of Aurora-A (at T288) and histone H3 (at S10) has been described previously. 19 Briefly, cells were transfected with the Myc-tagged Aurora-A wild type or mutants and treated with different concentrations of the inhibitors for 2 h followed by lysis in LDS sample buffer. The samples were sonicated and boiled, and the proteins were separated by LDS−PAGE and transferred to nitrocellulose membranes. The upper and lower parts of the membranes were probed with anti-P-T288 (Aurora-A autophosphorylation/activation) and anti-P-histone H3 (Aurora-B substrate phosphorylation) antibodies, respectively. The blots were quantified and IC 50 values determined using Graphpad Prism. Mammalian cell expression constructs encoding Aurora-A T217E, L215R, and R220K mutants with N-terminal Myc tags were produced using Quikchange mutagenesis (Stratagene).
Computational Chemistry. The protein−ligand cocrystal (PDB 2X6D) was prepared in MOE (Moleular Operating Environment) using Protonate3D, and the waters were removed. 48 GOLD (Genetic Optimization for Ligand Docking) was used for all docking experiments using the ChemScore (chemscore_kinase) scoring function. 49 The docked poses were analyzed on the basis of their ChemScore and also the consistency of the docked poses. The higher scored poses were selected as the preferred solutions if the majority of these solutions were consistent in binding mode. The binding site was defined as those atoms within 6 Å of the cocrystallized ligand in PDB 2X6D and the cocrystal ligand removed. The GOLD algorithm was set to explore an extensive binding mode space by setting the search eff iciency to 200%, or very f lexible. No constraints were used in the docking experiments.
Cocrystallization of Aurora-A with Ligand. Cocrystals of Aurora-A bound to compounds 21a and 21c were generated as previously described. 28 Data were integrated using XDS and processed using the CCP4 package. 50 Structures were solved by molecular replacement using Aurora-A (PDB code 1MQ4) as a model. Ligand fitting and model rebuilding were carried out using Coot, 50 and refinement was carried out using Phenix. 51 Mouse Liver Microsomal Stability. The compounds (10 μM) were incubated with male CD1 mouse liver microsome (1 mg mL −1 ) protein in the presence of NADPH (1 mM), UDPGA (2.5 mM), and MgCl 2 (3 mM) in phosphate-buffered saline (10 mM) at 37°C. Incubations were conducted for 0 and 30 min. Control incubations were generated by the omission of NADPH and UDPGA from the incubation reaction. The percentage of compound remaining was determined after analysis by LC/MS.
Human Liver Microsomal Stability. The compounds (10 μM) were incubated with mixed-gender pooled human liver microsome (1 mg mL −1 ) protein in the presence of NADPH (1 mM), UDPGA (2.5 mM), and MgCl 2 (3 mM) in phosphate-buffered saline (10 mM) at 37°C . Incubations were conducted for 0 and 30 min. Control incubations were generated by the omission of NADPH and UDPGA from the incubation reaction. The percentage of compound remaining was determined after analysis by LC/MS.
In Vivo Mouse PK (Compound 28c). Mice (female Balb/C) were dosed po or iv with 28c (5 mg kg −1 ) in 10% DMSO and 5% Tween 20 in saline. After administration, the mice were sacrificed at 5, 15, and 30 min and 1, 2, 4, 6, and 24 h. Blood was removed by cardiac puncture and centrifuged to obtain plasma samples. The plasma samples (100 μL) were added to the analytical internal standard (IS; olomoucine), followed by protein precipitation with 300 μL of methanol. Following centrifugation (1200g, 30 min, 4°C), the resulting supernatants were analyzed for 28c levels by LC/MS using a reversed-phase X-Bridge C18 (Waters, 50 × 2.1 mm) analytical column and positive ion mode ESI MRM on a Waters 2795 liquid chromatography system coupled to a Quattro Ultima triple-quadrupole mass spectrometer (Waters).
In Vivo Mouse PK (Compound 40f). Mice (female Balb/C) were dosed po or iv with 40f (5 mg kg −1 ) in 10% DMSO and 5% Tween 20 in saline. After administration, the mice were sacrificed at 5, 15, and 30 min and 1, 2, 4, 6, and 24 h. Blood was removed by cardiac puncture and centrifuged to obtain plasma samples. The plasma samples (100 μL) were added to the analytical IS (olomoucine), followed by protein precipitation with 300 μL of methanol. Following centrifugation (1200g, 30 min, 4°C), the resulting supernatants were analyzed for 40f levels by LC/MS using a reversed-phase Polar RP (Phenomenex, 50 × 2 mm) analytical column and positive ion mode ESI MRM on a Waters 2795 liquid chromatography system coupled to a Quattro Ultima triplequadrupole mass spectrometer (Waters).
Chemistry. General Procedures. Commercially available starting materials, reagents, and dry solvents were used as supplied.
Flash column chromatography was performed using Merck silica gel 60 (0.025−0.04 mm). Column chromatography was also performed on a FlashMaster personal unit using Isolute Flash silica columns or a Biotage SP1 purification system using Biotage Flash silica cartridges. Preparative TLC was performed on Analtech or Merck plates. Ion exchange chromatography was performed using acidic Isolute Flash SCX-II cartridges. 1 H NMR spectra were recorded on a Bruker Avance-500. Samples were prepared as solutions in a deuterated solvent and referenced to the appropriate internal nondeuterated solvent peak or tetramethylsilane. Chemical shifts were recorded in parts per million (δ) downfield of tetramethylsilane.
LC/MS analysis, method A: Analysis was performed on a Waters LCT with a Waters Alliance 2795 separation module and Waters 2487 dual-wavelength absorbance detector coupled to a Waters/Micromass LCT time-of-flight mass spectrometer with an ESI source. Analytical separation was carried out at 30°C on a Merck Chromolith SpeedROD column (RP-18e, 50 × 4.6 mm) using a flow rate of 2 mL/min in a 3.5 min gradient elution with detection at 254 nm. The mobile phase was a mixture of methanol (solvent A) and water containing formic acid at 0.1% (solvent B). Gradient elution was 1:9 Analytical HPLC analysis was performed on a Thermo-Finnigan Surveyor HPLC system or an Agilent Technologies 1200 series HPLC system at 30°C using a Phenomenex Gemini C 18 column (5 μm, 50 × 4.6 mm) and 10 min gradient of 10% → 90% MeOH/0.1% formic acid, visualizing at 254, 309, or 350 nm. The purity of the final compounds was determined by analytical HPLC as described above and is ≥95% unless specified otherwise.