Article pubs.acs.org/jmc
Rapid Development of Piperidine Carboxamides as Potent and Selective Anaplastic Lymphoma Kinase Inhibitors Marian C. Bryan,† Douglas A. Whittington,‡ Elizabeth M. Doherty,† James R. Falsey,† Alan C. Cheng,‡ Renee Emkey,‡ Rachael L. Brake,‡ and Richard T. Lewis*,‡ †
Medicinal Chemistry Research Technologies, Amgen Inc., One Amgen Center Drive, Thousand Oaks, California 91320, United States ‡ Amgen Inc., 360 Binney Street, Cambridge Massachusetts 02142, United States S Supporting Information *
ABSTRACT: Piperidine carboxamide 1 was identified as a novel inhibitor of anaplastic lymphoma kinase (ALK enzyme assay IC50 = 0.174 μM) during high throughput screening, with selectivity over the related kinase insulin-like growth factor-1 (IGF1R). The X-ray cocrystal structure of 1 with the ALK kinase domain revealed an unusual DFG-shifted conformation, allowing access to an extended hydrophobic pocket. Structure−activity relationship (SAR) studies were focused on the rapid parallel optimization of both the rightand left-hand side of the molecule, culminating in molecules with improved potency and selectivity over IGF1R.
■
INTRODUCTION Anaplastic lymphoma kinase (ALK) is a member of the insulin receptor superfamily of tyrosine kinases.1 Although not widely expressed in adult tissue, ALK is implicated in neuronal development, differentiation, and basal dopaminergic signaling.2,3 ALK is a putative oncogene and is expressed as part of aberrant fusion proteins in a number of cancers including anaplastic large-cell lymphomas (ALCL), inflammatory myofibroblastic tumors (IMT) and, more recently, a variety of solid tumor types.1,4 Additionally, abnormal expression of full-length ALK may play a role in oncogenesis given its aberrant expression in a number of different tumor types including diffuse large B-cell lymphoma, neuroblastoma, rhabdomyosarcomas, glioblastomas, and esophageal squamous cell carcinomas.1,5−8 In these settings, ALK is believed to support tumorigenesis through a variety of signaling mechanisms leading to cell-cycle progression, survival, cell migration, and cell shaping.9 Recently, the use of crizotinib for the targeted inhibition of ALK has been shown effective in early phase clinical trials against advanced non-small-cell lung cancers carrying activated ALK kinase.10 Coupling these compelling results with the knowledge that the protein is not widely expressed in adult tissue presents ALK as an attractive oncology target with a potentially large therapeutic window. We identified piperidine carboxamide 1 (Scheme 1) from a high-throughput screen of our proprietary sample collection against a recombinant, truncated ALK enzyme construct. Compound 1 inhibited the ALK construct with an IC50 = 0.174 μM and demonstrated comparable functional activity in a whole cell assay monitoring ALK phosphorylation (IC50 = 0.384 μM in ALK-positive Karpas-299 cells).11 The compound © 2012 American Chemical Society
also exhibited moderate to excellent selectivity over the related kinase family member insulin-like growth factor-1 (IGF1R, IC50 = 4.61 μM).12−14 Selectivity over IGF1R was of particular interest due to its critical functions as a mediator of cell proliferation, differentiation, and apoptosis, and broad expression of this kinase in normal tissue.15 Determination of the cocrystal structure of 1 bound to the ALK kinase domain (Figure 1) provided insights into possible improvements in potency and selectivity (PDB ID 4DCE).16 The cocrystal structure of 1 with ALK shows a “DFG-shifted” conformation, similar to a type 1 1/2 inhibitor conformation.17 When 1 is bound, Phe1271 of the DFG sequence, which normally occupies a hydrophobic pocket in the apoprotein,18 shifts and forms a lid on top of the benzyl group, allowing for an edge−face interaction with 1 and access to the extended hydrophobic pocket flanked by Phe1174 and Ile1179 (Figure 1A,B). The carboxamide carbonyl hydrogen bonds to catalytic Lys1150, which also hydrogen bonds to Glu1167. The amide NH, in turn, accepts a hydrogen bond from the backbone carbonyl of Gly1269, the residue preceding the DFG sequence. Additional interactions were observed between hinge residue Met1199 and N1 of the aminopyrimidine ring and the anilinic NH (Figure 1A). Finally, the trimethoxyphenyl group sat in a narrow groove sandwiched by Leu1122 and the hinge region. Comparison of the cocrystal structure of ALK with 1 and a crystal structure of a benzimidazole inhibitor-bound IGF1R19 shows residue differences in the region adjacent to the benzyl ring (Leu1196 vs Met1049 and Ile1171 vs Met1024 in IGF1R Received: November 21, 2011 Published: January 20, 2012 1698
dx.doi.org/10.1021/jm201565s | J. Med. Chem. 2012, 55, 1698−1705
Journal of Medicinal Chemistry
Article
Scheme 1. Retrosynthetic Analysis of 1
Figure 1. (A) Co-crystal structure of 1 with ALK at 2.0 Å resolution highlighting key interactions with the DFG residues (shown in green). (B) Cocrystal structure of 1 with ALK illustrating the extended hydrophobic pocket. (PDB ID 4DCE).
benzyl ring of compound 1. In addition, Asp1123 is oriented inward, where it impinges on the binding of compound 1. Substitution of Ile1171 (ALK) with Met1024 (IGF1R) also narrows the entrance to the pocket and modifies its shape. On the basis of these observations, we synthesized a series of analogues designed to probe the extended hydrophobic pocket and the linker binder region to optimize nonbonded interactions while maintaining or improving selectivity over IGF1R.
and ALK, respectively) as well as conformational differences in the DFG region (Figure 2).20 In the IGF1R crystal structure,
■
CHEMISTRY Retrosynthetic analysis of 1 (Scheme 1) reveals a lead amenable to a modular, parallel-synthesis approach. Arrays of analogues were prepared either via S N Ar with chloropyrimidine intermediate 2 (route A, Scheme 1) or by amide coupling to carboxylic acid intermediate 3 (route B, Scheme 1). Chloropyrimidine 2 was prepared from (S)-1-(tertbutoxycarbonyl)piperidine-3-carboxylic acid (4) and p-tolylmethylamine (5) according to Scheme 2. HATU-mediated coupling of 4 with 5 provided intermediate 6. Cleavage of the Boc protecting group under acidic conditions followed by reaction with 2,4-dichloropyrimidine afforded predominately the desired 4-substituted key intermediate 2. The three steps
Figure 2. Overlay of cocrystal structures of compound 1 with ALK (PDB ID 4DCE) and a structure of IGF1R bound to a benzimidazole inhibitor (PDB ID 2OJ9). ALK is in brown, 1 is in orange, and IGF1R is in green.
Phe1124 of the DFG region is flipped away from the pocket, precluding a putative favorable stacking interaction with the
Scheme 2. Synthesis of Compounds 7a−l via Chloropyrimidine Intermediate 2a
Reagents and conditions: (a) Et3N, HATU, CH2Cl2, 22 °C, 70%; (b) 1 M HCl, CH2Cl2, 22 °C, quant; (c) 2,4-dichloropyrimidine, Et3N, EtOH, 22 °C, 32%; (d) DMSO, 90 °C. a
1699
dx.doi.org/10.1021/jm201565s | J. Med. Chem. 2012, 55, 1698−1705
Journal of Medicinal Chemistry
Article
Scheme 3. Synthesis of Compounds 11a−v via Carboxylic Acid Intermediate 3a
a Reagents and conditions: (a) 2,4-dichloropyrimidine, Et3N, EtOH, 0−22 °C, 88%; (b) DMSO, 90 °C, 88%; (c) LiOH, H2O, dioxane, 22 °C, quant; (d) Et3N, HATU, DCE, 22 °C.
Scheme 4. Synthesis of R-Enantiomer of 1a
Reagents and conditions: (a) 2,4-dichloropyrimidine, Et3N, EtOH, 0−22 °C; (b) 3,4,5-trimethoxyaniline, DMSO, 90 °C; (c) TFA, CH2Cl2, 22 °C; (d) 5, HATU, Et3N, CH2Cl2, 22 °C.
a
Table 1. Hinge Region Modifications
IC50 (μM) (±SD)a
a
compd
R
ALK
IGF1R
1 7a 7b 7c 7d 7e 7f 7g 7h 7i 7j 7k 7l
3,4,5-OMePh Me Ph Bn 3-ClPh 4-ClPh 3-OMePh 4-OMePh 3-EtPh 4-EtPh 3,4-OMePh 3,4-benzodioxane 3,5-OMePh
0.174 (±0.012) >25c 2.32 (±0.13)b >25c 2.01 (±0.45) 6.30c 1.02 (±0.09) 3.10 (±2.06) 0.912 (±0.167) >25c 1.73 (±1.31) 1.48 (±0.48) 0.325 (±0.173)b
4.61 (±0.75) >25c >25c >25c >25c >25c 9.022 >25c >25c >25c >25c >25c 2.93 (±1.20)b
All values are the average of n ≥ 3 ± standard deviation unless indicated otherwise. bn = 2 ± standard deviation. cn = 1.
following high throughput mass-triggered preparative HPLC purification.22 In addition, the enantiomer of compound 1 (14) was generated from (R)-(−)-nipecotic acid ethyl ester (12) via carboxylic acid 13 using the method described for 3 (Scheme 4).
proceeded without epimerization, as determined by chiral supercritical fluid chromatography (SFC).21 Arrays of analogues were then rapidly assembled via SNAr reaction of 2 with various primary amines under thermal conditions. Final compounds were then obtained after high throughput masstriggered preparative HPLC purification.22 The synthesis of intermediate 3 is outlined in Scheme 3. Reaction of 2,4-dichloropyrimidine with ethyl-(S)-nipecotate (8) following the procedure for intermediate 2 afforded pyrimidine 9. Coupling with 3,4,5-trimethoxyaniline under thermal conditions gave aniline 10, which was saponified with LiOH to generate carboxylic acid 3. HATU-mediated coupling of 3 with various amines proceeded with minimal racemization of the chiral center.23 Final compounds were obtained as before
■
RESULTS AND DISCUSSION
Analogues were assayed for inhibition of ALK enzyme activity as well as selectivity over IGF1R.24 ALK functional activity was measured using a time-resolved fluorescence resonance energy transfer (TR-FRET) assay which measured the phosphorylation of a peptide substrate.11 FRET assays monitoring 1700
dx.doi.org/10.1021/jm201565s | J. Med. Chem. 2012, 55, 1698−1705
Journal of Medicinal Chemistry
Article
Table 2. Amide Bond Modifications
IC50 (μM) (±SD)a compd
X
Y
stereocenter
ALK
IGF1R
1 14
CO CO
NH NH
S R
0.174 (±0.01) 4.12 (±0.43)
4.61 (±0.75) >25
a
All values are the average of n ≥ 3 ± standard deviation.
Table 3. Right-Hand Side Modifications
IC50 (μM) (±SD)a
a
compd
R
ALK
IGF1R
1 11a 11b 11c 11d 11e 11f 11g
−CH2-(4-MePh) −Ph −CH2-Ph −(CH2)2-Ph −(CH2)3-Ph −CH2-CO-Ph −CH(S-Me)-Ph −CH(R-Me)-Ph
0.174 (±0.012) 0.833 (±0.063) 0.364 (±0.073) 0.358 (±0.415) 6.71 (±1.07) 2.94 (±0.83) 2.60c 2.01c
4.61 (±0.75) 6.59c 6.12 (±0.75)b >25b 6.39 (±1.17)b 4.10 10.54c >25c
All values are the average of n ≥ 3 ± standard deviation unless indicated otherwise. bn = 2. cn = 1.
Table 4. Right-Hand Side Substituent Effects
IC50 (μM) (±SD)a
a
compd
R
ALK
IGF1R
1 11h 11i 11j 11k 11l 11m 11n 11o 11p 11q 11r 11s 11t 11u
4-Me 2-Me 3-Me 3-OCF3 4-OCF3 2-Cl 3-Cl 4-Cl 3-NO2 4-NO2 3-CO2Me 4-CO2Me 3-Ph 4-Ph 3-OPh
0.174 (±0.012) 0.341 (±0.019) 0.083 (±0.018) 0.016 (±0.001)b 0.010 (±0.002) 0.587c 0.080 (±0.040) 0.095 (±0.012) 0.070 (±0.014) 2.93 (±0.59) 0.060 (±0.016)b 1.70 (±0.08) 0.019 (±0.01) 0.138 (±0.026) 0.192 (±0.054)
4.61 (±0.75) 8.32 (±2.62)b 0.722 (±0.046)b 0.116 (±0.015)b 0.798 (±0.088)b 7.08c 0.364 (±0.050)b 1.87 (±0.69)b 0.503 (±0.062)b >25b 8.89 (±4.29)b >25b 4.23c >25b 2.77 (±0.48)b
All values are the average of n ≥ 3 ± standard deviation unless indicated otherwise. bn = 2. cn = 1.
phosphorylation of a peptide substrate were also used for the counter-screening assay for IGF1R.
Structure−activity relationships specific to the hinge-binding region of the molecule are shown in Table 1, where 1701
dx.doi.org/10.1021/jm201565s | J. Med. Chem. 2012, 55, 1698−1705
Journal of Medicinal Chemistry
Article
Table 5. Comparison of 1, 11q, and 11s in Enzyme and Cellular Assays
IC50 (μM) (±SD)a
a
compd
R
cLogP
ALK
Karpas
cell shift
1 11s 11j 11m 11q 11v
4-Me 3-Ph 3-OCF3 3-Cl 3-CO2Me 3-morpholine
4.24 5.63 4.77 4.46 3.71 3.20
0.174 (±0.012) 0.019 (±0.011) 0.016 (±0.001)b 0.080 (±0.040) 0.060 (±0.016)b 0.031 (±0.002)b
0.384 (±0.126) 0.224 (±0.122) 0.059 (±0.021) 0.226c 0.061 (±0.032) 0.028 (±0.022)b
2.2 11.8 3.7 2.8 1.0 0.9
All values are the average of n ≥ 3 ± standard deviation unless indicated otherwise. bn = 2. cn = 1.
selectivity was observed (8-fold for 11i vs 17-fold for 11b). Introduction of a trifluoromethoxy group at C-3 or C-4 improved potency (11j IC50 = 0.016 μM and 11k IC50 = 0.010 μM, respectively). As with 11i, 11j showed a decrease in selectivity over IGF1R. C-3 or C-4 halo-derivatives showed improved potency relative to C-2 (11l, 11m, and 11n), although C-3 analogue 11m also exhibited increased activity for IGF1R. Strongly electron withdrawing groups such as nitro or methyl carboxylate maintained potency when appended to C3 (11o IC50 = 0.070 μM and 11q IC50 = 0.060 μM, respectively) and decreased potency at C4 (11p IC50 = 2.93 μM and 11r IC50 = 1.70 μM, respectively). Appending a less electron-withdrawing, hydrophobic phenyl ring to the C3 position (11s) increased potency for ALK and improved selectivity over IGF1R (ALK IC50 = 0.019 μM, IGF1R IC50 = 4.23 μM) to give the most selective analogue identified in these studies. C4-substituted phenyl 11t also maintained potency relative to methyl 1 (IC50 = 0.138 μM vs IC50 = 0.174 μM) without negatively impacting selectivity. Introduction of an ether spacer to 11s (11u) led to a 10-fold loss in ALK potency and decreased IGF1R selectivity relative to 11s. The functional activity of piperidine carboxamide 1 and potent analogues 11j, 11m, 11q, and 11s was subsequently measured in an ALK-positive Karpas-299 whole cell assay (Table 5). An apparent correlation between cell shift and the calculated logP (cLogP) was observed, such that as cLogP increased (11q < 1 < 11m < 11j < 11s), the cell shift also increased. These findings suggest that reducing clogP will lead to decreased cell shifts. To test this hypothesis, morpholine analogue 11v was generated using the method described in Scheme 3. Gratifyingly, the resulting compound proved highly potent in the ALK enzyme and cellular assays (ALK IC50 = 0.031 μM, cell IC50 = 0.028 μM). To evaluate morpholine 11v for kinase selectivity, the compound was analyzed in the IGF1R assay as well as tested at 1 μM against a panel of 99 kinases using the Ambit Biosciences KINOMEscan platform.25 Gratifyingly, morpholine 11v was >200-fold selective over IGF1R (IC50 = 7.39 μM) and showed no effect on IGF1R in the kinome panel. In addition, only 4 kinases, namely ALK, KIT, TRKA, and FLT3, have
