Discovery of a Potent and Orally Bioavailable CCR2 and CCR5 Dual Antagonist

Abstract

This report describes the discovery of a potent, orally bioavailable CC chemokine receptor 2 (CCR2) antagonist which, while optimized for CCR2 potency, also had potent CC chemokine receptor 5 (CCR5) activity.

Monocyte chemoattractant protein-1 (MCP-1, CCL2), included within the CC class of chemokines,¹ mediates chemotaxis of monocytes to inflammatory sites via interactions with its receptor, CCR2.² CCR2, a member of the G-protein-coupled seven-transmembrane receptor superfamily, is most abundantly expressed on monocytes. The CCL2/CCR2 axis has been implicated in various autoimmune or inflammation associated diseases³⁻¹² including rheumatoid arthritis (RA),⁵ multiple sclerosis,⁶ atherosclerosis,⁷ chronic obstructive pulmonary disease (COPD),⁸ asthma,⁹ diabetes/obesity,^10,11 and neuropathic pain.¹² A number of small molecule CCR2 antagonists have already been described by our group¹³⁻¹⁷ and by others.^18,19 Several small molecule CCR2 antagonists and a humanized monoclonal antibody targeting CCL2 have been advanced to clinical trials for treatment of RA, multiple sclerosis, atherosclerosis, COPD, pain, and diabetes.¹⁸⁻²² While the outcomes from the earliest trials for CCR2 blockade in RA have so far been disappointing,^21,22 a trial in patients at high risk for cardiovascular disease having elevated high-sensitivity C-reactive protein (CRP) levels with Millenium's CCL2 monoclonal antibody (MLN-1202²³) demonstrated lowering of CRP, consistent with a beneficial anti-inflammatory effect. This communication describes the discovery and detailed characteristics of 3-[(3R,4S)-1-((1R,3S)-3-isopropyl-3-{[6-(trifluoromethyl)-2H-1,3-benzoxazin-3(4H)-yl]carbonyl}cyclopentyl)-3-methylpiperidin-4-yl]benzoic acid, a potent and orally bioavailable CCR2 antagonist that was selected as a clinical candidate.

We have previously described CCR2 antagonists based upon an aminocyclopentane carboxamide scaffold typified by 1.^15,16 One drawback of compound 1 and its analogues was their high binding affinity at the outward delayed rectifier potassium channel (I_Kr). Blockade of I_Kr is associated with QTc prolongation in vivo, carrying the potential to induce cardiac arrhythmias.^24,25 Therefore, we set out to modify 1 to minimize the I_Kr inhibition, while retaining CCR2 potency. We previously reported that incorporation of a carboxyphenyl group onto the piperidine 4-position in a related series of CCR2 antagonists (2) dramatically decreased I_Kr binding affinity when compared to our standard 4-fluorophenylpiperidine-based analogues.¹⁷ We applied this observation to compound 1, giving target analogue 3 (Figure 1).

Figure 1.

Leads 1 and 2, and derived target analogue 3.

Compound 3 was prepared in an analogous fashion to that described previously for its close analogue 2.¹⁷ CCR2 binding affinities were determined by measuring inhibition of ¹²⁵I-MCP-1 binding to the endogenous CCR-2 receptor on human monocyte whole cells.¹³ I_Kr binding data was obtained as described previously.²⁶ While the I_Kr binding affinity for analogue 3 was successfully reduced by 600-fold (I_Kr IC₅₀ = 21 μM), the CCR2 binding affinity was also reduced, albeit by only 7-fold to 29 nM. In an effort retrieve the lost CCR2 potency, we decided to incorporate beneficial modifications found in other series explored within our program. In particular, we have observed that substitution of an oxygen atom in place of carbon at the 4-position of the 7-trifluoromethyltetrahydroisoquinoline subunit of 1 can improve potency. We decided to incorporate this feature, and the synthesis of the corresponding target compound 17 is described in Scheme 1. Commercially available fluoride 4 was treated with potassium tert-butoxide to give the ether 5.²⁷ Reduction with hydrogen and catalytic Raney Ni provided the corresponding benzyl amine 6. Commercially available ketoacid 7 was protected as its tert-butyl ester²⁸/dimethylacetal in two steps. Deprotonation of 8, followed by addition of 2-iodopropane, gave 9. Treatment with a 5:1 mixture of 4 M HCl in dioxane and water removed both protecting groups to afford ketoacid 10. Conversion to the corresponding acid chloride, followed by coupling with benzyl alcohol, provided the benzyl ester 11. The ester was resolved using preparative chiral HPLC (Chiralcel OD column, Chiral Technologies, Inc., 15% IPA/hexanes). The first peak to elute was identified as having the desired S absolute stereochemistry by elaboration to a known pure CCR2 analogue. Hydrogenolysis of the benzyl ester with hydrogen (1 atm) and Pd/C in methanol afforded (S)-ketoacid 10. Conversion of (S)-10 to acid chloride (S)-12, followed by addition of amine 6 and triethylamine, provided amide 13. Reduction of the ketone with sodium borohydride was followed by formation of the corresponding acetate ester. Cleavage of the tert-butyl ether to the phenol 14 was achieved by treatment with 4 M HCl in dioxane. Then, addition of an excess of paraformaldehyde and catalytic TsOH in toluene, followed by azeotropic removal of the generated water under reflux, provided the cyclized benzoxazine product 15. A more direct approach where cyclization was attempted in the presence of the cyclopentanone failed to give any product. Hydrolysis of the ester group in 15, followed by Swern oxidation, gave ketoamide 16. Reductive amination with ethyl 3-piperidin-4-yl benzoate¹⁷ gave a mixture (∼1:1) of cis- and trans-isomers which were separable by preparative TLC. The higher eluting compound was presumed to be the cis-isomer on the basis of the CCR2 binding affinity of both esters (higher eluting IC₅₀ 7.2 nM, lower eluting 41% inhibition at 1 μM). We have previously reported that, of the 4 possible 3-amino-1-carboxamide-cyclopentane stereoisomers, the cis-(3R,1S)-isomer alone has potent activity on CCR2.¹⁵ We have never observed an exception to this stereochemistry preference. Hydrolysis of the benzoate ester using excess lithium hydroxide gave the target analogue 17.

Scheme 1. Synthesis of Benzoxazine Analogue 17.

(a) KO-t-Bu (1.5 equiv), THF, 0 °C−rt; (b) H₂, Raney Ni, EtOH, 37% aq NH₄OH; (c) H₂SO₄, MgSO₄, t-BuOH, DCM; (d) TMOF, TsOH, MeOH, DCM; (e) LDA, THF; 2-iodopropane, −78 °C−rt; (f) 4 M HCl/dioxane, H₂O (5:1); (g) oxalyl chloride, DMF (cat.), DCM; BnOH, Et₃N, DCM, DMAP; (h) chiral HPLC, Chiralcel OD column, 15% IPA/hexanes, faster eluting peak; (i) H₂, Pd/C, MeOH, 1 atm; (j) oxalyl chloride, DMF (cat.), DCM; (k) Et₃N (2 equiv), 6, DCM; (l) NaBH₄, MeOH, 0 °C−rt; (m) Ac₂O, Et₃N, DMAP (cat.); (n) 4 M HCl/dioxane, rt; (o) paraformaldehyde (∼1 g/3 mmol substrate), TsOH, toluene, Dean−Stark trap, reflux 3.5 h; (p) LiOH, EtOH, H₂O; (q) oxalyl chloride, DMSO, substrate, Et₃N, −78 °C−rt; (r) ethyl 3-piperidin-4-ylbenzoate,¹⁷ NaB(OAc)₃H, 4 Å molecular sieves, 3−5 days; (s) LiOH (10 equiv), EtOH, water, 50 °C, 1.5 h.

As anticipated, benzoxazine analogue 17 was more potent compared to 3 in our binding assay (CCR2 IC₅₀ = 7 nM). In addition, 17 demonstrated weak binding on the I_Kr channel (IC₅₀ = 7.9 μM), corresponding to a >1000-fold selectivity window. In a functional assay measuring inhibition of MCP-1 mediated chemotaxis of monocytes¹³17 was a potent functional antagonist (IC₅₀ of 0.5 nM). In addition, compound 17 was orally bioavailable in rats (F = 21%), beagles (82%), and rhesus monkeys (F = 60%).

Encouraged, we decided to explore the possibility of further potency enhancements. In particular, we knew from earlier work that incorporation of a trans-(3R)-methyl group into the 4-arylpiperidine subunit can improve potency.^14,15 We decided to install the corresponding trans-4-(3-carboxyphenyl)-3-methylpiperidine 21 (Scheme 2) to determine if the same potency enhancement would apply in this series. The synthesis began with Boc protected ethyl 3-piperidin-4-ylbenzoate 18.^17,29 Oxidation with ruthenium(IV) oxide and sodium periodate proceeded over 11 days to afford the lactam 19.³⁰ Deprotonation, followed by treatment with iodomethane, gave the trans-3-methylpiperidinone 20. Removal of the Boc group, followed by reduction with borane−dimethylsulfide complex, gave the corresponding piperidine 21. Difficulties with the purification of 21 following the reduction step led us to protect the amine (Boc), whereupon purification was readily accomplished. The Boc group was then removed to provide clean 21 as a mixture of two trans-isomers. Reductive alkylation with ketone 16 afforded a mixture of all four possible diastereomeric products (∼5:5:6:6 ratio, 80% total yield). The four isomers were separated by preparative chiral HPLC using a Chiralcel OD column (Chiral Technologies, Inc.). The third (18%) and fourth (17%) peaks to elute were determined to be the desired cis-cyclopentane isomers. Hydrolysis of the esters collected from the third and fourth peaks afforded the target acids 22 and 23. We initially assigned the piperidine stereochemistries for 22 and 23 (as shown in Scheme 2) on the basis of their potencies; by analogy to an earlier series we knew that while both trans-isomers maintain potency, the (3R,4S)-4-aryl-3-methylpiperidine stereochemistry is optimal.¹⁴ We later definitively confirmed the tentatively assigned stereochemistry of 22 by application of an alternative synthetic route to the (4R)-carboxyphenyl-(3S)-methyl isomer of piperidine intermediate 21, where the stereochemistry was established though use of a chiral starting material of known absolute stereochemistry.³¹ Similarly, the cis-cyclopentane stereochemistry was confirmed by a later process synthesis using chiral starting materials of known absolute stereochemistry.

Scheme 2. Synthesis of Analogues 22 and 23.

(a) Ruthenium(IV) oxide hydrate (0.2 equiv), NaIO₄ (3.2 equiv), CHCl₃/H₂O, 11 days at rt;³¹ (b) KHMDS, THF; MeI, −78 °C; (c) 4 M HCl in dioxane; (d) BH₃ DMS, THF, rt; 0.5 M HCl/EtOH, 50 °C, 4 h; (e) Boc₂O, DIEA, DCM, DMAP (cat.); (f) 4 M HCl in dioxane; (g) 17, NaB(OAc)₃H, 4 Å mol sieves powder, DIEA, 4 days, rt; (h) preparative chiral HPLC (Chiralcel OD column, 8% EtOH/hexanes, cis-peaks were third and fourth to elute); (i) LiOH, EtOH, H₂O, rt, 22 h.

The CCR2 binding, functional, and I_Kr binding data for compounds 22, 23, as well as analogues 1, 3, and 17, are presented in Table 1. While analogues 22 and 23 are both potent, analogue 22 appears slightly more potent than both 23 and 17. Compound 22 exhibited similar binding potency for mouse CCR2 (IC₅₀ = 4 nM). Direct equilibrium binding experiments using ³H-labeled 22 and monocytes demonstrate that 22 has a K_D for the receptor of 0.7 nM. ³H-22 had a very slow association and disassociation at the receptor giving a receptor disassociation time (T_1/2) which was difficult to measure at room temperature, but was greater than 9 h. Analogue 22 is a potent inhibitor of MCP-1 mediated chemotaxis of monocytes (IC₅₀ = 0.3 nM). An in vitro human whole blood shape change assay³² measuring MCP-1-induced changes in monocyte cell shape was used to measure the potency of compound 22 to functionally antagonize CCR2 on monocytes in whole blood. In this assay compound 22 has an IC₅₀ of 15 nM with a preincubation of 30 min. Interestingly, when 22 was preincubated in whole blood for 24 h, the IC₅₀ improved to 0.1 nM, a shift in over 2 orders of magnitude. This is likely due to an underestimation of potency under standard 30 min preincubation conditions as a result of a combination of the slow receptor kinetics and the high level of plasma protein binding. Reversible binding to plasma proteins in vitro was 98, 98, 97, and 97% in rat, dog, monkey, and human, respectively (at 0.1 μM). When the compound was allowed ample time to come to equilibrium with the receptor on monocytes during the overnight incubation before whole blood stimulation with MCP-1, the compound's true potency could be determined.

Table 1. CCR2 Binding and Functional Data and I_Kr Binding Data for Analogues 1, 3, 17, 22, and 23^a.

	IC50 (nM)
analogue	CCR2	chemotaxis	IKr
1	4 (3)	2 (2)	49 (2)
3	29 (1)		21,000 (2)
17	7 (16)	0.5 (2)	7,900 (2)
22	4 (5)	0.3 (3)	33,000 (2)
23	6 (3)	2 (2)	7,300 (2)

^aNumbers in parentheses represent numbers of determinations. Standard deviations, when calculable, are always less than 25% of the measured value.

The 3-methyl group in 22 appears to confer improved selectivity against the I_Kr channel (I_Kr/CCR2 > 8,000). Compound 22 displayed potent activity against the closely related CCR5 receptor. However, the potency of 22 at CCR5 was at least 2.5-fold less than that observed for CCR2 and the compound exhibited much faster disassociation kinetics at CCR5.³³ No immune system deficits have been described in a group of individuals deficient for CCR5 (CCR5 delta32 homozygous individuals). Therefore, CCR5 blockade by 22 is unlikely to be a treatment liability. Interestingly, the IC₅₀ of 22 on murine CCR5 is >10 μM. Counterscreening against a panel of 136 receptors, enzymes, and ion channels (MDS Pharma Services) established 22 to be remarkably selective for CCR2. In addition to the known CCR5 activity, 22 displayed weak activity against the M2 and M4 subtypes of the muscarinic receptors (IC₅₀ ≥ 10 μM for both). Compound 22 did not inhibit any of five human CYP marker enzymes assayed (>100 μM on CYP3A4, CYP2C9, CYP2D6, CYP1A2, and CYP2C19). As shown in Table 2, 22 had excellent oral bioavailability in rats (F = 48%), dogs (F = 63%), and monkeys (F = 66%).

Table 2. Pharmacokinetic Properties of Compound 22^a.

species	dose (iv/po; mg/kg)	F (%)	AUC (po; μM·h)	Vdss (L/kg)	Clp (mL/min/ kg)	t1/2 (h)
rat	1/3	48	5.3	0.4	8.1	0.9
dog	1/2	63	33	0.5	1.1	5.5
rhesus	1/2	66	4.7	0.9	8.3	4.9

^aRat pk was analyzed in blood; dog and mouse were analyzed in plasma.

The in vivo efficacy of compound 22 was evaluated in an ex vivo rhesus whole blood shape change assay in which 22 was dosed at 2 mg/kg orally. Comparison of the plasma levels with the inhibition of monocyte shape change produced a concentration responsive curve that gave an estimated IC₅₀ of 0.9 nM (Figure 2).³⁴ This is consistent with the potency of 22 on rhesus CCR2 determined in in vitro whole blood spiking experiments in rhesus blood (IC₅₀'s of 30 nM and 0.2 nM with 30 min and 24 h preincubations, respectively). Our data suggest that subnanomolar potency can be achieved in vivo.

Figure 2.

In vivo potency of 22 determined by ex vivo rhesus whole blood shape change assay.³⁴

In summary, we have improved upon the I_Kr selectivity and CCR2 potency of antagonist lead 1 to give antagonists 17 and 22 which are orally bioavailable, potent functional antagonists of CCR2. Compound 22 was highly selective for CCR2, with the exception of related chemokine receptor CCR5, where it also had potent activity. The potent in vivo efficacy of compound 22 was demonstrated using an ex vivo rhesus whole blood shape change assay. On the basis of its potency, in vivo efficacy, safety, and oral bioavailability, compound 22 was selected as a clinical candidate for the CCR2 program at Merck.

Abbreviations

CCR2, CC chemokine receptor 2; CCL2, CC chemokine ligand 2; CCR5, CC chemokine receptor 5; TLC, thin layer chromatography.

Supporting Information Available

Experimental details for the synthesis and characterization of CCR2 antagonists 3, 17, and 22. This material is available free of charge via the Internet at http://pubs.acs.org.