Discovery of BMS-753426: A Potent Orally Bioavailable Antagonist of CC Chemokine Receptor 2

Abstract

To improve the metabolic stability profile of BMS-741672 (1a), we undertook a structure–activity relationship study in our trisubstituted cyclohexylamine series. This ultimately led to the identification of 2d (BMS-753426) as a potent and orally bioavailable antagonist of CCR2. Compared to previous clinical candidate 1a, the tert-butyl amine 2d showed significant improvements in pharmacokinetic properties, with lower clearance and higher oral bioavailability. Furthermore, compound 2d exhibited improved affinity for CCR5 and good activity in models of both monocyte migration and multiple sclerosis in the hCCR2 knock-in mouse. The synthesis of 2d was facilitated by the development of a simplified approach to key intermediate (4R)-9b that deployed a stereoselective reductive amination which may prove to be of general interest.

The CC chemokine receptor 2 (CCR2) is a G-protein coupled receptor that is mainly expressed on monocytes, macrophages, and a subset of T cells.^1,2 A number of ligands interact with CCR2, including monocyte chemoattractant protein-1 (MCP-1 or CCL2), MCP-2 (CCL8), MCP-3 (CCL7), and MCP-4 (CCL13). Of these, MCP-1 is the primary ligand for CCR2, and CCR2 is considered to be the exclusive receptor for MCP-1.³ CCR2 plays a nonredundant role in directing the emigration of monocytes from the bone marrow into circulation and a context-dependent role in their extravasation into resting and inflamed tissue.⁴⁻⁶ Given its fundamental importance in monocyte biology,⁷ numerous preclinical studies have been conducted in models of cardiovascular disease,⁸ cancer,⁹ fibrotic conditions,¹⁰ and a wide range of other diseases.¹¹⁻¹³

In light of these data, it is not surprising that many different institutions have pursued the discovery of CCR2 antagonists.^14,15 Recently, we reported our discovery of BMS-741672 (1a) as a potent, selective, and orally bioavailable antagonist of CCR2.¹⁶ In a pharmacokinetic study of 1a in cynomolgus monkeys, we found high levels of the N-demethylated metabolite 2a in the blood circulation. Indeed, the total AUC_0–24h exposure of 2a was 1.4-fold higher than that of the parent 1a (Figure 1, equiv 1).¹⁷ Unfortunately, the functional inhibition activity of 2a was ∼20× weaker than its parent 1a (vide infra). The propensity for this demethylation was confirmed in a similar study with N,N-dimethyl 1b (Figure 1, equiv 2), and the percent of metabolite formation (2b formation) was even more substantial. In this publication, we describe our efforts to address the issue of N-demethylation, which led to the discovery of N-tert-butyl amine as an effective replacement for the previous dialkylamines.

Figure 1.

In vivo cyno PK studies of 1a and 1b.

With the goal of finding a potent CCR2 antagonist with improved in vivo metabolic stability, we began by varying the N-alkylamine group (Table 1, compounds 1a–2d). Note that the assays used to evaluate the different potencies of these molecules, including CCR2 binding, CCR5 binding, and monocyte chemotaxis, have all been reported by us previously.^16,18 Compared to 1a, the demethylated 2a was 9- to 25-fold less active in inhibition of CCR2, as shown in the binding and chemotaxis assays, respectively. We found that a smaller R group had little effect on CCR2 potency, as the binding IC₅₀ values of 2b (R = Me) and 2c (R = H) were 4.6 and 2.9 nM, respectively; however, the primary amine 2c had markedly reduced affinity for CCR5 (Table 1). In previous studies, we had explored a range of exocyclic amines but had not found any that were superior to N-isopropyl, N-methyl.¹⁹ It was therefore of interest when we determined that tert-butyl amine 2d was 20 times more potent than 2a in the monocyte chemotaxis assay and also exhibited 60-fold higher affinity for CCR5 (CCR5 binding IC₅₀: 2a = 4959 vs 2d = 83 nM). The CCR5 binding for compound 2d was also assessed in primary T-cells, wherein it exhibited a binding value of 6.3 ± 1.5 nM. A brief survey of replacements of the 6-trifluoromethyl-quinazolin-4-yl moiety confirmed the SAR finding with the tert-butyl amine 2d (Table 1, compounds 3–6). The 6-trifluoromethoxyquinazoline analogue 3 retained potent binding to both CCR2 and CCR5. The simple 6-trifluoromethylbenzamide 4 exhibited weaker affinity for both CCR2 and CCR5 in the binding assays (Table 1), consistent with previous SAR studies.²⁰ Notably, the picolinamides 5a and 5b demonstrated good CCR2 potency in the binding and chemotaxis assays, and the tert-butyl amine 5b showed ca. 26-fold improved affinity for CCR5. The bicyclic analogue of 5b, that is, 6-tert-butyl-pyrido[3,2-d]pyrimidine (6), retained similar binding and functional antagonism at CCR2, with still further improved CCR5 affinity.

Table 1. In Vitro Characterization of CCR2 Antagonists 1–6^a.

compd	CCR2 binding IC50 (nM)	CCR5 binding IC50 (nM)	CCR5/2 ratio	chemotaxisb IC50 (nM)	Pampa PC (nm/s)
1a	1.1 ± 0.7 (n = 18)	780 ± 92 (n = 4)	∼ 710	0.7	443
1b	3.9 ± 1.5 (n = 3)	8056 ± 6212 (n = 3)	∼ 2100	4.5	557
2a	15 ± 7.2 (n = 6)	4959 ± 2258 (n = 3)	∼ 330	20	208
2b	4.6 ± 1.4 (n = 31)	4631 ± 1758 (n = 9)	∼ 1000	1.7	235
2c	2.9 ± 0.8 (n = 4)	18745 ± 7497 (n = 3)	∼ 6500	3.0	30
2d	2.7 ± 1.3 (n = 15)	83 ± 43 (n = 9)	∼ 31	0.8	560
3	7.6 ± 3.6 (n = 9)	34 ± 19 (n = 6)	∼ 4.5	NDc	507
4	15	2930	∼ 190	ND	ND
5a	3.0 ± 0.3 (n = 4)	3867 ± 1406 (n = 2)	∼ 1300	1.6	30
5b	1.3 ± 0.5 (n = 5)	147 ± 52 (n = 3)	∼ 110	0.3	332
6	6.6 ± 1.0 (n = 6)	18 ± 18 (n = 3)	∼ 2.7	0.6	251

^aIC₅₀ values reported as the average of two or more determinations, with the exception of compound 4.

^bAntagonism of chemotaxis of human peripheral blood mononuclear cells was induced by 10 nM MCP-1 at 37 °C (see ref (18)).

^cNot determined.

We explored the basis for the increase in CCR5 potency with the tert-butyl amine via structural modeling^21,22 and solvent mapping²³ of compounds 2c and 2d (Table 1), capitalizing on the availability of the previously reported crystal structures of CCR2 with BMS-687681 (RCSB 5T1A)²⁴ and CCR5 with maraviroc (RCSB 4MBS).²⁵ The predicted binding mode of 2c and 2d in CCR5 orients the ligand as previously observed in the CCR2/BMS-687681 crystal structure, with the quinazoline moiety sandwiched between TM1 and TM7 and the trisubstituted cyclohexyl projecting toward TM2 and TM3. The cyclohexyl 2-amino substituents are predicted to occupy a hydrophobic pocket formed by the top of TM3 and bottom of EC2 (Figure 2). WaterMap analysis of this pocket in the absence of ligand predicted the presence of entropically unfavorable waters whose displacement into bulk solvent would be required to achieve high potency. Two of these waters became trapped in the presence of the amino (Wat1 and Wat2 in Figure 2A) and aminomethyl substituents (data not shown), increasing their entropic unfavorability. The tert-butyl amine (Figure 2B) is predicted to fill this pocket, thereby displacing the unfavorable waters with the resulting free energy gain leading to increased binding affinity. WaterMap analysis of the equivalent pocket CCR2 crystal structure suggested the unfavorable waters would be displaced by the amino moiety or accessible to bulk solvent (Figure 2C), consistent with the relative insensitivity of CCR2 to amino substitution (Table 1).

Figure 2.

Structural models of compounds 2c (A) and 2d (B) in CCR5. Unstable waters occupying the hydrophobic pocket between TM3 and EC2 are shown as spheres in (A). (C) WaterMap predicted waters in the hydrophobic pocket between EC2 and TM3 from the CCR2/BMS-687681 crystal structure (RCSB 5T1A).

All of the compounds shown in Table 1 were also profiled for their binding to ion channels (a well-described issue for chemokine receptor antagonists), passive permeability, and metabolic stability. As shown in the Supporting Information, all of the tested compounds exhibited low Na+ channel binding and excellent liver microsome metabolic stability. As a result, it was difficult to triage compounds based on those data. We were able to disqualify two of the compounds based on their low passive permeability (2c, 5a; see Table 1), and elected to advance compounds 2b, 2d, 3, 5b, and 6 were further for in vivo PK profiling, as summarized in Table 2. Compounds 2b, 2d, and 3 showed low plasma clearances (25–36 mL/min/kg) and good oral bioavailability (79–100 F%) in their respective PK studies. The oral exposures of 2b, 2d, and 3 (both C_max and AUC_0–24) were consistent with high intestinal absorption and agreed well with the PAMPA permeability (Table 1). Compared to 2b, 2d, and 3, the clearance of 5b was high (75 mL/min/kg) after a single 2 mg/kg intravenous dose, and the oral bioavailability was moderate (F% = 66) after a single 10 mg/kg oral dose. Compound 6 exhibited a still higher clearance (157 mL/min/kg) and volume of distribution (20.7 L/kg) after a single 2 mg/kg intravenous dose; although its oral bioavailability was 100%, the oral exposure (both C_max and AUC_0–24) was the lowest of the compounds tested (Table 2). Although all five compounds exhibited excellent in vitro liver microsome stability profiles, the in vivo clearance data of 5b and 6 were high and not consistent with the in vitro observation. On the basis of the profiles of high exposures (both C_max and AUC_0–24) and low clearance shown in Table 2, 2b, 2d, and 3 became compounds of interest and were chosen for further advancement.

Table 2. Rat Pharmacokinetic Properties of 1a, 2b, 2d, 3, 5b, and 6.

compd/route/dose	Cmax (nM)	AUC0–24 (μM·h)	T1/2 (h)	Cl (mL/min/kg)	Vss (L/kg)	F (%)
1a/iv/2 mg/kg			5.1	43	11.3
1a/po/10 mg/kg	500	3.79				51
2b/iv/2 mg/kg			3.6	31	5.4
2b/po/10 mg/kg	1632	10.1				88
2d/iv/2 mg/kg			4.5	25	7.1
2d/po/10 mg/kg	1765	10.2				79
3/iv/2 mg/kg			5.8	36	13.5
3/po/11.5 mg/kg	1868	11.9				100
5b/iv/2 mg/kg			2.7	75	10
5b/po/10 mg/kg	720	3.10				66
6/iv/2 mg/kg			3.7	157	20.7
6/po/13 mg/kg	478	2.56				100

Compounds 2b, 2d, and 3 were profiled further (Table 3). None of the compounds exhibited meaningful inhibition of CYP450 isoforms 2D6, 3A4, 2C9, or 2C19 at concentrations up to 40 μM. Accordingly, drug interactions involving the inhibition of the major drug-metabolizing CYPs are not anticipated with these compounds at clinically relevant concentrations. Likewise, none of the three compounds showed meaningful inhibition in a hERG patch clamp assay at 30 μM (2d = 19% < 2b = 24% < 3 = 36%). However, compound 3 showed significantly higher % inhibition in the Na⁺ channel patch clamp assay at 30 μM compared to compounds 2b and 2d (2b = 2.5% < 2d = 13% < 3 = 65%). Relative to its comparators, the t-Bu amine 2d also showed a favorable PK profile in mouse and monkey. After a single 10 mg/kg oral dose to mice, the oral bioavailability of 2d was superior to that of 2b (F% = 93 vs 38%, respectively), and the oral exposure of 2d was almost three times higher than that of 2b (AUC_0–24: 5.4 vs 2.0 μM·h). In addition, examination of results from a cyno PK study of 2d, orally dosed at 1 mg/kg, clearly indicated that 2d had an excellent oral exposure in terms of AUC_0–24h and oral bioavailability (AUC_0–24h = 2352 nM·h and F% = 95); these values are substantially better than our previous lead compound 1a (AUC_0–24h = 836 nM·h and F% = 47, Figure 1). Importantly, the level of undesired primary N-dealkylated metabolite (M1) was significantly diminished in the case of 2d: the M1 (2c) to parent ratio was <0.01 after a single 1 mg/kg oral dose of 2d in cyno monkeys (vs 1a:M1 (2a)/P ratio = 1.4, Figure 1).

Table 3. In Vitro and In Vivo Characterization of 2b, 2d, and 3.

	2b	2d	3
CYPa IC50 (μM)
3A4/2D6	> 40 > 40	> 40 > 40	> 40 > 40
2C9/2C19	> 40 > 40	> 40 > 40	> 40 > 40
hERG patch clamp: % Inh @ 30 μM	24	19	36
Na+ patch clamp: % Inh @ 30 μM, 4 Hz	2.5	13	65
protein binding %free:
human/cyno/dog	50/65/78	59/49/51
rat/mouse	53/51	53/47
mouse PK data (IV: 2 mg/kg; PO: 10 mg/kg)
AUC0–24h (μM·h)/F (%)	2.0/38	5.4/93
Cl (mL/min/kg)/T1/2 (h)	70/2.6	63/4.1

^aCYP = cytochrome P450.

On the basis of its overall profile, compound 2d was evaluated further in pharmacodynamic and efficacy models. On account of its weaker affinity for the mouse homologue of CCR2, we conducted these studies in hCCR2 knock-in (KI) mice.²⁶ The thioglycolate (TG) peritonitis model was used to determine the activity of 2d in inhibiting monocyte/macrophage infiltration. Mice were administered TG and dosed orally with 2d at 1, 25, or 100 mg/kg, twice daily. Then 48 h post TG treatment, the peritoneal lavage was obtained for cellular infiltrate analysis. When administered at doses of 1, 25, and 100 mg/kg, compound 2d inhibited 28%, 74%, and 78% of monocyte/macrophage influx, respectively (Figure 3), at plasma concentrations corresponding to 51%, 91%, and 98% estimated receptor occupancy.¹⁸ On the basis of three separate studies with multiple doses, the average EC₅₀ for inhibition of monocyte/macrophage infiltration was estimated as 3.9 nM. This can be compared with the activity of compound 1a in the same model, which exhibited an EC₅₀ of 2.2 nM.

Figure 3.

Compound 2d in the hCCR2 KI mouse TG peritonitis model.

To assess the effect of 2d on chronic models of disease, we used the experimental autoimmune encephalomyelitis (EAE) model of multiple sclerosis in hCCR2 KI mice and included the earlier clinical compound 1a as a comparator. Oral dosing of 2d at 25 mg/kg (BID) and 1a at 55 mg/kg was initiated on day 1. Compound 2d reduced the area under curve (AUC) of the clinical score by 49% (p < 0.05) (Figure 4) and provided a comparable efficacy profile to 1a at the doses studied. Although the antagonist 2d did reduce clinical score, histological evaluation of the spinal cord on day 22 did not demonstrate a significant difference in total inflammatory cellular infiltrate relative to vehicle.

Figure 4.

Study the effect of 1a and 2d in EAE model of multiple sclerosis.

The key intermediate 7, which we have previously described,¹⁶ was used for the preparation of all the final compounds studied in this study. We initially investigated a direct approach to synthesizing t-butyl amines from 7, even though the available literature on the subject of converting a primary aliphatic amine (H₂N-R) to a tert-butylamine (t-BuHN-R) at the time was limited. Our two initial attempts at making the tert-butylamine adduct from 7 met with only partial success (Scheme 1): the route through aziridine 8 proceeded in <20% yield and required five transformations, and the path through N-alkylated oxime 10 took seven steps and provided (4R)-9b in <10% yield.

Scheme 1.

Reagents and conditions. (a) TFA, CH₂Cl₂, rt, 1 h, 93%. (b) Step 1: Me₂CHCHO, MgSO₄, CH₂Cl₂, rt, 1 h; step 2: NCS, MeCN, rt; step 3: NaBH₄, EtOH, rt, 20% for three steps. (c) H₂, Pd(OH)₂, EtOAc, 15 psi, 95%. (d) Step 1: TFA, CH₂Cl₂, rt, 1 h, 93%; step 2: BrCH₂CN, DIEA, MeCN, 40 °C, 16 h; step 3: m-CPBA, CH₂Cl₂, rt, 0.5 h, 55% for 3 three steps. (e) Step 1: NH₂OH·HCl, MeOH, 60 °C, 2 h; step 2: acetone, CH₂Cl₂, rt, 8 h, 58% for two steps. (f) Step 1: MeMgBr, PhH, rt; step 2: CS₂C, MeCN, rt, 30% for two steps.

In light of the poor results with direct routes, we pivoted to a counterintuitive approach in which we first destroyed the C4-stereocenter only to reintroduce the stereochemistry via a reductive amination step with tert-butylamine (Table 4). Conversion of carbamate 7 to ketone 11 proceeded through simple Boc deprotection, followed by treatment of the resulting primary amine with 3,5-di-tert-butylcyclohexa-3,5-diene-1,2-dione in MeOH and Dowex-50WX8-200 resin.²⁷ The crude solid isolated from the reaction was triturated with a mixture of hexane and ethyl acetate to provide the pure ketone 11 in 90% yield.

Table 4. Reductive Amination Approach to (4R)-9b^a.

entry	(b) step 1	(b) step 2	(b) solvent	(4R)-9b:(4S)-9b	yield of (4R)-9b (%)
1	2.5 equiv Ti(OPri)4	1.3 equiv NaBH3CN	MeOH	1:1
2	0.6 equiv TiCl4	0.2 equiv Pd/C, H2	MeOH	1:1
3	0.6 equiv TiCl4	0.2 equiv PtO2, H2	MeOH	2:1	30
4	0.6 equiv TiCl4	0.2 equiv PtO2, H2	CH2Cl2	4:1	50
5	0.7 equiv TiCl4	1.1 equiv BH3 (1 M THF)	CH2Cl2	1:1
6	0.7 equiv TiCl4	1.1 equiv BH3·SMe2	CH2Cl2	6:1	70
7	1.1 equiv Ti(OEt)4	1.1 equiv BH3·SMe2	CH2Cl2	1:1
8	0.6 equiv TiCl4 and 0.6 equiv Ti(OPri)4	1.1 equiv BH3·SMe2	CH2Cl2	8:1	78

^aReagents and conditions. (a) step 1: TFA, CH₂Cl₂, rt, 1 h, 93% yield; step 2: 3,5-di-tert-butylcyclohexa-3,5-diene-1,2-dione, Dowex-50WX8–200 resin, MeOH, 90% yield; (b) step 1 :2 equiv t-BuNH₂; step-2: rt 16 h.

With the ketone 11 in hand, we were in a position to explore the Titanium catalyst-based asymmetric reductive amination reactions with tert-butylamine. Representative reactions with selected reagents are summarized in Table 4. Neither reaction of 11 with Ti(OPr_i)₄ followed by NaBH₃CN nor with TiCl₄ followed by Pd/C/H₂ gave any appreciable level of diastereoselectivity (entries 1 and 2, R:S = 1:1). Using the combination of PtO₂ and H₂ as the reducing reagents, the TiCl₄-catalyzed reactions provided moderate diastereoselectivity, especially when CH₂Cl₂ was used as the solvent (entries 3 and 4, Table 4). The diastereoselectivity was much better for the reaction using the combination of TiCl₄ and BH₃·SMe₂ compared to the combination of TiCl₄ and BH₃·THF (entry 5 vs 6). After significant efforts to optimize the reaction conditions, we found that the combination of 0.6 equiv TiCl₄ and 0.6 equiv Ti(OPr_i)₄ in CH₂Cl₂ followed by BH₃·SMe₂ as the reducing reagent gave (4R)-9b in 78% yield with 8:1 (entry 8). After aqueous workup, the crude product was treated with an equal amount of CH₂Cl₂ and 2% aqueous NH₄Cl. The isomer (4R)-9b was found exclusively in the organic fraction, with the diastereomer (4S)-9b in the aqueous fraction, presumably because of the modulation of exposed polarity via conformational switching.¹⁶ Evaporation of the organic fraction provided (4R)-9b in >98% diastereomeric excess without any column chromatography purification (Supporting Information). Triethylamine-promoted coupling reaction of (4R)-9b and 4-chloro-6-(trifluoromethyl)quinazoline gave 2d in 88% yield with 99.7% purity after a simple recrystallization protocol. The absolute and relative stereochemistry of 2d was confirmed by single X-ray crystal structure analysis (see Supporting Information).

Our efforts to improve the in vivo metabolic stability of clinical candidate BMS-741672 (1a), a tertiary amine subject to N-demethylation, led to the SAR study of 2–6 described herein. The use of tert-butylamine as a hydrophobic secondary amine facilitated the identification of potent, permeable, and metabolically stable compounds. Pharmacokinetic studies in monkeys confirmed that key molecule 2d (C4 substituent = N(H)t-Bu) was metabolically stable in vivo, unlike 1a (C4 substituent = N(Me)i-Pr). To prepare some of the key compounds tested in this study, we developed a diastereoselective reductive amination protocol for the preparation of key intermediate (4R)-9; the diastereomeric purity was enriched further by capitalizing on the modulation of exposed polarity via conformational switching.¹⁶ As a result, we were able to prepare more than 100 g of (4R)-9d (de > 98%) from common intermediate 7 without a single column chromatography purification.

Unfortunately, as for so many of the chemokine receptors,²⁸⁻³⁰ the pursuit of CCR2 antagonists has not led to the registration of a novel drug, yet. Several authors, including us,³¹ have proposed that dual inhibition of both CCR2 and CCR5 could provide superior results, and studies in both fibrosis and cancer continue toward this end.³²⁻³⁴ As such, we note that an additional aspect associated with the discovery of clinical candidate BMS-753426 (2d) was the identification of a novel vector for enhancing CCR5 affinity in our series of trisubstituted cyclohexane derivatives. Indeed, the ability to improve CCR5 affinity while retaining the conformational plasticity engendered by the C2-acetamide¹⁶ and avoiding C2-hydrophobes²⁰ represented an important advance for our overall efforts. These studies will be reported in due course.

Glossary

Abbreviations

AUC

area under curve

CCR2

chemokine receptor 2

CYPs

cytochromes P450

EAE

experimental autoimmune encephalomyelositis

PAMPA

parallel artificial membrane permeability assay

PK

pharmacokinetics

TNFα

tumor necrosis factor α

MS

multiple sclerosis

MCP-1

monocyte chemoattractant protein-1

RA

rheumatoid arthritis

SAR

structure–activity relationships

TG

thioglycollate.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.1c00082.

• Analytic experimental HPLC methods; synthetic procedures and analytical data of 1–6; synthetic references for intermediates 7, 9, and compounds 11–16; reference for the biological methods; SMILES representation of compounds with key data (PDF)

Accession Codes

Accession code CCDC 2055911 for compound 2d contains the supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif.

The authors declare no competing financial interest.