Discovery of BMS-955176, a Second Generation HIV-1 Maturation Inhibitor with Broad Spectrum Antiviral Activity

Abstract

HIV-1 maturation inhibition (MI) has been clinically validated as an approach to the control of HIV-1 infection. However, identifying an MI with both broad polymorphic spectrum coverage and good oral exposure has been challenging. Herein, we describe the design, synthesis, and preclinical characterization of a potent, orally active, second generation HIV-1 MI, BMS-955176 (2), which is currently in Phase IIb clinical trials as part of a combination antiretroviral regimen.

HIV-1 infection can be effectively managed with current treatments that rely on long-term administration of multiple antiviral inhibitors designated as combination antiretroviral therapy (cART).¹ Nevertheless, the development of novel agents with improved safety profiles and different mechanisms of action (MOA) is still highly desirable due to long-term toxicity and resistance development with current therapies. New ARTs would preferentially exhibit minimal drug–drug interactions with a relatively low once-daily dose to facilitate combinations with other agents as part of a fixed dose regimen.²

In the latter stages of the virus life cycle, the HIV-1 matrix (MA), capsid (CA), nucleocapsid (NC), and p6 proteins are released from the Gag polyprotein via multiple cleavages made by the viral protease. The final and rate-limiting step in this maturation process is cleavage at the CA-SP1 junction, which induces a major structural change within the virion, leading to the formation of the conical core that is characteristic of an infectious virion. Maturation inhibitors (MIs) prevent replication by binding near the CA-SP1 segment of the Gag polyprotein, specifically blocking only this final rate-limiting cleavage event, leading to the production of immature, noninfectious particles.^3,4

Proof-of-concept for this MOA was demonstrated in a 10 day Phase I/II monotherapy trial with bevirimat,(1, Chart 1), an MI derived from betulinic acid (BA, 10).^5,6 At the highest doses (150 and 250 mg), a maximal viral load reduction (VLR) of 0.72 log₁₀ was observed. However, in the Phase IIa trial with highly experienced patients, only 45% of individuals responded to treatment with an HIV-1 RNA change from baseline of >0.5 log₁₀, and only 34% with >1 log₁₀ VLR.⁷ The restricted response to 1 was subsequently determined to be a consequence of naturally occurring polymorphic variation at Gag amino acid residues 369–371.⁸⁻¹⁰ The lack of polymorphic coverage in combination with difficulties in formulation ultimately led to a halt in the development of 1.^11,12 Three other modified triterpenoid MIs have been evaluated in humans, but none of these compounds have advanced beyond Phase I clinical studies.¹³⁻¹⁵

Chart 1. Selected HIV-1 Maturation Inhibitors.

In order to drive the discovery of a second generation MI with broad polymorphic coverage, it was important to maintain activity against the key polymorphic variation at V370.¹⁶ Therefore, we chose the V370A- and ΔV370-containing viruses as the primary screening viruses due to their high prevalence in B and non-B subtypes, respectively, and their high level resistance to BVM.¹⁷ Additional improvements in a second generation MI would be lower binding to human serum components that decreased the effective potency of 1 (human serum (HS) binding >99%)¹⁷ and an improvement in physical properties that would allow for an appropriate formulation that could deliver targeted concentrations in vivo. Herein, we describe the discovery of BMS-955176 (2), an orally efficacious, second generation HIV-1 maturation inhibitor that is currently in Phase IIb clinical trials.¹⁸ The complete virology profile of BMS-955176 (2) is described in a companion manuscript, and resistance studies have been completed and will be published elsewhere.¹⁷

The description of the effect of Gag polymorphisms on the antiviral response to bevirimat (1) stimulated our interest in maturation inhibition as a target, since this identified a screening paradigm, but there was no evidence in the literature that structural modification of 1 could address the fundamental problem. As the first step in our study of this chemotype, we explored changes to the succinic acid-based side chain at C-3, a moiety that appeared to have largely been accepted as a pharmacophoric element, with a view to revealing key aspects of the topography of the carboxylic acid while providing new opportunities for structural modification.^19,20 The benzoic acid derivative 3 emerged very quickly from that effort, exhibiting good antiviral activity against wild-type (WT) virus, a modest 7-fold shift in the presence of HS, and good oral exposure in rats (Table 1). The para-substitution pattern in 3 was critical for antiviral activity. However, potency against Gag V370A, the most prevalent polymorphism in the region of Gag associated with MI activity, was lacking.¹⁹

Table 1. HIV-1 Inhibition and Rat AUC_0-6h of Compounds 1–9.

	WT EC50(nM)a
	WT	WT + HSb	V370A	ΔV370	CC50(μM)	Rat AUC0–6h(nM*h)c
1	10 ± 11	1,291 ± 1,011	552 ± 633	>10,000	16.8 ± 14.5	6,390
2	1.9 ± 1.8	10.2 ± 6	2.7 ± 1.5	13 ± 11	9.2 ± 4	4,132 ± 685d
3	16 ± 13	105 ± 40	233 ± 305	>6,000	27.2 ± 38.5	5,724
4	2.3 ± 1.3	43 ± 29	7.8 ± 8.8	31 ± 19	2.61 ± 1.78	1, 623 ± 240d
5	2.2 ± 0.7	8.7	2.5 ± 2.2	20.6 ± 14	1.24 ± 0.16	153
6	2.2	-	12	61	2.36	-
7	0.7 ± 0.2	-	1.7 ± 0.7	9.9 ± 8.3	5.8 ± 0.8	1,044
8	1.0 ± 0.4	5.7 ± 3.1	2.5 ± 1.4	14.566 ± 8.4	3.5 ± 1.6	1,708
9	1.9 ± 1.2	-	9.9 ± 2.8	66 ± 53	8.48 ± 5.34	4,983

^aThe antiviral activity of the compounds was assessed in a multiple cycle assay in MT-2 cells in 10% fetal bovine serum (FBS).

^bCompounds evaluated for serum effects in 10% FBS + 40% human serum except for 2 (10% FBS + 40% human serum + 27 mg/mL human serum albumin); values are means from experiments performed a minimum of three times (n ≥ 3).

^cDose 5 mg/kg po, n = 2; see SI for vehicle information.

^dn = 3.

Attention was then directed toward derivatization of the C-28 position of 3, where an extensive series of analogues were prepared that incorporated a wide range of functionality. Amides such as 4, incorporating a basic side chain, provided excellent potency against both WT and V370A viruses, while maintaining a low human serum shift (∼4-fold). In addition, 4 exhibited an EC₅₀ of 31 nM against the ΔV370 Gag polymorphism. However, 4 exhibited low oral exposure, attributed to a combination of poor solubility and low membrane permeability, precluding its further advancement.²⁰ With further probing of this region, the dibasic C-28 amine 5 was found to retain good polymorphic coverage, but displayed even lower oral exposure than 4, while the more lipophilic monobasic isopropyl amine 6 exhibited 3- to 5-fold reduced activity toward the polymorphic viruses, demonstrating that a single amine installed close to the core was acceptable.²¹

Based on this discovery, the introduction of a nonbasic polar group in the side chain was probed with 7, which improved the antiviral profile of 6, with EC₅₀ values of <10 nM against the 3 screening viruses (Table 1).²² However, while this combination of basicity and polarity afforded good polymorphic coverage, poor oral exposure in rats (AUC_0–6h of 7 in rats =1,044 nM·h) remained an issue. The introduction of a moderately basic amine, 1,1-dioxidothiomorpholine heterocycle, pK_a = 4.5,²³ provided 8, which demonstrated targeted antiviral activity and improved oral exposure in rats, although still 2- to 3-fold less than for 1 and 3. We hypothesized that installing the more basic amine closer to the lipophilic core would have the effect of shielding the NH and may have a positive impact on the permeability properties of the molecule and, ultimately, on oral exposure. This concept was initially probed with the alcohol 9, which exhibited improved oral exposure (AUC_0–6h in rats = 4,983 nM·h), although at the expense of reduced activity toward the polymorphic viruses, particularly ΔV370.²⁴ With further experimentation, merging the C-17 amine with the 1,1-dioxidothiomorpholine heterocycle of 8 provided BMS-955176 (2), a compound with an optimal combination of antiviral potency and oral exposure. This compound exhibits excellent potency toward the key Gag mutations, with EC₅₀ values of 1.9, 2.7, and 13 nM against WT, V370A, and ΔV370 viruses, respectively. Furthermore, following oral dosing of 5 mg/kg of BMS-955176 to rats, the AUC_0–6h was 4,132 nM·h, demonstrating exposure comparable to 1 (Table 1). In the presence of HS, BMS-955176 exhibited a ∼5-fold reduction in antiviral activity, reflecting moderate serum binding (determined to be 84% using an ultracentrifugation method).¹⁷ The potent antiviral activity of 2 extended to a library of clinically relevant isolates (N = 87), representing sequence variability in the 362–370 Gag region of subtype B viruses, with a mean EC₅₀ value of 3.9 ± 3.4 nM.¹⁷ In addition, other site directed Gag polymorphisms known to be resistant to 1 were introduced into a laboratory strain, with the results for a select group compiled in Table 2, illustrating that 2 retains activity toward all of these viruses.¹⁷ BMS-955176 was shown to be an MI by a series of mechanistic experiments which ruled out entry, reverse transcription, and protease as targets, while demonstrating that 2 acts late in the viral life cycle. The compound binds saturably and reversibly to HIV-1 Gag pseudoparticles and this binding is specifically displaced by 1. In addition, 2 specifically inhibits cleavage of CA/SP1 p25, resulting in inhibition of the production of capsid p24.¹⁷

Table 2. Antiviral Activity of Bevirimat (1) and BMS-955176 (2) toward HIV-1 Viruses Containing Naturally-Occurring Polymorphisms in the Gag Protein^17,a.

	EC50
	V362I	Q369H	V370M	V370A/ΔT371	ΔV371A	ΔT371
bevirimat (1)	74 ± 59	7.0 ± 2.0	1,810 ± 190	1,114 ± 119	40 ± 48	77 ± 97
BMS-955176 (2)	4.5 ± 2.2	1.9 ± 0.9	2.8 ± 0.3	3.6 ± 3.9	2.0 ± 0.1	7.3 ± 3.9

^aValues are means from experiments performed a minimum of three times (n ≥ 3).

The metabolic stability of 2 in vitro was high, with a t_1/2 of >120 min when incubated in human, rat, and dog liver microsomes in the presence of NADPH, predicting low clearance in vivo. However, in monkey liver microsomes, 2 exhibited high metabolic turnover, with a t_1/2 of only 12 min. It was not possible to accurately assess the in vitro permeability of 2 due to limited solubility under the assay conditions, but it is presumed to be satisfactory, since 2 demonstrated good oral exposure in rats, with an AUC_total of 14,198 nM·h after a dose of 5 mg/kg (Table 3). In this experiment, the C₂₄ of 2 was 128 nM, 21-fold higher than that of 1 under comparable circumstances. The oral bioavailability (%F) of 2 ranged from low in the dog and cynomologus monkey, reflecting the reduced metabolic stability in the latter species, to moderate in mice and rats, with long t_1/2 values in all species tested. Allometric scaling and mean residence time methodology using data from all four species were used to predict the human PK profile of 2, which, combined with a targeted C_trough value of 281 nM (3-fold the protein binding-adjusted EC₉₀ vs the ΔV370 virus), indicated that 2 may be suitable for once-daily administration of reasonable low doses (≤120 mg). Efficacy at these doses has been confirmed in clinical studies.²⁵ Safety profiling of BMS-955176 showed no evidence of mutagenicity in an Ames reverse mutation assay, and low potential for off-target liabilities based on evaluation in a panel of receptor, ion channel, and enzyme activity assays. There was also a low potential for drug–drug interactions based on CYP450 human microsomal inhibition assays.

Table 3. In Vivo Pharmacokinetic Properties of BMS-955176 in Mice,^a Rats,^b Dogs,^c and Cynomolgus Monkeys^d.

	F (%)	AUCtotal (nM·h)	C24 (nM)	IV t1/2 (h)
Mouse	27	28,912	256 ± 78	8.9
Rat	26	14,198 ± 1,083	128 ± 27	6.6
Dog	8.5	29,358 ± 16,436	559 ± 345	31.7
Cyno	3.9	995 ± 318	17.4 ± 10.9	8.6

^aOral vehicle (solution): 90% polyethylene glycol 300 (PEG 300), 10% EtOH, n = 3; dose = 5 mg/kg PO.

^bOral vehicle (solution): 84.5% PEG 300, 10% EtOH, 50.1 N NaOH and 0.5% Tween 80 (TW80), n = 3; dose = 5 mg/kg PO.

^cOral vehicle (solution): 80% 50 mM citrate buffer; 20% sulfobutylether-beta-cyclodextrin (SBC), pH = 3, n = 3; dose = 1.85 mg/kg.

^dOral vehicle (solution): 80% 50 mM citrate buffer; 20% SBC, pH = 3, n = 3; dose = 2 mg/kg.

The synthesis of 2 (BMS-955176) is shown in Scheme 1.²² Commercially available 10 was treated with benzyl bromide in the presence of K₂CO₃, followed by oxidation with PCC to afford ketone 11, which was converted into the enol triflate 12. Suzuki coupling of 12 with (4-(methoxycarbonyl)phenyl)boronic acid afforded 13. The C-28 carboxylic acid in 13 was selectively deprotected using tert-butyldimethylsilane in the presence of Pd(OAc)₂ to afford silyl ester 14, which upon treatment with TBAF generated 15. Curtius rearrangement of 15 using DPPA afforded the C-17 primary amine 17 in a process that could be carried out either in a single operation or stepwise via isolation of the corresponding C-28 isocyanate 16. Derivatization of the primary amine in 17 turned out to be a challenge, presumably due to a combination of steric hindrance and decreased nucleophilicity of the N atom caused by its proximity to the steroid core. A 3-step procedure was initially developed to install the side chain present in 2 via reductive amination of 17 using tert-butyl 2-oxoethylcarbamate in the presence of NaBH(OAc)₃ and Ti(OiPr)₄, which rendered 18. The BOC protecting group was removed using HCl to provide the C-17 primary amine 19. In situ trapping of 19 via a double Michael addition to divinylsulfone afforded 20, with an overall yield of 38% for the 3-step procedure. Ultimately, the reductive amination/hydrolysis/Michael addition sequence was replaced by a one-step procedure in which 17 was alkylated with 4-(2-chloroethyl)thiomorpholine 1,1-dioxide to afford 20 in 73% yield. This combination of base and solvent appeared to be optimal in order to achieve the complete conversion of 17 to 20. Final hydrolysis of 20 to unmask the carboxylic acid was performed with aqueous NaOH to afford 2 (BMS-955176). Overall, 2 was prepared in 10 steps from 10 with a total yield of 20%.

Scheme 1. Synthesis of 2 (BMS-955176).

Reagents and conditions: (a) K₂CO₃, BnBr, DMF, 60 °C, 3.5 h, 99%; (b) PCC, CH₂Cl₂, 6 h, 96%; (c) KHMDS, PhNTf₂, THF, −78 °C, 4 h, 90%; (d) (4-(methoxycarbonyl)phenyl)boronic acid, Na₂CO₃, Pd(Ph₃P)₄, 1,4-dioxane/i-PrOH/H₂O, reflux, 14.5 h, 68.1%; (e) TBDMSH, Pd(OAc)₂, TEA, DCE, 60 °C, 2 h, 97%; (f) TBAF, H₂O/1,4-dioxane, rt, 4 h, 99%; (g) DPPA, TEA, 1,4-dioxane, 100 °C, 5 h, 93%; (h) conc. HCl, THF, rt, 72 h, quant.; (i) Na(OAc)₃BH, Ti(OiPr)₄, DCE, rt, 19 h 79%; (j) HCl, 1,4-dioxane, rt, 4 h, 95%; (k) TEA, 1,4-dioxane/EtOH, 85 °C, 3 h, 91%; (l) K₃PO₄, KI, MeCN, 115–125 °C, 48 h, 73%; (m) NaOH, 1,4-dioxane, 78 °C, 3 h, 77%.

In conclusion, BMS-955176 is a potent HIV-1 inhibitor in cell culture that, unlike 1, exhibits broad spectrum antiviral effects that encompass the V370A- and ΔV370-containing polymorphic viruses. In addition, BMS-955176 exhibits low serum binding, which translates into a modest 5-fold effect on potency in vitro, and preclinical PK predictive of once-daily dosing in humans. In a Phase IIa clinical trial, 10-days of monotherapy with 2 administered daily to treatment-naive and treatment-experienced subjects infected with HIV-1 subtypes B or C was generally safe and well-tolerated and demonstrated >1 log₁₀ reduction in viral RNA.^18,25 BMS-955176 is currently being evaluated in a Phase IIb clinical study as a part of a treatment regimen with mechanistically different antiretroviral agents.

Glossary

Abbreviations

AUC

area under the curve

BA

betulinic acid

CA

capsid

cART

combination antiretroviral therapy

Cyno

cynomologous monkey

DCE

dichloroethane

DCM

methylene chloride

DME

1,2-dimethoxyethane

DMF

dimethylformamide

F

oral bioavailability

FBS

fetal bovine serum

HS

human serum

KHMDS

potassium bis(trimethylsilyl)amide

LANL

Los Alamos National Laboratory

MA

matrix

MI

maturation inhibitor

MeOH

methanol

MOA

mechanism of action

NC

nucleocapsid

NFV

nelfinavir

PCC

pyridinium chlorochromate

PEG

polyethylene glycol

PK

pharmacokinetic

PO

per os

rt

room temperature

SP

spacer peptide

TBDMSH

tert-butyldimethylsilylhydride

TEA

trimethylamine

TFA

trifluoroacetic acid

THF

tetrahydrofuran

TW

tween

VLR

viral load reduction

WT

wild-type

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.6b00010.

• Characterization data for key compounds and experimental procedures for the preparation of 2, as well as descriptions of biologic and pharmacokinetic assays (PDF)

^# Deceased April eighth, 2016.

The authors declare the following competing financial interest: The authors were employees of Bristol-Myers Squibb at the time this work was completed.

Dedication

This article is dedicated to the memory of Beata Nowicka-Sans, who passed away on April 8th, 2016.