Molecular design and structure - Activity relationships leading to the potent, selective, and orally active thrombin active site inhibitor BMS-189664

Article abstract of DOI:10.1016/S0960-894X(01)00667-9

A series of structurally novel small molecule inhibitors of human α-thrombin was prepared to elucidate their structure-activity relationships (SARs), selectivity and activity in vivo. BMS-189664 (3) is identified as a potent, selective, and orally active reversible inhibitor of human α-thrombin which is efficacious in vivo in a mouse lethality model, and at inhibiting both arterial and venous thrombosis in cynomolgus monkey models.

Full text of DOI:10.1016/S0960-894X(01)00667-9

Bioorganic & Medicinal Chemistry Letters 12 (2002) 45–49
Molecular Design and Structure–Activity Relationships Leading
to the Potent, Selective, and Orally Active Thrombin Active Site
Inhibitor BMS-189664
Jagabandhu Das,* S. David Kimball,* Steven E. Hall,y Wen-Ching Han,
Edwin Iwanowicz, James Lin, Robert V. Moquin, Joyce A. Reid, John S. Sack,
Mary F. Malley, Chiehying Y. Chang, Saeho Chong, David B. Wang-Iverson,
Daniel G. M. Roberts, Steven M. Seiler, William A. Schumacher and Martin L. Ogletree
Bristol-Myers Squibb Pharmaceutical Research Institute, Princeton, NJ 08543-4000, USA
Received 16 July 2001; accepted 14 September 2001
Abstract—A series of structurally novel small molecule inhibitors of human a-thrombin was prepared to elucidate their structure–
activity relationships (SARs), selectivity and activity in vivo. BMS-189664 (3) is identified as a potent, selective, and orally active
reversible inhibitor of human a-thrombin which is efficacious in vivo in a mouse lethality model, and at inhibiting both arterial and
venous thrombosis in cynomolgus monkey models. # 2001 Elsevier Science Ltd. All rights reserved.
Development of direct acting thrombin inhibitors1 has
been an area of intense research over the past decades.
In the accompanying communication1 we outline our
structure based drug design approach leading to a novel
template as exemplified by 1 (Fig. 1). Further structural
modification led to the identification of 2 as a highly
potent, selective, and reversible thrombin active site
inhibitor. In this communication, we describe studies
based on the same template, that have culminated in the
discovery of 3 as a potent, selective, and orally active
thrombin inhibitor with excellent antithrombotic activ-
ity in vivo. The synthesis, SAR and biological studies
leading to 3 (BMS-189664) are described briefly.
The synthesis of this class of compounds follows a gen-
eral synthetic route that is illustrated in Scheme 1.
Coupling of BOC-d-phenylalanine 4 and benzyl l-pro-
linate 5 under standard amide coupling conditions fol-
lowed by hydrogenolysis of the benzyl ester afforded the
acid 6. The acid was coupled with bis-BOC-guanidine 72
Figure 1. Schematic representation of the inhibitor design from 1.
*Corresponding author. Tel.: +1-609-252-5068; fax: +1-609-252-
6804; e-mail: jagabandhu.das@bms.com
yCurrent address: Sphinx Pharmaceutucals, Eli Lilly & Co., Durham,
NC 27707, USA.
Scheme 1. Synthesis of 3: (a) EDAC, HOBt, DMF, NMM; (b) 10%
Pd–C, H2, EtOH; (c) CH2Cl2, CF3COOH; (d) CH2Cl2, Et3N,
CH3SO2Cl.
0960-894X/02/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00667-9

46
J. Das et al. / Bioorg. Med. Chem. Lett. 12 (2002) 45–49
Table 1. SAR of amino acid modifications at P3
Compd
X
Thrombin inhibition IC50 (mM)3
Compd
X
Thrombin inhibition IC50 (mM)3
9
1
10
11
12
13
Gly
l-Ser
l-Threo
l-Lys
l-Phe
l-Gln
2.9
2.8
3.0
4.9
12.0
2.9
14
15
16
17
18
19
l-allo-Thr
l-Asp
3.5
5.2
1.4
d-Phe
d-Gln
26.6
8.9
>333
d-Trp
d-Glu
Table 2. SAR from the sulfonamide modification at P3
Compd
R
Thrombin inhibition IC50 (mM)3
Compd
R
Thrombin inhibition IC50 (mM)3
20
21
H
CH3SO2
0.54
1.4
23
24
CH3CO
PhCH2OCO
42
0.67
22
CF3SO2
110
16
1.4
under standard conditions to form an amide, which was
treated with trifluoroacetic acid in dichloromethane to
form amino-guanidine 8. Reaction of 8 with methane-
sulfonyl chloride in dichloromethane in presence of
triethyl amine afforded 3 (BMS-189664) after purifica-
tion by reverse-phase preparative HPLC. This synthetic
route is quite general and was used for the synthesis of
other analogues in this series.
the d-Phe analogue 16 was selected for additional
structure–activity studies.
Table 2 outlines the SAR observed with the modifica-
tion of the sulfonamide functionality at P3. The data
demonstrate that the naphthalene sulfonamide group is
not an absolute requirement for thrombin inhibitory
activity. The corresponding amine 20 and the benzyl
carbamate 24 are roughly 2-fold more potent while the
methylsulfonamide 21 is equipotent to 16. Both
trifluoromethylsulfonamide (22) and acetamido (24)
groups are not tolerated at this position.
Compound 3 (BMS-189664) and its analogues were
tested for their ability to inhibit thrombin hydrolysis3 of
the chromogenic substrate S-2238 (IC50, Tables 1–5).
With 1 (BMS-185367) as a lead we studied the effect of
amino acid substitution at P3, summarized in Table 1.
Most of the amino acid modifications (1 and 10–18) are
well tolerated with minimal effect in thrombin inhibi-
tory activity in vitro. In contrast, the d-amino acid Glu
is detrimental to activity in vitro. Because of its potency,
In order to further optimize the potency, we investi-
gated the modification of the guanidine moiety that
binds to the Asp189 residue in the P1 specificity pocket
of thrombin (Table 3). Replacement of the guanidine
group by an imidazole (25, and 27) significantly reduced
the potency, while a benzyl amine replacement (26) was
Table 3. SAR from the P1–P2 linker modification
Compd
R
Thrombin inhibition IC50 (mM)3
Compd
R
Thrombin inhibition IC50 (mM)3
25
108
27
134
26
4.4
3
0.046

J. Das et al. / Bioorg. Med. Chem. Lett. 12 (2002) 45–49
Table 4. SAR from the sulfonamide and guanidine modification
47
Compd
R
R’
Thrombin inhibition IC50 (mM)3
Mouse ID50 (mpk4 iv)
Mouse ID50 (mpk4 po)
3
Me
PhCH2
Me
H
H
OH
OH
0.046
0.01
0.07
0.19
0.24
0.17
0.3
8.8
23
10.5
18
28
29
30
PhCH2
0.002
Table 5. Enzyme selectivity of selected thrombin inhibitors
Compd
Thrombin
IC50 (mM)4
Trypsin
(relative to thrombin)a
Plasmin
(relative to thrombin)a
tPA
(relative to thrombin)a
Factor Xa
(relative to thrombin)a
3
30
0.46
40
60
658
0.4
1000
700
4900
13,000
2100
700
6900
200
>7200
14,000
>1700
300
0.002
0.038
0.018
Argatroban
Efegatran
aSelectivity ratio: IC50/IC50 (thrombin).
tolerated. More importantly, introduction of a cyclic
guanidine moiety (3) resulted in a 30-fold increase in
potency, which we attribute to increased hydrophobic
interactions of the piperidine ring with residues in the
specificity pocket of thrombin.
analogue in this model in vivo when dosed by either the
oral or intravenous route. Despite its superior potency
in vitro, 30 is roughly 2-fold less potent than 3 in vivo.
The in vivo oral activity of 3 (BMS-189664, ID50=8.8
mpk) is significantly superior to that of Efegatran
(ID50=23 mpk) and Argatroban (ID50>100 mpk).
Further optimization of 3 focused on modification of
the sulfonamide and guanidine functionalities and is
reported in Table 4. Replacement of the methylsulfona-
mide group with benzylsulfonamide (28) resulted in a 5-
fold increase in potency, while the replacement of the
guanidine residue of 3 by hydroxyguanidine (29) had
minimal effect on potency. In contrast similar hydroxy-
guanidine replacement in the benzylsulfonamide analo-
gue 28 resulted in a 5-fold increase in thrombin
inhibitory activity (30, IC50=2 nM).
Separate studies5 were performed to verify the Ki of 3
and 30. Compound 30 (BMS-191032) displayed classic
competitive kinetics for thrombin inhibition with a Ki of
460 ꢀ 35 pM and was identified as the most potent
thrombin inhibitor in this series. The Ki of 3 (BMS-
189664) determined under these conditions was
8.17 ꢀ 0.78 nM, which is similar to 9.32 nM determined
from IC50.
The two most potent compounds in this series, 3 and 30,
were tested for their selectivity for thrombin over other
serine proteases.1 Both 3 and 30 displayed excellent
selectivity for thrombin relative to other serine proteases
important to hemostasis and thrombolysis (Table 5).
The selectivity for thrombin over trypsin varies from 40-
to 60-fold and the selectivities over plasmin and tPA
vary from 700- to greater than 10,000-fold.
Compounds 3 and 28–30 were evaluated further for
activity in vivo in a thrombin-induced mouse lethality
model.4 Compound 3 (BMS-189664) is the most potent
The crystal structures of the ternary complexes of
human a-thrombin with hirugen and thrombin inhibitors
3 (BMS-189664) and 30 (BMS-191032) were determined
at a resolution of 2.6 A and 2.4 A, respectively, by a
procedure described earlier6 (Fig. 2). Both inhibitors
bind to thrombin in an antiparallel fashion. In this
binding mode, the inhibitor backbone lies next to the
extended thrombin backbone Ser214-Glu217 and forms
a pair of hydrogen bonds to Gly216. This binding mode
is very similar to the ones reported for Argatroban7 and
d-Phe-Pro-ArgCH2Cl (PPACK).8 The proline group of
the inhibitors binds to the proximal P-pocket (also
referred to as S2 apolar hydrophobic pocket) and makes
contact with Tyr60A and Trp60D. The two nitrogens of
Figure 2. Superimposed conformations of 3 (yellow) and 30 (purple)
bound to the a-thrombin/hirugen complex.

48
J. Das et al. / Bioorg. Med. Chem. Lett. 12 (2002) 45–49
Table 6. Single-dose pharmacokinetic parameters of BMS-189664
Parameters
Monkeys
Beagle dogs
Parameters
Monkeys
Beagle dogs
IV Route
Dose (mmol/kg)
Half-life (h)
Clearance (mL/min/kg)
MRT (h)
Vss (L/kg)
13
(n=3)
6.4
6.4
1.8
12
(n=7)
9a
13.4
1.1
Oral route
Dose (mmol/kg)
Cmax (mM)
Tmax (h)
50
(n=3)
5.6
2.3
6
17
28
(n=7)
3
0.6
4.0
MRT (h)
F (%)
0.7
0.8
15
aValue derived from three dogs that had reliable terminal phase.
the guanidino moiety (3), are nestled in the specifity
pocket and make hydrogen bonding contacts with
Asp189. The hydroxyguanidine function in 30 binds in a
very similar fashion. The d-phenylalanine moieties lie in
the distal d-pocket formed by the side chains of Trp215,
Ile174 and Glu97A-Leu99, while the methyl sulfon-
amido (3) or the benzylsulfonamido (30) groups cover
the specificity pocket of the enzyme. The more pro-
nounced thrombin inhibitory potency of 30 (23-fold
more potent in vitro than 3) may be attributed in part to
the increased hydrophobic interactions between the
benzyl group and the residues at the surface of the
specificity pocket.
antagonist since it does not inhibit [3H]SFFLRR-NH2
binding (IC50> >100 mM) to human platelet mem-
branes. BMS-189664 also did not inhibit SFLLRNP-
stimulated platelet aggregation.
The pharmacokinetic profile of BMS-189664 was deter-
mined in several animal species (Table 6). Single iv and
oral doses of BMS-189664 were given to dogs and
cynomolgus monkeys. BMS-189664 concentrations in
plasma were determined using an LC/MS assay. Plasma
concentrations declined with a mean elimination half-
life of >6 h in both species. The oral bioavailability of
BMS-189664 based on the dose-normalized AUCs after
oral and iv dosing was 15 and 17% in dogs and monkeys,
respectively.
Because of its superior activity in vivo in the anesthe-
tized mouse model BMS-189664 (3) was characterized
further in vitro and in vivo. A modified thrombin time
(TT) was used to determine the direct inhibition of
thrombin activity in a protein-rich plasma environment.
In this assay, BMS-189664 doubled thrombin clotting
time in vitro at 51ꢀ9 nM (n=3). In studies of human
gel-filtered platelet aggregation stimulated by thrombin,
BMS-189664 inhibited thrombin induced platelet
aggregation with a pA2 value of 8.43ꢀ0.04 (n=3) with
a slope near unity (0.89ꢀ0.06). In these studies, BMS-
189664 is not likely to be acting as a thrombin receptor
In pentobarbital-anesthetized cynomolgus monkeys,
BMS-189664 inhibited arterial and venous thrombosis
at doses causing small increases in bleeding time and
more pronounced prolongation of ex vivo clotting time.
Thrombosis was induced in stenotic carotid arteries
using the Folts model9 as detailed in a representative
experiment (Fig. 3). Arterial thrombosis was disrupted
in six out of eight monkeys at a threshold dose of
1.5ꢀ0.2 mg/kg (cumulative iv dose), which increased
the APTT to 3.7ꢀ0.2 times control. Vehicle was with-
out effect in five monkeys. In other monkeys thrombosis
was induced by topical application of filter paper satu-
rated with 25% FeCl2 for 3 min to a vena cava in which
flow was impaired with a stenosis. Thrombus weight
was reduced by 46, 88, and 90% by BMS-189664 doses
of 0.5, 2, and 6 mg/kg, respectively (cumulative doses
achieved through continuous iv infusions for 1 h), while
Figure 3. BMS-189664 inhibited arterial thrombosis in monkeys. Pla-
telet-mediated reductions in blood flow were interrupted by momen-
tary removal of a stenosis on the crush-injured artery. This shaking
loose (SL) of the thrombus was required to prevent occlusion and
resulted in a transient recovery of blood flow. The stenosis was then
replaced to reinstate the cycle. BMS-189664 injections (0.2 mg/kg)
were repeated until spontaneous blood flow restoration (SP) negated
the need for SLto maintain stable blood flow. Once several SP were
observed, the stenosis was reset (RS) in an attempt to reestablish the
cyclic pattern. In this monkey a cumulative iv threshold dose of 1 mg/
kg produced a persistent SP pattern.
Figure 4. Dose-dependent effect of BMS-189664 on venous thrombo-
sis, bleeding time and ex vivo clotting time was determined in anes-
thetized monkeys. BMS-189664 was infused at 9, 25, and 100 mg/kg/
min for 1 h to achieve 0.5, 2, and 6 mg/kg, respectively.

J. Das et al. / Bioorg. Med. Chem. Lett. 12 (2002) 45–49
2. Bis-BOC-guanidine
49
the APTT was increased by 2.4, 3.2, and 4.3 times con-
7
was synthesized from 4-amino-
methylpiperidine (93% overall yield) by its reaction with ben-
zaldehyde (1.1 equiv, toluene, Á), followed by addition of bis-
BOC-amidinopyrazole (1 equiv, rt) and subsequent hydrolysis
(1 N aq KHSO4, rt).
trol, respectively (Fig. 4). Renal cortex bleeding time
was also determined in these monkeys before and after
thrombus formation using a Surgicutt1 device (Inter-
national Technidyne Corp., Edison, NJ, USA) to pro-
duce a 1-mm deep incision. The same BMS-189664
doses prolonged bleeding time by 8, 62, and 115%,
respectively (Fig. 4). The APTT and bleeding time were
not significantly affected in vehicle-treated monkeys.
3. In vitro inhibition of thrombin catalytic activity using 10
mM substrate s-2238 (d-Phe-Pip-Arg-pNA) was measured at rt
after 3 min incubation with inhibitor. For a description of the
assay, see: Balasubramanian, N.; St. Laurent, D. R.; Federici,
M. E.; Meanwell, N. A.; Wright, J. J.; Schumacher, W. A.;
Seiler, S. M. J. Med. Chem. 1993, 36, 300. The Ki values for
competitive inhibitors can be determined from IC50 values,
see: Cheng, Y. C.; Prusoff, W. H. Biochem. Pharm. 1973, 22,
3099. Each value represents an average of at least three deter-
minations.
4. The model involved iv challenge with human a-thrombin
(10–20 units/mouse) in anesthetized mice treated with the
inhibitor either 10 min (iv) or 60 min (po) prior to thrombin
challenge. ID50 is the dose of the inhibitor required for 50%
mice survival.
In conclusion, we have described the SAR of a structu-
rally novel series of compounds leading to BMS-189664
as a potent, highly selective, and orally active reversible
thrombin inhibitor. In addition, we have identified
BMS-191032 as the most potent thrombin inhibitor in
vitro (Ki=460ꢀ35 pM) in this series. The data provided
show that BMS-189664 is efficacious at protecting mice
from thrombin-induced lethality in vivo and at inhibit-
ing arterial and venous thrombosis in monkeys. Based
on these findings and additional pharmacokinetic and
efficacy studies in other animal models, BMS-189664
was selected for further development.
5. Ki was determined by measuring enzyme activity at three
different substrate (S) concentrations and six different inhi-
bitor (I) concentrations, curve fitting the data to V=Vmax[S]/
{Km(1+[I]/Ki)+[S]}.
6. Malley, M. F.; Tabernero, L.; Chang, C. Y.; Ohringer, S.;
Roberts, D. G. M.; Das, J.; Sack, J. S. Protein Sci. 1996, 5,
221.
7. Banner, D. W.; Hadvary, P. J. Biol.Chem. 1991, 266,
20085.
References and Notes
8. Bode, W.; Turk, D.; Karshikov, A. Protein Sci. 1992, 1,
426.
9. Schumacher, W. A.; Heran, C. L.; Goldenberg, H. J.; Har-
ris, D. N.; Ogletree, M. L. Am. J. Physiol. 1989, 256, H726.
1. Das, J.; Kimball, S. D.; Reid, J. A.; Wang, T. C.; Lau,
W. F.; Roberts, D. G. M.; Seiler, S. M.; Schumacher, W. A.;
Ogletree, M. L. Bioorg. Med. Chem. Lett. 2002, 12, 41.