Retro-binding thrombin active site inhibitors: Identification of an orally active inhibitor of thrombin catalytic activity

Article abstract of DOI:10.1016/S0960-894X(02)00651-0

A series of retro-binding inhibitors of human α-thrombin was prepared to elucidate structure-activity relationships (SAR) and optimize in vivo performance. Compounds 9 and 11, orally active inhibitors of thrombin catalytic activity, were identified to be efficacious in a thrombin-induced lethality model in mice.

Full text of DOI:10.1016/S0960-894X(02)00651-0

Bioorganic & Medicinal ChemistryLetters 12 (2002) 3183–3186
Retro-Binding Thrombin Active Site Inhibitors:
Identification of an Orally Active Inhibitor
of Thrombin Catalytic Activity
Edwin J. Iwanowicz,* S. David Kimball, James Lin, Wan F. Lau, W.-C. Han,
TammyC. Wang, Daniel G. M. Roberts, W. A. Schumacher, Martin L. Ogletree
and Steven M. Seiler
Bristol-Myers Squibb Pharmaceutical Research Institute, Princeton, NJ 08543-4000, USA
Received 15 March 2002; accepted 23 July2002
Abstract—A series of retro-binding inhibitors of human a-thrombin was prepared to elucidate structure–activityrelationships
(SAR) and optimize in vivo performance. Compounds 9 and 11, orally active inhibitors of thrombin catalytic activity, were iden-
tified to be efficacious in a thrombin-induced lethalitymodel in mice.
# 2002 Elsevier Science Ltd. All rights reserved.
Thrombin, a serine protease product in the blood coa-
gulation cascade, plays a central role in hemaostasis and
thrombosis.1 Thrombin cleaves fibrinogen to form fibrin
and activates Factor XIII, which crosslinks and stabilizes
the formed clot. Bylimited proteolysis, thrombin activates
Factors V, VIII, and XI, which promote further thrombin
production. Through the activation of protease-activated
receptors (PARs), thrombin induces platelet shape
change, aggregation, and secretion.2 In addition,
thrombin is mitogenic for vascular smooth muscle and
can activate endothelial cells, therebypromoting cell
adhesion. Through proteolytic activity, thrombin plays
a major role in arterial and venous thrombosis; there-
fore, inhibition of thrombin catalytic activity continues
to be a target for developing new therapeutic agents.3
In this report, we describe studies that led to the identi-
fication of 11, a retro-binding orallyactive inhibitor of
thrombin catalytic activity.
The synthesis of compound 1 is representative of the
syntheses carried out to prepare compounds 2–10
(Scheme 1).
The studies reported herein focus on the enhancement
of the metabolic stabilityfor the retro-binding series of
inhibitors. The C-terminal methyl ester was viewed as a
significant metabolic liabilityand was the focus of our
primaryefforts. Compound 1, with optimal residues for
N1 and N2, was one of the most potent analogues pre-
pared in this series.5,9 In earlystudies with less potent
analogues, replacement of the N3 phenyl moiety with
cyclohexyl yielded a 10-fold enhancement in potency. In
combination with p-nitrophenylanaline (N1) and allo-
threonine (N2), replacement of the N3 phenyl with
cyclohexyl gave analogues that had similar binding affi-
nityfor thrombin (Table 1). In addition, a varietyof
functionalities at the C-terminus were tolerated calling
into question the need for functionalityat this site.
Earlyefforts directed toward preparing inhibitors of
thrombin catalytic activity were patterned after sub-
strates and naturallyoccurring inhibitors. Notabl,y
potent small molecule inhibitors that bound in a similar
fashion to the N1–N3 residues of hirudin were unknown
4,5
until disclosure byour group.
Molecular modeling
studies supported the binding hypothesis that was later
proven bysolving the crystallographic structure of the
ternarycomplex among human a-thrombin, hirugen,
and BMS-183,507.6ꢀ8
Removal of the N3 residue, affording compound 6, led
to a four order of magnitude loss in potency, (Table 2).
Analogues prepared from both optical antipodes of
a-methyl benzyl amine and a-methyl phenethylamine
were examined for thrombin inhibition. Compounds 7
*Corresponding author. Tel.:+1-609-252-5416; fax:+1-609-252-6601;
e-mail: iwanowie@bms.com
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00651-0

3184
E. J. Iwanowicz et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3183–3186
nine side chain to be the lowest energyof the available
conformations. Subsequent modeling studies concluded
that the N-methylation of the amide nitrogen of the Phe
residue would be expected to render as lowest energythe
bound rotomeric conformation of the allo-thronine side
chain while introducing two energeticallysimilar roto-
meric populations about the amide bond (Fig. 1).
Compound 11 was prepared as shown in Scheme 1 uti-
lizing 13 and was shown to be is a potent inhibitor of
thrombin catalytic activity displaying classic competitive
kinetics with a Ki of 7.0 nM.11 In addition, this analogue
displays excellent selectivity for the thrombolytic
enzymes plasmin and t-PA (Table 3).12 Data for GYKI-
14,766 is given for comparison. In addition, a modified
thrombin time (TT) was used to determine the direct
inhibition of thrombin activityin a protein rich environ-
ment.13 In this assay, 11 doubled clotting time (in vitro) at
0.085 mM (GYKI-14,776: 0.40 mM) (Table 4).14 Com-
pound 11 was chosen for further evaluation in vivo.
Scheme 1. Synthesis of compound 1: (a) EDAC, HOBt, l-4-Nitro-
Phe(OMe), DMF, 0 ꢁC, 82%; (b) TFA, CH2Cl2, 0 ꢁC, 95%; (c)
EDAC, HOBt, l-4-Nitro-Phe(OMe), DMF, 0 ꢁC, 89%; (d) TFA,
CH2Cl2, 0 ꢁC, 95%; (e) EDAC, HOBt, BocNH(CH2)3C(O)OH, DMF,
0 ꢁC, 88%; (f) TFA, CH2Cl2, 0 ꢁC, quant; (g) Et3N, H2NC(SO3H)NH,
DMF, 0–20 ꢁC, 79%.
GYKI-14,776 and 11 were examined in a thrombin-
13
induced lethalitymodel in anesthetized mice.
Two
modes of administration were examined, iv and oral.
Compound 11 is a veryeffective agent at 10 min and 1 h
when given byiv administration. This analogue is also
effective when dosed orallywith an ED 50 and 44 mg/kg.
and 9 having the same relative stereochemistryas l-
phenylalanine were found to be the more potent dia-
steromers. However, neither analogue was as potent as
the preferred analogues described in Table 1. Metabolic
stabilitystudies utilizing human and mouse liver S-9
fractions identified cleavage of the amide bond between
the N1 and N2 residues as the predominate mode of
degradation for compound 7.
In conclusion, combinations of the preferred N1–N3
residues, previouslyidentified, led to the most potent
analogues in this retro-binding series. The terminal
methyl ester may be removed with a modest loss in
potencyand the major metabolic liability, cleavage of
the amide bond between N1 and N2, was effectively
The crystallographic structure of the ternary complex
among human a-thrombin, hirugen and BMS-183,507,
showed the NH of the amide bond linking N1 and N2 is
directed toward solvent. Potentiallythe amide linkage
maybe alkylated in an attempt to reduce the rate of
enzymatic cleavage. The impact of this change, upon the
bound conformation of the inhibitor, must be considered.
Molecular modeling studies of BMS-183,507 had shown
that the bound rotomeric configuration of the allo-threo-
Table 2. SAR from modification of the N3 position
Table 1. N3 SAR
Compd
R
Thrombin inhibition IC50 (nM)
230,000
6
-H
7
8
43ꢂ6.0
260ꢂ33
Compd
R
X
Thrombin inhibition10
IC50 (nM)
9
72ꢂ6.0
310ꢂ31
1
2
3
4
5
-Ph
-Ph
-Chx
-Chx
-Chx
-C(O)OMe
-CH2OH
-C(O)OMe
-CH2OH
-CO2H
8.3ꢂ2.0
19ꢂ2.0
16ꢂ3.0
19ꢂ2.0
55ꢂ22
10

E. J. Iwanowicz et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3183–3186
3185
Salzman, E. W., Hirsh, J. Eds. Lippincott: Philadelphia, 1994;
pp 3–18.
2. Coughlin, S. R. Nature 2000, 407, 258.
3. Vacca, J. P. Curr. Opin. Chem. Biol. 2000, 4, 394.
4. Sanderson, P. E. J. Annu. Rep. Med. Chem. 2001, 36, 79.
5. Iwanowicz, E. J.; Lau, W. F.; Lin, J.; Roberts, D. G. M.;
Seiler, S. M. J. Med. Chem. 1994, 37, 2122.
6. Lau, W. F.; Tabernero, L.; Sack, J. S.; Iwanowicz, E. J.
Bioorg. Med. Chem. Lett. 1995, 3, 1039.
7. Tabernero, L.; Chang, C. Y.; Ohringer, S. L.; Lau, W. F.;
Iwanowicz, E. J.; Han, W.-C.; Wang, T. C.; Seiler, S. M.;
Roberts, D. G. M.; Sack, J. S. J. Mol. Biol. 1995, 246, 14.
8. The majorityof thrombin inhibitors and substrates bind in
a manner that allows the formation of a pair of hydrogen-
bonds to Gly216 in a substrate-like or anti-parallel orienta-
tion. d-Phe-Pro-Arg-derived inhibitors such as GYKI-14,766,
bind to thrombin forming this antiparallel arrangement with
Gly216. In contrast, 1 forms a parallel hydrogen-bond
arrangement, a keyfeature of the retro mode of binding.
Figure 1. Newman projection of the b-hydroxy side chain of the N2
residue as determined from the inhibitor/protein crystallographic
studies. The curve arrow signifies the two energeticallysimilar roto-
meric populations about the amide bond when the amide nitrogen is
alkylated with a methyl residue.
Table 3. Inhibition of keyserine proteases by11 and GYKI-14,766
[IC50 (mM)]
Compd
Thrombin
Factor Plasmin t-PA Trypsin
Xa
11
0.030ꢂ5
(Ki=7.0 nM)
0.008b
16
25
34
0.045
GYKI-14,766
4.6
0.23
0.70
0.008a
aGYKI-14,766 was found to exhibit slow tight binding kinetics for the
inhibition of thrombin and trypsin.15
9. N1-3 terminologyis used to signifythe three amino acids
bound in the active site of thrombin where the N terminus of
N1 is proximal to Ser 195.
Table 4. Efficacy(ED 50) in the mouse thrombin challenge model for
BMS-190,623 and GYKI-14,766. Dosages are given in mg/kg
10. Chromogenic substrate assay: The inhibitors were incu-
bated in an assaybuffer (145 mM NaCl, 5 mM KCl, 1 mg/mL
poly(ethylene glycol) (PEG-8000), 30 mM N-(2-hydroxy-
ethyl)piperazine-N0-2-ethanesulfonic acid, pH 7.4) with human
a-thrombin (0.1 unit/mL) at room temperature for 3 min. The
change in optical densitywas measured at 405 nm. A kinetic
microplate reader (Molecular Devices Corporation Vmax) was
used to measure the change in optical densityover time. The Km
for S-2238 under these conditions is 2.5 mM. Each value repre-
sents the average of at least triplicate determinations.
11. N-Methyl allo-threonine containing intermediate 13 was
prepared as given below:
Compd
iv
10 min
iv
1 h
oral
1 h
11
GYKI-14,766
0.04
0.21
3.0
2.2
44
23
addressed through methylation of the amide nitrogen.
Compound 11 is a potent inhibitor of thrombin catalytic
activitywith a Ki of 7.0 nM. Notablythis analogue is
highly selective over the thrombolytic enzymes plasmin
and t-PA. In vivo studies in a mouse model utilizing
human a-thrombin showed that 11 was effective when
administered iv or orally. We are continuing to explore
the implications of the retro-binding motif to the design
of new inhibitors. In the future, we will report on
extensions of this work to other serine proteases.
A solution of 12 (4.04 g, 12.0 mmol, 1.0 equiv), formaldehyde
(0.024 M of 37% an aqueous solution in methanol, 500 mL,
1.0 equiv), diisopropylethylamine (2.4 mL, 13.8 mmol, 1.2
equiv) and palladium hydroxide on carbon (1.24 g, 31% by
weight) was stirred inder an atmosphere of hydrogen for 3.5 h.
After flushing with nitrogen and filtering though a pad of
References and Notes
1. Colman, R. W.; Marder, V. J.; Salzman, E. W.; Hirsh, J. In
Hemostasis and Thrombosis. Colman, R. W., Marder, V. J.,

3186
E. J. Iwanowicz et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3183–3186
Celite, the solution was concentrated in vacuo, dissolved in
chloroform and washed twice with saturated aqueous sodium
bicarbonate. The organic layer was dried over sodium sulfate
and concentrated in vacuo to give a white solid (2.68 g, 95%
yield as a 9:1 mixture of the mono- and di-methylated amines.
This mixture was utilized without further purification.
12. Inhibition of bovine pancreatic trypsin was measured in an
assaycontaining 2 mM CaCl 2, 50 mM Tris pH 8.0 and 30 mM
carbobenzyloxy-Val-Gly-Arg-pNA as substrate. The inhibitor
was combined with trypsin assay buffer and incubated for 3
min, after which the substrate was added and absorption was
mg/mL polyethylene glycol (PEG-8000), 30 mM N-(2-hydro-
xyethyl)piperazine-N0-ethanesulfonic acid, pH 7.4 and 100 mM
methanesulfonyl-d-cyclohexyltyrosyl-Gly-Arg-pNA as sub-
strate. The inhibitor was incubated with the enzyme for 3 min,
after which the substrate was added and the rate of hydrolysis
was measured as for thrombin. Inhibition of Factor Xa was
measured in an assaycontaining 145 mM NaCl, 5 mM KCl, 1
mg/mL polyethylene glycol (PEG-8000), 30 mM N-(2-hydro-
xyethyl)piperazine-N0-ethanesulfonic acid, pH 7.4 and 100 mM
S-2222. The inhibitor was incubated with the enzyme for 3
min, after which the substrate S-2222 was added and the rate
of hydrolysis was measured as for thrombin. The IC50 values
are reported with corresponding standard deviations.
13. 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.
14. Concentration of inhibitor required to double clotting
time in human plasma. Tests fibrinogen conversion to fibrin
13
and is performed byadding thrombin to citrated plasma.
measured using a microplate reader (Molecular devices Vmax
)
at 405 nM. Inhibition of human plasmin was measured in an
assaycontaining 145 mM NaCl, 5 mM KCl, 1 mg/mL poly-
ethylene glycol (PEG-8000), 30 mM N-(2-hydroxy-
ethyl)piperazine-N0-ethanesulfonic acid, pH 7.4 and 100 mM S-
2251 (d-Val-Leu-Lys-pNA). The inhibitor was incubated with
the enzyme for 3 min, after which the substrate was added and
the rate of hydrolysis was measured as for thrombin and tryp-
sin. Inhibition of tissue plasminogen activator (t-PA) was
measured in an assaycontaining 145 mM NaCl, 5 mM KCl, 1
15. Morrison, J. F.; Walsh, C. T. Advances in Enzymology;
Meister, A., Ed.; Wiley: New York, 1988; Vol. 61, p 210–301.