Structure-based design of novel groups for use in the P1 position of thrombin inhibitor scaffolds. Part 2: N-acetamidoimidazoles

Article abstract of DOI:10.1016/j.bmcl.2008.01.098

Guided by X-ray crystallography of thrombin-inhibitor complexes and molecular modeling, alkylation of the N1 nitrogen of the imidazole P1 ligand of the pyridinoneacetamide thrombin inhibitor 1 with various acetamide moieties furnished inhibitors with sign

Full text of DOI:10.1016/j.bmcl.2008.01.098

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 2062–2066
Structure-based design of novel groups for use in the P1 position
of thrombin inhibitor scaffolds. Part 2: N-acetamidoimidazoles
Richard C. A. Isaacs,a,* Mark G. Solinsky,a Kellie J. Cutrona,a Christina L. Newton,a
Adel M. Naylor-Olsen,b Daniel R. McMasters,b Julie A. Krueger,c S. Dale Lewis,c
Bobby J. Lucas,c Lawrence C. Kuo,d Youwei Yan,d J. J. Lynche and E. A. Lylee
aDepartment of Medicinal Chemistry, Merck Research Laboratories, West Point, PA 19486, USA
bDepartment of Molecular Systems, Merck Research Laboratories, West Point, PA 19486, USA
cDepartment of Biological Chemistry, Merck Research Laboratories, West Point, PA 19486, USA
dDepartment of Structural Biology, Merck Research Laboratories, West Point, PA 19486, USA
eDepartment of Pharmacology, Merck Research Laboratories, West Point, PA 19486, USA
Received 1 November 2007; revised 14 January 2008; accepted 25 January 2008
Available online 30 January 2008
Abstract—Guided by X-ray crystallography of thrombin-inhibitor complexes and molecular modeling, alkylation of the N1 nitro-
gen of the imidazole P1 ligand of the pyridinoneacetamide thrombin inhibitor 1 with various acetamide moieties furnished inhibitors
with significantly improved thrombin potency, trypsin selectivity, functional in vitro anticoagulant potency and in vivo antithrom-
botic efficacy. In the pyrazinoneacetamide series, oral bioavailability was also improved.
ꢀ 2008 Elsevier Ltd. All rights reserved.
In a previous disclosure1 we demonstrated that simple
azoles and in particular, imidazoles, in spite of their rel-
atively weak basicity (pKa ꢀ 7) could be employed as the
S1-binding element in the design of both noncovalent
peptide and nonpeptide thrombin inhibitors by optimiz-
ing the proximity of the inhibitor’s P1 imidazole nitro-
gen to the carboxylate of Asp189 in the thrombin
active site. Additional potency enhancement was
achieved by incorporating a methyl group on the imid-
azole, putatively in part by modestly increasing the basi-
city of the imidazole and also partly by making
lipophilic contact with the aliphatic side chain of
Val213. This strategy led to the development of pyridi-
none imidazole compound 1 (Fig. 1), a potent thrombin
inhibitor (Ki = 8 nM) which is 5400-fold selective for
thrombin versus trypsin (Ki = 43 lM). The functional
in vitro anticoagulant potency of inhibitor 1 was as-
sessed by measuring the concentration of the inhibitor
required to double the activated partial thromboplastin
time (APTT) in human plasma (1.1 lM).2 It also per-
formed poorly in vivo in the rat ferric chloride efficacy
assay (3/6 animals occluded at an iv infusion rate of
10 lg/kg/min).3 By comparison, the aminopyridine
analog 2 (Fig. 1) is not only potent (Ki = 0.5 nM) and
selective (6400-fold) for thrombin versus trypsin (Ki =
3.2 lM) but shows 5-fold higher functional anticoagu-
lant activity in the 2· APTT assay (0.21 lM) and full
in vivo efficacy in the rat ferric chloride assay (0/6 occlu-
O
O2
S
N
N
H
N
H
R
O
Thrombin
Ki (nM)
8
0.04
O
0.5
H
N
H
O
N
N
R
N
NH2
Cl
1
2
3
Figure 1. Comparison of binding potency of imidazole P1-bearing
thrombin inhibitor 1 with analogs bearing more potent but structurally
distinct P1 ligands.
Keywords: Thrombin; Anticoagulant.
*
Corresponding author. Tel.: +1 215 652 0880; fax: +1 215 652
3971; e-mail: richard_isaacs@merck.com
0960-894X/$ - see front matter ꢀ 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.01.098

R. C. A. Isaacs et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2062–2066
2063
sions at an iv infusion rate of 10 lg/kg/min).4 The over-
all portfolio of properties embodied in inhibitor 2 made
it attractive as a drug development candidate. However,
the aminopyridine P1 ligand imparted unacceptable off
target activity which the imidazole P1 ligand did not.
We therefore sought to improve upon inhibitor 1. In this
paper we will describe how we used P1 targeted struc-
ture-based design techniques to further optimize 1, the
objective being to more closely match the mix of desir-
able activities and properties found in inhibitor 2.
which contain a neutral 2,5-disubstituted benzylamine
motif are very potent thrombin inhibitors. X-ray analy-
sis of thrombin bound inhibitors bearing this P1 li-
gand5b,c revealed that the lack of a direct binding
interaction with Asp189 is completely compensated for
by the presence of hydrogen bonding interactions be-
tween the oxyacetamide moiety and the NAH of
Gly219 on the thrombin b sheet. In principle, this sem-
inal finding when combined with the X-ray data from
inhibitor 2 offered a potentially attractive approach to
improving the potency of the imidazole series through
incorporating an acetamide residue onto the imidazole
ring.
We first focused on improving binding potency and
sought to take advantage of the wealth of enzyme–inhib-
itor X-ray structural information and molecular model-
ing data which we had generated in-house over time
during our work on a number of different thrombin
inhibitor scaffolds.
An overlay of the bound crystal structures of inhibitors
2 and 3 with an active site model of inhibitor 1 indicated
that there were two potential sites for incorporation of
an acetamide moiety onto the imidazole, namely N1
and C4 (Fig. 2). Based on this analysis we initially pro-
posed hybrids 4 and 5 for consideration. Prior to initiat-
ing any actual synthetic work, molecular modeling
studies were carried out on proposed hybrids 4 and 5,
the minimized structures each being compared in turn
to the X-ray structures of inhibitors 2 and 3 and the ac-
tive site model of 1. In so doing it was determined that
because of the smaller internal 108 degree bond angle
of an imidazole (as in the imidazole P1 of inhibitor 1)
relative to the larger internal 120ꢁ bond angle of a ben-
zene ring (as in the phenyl P1 of inhibitor 3), a methyl
group attached to imidazole (at either N1 or C4) occu-
pies approximately the same region of space as the
methylene group in the oxyacetamide portion of the 2-
phenoxyacetamide in 3. This led to a further refinement
A detailed X-ray analysis of the potent methylamino-
pyridine inhibitor 2 bound to the a-thrombin–hirugen
complex has been published.4 For the purposes of the
current discussion it is instructive to note that the
methyl group of the methylaminopyridine points toward
S1, making contact with the aliphatic side chain of
Val213. The amino group makes a direct ionic interac-
tion with the Asp189 carboxylate. An ordered water
molecule forms a bridge between the pyridine nitrogen
and the Asp189.
Previous disclosures from these laboratories have dem-
onstrated that it is possible to design potent P1 ligands
which bind without any ionic interaction with
Asp189.5 In particular, inhibitors such as 3 (Fig. 1)
O
O2
H
N 1
S
N
H
N
N
H
O
1
N
4
H
N
O
O
make hybrids to
increase potency
N
2
NH2
(X-ray driven design)
3
Cl
H
N
R
H
N
O
O
R
O2
S
O2
S
N
N
N
N
O
O
N
H
N
H
N
H
N
N
H
O
O
HN
Proposed propionamide Hybrid 5
(assumes "neutral P1 (3) binding mode")
Proposed propionamide Hybrid 4
(assumes "aminopyridine P1 (2) binding mode")
Modeling driven modification
Modeling driven modification
R
HN
R
HN
O
O
O
O2
S
O2
S
O
N
N
N
N
N
H
N
H
N
H
N
H
N
O
O
HN
Refined acetamide Hybrid 7
(assumes "neutral P1 (3) binding mode")
Refined acetamide Hybrid 6
(assumes "aminopyridine P1 (2) binding mode")
Figure 2. Proposed X-ray inhibitor-complex structure driven modification of the imidazole portion of inhibitor 1 targeting increased binding affinity
for thrombin.

2064
R. C. A. Isaacs et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2062–2066
of our proposed hybrids from 4 and 5, respectively, to 6
and 7. We chose to test our hypothesis first with hybrid
6 as opposed to hybrid 7 because the chemistry to make
N-alkylated imidazoles was more straightforward than
the chemistry to make C-alkylated imidazoles.
4-methanol 8 with tert-butyl bromoacetate gave a 2:1
mixture of alkylated isomers 9a and 9b which were sep-
arable by column chromatography. The regiochemistry
of the major alcohol isomer 9a was confirmed by NOE
studies on a subsequent derivative. Alcohol 9a was con-
verted to azide 10 which was then reduced to amine 11.
Amine 11 was incorporated into the pyridinone scaffold
via standard amide coupling to give the imidazole-N1-
acetate 12. We were gratified to find that 12, an interme-
diate towards the first compound in our proposed series
of alkylated imidazoles was a very potent thrombin
inhibitor (Ki = 0.48 nM), 16-fold more potent than the
parent imidazole 1 (Ki = 8 nM), and equipotent with
the methylaminopyridine 2 (Ki = 0.5 nM) whose overall
activity profile was being targeted. Ester 12 was con-
verted to acid 13 (Ki = 31 nM) which was then parlayed
into a number of different amides (Table 1). Alterna-
tively, removal of the tert-butyl group from the interme-
diate azide 10, coupling to an appropriate amine,
reduction of the azide and subsequent coupling to the
pyridinone acid also provided access to these N-acetam-
idoimidazole hybrid analogs.
The synthesis of hybrid 6 type compounds proceeded as
outlined in Figure 3.6 Alkylation of 5-methylimidazole-
tBu
tBu
O
O
O
.HCl
HO
O
H
N
c
N
N
N
a, b
HO
N3
N
N
8
10
9a
f,g,d
HN
d, e
R
O
tBu
O
O
O
O2
S
N
N
N
H2N
N
N
H
N
H
O
N
14
Relative to the unalkylated imidazole lead 1, the imi-
dazoleacetamides as a class are very potent (low and
sub-nanomolar) thrombin inhibitors which show a high
degree of tolerance for structural diversity (lipophilic,
hydrophilic, cyclic, and acyclic substituents) at the
amide position. This is in contrast to the original neutral
phenoxyacetamide series exemplified by 3 where only
small lipophilic amides (e.g., ethyl and cyclopropyl) were
tolerated.5b
12
e
R
HN
O
O
O2
S
N
N
N
H
N
H
O
N
Hybrids 6
Figure 3. Synthesis of hybrids 6 and analog 12. Reagents and
conditions: (a) BrCH2CO2tBu, K2CO3, DMF; (b) chromatographic
separation; (c) (PhO)2PON3, DBU, DMF; (d) H2, 10% Pd/C, EtOAc;
(e) RCO2H, EDC, HOBt, Et3N, DMF; (f) HCl, EtOAc, 0 ꢁC; (g)
RR0NH, EDC, HOBt, Et3N, DMF.
Another improvement over the unalkylated imidazole 1
is that the imidazoleacetamide class of inhibitors is uni-
formly very highly selective for thrombin versus trypsin
(>40,000-fold). This high degree of selectivity is also ob-
Table 1. Thrombin and trypsin inhibition constants, in vitro anticoagulant potency, and in vivo antithrombotic efficacy for compounds 12 and 15–29
R
HN
O
O
O2
S
N
N
N
N
H
N
H
O
Compound
R
Thrombin Ki (nM)
Trypsin Ki (lM)
Fold selectivity
2· APTT (lM)
FeCl3
12
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
t-Bu ester
t-Bu
0.48
0.36
2
50
68
104,000
190,000
70,000
103,000
52,000
180,000
59,000
53,000
48,000
48,000
94,000
86,000
355,000
233,000
41,000
118,000
ND
0.52
0.7
4/6
2/6
Et
c-Pr
137
185
57
4/6
1.8
1.1
0.19
0.83
1.9
2.8
2.4
0.9
0.65
0.31
0.3
3
0.68
0.53
0.56
0.53
0.51
0.66
0.85
0.39
0.37
0.35
0.32
0.65
0.26
ND
ND
ND
1/6
c-PrCH2
EtCMe2–
CF3CH2
34
49
Morpholine
4-HO-azetidine–
HOCH2CH2
H2NCH2CH2
4-Piperidine
3-Piperidine
H2NCH2CMe2–
4-H2N-azetidine–
H2NCMe2CH2–
101
167
115
85
ND
ND
ND
0/6
56
0/6
0/6
110
70
ND
ND
ND
122
80
0.68

R. C. A. Isaacs et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2062–2066
2065
served relative to the full panel of plasma and intestinal
lumen proteases against which our thrombin inhibitors
are routinely counter-screened (factor Xa, chymotryp-
sin, tPA, p-Kal).
are indeed made. However, the S1 binding mode of
the imidazole-N-acetamide in 15 is different from that
of the noncharged oxyacetamide ligand in 3 in that the
amide carbonyl of 15 makes a hydrogen bond to a nitro-
gen of Glu192 instead of Gly219. This 180ꢁ reversal then
allows the acetamide NH to participate in hydrogen
bonding to an ordered water molecule that also hydro-
gen-bonds to the nitrogen of Gly219 and the oxygen
of Gly216. The tert-butyl group is still able to fill the
lipophilic S4 pocket. However, with this mode of access,
the S4 substituent is partially solvent exposed thereby
allowing accommodation of more structurally diverse
groups (hydrophilic as well as hydrophobic) unlike in
the original phenoxyacetamide series where the S4 sub-
stituent is more deeply buried in the S4 pocket and tol-
erant of small hydrophobic groups.
Generally speaking, increased intrinsic thrombin po-
tency, results in improved in vitro functional potency
in the 2· APTT anticoagulant assay and in vivo efficacy
in the rat ferric chloride efficacy model. However, this
trend is often attenuated by physical properties such as
protein binding and solubility.7
The performance of the imidazoleacetamide class of
inhibitors in the 2· APTT functional clotting assay
was significantly improved (routinely < 1 lM) relative
to unalkylated imidazoles such as
1
(2·
APTT = 1.1 lM) regardless of the nature of the amide
moiety. Analogs in which the amide moiety contained
polar functionality (e.g., 26 and 27) displayed highest
functional potency. This is consistent with prior obser-
vations that increasing polarity and hence higher plasma
free fraction, confers increased functional anticoagulant
potency.7 Several analogs (again, particularly the ones
such as 24 and 26 having more polar amides) displayed
full efficacy in the rat ferric chloride antithrombotic as-
say. This trend is also consistent with prior
observations.7
Another view of the X-ray structure of 15 bound to
thrombin (Fig. 5) demonstrates that the unalkylated
nitrogen atom of the imidazole binds to the carboxyl
of Asp189 via an ordered water molecule. In our previ-
ous paper1 where we described the development of unal-
kylated imidazole inhibitor 1, we noted that in the
absence of a crystal structure of 1 bound to thrombin,
active site modeling suggested that the imidazole group
while being geometrically incapable of directly binding
to Asp 189 was likely capable of doing so via an ordered
water molecule. We had based that supposition on the
imidazole–water–Asp189 interaction seen here with
inhibitor 15.
The X-ray crystal structure of 15 bound to the a-throm-
bin–hirugen complex was determined (Fig. 4) and pro-
vided partial validation for our molecular modeling-
driven approach to improving binding potency. The
benzylsulfonamidopyridinone moiety of the inhibitor
occupies the S3 and S2 pockets of the enzyme as previ-
ously detailed.4,5a,b However, the key binding interac-
tions in the S1 specificity pocket, while somewhat
different from what was predicted based on modeling,
still provided grounds for rationalization of the ob-
served increase in binding potency. Additional produc-
tive hydrogen bonding interactions with the enzyme
Having achieved significant improvements in intrinsic
thrombin potency, trypsin selectivity, in vitro functional
anticoagulant potency, and in vivo antithrombotic effi-
cacy, we turned our attention to an evaluation of oral
bioavailability in this series of inhibitors. Inhibitor 2
when dosed orally in dogs at 5 mpk displayed the fol-
lowing pharmacokinetic parameters: Cmax = 5.31 lM,
Figure 4. X-ray crystal structure of inhibitor 15 bound to the active
site of human a-thrombin showing key hydrogen bonding S1 interac-
tions with the imidazoleacetamide moiety (see Ref. 10).
Figure 5. X-ray crystal structure of inhibitor 15 bound to the active
site of human a-thrombin showing a water-mediated interaction of the
imidazole nitrogen with the Asp189 carboxylate (see Ref. 10).

2066
R. C. A. Isaacs et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2062–2066
lower concentrations being predictive of greater efficacy.
We use this 1 lM mark routinely as a gate for determining
whether to evaluate a compound further in vivo in the rat
ferric chloride efficacy assay (see Ref. 3). In practice, in
optimizing compounds, we strive to achieve 2· APTT
values of <0.5 lM.
t1/2 = 2.5 h, AUC = 18.8 lM h. By comparison, opti-
mized N-acetamidoimidazole inhibitors such as 26
showed no oral absorption at the same dose. Although
the bulk of this optimization was carried out in the
pyridinone P2 series of inhibitors, while this work was
in progress, modifications to the aminopyridinone core
directed toward improving metabolic stability and oral
bioavailability were under investigation. This effort led
to the development of the aminopyrazinone template.8
The piperidino-acetamidoimidazole P1 group present
in 26 was incorporated into the aminopyrazinone scaf-
fold. The resulting inhibitor 30 had an attractively bal-
anced overall activity profile (thrombin Ki = 1.2 nM,
trypsin Ki = 18 lM, 2· APTT = 0.7 lM, full efficacy in
the rat ferric chloride assay, 0/6 occlusions at 10 lg/kg/
min) albeit not as good as 26 but had better pharmaco-
kinetic profile in dogs at an oral dose of 5 mpk
(Cmax = 1.46 lM, t1/2 = 1.6 h) and rhesus monkeys, also
at an oral dose of 5 mpk (Cmax = 0.36 lM, t1/2 = 1.1 h).
None of these compounds however surpassed com-
pound 2 in terms of full in vitro and in vivo profiles.9
3. (a) Kurz, K. D.; Main, B. W.; Sandusky, G. E. Thromb.
Res. 1990, 60, 269; (b) Lewis, S. D.; Ng, A. S.; Lyle, E. A.;
Mellott, M. J.; Appleby, S. D.; Brady, S. F.; Stauffer, K.
J.; Sisko, J. T.; Mao, S.-S.; Veber, D. F.; Nutt, R. F.;
Lynch, J. J., Jr.; Cook, J. J.; Gardell, S. J.; Shafer, J. A.
Thromb. Haemostasis 1995, 74, 1107; (c) In a typical
experiment, six rats are evaluated. Full efficacy is defined
as when none of the animals occlude i.e., 0/6 occlusions.
4. Sanderson, P. E. J.; Cutrona, K. J.; Dorsey, B. D.; Dyer,
D. L.; McDonough, C.; Naylor-Olsen, A. M.; Chen, I.-
W.; Chen, Z.; Cook, J. J.; Gardell, S. J.; Krueger, J. A.;
Lewis, S. D.; Lin, J. H.; Lucas, R. J.; Lyle, E. A.; Lynch, J.
J.; Stranieri, M. T.; Vastag, K.; Shafer, J. A.; Vacca, J. P.
Bioorg. Med. Chem. Lett. 1998, 8, 817.
5. The following references are particularly germane to the
subject matter at hand: (a) Lumma, W. C.; Witherup, K.
M.; Tucker, T. J.; Brady, S. F.; Sisko, J. T.; Naylor-Olsen,
A. M.; Lewis, S. D.; Freidinger, R. M. J. Med. Chem.
1998, 41, 1011; (b) Tucker, T. J.; Brady, S. F.; Lumma, W.
C.; Lewis, S. D.; Gardell, S. J.; Naylor-Olsen, A. M.; Yan,
Y.; Sisko, J. T.; Stauffer, K. J.; Lucas, B. J.; Lynch, J. J.;
Cook, J. J.; Stranieri, M. T.; Holahan, M. A.; Lyle, E. A.;
Baskin, E. P.; Chen, I. W.; Dancheck, K. B.; Krueger, J.
A.; Cooper, C. M.; Vacca, J. P. J. Med. Chem. 1998, 41,
3210; (c) Sanderson, P. E. J.; Cutrona, K. J.; Dyer, D. L.;
Krueger, J. A.; Kuo, L. C.; Lewis, S. D.; Lucas, B. J.; Yan,
Y. Bioorg. Med. Chem. Lett. 2003, 13, 161.
HN
N
O
NH
N
O
N
N
N
H
N
H
O
30
To summarise, by taking advantage of X-ray crystallo-
graphic data generated from a ‘non-charged P1’ series
of thrombin inhibitors, we have further refined our ser-
ies of imidazole P1 thrombin inhibitors. The resulting N-
acetamidoimidazole inhibitors are very potent, selective
and efficacious anticoagulants. Oral bioavailability thus
far has been observed only in conjunction with our P3P2
pyrazinone scaffold.
6. For full experimental details, see: Isaacs, R. C. A.; Naylor-
Olsen, A. M.; Dorsey, B. D.; Newton, C. L. PCT Intl.
Appl. WO 9842342 A1, 1998.
7. Tucker, T. J.; Lumma, W. C.; Mulichak, A. M.; Chen, Z.;
Naylor-Olsen, A. M.; Lewis, S. D.; Lucas, R.; Freidinger,
R. M.; Kuo, L. C. J. Med. Chem. 1997, 40, 830.
8. Sanderson, P. E. J.; Lyle, T. A.; Cutrona, K. J.; Dyer, D.
L.; Dorsey, B. D.; McDonough, C. M.; Naylor-Olsen, A.
W.; Chen, I-W.; Chen, Z.; Cook, J. J.; Cooper, C. M.;
Gardell, S. J.; Hare, T. R.; Krueger, J. A.; Lewis, S. D.;
Lin, J. H.; Lucas, B. J.; Lyle, E. A.; Lynch, J. J.; Stranieri,
M. T.; Vastag, K.; Yan, Y.; Shafer, J. A.; Vacca, J. P. J.
Med. Chem. 1998, 41, 4466.
9. The in vitro profile of compound 2 was presented at the
beginning of the manuscript. When dosed orally in dogs at
1 mpk the observed Cmax was 1 lM and the t1/2 was 2.5 h).
For the corresponding aminopyrazinone analog, the
in vitro profile was as follows: thrombin Ki = 0.8 nM,
trypsin Ki = 1.8 lM, 2· APTT = 0.41 lM, full efficacy in
the rat ferric chloride assay, 0/6 occlusions at 10 lg/kg/
min. When dosed orally in dogs at 0.5 mpk the observed
Cmax was 2 lM and the t1/2 was 3.8 h).
Acknowledgment
We express our gratitude to John Wai, Neville Anthony
and Thomas Tucker for helpful discussions during the
preparation of this manuscript.
References and notes
1. Isaacs, R. C. A.; Solinsky, M. G.; Cutrona, K.; Newton,
C. L.; Naylor-Olsen, A. M.; Krueger, J. A.; Lewis, S. D.;
Lucas, B. J. Bioorg. Med. Chem. Lett. 2006, 16, 338.
2. We consider human plasma 2· APTT values of 1 lM or
less to be generally predictive of good in vivo efficacy with
10. The X-ray coordinates for inhibitor 15 have been depos-
ited with the RCSB Protein Data Bank database (Depo-
sition No. 3C1K).