Oxyguanidines. Part 2: Discovery of a novel orally active thrombin inhibitor through structure-based drug design and parallel synthesis

Article abstract of DOI:10.1016/j.bmcl.2004.05.002

Through structure-based drug design and parallel synthesis, we have discovered a novel series of nonpeptidic phenyl-based thrombin inhibitors using oxyguanidines as guanidine bioisosteres. These compounds have been found to be highly potent, highly selective, and orally bioavailable.

Full text of DOI:10.1016/j.bmcl.2004.05.002

Bioorganic & Medicinal Chemistry Letters 14 (2004) 3727–3731
Oxyguanidines. Part 2: Discovery of a novel orally active thrombin
inhibitor through structure-based drug design and parallel synthesis
Tianbao Lu,a,* Thomas Markotan,a Frank Coppo,a Bruce Tomczuk,a Carl Crysler,a
Stephen Eisennagel,a John Spurlino,a Lisa Gremminger,a Richard M. Soll,a
Edward C. Giardinob and Roger Bonea
a3-Dimensional Pharmaceuticals, Inc., 665 Stockton Drive, Exton, PA 19341, USA
bDrug Discovery, Johnson & Johnson Pharmaceutical Research & Development, Spring House, PA 19477, USA
Received 12 March 2004; revised 3 May 2004; accepted 4 May 2004
Available online 28 May 2004
Abstract—Through structure-based drug design and parallel synthesis, we have discovered a novel series of nonpeptidic phenyl-
based thrombin inhibitors using oxyguanidines as guanidine bioisosteres. These compounds have been found to be highly potent,
highly selective, and orally bioavailable.
ꢀ 2004 Elsevier Ltd. All rights reserved.
active thrombin inhibitors utilizing this combined ap-
The serine protease, thrombin, a key mediator in the
proach.
blood coagulation system, is involved in the cleavage of
fibrinogen to form fibrin, the cross-linking of fibrin to
form clots, the stimulation of platelet aggregation, and
In a previous paper,8 we reported a novel series of po-
tent orally active thrombin inhibitors using oxyguan-
the autoamplification of the coagulation cascade to
produce additional active thrombin.1 The orally bio-
available antithrombotic agent, warfarin, is the current
treatment but possesses a number of limitations, such as
(a) the indirect mechanism of action, (b) the need for
constant monitoring to assure effective drug plasma
levels or avoidance of bleeding complications, and (c)
potential drug–drug interactions. Other antithrombotics
idines as guanidine bioisosteres exemplified by 1 (Fig. 1).
Compound 29 has been disclosed by Glaxo as a potent
thrombin inhibitor. By combining the left side of 2 with
the right side of 1, we have prepared compound 8a,
which has a Ki ¼ 21 nM for thrombin. X-ray crystallo-
graphic studies revealed that the right side oxyguanidine
of 8a binds to the thrombin S1 pocket in the same
such as heparin and low-molecular-weight heparin
(LMWH), lack intrinsic oral bioavailability.2 There
have been very few reports of orally active thrombin
inhibitors, which are efficacious in animal models of
thrombosis.3–6 We have reported a series of orally active
thrombin inhibitors using amidinohydrazones as gua-
nidine bioisosteres.7 It is now widely recognized that
structure-based drug design coupled with parallel syn-
thesis is a powerful tool and valuable asset in the drug
discovery process for lead optimization. We report here
the discovery and development of novel potent orally
Keywords: Oxyguanidine; Thrombin inhibitor; Structure-based drug
design; Parallel synthesis.
* Corresponding author. Tel.: +1-610-458-6068; fax: +1-610-458-8249;
Figure 1. Chemical structures of compounds 1, 2, and 8a.
e-mail: tlu3@prdus.jnj.com
0960-894X/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.05.002

3728
T. Lu et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3727–3731
hexafluorophosphate (BOP), triethylamine, and dime-
thylaminomethyl-polystyrene resin in CH2Cl2 followed
by deprotection with trifluoroacetic acid/methylene
chloride. The crude products were purified by silica gel
chromatography using IST VacMaster manifold and
SPE cartridges to give 8 >90% pure by 1H NMR,
LCMS, and HPLC.
Table 1 summarizes the thrombin inhibition for selected
analogs 8. We have determined the X-ray crystal struc-
ture of several of these analogs and invariably the aryl,
heteroaryl, or cycloalkyl groups with 6 1 carbon atom
attachment are determined to occupy the aryl binding
pocket (8a,f, 13d,o, 14). In the case of two preferred
groups as in 8p, the aryl or heteroaryl group is the
preferential group binding in the aryl binding pocket.
Based on this analysis, the following SAR will be dis-
cussed. While none of these analogs were demonstrably
better than 8a, several analogs had a similar potency to
8a (8e,f,o,p). Two of these (8e and 8f) contain a small
cycloalkylmethylene substituent and a small alkyl chain
(propyl). The other two (8o and 8p) share a 3-furyl-
methylene substituent with an allyl or cycloalkylmeth-
ylene substituent. Compounds that contain heteroatoms
(8h), polar groups (8g,h), heterocyclic rings (8j,n), or
heteroaryl rings (8l,m,n) reduce activity. The 3-furylm-
ethylene group does appear to be significant since other
aryl or heteroaryl methylene groups are less potent
(8q–s). Cyclized secondary amines (8t–z) are not
accommodated in the aryl binding pocket.
Figure 2. Overlap of crystal structures of compounds 1 (blue) and 8a
(gray) bound to thrombin.
manner as the right side of 1, the cyclopentyl ring binds
to the aryl binding pocket, and the allyl group is exposed
to the solvent (Fig. 2).10 Based on this structure and
previous structure–activity relationship (SAR) infor-
mation, we devised a solution-based parallel synthesis
approach to explore the SAR of the aryl binding pocket.
A solution-phase library of analogs of 8 was prepared in
multi-milligram quantities according to Scheme 1. The
key intermediate 7 was prepared from 3,5-dichlorophe-
nol in nine steps in overall 23% yield. 3,5-Dichlorophe-
nol 4 was first protected with tert-butyldimethylsilyl
chloride, treated with Rieke Mg and carbon dioxide at
)78 ꢁC, followed by oxalyl chloride and benzyl alcohol,
and deprotected with tetrabutylammonium fluoride to
give compound 5. The ester 5 was reacted with 3-
bromopropanol in the presence of cesium carbonate
followed by Mitsunobu reaction with N-hydroxyph-
thalimide. The phthalimide 6 was deprotected with
aqueous methylamine in tetrahydrofuran, treated with
bis-Boc amidinopyrazole in DMF, and hydrolyzed
with base (NaOH) in ethanol–water to give the inter-
mediate 7. Compound 7 was treated with R1–HN–R2,
benzotriazole-1-yl-oxy-tri-(dimethylamino)phosphonium
In order to assess the preference of methyl in place of
chloro of 8, compounds 13 (Scheme 2) were prepared by
a similar route as compounds 8. The requisite hydroxy
benzoic acid 10 was prepared via the literature meth-
od.11 The noncommercially available secondary amines
were made through reductive amination of aldehydes
and primary amines. Table 2 summarizes the thrombin
inhibition for the selected analogs 13. In general, com-
pounds 13 are more potent than compounds 8, and have
a narrower SAR. Still, similar conclusions can be drawn
from this series. The most potent compounds with Ki
less than 10 nM contain either a cyclohexyl, cyclo-
hexylmethylene, or furylmethylene substituent as R2 and
Scheme 1. Reagents and conditions: (a) TBDMSCl, DMAP, DIEA, CH2Cl2, 100%; (b) Rieke Mg, CO2, THF, )78 ꢁC, 64%; (c) oxalyl chloride,
C6H5CH2OH, cat. DMF, CH2Cl2, 96%; (d) TBAF, THF, 91%; (e) 3-bromopropanol, Cs2CO3, CH3CN, 50 ꢁC, 83%; (f) N-hydroxyphthalimide,
triphenylphosphine, DEAD, THF, 95%; (g) 40% aq CH3NH2, THF, 64%; (h) N,N0-bis(tert-butoxycarbonyl)amidinopyrazole, DMF, 85%; (i) NaOH,
EtOH–H2O, 70%; (j) R1–NH–R2, benzotriazole-1-yl-oxy-tri(dimethylamino)phosphonium hexafluorophosphate (BOP), dimethylaminomethyl-
polystyrene resin, triethylamine, CH2Cl2; (k) TFA, CH2Cl2, 55–85% over last two steps.

T. Lu et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3727–3731
Table 1. Thrombin inhibition for selected analogs 8
3729
Compds
R1–N–R2
Ki, lM
Compds
R1–N–R2
Ki, lM
N
N
8a
0.021
8b
0.033
N
8c
8e
8g
8i
0.373
0.016
0.213
0.156
0.159
0.047
0.020
0.147
0.091
7.62
8d
8f
0.038
0.023
0.038
0.204
0.059
0.052
0.017
1.60
N
N
N
8h
8j
N
N
H3CO2C
HOOC
O
N
N
N
N
N
N
8k
8m
8o
8q
8s
8l
N
O
N
N
N
8n
8p
8r
8t
N
N
N
N
O
O
N
N
N
S
0.880
0.470
2.900
0.171
N
8u
8w
8y
8v
8x
8z
N
O
N
1.500
0.640
N
N
N
N
N
N
N
lower activity; replacing furylmethylene with 2-pyr-
idylmethylene gives a 8-fold less active compound (13z
vs 13j). In general, nonpolar alkyl chains in R1, and
smaller aromatic, heterocyclic, or cycloalkyl rings in R2
lead to the most potent compounds. In addition, the
introduction of a cyclopropyl group in the S1 binding
pocket increases potency by 2–3-fold (14 vs 13g and 15
vs 13j, Fig. 3 and Table 3).
These thrombin inhibitors retain the excellent selectivity
against other serine proteases (>103-fold) that was ob-
served in the previous set of phenyl-based thrombin
inhibitors.7;8;12 The oxyguanidine unit of compound 13g
has a measured pKa of 7.23. The apparent permeability
coefficient (Papp) from the apical to basolateral side in
the human Caco-2 monolayer was determined to be
4.05 · 10ꢀ6 cm/s that would be considered highly per-
meable. When administered as a 2 mg/kg iv infusion
over 5 min to beagle dogs, the terminal half-life was
3.8 h. The Cmax reached 4.5 lM at 1 h after 10 mg/kg by
oral gavage, and the dose-corrected oral bioavailability
(F ) was 62% (Fig. 4).
Scheme 2. Reagents and conditions: (a) AcOH, H2O, NaOH, 100%;
(b) MgO, H2O, 100 ꢁC, 74%; (c) C6H5CH2Br, NaHCO3, DMF, 86%;
(d) 3-bromopropanol, Cs2CO3, CH3CN, 50 ꢁC, 75%; (e) N-hydroxy-
phthalimide, triphenylphosphine, DEAD, THF, 95%; (f) 40% aq
CH3NH2, THF, 82%; (g) N,N0-bis(tert-butoxycarbonyl)amidinopy-
razole, DMF, 90%; (h) NaOH, EtOH–H2O, 85%; (i) R1–NH–R2,
benzotriazole-1-yl-oxy-tri(dimethylamino)-phosphonium hexafluoro-
phosphate (BOP), dimethylaminomethyl-polystyrene resin, triethyl-
amine, CH2Cl2; (j) TFA, CH2Cl2, 54–85% over last two steps.
The pharmacokinetics of several analogs in dogs are
compared in Table 3. The parameters of all of them are
quite similar and would be considered moderate for the
dog. Firstly, this series is not plagued by low absorption
potential as judged by the high permeability data in the
human Caco-2 cell line. In the dog, the clearance values
an alkyl or S-alkyl chain in the R1 position
(13c,e,g,h,i,p). Pyridines in the R2 position tend to have

3730
T. Lu et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3727–3731
Table 2. Thrombin inhibition for selected analogs 13
Compds
R1–N–R2
Ki, lM
Compds
R1–N–R2
Ki, lM
0.017
N
13a
0.010
13b
N
O
13c
13e
13g
13i
0.006
0.009
0.009
0.007
0.068
0.021
0.110
0.048
0.021
0.195
0.094
0.073
13d
13f
13h
13j
0.008
0.065
0.008
0.011
0.022
0.028
0.006
0.052
0.024
0.031
0.034
0.086
N
H2N
N
O
N
O
N
C2H5O2C
O
N
N
O
O
O
N
N
N
N
O
13k
13m
13o
13q
13s
13u
13w
13y
13l
N
O
13n
13p
13r
13t
13v
13x
13z
N
O
O
O
O
N
N
S
N
N
N
H2N
N
N
S
F
N
N
S
N
S
N
N
N
N
N
N
N
N
N
(CL) range from 1/10 to 1/3 of hepatic blood flow
(HBFdog ¼ 38 mL/min/kg), which would be considered in
the medium range. The volume of distribution (Vss) is
also considered in the medium range for the dog
(2–4-fold total body water of approximately 7 L). The
terminal half-lives are in the medium range for the dog.
As a result, the series is characterized as having very
good, dose-corrected oral bioavailability (F ). Based on
the comparable metabolic stability in dog and human
microsomes (>98%), we would expect that this series has
a high potential for oral bioavailability in human.
O
H
In regard to a biomarker in human, the apTT (activated
partial thromboplastin time) anticoagulant assay using
human plasma is a reasonable measure of efficacy. The
potencies of these analogs (Table 3) in the apTT assay
differ by three orders of magnitude from their thrombin
inhibition potencies. The most likely explanation for this
difference is the high plasma protein binding for this
series (>98%). The plasma protein binding may limit the
antithrombotic efficacy of this series in vivo.
N
N
N
NH
2
14
15
O
O
O
O
NH
O
H
N
NH
NH
2
O
O
Figure 3. Chemical structures of compounds 14 and 15.
Table 3. In vitro anticoagulant potency (2 · apTT: concentration of compound required to double the human activated partial thromboplastin time)
and pharmacokinetic parameters in dogs after 2 mg/kg infusion over 5 min and 10 mg/kg oral administration
Compds
Ki (lM)
Caco-2 2 · apTT
(A to B, 10ꢀ6 cm/s) (lM)
PK (dog)
Cmax (po, lM) T1=2 (iv, h)
CL (iv, mL/min/kg)
Vss (iv, L/kg) F (%)
13a
13g
13j
14
0.010
0.009
0.011
0.004
0.004
4.86
4.05
3.00
4.11
2.17
11.3
11.5
8.4
3.7
4.5
5.1
5.8
4.1
3.7
3.8
2.5
4.4
5.9
11.2
10.1
7.8
2.8
2.4
1.6
1.7
2.5
89
62
68
73
77
7.9
8.7
4.2
5.7
15

T. Lu et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3727–3731
3731
J., Salzman, E. W., Eds.; J.B. Lippincott: Philadelphia,
1994; p 3; (b) Coleman, R. W.; Cook, J.; Niewiarowski, S.
In Hemostasis and Thrombosis: Basic Principles and
Clinical Practice, 3rd ed; Coleman, R. W., Marder, V.
J., Salzman, E. W., Eds.; J.B. Lippincott: Philadelphia,
1994; p 508.
10.00
1.00
0.10
0.01
2mg/kg iv infusion for
5min
10mg/kg PO
2. Tapparelli, C.; Metternich, R.; Ehrhardt, C.; Cook, N. S.
Trends Pharmacol. Sci. 1993, 14, 366.
3. Brady, S. F.; Stauffer, K. J.; Lumma, W. C.; Smith, G. M.;
Ramjit, H.; Lewis, S. D.; Lucas, B. J.; Gardell, S. J.; Lyle,
E. A.; Appleby, S. D.; Cook, J. J.; Holahan, M. A.;
Stranieri, M. T.; Lynch, J. J.; Lin, J. H.; Chen, I. W.;
Varstag, K.; Naylor-Olson, A. M.; Vacca, J. P. J. Med.
Chem. 1998, 41, 401.
0
200
400
Time (min)
600
800
4. Gustafsson, D.; Antonsson, T.; Bylund, R.; Eriksson, U.;
Gyzander, E.; Nilsson, I.; Elg, M.; Mattsson, C.; Deinim,
J.; Pehrsson, S.; Karlsson, O.; Nilsson, A.; Sorensen, H.
Thromb. Haemost. 1998, 79, 110.
Figure 4. In vivo pharmacokinetics in dog (n ¼ 3) after 2 mg/kg iv over
5 min and 10 mg/kg po of compound 13g.
5. Rebello, S. S.; Miller, B. V.; Basler, G. C.; Lucchesi, B. R.
J. Cardiovasc. Pharmacol. 1997, 29, 240.
In summary, we have optimized a novel series of non-
peptidic phenyl-based, highly potent, and highly selec-
tive thrombin inhibitors using oxyguanidines as
guanidine bioisosteres through structure-based design
and parallel synthesis. The compounds of this series are
Caco-2 monolayer permeable and orally bioavailable in
dog. Additional SAR and in vivo efficacy results will be
reported in due course.
6. Sanderson, P. E. J.; Lyle, T. A.; Cutrona, K. J.; Dyer, D.
L.; Dorsey, B. D.; McDonough, C. M.; Naylor-Olson, A.
M.; 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.
7. Soll, R. M.; Lu, T.; Tomczuk, B.; Illig, C. R.; Fedde, C.;
Eisennagel, S.; Bone, R.; Murphy, L.; Spurlino, J.;
Salemme, F. R. Bioorg. Med. Chem. Lett. 2000, 10, 1.
8. Tomczuk, B.; Lu, T.; Soll, R. M.; Wang, A.; Murphy, L.;
Crysler, C.; Dasgupta, M.; Eisennagel, S.; Spurlino, J.;
Bone, R. Bioorg. Med. Chem. Lett. 2003, 13, 1495.
9. Watson, N. S.; Pass, M.; Patel, V. WO 9722589 A1, 1997.
10. The atomic coordinates described in this paper have been
deposited with the Protein Data Bank (entries 1T4U and
1T4V).
Acknowledgements
The authors gratefully acknowledge Mike Kolpak for
NMR spectra and Karen DiLoreto for LC–MS.
References and notes
11. Turner, F. A.; Gearien, J. E. J. Org. Chem. 1959, 24, 1952.
12. Lu, T.; Tomczuk, B.; Illig, C. R.; Bone, R.; Murphy, L.;
Spurlino, J.; Salemme, F. R.; Soll, R. M. Bioorg. Med.
Chem. Lett. 1998, 8, 1595.
1. (a) Coleman, R. W.; Marder, V. J.; Salzman, E. W.;
Hirsch, J. In Hemostasis and Thrombosis: Basic Principles
and Clinical Practice, 3rd ed; Coleman, R. W., Marder, V.