Design, synthesis, and structure-activity relationship of a new class of amidinophenylurea-based factor VIIa inhibitors

Article abstract of DOI:10.1016/S0960-894X(03)00168-9

Selective inhibition of coagulation factor VIIa has recently gained attraction as interesting approach towards antithrombotic treatment. Using parallel synthesis supported by structure-based design and X-ray crystallography, we were able to identify a novel series of amidinophenylurea derivatives with remarkable affinity for factor VIIa. The most potent compound displays a K_i value of 23 nM for factor VIIa.

Full text of DOI:10.1016/S0960-894X(03)00168-9

Bioorganic & Medicinal Chemistry Letters 13 (2003) 1463–1467
Design, Synthesis, and Structure–Activity Relationship of a New
Class of Amidinophenylurea-Based Factor VIIa Inhibitors
Otmar Klingler,a,* Hans Matter,a Manfred Schudok,a S. Paul Bajaj,b Joerg Czech,a
Martin Lorenz,a Hans Peter Nestler,a Herman Schreudera and Peter Wildgoosea
aAventis Pharma Deutschland GmbH, Building G878, D-65926 Frankfurt, Germany
bDepartment of Pharmacological & Physiological Sciences and Internal Medicine, Saint Louis University Health Sciences Center,
St. Louis, MI 63110, USA
Received 5 November 2002; revised 23 December 2002; accepted 3 February 2003
Abstract—Selective inhibition of coagulation factor VIIa has recently gained attraction as interesting approach towards antith-
rombotic treatment. Using parallel synthesis supported by structure-based design and X-ray crystallography, we were able to
identify a novel series of amidinophenylurea derivatives with remarkable affinity for factor VIIa. The most potent compound dis-
plays a Ki value of 23 nM for factor VIIa.
# 2003 Elsevier Science Ltd. All rights reserved.
Activated factor VII (fVIIa) plays a pivotal role in the
initiation of blood coagulation. Tissue factor (TF), a
cofactor and allosteric regulator of fVIIa which is loca-
lized on the surface of subendothelial cells, is another
important protein in the early steps of the coagulation
cascade.1ꢀ4 The fVIIa/TF complex efficiently hydrolyzes
the zymogen coagulation factors (fX, fIX and fVII) to
the corresponding active serine protease forms, culmi-
nating in the conversion of prothrombin to thrombin
which eventually transforms fibrinogen to fibrin in the
final stage of clotting.5,6
structure activity relationship studies of a novel series of
amidinophenylurea derivatives as potent inhibitors of
the fVIIa/TF complex.18
The initial representatives of the amidinophenylurea ser-
ies, compounds 1 and 2, were discovered from focused
screening of serine protease directed compounds from our
internal collection.19 The selection of the screening set was
guided by the identification of relevant protein–ligand
interactions from the public X-ray structure of the fVIIa/
TF complex with a covalent inhibitor (1DAN).20
The pathophysiological role of an excessively active
coagulation cascade and subsequent inappropriate
thrombus formation in blood vessels has been well
recognized. Currently available anticoagulants show
undesirable properties and side effects, like bleeding,
that limit their usefulness in the treatment of thrombotic
disorders and encourage efforts to discover new low
molecular weight compounds which inhibit the enzymes
of the blood coagulation cascade.7,8 Several reports
point out that inhibiting the early blood coagulation
stages cause significantly fewer bleeding events.9 There-
fore, the fVIIa/TF complex is seen as promising target
for developing novel therapeutic anticoagulants.10ꢀ17 In
this publication, we describe the identification and
Figure 1. Chemical structures of initial amidinophenylureas.
Consistent manual and automated docking results
based on this structure plus structural experience from
other serine proteases led us to postulate that the ben-
zamidine is situated in the fVIIa catalytic domain S1
pocket, interacting with the carboxylate side chain of
Asp189 at the bottom of this pocket, the Og atom of
Ser190 and to the carbonyl oxygen of Gly219 by
*Corresponding author. Tel.: +49-69-305-83313; fax: +49-69-305-
21618; e-mail: otmar.klingler@aventis.com
0960-894X/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00168-9

1464
O. Klingler et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1463–1467
additional hydrogen bonds.21ꢀ24 Those studies also led
to the hypothesis that both urea NH protons interact
with the active triad residue Ser195. The second aro-
matic ring was postulated to occupy the more hydro-
phobic, shallow binding site area bridging the S2 and S4
pockets above the Trp215-Gly216 beta-sheet region. This
binding site hypothesis prompted for a rational design
using the amidinophenylurea scaffold as privileged tem-
plate for exploring the both the S2 and S4 pockets.
Table 1. In vitro activity for compounds 8–19
Ki (mM)
Compd
R4
fVIIa
0.97
fXa
Thrombin
10.1
Compounds 1–19, 21, and 24–27 were readily prepared
according to methods outlined in Scheme 1. Reaction of
aminobenzonitrile derivatives with various isocyanates
in DMF yielded nitrile intermediates 3. Of the different
procedures for amidine formation utilized, the two step
Pinner reaction proved to be the most convenient
method.25 Thus, intermediates 3 were sequentially trea-
ted with HCl/ethanol and NH3/ethanol to afford the
benzamidines 4, which were converted to the key inter-
mediates 5. The final step in the synthesis sequence was
partly carried out by automated parallel synthesis cou-
pling the acids 5 (TOTU/DMF)26 with amino building
blocks H2N-R4. Protection of the amidino groups was
not necessary. The amino building blocks were com-
mercially available or prepared from ketone precursors
by reductive amination.
8
14.5
nd
9
0.73
nd
10
11
7.62
0.43
>10
>10
>50
50.7
12
13
14
15
16
20.2
>10
>10
nd
>50
15.9
nd
0.027
31.0
1.51
7.8
>50
Scheme 1. Synthesis of compounds 1–19, 21, 24–27. Reagents and
conditions: (a) ethoxycarbonyl-CHR3 isocyanate, DMF; (b) HCl,
EtOH; (c) NH3, EtOH; (d) NaOH, H2O; (e) H2N-R4, TOTU, DMF,
N-ethylmorpholine.
0.023
>10
16.4
17
0.027
>10
>100
18
19
0.035
0.067
>10
>10
18.2
11.2
Scheme 2. Synthesis of compounds 20, 22 and 23. Reagents and
conditions: (a) carbonyldiimidazole, DMF, 80 ꢁC; (b) HCl, EtOH;
(c) NH3, EtOH; (d) NaOH, H2O; (e) H2N-R4, TOTU, DMF,
N-ethylmorpholine.
nd: Not determined.

O. Klingler et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1463–1467
1465
For synthesis of compounds 20, 22 and 23, an alter-
native route depicted in Scheme 2 proved more efficient.
Reaction of aminobenzonitrile with carbonyldiimida-
zole at 80 ꢁC provided the intermediate 6 in situ, which
was converted to the urea building blocks 7 by addition
of amino acid derivatives. The final reaction steps were
conducted as described in Scheme 1.
Ser195-Og (3.1 A), the other via a structurally con-
served water moleule (3.2 A). Especially remarkable
were the interactions of one distal phenyl ring with the
protease S2 pocket. This ring is stacked on top of the
imidazole ring from the active triad His57 (ꢂ3.3 A), its
para-carbon atom is in close distance to the active triad
Asp102 carboxylate (3.2 A) and the hydrophobic por-
tion of Thr99 (3.5 A to Ca), which itself separates S2
from the open S4 pocket. The other phenyl ring is situ-
ated ꢂ3.5 A on top of the Gly216 carbonyl oxygen in
the hydrophobic binding site area next to S4.
Initial variations of the R4 phenyl group did not sig-
nificantly improve fVIIa affinity (Fig. 1, 1, 2, Table 1, 8,
15). Subsequently, a branching point was introduced at
the a-position of the R4 substituent. Incorporation of a
second phenyl ring yielded in compound 9 with
enhanced affinity (Ki=0.73 mM).
The replacement of the S2 directed phenyl substituent
by a smaller, hydrophobic methyl group led to a sig-
nificant increase in binding affinity (11, Ki=0.43 mM).
This was supported by modeling studies, suggesting an
improved complementarity of hydrophobic protein–
ligand interactions in this area, if the remaining phenyl
ring comes closer to the protein surface region. There-
fore our further optimization was based on the X-ray
crystal structure of compound 9. In contrast, introduc-
tion of a second methyl group in a-position of com-
pound 10 resulted in a decreased affinity. Among the
mono-a-methyl substituted derivatives, compounds 13
and 16–18 exhibited particularly high biological activity
(Ki=0.023–0.035 mM). The most active fVIIa inhibitors
were found to be highly selective against related pro-
teases like factor Xa and thrombin, as summarized in
Table 1. Several enantiomeric pairs of R/S analogues
were prepared to investigate the stereochemical orienta-
tion of the methyl group at the a-position of R4. For
example, the S-phenyl(methyl) analogue 11 is 50-fold
more potent than its R-phenyl(methyl) counterpart 12,
and S-3-bromophenyl(methyl) inhibitor 13 is 1150-fold
more potent than its R-3-bromophenyl-(methyl) enan-
tiomer 14. This suggests in accord with our binding site
model the importance of filling the S2 pocket with small
hydrophobic substituents.
For compound 9 we solved its complex with human
fVIIa/TF by X-ray structure analysis at 2.2 A (Fig. 2),
thus validating our initial model about favorable pro-
tein–ligand interactions for biological affinity.27 In
addition to the expected amidinophenyl binding to S1,
one urea NH interacts directly with the active triad
To determine how variations at the central unit of the
urea scaffold would affect activity, the glycyl moiety was
replaced by several amino acids, as shown in Table 2.
Incorporation of alanine generated compound 20 which
is about 50-fold less potent than the corresponding gly-
cine derivate 8 (Table 1). Compound 21 with phenyl-
alanine moiety is 10-fold less potent than glycine
derivative 2. Affinity is also decreased by a phenylgly-
cine at the central core. The glycine derivative 13 (Table
1) is 120-fold more potent than the phenylglycine ana-
logue 23. Glutamic acid derivative 22 was the most
active compound within this series, but also showed a
20-fold decrease in potency relative to compound 13.
Thus, within this series the glycyl residue proved to be
the most preferred residue at this position. Obviously,
none of the introduced substituents led to favorable
interactions to fVIIa, as both Gly Ca protons are
directed towards the solvent exposed area of the binding
site.
Figure 2a. Comparison of X-ray structure (white carbon atoms) to
proposed binding mode (magenta) for compound 9 in factor VIIa. The
protein is displayed with a MOLCAD surface indicating local curva-
ture. The conformational model for 9 was generated using flexible
docking with QXP.
Finally, our interest was to discern electronic effects of a
substituent in the benzamidine core moiety, as a lower
basicity of this fragment is generally accepted to pro-
duce a more favorable oral bioavailability.29 Hence,
Figure 2b. Schematic interactions of 9 with key residues of human
factor VIIa.

1466
O. Klingler et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1463–1467
Table 2. In vitro activity for compounds 20–23
attributed to unfavorable steric interactions between
one amidinophenyl NH2 group and the ortho-halogen
atoms, as compared to the less bulky hydrogen.
Obviously, the favorable dihedral angle of ꢂ30ꢁ for the
amidine plane relative to the phenyl ring could not be
maintained in these derivatives. As expected, the
experimental pKa values decrease from 11.2 to 11.3 of
compounds 13, 16 and 19 to 10.7 to 10.9 of compounds
24, 25, 26 and 27 (Table 3).28
Compd
Structure
Ki (mM)
fVIIa
20
21
45.8
20.9
In summary, we have described the SAR of a series of
factor VIIa inhibitors with various modifications
around an amidinophenylurea template. The incor-
poration of branched amines attached to the core tem-
plate led to a significant improvement of biological
activity and high selectivity against related proteases.
The best compounds (13 and 16–18) have Ki values
between 23 and 35 nM against fVIIa without significant
inhibition of factor Xa or thrombin. Hence, this series
constitutes a promising area for further studies sup-
ported by structure-based design.
22
23
0.54
3.2
Acknowledgements
The authors would like to thank Peter Below, Silke
Herok, Iris Hessler, Alexander Liesum and Andrea
Struppmann for their technical assistance.
halogen atoms were introduced into the ortho-position
to the amidino group, resulting in compounds 24–27.
The modifications, which included incorporation of a
chlorine, afforded analogues 24 and 26 with 140-fold
and 60-fold loss of activity relative to the parent deri-
vatives 13 and 19. The effect of a fluorine substituent is
less marked. However, compounds 25 and 27 are 4-fold
and 7-fold less active than the unsubstituted analogues
13 and 16. According to modeling studies, this reduced
biological affinity by substitution in the S1 pocket was
References and Notes
1. Kirchhofer, D.; Banner, D. W. Trends in Cardiovascular
Medicine 1997, 7, 316.
2. Petersen, L. C.; Hedner, U.; Wildgoose, P. In Mol. Basis
Thromb. Hemostasis, High, K. A., Roberts, H. R., Eds., Dek-
ker, New York, 1995, pp 147–165.
3. Gallagher, K. P.; Mertz, T. E.; Chi, L.; Rubin, J. R.;
Uprichard, A. C. G. Handbookof Experimental Pharmacology
1999, 132, 421.
Table 3. In vitro activity for compounds 24–27
4. Morrisey, J. H. Thromb. Haemostasis 2001, 86, 66.
5. Kirchhofer, D.; Nemerson, Y. Curr. Opin. Biotechnol. 1996,
7, 386.
Compd
Structure
Ki (mM)
fVIIa
6. Eigenbrot, C.; Kirchhofer, D. Trends in Cardiovascular
Medicine 2002, 12, 19.
7. Hirsh, J.; Weitz, J. I. Lancet 1999, 353, 1431.
8. Adang, A. E. P.; Rewinkel, J. B. M. Drugs Future 2000, 25,
369.
9. Girard, T. J.; Nicholson, N. S. Curr. Opin. Pharmacol.
2001, 1, 159.
10. Young, W. B.; Kolesnikov, A.; Rai, R.; Sprengeler, P. A.;
Leahy, E. M.; Shrader, W. D.; Sangalang, J.; Burgess-Henry,
J.; Spencer, J.; Elrod, K.; Cregar, L. Bioorg. Med. Chem. Lett.
2001, 11, 2253.
24
25
3.76
0.12
11. Jakobsen, P.; Horneman, A. M.; Persson, E. Bioorg. Med.
Chem. 2000, 8, 2803.
12. Jakobsen, P.; Ritsmar Petersen, B.; Persson, E. Bioorg.
Med. Chem. 2000, 8, 2095.
13. Kohrt, J. T.; Filipski, K. J.; Rapundalo, S. T.; Cody,
W. L.; Edmunds, J. J. Tetrahedron Lett. 2000, 41, 6041.
14. Roussel, P.; Bradley, M.; Kane, P.; Bailey, C.; Arnold, R.;
Cross, A. Tetrahedron 1999, 55, 6219.
26
27
3.94
0.47
15. Aliagas-Martin, I.; Artis, D.R.; Dina, M. S.; Flygare,
J. A.; Goldsmith, R. A.; Munroe, R. A.; Olivero, A. G.; Pas-
tor, R.; Rawson, T. E.; Robarge, K. D.; Sutherlin, D. P.;

O. Klingler et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1463–1467
1467
Weese, K. J.; Zhou, Z.; Zhu, Y. PCT Patent WO 0041531.
Chem. Abstr. 2000, 133, 101383.
16. Senokuchi, K.; Ogawa, K. PCT Patent WO 9941231.
Chem. Abstr. 1999, 131, 184864.
25. (a) Pinner, A.; Klein, F. Chem. Ber. 1878, 11, 1825. (b)
Gautier, J. A.; Miocque, M.; Farnoux, C. C. In The Chemistry
of Amidines and Imidates; Patai, S., Ed.; Wiley: New York,
1975; pp 283–348.
17. Ackermann, J.; Alig, L.; Chucholowski, A.; Groebke, K.;
Hilpert, K.; Kuehne, H.; Obst, U.; Weber, L.; Wessel, H. P.
PCT Patent WO 0035858. Chem. Abstr. 2000, 133, 59090.
18. Klingler, O.; Schudok, M.; Nestler, H. P.; Matter, H.;
Schreuder, H. EP Patent 1162194. Chem. Abstr. 2001, 136, 19952.
19. Inhibitory activity expressed as inhibition constant Ki
toward fVIIa/tissue factor was determined using a chromo-
genic assay as described in: Ostrem, J. A.; Al-Obeidi, F.; Safar,
P.; Safarova, A.; Stringer, S. K.; Patek, M.; Cross, M. T.;
Spoonamore, J. C.; LoCascio, J. C.; Kasireddy, P.; Thorpe,
D. S.; Sepetov, N.; Lebl, M.; Wildgoose, P.; Strop, P. Bio-
chemistry 1998, 37, 1053.
20. Banner, D.; D’Arcy, A.; Chene, C.; Winkler, F. K.; Guha,
A.; Konigsberg, W. H.; Nemerson, Y.; Kirchhofer, D. Nature
1996, 380, 41.
21. Rarey, M.; Kramer, B.; Lengauer, T.; Klebe, G. J. Mol.
Biol. 1996, 261, 470.
22. McMartin, C.; Bohacek, R. S. J. Comp. Aided Mol. Des
1997, 11, 333.
23. Heiden, W.; Goetze, T.; Brickmann, J. J. Comp. Chem.
1993, 14, 246.
24. Modeling was done using Sybyl 6.5-6.8 (Tripos). Energy
calculations were based on the MMFF94s force field (Halgren, T.
J. Comp. Chem. 1999, 20, 720–729.). Protein-ligand complexes
were minimized using the quasi-Newton-Raphson procedure.
For manual docking, ligands were placed in the fVIIa binding site
and optimized treating all side chains within 4 A flexible. Auto-
mated docking was carried out using FlexX (21) and QXP (22).
26. TOTU: O-[Ethoxycarbonyl)cyanomethylenamino]-N,N,N0,
N0-tetramethyl-uronium tetrafluoro-borate; Konig, W.; Brei-
pohl, G.; Pokorny, P.; Birkner, M. Proceedings of the 21st
European Peptide Symposium 1990; Giralt, E., Andren, D,
Eds.; ESCOM, Leiden, 1990, p 143.
27. The factor VIIa-tissue factor complex was crystallized
in hanging drops at pH 6.5 using conditions described by
Kirchhofer, D.; Guha, A.; Nemerson, Y.; Konigsberg, W.
H.; Vilbois, F.; Chene, C.; Banner, D.; D’Arcy, A. Prot.
Struct. Func. Gen. 1995, 22, 419. Data were collected at
beam-line ID 14-3 of the European Synchrotron Radiation
Facility (ESRF) in Grenoble. The data were processed with
MOSFLM: Leslie, A.G.W.; Powell, H.R. NATO Science
Series, Series I: Life and Behavioural Sciences 2001, 325, 61.
The crystal diffracted to 2.2 A with an overall Rsym of
5.7%. The space group was P212121 with a=70.08 A, b=81.74
A and c=126.17 A. Initial phases were calculated using the
structure by Banner (20) as starting model. The inhibitor was
fitted using Quanta and refined using CNX (Accelrys). The
final R-factor was 20.0% (R-free=25.0%). The benzamidine
and urea parts of the inhibitor are well defined in the electron
density maps, while the two phenyl groups and the peptide
linker have poor electron density indicating some disorder.
28. The pKa values were measured by UV difference spectro-
scopy on a Sirius SGA profiler.
29. Linusson, A.; Gottfries, J.; Olsson, T.; Ornskov, E.;
´
Folestad, S.; Norden, B.; Wold, S. J. Med. Chem. 2001, 44,
3424.