Thrombin inhibitors built on an azaphenylalanine scaffold

Article abstract of DOI:10.1016/j.bmcl.2003.12.083

A series of azaphenylalanine derivatives were investigated as novel thrombin inhibitors based on the prodrug principle. By systematic structural modifications we have identified optimal groups for this series that led us to potent inhibitors of thrombin i

Full text of DOI:10.1016/j.bmcl.2003.12.083

Bioorganic & Medicinal Chemistry Letters 14 (2004) 1563–1567
Thrombin inhibitors built on an azaphenylalanine scaffold
Anamarija Zega,a Gregor Mlinsek,b Tomaz Solmajer,b,c Alenka Trampus-Bakija,d
Mojca Stegnard and Uros Urleba,c,*
aFaculty of Pharmacy, University of Ljubljana, Asˇkercˇeva 7, 1000 Ljubljana, Slovenia
bNational Institute of Chemistry, POB 660, Hajdrihova 19, 1001 Ljubljana, Slovenia
cLek Pharmaceuticals d.d., Drug Discovery, Verovsˇkova 57, 1526 Ljubljana, Slovenia
dUniversity Medical Centre, Department of Angiology, Zalosˇka 7, 1525 Ljubljana, Slovenia
Received 1 July 2003; revised 22 August 2003; accepted 22 December 2003
Abstract—A series of azaphenylalanine derivatives were investigated as novel thrombin inhibitors based on the prodrug principle.
By systematic structural modifications we have identified optimal groups for this series that led us to potent inhibitors of thrombin
incorporating the benzamidine fragment at the P1 position, and their potentially orally active benzamidoxime prodrugs. The
binding modes in the thrombin active site of two representative compounds were identified by X-ray crystallographic analysis.
# 2004 Published by Elsevier Ltd.
Thromboembolic disorders are the major cause of mor-
bidity and mortality in Western societies. At present,
clinical regulation of thrombosis still involves the
administration of heparin and its derivatives, or oral
anticoagulants of the dicumarol type, which all indir-
ectly inhibit thrombin, a trypsin-like serine protease that
plays a key role in blood coagulation processes. These
drugs have several limitations in both their effectiveness
and bleeding risk, leading to the need for extensive
monitoring. An injectable form of the small-molecule
direct thrombin inhibitor, argatroban, has been
approved by the FDA, but only for the relatively rare
condition of heparin-induced thrombocytopenia.1
Orally active thrombin inhibitors have been reported
recently, for example, the most advanced oral thrombin
inhibitor ximelagatran.2 However, an ideal, clinically
useful direct thrombin inhibitor is still a high priority in
medicinal chemistry research.3
Most direct thrombin inhibitors have a strong base that
fits in the S1 specificity pocket of thrombin. The strong
base, often a guanidine or amidine, is protonated at the
pH of the intestinal tract, thus being a poor candidate
for intestinal absorption. Development of new inhibi-
tors has focused on introducing less basic substituents
to improve intestinal absorption.3f,g The N-hydroxyl-
ated derivatives, amidoximes are less basic because of
the introduced the oxygen atom. They are not proto-
nated under physiological conditions and should lead to
sufficient oral absorption and therefore to improved
bioavailability.
Because of these properties, and taking advantage of the
well-investigated reduction of benzamidoximes to ben-
zamidines,9 we have designed compounds with benza-
midoxime moieties instead of benzamidine.
In this report we describe the synthesis, SAR(structure–
activity relationship) and a binding mode of thrombin
inhibitors with an azaphenylalanine scaffold, which are
based on the prodrug principle.
In earlier studies we identified a novel series of non-
covalent inhibitors built on the conformationally
restricted azaphenylalanine scaffold.4,5 In this type of
compound, the a-carbon of the original peptidomimetic
structure6 was replaced by nitrogen and the stereogenic
centre of the central amino acid was omitted, with the
result that the overall conformation was changed.7,8
All target compounds in Tables 1 and 2 were prepared
employing essentially the same strategy as described in
Scheme 1. Commercially available 4-cyanobenzalde-
hyde and 3-cyanobenzaldehyde 1 were transformed into
the corresponding Boc-protected hydrazones 2 using
tert-butylcarbazate followed by catalytic hydrogenation
on Pd/carbon to give intermediates 3. Coupling of these
Boc-protected hydrazines 3 with cyclic amines gave the
Keywords: Thrombin inhibitors; Azapeptidomimetics; Prodrug.
* Corresponding author. Tel.: +386-1-476-9500; +386-1-425-8031;
e-mail: urlebu@ffa.uni-1j.si
0960-894X/$ - see front matter # 2004 Published by Elsevier Ltd.
doi:10.1016/j.bmcl.2003.12.083

1564
A. Zega et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1563–1567
Table 1. Inhibitory activities of para substituted benzamidines and benzamidoximes12
Ki (mM)
Compd
R1
R2
R3
Thrombin
1.9
Trypsin
Factor Xa
6a
OH
>68
13
>75
7a
6b
7b
6c
6d
7d
6e
7e
6f
H
OH
H
0.095
0.43
0.085
0.77
0.38
0.045
0.21
0.032
1.6
66
>75
53
45
26
OH
OH
H
>68
>68
18
>75
>75
>75
>75
69
OH
H
13
10
OH
H
>400
23
ND
ND
7f
0.15
ND not determined.
required Boc-protected carbazamides 4 in a one-pot
synthesis, utilizing commercially available triphosgene.
Deprotecting the Boc group in compounds 4 by treat-
ment with gaseous HCl in AcOH and subsequent reac-
tion with aryl sulfonyl chlorides led to compounds 5.
Methoxy-substituted naphthalenesulfonyl chloride was
prepared from 6-hydroxy-2-naphthalenesulfonic acid
sodium salt and thionyl chloride according to literature
procedures.10,11 Finally, hydroxylamine was used to
convert the nitriles 5 to the target benzamidoximes 6,
and cyano compounds were treated with ethanolic HCl
and reacted with ammonium acetate to give the corres-
ponding amidines 7.
The enzyme inhibition constants of the target benzami-
dines (7) and their benzamidoxime analogues (6) toward
thrombin and the structurally related serine proteases
FXa and trypsin are listed in Tables 1 and 2.
Comparing the in vitro inhibitory potencies of the para
substituted (Table 1) and meta substituted (Table 2)
moieties on the aromatic ring in the P1 part of the
molecule, it is evident that, of the benzamidines, meta
isomers are preferred for binding in the S1 specificity
pocket. In contrast, the most potent of the benzami-
doximes is the para substituted compound 6e, with a Ki
of 210 nM.
Although we propose in vivo reduction of benzamidoxime
to the benzamidine, we have also investigated the SAR
of benzamidoximes.
In evaluation for trypsin-selectivity, para substituted
analogues exhibited better selectivity ratio than meta
analogues, particularly among benzamidines (e.g., the

A. Zega et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1563–1567
Table 2. Inhibitory activities of meta substituted benzamidines and benzamidoximes12
1565
Ki (mM)
Compd
R1
R2
R3
Thrombin
12
Trypsin
Factor Xa
5.4
6g
OH
67
47
6h
6i
OH
OH
H
4.2
111
>75
5.3
0.47
>68
0.11
ND
32
7i
0.004
0.011
0.59
7j
H
ND
>75
3.9
6k
7k
6l
OH
H
0.005
0.61
0.14
17
OH
H
103
3.0
7l
0.009
0.053
ND not determined.
para isomer 7e (312-fold) versus the meta isomer 7k (28
fold)).
piperidine ring, or introduction of the azepine ring, led to
compounds with moderate thrombin inhibitory potency
(6i, Ki=469 nM; 6k, Ki=594 nM; 6l, Ki=610 nM).
Methylation and enlargement of the piperidine ring
resulted in increased hydrophobicity at the P2 part of
the molecule, with increased potency of the benzami-
dines and benzamidoximes. In the series of para sub-
stituted benzamidoximes, introduction of a 2- or 4-
methyl group on the piperidine ring led to 4,5 fold (6b,
Ki=427 nM) and 2.5-fold (6c, Ki=768 nM) increase in
activity, respectively. Incorporation of an azepine ring
in this part of the molecule yielded the most potent
benzamidoxime compound of the series, 6e, with a Ki of
210 nM. The favourable effect of this modification was
not so evident in the series of para benzamidines where
the maximum increase in potency was 3-fold (7a,
Ki=950 nM versus 7e, Ki=320 nM). Introduction of a
2-methyl group on the piperidine in the series of meta
substituted benzamidoximes resulted in compounds
with low activity (6g, Ki=12 mM; 6h, Ki=4 mM). In
contrast, a methyl group on the p-position on the
Thus, among the benzamidines, meta isomers are
preferred for binding in the S1 specificity pocket and
substitution in the P2 part of the molecule led us to
potent inhibitors of thrombin with Ki values lowered
to 4 nM (7i).
Variation of the sulfonamide aryl moiety, which fits into
the S3 binding pocket, led to improvements in potency
in the series of para substituted benzamidoximes.
Replacement of the naphthyl group with the 6-methoxy-
2-naphthyl group led to compound 6d which showed
enhanced potency over 6c. 4-Substituted biaryl replace-
ment (compounds 6f and 7f) resulted in significant
losses of potency. Presumably, the biaryl groups of this
series reach deeper in the S3 pocket of the thrombin
active site, and probably bump into side wall of the
pocket, causing less favourable binding affinity.

1566
A. Zega et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1563–1567
In order to understand the in vitro structure–activity
relation, the binding conformations in the active site of
human a-thrombin of the meta-substituted benzami-
dine, compound 7k, and the para-substituted benzami-
doxime, compound 6d, were determined by X-ray
crystallography. A simple superposition of the struc-
tures of 6d and 7k in the active site of thrombin is
shown in Figure 1.
The complexes, refined at a resolution 1.73 A, showed
that 6d and 7k bind in a compact form similar to that of
other molecules of the Argatroban/NAPAP family.13
Benzamidine or benzamidoxime occupies the S1 selec-
tivity pocket, the 4-methylpiperidine ring nestles under
the YPPW loop of the lipophilic S2 binding pocket and
the naphthalene group enters the aryl binding site.
The meta-substituted benzamidine of compound 7k
goes straight to the bottom of the S1 pocket (Fig. 2) in
the position that facilitates optimal twined contacts with
Asp 189 of thrombin, while the para-substituted benza-
midoxime in 6d is directed to the side wall of the S1
pocket. The benzamidoxime oxygen points towards the
plane of the benzyl group of Tyr 228. The distance
between them is 3.5 A, suggesting an interaction
between hydrogen of the benzamidoxime hydroxyl
group and the p-electron system of the ring.14 Only one
nitrogen of benzamidoxime forms a hydrogen bond
with Asp 189 and two weak bonds with the carbonyl
oxygens of Trp 215 and Ala 190. In 7k there is a total of
seven hydrogen bonds at the bottom of the S1 pocket.
Two of them are with Asp 189, three with other residues
and two with water molecules. 6d additionally lacks a
hydrogen bond with the backbone carbonyl oxygen of
Gly 216. This is probably a consequence of the scaffold
position of the azaphenylalanine functionality in the
active site pocket and para substitution of the benzami-
doxime ring. The third reason for the observed reduced
affinity of 6d versus 7k is probably the absence of any
hydrogen bond with water molecules that are in contact
with protein. 7k on the other hand forms two such
contacts and is thus indirectly bonded to six surround-
ing residues in the active site through a dense hydrogen
bond network of water molecules. Furthermore, the
difference in pKa values of the benzamidine (pKa=11.6)
and benzamidoxime (pKa=4.5 or 7.0 depending on the
tautomeric form) plays a role in binding of the inhibitor
to the aspartate Asp 189 of the S1 binding pocket.15
Scheme 1. (a) Boc-NHNH2, EtOH, reflux; (b) H2, Pd/C, MeOH; (c)
(1) (Cl3CO)2CO, CH2Cl2; (2) cyclic amine, DIEA; (d) HCl (g), AcOH;
(e) aryl sulfonyl chloride, CH2Cl2, Et3N; (f) NH2OH, EtOH, reflux; (g)
HCl/EtOH, NH4OAc.
Figure 1. Binding conformation of 6d (gray) superimposed onto 7i
(green) as bound to the active site of thrombin.
Figure 2. Schematic representation of inhibitor 7i bound in the active
site of thrombin. Dashed lines indicate hydrogen bonds.

A. Zega et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1563–1567
1567
In summary, we have prepared and evaluated a series of
thrombin inhibitors with an azaphenylalanine scaffold
and reduced stereogenicity that incorporate a benzami-
dine or weakly basic benzamidoxime element at the
P1 position. Several benzamidine based thrombin inhib-
itors with low nanomolar potency and their benza-
midoxime analogues, for which we propose in vivo
reduction, were identified.
S. D. Curr. Pharm. Des. 1995, 1, 441. (c) Ripka, W. C.;
Vlasuk, G. P. Ann. Rep. Med. Chem. 1997, 32, 71. (d)
Wiley, M. R.; Fisher, M. J. Exp. Opin. Ther. Patents 1997,
7, 1265. (e) Sanderson, P. E. J.; Naylor-Olsen, A. M.
Curr. Med. Chem. 1998, 5, 289. (f) Menear, K. Curr. Med.
Chem. 1998, 5, 457. (g) Rewinkel, J. B. M.; Adang, A. E. P.
Curr. Pharm. Design 1999, 5, 1043. (h) Coburn, C. A.
Exp. Opin. Ther. Patents 2001, 11, 1. (i) Sanderson, P. E. J.
Annu. Rep. Med. Chem. 2001, 36, 79.
4. Zega, A.; Mlinsek, G.; Sepic, P.; Grdadolnik, S. G.;
Solmajer, T.; Tschopp, T. B.; Steiner, B.; Kikelj, D.;
Urleb, U. Bioorg. Med. Chem. 2001, 9, 2745.
5. Zega, A.; Trampus-Bakija, A.; Fortuna, M.; Stegnar, M.;
Tschopp, T. B.; Steiner, B.; Urleb, U. Pharmazie 2001, 56,
683.
Further SARdevelopment and studies directed to
demonstrate antithrombotic activity in animal models
following oral administration are in progress; the results
will be published in due course.
6. (a) Kim, S.; Hwang, S. Y.; Kim, Y. K.; Yun, M.; Oh, Y. S.
Bioorg. Med. Chem. Lett. 1997, 7, 769. (b) Oh, Y. S.; Yun,
M.; Hwang, S. Y.; Hong, S.; Shin, Y.; Lee, K.; Yoon,
K. H.; Yoo, Y. J.; Kim, D. S.; Lee, S. H.; Lee, Y. H.;
Park, H. D.; Lee, C. H.; Lee, S. K.; Kim, S. Bioorg. Med.
Chem. Lett. 1998, 8, 631. (c) Lee, K.; Hwang, S. Y.;
Hong, S.; Hong, C. Y.; Lee, C.-S.; Shin, Y.; Kim, S.; Yun,
M.; Yoo, Y. J.; Kang, M.; Oh, Y. S. Bioorg. Med. Chem.
Lett. 1998, 6, 869.
Acknowledgements
The authors thank Dr. T. B. Tschopp and Dr. B. Steiner
(Pharma Division, Preclinical research, F. Hoffman La
Roche Ltd, Basel, Switzerland) for biological testing of
some compounds, Drs. W. Bode and R. Friedrich (Max
Planck institut fur Biochemie, Martinsread, Germany)
¨
7. Gante, J. Synthesis 1989, 405.
for help in the determination of the X-ray structures
and Dr. Roger Pain (Jozef Stefan Institute, Ljubljana,
Slovenia) for critical reading of the manuscript. Finan-
cial support of this work by the Ministry of Education,
Science and Sport of the republic of Slovenia (Grant No
P07087-502 and L3-5084) and by Lek Pharmaceuticals
d.d., Ljubljana is gratefully acknowledged.
8. Zega, A.; Urleb, U. Acta Chim. Slov. 2002, 49, 649.
9. Clement, B. Drug Metab. Rev. 2002, 34, 565.
10. Gravett, D. M.; Guillet, J. E. J. Am. Chem. Soc. 1993,
115, 5970.
11. Loewenthal, H. J. E.; Gottlieb, L. J. Org. Chem. 1992, 9,
2631.
12. The detailed assay conditions are described in refs 4 and 5.
13. Mlinsek, G.; Friedrich, R.; Oblak, M.; Zega, A.; Marinko,
`
P.; Peterlin-Masie, L.; Kikelj, D.; Urleb, U.; Hodoscek,
References and notes
M.; Solmajer, T.; Bode, W. Thrombin inhibitors with
novel azaphenylalanine scaffolds and a new P1 binding
pocket functionality: structural and energetic analysis of
binding, In preparation.
1. Matsuo, T.; Koide, M.; Kario, K. Semin. Thromb.
Haemost. 1997, 23, 517.
2. Gustafsson, D.; Elg, M. Thrombosis Research 2003, 109,
9.
3. (a) For reviews see: Taparelli, C.; Metternich, R.;
Erhardt, C.; Cook, N. S. TIPS 1993, 14, 366. (b) Kimball,
14. Levitt, M.; Perutz, M. F. J. Mol. Biol. 1988, 201, 751.
15. pKa values were calculated using ACD/pKa DB software
(Version 6.0). Information about the program is available
at the internet address http://www.acdlabs.com