Design and structure-activity relationship of thrombin inhibitors with an azaphenylalanine scaffold: Potency and selectivity enhancements via P2 optimization

Article abstract of DOI:10.1016/S0968-0896(01)00202-4

Theoretical and structural studies followed by the directed synthesis and in vitro biological tests lead us to novel noncovalent thrombin pseudopeptide inhibitors. We have incorporated an azapeptide scaffold into the central part of the classical tripeptide D-Phe-Pro-Arg inhibitor structure thus eliminating one stereogenic center from the molecule. A series of compounds has been designed to optimize the occupancy of the S2 pocket of thrombin. Increased hydrophobicity at P2 provides an enhanced fit into this active site S2 pocket. In the present paper, we also report on the structure of these inhibitors in solution and conformational analysis of inhibitors in the active site in order to asses the consequences of the replacement of the central α-CH by a nitrogen functionality. In vitro biological testing of the designed inhibitors shows that elimination of R, S stereoisomerism and restriction of conformational freedom influences the binding of inhibitors in a favorable fashion.

Full text of DOI:10.1016/S0968-0896(01)00202-4

Bioorganic & Medicinal Chemistry 9 (2001) 2745–2756
Design and Structure–Activity Relationship of Thrombin
Inhibitors with an Azaphenylalanine Scaffold: Potency and
Selectivity Enhancements Via P2 Optimization
A. Zega,a G. Mlinsek,b P. Sepic,a S. Golic Grdadolnik,b T. Solmajer,b,c
T.B. Tschopp,d B. Steiner,d D. Kikelja and U. Urleba,c,*
aFaculty of Pharmacy, University of Ljubljana, Asˇkerœˇ va 7, 1000 Ljubljana, Slovenia
bNational Institute of Chemistry, POB 3430, Hajdrihova 19, 1115 Ljubljana, Slovenia
cLek, d.d., Pharmaceutical and Chemical Company, 1526 Ljubljana, Slovenia
dPharma Division, Preclinical Research, F. Hoffmann La Roche Ltd, 4002 Basel, Switzerland
This paper is dedicated to Professor Dr. M. Tisler (Ljubljana) in honor of his 75th birthday
Received 6 February 2001; accepted 15 June 2001
Abstract—Theoretical and structural studies followed by the directed synthesis and in vitro biological tests lead us to novel non-
covalent thrombin pseudopeptide inhibitors. We have incorporated an azapeptide scaffold into the central part of the classical tri-
peptide d-Phe-Pro-Arg inhibitor structure thus eliminating one stereogenic center from the molecule. A series of compounds has
been designed to optimize the occupancy of the S2 pocket of thrombin. Increased hydrophobicity at P2 provides an enhanced fit
into this active site S2 pocket. In the present paper, we also report on the structure of these inhibitors in solution and conforma-
tional analysis of inhibitors in the active site in order to asses the consequences of the replacement of the central a-CH by a nitrogen
functionality. In vitro biological testing of the designed inhibitors shows that elimination of R, S stereoisomerism and restriction of
conformational freedom influences the binding of inhibitors in a favorable fashion. # 2001 Elsevier Science Ltd. All rights reserved.
Introduction
of the P1 residue of these inhibitors which are com-
pletely protonated under all physiologcal conditions
hamper the passive diffusion across intestinal barriers
that is the main route of absorption for most drugs.4ꢀ17
Closely related derivatives of Argatroban, such as the
amidrazone derivative LB-30,05718ꢀ20 (2) and the ben-
zamidine derivative UK 156,40621 (3) have recently been
identified as orally active thrombin inhibitors (Fig. 1).
Thrombin is a trypsin-like serine protease that plays a
central role in the blood coagulation process. It cleaves
and thus transforms fibrinogen to insoluble fibrin which
is the major component of a blood clot.1 Unbalanced
hemostasis can lead to arterial and venous thrombosis
which are important causes of morbidity and mortality.
An effective, especially orally active, thrombin inhibitor
that would cause fewer side effects than the currently
used heparin and warfarin would therefore be of clinical
importance.2,3
Replacement of the strongly basic groups in potent
thrombin inhibitors by less basic ones is usually accom-
panied by a decreased potency.9,10 Thus, a compromise
is being sought in order to simultaneously obtain suffi-
cient affinity, selectivity and bioavailability. In this
paper, we present a novel series of noncovalent inhibi-
tors which have been optimized for reduced basicity at
the S1 pocket and contain a conformationally restricted
substituent aimed at the S2 binding pocket in the active
site of thrombin (Fig. 2).
Thrombin inhibitors are intensively sought after world-
wide. A therapeutically useful protease inhibitor should
be potent, selective and orally bioavailable. The best
studied inhibitors are derived from the sequence d-Phe-
Pro-Arg as found in the clinically used compound
Argatroban (1, Novastan). However, the strong basicity
Azaamino acids are important constituents of pseudo-
peptides and are particulary important as protease inhi-
bitors and inhibitors of other enzymes.22ꢀ27 Replacement
*Corresponding author. Tel.: +386-1-476-9500; fax: +386-1-425-
8031; e-mail: urlebu@ffa.uni-lj.si
0968-0896/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(01)00202-4

2746
A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
Figure 1. Structures of thrombin inhibitors.
Commercially available 4-cyanobenzaldehyde 1 was
transformed into the corresponding hydrazone 2 using
tert-butylcarbazate followed by catalytic hydrogenation
on Pd/carbon to give intermediate 3. Coupling of this
aryl tert-butyl hydrazinecarboxylate with cyclic amines
yielded the required Boc protected carbazamides 4a–d in
a one-pot synthesis utilizing commercially available tri-
phosgene. Deprotecting the Boc group in compounds
4a–d by treatment with gaseous HCl in AcOH and sub-
sequent reaction with naphthalene-2-sulfonylchloride
led to compounds 5a–d. Finally, hydroxylamine was
used to convert the nitriles 5a–d to the target benzami-
doximes 6a–d.
With the Na-substituted-Nb-protected hydrazines tri-
phosgene gives the corresponding carbazic acid chlo-
rides in situ which cannot be isolated. These react
rapidly at low temperature with the amine to yield Nb
protected carbazamide.29
Figure 2. Schematic drawing of the thrombin inhibitor in the active
site of thrombin (R moieties are defined in Table 4).
of the central a-CH of the amino acid residue by nitro-
gen has little effect on the overall polarity of the mol-
ecule but would be expected to provide for novel
chemical and biological properties, by virtue of reduced
conformational flexibility. The incorporation of an aza-
peptide scaffold into the argatroban-like thrombin inhi-
bitor structure9 was expected to be beneficial for two
reasons. Firstly, the change in the overall conformation
might lead to higher affinity for the binding site. Sec-
ondly, the resulting compound might be more stable to
enzymatic degradation and thus have longer duration of
action.22,28 In addition, the absorption and transport
properties might be favorably altered.
Solution structures determined by 2D NMR
The proton chemical shifts assignment and proton–
proton distance determination of 6a–d were determined
following standard procedures, that is homonuclear
DQF-COSY, TOCSY, and NOESY experiments.
The experimentally derived distances were incorporated
as conformational restraints in extensive molecular
dynamics simulations carried out in explicit DMSO.30
By the application of the force field used in the molec-
ular dynamics simulations, important energetic features
of the molecule are included in the analysis. Moreover,
the inclusion of explicit solvent molecules improves the
energetic refinement given the large surface area of the
molecules.
We have performed both a careful molecular modeling
study in the thrombin active site based on published X-
ray structures of various inhibitor–thrombin complexes,
and a structural study of inhibitors in solution using
2-D NMR spectroscopy to explore how these changes
alter the hydrophobic and electrostatic interactions with
the enzyme.
Even with different starting structures, the molecular
dynamics (MD) simulations of all compounds always
resulted in large Nuclear Overhauser Enhacement
(NOE) violations: sometimes with violations as high as
1.5 A. This is an indication that the experimental con-
straints cannot be satisfied by a single conformation.
Therefore we carried out MD simulations with different
sets of experimental constraints. After careful examina-
tion of distance violations we found that there are four
main groups of conformations which differ from one
another in the orientation of the naphthalene, the
Results and Discussion
Chemistry
All target compounds shown in Table 1 were prepared
according to Scheme 1.

A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
2747
Table 1. Conformational constraints (A) of compounds 6a–d determined from NOESY or ROESY spectra and distance violations (A) from
restrained simulations
Conformational constraints
6a
c
6b
6c
6d
Conf.:
Dist.
a
b
d
e
f
Conf. A
Conf. A
Conf. A
Viol.
Viol.
Viol.
Viol.
Viol.
Viol.
Dist.
Viol.
Dist.
Viol.
Dist.
Viol.
0.02
Pip-H2–H3
2.410
0
Pip-H2 –H3
2.325
2.424
0.00
0.00
0
0
Pip-H2 –H3
2.339
2.410
2.339
0.07
0.00
0.01
Pip-H6–H7
0
0
0
0
Pip-H5 (H6 )0–H6 (H7 )
2.569
2.805
3.060
3.245
3.730
2.581
2.424
0.00
0.05
0.00
0.01
0.13
0.04
0.00
CH3–Pip-H2
0
CH3–Pip-H3
CH3–Pip-H4
CH3–Pip-H6
Pip-H6–H4
0
Pip-H6–H5
Pip-H6(H7) –NH
3.949
0.39
0.25
3.448
2.789
0.11
0.03
3.257
0.21
0
Pip-H6 –NH
3.942
0.33
Pip-H6(H7)–CH21
3.138
2.555
0.13
0.31
0.17
0.29
3.054
2.993
0.12
0.21
Pip-H6(H7)–CH22
0
Pip-H6 –CH21
0
Pip-H2 –CH21
0.27
0.25
0.30
0.13
0.18
0.29
0.27
0.29
0
Pip-H2 –CH22
0
Pip-H2 –NH
0
Nph-H1–Pip-H2
Nph-H1–NH
3.333
3.787
0.10
0.11
0.12
0.08
3.835
3.322
2.920
0.11
0.18
0.13
3.100
3.543
0.23
0.11
3.137
3.540
0.13
0.16
Nph-H1–Pip-H6(H7)
0
Nph-H1–Pip-H6
0
Nph-H1–Pip-H4,4 (H5)
4.839
0.09
0.07
3.579
0.15
Nph-H1–CH3
Nph-H1–CH21
6.240
0.10
4.031
2.836
0.39
0.43
Nph-H3–NH
0.36
0.15
0.09
0.34
0.13
0
Nph-H3–Pip-H2
Nph-H3–Pip-H6
4.103
4.301
5.050
Nph-H3–CH21
0.29
0.27
0
Nph-H4–Pip-H4,4
0.13
0.25
0.18
0.05
0
Nph-H4–Pip-H2
0
Nph-H8–Pip-H4,4 (H5)
4.410
4.320
0.11
0.47
0.16
3.557
0.16
Nph-H8–Pip-H6
3.143
4.239
0.24
0.15
0
Pip-H8–Pip-H6
Nph-H8–CH3
4.625
0.16
0
Nph-H8–Pip-H2
0.19
0.06
NH–CH21
NH–CH22
3.256
0.05
0.05
0.06
0.38
0.07
3.183
3.230
0.31
0.35
Ar-H2–Pip-H2
0.34
0
Ar-H2–Pip-H2
4.025
2.849
3.412
0.13
0.07
0.12
0
0
Ar-H2–Pip-H6 (H7 )
Ar-H2–CH21
Ar-H6–CH22
Ar-H2–CH22
Ar-H6–CH21
Ar-H6–NH
4.000
2.868
2.896
2.896
2.868
3.415
0.27
0.12
0.08
0.35
0.10
0.09
3.963
2.879
2.875
0.30
0.12
0.07
3.966
3.271
3.368
0.27
0.45
0.32
0.10
0.11
0.08
0.10
0.13
0.06
0.10
0.07
0.11
0.14
0.12
0.14
0.10
3.395
5.369
2.777
0.15
0.00
0.00
3.341
0.16
Ar-H2,6–NH2
Ar-H3,5–NH2
Ar-H3,5–OH
OH–NH2
2.844
4.257
3.858
0.00
0.24
0.32
0.00
0.24
0.36
0.00
0.19
0.33
0.00
0.35
0.30
0.00
0.25
0.34
0.00
0.26
0.29
Average violation
0.18
0.19
0.17
0.21
0.16
0.19
0.09
0.11
0.17
piperidine or hexahydro-1H-azepin ring and the benza-
midoxime spacer CH2 group (Fig. 3a and b). When only
NOEs which are consistent with a particular orientation
of the above-mentioned groups were included in MD
simulations, the resulting structures had low average
distance restraint violations between 0.1 and 0.2 A and
all NOE violations were less than 0.5 A (Table 1).
we were interested in the influence of different sub-
stituents on the conformation of molecules only simu-
lations with one particular orientation of flexible groups
were performed for the other molecules (6b–c) (Fig. 3b).
The resulting conformations of analogues are similar
and have a typical Y-conformation. The piperidine
and naphthalene rings are forming the upper part of
Y while the benzamidoxime moiety is perpendicular
to the other two rings forming the lower part of Y.
Analogue 6b is an exception, it has a perpendicular
orientation of the piperidine and naphthalene rings.
The different orientation of the above-mentioned rings
and the CH2 group do not significantly influence the
overall conformation of 6a (Fig. 3a). Therefore, since

2748
A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
Solution conformations of the molecules do not change
significantly after their energy minima were determined
in the thrombin active site (Fig. 3a and b).
ligands in the active site of thrombin in order to gather
information on their binding properties.
We have chosen BMS 189,090 for comparison since it
has a low affinity constant (Ki=3.6 nM) and a 3-D
structure similar to the solution structure of 6a [the root
mean square deviation (RMS) obtained from super-
position of 40 common heavy atoms of the two com-
pounds is 1.3 A]. Partial conformational analysis was
performed by rotating parts of the inhibitor molecule
around selected flexible bonds. Flexible bonds were
Conformational search by systematic rotations
Based on the X-ray structure of the complex of throm-
bin inhibitor BMS 189,09031 with thrombin and the
solution structure of N0-hydroxy-4-{[2-(2-naphtylsulfo-
nyl)-1-(1-piperidinylcarbonyl)hydrazino]methyl} benzene-
carboximidamide (6a), we analysed the flexibility of the
Scheme 1. (a) Boc-NHNH2, EtOH, reflux, 95%; (b) H2, Pd/C, MeOH, 97%; (c) (1) (Cl3CO)2CO, CH2Cl2, (2) HNR1 R2, DIEA, 35–81%; (d) HCl
(g), AcOH; (e) naphtalene-2-sulfonyl chloride, CH2Cl2, Et3N, 22–33%; (f) NH2OH, EtOH, reflux, 38–57%.
Figure 3. (a) Comparison of solution conformations of 6a (thin lines) with the conformation of 6a (heavy lines) as modeled in the thrombin active
site. The various orientations of the naphthalene, piperidine ring and benzamidoxime are shown. (b) Comparison of solution conformations of 6b–c
(thin lines) with the structure of 6a (heavy lines). The various orientations of the naphthalene, piperidine and benzamidoxime moiety are shown.

A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
2749
defined as given in Fig. 4 and a total of two single
dihedral angle rotations and three coupled dihedral
angle rotations were performed. These rotations were
chosen as a reasonable compromise between the combi-
natorial blow up in the case of a systematic conforma-
tional analysis for all eight bonds (360/10**8 rotational
conformers) and taking into account the Y structure of
the inhibitor molecule. The resulting conformations
were compared and possible reasons for the different
binding affinities were elucidated.
stereoisomer constraint by having nitrogen on the pla-
nar aza skeletal element enhances the possibility that the
inhibitor ligand will fit into the active site pocket.
Our initial hypothesis based on the study of model
peptides containing aza aminoacids was that the central
nitrogen atom influences surrounding bonds and makes
them more rigid. Whether the introduction of a rigid
structural element favors the complexation of the inhi-
bitor with the target protein was determined by com-
parison of identical inhibitors differing only in that the
central a-CH was substituted by nitrogen. Previous
work done by Bode et al.32 shows the importance of R,
S stereoisomerism in a-CH analogues. Since the free
electron pair of nitrogen means that the molecule has no
chiral properties compared to chiral a-CH carbon atom
this substitution should facilitate the more favorable
‘planar configuration’ of the inhibitor in the protein.
Data obtained through systematic conformational
search show that the dihedral a2 of the studied inhibi-
tors has only roughly half of the favorable conforma-
tions observed for the rotation of dihedral a4. There are
several possible explanations for this. One might be that
the rigidity introduced by the nitrogen atom acts as a
barrier to rotation thus precluding successful adaptation
of dihedrals to the furrowed surface of the thrombin
active site and increasing fluctuations. Another possible
explanation is the planarity around the central N atom
compared to the C atom. It could also be that the
backbone of the inhibitor gets displaced due to the
higher electronegativity of the N compared to the C
atom although a direct hydrogen bond between the
central N atom and surrounding residues was not
observed in our simulations (Table 3).
The following structural properties were addressed by in
our series of ligands: the role of the central nitrogen
atom in 6a–d, the effect of the addition of a CH3 group
to the piperidine ring and the replacement of the piper-
idine ring by a seven-membered ring. A general result
observed from the data in Table 2 is that in the case of
coupled dihedral angle rotations only a small fraction (5–
10%) of the conformations are accessible. From this we
can deduce that the inhibitor fits reasonably well into all
three active site pockets S1–S3 of the enzyme. Further-
more, the position and size of substituent at S2 influences
the number of possible energetically favorable conformers.
Flexibility of the piperidinyl carbonyl substituent is much
smaller than for the benzamidoxime. This can be atrib-
uted to the flexible methylene group connecting the
benzamidoxime ring with the central nitrogen atom.
In the case of single dihedral angle a1 which defines the
rotations of amidoxime functionality found, nearly all
available conformers (between 65 and 71 out of a pos-
sible 72) are possible. This is drastically different from
rotations about angle a0 which defines the rotations of
the P2 moiety where only between 9 and 17 conformers
are available in the inhibitors (Fig. 5a and b). This
finding can be further elaborated: in the case of angle a4
which distinguishes between the central N and the a-
CH, the possible number of accessible conformers is
significantly larger for nitrogen. Thus, releasing of the
The distal part of the ligands projecting into the S3
pocket (2-naphthalenesulfonylhydrazide) is much less
flexible than the central part of the molecule around the
N–N bond. This can be attributed to the tight fit of the
ligand into the active site via multiple hydophobic and
steric contact points at this part of the inhibitor surface.
The ortho (structure 6bC) and para (structure 6cC)
methyl substituent of the piperidine ring it is limiting the
a-CH analogues much less than the nitrogen analogues
6b and 6c.
This is further corroborated by data presented in Table
3.
Table 2. The total number of favorable conformations of inhibitors
6a–d and their carbon analogues as obtained by rotations around
dihedrals in active site of thrombin (dihedral angles are defined in Fig.
4). The number of favorable conformations in all potential minima are
also included
Compd
a0
a1
a2
a3
a4
6a
6aC
6b
6bC
6c
6cC
6d
12
10
9
66
69
67
66
71
68
66
65
36
25
13
25
14
27
39
13
71
80
50
49
53
53
60
48
131
94
98
86
102
88
118
98
9
17
11
13
15
6dC
Figure 4. Torsion angles definitions in molecules 6a–d.

2750
A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
Figure 5. (a) Relative energy profiles for conformational dependence of the inhibitor–thrombin complex on a0. (b) Relative energy profiles for
conformational dependence of the inhibitor–thrombin complex on a1.
The hydrogen bonds observed for the aza analogues in
the thrombin active site are as expected. It may be of
importance for the interpretation of the inhibitory
activity of methyl substituted P2 moieties that for these
molecules there are more H bonds (5 and 4) than for the
unsubstituted analogue 6a (2).
The most important difference in the active site of
thrombin compared with that of trypsin is the YPPW
loop (Tyr60a–Trp60d) which serves to bind small
lipophylic groups in thrombin but is absent in trypsin.
We therefore expected to see an increased binding of the
substituted piperidine moiety to the thrombin S2

A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
2751
pocket. We found that the increased number of con-
formers for dihedral angle a0 of the para methyl sub-
stituted N analogue 6c favorably influences the binding
activity when compared with the carbon analogue 6cC.
binding. Compound 6d (Ki=0.201 mM) exhibited 6-fold
potency enhancement over the corresponding a-CH
analogue (Ki=1.140 mM).18
We have focused our attention principally on the S2
pocket. Methylation and enlargement of the piperidine
ring resulted in increased hydrophobicity at the P2 part
of the molecule, resulting in a 10-fold increase in 6d
potency compared to 6a.
Biological evaluation
The in vitro data for the series 6a–d are summarized in
Table 4. Chromogenic assays were performed with
standard set of serine proteases from thrombin family
including thrombin, fXa, fVIIa and trypsin. The selec-
tivity ratio of trypsin/thrombin is given in Table 4.
The pKa values of benzamidoxime, that is a pKa of
approximately 13 (deprotonation) and a pKa of
1
2
approximately 5 (protonation), indicate that our com-
pounds remain uncharged over a wide pH range. This
property could enhance oral absorption.33
Introduction of a nitrogen funcionality into the central
part of the inhibitor series has had a favorable effect on
Table 3. Hydrogen bonds between inhibitors and thrombin residues
Compd
Number of
H bonds
Distance
(A)
Angle (ꢁ)
Residue
Compd atom
6a
6b
1
2
1
2
3
4
5
1
2
3
4
1
2
3
2.08
2.49
2.44
2.22
2.36
1.67
1.99
2.25
2.12
1.58
1.88
2.05
2.46
1.63
165.95
138.91
144.82
138.35
175.29
168.26
164.09
151.83
160.14
167.30
165.82
137.51
154.82
175.06
cys 191 N
gly 216 N
glu 192 OE1
ser 195 OG
gly 216 N
ala 190 O
ala 190 O
gly 216 N
gly 219 O
asp 189 OD2
asp 189 OD2
asp 189 OD2
asp 221 N
ala 190 O
N1 (amide)
N (oxime)
N1 (amide)
N1 (oxime)
N1 (oxime)
OH (oxime)
N1 (amide)
O3 (sulphur)
N1 (amide)
OH (oxime)
N1 (oxime)
N1 (amide)
O3 (oxime)
O3 (oxime)
6c
6d
Table 4. Structure–activity relationship of P2 directed aza inhibitors
Compd
R
Ki (mM)
fVIIa
Thr
fXa
Tryp
Tryp/Thr
6a
6b
1.938
>69.4
>69.4
>75.4
>75.4
>68.3
>35.2
0.427
45.308
107.8
6c
6d
0.768
0.201
>69.4
>69.4
>75.4
>75.4
>68.3
>88.9
13.341
66.5

2752
A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
Conclusion
tert-Butyl 2-(4-cyanobenzyl)hydrazinecarboxylate (3).
tert-Butyl 2-(4-cyanobenzylidene) hydrazinecarboxylate
2 (14.88 g, 60.7 mmol) was dissolved in MeOH (250 mL)
and 10% Pd/C (1.49 g, 10 w/w%) was added. The mix-
ture was hydrogenated for 6 h. The catalyst was filtered
off and the filtrate was evaporated in vacuo to yield oily
3 (14.52 g, 97%) which crystallized overnight. Mp 58–
64 ꢁC (lit. 66–69 ꢁC);34 IR (KBr) 3366, 3236, 2984, 2932,
By using a structure-based approach to design novel
thrombin inhibitors, we developed a novel class of direct
noncovalent inhibitors with moderate affinities for throm-
bin and good selectivity over trypsin. The main character-
istic of our compounds is the central nitrogen functionality
which reduces the flexibility of the ligand. Using molecular
modeling, we have explored this flexibility and found that
the planar nitrogen does not interfere with the S2 pocket
steric restrictions and that the experimental potency of the
described inhibitors is higher compared to that of the C
analogues. However, the interference is different in the
case of the P1 substituent where benzamidoxime moiety
of molecules 6a–d has more accessible conformers than
its carbon analogues 6aC–6dC.
2225, 1685, 1484, 1368, 1284, 1166, 1023, 818 cmꢀ1
;
1
300.15 MHz H NMR (CDCl3): 1.45 (s, 9H, Boc-H),
4.05 (d, J=4.5 Hz, 2H, CH2), 4.31 (br s, 1H, NH), 6.18
(br s, 1H, NH), 7.46 (d, J=8.3 Hz, 2H, Ar–H2,6), 7.61
(d, J=8.3 Hz, 2H, Ar–H3,5); FAB-MS: MH+=248.
tert-Butyl 2-(4-cyanobenzyl)-2-(1-piperidinylcarbonyl)-
hydrazinecarboxylate
23.49 mmol) was dissolved in CH2Cl2 (40 mL) and a
(4a).
Triphosgene
(6.98 g,
Further optimization of the ligands at the S1 pocket is
under way in this laboratory.
mixture of tert-butyl 2-(4-cyanobenzyl)hydrazine-
carboxilate
3
(11.64 g, 47.06 mmol) and DIEA
(12.28 mL, 70.59 mmol) in CH2Cl2 (80 mL) was slowly
added to the stirred solution of triphosgene over a per-
iod of 30 min. After a further 5 min of stirring, a solu-
tion of piperidine (4.64 mL, 46.94 mmol) and DIEA
(12.28 mL, 70.59 mmol) in CH2Cl2 (80 mL) was added
in one lot. The reaction mixture was stirred for 30 min,
evaporated to dryness, diluted with EtOAc (100 mL),
washed with 10% aqueous citric acid 50 mL), 10%
aqueous NaHCO3 (50 mL) and saline solution(50 mL),
dried and evaporated under reduced pressure to give
oily 4a. The product was crystallized from diethylether
as white crystals (6.59 g, 42%). Mp 50–55 ꢁC; IR (KBr)
3294, 2942, 2229, 1730, 1638, 1519 cmꢀ1; 300.15 MHz
1H NMR (DMSO-d6): 1.28–1.60 (m, 6H, Pip–H3,4,5),
1.37 (s, 9H, Boc), 2.97–3.20 (m, 4H, Pip–H2,6), 4.45 (br
s, 2H, CH2), 7.50 (d, J=8.3 Hz, 2H, Ar–H2,6), 7.78 (d,
J=8.3 Hz, 2H, Ar–H3,5), 9.27 (s, 1H, NH); FAB-MS:
MH+=359. Anal. calcd for C19H26N4O3: C, 63.67; H,
7.31; N, 15.63; found: C, 63.29; H, 6.95; N, 15.34.
Experimental
Chemistry
Melting points were determined on a Reichter hot stage
microscope and are uncorrected. 1H NMR spectra were
recorded on a Bruker Avance DPX300 spectrometer or
on a Varian INOVA 600 MHz spectrometer. The pro-
ton chemical shifts of 6a–d were assigned following
standard procedures using homonuclear DQF-COSY,
TOCSY, and NOESY experiments and are reported in d
(ppm) relative to tetramethylsilane. In NMR assign-
ments the symbol ‘Pip’ refers to piperidine or in com-
pounds 4d, 5d and 6d to the hexahydro-1H-azepine ring,
‘Ar’ to phenyl and ‘Nph’ to naphthalene ring. The IR
spectra were recorded on a Perkin-Elmer FTIR 1600
spectrophotometer. Mass spectroscopy was performed
on Varian-MAT 311A mass spectrometer. Elemental
analyses for all new compounds were performed on a
Perkin-Elmer 240 C C, H, N analyzer. All solvents and
reagents used in the syntheses were of commercial syn-
thetic grade and when required were further purified
and dried by standard methods. All reactions with air-
or moisture-sensitive reactants were carried out under
an argon atmosphere. Reagent abreviations: DIEA
stands for N,N-diisopropylethylamine.
4b, 4c and 4d were prepared by the same methods
described above for the synthesis of 4a using appro-
priate starting compounds.
tert-Butyl 2-(4-cyanobenzyl)-2-[(2-methyl-1-piperidinyl)-
carbonyl]hydrazine carboxylate (4b). Yield: 3.99 g
(38%). Mp 43–47 ꢁC; IR (KBr) 3284, 2978, 2229, 1727,
1
1639 cmꢀ1; 300.15 MHz H NMR (DMSO-d6): 1.10 (d,
J=6.8 Hz, 3H, CH3), 1.36 (s, 9H, Boc), 1.36–1.53 (m,
1H, Pip–H) 1.48–1.53 (m, 5H, Pip–H), 2.88 (t,
J=12.4 Hz, 1H, Pip–H), 3.50–3.59 (m, 1H, Pip–H),
3.99–4.20 (m, 1H, Pip–H), 4.46 (br s, 2H, CH2), 7.50 (d,
J=8.3 Hz, 2H, Ar–H2,6), 7.78 (d, J=8.3 Hz, 2H, Ar–
tert-Butyl-2-(4-cyanobenzylidene)hydrazinecarboxylate (2).
4-Cyanobenzaldehyde (5.11 g, 39.0 mmol) 1 suspended
in EtOH (100 mL) was added to a stirred solution of
tert-butylcarbazate (5.29 g, 40.0 mmol) in EtOH. The
mixture was heated at reflux temperature and after 4 h
the EtOH was partially evaporated in vacuo. Water
(100 mL) was added and the precipitated product was
collected by filtration and washed with diethylether to
give 9.12 g (95%) of the title compound 2 as a white
solid. Mp 158–160 ꢁC (lit. 154–158 ꢁC);34 IR (KBr)
3297, 2996, 2227, 1701, 1531, 1459, 1374, 1258, 1149,
H
3,5), 9.27 (s, 1H, NH); FAB-MS: MH+=373. Anal.
calcd for C20 H28 N4 O3: C, 64.49; H, 7.58; N, 15.04;
found C, 64.06; H, 7.64; N, 14.85.
tert-Butyl 2-(4-cyanobenzyl)-2-[(4-methyl-1 piperidinyl)-
carbonyl]hydrazine carboxylate (4c). Yield: 3.67 g
(81%). Mp 55–58 ꢁC; IR (KBr) 3284, 2977, 2228, 1728,
1636 cmꢀ1; 300.15 MHz 1H NMR (CDCl3): 0.96 (d,
J=6.4 Hz, 3H, CH3), 0.90–1.05 (m, 2H, Pip–H3,5), 1.45
1
1058, 946, 833, 772, 610 cmꢀ1; 300.15 MHz H NMR
(CDCl3): 1.56 (s, 9H, Boc-H), 7.65 (d, J=8.3 Hz, 2H,
Ar–H2,3), 7.77 (d, J=8.3 Hz, 2H, Ar–H5,6), 7.93 (s, 1H,
CH ), 8.13 (s, 1H, NH); FAB-MS: MH+=246.
0
0
(s, 9H, Boc), 1.45–1.60 (m, 4H, Pip-H3 ,5 ), 2.60–2.75 (m,
0
0
2H, Pip–H2,6), 3.70–3.85(m, 2H, Pip–H2 ,6 ), 4.52 (br s,

A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
2753
2H, CH2), 7.48 (d, J=8.3 Hz, 2H, Ar–H2,6), 7.61 (d,
J=8.3 Hz, 2H, Ar-H3,5), 9.32 (s, 1H, NH); FAB-MS:
MH+=373. Anal. calcd for C20H28N4O3: C, 64.49; H,
7.58; N, 15.04; found: C, 64.89; H, 7.66; N, 15.30.
(22%). Mp 148–151 ꢁC; IR (KBr) 3191, 3054, 2913,
2224, 1686 cmꢀ1; 300.15 MHz 1H NMR (DMSO-d6):
0
0
0.10–0.35 (m, 2H, Pip–H3,3 ,5,5 ), 0.47 (d, 3H, J=6.0 Hz,
0
0
0
CH3), 1.22–1.26 (m, 3H, Pip–H3,3 ,4,4 ,5,5 ), 2.30–2.45 (m,
0
0
1H, Pip–H2,2 ), 2.65–2.80 (m, 1H, Pip–H6,6 ), 3.45–3.65
0
0
tert-butyl 2-(1-azepanylcarbonyl)-2-(4-cyanobenzyl)hy-
drazinecarboxylate (4d). Yield: 1.58 g, (35%). Mp 136–
(m, 2H, Pip–H2,2 , 6,6 ), 4,38 (br d, 2H, CH2), 7.34 (d,
2H, J=8.3, Ar–H2,6), 7.64–7.80 (m, 5H, Ar–H3,5 and
Nph–H3,6,7), 8.02–8.18 (m, 3H, Nph–H4,5,8), 8.42 (s, 1H,
Nph–H1), 9.55 (s, 1H, NH) (two sets of signals); FAB-
MS: MH+=463. Anal. calcd for C25 H26 N4 O3 S: C,
64.91; H, 5.67; N, 12.11; found: C, 64.86; H, 5.58; N,
12.17.
141 ꢁC; IR (KBr) 3287, 2925, 2223, 1726, 1634 cmꢀ1
;
300.15 MHz 1H NMR (CDCl3): 1.44 (s, 9H, Boc), 1.51–
1.62 (m, 4H, Pip–H4,5), 1.65–1.78 ( m, 4H, Pip–H3,6),
3.30–3.49 (m, 4H, Pip–H2,7), 4.53 (s, 2H, CH2), 6.24 (br
s, 1H, NH), 7.51 (d, J=8.3 Hz, 2H, Ar–H2,6), 7.63 (d,
J=8.3 Hz, 2H, Ar–H3,5); FAB-MS: MH+=373. Anal.
calcd for C20H28N4O3: C, 64.49; H, 7.58; N, 15.04;
found: C, 64.79; H, 7.56; N, 15.34.
N0-(1-Azepanylcarbonyl)-N0-(4-cyanobenzyl)-2-naphtha-
lenesulfonohydrazide (5d). Yield: 0.26 g (22%). Mp 192–
194 ꢁC; IR (KBr) 3195, 2931, 2221, 1676 cmꢀ1
;
N0-(4-Cyanobenzyl)-N-(1-piperidinylcarbonyl)-2-naphtha-
lenesulfonylhydrazide (5a). HCl gas was bubbled for
45 min through a solution of tert-butyl 2-(4-cyano-
benzyl)-2-(1-piperidinylcarbonyl)hydrazinecarboxylate
4a (1.38 g, 3.86 mmol) in AcOH (15 mL) at room tem-
perature. AcOH was evaporated in vacuo to yield oily
2-(4-cyanobenzyl)-2-(1-piperidinylcarbonyl)hydrazinum
chloride (1.04 g, 3.53 mmol). To a solution of this com-
pound in CH2Cl2 (50 mL) cooled to 0 ꢁC were added
DIEA (1.23 mL, 7.06 mmol) and naphthalene-2-sulpho-
nylchloride (0.80 g, 3.53 mmol). The mixture was stirred
overnight at rt and concentrated in vacuo. The residue
was dissolved in EtOAc (100 mL) and washed with H2O
(50 mL), 1 M HCl (50 mL) and brine (50 mL). The
organic phase was dried over Na2SO4. After evapora-
tion of the solvent under reduced pressure, the product
was crystallized from diethylether as white crystals
(0.58 g, 33%). Mp 155–157 ꢁC; IR (KBr) 2940, 2233,
1801, 1630 cmꢀ1; 300.15MHz 1H NMR (DMSO-d6):
1.04–1.45 (m, 8H, Pip–H3,4,5), 2.90–3.20 (m, 4H, Pip-H2,6),
4.29 (br d, 2H, CH2), 7.16 (d, 2H, J=8.6 Hz, Ar–H2,6),
7.65–7.77 (m, 5H, Ar–H3,5 and Nph–H3,6,7), 8.05–8.17 (m,
3H, Nph–H4,5,8), 8.43 (s, 1H, Nph–H1), 9.59 (s, 1H, NH);
FAB-MS: MH+=449. Anal. calcd for C24H24N4O3S: C,
64.27; H, 5.39; N, 12.49; found: C, 63.98; H, 5.19; N, 12.49.
300.15 MHz H NMR (DMSO-d6): 0.63–1.48 (m, 8H,
Pip–H3,4,5,6), 3.18 (m, 4H, Pip–H2,7), 4.35 (br d, 2H,
CH2), 7.31 (d, J=8.3 Hz, 2H, Ar–H2,6), 7.55 (d,
J=8.3 Hz, 2H, Ar–H3,5), 7.64 (m, 1H, Nph–H7), 7.69
(m, 1H, Nph–H6), 7.77 (d, 1H, J=7.8, Nph–H3), 8.01
(d, 1H, J=7.8, Nph–H5), 8.05 (d, 1H, J=7.8, Nph–H4),
8.13 (d, 1H, J=7.8, Nph–H8), 8.45 (s, 1H, Nph–H1),
9.35 (s, 1H, NH); FAB-MS: MH+=463. Anal. calcd
for C25H26N4O3S: C, 64.91; H, 5.67; N, 12.11; found: C,
64.93; H, 5.71; N, 12.26.
1
N0-Hydroxy-4-{[2-(2-naphtylsulfonyl)-1-(1-piperidinylcar-
bonyl)hydrazino]methyl}benzenecarboximidamide
(6a).
NH2OH (0.54 g, 16.4 mmol) was added to the solution
of N0-(4-cyanobenzyl)-N-(1-piperidinylcarbonyl)-2-naph-
thalenesulfonylhydrazide 5a (3.68 g, 8.20 mmol) in
EtOH (100 mL). The mixture was heated at reflux tem-
perature overnight and evaporated under reduced pres-
sure to yield white foam 6a. Product was purified by
column chromatography on silica gel (CHCl3/MeOH,
50:1). Yield: 1.49 g (38%). Mp 194–197 ꢁC; IR (KBr)
3383, 1660, 1427, 1335 cmꢀ1; 600 MHz 1H NMR
(DMSO-d6): 1.03 (br s, 4H, Pip–H3,5), 1.24 (m, 2H,
J=5.75 Hz, Pip–H4), 2.93 (br s, 2H, Pip–H2,6), 3.06 (br
0
0
s, 2H, Pip–H2 ,6 ), 4.18 (br s, 1H, CH21), 4.38 (br s, 1H,
CH22), 5.78 (s, 2H, NH2), 7.16 (d, 2H, J=8.2 Hz, Ar–
5b, 5c and 5d were prepared by the same methods
described above for the synthesis of 5a using appro-
priate starting compounds.
H
2,6), 7.58 (d, 2H, J=8.2 Hz, Ar–H3,5), 7.68 (dd, 1H,
J=8.0, Nph–H7), 7.71 (dd, 1H, J=8.0 Hz, Nph–H6),
7.77 (d, 1H, J=8.6 Hz, Nph–H3), 8.04 (d, 1H,
J=8.0 Hz, Nph–H5), 8.09 (d, 1H, J=8.0 Hz, Nph–H4),
8.16 (d, 1H, J=8.0 Hz, Nph–H8), 8.46 (s, 1H, Nph–H1),
9.41 (s, 1H, NH), 9.61 (s, 1H, OH); FAB-MS:
MH+=482. Anal. calcd for C24H27N5O4S: C, 59.86; H,
5.65; N, 14.54; found C, 59.49; H, 5.33; N, 14.25.
N0-(4-Cyanobenzyl)-N0-[(2-methyl-1-piperidinyl)carbonyl]-2
naphthalenesulfonohydrazide (5b). Yield: 0.50 g (27%).
Mp 128–132 ꢁC; IR (KBr) 3240, 2950, 2228, 1652 cmꢀ1
;
300.15 MHz H NMR (DMSO-d6): 0.35–0.54 (m, 2H,
1
0
0
0
0
Pip–H3,3 ,5,5 ), 0.60–0.90 (m, 2H, Pip-H3,3 ,5,5 ), 1.15–1.50
0
(m, 5H, Pip–H4,4 and CH3), 2.55–2.70, 2.75–2.95, 3.34–
The following compounds were prepared by the same
methods described above for the synthesis of 6a, using
appropriate starting compounds.
0
3.36 (m,0 2H, Pip–H6,6 ), 3.65–3.80, 3.85–4.15 (m, 1H,
Pip–H2,2 ), 4.33 (br s, 2H, CH2), 7.36 (d, 2H,
J=8.3 Hz, Ar–H2,6), 7.67–7.75 (m, 5H, Ar–H3,5 and
Nph–H3,6,7), 8.02–8.16 (m, 3H, Nph–H4,5,8), 8.42 (s,
1H, Nph–H1), 9.44 (br s, 1H, NH) (two sets of signals);
FAB-MS: MH+=463. Anal. calcd for C25H26N4O3S:
C, 64.91; H, 5.67; N, 12.11; found: C, 65.17; H, 5.70;
N, 12.15.
N0 -Hydroxy-4-{[1-[(2-methyl-1-piperidinyl)carbonyl]-2-
(2-naphthylsulfonyl)hydrazino]methyl}benzenecarboximi-
damide (6b). Yield: 0.15 g (47%). Mp 174–176 ꢁC; IR
(KBr) 3369, 2936, 1647 cmꢀ1; 600 MHz 1H NMR
(DMSO-d6): 0.23 and 0.69 (br s, 3H, CH3), 0.54 and
1.01 ( br s, 1H, Pip–H5), 0.74 and 1.03 (br s, 1H, Pip–
0
H3), 1.12 and 1.14 (br s, 1H, Pip–H3 ), 1.26 and 1.31 (br
0
N0 -(4-Cyanobenzyl)-N0 -[(4-methyl-1-piperidinyl)carbo-
nyl]-2-naphthalenesulfonohydrazide (5c). Yield: 0.43 g
s, 1H, Pip–H4), 1.32 and 1.41 (br s, 1H, Pip–H5 ), 2.52

2754
A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
and 2.86 (br s, 1H, Pip–H6), 3.34 and 3.52 (br s, 1H,
S2 and S3 pocket. The benzamidoxime group was posi-
tioned in the S1 site, the piperidine or hexahydro-1H-
azepine ring in the S2 and the naphthalene-2-sulfonyl
group in the S3. As RMS values of the superimposed
active sites of proteins were in the range of 1.07–1.28 A,
we concluded that the binding of different inhibitors
does not change the conformation of the active site to
the extent that would preclude the techniques we used.
A sphere of water molecules (2R=40 A) was added
around the inhibitor in order to simulate the effects of
solvent. Minimal energies were computed for these
complexes using the Discover program until the max-
imal gradients were lower than 0.001 kcal/molA. After
minimization of these complexes, we performed a sys-
tematic conformational analysis of the flexible angles in
the inhibitor backbone. Bearing in mind that the con-
formational space of the inhibitors is restricted by three
specificity pockets (S1, S2 and S3), our search was
aimed at determining the differential flexibility of the
aza versus the a-CH structural elements in the enzyme
active site. Ten-degree steps rotations around angles a0–
a4 (Fig. 5a and b) resulted in 37 conformations for these
rotations around a single bond and in 1369 conforma-
tions for rotations around two subsequent bonds. The
potential energy of the 37 and 1369 inhibitor–enzyme
complexes was calculated. Since the active site of the
inhibitor was kept rigid, the majority of conformations
showed repulsion contacts and thus high positive energy
values. The conformations with favorable (negative)
energy values were numbered, and are presented in
Table 1. This procedure was also performed with the
carbon analogues.
0
0
Pip–H6 ), 3.57 and 3.87 (br s, 1H, Pip–H2 ), 4.11 and
4.19 (br s, 1H, CH22), 3.30 (br s, 1H, CH21), 5.85 (s,
2H, NH2), 7.13 and 7.15 (d, 1H, J=8.2 Hz, Ar–H2,6),
7.49 and 7.51 (d, 1H, J=8.2 Hz, Ar–H3,5), 7.64 (dd, 1H,
Nph–H7), 7.78 (dd, 1H, Nph–H6), 7.70 (d, 1H, Nph–
H3), 7.97 (d, 1H, Nph–H5), 8.03 (d, 1H, Nph–H4), 8.07
(d, 1H, Nph–H8), 8.36 and 8.37 (s, 1H, Nph–H1), 9.16
and 9.39 (br s, 1H, NH), 9.62 (s, 1H, OH); FAB-MS:
M+=496. Anal. calcd for C25H29N5O4S: C, 60.59; H,
5.90; N, 14.13; found: C, 60.27; H, 5.81; N, 13.91.
N0 -Hydroxy-4-{[1-[(4-methyl-1-piperidinyl)carbonyl]-2-
(2-naphthylsulfonyl)hydrazino]methyl}benzenecarboximi-
damide (6c). Yield: 0.26 g (56%). Mp 195–197 ꢁC; IR
(KBr) 3414, 2947, 1654 cmꢀ1; 600 MHz 1H NMR
(DMSO-d6): 0.06 (br s, 1H, Pip–H5), 0.25 (br s, 1H,
0
0
Pip–H3), 0.43 (s, 3H, CH3), 1.20 (br s, 3H, Pip–H3 ,4,5 ),
0
0
2.35 (br s, 1H, Pip–H2 ), 2.65 (br s, 1H, Pip–H6 ), 3.50
(br s, 1H, Pip–H6), 3.55 (br s, 1H, Pip–H2), 4.22 (br s,
1H, CH22), 4.39 (br s, 1H, CH21), 5.78 (s, 2H, NH2),
7.16 and 7.23 (d, 1H, J=8.2 Hz, Ar–H2,6), 7.59 and 7.79
(d, 1H, J=8.2 Hz, Ar–H3,5), 7.66 (dd, 1H, J=8.0, Nph–
H7), 7.71 (dd, 1H, J=8.0, Nph–H6), 7.77 (d, 1H, J=8.0,
Nph–H3), 8.03 (d, 1H, J=8.0 Hz, Nph–H5), 8.07 (d,
1H, J=8.0 Hz, Nph–H4), 8.14 (d, 1H, J=8.0 Hz, Nph–
H8), 8.45 (s, 1H, Nph–H1), 9.39 and 9.46 (br s, 1H,
NH), 9.62 (s, 1H, OH); FAB-MS: MH+=496. Anal.
calcd for C25H29N5O4S: C, 60.59; H, 5.90; N, 14.13;
found: C, 60.18; H, 5.93; N, 13.65.
4-{[1-(1-Azepanylcarbonyl)-2-(2-naphthylsulfonyl)hydra-
zino]methyl}-N0-hydroxybenzenecarboximidamide (6d).
Yield: 0.07 g (57%). Mp 193–195 ꢁC; IR (KBr) 3402,
Solution structure determination by 2-D NMR
1
3255, 2921, 1648 cmꢀ1; 600 MHz H NMR (DMSO-d6):
0
0
0.65 (br s, 2H, Pip–H4,5), 1.00 (br s, 2H, Pip–H4 ,5 ), 1.14
Nuclear magnetic resonance. NMR spectra were recor-
ded on Varian INOVA 600 MHz in DMSO-d6 at 298 K
and in mixture of DMSO-d6/H2O (80%/20%) H2O at
273 K. The sample concentrations were 10 mM. The
DQF-COSY,37 WET-DQF-COSY and WG-NOESY
were performed using gradients. The TOCSY,38,39
NOESY40 and ROESY experiments were recorded
using standard pulse sequences in the phase-sensitive
mode. The TOCSY spectra were recorded with a
MLEV-1739 mixing sequence of 60 ms and a 10-kHz
spin-lock field strength. NOESY spectra of 6a were
measured at four different mixing times 75, 150, 300,
and 450 ms. For the other compounds, a mixing time of
0
0
(br s, 2H, Pip–H3,6), 1.32 (br s, 2H, Pip–H3 ,6 ), 2.97 (br
0
0
s, 2H, Pip–H2,7), 3.13 (br s, 2H, Pip–H2 ,7 ), 4.17 (s, 1H,
CH21), 4.41 (s, 1H, CH22), 5.77 (s, 2H, NH2), 7.16 (d,
2H, J=8.1 Hz, Ar–H2,6), 7.58 (d, 2H, J=8.1 Hz, Ar–
H
3,5), 7.65 (dd, 1H, J=8.1 Hz, Nph–H7), 7.69 (dd, 1H,
J=8.1 Hz, Nph–H6), 7.77 (d, 1H, J=8.7 Hz, Nph–H3),
8.01 (d, 1H, J=8.1 Hz, Nph–H5), 8.05 (d, 1H,
J=8.1 Hz, Nph–H4), 8.13 (d, 1H, J=8.1 Hz, Nph–H8),
8.45 (s, 1H, Nph–H1), 9.35 (s, 1H, NH), 9.60 (s, 1H,
OH); FAB-MS: MH+=466. Anal. calcd for
C25H29N5O4S: C, 60.59; H, 5.90; N, 14.13; found: C,
60.27; H, 5.78; N, 14.26.
1
75 ms was used. The H sweep width was from 6800 to
8000 Hz for different compounds. Typically, the homo-
nuclear proton spectra were acquired with 4096 data
points in t2, 16–32 scans, 256–512 complex points in t1,
and a relaxation delay of 1–1.5 s. Data were processed
and analyzed with the FELIX software package from
Biosym Technologies. Spectra were zero-filled two times
and apodizied with a squared sine bell function shifted
by p/2 in both dimensions.
Molecular modelling
Molecular simulations were performed on a Silicon
Graphics workstation using Insight/Discover pro-
grams.35,36 Our starting point was X-ray structural data
of the human a thrombin–BMS189090 inhibitor com-
plex. Structural data for several other complexes of
human a thrombin with the inhibitors BMS186282,
PPACK, modified PPACK, p-amidinophenylpyruvate
and Ac-d-Phe-Pro-boroarg-OH were also used in the
following manner. By superimposing 21 amino acids
from the active site onto corresponding amino acids of
the BMS-189090 inhibitor complex, we found that the
position of various inhibitors was restricted to the S1,
The proton–proton distances were calculated from the
NOESY spectra measured at 75 ms using the two-spin
approximation and the integrated intensity of a geminal
pair of protons assumed to have a distance of 1.78 A.
All compounds are in the positive NOE regime when

A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
2755
studied in DMSO at 298 K and in negative NOE regime
when studied in mixture of DMSO/H2O at 273 K The
ROESY spectra of LK-632 in DMSO display significant
spin-diffusion effects, which were identified by the
examination of ROESY spectra collected at different
mixing times (75, 150, 200, and 250 ms). Based on this
the mixing time of 75 ms was chosen for determination
of distances of other compounds. A pseudo-atom cor-
rection was added for the methyl and methylene groups
and a ꢂ10% correction were applied to the distance in
order to produce the upper and lower limit constraints.
and the respective Km, which had been previously
determined [Ki=IC50/(1+S/Km)]. The Km for the sub-
strates used were determined under the conditions of the
test with at least five substrate concentrations ranging
from 0.5 to 15 times Km (Lottenberg et al.)42 according
to Eadie.43 The Km was 1132 mM for Chromozym-tPA,
108 mM for S-2366, 613 mM for S-2222 and 430 mM for
S-2251.
Acknowledgements
Molecular dynamics. MD simulations were performed
using DISCOVER (Consistent Valence Force Field)
and INSIGHT II software from Biosym Technologies.
Several simulations with different starting structures for
each of the compounds were carried out in DMSO.30
The molecule with all atoms treated explicitly was
centred in a cubic box (x=y=z=40 A) using three-
dimensional periodic boundary conditions. The dielec-
tric constant was set to 1.0. Neighbor lists for calcula-
tion of non-bonded interactions were updated every 10
fs within a radius of 14 A. The actual calculation of
non-bonded interactions was carried out up to a radius
of 12 A without the use of switching functions. A time
step of 1 fs was employed for the MD simulation. The
simulation protocol consisted of two minimization
cycles (steepest descent and conjugate gradients), first
with the solute fixed and then with all the atoms allowed
to move freely. The convergence criterion was
1 kcal Aꢀ1. The initial MD phase of the calculation
involved a gradual heating, starting from 100 and then
increasing to 150, 200, 250 and finally to 300 K in steps
of 0.5, 0.5, 5, 1, and 5 ps, respectively, each by direct
scaling of the velocities. The NMR-derived distance
restraints with a force constant of 10 kcal molꢀ1 Aꢀ1
were applied during the entire simulation. Configura-
tions were saved every 1 ps for another 200 ps of
dynamics. The last 100 ps of the trajectory was used for
analysis of NOE distance violations and dihedral
angles.
The authors are grateful to Ministry of Science and
Technology of Republic Slovenia for partial financial
support (grant No. J1-8889) and Lek, d.d. for support
of T.S. and U.U. We cordially thank Ms. Charlotte Taft
for linguistic help with the manuscript.
References and Notes
1. Stryer, L. Biochemistry, 4th ed.; W. H. Freeman: New
York, 1995; pp 252–258.
2. Mutschler, E. Arzneimittelwirkungen: Lehrbuch der Phar-
makologie und Toxikologie: mit einf’’hrenden Kapiteln in die
Anatomie, Physiologie und Patophysiologie, 7th ed.; Wis-
senschaftliche Verlagsgellschaft: Stuttgart, 1996; pp 413–414,
425–430.
3. (a) Hirsh, J.; Fuster, V. Circulation 1994, 89, 1449. (b)
Hirsh, J.; Fuster, V. Circulation 1994, 89, 1469.
4. Taparelli, C.; Metternich, R.; Erhardt, C.; Cook, N. S.
TIPS 1993, 14, 366.
5. Kimball, S. D. Curr. Pharm. Design 1995, 1, 441.
6. Ripka, W. C.; Vlasuk, G. P. Ann. Rep. Med. Chem. 1997,
32, 71.
7. Wiley, M. R.; Fisher, M. J. Exp. Opin. Ther. Patents 1997,
7, 1265.
8. Sanderson, P. E. J.; Naylor-Olsen, A. M. Current Med.
Chem. 1998, 5, 289.
9. Rewinkel, J. B. M.; Adang, A. E. P. Curr. Pharm. Design
1999, 5, 1043.
10. Menear, K. Current Med. Chem. 1998, 5, 457.
11. Renatus, M.; Bode, W.; Huber, R.; Stuerzebecher, J.;
Stubbs, M. J. Med. Chem. 1998, 41, 5445.
12. Gardell, S. J.; Sanderson, P. E. J. Coron. Artery Dis. 1998,
9, 75.
Enzyme assay
13. Wagner, J.; Kallen, J.; Ehrhardt, C.; Evenou, J.-P.;
Wagner, D. J. Med. Chem. 1998, 41, 3664.
14. Misra, R. N.; Kelly, Y. F.; Brown, B. R.; Roberts,
D. G. M.; Chong, S.; Seiler, S. M. Biorg. Med. Chem. Lett.
1994, 4, 2165.
15. Balasubramanian, B. N. Bioorg. Med. Chem. 1995, 3, 999.
16. Kikomoto, R.; Tonomura, S.; Hara, H.; Ninomya, K.;
Maruyama, A.; Sugano, M.; Tamao, Y. Biochem. Biophys.
Res. Commun. 1981, 101, 440.
17. Kikumoto, R.; Tamao, Y.; Tezuka, T.; Tonomura, S.;
Hara, H.; Ninomiya, K.; Hijikata, A.; Okamoto, S. Biochem-
istry 1984, 23, 85.
18. Kim, S.; Hwang, S. Y.; Kim, Y. K.; Yun, M.; Oh, Y. S.
Bioorg. Med. Chem. Lett. 1997, 7, 769.
19. 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.
20. 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.
Spectrophotometric enzyme tests were run in microtiter
plates in a final volume of 125 mL using the following
conditions: the inhibition of factor VIIa was tested with
TF/FVIIa/Ca2+ at a final concentration of 15 nM/
5 nM/5 mM with the substrate Cromozym-tPA at
500 mM f.c. Thrombin was tested at 1 nM with the sub-
strate S-2366 at 200 mM f.c. Factor Xa was tested at
3 nM with S-2222 at 200 nM. Trypsin was tested at
6 nM using the substrate S-2251 at 200 nM. The reac-
tion kinetics between enzymes and substrates were lin-
ear both with time and the concentration of the enzyme
chosen. Inhibitors were dissolved in DMSO and tested
at 100 mM as the highest concentration. Inhibitors were
diluted using HNPT buffer consisting of HEPES
100 mM, NaCl 140 mM, PEG 6000 0.1% and Tween 80
0.02%, pH 7.8. Full concentration dose–response curves
were run for inhibitors showing inhibition >50% for
any one enzyme at 100 mM. Apparent Ki were calculated
according to Cheng and Prusoff 41 based on the IC50

2756
A. Zega et al. / Bioorg. Med. Chem. 9 (2001) 2745–2756
21. (a) Danilewitcz, J. C.; Kobylecki, R. J.; Ellis, D. Inter-
national Patent Aplication WO 9513274, 1995. (b) Allen, M.;
Abel, S. M.; Barber, C. G.; Cussans, N. J.; Danilewicz, J. C.;
Ellis, D.; Hawkeswood, E.; Herron, M.; Holland, S.; Fox, D.
N. A.; James, K.; Kobylecki, R. J.; Overington, J. P.; Pandit,
J.; Parmar, H.; Powling, M. J.; Rance, D. J.; Taylor, W.;
Shepperson, N. B. 215th ACS National Meeting, 1998, Dallas,
TX; Abstract MEDI 64.
31. Malley, F. M.; Tabernero, L.; Chang, C. Y.; Ohringer,
S. L.; Roberts, D. G. M.; Das, J.; Sack, J. S. Protein Sci. 1996,
5, 221.
32. Bauer, M.; Brandstetter, H.; Turk, D.; Stuerzebecher, J.;
Bode, W. Semin. Throm. Hemos. 1993, 19, 352.
33. Pearse, G. A., Jr.; Pflaum, R. T. J. Am. Chem. Soc. 1959,
81, 6505.
34. Fassler, A.; Bold, G.; Capraro, H. G.; Cozens, R.;
¨
22. Gante, J. Synthesis 1989, 259, 405.
Mestan, J.; Poncioni, B.; Rosel, J.; Titelnot-Blomley, M.;
¨
Lang, M. J. Med. Chem. 1996, 39, 3203.
35. InsightII Version 97.0; MSI Inc.: San Diego, USA, 1997.
36. Discover, Version 95.0; MSI Inc.: San Diego, USA, 1995.
37. Rance, M.; Sorensen, O. W.; Bodenhausen, G.; Wagner,
23. Gante, J. Angew. Chem. 1994, 106, 1780.
24. Powers, J. C.; Gupton, B. F. Methods Enzymol. 1977, 46,
208.
25. Kurtz, A. N.; Niemann, C. J. Org. Chem. 1961, 26, 1843.
26. Dorn, C. P.; Zimmerman, M.; Yang, S. S.; Yurewicz,
E. C.; Ashe, B. M.; Frankshun, R.; Jones, H. J. Med. Chem.
1977, 20, 1464.
27. Spatola, A. F. Chemistry and Biochemistry of Amino
Acids, Peptides, and Proteins; Weinstein, B., Ed.; Marcel Dek-
ker: New York, 1983; pp 267–357.
28. Dutta, A. S.; Morley, J. S. J. Chem. Soc., Perkin Trans. 1
1976, 244.
G.; Ernst, R. R.; Wutrich, K. Biochem. Biophys. Res. Com-
mun. 1983, 117, 479.
¨
38. Braunschweiler, L.; Ernst, R. R. J. Magn. Reson. 1983, 53,
521.
39. Bax, A.; Davis, D. G. J. Magn. Reson. 1985, 65, 355.
40. Jeener, J.; Meier, B. H.; Bachmann, P.; Ernst, R. R. J.
Chem. Phys. 1979, 71, 4546.
41. Cheng, Y. C.; Prusoff, W. H. Biochem. Pharmacol. 1973,
22, 3099.
42. Lottenberg, R.; Hall, J. A.; Blinder, M.; Binder, E. P.;
Jackson, C. R. Biochem. Biophys. Acta 1983, 742, 539.
43. Eadie, G. S. J. Biol. Chem. 1942, 146, 85.
29. Eckert, H.; Foster, B. Angew. Chem. Int. Ed. 1987, 26,
894.
30. Mierke, D. F.; Kessler, H. J. Am. Chem. Soc. 1991, 113,
9466.