Design and synthesis of highly constrained factor Xa inhibitors: Amidine-Substituted bis(benzoyl)-[1,3]-diazepan-2-ones and bis(benzylidene)-bis(gem-dimethyl)cycloketones

Article abstract of DOI:10.1016/S0968-0896(03)00332-8

Two conformationally constrained templates have been designed to provide selective inhibitors of the coagulation cascade serine protease, Factor Xa (FXa). The most active inhibitor, 2,7-bis[(Z)-p-amidinobenzylidene)]-3,3,6,6-tetramethylcycloheptanone, exh

Full text of DOI:10.1016/S0968-0896(03)00332-8

Bioorganic & Medicinal Chemistry 11 (2003) 3379–3392
Design and Synthesis of Highly Constrained Factor Xa Inhibitors:
Amidine-Substituted Bis(benzoyl)-[1,3]-diazepan-2-ones
and Bis(benzylidene)-bis(gem-dimethyl)cycloketones
Jian Cui,a,b David Crich,b Donald Wink,b Matthew Lam,c Arnold L. Rheingold,c
David A. Case,d WenTao Fu,a Yasheen Zhou,a Mohan Rao,d Arthur J. Olsond
and Michael E. Johnsona,*
aCenter for Pharmaceutical Biotechnology, University of Illinois at Chicago, 900 S. Ashland Avenue, M/C 870,
Chicago, IL 60607-7173, USA
bDepartment of Chemistry, University of Illinois at Chicago, 845 West Taylor Street, Chicago, IL 60607, USA
cDepartment of Chemistry, University of Delaware, Newark, DE 19716, USA
dDepartment of Molecular Biology, The Scripps Research Institute, La Jolla, CA 92037, USA
Received 7 November 2002; revised 10 May 2003; accepted 13 May 2003
Abstract—Two conformationally constrained templates have been designed to provide selective inhibitors of the coagulation cas-
cade serine protease, Factor Xa (FXa). The most active inhibitor, 2,7-bis[(Z)-p-amidinobenzylidene)]-3,3,6,6-tetra-
methylcycloheptanone, exhibits a Ki of 42 nM against FXa, with strong selectivity against thrombin (1000-fold), trypsin (300-fold)
and plasmin (900-fold). With only two freely rotatable bonds, molecular modeling suggests that one amidine group is positioned
into the S1 pocket, forming hydrogen bonds with the side chain of Asp189, similar to other amidine-based inhibitors, with the
second benzamidine positioned into the S4 pocket in a position to form strong cation–pi bonding with the S4 aryl cage. We suggest
that this interaction plays an important role in the specificity of these inhibitors against other serine proteases.
# 2003 Elsevier Ltd. All rights reserved.
Introduction
are expected to become the leading cause of death
worldwide within two decades.1ꢀ7
Activation of the blood clotting or coagulation system is
necessary to prevent blood loss after injury, but uncon-
trolled intravascular activation of coagulation can cause
pathological thrombosis. This can lead to the serious
clinical consequences of myocardial infarction, stroke,
pulmonary embolism, deep-vein thrombosis and dis-
seminated intravascular coagulation. Thromboembolic
diseases are the leading causes of morbidity and mor-
tality in most industrialized societies. Genesis of throm-
bosis can be induced by endothelial injury,
hyperviscosity and decrease in fibrinolysis. The anato-
mical origin of thrombosis disorders differs according to
diverse types of pathophysiology. Thrombotic disorders
are a major cause of disability and death in the US, and
Thrombin and Factor Xa (FXa) play critical roles in
thrombosis and hemostasis, with thrombin initiating
clot formation, and FXa providing the sole mechanism
for thrombin activation. Thrombin and related pro-
teases involved in blood coagulation have been studied
extensively over many years. Although heparin and the
coumarin anticoagulants are long-established treat-
ments in coagulation disorders, side effects, limitations
of efficacy and bleeding complications with these agents
have stimulated intense searches for more direct and
specific inhibitors of key enzymes in the coagulation
cascade.1ꢀ4,6
There is substantial evidence suggesting that FXa
should be an attractive target for inhibitor design. Sev-
eral organisms have developed specific peptidic inhibi-
tors that selectively target FXa, and provide very potent
*Corresponding author. Tel.: +1-312-996-9114; fax: +1-312-413-
9303; e-mail: mjohnson@uic.edu
0968-0896/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0968-0896(03)00332-8

3380
J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
anticoagulant activity. For example, tick anticoagulant
peptide exhibits very selective inhibition of FXa with a
Ki of 180 pM, and, in a variety of animal models, has
been shown to be equal to, and in many cases, superior
to traditional heparin therapy or recombinant hirudin
as an antithrombotic agent.8 Similarly, the leech protein
antistasin exhibits selective inhibition of FXa, and has
been shown to be substantially more effective than
heparin in models of arterial thrombosis and in reoc-
clusion prophylaxis.6,9,10 A series of anticoagulant pep-
tides produced by the hookworm Ancylostoma caninum
also exhibit selective inhibition of both FXa and Factor
VIIa.11 Additionally, several comparisons between
direct thrombin inhibitors and direct FXa inhibitors in
animal models have shown inhibitors of FXa to be
superior antithrombotic agents12ꢀ14 with lower bleeding
risk–antithrombotic benefit ratios.15,16
Inhibitor Design
Owing to the two double bonds adjoining the central ring,
BABCH has three possible isomers (Z,Z), (E,Z) and (E,E)
(Fig. 1). As a first step in developing alternative templates,
we developed a detailed conformational analysis of
BABCH. To obtain computationally reliable results, we
found it necessary to utilize Gaussian ab initio calcula-
tions. Our Gaussian calculations, performed at the HF/3-
21G level to provide reliable energies and geometries,
indicate that the extended (E,E) isomer has a con-
formational energy about 2 kcal/mol lower than the V-
like (Z,Z) isomer, which is consistent with the con-
formational population distribution that can be inferred
from the observed Ki values.20,21 The lowest energy
conformers of all three isomers have the same central
ring conformation, a pseudo-chair conformation. The
relative positioning of the benzamidine rings in the cal-
culated (Z,Z)BABCH conformation is also very close to
that experimentally observed in the BABCH/tissue-type
plasminogen activator (tPA) complex.23
A number of FXa inhibitor designs have been described
recently, with most based on moderately flexible linkers
connecting various substituents that exhibit affinity for
the FXa S1 and S4 subsites.1ꢀ5,7,17,18 Docking studies
have suggested that binding in an L-shaped configur-
ation, with a ‘recognition’ moiety inserted into the S1
specificity pocket, and an aromatic or aryl cationic group
inserted into the aromatic cage formed by Tyr99, Phe174
and Trp215 is especially favored.19 The BABCH inhibi-
tors, shown in Figure 1, are particularly intriguing probes
of specificity between FXa and related serine pro-
teases.20,21 With essentially only two degrees of freedom
within the molecule (rotation of the benzamidine groups),
the most active structure (the Z,Z or syn isomer shown
here) exhibits very high FXa inhibitory activity, with a
Kiꢁ0.66 nm, and an approximate 800-fold difference in
inhibitory activity between FXa and thrombin.20,21 The
initial description of these inhibitors suggested that they
were in the extended E,E configuration.22 More recent
work, however, has shown that there is photo conversion
between the E,E, E,Z and Z,Z isomers in the presence of
light, and that the Z,Z isomer is 25,000 times more active
than the E,E isomer, but is present at only trace levels at
equilibrium in the presence of light.20,21
The observed binding mode of BABCH with tPA23 dif-
fers from that initially predicted for the (Z,Z)BABCH-
FXa complex.20 Accordingly, to determine the most
probable binding mode, we docked a molecular model
of the calculated (Z,Z)BABCH into the active site of the
FXa crystal structure24 with the Autodock pro-
gram,25,26 using the calculated (Z,Z)BABCH con-
formation, and allowing free rotation only for the
benzamidine groups and limited rotation of ꢂ25ꢃ for
the amidine groups. We find that the most probable
binding mode positions one benzamidine group into the
S1 pocket, and the second into the S4 pocket, as shown
in Figure 2. In this model, the benzamidine in the S1
specificity pocket exhibits bidentate hydrogen bonding
with the Asp189 carboxylate. The heptanone ring car-
bonyl is positioned to permit hydrogen bonding to the
Gly216 amide, shifted somewhat further away from the
Tyr99 wall of the S4 site than observed for the BABCH–
tPA complex.23 The distal benzamidine is inserted at a
slight angle into the S4 pocket, with the amidine slightly
In the present work, we have devised two alternative,
conformationally constrained structural templates that
stabilize the Z,Z or syn configuration, and have devel-
oped an initial lead that has an FXa dissociation con-
stant of ꢁ42 nM, and 250–1000-fold selectivity against
trypsin, plasmin and thrombin.
Figure 2. Relaxed stereo view of the computational model of
(Z,Z)BABCH (1) docked into the active site of Factor Xa. The
BABCH model conformation was determined by Gaussian calcula-
tions. In docking, the benzamidine groups were allowed to rotate
freely and the amidines were allowed ꢂ25ꢃ rotation with respect to the
plane of the phenyl ring.
Figure 1. BABCH isomers.

J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
3381
pointed up toward the carbonyl cluster,27 forming a
Chemistry
hydrogen bond with the carbonyl of Thr98, and nearly
centered in the aryl cage formed by the Tyr99 phenole,
the Trp215 indole and the Phe174 phenyl rings. Overall,
the modeled binding mode of BABCH with FXa is
similar to that experimentally observed for the
BABCH–tPA complex,23 despite the differences in the
S4 subsite between the two enzymes. The distal amidine
also appears to be almost ideally positioned within the
S4 aryl cage to provide additional cation–pi stabili-
zation for this binding mode that should be unique to
Fxa.28 Based on this analysis, an alternative template
that can stabilize the (Z,Z)BABCH configuration
should provide an effective FXa inhibitor with good
enzymatic specificity.
Template I
The synthesis of 1,3-bis(3-amidinobenzoyl)-[1,3]-diaza-
pan-2-one (2) is shown in Scheme 1. Deprotonation of
the bis-amide (3) with a single equivalent of lithium
hexamethyldisilazide (LiHMDS), followed by treatment
with 4-nitrophenyl chloroformate and then a second
equivalent of LiHMDS furnished the N,N0-diacyl urea
(4) in 45% overall yield. This was most effectively con-
verted to the bis amidine (2) via the symmetrical thioa-
mide (5) and the bis-thioimidate (6) by treatment with
To take advantage of the Z,Z-BABCH structure, but at
the same time avoid its stability problems, we focused
on modifying the central skeleton. Two potential isos-
teric scaffolds were designed to render the symmetrical
cycloheptadienone core of (Z,Z)BABCH into a less
photochemically sensitive and more synthetically fea-
sible pharmacophore (Fig. 3). The first template is a
seven-membered cyclic urea in which two amide bonds
were chosen to replace the problematic double bonds.
The new amide bonds have the advantages of keeping
the original double bond characteristics, thereby main-
taining a similar geometry, and avoiding the isomer
problem. The energy difference between the syn and anti
configuration is calculated to be about 8.4 kcal/mol,
favoring the folded syn configuration (Fig. 4). The sec-
ond template is a bis-gem-tetramethylcycloheptanone.
In this system,1,3 A-strain between the four methyl
groups and the phenyl rings should destabilize the (E,E)
isomer and so favor a syn-conformation similar to the
(Z,Z)BABCH. The energy difference between the (Z,Z)
and the (E,E) isomers is calculated to be about 2.8 kcal/
mol, favoring a syn-conformation similar to that of
(Z,Z)BABCH. We therefore predicted that the designed
templates should maintain the same geometry as
(Z,Z)BABCH, with a substantially decreased possibility
of isomerization. Based on this analysis, the synthesis of
both templates I and II was undertaken with several
variations.
Figure 4. Comparative energies for the folded (syn) conformations
relative to the extended (anti) conformations for (Z,Z)BABCH and
templates I and II, as calculated by Gaussian (HF/3,21).
Scheme 1. (a) 1 equiv LiHMDS, THF, ꢀ78 ꢃC; p-NO2C6H4OC
(¼O)Cl; 1 equiv LiHMDS, ꢀ78 ꢃC to rt; (b) (EtO)2P(¼S)SH, rt; (c)
MeI, Á; (d) NH3–HOAc, 0 ꢃC (two steps).
Figure 3. Alternate templates for the central heptanone ring of
BABCH expected to stabilize the folded syn configuration.

3382
J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
diethyl dithiophosphate,29ꢀ31 methyl iodide, and
ammonium acetate. The attempted conversion of the
bis-nitrile 4 to the bis-amidine 2 by the more usual Pin-
ner reaction,32ꢀ34 involving sequential treatment with
Hal in methanol and then ammonia, served only to
illustrate the relative susceptibility of the system toward
nucleophilic attack on the tricarbonyl system, and was a
harbinger of the problems subsequently encountered in
the corresponding p-substituted series.
At the same time, the commodity substance isophorone
was converted to the lower homologue 3,3,5,5-tetra-
methylcyclohexanone (17), also by conjugate methyl-
ation, in order to provide a second analogue for testing
and a more readily available substance with which to
develop the double aldol chemistry.
Thus, 2,6 - bis[(Z) - p - cyanobenzylidene] - 3,3,5,5 - tetra-
methylcyclohexanone (16) was prepared through a
modified Mukaiyama reaction as shown in Scheme 4. The
TMS enol ether 18 was reacted with acetal 19 and tita-
nium tetrachloride in refluxing dichloromethane to give
20 in 67% yield along with 10% of the dicoupled product
16. Iteration of the Mukaiyama reaction sequence on the
isolated mono-coupled product 20 provided first the silyl
enol ether 21 and then the desired doubly acylated pro-
duct 16 in yields of 95 and 62%, respectively. Analogues,
2-[(Z)-m-cyanobenzylidene]-6-[(Z)-p-cyanobenzylidene]-
3,3,5,5-tetramethylcyclohexanone (22) and 2,6-bis[(Z)-
m-cyanobenzylidene]-3,3,5,5-tetramethylcyclohexanone
(23), were synthesized following the same procedure.
NOESYspectra were obtained and the ( Z,Z)-configur-
ations were confirmed for all. However, after exposure
to ambient light for several days, olefin isomerization
occurred, and 20 was converted into a mixture of E and
Z isomers. Similarly, light stimulated the isomerization
of 16 into a mixture of (Z,Z) and (Z,E) isomers. This
may due to the relatively low energy difference (ꢁ1
kcal/mol) between the Z- and E-configurations of these
particular cyclohexanone compounds.
All attempts at application of the same sequence to
preparation of the corresponding 1,3-bis(4-amidino-
benzoyl)-[1,3]-diazapan-2-one (7) failed at the last step
(Scheme 2), apparently because of the enhanced elec-
trophilicity of the diacyl urea system resulting from
conjugation of the powerfully electron-withdrawing
amidinium groups. An alternative synthesis involving
hydrogenolysis of the less basic, stable N,N0-diacetoxy
bisamidine (13), derived by reaction of the bisthioimi-
date 11 with hydroxylamine followed by acetylation,
also failed to provide 7.
Template II
It was envisaged that the target compound, 2,7-bis[(Z)-p-
amidinobenzylidene)]-3,3,6,6-tetramethylcycloheptanone,
could be synthesized by double aldol condensation of
two equivalents of 4-cyanobenzaldehyde with
3,3,6,6-tetramethylcycloheptanone. The literature ap-
proaches35,36 to 3,3,6,6-tetramethylheptanone (15) were
deemed inappropriate for our purposes, and a route
(Scheme 3) was developed in which 4,4-dimethylcyclo-
hexanone was converted to 3,3,6-trimethylcyclohepte-
none (14), according to a literature protocol,37,38
followed by conjugate addition in the standard manner.
Scheme 3. (a) CuBr–Me2S, MeLi.
Scheme 2. (a) 1 equiv LiHMDS, THF, ꢀ78 ꢃC; p-NO2C6H4OC
(¼O)Cl; 1 equiv LiHMDS, THF, ꢀ78 ꢃC to rt; (b) (EtO)2P(¼S)SH, rt;
(c) MeI, Á; (d) NH2OH–HCl, Et3N, 0 ꢃC; (e) CH3COCl, Et3N, 0 ꢃC;
(g) Pd/C, H2; (g) NH3–HOAc, 0 ꢃC.
Scheme 4. (a) CuBr–Me2S, MeLi/Et2O; (b) TMSCl, Et3N, NaI;
(c) TiCl4/CH2Cl2; (d) TMSCl, Et3N, NaI; (e) TiCl4/CH2Cl2, 19, Á.

J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
3383
In the tetramethylcycloheptanone series, the aldol con-
densation was significantly more difficult and it was
found necessary to use the higher boiling point solvent
1,2-dichloroethane (bp 85 ꢃC) to increase the reaction
temperature. After heating 24 with three equivalents of
25, a more reactive acetal of p-cyanobenzaldehyde, and
two equivalents of titanium tetrachloride in refluxing
1,2-dichloroethane overnight, a mixture of mono-
coupling product 26 and di-coupling product 28 were
obtained in yields of 41 and 5%, respectively. After a
further Mukaiyama reaction on 26, 2,7-bis[(Z)-p-
cyanobenzylidene] - 3,3,6,6 - tetramethylcycloheptanone
(28) was finally obtained in 43% yield from 26 (Scheme
5). Analogues, 2-[(Z)-m-cyanobenzylidene]-7-[(Z)-p-
cyanobenzylidene] - 3,3,6,6 - tetramethylcycloheptanone
(29) and 2,7-bis[(Z)-m-cyanobenzylidene]-3,3,6,6-tetra-
methylcycloheptanone (30), were synthesized following
the same procedure. NOESYspectra were again
obtained and (Z,Z)-configurations were confirmed for
all compounds. Compared to the cyclohexanones 20
and 16, 26 and 28 are photochemically stable, and no
significant isomerization occurred after exposure to
light. This obviously reflects the more significant energy
difference between the E- and Z-configurations, as
determined computationally, between the cyclohepta-
none and cyclohexanone series. The more hindered
nature of the cycloheptanone is also reflected in the
considerably more forcing aldol conditions required.
Finally, the complete series of nitriles was converted to
the corresponding amidines by Pinner’s method
(Scheme 6). The final products were isolated by HPLC.
NOESYspectra were obtained and ( Z,Z)-configur-
ations were confirmed for all (Tables 1 and 2; additional
data in Supporting Information).
Results and Discussion
The crystal structure of the cyclic urea template cyano
intermediate 9 is shown in Figure 5, from which it can be
Scheme 6. (a) HCl(g), CH3OH; (b) NH4OAc, CH3OH.
Scheme 5. (a) TMSCl, Et3N, NaI; (b) TiCl4, Á, ClCH2CH2Cl;
(c) TMSCl, Et3N, NaI; (d) 25, TiCl4, Á, ClCH2CH2Cl.
Table 2. 1H and NOESYdata for bisamidine compounds 34–36
Table 1. 1H and NOESYdata for bisamidine compounds 31–33
Proton
34
35
36
Proton
31
32
33
1H NOESY
1H
NOESY
1H
NOESY
1H NOESY
1H
NOESY
1H
NOESY
1
2
3
4
5
6
7
10
20
30
40
50
60
70
1.81 (s) H2
1.29 (s) H1,3
6.71 (s) H2,4
1.82 (s)
1.30 (s)
6.73 (s)
H2
H1,3
H2,4
1.81 (s)
1.30 (s)
6.72 (s)
H2
H1,3
H4,7
1
2
3
4
5
6
7
20
30
40
50
60
70
1.90 (s) H2
1.39 (s) H1,3
6.86 (s) H2,4
1.87 (s)
1.37 (s)
6.84 (s)
H2,20
H1,3
H2,4
1.86 (s)
1.38 (s)
6.83 (s)
H2
H1,3
H4,7
7.30 (d) H3 7.88–7.25 (m) H3 7.67–7.21 (m) H3
7.52 (d)
7.88–7.25 (m)
7.67–7.21 (m)
7.67–7.21 (m)
7.67–7.21 (m) H3
7.43 (d) H3 7.70–7.30 (m) H3 7.71–7.26 (m) H3
7.70–7.30 (m)
7.63 (d)
7.71–7.26 (m)
7.71–7.26 (m)
7.71–7.26 (m) H3
1.82 (s)
1.29 (s)
6.72 (s)
H20
H1,30
1.37 (s)
6.84 (s)
H1,30
H20,40,70
H20,40,70
7.88–7.25 (m) H30
7.88–7.25 (m)
7.88–7.25 (m)
7.70–7.30 (m) H30
7.70–7.30 (m)
7.70–7.30 (m)
7.88–7.25 (m) H30
7.70–7.30 (m) H30

3384
J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
Table 3. In vitro inhibitory activities of cyclic ureas
Ki (mM)
Compd
R1
H
R2
Factor Xa Thrombin Trypsin Plasmin
5
13.9
110
260
120
>500
>500
24
10
2
H
0.82
15.2
H
0.15
8.8
1.7
Figure 5. ORTEP plot for the crystal structure of 9.
seen that the predicted twist structure for the central ring
is observed, with the two ‘arms’ in the folded syn con-
figuration, analogous to the Z,Z configuration of
BABCH, also as predicted, positioning the two phenyl
rings with a separation almost identical to that of the
separation observed for BABCH crystallized with tPA,23
indicating that the cyclic urea template is an excellent
mimic for the central heptanone ring of BABCH.
position also enters the S4 subsite at an unfavorable
angle, generating a bumping collision with the Tyr99
side chain that forces the amidine up away from the
Trp215 indole ring. These unfavorable interactions at
both the S1 and S4 subsites are a probable source for
the sharply reduced activities of analogues containing
amidines in the 3-position.
FXa, thrombin, trypsin and plasmin Ki’s were obtained
from in vitro enzyme affinity assays at 37 ꢃC (Table 3).
Bisthioamides 5 and 10 were tested together with
3,30-bisamidine 2. Surprisingly, the 4,40-bisthioamide 10
showed a FXa Ki of 0.82 mM, which is relatively potent
considering there are no positive charges on it and the
thioamide could only form weak hydrogen bonds.
3,30-Bisamidine 2 showed FXa Ki at 150 nM. All three
compounds were selective toward FXa over thrombin,
trypsin and plasmin.
In order to experimentally check the geometry of the
designed template, the structure of one of the synthetic
intermediates, 2,7-bis[(Z)-p-cyanobenzylidene]-3,3,6,6-
tetramethylcycloheptanone (28), was determined by X-
ray crystallography (Fig. 6). As expected, the four
methyl groups at the b positions were able to force the
two phenyl rings up and maintain the olefin double
bonds in the (Z,Z) configuration. Thus, the crystal
structure is in very good agreement with the computa-
tionally derived structure and is confirmed to be in the
same geometry as (Z,Z)BABCH.
All six gem-dimethyl amidine analogues were tested to
obtain their FXa, thrombin, trypsin and plasmin Ki’s
from in vitro enzyme affinity assays at 37 ꢃC (Table 4).
Km and nmax values were also obtained under the same
conditions from the Michaelis–Menten equation.39 Ki
values were calculated using Dixon analysis.40 Among
the various analogues, 2,7-bis[(Z)-p-amidinobenzyli-
dene)]-3,3,6,6-tetramethylcycloheptanone (34) is the
most active, with an FXa Ki of 42 nM. It also shows
high selectivity over thrombin (1000-fold), trypsin (300-
fold) and plasmin (900-fold).
The other bis(gem-dimethyl)cycloketone compounds
with amidino groups in the meta-positions or mixed
meta-, para-positions showed neither good activity nor
selectivity. They are all in the same activity range as
(E,E)BABCH (FXa Ki=17 mM) and showed different
inhibitory characteristics than the 4,40-bisamidines, 34
and 31. Modeling suggests that a 3-benzamidine in the
proximal position has only one NH2 group that can
hydrogen bond to the Asp189 carboxylic acid, losing the
bidentate hydrogen bonding characteristic of most ami-
dine-based inhibitors. A 3-benzamidine in the distal
Figure 6. ORTEP plot for the crystal structure of 28.

J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
3385
Table 4. In vitro inhibitory activities of the amidine-substituted (bis)benzylidene-(bis)gem-tetramethylcycloketones
Ki (mM)
Compd
n
R1
H
R2
R3
R4
H
Factor Xa
0.13
Thrombin
8.3
Trypsin
3.1
Plasmin
38
31
1
32
33
34
35
36
1
1
2
2
2
H
H
H
3.7
12
55
42
7.9
11
32
290
>500
25
H
5.3
9.5
11
H
H
H
0.042
3.2
H
H
2.3
11
55
H
6.5
>500
The inhibitory activity of 34, the gem-dimethyl analogue
of (Z,Z)BABCH is quite promising, but is substantially
lower than that of (Z,Z)BABCH itself. To better
understand the interactions of the various analogues
with FXa, computational models based on the crystal
structures of the intermediates for both templates (Figs
5 and 6) were docked with FXa. For the cyclic urea
template (I), the amidine group within the S1 site for the
cyclic urea model (Fig. 7) also forms bidentate hydrogen
bonds with Asp189. The central carbonyl of the cyclic
urea ring is positioned similar to that of the
(Z,Z)BABCH model, permitting hydrogen bonding to
the Gly216 amide. The distal benzamidine group is
inserted into the S4 aryl cage in a fashion similar to that
observed for the BABCH–tPA complex,23 with the
benzamidine ring nearly perpendicular to the indole ring
of Trp215, and one side of the amidine positioned to
form a hydrogen bond with the carbonyl of Thr98. With
pharmacophores other than the benzamidines used
here, this template shows promise for other inhibitor
designs.
For the gem-dimethyl heptanone template (II), the ben-
zamidine group is rotated within the S1 subsite to form
a bridged hydrogen bond between one side of the ami-
dine and the two Asp189 carboxyl oxygens, in contrast
to the bidentate positioning of Z,Z-BABCH (Fig. 8).
The other side of the amidine forms a hydrogen bond
with the hydroxyl oxygen of Tyr228. The heptanone
ring is consequently rotated away from the backbone
chain, with the carbonyl roughly perpendicular to the
peptide backbone, but close to the amide of Trp215.
The distal benzamidine is inserted into the S4 aryl cage
in an orientation similar to that of the (Z,Z)BABCH
model, although tilted up somewhat more with respect
to the plane of the Trp215 indole ring, and not within
hydrogen bonding proximity of the Thr98 carbonyl.
The interaction between the Tyr99 ring and the axial
methyl of this inhibitor pushes the heptanone ring back
from the S4 pocket, and appears to produce a less opti-
mal binding mode, which may be the source of the
reduced inhibitory activity, as compared to
(Z,Z)BABCH. The distal benzamidine is positioned
Figure 7. Relaxed stereo view of a model of 7 docked into the active
site of Factor Xa using Autodock. The model of 7 was constructed by
substituting amidines for the cyano groups in the intermediate crystal
structure 9 (Fig. 5). The benzamidine groups were allowed to rotate
freely, and the amidines allowed ꢂ25ꢃ rotation with respect to the
plane of the phenyl ring. All other angles were maintained at the
values found in the crystal structure.

3386
J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
Figure 8. Relaxed stereo view of a superposition model of 34 (red) and (Z,Z)-BABCH (green) docked into the active site of factor Xa using
Autodock. The model of 34 was constructed by substituting amidines for the cyano groups in the intermediate crystal structure 28 (Fig. 6). The
benzamidine groups were allowed to rotate freely; the amidines were allowed ꢂ25ꢃ rotation with respect to the plane of the phenyl ring. All other
angles were maintained at the values found in the crystal structure.
within the S4 aryl cage in a manner nearly ideal for
maximizing cation–pi interactions.28 Comparative
Gaussian calculations for a benzamidine centered in an
aryl cage versus a similar cage with aliphatic side chains
replacing the Phe and Tyr rings show a difference of
about 3 kcal/mol in favor of the aryl cage, and suggest
that cation–pi stabilization of the amidine within the
aryl cage is the most probable source for the observed
enzymatic specificity (Table 4).
trends, in relation to structure, are also intriguing. There
is only modest variation in activity for all of these struc-
tures with thrombin, trypsin and plasmin, suggesting
that there is specificity pocket (Asp189) recognition of
the benzamidine group, but only modest affinity for
other moieties of the inhibitor, consistent with our design
goals. For FXa, however, there is a clear preference for
placing both amidines in the 4-position, for both the six-
and seven-member central rings. Likewise, just as for
the parent (Z,Z)BABCH, there is a significant increase
in activity in going from the six-member cyclohexane
analogue to the seven-member cycloheptane analogue.
(Going to larger ring sizes is not likely to be useful, as
larger ring systems for the parent BABCH structure
exhibit significantly reduced activity.) These results
provide a strong foundation for further work with this
series of conformationally highly constrained inhibitors.
Conclusions
The cyclic urea template is structurally quite attractive,
but the central ring lacks the stability necessary for
benzamidine substituents with p-amidines. It may,
however, be a useful template for less basic groups that
have been used in a variety of other designs. This will be
explored in future work.
Experimental
General
With only two freely rotatable bonds, the most active
structure, 34, has an FXa Ki of 42 nM. It also shows
strong selectivity against thrombin (1000-fold), trypsin
(300-fold) and plasmin (900-fold). Modeling indicates
that it should bind in a mode qualitatively similar to
that of (Z,Z)BABCH, with cation–pi interactions pro-
viding enzymatic selectivity, but with subtle differences
in positioning within both the S1 and S4 subsites. A
bumping collision between the Tyr99 phenolic ring and
the adjacent methyl of the gem-dimethyl analogue is the
most probable source for the difference in activity
between 34 and (Z,Z)-BABCH. The high selectivities
against thrombin and plasmin are particularly
encouraging, indicating that we have, in fact, achieved
our goal of selective inhibition of FXa. The activity
All solvents were dried and distilled by standard meth-
ods. All reactions were carried out in dried solvents
1
under an atmosphere of argon. H NMR spectra were
acquired at either 300 or 400 MHz and 13C NMR spec-
tra at 75.5 MHz. Chemical ionization (CI) and electro-
spray (ESI) mass spectra were recorded on a Finnigan
MAT 90 or LCQ mass spectrometer, respectively.
FABHRMS were recorded with a VG 7070-HF mass
spectrometer at the University of Minnesota mass spec-
trometry laboratory. Melting points are uncorrected.
Elemental analyses were performed by Midwest Micro-
lab (Indianapolis, IN, USA).

J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
3387
N,N0-Bis(4-cyanobenzoyl)-1,4-diaminobutane (8). A sus-
pension of 4-cyano-benzoic acid (25.0 g, 0.17 mol) in
thionyl chloride was heated to reflux until it became
clear. Thionyl chloride was distilled off under reduced
pressure. The residue was dissolved in anhydrous DMF
(100 mL). The resulting solution was cooled to 0 ꢃC and
triethylamine (30.0 mL, 0.19 mol) was added slowly.
Then 1,4-diaminobutane (8.5 mL, 0.085 mol) was added
dropwise via syringe. The reaction mixture was warmed
up to room temperature and stirred overnight. The
reaction mixture was poured into ice-cold 10% HCl
solution. The resulting precipitate was filtered and
washed twice with ice cold 10% NaOH solution and
then several times with water. The solid was dried at
50 ꢃC under vacuum to give N,N0-bis(4-cyanobenzoyl)-
1,4-diaminobutane (26.4 g, 90%) as a slightly yellow
(down), 128.7 (down), 119.0 (up), 114.1 (up), 45.4 (up),
27.2 (up); CI–MS m/e 373 (M+1)+; IR (CHCl3, cast) n
2230, 1692 cmꢀ1. Anal. calcd for C21H16N4O3: C, 67.73;
H, 4.33; N, 15.05. Found: C, 67.72; H, 4.35; N, 14.98.
1,3-Bis(3-cyanobenzoyl)-[1,3]-diazepan-2-one (4). 4 was
prepared by the same method described above for the
synthesis of 9 using 3 as starting material (45% from 3,
mp 204–205 ꢃC). 1H NMR (300 MHz, DMSO-d6) d
7.97–7.56 (8H, m), 4.24 (4H, bs), 1.88 (4H, bs); APT 13C
NMR (75.5 MHz, DMSO-d6) d 170.2 (up), 159.9 (up),
137.6 (up), 135.4 (down), 132.5 (down), 131.4 (down),
130.5 (down), 118.9 (up), 112.4 (up), 45.5 (up), 27.2
(up); CI–MS 373 (M+1)+; IR (CHCl3, cast) n 2233,
1691 cmꢀ1. Anal. calcd for C21H16N4O3: C, 67.73; H,
4.33; N, 15.05. Found: C, 67.45; H, 4.38; N, 14.89.
powder (mp 264–265 ꢃC). H NMR (300 MHz, DMSO-
1
d6) d 8.72 (2H, t, J=5.4 Hz), 7.99 (4H, d, J=8.7 Hz),
7.95 (4H, d, J=8.7 Hz), 3.31 (4H, m), 1.58 (4H, m);
APT 13C NMR (75.5 MHz, DMSO-d6) d 165.6 (up),
139.4 (up), 133.3 (down), 128.9 (down), 119.2 (up),
1,3-Bis(4-thiocarbamylbenzoyl)-[1,3]-diazepan-2-one (10).
Several drops of H2O were added to a suspension of 9
(0.8 g, 2.2 mmol) in diethyl dithiophosphate (3 mL).
The reaction mixture was stirred at room temperature
for 2 days, and then quenched with H2O (50 mL). The
resulting precipitate was filtered and washed with a large
volume of H2O, EtOAc (10 mL) and hexanes (50 mL) to
give 1,3-bis(4-thiocarbamylbenzoyl)-1,3-diazepaꢃn-2-one
114.3 (up), 39.9 (up), 27.3 (up); CI–MS m/e 347
+
.
(M+1) . Anal. calcd for C20H18N4O2 0.2H2O C, 68.64;
H, 5.30; N, 16.01. Found: C, 68.56; H, 5.43; N, 16.01.
N,N0-Bis(3-cyanobenzoyl)-1,4-diaminobutane (3). 3 was
prepared by the same method described above for the
synthesis of 8 using 3-cyanobenzoic acid as starting
material (90% from 3-cyanobenzoic acid, mp 203–
(0.9 g, 95%) as a yellow powder (mp 235–238 C). H
1
NMR (300 MHz, DMSO-d6) d 9.98 (2H, s), 9.57 (2H, s),
7.81 (4H, d, J=8.1 Hz), 7.50 (4H, d, J=8.1 Hz), 4.21
(4H, m), 1.87 (4H, m); APT 13C NMR (75.5 MHz,
DMSO-d6) d 200.1 (up), 171.5 (up), 160.0 (up), 142.4
(up), 138.3 (up), 128.0 (down), 127.6 (down), 45.3 (up),
27.6 (up); CI–MS m/e 373 (M+1–2H2S)+. Anal. calcd
for C21H20N4O3S20.4H2O: C, 56.33; H, 4.68; N, 12.51; S,
14.32. Found: C, 56.64; H, 4.62; N, 12.34; S, 14.37.
204 ꢃC). H NMR (300 MHz, DMSO-d6) d 8.70 (2H, t,
1
J=5.6 Hz), 8.25 (2H, s), 8.14 (2H, d, J=7.8 Hz), 7.99
(2H, d, J=7.8 Hz), 7.68 (2H, t, J=7.8 Hz), 3.32 (4H,
bs), 1.58 (4H, bs); APT 13C NMR (75.5 MHz, DMSO-
d6) d 165.1 (up), 136.4 (up), 135.4 (down), 132.9 (down),
131.6 (down), 130.6 (down), 119.2 (up), 112.5 (up), 39.9
(up), 27.3 (up); CI–MS m/e 347 (M+1)+. Anal. calcd
for C20H18N4O2 1/4H2O: C, 68.46; H, 5.31; N, 15.97.
Found: C, 68.51; H, 5.31; N, 15.82.
1,3-Bis(3-thiocarbamylbenzoyl)-[1,3]-diazepan-2-one (5).
5 was prepared by the same methods described above
for the synthesis of 10 using 4 as starting material (95%
ꢃ
1
from 4, mp 228–230 C). H NMR (300 MHz, DMSO-
d6) d 9.99 (2H, s), 9.61 (2H, s), 8.02–7.41 (8H, m), 4.24
(4H, bs), 1.89 (4H, bs); APT 13C NMR (75.5 MHz,
DMSO-d6) d 199.5 (up), 171.5 (up), 160.1 (up), 139.9
(up), 135.8 (up), 130.6 (down), 130.3 (down), 128.9
(down), 127.3 (down), 45.4 (up), 27.6 (up); CI–MS m/e
373 (M+1–2H2S)+. Anal. calcd for C21H20N4O3S2: C,
57.25; H, 4.58; N, 12.72; S, 14.56. Found: C, 56.96; H,
4.60, N, 12.45; S, 14.56.
1,3-Bis(4-cyanobenzoyl)-[1,3]-diazepan-2-one (9). Under
a nitrogen atmosphere, a solution of LiHMDS in THF
(1.0 M, 6.3 mL) was added dropwise to a suspension of
8 (2.0 g, 5.8 mmol) in 250 mL THF at ꢀ78 ꢃC. The
reaction mixture was stirred for 30 min, and 4-nitro-
phenyl chloroformate (1.3 g, 6.3 mmol) was added in.
The mixture was warmed up to room temperature and
stirred for another 30 min. Then the reaction mixture
was cooled to ꢀ78 ꢃC again and a solution of LiHMDS
in THF (1.0 M, 6.3 mL) was added in dropwise. After
stirring at room temperature overnight, the mixture was
concentrated under reduced pressure. Ethyl acetate (250
mL) was added to the residue and the resulting pre-
cipitate was separated by filtration. The filtrate was
washed quickly with ice-cold satd NaHCO3 solution
(250 mL) and H2O (250 mLꢄ3), dried over Na2SO4 and
concentrated. The residue was purified by column
chromatography on silica gel eluted with EtOAc/hex-
anes (1:1, v/v, Rf=0.45) to give 1,3-bis(4-cyano-
benzoyl)-1,3-diazepan-2-one (0.86 g, 40%) as a white
1,3-Bis(3-amidinobenzoyl)-[1,3]-diazepan-2-one (2).
A
suspension of 5 (230 mg, 0.523 mmol) in acetone (10
mL) was treated with MeI and the yellow mixture was
heated to reflux for 1 h. The resulting brown mixture
was concentrated under reduced pressure to give the
crude product 6 that was used for the next step without
further purification.
The brown solid was suspended in ice-cold anhydrous
acetonitrile (15 mL) and treated with ammonium acetate
(85 mg, 1.1 mmol) at 0 ꢃC. The reaction mixture was
stirred at room temperature overnight and turned to a
white cloudy mixture. The resulting precipitate was fil-
tered out, washed with ice-cold acetonitrile and ethyl
ether, and dried under vacuum to give 1,3-bis(3-amidi-
solid (mp 219–221 ꢃC). H NMR (300 MHz, DMSO-d6)
1
d 7.85 (4H, d, J=8.1 Hz), 7.65 (4H, d, J=8.1 Hz), 4.21
(4H, m), 1.88 (4H, m); APT 13C NMR (75.5 MHz,
DMSO-d6) d 170.6 (up), 159.8 (up), 140.4 (up), 133.2

3388
J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
nobenzoyl)-1,3-diazepan-2-one, hydroiodide salt (220
mg, 64%) as a white powder (mp 243–245 ꢃC). 1H NMR
(300 MHz, DMSO-d6) d 9.20 (6H, bs), 7.88–7.60 (8H, m),
4.26 (4H, bs), 1.91 (4H, bs); APT 13C NMR (75.5 MHz,
DMSO-d6) d 170.9 (up), 165.4 (up), 160.0 (up), 136.9
(up), 132.8 (down), 131.5 (down), 129.9 (down), 129.2
(up), 127.5 (down), 45.6 (up), 27.5 (up). Anal. calcd for
ether (3ꢄ50 mL). The combined extracts were washed
with saturated aqueous NaCl and dried (Na2SO4). After
concentration under reduced pressure, 15 (1.08 g, 99%)
was collected as a colorless liquid.13 1H NMR (CDCl3)
d 2.36 (4H, s), 1.55 (4H, s), 1.00 (12H, s); 13C NMR
(75.5 MHz, CDCl3) d 212.0, 57.2, 37.7, 32.1, 30.2; ES–
MS m/e 169.1 (M+1)+.
.
.
C21H22N6O3 1.9HI 0.4CH3COOH): C, 38.88; H, 3.82;
N, 12.48. Found: C, 39.14; H, 3.80; N, 12.52.
3,3,6,6 - Tetramethyl - 1 - trimethylsilyloxycycloheptene
(24). Standard protocol for the formation of trimethylsi-
lylenol ethers. Sodium iodide was heated to 140 ꢃC for 3
h then dissolved in acetonitrile to give a 1 M solution.
Chlorotrimethylsilane (3.9 mL, 30.8 mmol) was added
dropwise at 0 ꢃC to a stirred mixture of 15 (3.45 g, 20.5
mmol) and triethylamine (8.6 mL, 61.5 mmol). After
warming to room temperature the above sodium iodide
solution (26 mL, 26 mmol) was added dropwise to the
reaction mixture resulting in an exothermic reaction and
the formation of a white precipitate in the light brown
solution. After stirring overnight the reaction was quen-
ched by addition of ice-water and extracted with pen-
tane. The extracts were quickly washed sequentially with
ice-cold saturated aqueous solutions of ammonium
chloride and sodium hydrogen carbonate, then with
brine, dried (Na2SO4) and concentrated under reduced
pressure to give the silyl enol ether as a brown liquid
(4.86 g, 99%). 1H NMR (CDCl3) d 4.69 (1H, s), 2.08 (2H,
s), 1.45 (4H, m), 0.99 (6H, s), 0.95 (6H, s), 0.16 (9H, s).
No further purification was attempted on this compound
which was used directly in the next step.
1,3-Bis(4-N-hydroxyamidinobenzoyl)-[1,3]-diazepan-2-one
(12). 11 was prepared by the method described above for
the synthesis of 6 using 10 (250 mg, 0.57 mmol) as start-
ing material. The brown solid was dissolved in anhy-
drous DMF (10 mL). At 0 ꢃC, Et3N (0.18 mL, 1.25
.
mmol) and NH2OH HCl (90 mg, 1.25 mmol) were added
respectively. The reaction mixture was stirred at room
temperature overnight, then neutralized with Et3N (0.18
mL, 1.25 mmol) and diluted with ethyl acetate. The
organic phase was washed with H2O, dried over Na2SO4
and passed through a short silica gel column. After con-
centrating under vacuum, 1,3-bis(4-N-hydroxyami-
dinobenzoyl)-1,3-diazepan-2-one (170 mg, 68%) was
collected as a white powder (mp >280 ꢃC dec). 1H NMR
(300 MHz, DMSO-d6) d 9.82 (2H, s), 7.67 (4H, d, J=8.4
Hz), 7.48 (4H, d, J=8.4 Hz), 5.86 (4H, s), 4.22 (4H, bs),
1.86 (4H, bs); APT 13C NMR (75.5 MHz, DMSO-d6) d
171.8 (up), 160.1 (up), 151.0 (up), 136.8 (up), 136.3 (up),
128.0 (down), 126.1 (down), 45.4 (up), 27.6 (up). Anal.
calcd for C21H22N6O5: C, 57.53; H, 5.05; N, 19.17.
Found: C, 57.85; H, 5.26; N, 19.05.
1-Trimethylsilyloxy-3,3,5,5-tetramethylcyclohexene (18).
This was prepared by the general protocol for the for-
mation of trimethylsilyl enol ethers in 99% yield from
3,3,5,5-tetramethylcyclohexanone42,43 (17). 1H NMR
(CDCl3) d 4.66 (1H, s), 1.76 (2H, s), 1.26 (2H, s), 1.00
(6H, s), 0.98 (6H, s), 0.18 (9H, s). No further purifica-
tion was attempted on this compound which was used
directly in the next step.
1,3-Bis(4-N-acetoxyamidinobenzoyl)-[1,3]-diazepan-2-one
(13). 12 (150 mg, 0.34 mmol) was dissolved in DMF (3
mL). At0 ꢃC, Et3N (105 mL, 0.75mmol)and CH3COCl (55
mL, 0.75 mmol) were added, respectively. The reaction
mixture was stirred at room temperature overnight, then
diluted with ethyl acetate and washed with H2O. The
organic phase was dried over Na2SO4 and then con-
centrated under vacuum. The residue was purified by col-
umn chromatography on silica gel eluted with EtOAc/
hexanes (v/v=2:1, Rf=0.65) to give 1,3-bis(4-N-acetox-
yamidinobenzoyl)-1,3-diazepan-2-one (160 mg, 90%) as
2-[(Z)-p-Cyanobenzylidene]-3,3,5,5-tetramethylcyclohex-
anone (20). At 0 ꢃC, to a solution of 18 (226 mg, 1.0
mmol) and 1944,45 (190 mg, 1.1 mmol) in dichloro-
methane (5 mL) was added dropwise, a solution of
TiCl4 in dichloromethane (1.2 mL, 1.0 M, 1.2 mmol).
The resulting dark mixture was stirred at room tem-
perature overnight, then quenched with water (3 mL).
The aqueous phase was extracted with ethyl acetate
(3ꢄ5 mL) and the combined extracts were washed with
saturated aqueous NaCl, dried (Na2SO4) and con-
centrated under reduced pressure. The residue was pur-
ified by column chromatography to give 16 (40 mg,
10%) and the title product, 20 (180 mg, 67%) as a yel-
low liquid which turned to a wax-like solid on standing.
a white solid (mp >280 ꢃC dec). H NMR (300 MHz,
1
DMSO - d6) d 7.69 (4H, d, J=8.0 Hz), 7.54 (4H, d,
J=8.0 Hz), 6.87 (4H, s), 4.22 (4H, bs), 2.12 (6H, s), 1.88
(4H, bs); APT 13C NMR (75.5 MHz, DMSO-d6) d 171.6
(up), 169.3 (up), 156.6 (up), 138.0 (up), 134.9 (up), 128.0
(down), 127.6 (down), 45.3 (up), 27.5 (up), 20.7 (down).
Anal. calcd for C25H26N6O7 H2O: C, 55.55; H, 5.22; N,
15.55. Found: C, 55.37; H, 5.09; N, 15.80.
3,3,6,6-Tetramethylcycloheptanone (15).35,36 At 0 ꢃC, to
a stirred suspension of cuprous bromide–dimethyl sul-
fide complex (1.4 g, 6.8 mmol) in ether (15 mL) was
added dropwise, an ethereal solution of methyllithium
(11.5 mL, 1.4 M, 16.1 mmol) resulting in a colorless
solution. After stirring for 5 min, a solution of 1441 (1.0
g, 6.5 mmol) in ether (15 mL) was added dropwise after
which the reaction mixture, from which yellow pre-
cipitate was formed, was stirred at room temperature
for 40 min, then poured into 100 mL NH4Cl/NH4OH
buffer (pH ꢅ8). The aqueous phase was extracted with
Mp 55–58 ꢃC; H NMR (CDCl3) d 7.54 (2H, d, J=6.2
1
Hz), 7.32 (2H, d, J=6.3 Hz), 6.39 (1H, s), 2.43 (2H, s),
1.73 (2H, s), 1.26 (6H, s), 1.11 (6H, s); IR (CHCl3, cast)
n 2225, 1734, 1694, 1603 cmꢀ1. No further purification
was attempted on this compound which was used
directly in the next step.
On exposure to ambient laboratory light for more than
one week, 20 was converted into a mixture of E- and

J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
3389
Z-isomers. Separation of the isomers by column chro-
matography was difficult due to their close Rf values.
The NMR spectrum of the E-isomer was obtained
7.63 (4H, d, J=8.4 Hz), 7.43 (4H, d, J=8.4 Hz), 6.86
(2H, s), 1.90 (2H, s), 1.39 (12H, s); 13C NMR (metha-
nol-d4) d 211.7, 166.9, 151.2, 143.2, 129.9, 129.8, 127.5,
127.0, 52.4, 38.1, 30.3; ES–MS m/e 415.3 (M+1)+.
Anal. calcd for C26H30N4O 2CF3COOH 0.5H2O: C,
55.30; H, 5.10; N, 8.60. Found: C, 55.30; H, 5.40; N,
8.76.
1
from a mixture with the Z-isomer. H NMR (CDCl3) d
.
.
7.53 (2H, d, J=8.4 Hz), 7.31 (2H, d, J=8.1 Hz), 6.38
(1H, s), 2.42 (2H, s), 1.72 (2H, s), 1.26 (6H, s), 1.11
(6H, s).
1-Trimethylsilyloxy-2-(p-cyanobenzylidene)-3,3,5,5-tetra-
methyl-cyclohexanone (21). This was obtained in 95%
yield from 20 by the standard protocol for the formation
2,6-Bis[(Z)- m-cyanobenzylidene]-3,3,5,5-tetramethylcy-
clohexanone (23). This was prepared in 50% yield from
18 and 3-dimethoxymethyl-benzonitrile46,47 by the
method described above for the synthesis of 16. Mp
107–108 ꢃC; 1H NMR (CDCl3) d 7.47–7.20 (8H, m),
6.55 (2H, s), 1.80 (2H, s), 1.31 (12H, s); 13C NMR
(CDCl3) d 197.1, 150.9, 137.3, 133.5, 132.5, 131.3, 128.9,
128.2, 118.8, 112.3, 52.6, 38.6, 31.1; ES–MS m/e 381.2
(M+1)+. Anal. calcd for C26H24N2O: C, 82.07; H, 6.36;
N, 7.36. Found: C, 82.20; H, 6.51; N, 7.34.
1
of trimethylsilyl enol ethers. H NMR (CDCl3) d 7.50
(2H, d, J=8.3 Hz), 7.29 (2H, d, J=8.5 Hz), 6.28 (1H, s),
4.94 (1H, s), 1.53 (2H, s), 1.24 (6H, s), 1.11 (6H, s), ꢀ0.19
(9H, s). No further purification was attempted on this
compound which was used directly in the next step.
2,6-Bis[(Z)-p-cyanobenzylidene]-3,3,5,5-tetramethylcyclo-
hexanone (16). At 0 ꢃC, to a solution of 21 (210 mg, 0.62
mmol) and 8a44,45 (120 mg, 0.68 mmol) in dichloro-
methane (5 mL) was added dropwise a solution of TiCl4
in dichloromethane (0.75 mL, 1.0 M, 0.75 mmol). The
resulting dark mixture was heated at reflux for 3 h, and
then quenched with water (3 mL). The aqueous phase
was extracted with ethyl acetate (3ꢄ5 mL) and the
combined extracts were washed with brine, dried
(Na2SO4) and concentrated under reduced pressure.
The residue was purified by column chromatography to
give 16 (145 mg, 62%) as a yellow solid. Mp 151–
2,6-Bis[(Z)-m-amidinobenzylidene]-3,3,5,5-tetramethylcy-
clohexanone (33). This was prepared in 70% yield from
23 by the method described above for the synthesis of
31. Mp 170–173 ꢃC; 1H NMR (methanol-d4) d 7.71–7.26
(8H, m), 6.83 (2H, s), 1.86 (2H, s), 1.38 (12H, s); 13C
NMR (methanol-d4) d 209.3, 162.3, 151.2, 138.1, 134.2,
129.6, 128.9, 128.6, 126.8, 121.9, 52.2, 38.4, 30.4; ES–
MS m/e 208.2 1/2(M+2)2+, 415.3 (M+1)+. Anal.
.
.
calcd for C26H30N4O 2CF3COOH 0.75H2O: C, 54.92;
H, 5.15; N, 8.54. Found: C, 55.02; H, 5.30; N, 8.95.
154 ꢃC; H NMR (CDCl3) d 7.43 (4H, d, J=8.3 Hz),
1
7.25 (4H, d, J=8.1 Hz), 6.60 (2H, s), 1.82 (2H, s), 1.33
(12H, s); 13C NMR (CDCl3) d 197.0, 151.3, 141.2, 132.0,
130.0, 129.5, 119.1, 111.4, 52.9, 38.8, 31.3; ES–MS m/e
2-[(Z)-m-Cyanobenzylidene]-6-[(Z)-p-cyanobenzylidene]-
3,3,5,5-tetramethylcyclohexanone (22). This was pre-
pared in 65% yield from 21 and 3-dimethoxymethyl-
benzonitrile46,47 by the method described above for the
381.2 (M+1)+; IR (cast) n 2229, 1676, 1615 cmꢀ1
.
synthesis of 16. Mp 142–143 ꢃC; H NMR (CDCl3) d
1
Anal. calcd for C26H24N2O 0.2H2O: C, 81.31; H, 6.40;
N, 7.29. Found: C, 81.56; H, 6.30; N, 7.23.
7.51–7.17 (8H, m), 6.56 (1H, s), 6.55 (1H, s), 1.80 (2H,
s), 1.32 (12H, s); 13C NMR (CDCl3) d 197.4, 151.6,
151.3, 141.1, 137.6, 133.7, 133.0, 132.0, 131.4, 129.9,
129.0, 128.9, 128.3, 119.2, 118.9, 112.6, 111.4, 52.9, 39.0,
38.9, 31.3, 31.3; ES–MS m/e 381.2 (M+1)+. Anal. calcd
On exposure to ambient laboratory light for more than
one week, 16 was converted into a mixture of Z,Z- and
Z,E-isomers. The separation of the isomers by column
chromatography was difficult due to their close Rf
.
for C26H24N2O 0.6H2O: C, 79.81; H, 6.49; N, 7.16.
Found: C, 79.64; H, 6.23; N, 7.41.
1
values. The H NMR spectrum of the Z,E-isomer was
therefore obtained from a mixture with the Z,Z-isomer.
1H NMR (CDCl3) d 7.62–7.23 (8H, m), 6.61 (1H, s),
6.60 (1H, s), 1.69 (2H, s), 1.32 (6H, s), 1.13 (6H, s).
2-[(Z)-m-Amidinobenzylidene]-6-[(Z)-p-amidinobenzyl-
idene]-3,3,5,5-tetramethylcyclohexanone (32). This was
prepared in 56% yield from 22 by the method described
above for the synthesis of 31. Mp 163–165 ꢃC; 1H NMR
(methanol-d4) d 7.70–7.30 (8H, m), 6.84 (2H, s), 1.87 (2H,
s), 1.37 (12H); 13C NMR (methanol-d4) d 212.4, 165.7,
164.4, 153.5, 151.2, 150.8, 138.3, 134.2, 133.9, 132.9,
130.1, 129.8, 128.9, 128.5, 127.4, 126.7, 52.3, 38.1, 30.4;
ES–MS m/e 208.3 1/2(M+2)2+, 415.3 (M+1)+. Anal.
2,6-Bis[(Z)-p-amidinobenzylidene]-3,3,5,5-tetramethylcy-
clohexanone (31). Dry HCl gas was passed at 0 ꢃC
through a solution of 21 (100 mg, 0.26 mmol) in anhy-
drous ethyl ether (5 mL) and methanol (2 mL) for 5
min. The reaction vessel was then sealed and stored at
0 ꢃC for 2 days. After removal of the solvents under
reduced pressure, the residue was dissolved in methanol
and treated with ammonium acetate (25 mg, 0.31 mmol)
at 0 ꢃC followed by stirring at room temperature over-
night. The resulting mixture was diluted with water to a
final volume of 5 mL. Ten 50-mL aliquots of this solu-
tion were purified by semi-preparative HPLC (Beckman
Ultrasphere C-18 column) using a 25–30% gradient of
solvent A (0.1% TFA in acetonitrile) in solvent B (0.1%
TFA in water) to give 31 (12 mg, 70% from 16), which
was collected as a white solid after concentration under
vacuum. Mp 185–187 ꢃC; 1H NMR (methanol-d4) d
.
.
calcd for C26H30N4O 2CF3COOH 0.5H2O: C, 55.30; H,
5.10; N, 8.60. Found: C, 55.44; H, 5.52; N, 8.77.
2-[(Z)-p-Cyanobenzylidene]-3,3,6,6-tetramethylcyclohep-
tanone (26). A solution of TiCl4 in dichloromethane (2.0
mL, 1.0 M, 2.0 mmol) was added dropwise at room tem-
perature to a solution of 24 (240 mg, 1.0 mmol) and 2548
(530 mg, 3.0 mmol) in 1,2-dichloroethane (3 mL) in a
flask equipped with a reflux condenser. The resulting
dark mixture was heated at reflux overnight, then cooled
to room temperature and quenched with water (10 mL)

3390
J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
and the aqueous phase was extracted with ethyl acetate
(3ꢄ10 mL). The combined organic layers were washed
with brine, dried (Na2SO4) and concentrated under
reduced pressure. The residue was purified by column
chromatography to give 28 (20 mg, 5%) as a yellow solid
and the title product, 26 (113 mg, 41%) as a yellow oil
which turned to a wax-like solid on standing. Mp 84–
85 ꢃC; 1H NMR (CDCl3) d 7.55 (2H, d, J=6.3 Hz), 7.31
(2H, d, J=6.2 Hz), 6.33 (1H, s), 2.19 (2H, s), 1.51 (4H,
m), 1.24 (6H, s), 0.97 (6H, s); 13C NMR (CDCl3) d 210.9,
158.0, 140.9, 132.6, 129.2, 122.7, 119.2, 111.2, 57.6, 37.9,
37.4, 37.0, 33.9, 29.9, 29.4; ES–MS m/e 282.2 (M+1)+.
No further purification was attempted on this compound
which was used directly in the next step.
(1H, s), 1.82 (4H, s), 1.30 (6H, s), 1.29 (6H, s); 13C
NMR (methanol-d4) d 190.5, 153.6, 153.0,143.4, 138.5,
138.1, 134.2, 130.6, 129.8, 129.7, 129.4, 128.9, 128.8, 128.1,
127.4, 126.7, 126.4, 38.8, 38.8, 38.5, 38.3, 28.6, 28.3; ES–
MS m/e 215.2 1/2(M+2)2+, 429.3 (M+1)+; FABHRMS
m/e 429.2652 (M+H)+, calcd for C27H33N4O 429.2654.
Anal. calcd for C27H32N4O 1.7CF3COOH 0.75H2O: C,
57.42; H, 5.58; N, 8.81. Found: C, 57.49; H, 5.52; N,
8.89.
.
.
2-[(Z)-m-Cyanobenzylidene]-3,3,6,6-tetramethylcyclohep-
tanone (37). This was prepared in 32% yield from 24
and 3-dimethoxymethyl-benzonitrile46,47 by the method
described above for the synthesis of 26. Mp 65–68 ꢃC;
1H NMR (CDCl3) d 7.50–7.34 (4H, m), 6.31 (1H, s),
2.19 (2H, s), 1.58–1.49 (4H, m), 1.24 (6H, s), 0.98 (6H,
s); ES–MS m/e 282.1 (M+1)+. No further purification
was attempted on this compound which was used
directly in the next step.
1-Trimethylsilyloxy-2-[(Z)-p-cyanobenzylidene]-3,3,6,6-
tetramethylcycloheptanone (27). This was prepared in
quantitative yield by the standard protocol for the for-
mation of trimethylsilyl enol ethers. 1H NMR (CDCl3) d
7.53 (2H, d, J=7.6 Hz), 7.24 (2H, d, J=7.6 Hz), 6.26
(1H, s), 4.69 (1H, s), 1.50 (4H, m), 1.19 (6H, s), 1.02 (6H,
s), 0.05 (9H, s). No further purification was attempted on
this compound which was used directly in the next step.
2,7-Bis[(Z)-m-cyanobenzylidene]-3,3,6,6-tetramethylcy-
cloheptanone (38). was prepared in 30% yield from 24
and 3-dimethoxymethyl-benzonitrile46,47 by the method
described above for the synthesis of 26. Mp 157–160 ꢃC;
1H NMR (CDCl3) d 7.39–7.13 (8H, m), 6.42 (2H, s), 1.76
(4H, s), 1.28 (12H, s); 13C NMR (CDCl3) d 201.4, 153.8,
137.6, 133.9, 132.8, 131.1, 128.9, 127.3, 119.0, 112.3, 38.6,
2,7-Bis[(Z)-p-cyanobenzylidene]-3,3,6,6-tetramethylcyclo-
heptanone (28). This was prepared in 43% yield from 27
by the method described above for the synthesis of 26.
Mp 167–168 ꢃC; 1H NMR (CDCl3) d 7.34 (4H, d, J=5.0
Hz), 7.13 (4H, d, J=4.9 Hz), 6.48 (2H, s), 1.76 (4H, s),
1.28 (12H, s); 13C NMR (CDCl3) d 200.8, 153.5, 141.0,
130.9, 129.8, 128.1, 118.8, 110.9, 38.5, 38.0, 29.3; ES–MS
m/e 395.2 (M+1)+; IR (CHCl3, cast) n 2226, 1668, 1604
.
38.5, 29.7. Anal. calcd for C27H26N2O 0.4H2O: C, 80.73;
H, 6.72; N, 6.97. Found: C, 80.67; H, 6.65; N, 6.97.
2,7-Bis[(Z)-m-amidinobenzylidene]-3,3,6,6-tetramethylcy-
cloheptanone (39). was prepared in 59% yield from 38
by the method described above for the synthesis of 31.
Mp 194–196 ꢃC; 1H NMR (methanol-d4) d 7.67–7.21
(8H, m), 6.72 (2H, s), 1.81 (4H, s), 1.30 (12H, s); 13C
NMR (methanol-d4) d 208.7, 161.8, 153.1, 138.4, 134.2,
129.4, 129.0, 128.8, 128.1, 126.5, 38.6, 38.4, 28.6; ES–MS
m/e 215.2 1/2(M+2)2+, 429.3 (M+1)+; FABHRMS m/e
429.2634 (M+H)+, calcd for C27H33N4O 429.2654 Anal.
cmꢀ1. Anal. calcd for C27H26N2O 0.75H2O: C, 79.48; H,
.
6.79; N, 6.87. Found: C, 79.48; H, 6.53; N, 6.79.
2,7-Bis[(Z)-p-amidinobenzylidene)]-3,3,6,6-tetramethylcy-
cloheptanone (34). This was prepared in 65% yield from
28 by the method described above for the synthesis of
31. Mp 205–207 ꢃC; 1H NMR (methanol-d4) d 7.52 (4H,
d, J=8.4 Hz), 7.30 (4H, d, J=8.4 Hz), 6.71 (2H, s), 1.81
(4H, s), 1.29 (12H, s); 13C NMR (methanol-d4) d 212.8,
166.6, 153.7, 143.4, 129.9, 129.3, 127.4, 126.4, 38.7, 38.4,
28.5; ES–MS m/e 215.3 1/2(M+2)2+, 429.2 (M+1)+;
FABHRMS m/e 429.2652 (M+H)+, calcd for
.
.
calcd for C27H32N4O 1.9CF3COOH H2O: C, 55.78; H,
5.46; N, 8.45. Found: C, 56.01; H, 5.66; N, 8.27.
Enzyme assays
C27H33N4O
429.2654.
Anal.
calcd
for
Human FXa was obtained from American Diagnostica
Inc., Greenwich, CT, USA. Human thrombin, bovine
trypsin and human plasmin were from Sigma, St. Louis,
MO, USA. The activities of human FXa, human throm-
bin, bovine trypsin and human plasmin were determined
as the initial rates of the cleavage of peptide p-nitroani-
lide by the enzymes. Initial rates for the control assays
(no inhibitor) were measured at different substrate con-
centrations and used in the Michaelis–Menten equation
to determine Km and nmax39 under the present assay con-
ditions (TableCurve 2D Windows v4.07 from AISN
Software, Inc.). Competitive inhibition was assumed, and
initial rates for the inhibition assays were measured at dif-
ferent inhibitor concentrations and Ki values were calcu-
lated using Dixon analysis.40 Data analysis was performed
with the commercial graphing package Origin 5.0 (Micro-
cal Software, Inc.). The assay was performed at 37ꢃC in
semimicrocuvette. Reaction rates were determined by
measuring the rate of the absorbance change at 405 nM in
.
.
C27H32N4O 2.75CF3COOH 2.5H2O: C, 49.59; H, 5.09;
N, 7.12. Found: C, 49.62; H, 4.92; N, 7.10.
2-[(Z)-m-Cyanobenzylidene]-7-[(Z)-p-cyanobenzylidene]-
3,3,6,6-tetramethylcycloheptanone (35). This was pre-
pared in 34% yield from 27 and 3-dimethoxymethyl-
benzonitrile48 by the method described above for the
synthesis of 26. Mp 125–127 ꢃC; H NMR (CDCl3) d
1
7.56–7.06 (8H, s), 6.45 (1H, s), 6.40 (1H, s), 1.74 (4H, s),
1.27 (6H, s), 1.25 (6H, s); ES–MS m/e 395.3 (M+1)+.
.
Anal. calcd for C27H26N2O 0.3H2O: C, 81.09; H, 6.70;
N, 7.00. Found: C, 81.09; H, 6.43; N, 7.20.
2-[(Z)-m-Amidinobenzylidene]-7-[(Z)-p-amidinobenzyl-
idene]-3,3,6,6-tetramethylcycloheptanone (36). This was
prepared from 35 in 62% yield by the method described
above for the synthesis of 31. Mp 176–179 ꢃC; 1H NMR
(methanol-d4) d 7.88–7.25 (8H, m), 6.73 (1H, s), 6.72

J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
3391
a Beckman 2400 spectrophotometer or a Shimadzu
UV-2401PC UV–vis recording spectrophotometer.
Molecular modeling
We utilized the X-ray protein structure coordinates of
human FXa reported by Padmanabhan et al.24 (PDB
accession code: 1hcg). The standard chymotrypsinogen
residue numbering51 was maintained. Comparative con-
formational energies were calculated by first building the
molecular model in Sybyl (Tripos, Inc.), then optimizing
the geometry using the AM1 module in Gaussian98
(Gaussian, Inc.), and doing final geometry optimization/
energy minimization using the Gaussian 94/98 (3,21)
basis set. Docking calculations utilized the Autodock
program developed by Olson and colleagues25,26 (The
Scripps Research Inst) with 10 docking runs for each
ligand. Grid maps were centered on the active site of FXa
Anti-FXa activity
Anti-FXa activities were measured by using the chro-
mogenic substrate MeO-CO-d-CHG-Gly-Arg-pNA
(American Diagnostica Inc.) and human FXa. DMSO
(20 mL) or a DMSO solution of the inhibitor (20 mL) and
a solution of substrate (100 mL) were mixed with 0.1 M
Tris–0.2 M NaCl buffer pH 8.4 (360 mL). The reaction
was started with the addition of 0.31 unit/mL (or 7.0 nM)
human FXa solution (20 mL). The inhibition assays were
run with 350 mM substrate concentration.
˚
with a grid-point spacing of 0.375 A and 61ꢄ61ꢄ61
Anti-thrombin activity
points in the grid maps. The genetic algorithm docking
method was used with a population size of 50; 2.5 106
energy evaluations; 2.7 104 maximum number of gen-
erations; an elitism value of 1; a mutation rate of 0.02; a
crossover rate of 0.80; a GA window size=10; and cau-
chy distribution parameters, alpha=0 and beta=1.
Anti-thrombin activities were measured by using the
chromogenic substrate N-p-tosyl-Gly-Pro-Arg-pNA
(Sigma) and human thrombin. DMSO (15 mL) or a
DMSO solution of the inhibitor (15 mL) and a solution
of substrate (15 mL) were mixed with 0.14 M NaCl–5
mM KCl–30 mM HEPES–125 mM PEG-8000 pH 7.4
(405 mL). The reaction was started with the addition of
0.1 unit/mL (or 2.7 nM) human thrombin solution (15
mL). The inhibition assays were run with 20.1 mM sub-
strate concentration.
Supporting Information available
Tables of 1H and NOESYdata for 16, 22, 23, 28, 35 and
38; elemental analyses for compounds 2–5, 8, 9, 12, 13,
16, 22, 23, 28, 31–36, 38 and 39; crystallographic data
for 9 and 28. This material is available free of charge via
the Internet at http://www.elsevier.com.
Anti-trypsin activity
Anti-trypsin activities were measured by using chromo-
genic substrate N-p-tosyl-Gly-Pro-Arg-pNA (Sigma)
and bovine trypsin. DMSO (15 mL) or a DMSO solu-
tion of the inhibitor (15 mL) and a solution of substrate
(15 mL) were mixed with 0.14 M NaCl–5 mM KCl–30
mM HEPES–125 mM PEG-8000 pH 7.4 (405 mL). The
reaction was started with the addition of 4.2 nM bovine
trypsin solution (15 mL). The inhibition assays were run
with 40.2 mM substrate concentration.
Acknowledgements
Supported in part by grants from the NIH and Amer-
ican Heart Association Midwest Affiliate. We thank Dr.
Bernard Santarsiero for assistance with the Ortep plots
and small molecule crystallographic data conversion.
References and Notes
Anti-plasmin activity
Anti-plasmin activities were measured by using chro-
mogenic substrate d-Val-Leu-Lys-pNA (Sigma) and
human plasmin. DMSO (15 mL) or a DMSO solution of
the inhibitor (15 mL) and a solution of substrate (15 mL)
were mixed with 0.14 M NaCl–5 mM KCl–30 mM
HEPES–125 mM PEG-8000 pH 7.4 (405 mL). The
reaction was started with the addition of 0.1 unit/mL
human plasmin solution (15 mL). The inhibition assays
were run with both 302 mM and 206 mM substrate
concentration.
1. Rai, R.; Sprengeler, P. A.; Elrod, K. C.; Young, W. B.
Curr. Med. Chem. 2001, 8, 101.
2. Vacca, J. P. Annu. Rep. Med. Chem. 1998, 33, 81.
3. Vacca, J. P. Curr. Opin. Chem. Biol. 2000, 4, 394.
4. Zhu, B. Y.; Scarborough, R. M. Annu. Rep. Med. Chem.
2000, 35, 83.
5. Fevig, J. M.; Wexler, R. R. Annu. Rep. Med. Chem. 1999,
34, 81.
6. Hauptmann, J.; Sturzebecher, J. Thromb. Res. 1999, 93,
203.
7. Ripka, W. C.; Vlasuk, G. P. Annu. Rep. Med. Chem. 1997,
32, 71.
8. Vlasuk, G. P. Thromb. Haemost. 1993, 70, 212.
9. Hauptmann, J.; Kaiser, B. Blood Coagul. Fibrinolysis 1993,
4, 577.
10. Lapatto, R.; Krengel, U.; Schreuder, H. A.; Arkema, A.;
de Boer, B.; Kalk, K. H.; Hol, W. G.; Grootenhuis, P. D.;
Mulders, J. W.; Dijkema, R.; Theunissen, H. J.; Dijkstra,
B. W. EMBO J. 1997, 16, 5151.
11. Stassens, P.; Bergum, P. W.; Gansemans, Y.; Jespers, L.;
Laroche, Y.; Huang, S.; Maki, S.; Messens, J.; Lauwereys, M.;
Cappello, M.; Hotez, P. J.; Lasters, I.; Vlasuk, G. P. Proc.
Natl. Acad. Sci. U.S.A. 1996, 93, 2149.
X-ray crystallographic analysis
Data were processed and the structures were refined on F
using the XTAL 3.749 suite of programs. The structures
were solved using SHELXS-8650 in the Pcan space group.
All non-hydrogen atoms were refined with anisotropic
thermal parameters. Hydrogen atoms were located in
idealized positions based on the carbon backbone and
were given
U=0.070.
a fixed isotropic thermal parameter

3392
J. Cui et al. / Bioorg. Med. Chem. 11 (2003) 3379–3392
12. Fioravanti, C.; Burkholder, D.; Francis, B.; Siegl, P. K.;
Gibson, R. E. Thromb. Res. 1993, 71, 317.
30. Cherkasov, R. A.; Kutyrev, G. A.; Pudovik, A. N. Tetra-
hedron 1985, 41, 2567.
13. Lynch, J. J., Jr.; Sitko, G. R.; Lehman, E. D.; Vlasuk,
G. P. Thromb. Haemost. 1995, 74, 640.
31. Pedersen, B. S.; Lawesson, S. O. Tetrahedron 1979, 35,
2433.
14. Nicolini, F. A.; Lee, P.; Malycky, J. L.; Lefkovits, J.;
Kottke-Marchant, K.; Plow, E. F.; Topol, E. J. Blood
Coagul. Fibrinolysis 1996, 7, 39.
15. Sato, K.; Taniuchi, Y.; Kawasaki, T.; Hirayama, F.;
Koshio, H.; Matsumoto, Y.; Iizumi, Y. Jpn. J. Pharmacol.
1998, 78, 191.
16. Harker, L. A.; Hanson, S. R.; Kelly, A. B. Thromb. Hae-
most. 1995, 74, 464.
17. Kaiser, B. Drugs Future 1998, 23, 423.
18. Sanderson, P. E. J. Annu. Rep. Med. Chem. 2001, 36, 79.
19. Rao, M. S.; Olson, A. J. Proteins 1999, 34, 173.
20. Shaw, K. J.; Guilford, W. J.; Dallas, J. L.; Koovakkaat,
S. K.; McCarrick, M. A.; Liang, A.; Light, D. R.; Morrissey,
M. M. J. Med. Chem. 1998, 41, 3551.
32. Dann, O.; Char, H.; Griessmeier, H. Liebigs Ann. Chem.
1982, 1836.
33. Pinner, A.; Klein, F. Chem. Ber. 1877, 10, 1889.
34. Pinner, A.; Oppenheim: Berlin, 1892.
35. Liberman, A. L.; Vasina, T. V. Zh. Org. Khim. 1967, 3, 690.
36. Borgen, G.; Dale, J. Acta Chem. Scand. 1972, 26, 952.
37. Ito, Y.; Fuji, S.; Nakatsuka, M.; Kawamoto, F.; Saegusa,
T. Org. Synth. 1976, Collective Volume VI, 327–333.
38. Sugimura, T.; Futagawa, T.; Tai, A. Chem. Lett. 1990,
2291.
39. Lineweaver, H.; Burk, D. J. Am. Chem. Soc. 1934, 56, 658.
40. Dixon, M. Biochem. J. 1953, 55, 170.
41. Dowd, P.; Zhang, W.; Geib, S. J. Tetrahedron 1995, 51,
3435.
21. Guilford, W. J.; Shaw, K. J.; Dallas, J. L.; Koovakkat,
S.; Lee, W.; Liang, A.; Light, D. R.; McCarrick, M. A.;
Whitlow, M.; Ye, B.; Morrissey, M. M. J. Med. Chem. 1999,
42, 5415.
42. Batey, R. A.; Harling, J. D.; Motherwell, W. B. Tetra-
hedron 1996, 52, 11421.
43. Kharasch, M. S.; Tawney, P. O. J. Am. Chem. Soc. 1941,
63, 2308.
22. Sturzebecher, J.; Sturzebecher, U.; Vieweg, H.; Wagner,
G.; Hauptmann, J.; Markwardt, F. Thromb. Res. 1989, 54,
245.
23. Renatus, M.; Bode, W.; Huber, R.; Sturzebecher, J.;
Prasa, D.; Fischer, S.; Kohnert, U.; Stubbs, M. T. J. Biol.
Chem. 1997, 272, 21713.
24. Padmanabhan, K.; Padmanabhan, K. P.; Tulinsky, A.;
Park, C. H.; Bode, W.; Huber, R.; Blankenship, D. T.; Cardin,
A. D.; Kisiel, W. J. Mol. Biol. 1993, 232, 947.
25. Goodsell, D. S.; Olson, A. J. Proteins 1990, 8, 195.
26. Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.;
Hart, W. E.; Belew, R. K.; Olson, A. J. J. Comput. Chem.
1998, 19, 1639.
27. Brandstetter, H.; Kuhne, A.; Bode, W.; Huber, R.; von
der, S. W.; Wirthensohn, K.; Engh, R. A. J. Biol. Chem. 1996,
271, 29988.
44. Ouari, O.; Chalier, F.; Bonaly, R.; Pucci, B.; Tordo, P.
J. Chem. Soc., Perkin Trans. 2 1998, 2299.
45. Robson, C.; Meek, M. A.; Grunwaldt, J. D.; Lambert,
P. A.; Queener, S. F.; Schmidt, D.; Griffin, R. J. J. Med.
Chem. 1997, 40, 3040.
46. Harding, K. E.; Ligon, R. C.; Tseng, C. Y.; Wu, T. C.
J. Org. Chem. 1973, 38, 3478.
47. House, H. O.; Kinloch, E. F. J. Org. Chem. 1974, 39, 1173.
48. Sheldon, J. C.; Taylor, M. S.; Bowie, J. H.; Dua, S.; Chia,
C. S. B.; Eichinger, P. C. H. J. Chem. Soc., Perkin Trans. 2
1999, 333.
49. Hall, S. R. d. B.D.J.; Olthof-Hazekamp, R. Xtal3.7 Sys-
tem; University of Western Australia: Crawley, Western Aus-
tralia, 2000.
50. Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Crys-
tallogr. 1990, A46, 467.
28. Lin, Z.; Johnson, M. E. FEBS Lett. 1995, 370, 1.
29. Benner, S. A. Tetrahedron Lett. 1981, 22, 1851.
51. Bode, W.; Turk, D.; Karshikov, A. Protein Sci. 1992, 1,
426.