9-Hydroxyazafluorenes and their use in thrombin inhibitors

Article abstract of DOI:10.1021/jm049423s

Optimization of a previously reported thrombin inhibitor, 9-hydroxy-9-fluorenylcarbonyl-L-prolyl-trans-4-aminocyclohexylmethylamide (1), by replacing the aminocyclohexyl P1 group provided a new lead structure, 9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-aminomethyl-5-chlorobenzylamide (2), with improved potency (K_i = 0.49 nM for human thrombin, 2× APTT = 0.37 μM in human plasma) and pharmacokinetic properties (F = 39%, iv T_1/2 = 13 h in dogs). An effective strategy for reducing plasma protein binding of 2 and improving efficacy in an in vivo thrombosis model in rats was to replace the lipophilic fluorenyl group in P3 with an azafluorenyl group. Systematic investigation of all possible azafluorenyl P3 isomers and azafluorenyl-N-oxide analogues of 2 led to the identification of an optimal compound, 3-aza-9-hydroxyfluoren-9(R)-ylcarbonyl-L-prolyl-2-aminomethyl-5- chlorobenzylamide (19b), with high potency (K_i = 0.40 nM, 2× APTT = 0.18 μM), excellent pharmacokinetic properties (F = 55%, T _1/2 = 14 h in dogs), and complete efficacy in the in vivo thrombosis model in rats (inhibition of FeCl₃-induced vessel occlusions in six of six rats receiving an intravenous infusion of 10 μg/kg/min of 19b). The stereochemistry of the azafluorenyl group in 19b was determined by X-ray crystallographic analysis of its N-oxide derivative (23b) bound in the active site of human thrombin.

Full text of DOI:10.1021/jm049423s

2282
J. Med. Chem. 2005, 48, 2282-2293
Articles
9-Hydroxyazafluorenes and Their Use in Thrombin Inhibitors
Kenneth J. Stauffer,*,† Peter D. Williams,† Harold G. Selnick,† Philippe G. Nantermet,† Christina L. Newton,†,∇
Carl F. Homnick,† Matthew M. Zrada,† S. Dale Lewis,‡ Bobby J. Lucas,‡ Julie A. Krueger,‡ Beth L. Pietrak,‡
Elizabeth A. Lyle,§ Rominder Singh,| Cynthia Miller-Stein,| Rebecca B. White,| Bradley Wong,|
Audrey A. Wallace,§ Gary R. Sitko,§ Jacquelyn J. Cook,§ Marie A. Holahan,§ Maria Stranieri-Michener,§
Yvonne M. Leonard,§ Joseph J. Lynch, Jr.,§ Daniel R. McMasters, and Youwei YanX
Departments of Medicinal Chemistry, Biological Chemistry, Pharmacology, Drug Metabolism, and Molecular Systems,
Merck Research Laboratories, West Point, Pennsylvania 19486
Received July 20, 2004
Optimization of a previously reported thrombin inhibitor, 9-hydroxy-9-fluorenylcarbonyl-L-
prolyl-trans-4-aminocyclohexylmethylamide (1), by replacing the aminocyclohexyl P1 group
provided a new lead structure, 9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-aminomethyl-5-chlo-
robenzylamide (2), with improved potency (Ki ) 0.49 nM for human thrombin, 2× APTT )
0.37 µM in human plasma) and pharmacokinetic properties (F ) 39%, iv T1/2 ) 13 h in dogs).
An effective strategy for reducing plasma protein binding of 2 and improving efficacy in an in
vivo thrombosis model in rats was to replace the lipophilic fluorenyl group in P3 with an
azafluorenyl group. Systematic investigation of all possible azafluorenyl P3 isomers and
azafluorenyl-N-oxide analogues of 2 led to the identification of an optimal compound, 3-aza-
9-hydroxyfluoren-9(R)-ylcarbonyl-L-prolyl-2-aminomethyl-5-chlorobenzylamide (19b), with high
potency (Ki ) 0.40 nM, 2× APTT ) 0.18 µM), excellent pharmacokinetic properties (F ) 55%,
T1/2 ) 14 h in dogs), and complete efficacy in the in vivo thrombosis model in rats (inhibition
of FeCl3-induced vessel occlusions in six of six rats receiving an intravenous infusion of 10
µg/kg/min of 19b). The stereochemistry of the azafluorenyl group in 19b was determined by
X-ray crystallographic analysis of its N-oxide derivative (23b) bound in the active site of human
thrombin.
Introduction
(frequent patient monitoring is required because of its
slow onset of action, variable interpatient dose-
response, and significant food and drug interactions).
The serine protease, thrombin, is a critical enzyme
in the coagulation cascade. The primary components of
a vascular thrombus, activated platelets and fibrin, are
both produced by mechanisms involving proteolysis
catalyzed by thrombin. Inhibitors of thrombin have long
been recognized as potential therapeutic agents for the
treatment of a variety of thrombotic disorders, and
indeed, intravenous and oral thrombin inhibitors have
recently progressed through human clinical trials and
reached the marketplace.1,2 Our goal has been to
identify a once a day oral thrombin inhibitor with
predictable pharmacokinetics and no clinically mean-
ingful food or drug interactions. Such a compound would
offer significant advantages over the two most widely
employed antithrombotic drugs, low molecular weight
heparin (the requirement for parenteral administration
precludes its use on a chronic basis) and warfarin
Several years ago we reported the optimization of a
series of thrombin inhibitors containing an L-prolyl-
trans-4-aminocyclohexylmethylamide P2-P1 scaffold,
using parallel solid-phase synthesis methods to vary the
P3 portion of the molecule.3 An interesting inhibitor
which incorporates a 9-hydroxyfluorenyl group in P3
was identified from this work (1, see Figure 1). In an
isolated enzyme assay 1 inhibited human thrombin with
a Ki value of 1.5 nM and 1 had good selectivity for
inhibiting thrombin versus a panel of other human
serine proteases (Ki ) 860 nM trypsin; Ki g 20 µM for
plasmin, tPA, activated protein C, plasma kallikrein,
and chymotrypsin). In functional assays, 1 inhibited
coagulation of human plasma triggered by the intrinsic
pathway with a doubling of the activated partial throm-
boplastin time (2× APTT) at a concentration of 0.85 µM,
and in an arterial thrombosis model in rats, an intra-
venous infusion of 10 µg/kg/min of 1 inhibited FeCl3-
induced vessel occlusion in six of six animals. The
9-hydroxyfluorenyl group also engendered 1 with supe-
rior pharmacokinetic properties compared to numerous
P3 variants in this P2-P1 structural series (F ) 74%,
* To whom correspondence should be addressed. Phone: 215-652-
5264, fax 215-652-3971, e-mail ken_stauffer@merck.com.
† Medicinal Chemistry.
‡ Biological Chemistry.
§ Pharmacology.
| Drug Metabolism.
Molecular Systems.
X Structural Biology.
∇ Deceased June 29, 2001.
10.1021/jm049423s CCC: $30.25 © 2005 American Chemical Society
Published on Web 02/24/2005

9-Hydroxyazafluorenes
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7 2283
exposure and plasma half-life after a 1 mg/kg oral
dose in dogs (AUC ) 13 µM‚h, T1/2 ) 4.2 h) compared
to 2.
Both 2 and 3 showed reduced efficacy for inhibiting
FeCl3-induced thrombosis in rats compared to 1. While
an intravenous infusion of 10 µg/kg/min of 1 was fully
efficacious in six rats (0/6 occlusions), the same dose of
2 and 3 resulted in vessel occlusions in 4/5 and 6/6
animals, respectively. Possibly contributing to the poor
efficacy was the fact that both 2 and 3 were more highly
bound to blood plasma proteins (2, 1% free in human
plasma, 9% free in rat plasma; 3, 3% free in human
plasma, 15% free in rat plasma) than 1 (8% free in
human plasma, 27% free in rat plasma). From our
previous work, we were aware that reducing inhibitor
lipophilicity can lessen plasma protein binding and
improve efficacy in both the 2× APTT assay and the
FeCl3-induced thrombosis assay in rats.6 One way to
accomplish this in the present series was to replace the
fluorenyl ring in P3 with an azafluorenyl ring. It was
thought that this type of modification would reduce
lipophilicity while maintaining, as much as possible, the
structural features important for good intrinsic potency
and good pharmacokinetic properties. Additionally,
since it is well-known that pyridine derivatives are
capable of being metabolized to their corresponding
N-oxides, analogues containing an azafluorenyl-N-oxide
group in P3 were also of interest. Herein we detail the
synthesis and biological properties of analogues of lead
compound 2 which contain all possible isomers of
9-hydroxyazafluorene and the corresponding N-oxides
in the P3 position.
Figure 1.
iv T1/2 ) 2 h in dogs; F ) 39%, iv T1/2 ) 4 h in
cynomologus monkeys).
X-ray crystallographic analysis of 1 bound to the
active site of human thrombin showed one of the
benzene rings of the P3 hydroxyfluorenyl group to
be binding in the hydrophobic S3 pocket of the enzyme,
the other benzene ring contacting the hydrophobic
residues Tyr-60A and Trp-60D contained in the loop
which also contacts the P2 proline, the hydroxyl group
hydrogen bonding to the carbonyl of Gly-216, and the
P1 aminocyclohexyl moiety binding in the S1 pocket of
the enzyme with the amino group interacting with Asp-
189.
Having optimized the P3-P2 portion of the molecule
with the 9-hydroxyfluorenyl-prolyl moiety, we began to
screen new P1 group analogues of 1 for improved
performance. Our laboratories were also developing new
P1 structures based on the previously reported 2,5-
dichlorobenzylamide template,4 replacing the chlorine
at position 2 with a variety of new substituents. An
interesting finding from this work was that an ami-
nomethyl or aminoethyl group at position 2 brought
about a large increase in binding potency by forming a
salt bridge with a surface glutamate on the enzyme as
well as making several other inter- and intramolecular
hydrogen bonds.5 Incorporation of these new P1 groups
into the 9-hydroxyfluorenyl-prolyl P3-P2 scaffold gave
2 and 3 (Figure 1). The aminomethyl analogue 2
displayed superior potency compared to the aminoethyl
analogue 3 and 1 in both the isolated enzyme and the
2× APTT assays. Both 2 and 3 exhibited excellent
pharmacokinetic properties in dogs (2, F ) 39%; iv T1/2
) 13 h; 3, F ) 39%; iv T1/2 ) 14 h) with plasma half-
lives improved by 6-fold compared to 1.
Synthetic Chemistry
The Boc protected prolylchlorobenzylamide derivative
7 served as a key common intermediate for the com-
pounds in Table 1. Compound 7 was prepared in a
straightforward fashion using 2-(tert-butyloxycarbonyl-
aminomethyl)-5-chlorobenzylamine 67 as depicted in
eq 1 of Scheme 1. Acylation of 7 with 9-hydroxyfluorene-
9-carboxylic acid using standard peptide coupling con-
ditions followed by removal of the Boc protecting
group under acidic conditions gave 2 in very good overall
yield.
Synthesis of azafluorenyl P3 analogues of 2 was
envisioned as utilizing azafluorenones as key intermedi-
ates. The keto group of the azafluorenone could in
principle be converted to a cyanohydrin and then
hydrolyzed to a hydroxy acid for coupling to amine 7.
Several methods for preparing azafluorenones have
been reported,8 and the method which proved to be the
best suited for our purposes was that which utilizes an
intramolecular Friedel-Crafts reaction of a pyridine
substrate containing adjacent carboxy and phenyl sub-
stituents. Synthesis of the 4-azafluorene-containing
inhibitors, 20a,b and 24a,b, is shown in eq 2 of Scheme
1. 2-Chloro-3-cyanopyridine underwent efficient Suzuki
coupling with phenylboronic acid followed by acid hy-
drolysis to give the Friedel-Crafts substrate 8. Hot
polyphosphoric acid induced efficient cyclization of 8 to
4-azafluorenone 9. A more direct method of synthesizing
9 was also developed which involves a one-step oxidative
ring contraction of 7,8-benzoquinoline using basic per-
manganate, reaction conditions analogous to what has
The free P1 amino group and the P3 fluorenyl group
in 2 were found to be important structural features for
obtaining favorable pharmacokinetic properties. For
example, replacing the basic P1 amino group with an
acetamido group (4, Figure 1), in addition to reducing
potency, brought about a dramatic reduction of exposure
in the systemic circulation and plasma half-life after an
oral dose of 1 mg/kg in dogs (AUC ) 0.04 µM‚h, T1/2
)
0.48 h) compared to 2 (AUC ) 32 µM‚h, T1/2 ) 13 h).
Likewise, a simple alteration of the P3 group involving
opening of the tricyclic fluorenyl ring (5, Figure 1), while
having little effect on potency, also reduced systemic

2284 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7
Stauffer et al.
Table 1.
plasma protein binding
(% free)
dog pharmacokineticsb
1 mg/kg po
position of
aza or aza-N-oxide
substitution
in vivo rat
thrombosis model
occlusions
thrombin Ki 2× APTT
X
(nM)a
(µM)a
human
rat
AUC (µM‚h)
T1/2 (h)
2
-
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Cl
Cl
0.49
3.3
0.20
3.2
0.30
2.0
0.40
1.4
0.78
140
0.25
37
0.98
4.4
0.77
3.3
2.1
1.1
0.042
0.37
0.50
0.13
0.94
0.12
0.29
0.18
0.23
0.17
-
1
7
9
10
32
34
30
18
20
13
36
-
55
-
70
-
4/5
3/6
0/3
0/5
0/6
3/5
0/6
0/6
0/5
-
0/3
-
0/6
-
32
22
2.8
12
24
2.4
28
14
48
-
0.029
-
0.28
-
13c
6.0
2.5
2.5
10
2.0
14c
6.1
5.5c
-
17a
17b
18a
18b
19a
19b
20a
20b
21a
21b
22a
22b
23a
23b
24a
24b
25a
25b
1N
1N
9
2N
13
27
22
8
12
12
-
35
-
59
-
16
-
56
6
11
2N
3N
3N
4N
4N
1N-O
1N-O
2N-O
2N-O
3N-O
3N-O
4N-O
4N-O
4N
0.085
-
1.3
-
1.8
-
0.13
0.71
0.19
0.43
0.27
0.22
0.13
41
-
42
6
18
1/5
-
0/6
1/6
0/6
1.5
-
2.9
7.0
1.7
1.8
-
5.7c
3.1
3.8
4N
a The reported thrombin Ki values and 2× APTT values are from one or two determinations; typical standard error in these assays is
(20%. b Compounds were dosed as TFA salts in 1% methocel suspension in male beagle dogs (n ) 2); typical standard error in the
measurement of AUC is (20%. c Plasma half-life after a 1 mg/kg iv dose in male beagle dogs (n ) 2).
Scheme 1a
a Reagents and conditions: a. Fmoc-L-proline, EDC, HOBT (94%); b. piperidine, DMF (90%); c. 9-hydroxy-9-fluorenecarboxylic acid,
EDC, HOBT (90%); d. HCl(g), EtOAc (89%); e. PhB(OH)2, Pd(PPh3)4, K2CO3 (94%); f. concentrated HCl, reflux (99%); g. polyphosphoric
acid, 190 °C (92%); h. (CH3)3SiCN, ZnI2, reflux; i. concentrated HCl, HOAc (20%); j. HCl(g), MeOH (70%); k. hydrazine, 50 °C (85%); l.
amyl nitrite, 7, -20 °C (85%); m. TFA, CH2Cl2, chromatographic separation of diasteromers (36%); n. (tBuO2C)2O (91%); o. MCPBA,
CH2Cl2 (44%); p. TFA, CH2Cl2 (95%).
been reported for the preparation of 4,5-diazafluo-
renone.9 4-Azafluorenone 9 was readily converted to
trimethylsilyl cyanohydrin 10 upon heating in trimeth-
ylsilyl cyanide with zinc iodide catalysis.10 Hydrolysis11
of 10 to the corresponding hydroxy acid 11, however,
proved to be problematic due to competing decarboxy-

9-Hydroxyazafluorenes
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7 2285
lation to alcohol 12. Additionally, the coupling of 11 to
proline derivative 7 under a variety of conditions
produced the desired product 16 in very low yields
(<5%) due to competitive decarboxylation of 11 during
the reaction. While the parent, 9-hydroxyfluorene-9-
carboxylic acid, is chemically stable and well-behaved
in peptide coupling reactions, the electron-withdrawing
effect of the nitrogen in the 4-azafluorenyl analogue 11
was found to vastly increase the propensity for decar-
boxylation, making peptide couplings extremely dif-
ficult. At the time this work was ongoing, a report on
the preparation of 9-amino-4,5-di-azafluorenyl-9-car-
boxamide peptide derivatives appeared.12 The coupling
procedure in this report avoided the intermediacy of a
carboxylic acid and instead employed a hydrazide which
was converted to an acyl azide as the reactive acylating
species. A similar approach was adopted in our work
and resulted in greatly improved yields. The treatment
of trimethylsilyl cyanohydrin 10 with HCl gas in
methanol provided hydroxy ester 13, and heating 13
with excess hydrazine produced hydrazide 14. Treat-
ment of 14 with amyl nitrite at low temperature
produced acyl azide 15, the presence of which could be
detected by LC-MS, and 15 slowly acylated 7 at low
temperature to provide the desired coupling product 16
in 85% isolated yield. Byproducts observed in this
reaction were ketone 9 and alcohol 12 which presumably
had arisen from Curtius rearrangement and hydrolysis/
decarboxylation reactions of 15, respectively. Removal
of the Boc group in 16 followed by chromatographic
separation provided the individual 4-azafluorenyl P3
diastereomers 20a and 20b. A portion of each diaste-
reomer was reprotected with di-tert-butyl dicarbonate,
oxidized with MCPBA, and then deprotected with acid
to give the corresponding N-oxide derivatives 24a and
24b.
The remainder of the compounds in Table 1 were
prepared in a manner analogous to that shown in eq 2
of Scheme 1, starting with the different isomers of 8.
The carboxylic acid obtained from methyl 4-phenylpy-
ridine-3-carboxylate,13 was employed as the Friedel-
Crafts substrate for the preparation of compounds in
the 2-azafluorene series (18a,b and 22a,b). Methyl
3-bromopyridine-2-carboxylate and methyl 3-bromo-
pyridine-4-carboxylate, prepared by the method of
Epsztajn et al.,14 underwent efficient Suzuki coupling
with phenylboronic acid, and following ester group
hydrolysis, gave the appropriate Freidel-Crafts sub-
strates for the preparation of compounds in the 1-azaflu-
orene series (17a,b and 21a,b), and 3-azafluorene series
(19a,b and 23a,b), respectively. 7-Chloro-4-azafluo-
renone was prepared by the method of DuPriest et al.15
and was used to obtain 25a and 25b.
Compounds 17a,b-20a,b, and 25a,b in Table 1 were
prepared as mixtures of P3 diastereomers. For each
mixture, chromatographic conditions were identified
which provided the individual diastereomers in pure
form. The diastereomers with the 3-azafluorenyl P3
group, 19a,b, although separable, were very closely
eluting under a variety of chromatographic conditions,
thus making isolation of larger quantities difficult. A
more efficient separation was possible at the Boc-
protected precursor stage. Alternatively, efficient sepa-
ration of enantiomers at the hydroxy ester stage (the
3-azafluorene isomer of 13) was accomplished using
HPLC on a chiral stationary phase, and further pro-
cessing of the (-) enantiomer provided access to larger
quantities of 19b.
Biological Assays
Inhibition constants (Ki values) were determined for
test compounds in an assay using human derived
thrombin, trypsin, urokinase, plasmin, kallikrein, t-PA
and chymotrypsin with the corresponding fluorescent
or chromogenic substrate under steady state conditions
as previously described.17 The concentration of test
compound required to double the activated partial
thromboplastin time in human plasma, the 2× APTT
value, was determined under conditions involving a
3-fold dilution of human plasma as previously re-
ported.18 The antithrombotic activities of intravenously
administered test compounds were determined in an
anesthetized rat model of topical FeCl3-induced arterial
thrombosis.19 Test compounds were administered by
intravenous infusion at a dose of 10 µg/kg/min for 180
min, FeCl3 was applied to the carotid artery at 120 min
post dose, and carotid artery blood flow was monitored
for 60 min. Results are expressed as number of animals
which had occluded vessels (n ) 5 or 6). For plasma
pharmacokinetic assays, amorphous TFA salts of test
compounds were dosed orally in 1% methocel solution
to beagle dogs (n ) 2) at 1 mg/kg. Blood samples were
obtained over 8 h postdose, and a final blood sample
was taken at 24 h. At each time point, test compound
was isolated from the plasma fraction by solid-phase
extraction and quantified by LC-MS/MS against a
standard curve. Area under the curve (AUC) and plasma
half-life (T1/2) were determined from analysis of the
concentration versus time curve. Plasma protein binding
was determined as previously described.6 Standard
error in the measurement of thrombin Ki values, 2×
APTT values, and oral AUC values is (20%.
Results and Discussion
Assay results for the diastereomeric pairs of each of
the four azafluorenyl positional isomers and selected
N-oxide derivatives are collected in Table 1. Comparing
the intrinsic potency (Ki vs human thrombin) of the aza
analogues 17a,b-20a,b to lead compound 2, it is seen
that introduction of nitrogen into the fluorenyl ring
system is generally well-tolerated, with potencies rang-
ing from 6-fold less potent (17a, 18a) to 2.5-fold more
potent (17b) than 2. The difference in potency between
the two diastereomers for each of the four positional
isomers progressively decreased as the nitrogen was
moved around the fluorenyl ring from the 1-position
(17a/17b, difference of 17-fold) to the 4-position (20a/
Yields in the acyl azide coupling reaction with proline
derivative 7 were found to be substrate dependent.
Higher yields were obtained with the 1- and 4-azafluo-
renyl isomers (62% and 85%, respectively), and lower
yields resulted with the 2- and 3-azafluorenyl isomers
(8% and 13%, respectively). Yields with the 2- and
3-azafluorenyl isomers, however, were substantially
improved by adding HOAT to the acyl azide to make
an active ester,16 prior to the addition of proline coupling
partner 7 (30% and 44% yields, respectively, in the
presence of HOAT).

2286 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7
Stauffer et al.
N-oxide oxygen to be located 2.6 Å from the backbone
carbonyl oxygen of Glu-97A which is presumably a
destabilizing electrostatic interaction. However, a crys-
tallographically resolved water molecule, shown as a red
sphere in Figure 2, is within hydrogen bonding distance
(3.1 Å) of the aza-fluorenyl-N-oxide oxygen and may help
to stabilize positioning of the N-oxide group in the S3
pocket. This water molecule is also within hydrogen
bonding distance to the carbonyl oxygen of Trp-96 (2.5
Å) and the phenolic hydroxyl group of Tyr-60A (2.8 Å).
Also, there is an aromatic ring edge-to-face interaction
involving the indole side chain of Trp-60D (edge) and
the benzenoid ring portion of the tricyclic azafluorenyl-
N-oxide (face). These two rings are separated by a
distance of approximately 3.6 Å as indicated in Figure
2. Of the two diastereomers 23a and 23b, we speculate
that the electropositive edge of the Trp-60D side chain
would engage in a more favorable electrostatic interac-
tion with the π-face of the more electron rich benzenoid
ring portion of the tricyclic azafluorenyl-N-oxide ring
system in diastereomer 23b than it would with the
π-face of the more electron deficient N-oxide ring portion
of the azafluorenyl-N-oxide group upon binding of
diastereomer 23a. Crystal structures of other positional
azafluorenyl-N-oxide isomers are needed to more fully
understand the relationship between inhibitor P3 in-
teractions with the enzyme and binding potency in this
series of compounds.
Compared to 2, azafluorenyl analogues 17a,b-20a,b
all showed reduced binding to plasma proteins in both
human and rat plasma. Human plasma protein binding
was particularly high with 2, and as predicted, the free
fraction was significantly increased with the introduc-
tion of nitrogen into the fluorenyl ring. Conversion of
the azafluorenyl analogues to their corresponding N-
oxides increased the free fraction to an even greater
extent. The 4-azafluorenyl analogue 20b and its corre-
sponding N-oxide 24b illustrate this point, with 10-fold
and 50-fold increases in the free fraction in human
plasma, respectively, compared to 2. Reducing human
plasma protein binding had beneficial effects on efficacy
in the 2× APTT assay. For example, although the
intrinsic potencies of 20b, 22b, 23b, 24b, and 25a are
all less than that of 2, the former are all more potent
than 2 in the 2× APTT assay. Azafluorenes 17a,b-
20a,b and N-oxides 21b-24b all showed reduced plasma
protein binding in rat plasma compared to 2. The
greater free fraction in rat plasma resulted in improved
performance in the FeCl3-induced thrombosis assay,
even with several compounds (19a,b, 20a,b, 22b, 23b,
24b) of similar or lesser intrinsic potency than 2.
Figure 2. Active site region from the X-ray structure of the
23b-thrombin complex highlighting several interactions of the
azafluorenyl N-oxide P3 moiety with the enzyme.
20b, difference of 1.8-fold). Comparing the more potent
diastereomers from each pair of positional isomers, 17b-
20b, it is seen that potency progressively increased as
the nitrogen was moved around the fluorenyl ring from
the 4-position (20b, Ki ) 0.78 nM) to the 1-position (17b,
Ki ) 0.20 nM). Conversion of each of the azafluorenyl
diastereomers 17a,b-20a,b to its corresponding N-
oxide derivative, 21a,b-24a,b, resulted in a small loss
of potency (1.3-fold to 3.3-fold) except for azafluorenes
17a and 18a, where the loss in potency was larger (42-
fold and 12-fold loss of potency for N-oxides 21a and
22a, respectively). Addition of a chlorine at the 7-posi-
tion of the 4-azafluorenyl ring produced a very potent
inhibitor with diastereomer 25b. For all the compounds
in Table 1, excellent selectivity for inhibiting thrombin
vs the related serine protease, trypsin, was observed,
with selectivity ratios ranging from 1000-fold to 7000-
fold. Compound 25b displayed the greatest absolute
potency for trypsin in the group, with a Ki value of 300
nM.
The structure of compound 23b bound in the active
site of human thrombin was determined by X-ray
crystallographic analysis (see Figure 2). The positioning
of the P1 aminomethyl chlorobenzyl group in the S1
pocket was found to be virtually identical to that
previously observed for a thrombin inhibitor utilizing a
pyridinone P2 template,5 with the aminomethyl group
in 23b forming a salt bridge with Glu-192 as well as
being within hydrogen-bonding distance of four ad-
ditional H-bond acceptors: the Gly-216 carbonyl oxygen,
the proline carbonyl oxygen, the carbonyl oxygen of the
azafluorenyl-9-carboxamide, and a crystallographically
resolved water molecule. We were somewhat surprised,
however, to find that 23b, the more potent of the two
diastereomerically related 3-azafluorenyl-N-oxides 23a
and 23b, has the R configuration in P3, i.e., the isomer
which has the aza-N-oxide ring pointed in toward the
S3 pocket. Prior to determination of this structure,
visual inspection of the fluorenyl group in the X-ray
structure of lead compound 1 bound in the thrombin
active site led us to speculate that the R-isomers of the
position 1-, 2-, and 3-azafluorenyl-N-oxides might bind
less tightly than the corresponding S-isomers because
of unfavorable steric interactions of the N-oxide group
with the enzyme. The structure of 23b shows the
Initial screening for pharmacokinetic properties was
done in dogs (n ) 2) using a 1 mg/kg oral dose of
amorphous test compound as its trifluoroacetate salt in
methocel suspension. Area under the concentration
versus time curve (AUC) and plasma half-life data are
given in Table 1. In comparing oral AUC values, it is
seen that plasma exposures for all of the azafluorenes
17a,b-20a,b with the exception of 17b and 19a were
similar to that of the parent compound 2. Azafluorene-
N-oxides 21b-24b, however, showed significantly less
plasma exposure after oral dosing compared to 2, and
there was a significant drop (10-fold or greater) in AUC
for each azafluorene-N-oxide compared to its corre-

9-Hydroxyazafluorenes
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7 2287
Table 2. Pharmacokinetic Parametersafor Selected Compounds
in the rat was high, and oral bioavailability was on the
low side. But in dogs and rhesus monkeys, the oral
bioavailability and plasma half-life of 19b were very
favorable. In addition, inhibition of other human serine
proteases by 19b was minimal (Ki trypsin ) 2.5 µM;
factor Xa ) 8.1 µM; chymotrypsin, kallikrein, plasmin,
TPA, urokinase >100 µM). Taking into consideration
all of our key assays (2xAPTT potency, FeCl3-induced
thrombosis assay in rats, oral bioavailability and plasma
half-life in three animal species, stability in human liver
microsomes, human plasma protein binding), 19b
emerged as the best compound in this series of azafluo-
renyl P3 inhibitors.
compd species F (%) CL (mL/min/kg) Vd (L/kg)
T1/2 (h)
5.5
7.7
20b dog
88
2.5
1.2
1.2
monkey 35
1.9
rat
24b dog
15
9
5
1
55
53 ( 5
1.0
6.4
9.4 ( 2
0.39
4.1
4.5 ( 0.2 1.8 ( 0.2
0.52
2.0
2.9 ( 1
0.43
2.2
5.7
6.3
7.0 ( 0.06
14
8.2
2.7 ( 0.8
monkey
rat
19b dog
monkey 57
rat
22
56 ( 9.5
5.8 ( 3
a male beagle dogs (n ) 2) were dosed 1 mg/kg iv and po, male
rhesus macaques (n ) 2) were dosed 1 mg/kg iv and 2 mg/kg po,
male Sprague-Dawley rats (n ) 3) were dosed 2 mg/kg iv and 10
mg/kg po.
Summary and Conclusion
sponding azafluorene analogue. The azafluorene-N-
oxides were found to be significantly more polar than
the azafluorenes as determined by log P measurements
(values of -1.2 to -0.50 for the azafluorene-N-oxides
vs 0.2 to 0.5 for the azafluorenes), and so we speculate
that the lower plasma exposure of the former may be
the result of poor absorption in the gut due to their
highly polar nature. Contrary to this argument, how-
ever, the plasma exposures of 4-aza-7-chlorofluorenes
25a,b were modest despite their polarities being similar
to that of fluorene 2 (log P values of 1.1, 1.3, and 1.5,
respectively). Plasma half-lives for azafluorenes 17a,b-
20a,b and 25a,b fell into three ranges, short (2-4 h,
17b, 18a, 19a, 25a, 25b), intermediate (5-6 h, 17a, 20a,
20b), and long (10-14 h, 18b and 19b). For two of the
N-oxides, 21b and 24b, the plasma half-lives were very
similar to their corresponding azafluorenyl analogues,
and for the other two N-oxides, 22b and 23b, there was
a significant reduction in half-life compared to their
corresponding azafluorenyl analogues. The pharmaco-
kinetic behavior of azafluorene 20b and its N-oxide
derivative 24b was examined in greater detail (see
Table 2). In three species, the N-oxide showed signifi-
cantly reduced (ca. 10-fold) oral bioavailability compared
to the unoxidized azafluorene. In dogs and monkeys, the
oral bioavailability and half-life of 20b were very good.
In rats, however, 20b showed high clearance with a
value approaching that of hepatic blood flow, and poor
oral bioavailability. In vitro metabolism studies with
20b using human liver microsomes showed efficient
conversion to its N-oxide derivative, 24b. The good oral
bioavailability and half-life of 20b in two species and
the good half-life of its N-oxide metabolite 24b in three
species lead one to speculate that 20b dosed in humans
might exhibit a pharmacodynamic half-life longer than
its pharmacokinetic half-life due to the formation of 24b
as a low clearance active metabolite.
Modification of a promising series of previously
described P2 proline-based thrombin inhibitors by in-
corporation of a recently discovered P1 group brought
about significant improvements in potency and phar-
macokinetic properties. The prototype in this new series,
2, however, was found to lack full efficacy for inhibiting
clot formation in an in vivo thrombosis model in rats.
Introduction of nitrogen into the tricyclic fluorenyl P3
group of 2 to produce an azafluorenyl ring system
proved to be an effective solution to reducing plasma
protein binding and restoring full efficacy in the rat
thrombosis model. A systematic study of inhibitors
containing all possible azafluorene positional isomers,
their diastereomers, and N-oxides was undertaken. An
optimal compound was identified, 19b, which contains
a 3-aza-9-hydroxyfluoren-9(R)-yl P3 group. Determina-
tion of the stereochemistry of the azafluorenyl P3 group
in 19b was accomplished by X-ray crystallographic
analysis of the corresponding azafluorenyl-N-oxide de-
rivative 23b bound in the active site of human thrombin.
Compound 19b was found to be a highly potent inhibitor
of thrombin in an isolated enzyme assay (Ki ) 0.4 nM)
with greater than 5000-fold selectivity for human
thrombin vs a panel of other serine proteases. In
functional assays, 19b doubled the activated partial
thromboplastin time in human plasma at a concentra-
tion of 0.18 µM and fully inhibited FeCl3-induced vessel
occlusion in rats with an intravenous infusion of 10 µg/
kg/min. Compound 19b showed excellent pharmacoki-
netic properties in dogs (F ) 55%, iv T1/2 ) 14 h) and
rhesus monkeys (F ) 57%, iv T1/2 ) 8.2 h) and more
modest pharmacokinetic properties in rats (F ) 22%,
iv T1/2 ) 2.7 h). Compound 19b also exhibited good
stability in the presence of human liver microsomal
preparations and maintained a significant free fraction
in human plasma protein binding experiments. Taking
into consideration all of these key preclinical properties,
19b represents the best compound to have emerged from
our thrombin inhibitor discovery effort to date, and as
such offers considerable potential for development as
an oral antithrombotic agent.
Gratifyingly, the two azafluorenes which exhibited
long plasma half-lives in dogs, 18b and 19b, were also
among the most potent thrombin inhibitors in this series
of compounds, and both were fully efficacious for
inhibiting FeCl3-induced thrombosis in rats. These two
compounds, however, were distinguishable from one
another in terms of their stability in human liver
microsomes, with 18b being degraded rapidly and 19b
having good stability. Since this result suggests the
potential for a longer half-life for 19b compared to 18b
in humans, we decided to pursue 19b further and obtain
a more complete set of pharmacokinetic data (see Table
2). As was seen with azafluorene 20b, clearance of 19b
Experimental Section
All reactions were carried out using commercial grade
reagents and solvents. Analytical HPLC data was obtained
using an Agilent Zorbax SB-C8 4.6 mm ID × 75 mm 3.5 µm
column with a 4.5 min linear gradient from 95:5 to 0:100 A:B
(A ) 0.1% TFA in water, B ) 0.1% TFA in acetonitrile), flow
rate ) 3 mL/min, UV detection at 215 nm. Reverse phase

2288 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7
Stauffer et al.
preparative HPLC was performed on a Waters Prep LC 4000
using three Waters C18 PrepPak 40 × 100 mm columns
connected in series, mobile phase was 0.1% TFA in water and
0.1% TFA in acetonitrile with gradients chosen based on the
degree of separation and polarity, flow rate ) 60 mL/min, UV
detection at 215 nm. LC-MS data was obtained on a Waters
2690 analytical HPLC (4 min linear gradient of 92:8 to 0:100
A:B where A ) 0.05% TFA in water and B ) acetonitrile), flow
rate ) 1.5 mL/min with detection using a Micromass ZMD
mass spectrometer (positive ion electrospray ionization). The
1H NMR spectra were recorded on a Varian Unity Inova 400
MHz spectrometer at 400 MHz with chemical shifts (δ)
reported in ppm downfield relative to tetramethylsilane
internal standard. High-resolution mass spectroscopy was
performed on a Bruker Bio-APEX-11 FTICR/MS with an
electrospray ionization mode. The solvent system consisted of
acetonitrile/water (50:50% v/v) with 0.1% formic or acetic acid
delivered by either direct infusion or auto injection using an
HP-1100 sample delivery system. Chemical abbreviations:
DIEA ) N,N-diisopropylethylamine; EDC ) 1-[3-(dimethyl-
amino)propyl]-3-ethylcarbodiimide hydrochloride; HOBT )
1-hydroxy-benzotriazole hydrate; HOAT ) 1-hydroxy-7-aza-
benzotriazole; MCPBA ) meta-chloroperoxybenzoic acid.
L-Prolyl-2-(tert-butyloxycarbonylaminomethyl)-5-chlo-
robenzylamide (7). Step 1. To a stirred solution of 67 (3.80
g, 14.1 mmol, HPLC tR ) 2.63 min), Fmoc-L-Proline (4.98 g,
14.8 mmol, HPLC tR ) 3.26 min), and HOBT (2.15 g, 14.1
mmol) in DMF (30 mL) was added EDC (3.51 g, 18.3 mmol).
The pH of the solution was slowly raised to pH 6 (as measured
on wetted E. Merck pH indicator strips) by the gradual
addition of DIEA (∼2 mL). At 2 h reaction time, HPLC analysis
indicated complete consumption of the benzylamine starting
material. The solvent was removed on a rotary evaporator and
the residue was partitioned between EtOAc (200 mL) and
water (100 mL). The organic phase was dried (MgSO4), filtered,
and the solvent was removed in vacuo. The residue was stirred
in ether (100 mL) for several hours and the solid was collected
by filtration, washed with ether, and dried to give N-fluore-
nylmethoxycarbonyl-L-prolyl-2-(tert-butyloxycarbonyl-amino-
methyl)-5-chlorobenzylamide as a white crystalline solid (7.8
zylamide as a gum (1.4 g, 90% yield). HPLC tR ) 3.69 min;
LC-MS m/z ) 576.
Step 2. A stirred solution of the product from the previous
step (1.3 g, 2.2 mmol) in EtOAc (100 mL) was cooled to 0 °C.
HCl gas was bubbled through the solution for 15 min, and the
mixture was stirred at 0 °C for 1 h. Nitrogen was bubbled
through the solution for 15 min to purge excess HCl, and the
solvent was removed in vacuo. The resulting solid was
suspended in ether (75 mL) and stirred vigorously for 1 h. The
solid was collected by filtration to give the HCl salt of 2 (1.0
g, 89% yield). HPLC tR ) 2.83 min; LC-MS m/z ) 476; 1H NMR,
DMSO-d6: δ 8.69 (br t, 1H), 8.42 (s, 3H), 7.83 (dd, J ) 7, 2
Hz, 2H), 7.2-7.5 (m, 9H), 6.10 (s, 1H), 4.42 (t, J ) 5 Hz, 1H),
4.40 (ABX, J ) 15, 5 Hz, 2H), 4.10 (s, 2H), 2.41 (m, 2H), 1.90
(m, 1H), 1.54 (m, 2H), 1.40 (m, 1H); HRMS (ES/FTMS) C27H26-
ClN3O3 calcd 476.1735 (M + 1), found: 476.1738.
4-Aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide (20a,b). Step 1. To a solu-
tion of 2-chloro-3-cyanopyridine (4.06 g, 29.3 mmol) in toluene
(60 mL) were added phenylboronic acid (5.4 g, 43.9 mmol),
potassium carbonate (6.1 g, 44.1 mmol) and tetrakis(tri-
phenylphosphine)palladium (1.7 g, 1.5 mmol). The mixture was
refluxed under nitrogen for 24 h, and the solvent was removed
in vacuo. The residue was purified on a silica gel column using
100% CH2Cl2 then 99:1 CH2Cl2:MeOH as the mobile phase.
2-Phenyl-3-cyanopyridine was obtained as a solid (5.0 g, 94%
yield). HPLC tR ) 2.98 min; LC-MS m/z ) 181; HRMS (ES/
FTMS) C12H8N2 calcd 181.0760 (M + 1), found: 181.0763.
Step 2. The product from the previous step (5.0 g, 27.9 mmol)
was dissolved in concentrated hydrochloric acid (140 mL) and
heated to reflux for 6 days. The solvent was evaporated in
vacuo to yield 2-phenylpyridine-3-carboxylic acid 8 as a solid
(7.97 g, 100% yield). HPLC tR ) 1.08 min; LC-MS m/z ) 200).
1H NMR, 400 MHz (CD3OD): δ 9.03 (d, J ) 7.97 Hz, 1H),
8.97-8.95 (m, 1H), 8.18-8.14 (m, 1H), 7.68-7.62 (m, 5H);
HRMS (ES/FTMS) C12H9NO2 calcd 200.0706 (M + 1), found:
200.0683.
Step 3. Acid 8 (7.97 g, 40.0 mmol) and polyphosphoric acid
(162 g) were combined and the mixture was heated slowly to
200 °C with mechanical stirring until the acid was fully
consumed (7 h). The mixture was cooled to 140 °C, poured
carefully over crushed ice, and stirred for 5 h. The solution
was extracted several times with CH2Cl2. The combined
extracts were washed with aqueous sodium bicarbonate, dried
over Na2SO4, and filtered, and the solvent was removed in
vacuo to yield 4-aza-9-fluorenone 9 as a yellow solid (4.7 g,
g, 94% yield). TLC Rf ) 0.5 (1:1 EtOAc:hexanes); HPLC tR
3.87 min; LC-MS m/z ) 590.
)
Step 2. To a stirred solution of the product from the previous
step (7.0 g, 12 mmol) in DMF (50 mL) was added piperidine
(7.5 mL). At 15 min reaction time, HPLC analysis indicated
complete consumption of the Fmoc-Proline derivative with
formation of two new closely eluting components (HPLC tR
)
92% yield over two steps). HPLC tR
) 2.91 min; LC-MS m/z )
1
2.76 min desired product, 2.80 min Fmoc derived byproduct).
The solvent and excess piperidine were removed on a rotary
evaporator (bath temp 40°C, ∼0.5 Torr). The residue was
purified on a silica gel column using a gradient elution of 4%,
8%, 12% A in CH2Cl2 (A ) 95:5 MeOH:NH4OH) to give 7 as a
gum (3.9 g, 90% yield). TLC Rf ) 0.4 (90:10:0.5 CH2Cl2:MeOH:
NH4OH; iodine visualization); HPLC tR ) 2.77 min; LC-MS
m/z ) 368; 1H NMR, CDCl3, δ 8.04 (br s, 1H), 7.2-7.3 (m, 3H),
5.24 (br s, 1H), 4.4-4.5 (ABX, 2H), 4.28 (d, 2H), 3.80 (dd, J )
5.4, 9.4 Hz, 1H), 2.9-3.0 (ABXY, 2H), 2.1-2.2 (m, 1H), 1.9-
2.0 (m, 1H), 1.7-1.8 (m, 2H), 1.44 (s, 3H).
182; H NMR, CDCl3: δ 8.62 (dd, J ) 5.1, 1.6 Hz, 1H), 7.90
(dd, 7.3, 1.6 Hz, 1H), 7.86 (d, J ) 7.5 Hz, 1H), 7.74 (d, J ) 7.5
Hz, 1H), 7.61 (dt, J ) 1.1, 7.4 Hz, 1H), 7.44 (t, J ) 7.5 Hz,
1H), 7.22 (dd, J ) 7.3, 2.1 Hz, 1H); HRMS (ES/FTMS) C12H7-
NO calcd 182.0601 (M + 1), found: 182.0588.
Step 4. To a stirred solution of 9 (4.7 g, 25.9 mmol) in CH2-
Cl2 (75 mL) was added zinc iodide (0.81 g, 2.5 mmol) and
trimethylsilylcyanide (31 mL, 233 mmol). The mixture was
stirred at ambient temperature for 2 h and then heated to
reflux for 24 h. The mixture was then blown dry with a stream
of nitrogen gas to yield crude 4-aza-9-trimethylsiloxy-9-cy-
anofluorene (HPLC tR ) 3.71 min; LC-MS m/z ) 281). This
material was dissolved in MeOH (40 mL) and cooled to 0 °C
with stirring. The solution was saturated with HCl gas and
stirred for 4 h with warming to ambient temperature. The
solvent was removed in vacuo, and the residue was partitioned
between CH2Cl2 and aqueous sodium bicarbonate solution. The
CH2Cl2 layer was evaporated in vacuo, and the residue was
purified on silica gel using 95:5 CH2Cl2:MeOH as the mobile
phase. 4-Aza-9-hydroxy-9-fluorene carboxylic acid methyl ester
13 was obtained as a solid (4.4 g, 70% yield). 1H NMR, CDCl3:
δ 8.58 (dd, J ) 5.0, 1.2 Hz, 1H), 7.97 (d, J ) 8.3 Hz, 1H), 7.72
(dd, J ) 7.5, 1.5 Hz, 1H), 7.50 (dt, J ) 7.6, 1.6 Hz, 2H), 7.43
(dt, J ) 7.6, 1.4 Hz, 2H), 7.18 (dd, J ) 7.6, 5.1 Hz, 1H), 4.45
(s, 1H), 3.63 (s, 3H); HPLC tR ) 2.08 min; LC-MS m/z ) 242;
HRMS (ES/FTMS) C14H11NO3 calcd 242.0812 (M + 1), found:
242.0790.
9-Hydroxy-9-fluorenylcarbonyl-L-prolyl-2-aminomethyl-
5-chlorobenzylamide (2). Step 1. To a stirred solution of 7
(1.0 g, 2.7 mmol), 9-hydroxyfluoren-9-yl carboxylic acid (0.66
g, 2.9 mmol), and HOBT (0.44 g, 2.9 mmol) in DMF (60 mL)
was added EDC (0.67 g, 3.5 mmol). DIEA (∼0.5 mL) was added
dropwise until an aliquot of the reaction spotted on wetted E.
Merck pH indicator strips produced a reading of pH 6. The
mixture was stirred at ambient temperature for 24 h and then
concentrated in vacuo. The residue was partitioned between
EtOAc (100 mL) and saturated aqueous NaHCO3 solution (75
mL). The organic phase was dried (MgSO4) and filtered, and
the solvent was removed in vacuo. The residue was purified
on a silica gel column using 1:1 EtOAc:hexanes as eluant.
Product-containing fractions were combined, and the solvent
was removed in vacuo to give 9-hydroxy-9-fluorenylcarbonyl-
L-prolyl-2-(tert-butyloxycarbonylaminomethyl)-5-chloroben-

9-Hydroxyazafluorenes
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7 2289
Step 5. Hydroxy ester 13 (1.9 g, 7.9 mmol) was stirred in
hydrazine (10 mL) and the mixture was warmed to 50 °C for
30 min. The reaction mixture was cooled to ambient temper-
ature, and the excess hydrazine was removed in vacuo. The
residue was suspended in ether (50 mL) and vigorously stirred
for 30 min. The solid was collected by filtration and dried in
vacuo to give hydrazide 14 (1.5 g, 85% yield). 1H NMR, DMSO-
d6: δ 9.51 (br s, 1H), 8.52 (dd, J ) 5.0, 1.3 Hz, 1H), 7.81 (d, J
) 7.0 Hz, 1H), 7.77 (dd, J ) 7.5, 1.3 Hz, 1H), 7.4-7.5 (m, 3H),
7.27 (dd, J ) 7.5, 5.0 Hz, 1H), 6.70 (s, 1H), 4.30 (br s, 2H);
LC-MS m/z ) 242; HRMS (ES/FTMS) C13H11N3O2 calcd
242.0924 (M + 1), found: 242.0928.
Step 6. Hydrazide 14 (1.67 g, 6.9 mmol) was dissolved in
DMF (10 mL) and cooled to -20 °C. THF saturated with HCl
gas was then added until a drop of the reaction mixture applied
to wetted E. Merck pH indicator strip produced a reading of
pH 1-2. Amyl nitrite (0.9 mL, 6.9 mmol) was added, and the
reaction was stirred at -20 °C until all of the hydrazide was
converted to the acyl azide as determined by HPLC and LC-
MS analysis (6 h). DIEA was then added until a drop of the
reaction mixture applied to wetted E. Merck pH indicator strip
produced a reading of pH 6.5. Proline derivative 7 (2.54 g, 6.9
mmol) was then added, and more DIEA was added until a drop
of the reaction mixture applied to wetted E. Merck pH
indicator strip produced a reading of pH 8. The reaction
mixture was stirred at -20 °C until the acyl azide had been
completely consumed as determined by HPLC and LC-MS
analysis (24 h). The mixture was warmed to ambient temper-
ature and the solvent was removed in vacuo. The residue was
purified on a silica gel column using 90:10 CH2Cl2:MeOH as
the mobile phase. Pure fractions were combined and the
solvent was removed in vacuo to give 4-aza-9-hydroxy-9-
fluorenylcarbonyl-L-prolyl-2-(tert-butyloxycarbonylamino-
methyl)-5-chlorobenzylamide 16 as a solid (3.4 g, 85% yield).
HPLC tR ) 3.1 min; LC-MS m/z ) 577.
Step 7. To a stirred solution of 16 (3.4 g, 5.9 mmol) in CH2-
Cl2 (6 mL) was added TFA (2 mL). The mixture was stirred at
ambient temperature for 2 h, and then the solvents were
evaporated in vacuo. The residue was purified by preparative
reverse phase HPLC using a TFA buffered water:acetonitrile
gradient. The resulting mixture of diastereomers was then
separated on a Chiralpak AD column (5 × 50 cm, mobile phase
of 35% A (A ) 0.1% diethylamine in hexane) and 65% EtOH,
60 mL/min flow rate). Using an analytical Chiralpak AD
column, 250 × 4.6 mm, 1.5 mL/min flow rate, the two
diastereomers had retention times of 5.7 min (20a) and 7.9
min (20b). The individual diastereomers were run through a
preparative reverse phase HPLC using a TFA buffered water:
acetonitrile gradient to give, after lyophilization, solid TFA
salts of 4-aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide. 20a (0.56 g, 20% yield) HPLC
tR ) 2.35 min; >99% purity, LC-MS m/z ) 477; 1H NMR,
CDCl3: δ 8.63 (d, J ) 5.04 Hz, 1H), 8.52 (s, 2H), 8.41 (d, J )
5.12 Hz, 1H), 8.02 (d, J ) 6.51 Hz, 1H), 7.70 (d, J ) 7.78 Hz,
1H), 7.54-7.51 (m, 2H), 7.39 (d, J ) 7.33 Hz, 1H), 7.35 (s,
1H), 7.33-7.27 (s,1H), 7.22 (m, 1H),4.91-4.85 (m,1H), 4.45-
4.42 (m, 1H), 4.19-4.09 (m,1H), 4.03-3.99 (d, J ) 14.74 Hz,
1H), 2.44-2.36 (m, 2H), 1.97-1.88 (m, 1H), 1.80-1.70 (m, 1H),
1.68-1.63 (m, 1H), 1.53-1.45 (m,1H); HRMS (ES/FTMS)
C26H25ClN4O3 calcd 477.1688 (M + 1), found: 477.1681. 20b
(0.45 g, 16% yield) HPLC tR ) 2.35 min; >99% purity, LC-MS
m/z ) 477; 1H NMR, CDCl3: δ 8.66-8.65 (d, J ) 5.31 Hz, 1H),
8.52-8.51 (s, 2H), 8.10 (d, J ) 7.60 Hz, 1H), 7.76 (d, J ) 7.60
Hz, 1H), 7.57-7.54 (m,1H), 7.47-7.40 (m,2H), 7.37-7.26 (m,
4H), 4.88-4.83 (m, 1H), 4.47-4.44 (m,1H), 4.23-4.00 (m, 2H),
2.32-2.17 (m, 2H), 2.01-1.92 (m,1H), 1.76-1.63 (m, 2H),
1.61-1.45 (m, 1H); HRMS (ES/FTMS) C26H25ClN4O3 calcd
477.1688 (M + 1), found: 477.1686.
water (40 mL) was added over 3 h. The hot mixture was
filtered. The filtrate was cooled to ambient temperature,
extracted with CH2Cl2, and dried over MgSO4, and the solvent
was removed in vacuo to give 9 (0.2 g, 20% yield).
1-Aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide (17a,b). Step 1. A solution
of 1,4-bis(diphenylphosphino)butane (0.75 g, 1.8 mmol) and bis-
(benzonitrile)dichloropalladium (0.67 g, 1.8 mmol) in CH2Cl2
(30 mL) was stirred at ambient temperature for 20 min, and
then the solvent was removed in vacuo. To the residue were
added methyl 3-bromopyridine-2-carboxylate14 (7.4 g, 34.4
mmol), phenylboronic acid (5.5 g, 45.3 mmol), sodium carbon-
ate (7.4 g, 69.8 mmol), and toluene (150 mL). The mixture was
heated to reflux for 18 h. The mixture was cooled to ambient
temperature, the solids were removed by filtration, and the
filtrate solvent was removed in vacuo. The residue was purified
on silica gel using 1:2 EtOAc:hexanes as the mobile phase.
Pure fractions were combined, and the solvents were removed
in vacuo to yield methyl 2-phenyl-1-pyridinecarboxylic acid (5.3
g, 72% yield). HPLC tR ) 2.9 min; LC-MS m/z ) 214.
Step 2. To the product from the previous step (5.3 g, 24.9
mmol) dissolved in methanol (100 mL) was added NaOH
(1.2 g, 29.9 mmol), and the mixture was stirred at ambient
temperature for 1 h. The mixture was neutralized with 1 N
HCl (30 mL), and the solvents were removed in vacuo to
yield 2-phenylpyridine-1-carboxylic acid (>4.9 g, contains
NaCl, 100% yield). HPLC tR ) 1.85 min; LC-MS m/z ) 200;
HRMS (ES/FTMS) C12H9NO2 calcd 200.0706 (M + 1), found:
200.0718.
Step 3. The product from the previous step (4.9 g theoretical,
24.6 mmol) was converted to 1-aza-9-fluorenone using the
procedure given in Step 3 for 20a,b (4.2 g, 94% yield). HPLC
tR ) 2.83 min; LC-MS m/z ) 182; 1H NMR, CD3OD: δ 8.50 (d,
J ) 6.14 Hz, 1H), 8.13 (d, J ) 8.88 Hz, 1H), 7.74 (d, J ) 7.42
Hz, 1H), 7.64 (d, J ) 7.51 Hz, 1H), 7.62 (m, 1H), 7.52-7.49
(m, 1H), 7.44 (m, 1H); HRMS (ES/FTMS) C12H7NO calcd
182.0601 (M + 1), found: 182.0592.
Step 4. The product from the previous step (4.2 g, 23.2 mmol)
was converted to 1-aza-9-hydroxy-9-fluorene carboxylic acid
methyl ester using the procedure given in Step 4 for 20a,b
except that acetonitrile was used in place of CH2Cl2 as the
solvent in the reaction with trimethylsilylcyanide. The product
was purified on a silica gel column using 98:2 CH2Cl2:MeOH
as the mobile phase (4.6 g, 82% yield). HPLC tR ) 2.44 min;
1
LC-MS m/z ) 242; H NMR, CDCl3: δ 8.48 (d, J ) 6.23 Hz,
1H), 7.93 (d, J ) 9.16 Hz, 1H), 7.69 (d, J ) 7.51 Hz, 1H), 7.55-
7.51 (d, J ) 7.33 Hz, 1H), 7.47 (m, 1 H), 7.45-7.40 (m, 1H),
7.39-7.30 (m,1H), 3.64 (s, 3H).
Step 5. The product from the previous step (4.6 g, 18.9 mmol)
was dissolved in hydrazine (20 mL) and the mixture was
stirred at ambient temperature for 2 h. The mixture was
evaporated in vacuo and the residue was suspended in CH2-
Cl2. The solid was collected by filtration, washed with CH2-
Cl2, and dried in vacuo to give 1-aza-9-hydroxy-9-fluorene
carboxylic acid hydrazide (4.4 g, 96% yield). HPLC tR ) 1.71
min; LC-MS m/z ) 242; 1H NMR, CD3OD: δ 8.37 (d, J ) 6.41
Hz, 1H), 8.13 (d, J ) 9.16 Hz, 1H), 7.79 (d, J ) 7.50 Hz, 1H),
7.53 (d, J ) 7.41 Hz, 1H), 7.47-7.36 (m, 3H); HRMS (ES/
FTMS) C13H11N3O2 calcd 242.0924 (M + 1), found: 242.0891.
Step 6. The compound from the previous step (1.2 g, 5 mmol)
was converted to 1-aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-
2-(tert-butyloxycarbonylaminomethyl)-5-chlorobenzylamide
using the procedure given in Step 6 for 20a,b (1.8 g, 62% yield).
HPLC tR ) 3.29 min; LC-MS m/z ) 577.
Step 7. The product from the previous step (1.8 g, 3.1 mmol)
was deprotected using the same procedure as given in Step 7
for 20a,b. The diastereomers were separated by preparative
reverse phase HPLC using a TFA buffered water:acetonitrile
gradient. Lyophilization gave solid TFA salts of 1-aza-9-
hydroxy-9-fluorenylcarbonyl-L-prolyl-2-aminomethyl-5-chlo-
robenzylamide. First-eluting diastereomer 17a (0.5 g, 32%
yield) HPLC tR ) 2.53 min; >99% purity, LC-MS m/z ) 477;
1H NMR, CDCl3: δ 8.64 (d, J ) 4.67 Hz, 1H), 8.3 (s, 2H), 8.07-
8.11 (m, 2H), 7.69 (m, 1H), 7.51-7.45 (m, 2H), 7.43-7.41 (m,
Alternate Synthesis of 4-Aza-9-fluorenone (9). To a
three-neck 500 mL round-bottom flask equipped with con-
denser and addition funnel were added 7,8-benzoquinoline
(1.02 g, 5.7 mmol), water (75 mL), and potassium hydroxide
(1.1 g, 20.2 mmol). The mixture was heated to reflux, and a
hot solution of potassium permanganate (2.6 g, 16.5 mmol) in

2290 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7
Stauffer et al.
1H), 7.32 (s, 1H), 7.22 (d, J ) 8.15 Hz, 1H), 7.12 (d, J ) 10.07
Hz, 1H), 4.8 (m, 1H), 4.44 (m, 1H), 4.15 (s, 1H), 4.05 (d, J )
12.64 Hz, 1H), 2.64 (m, 2H), 2.04-1.98 (m,1H), 1.75-1.71
(m, 2H), 1.56-1.53 (m, 1H); HRMS (ES/FTMS) C26H25ClN4O3
calcd 477.1688 (M + 1), found: 477.1683. Second-eluting
diastereomer 17b (0.25 g, 17% yield) HPLC tR ) 2.62 min;
>99% purity, LC-MS m/z ) 477; 1H NMR, CDCl3: δ 8.49-
8.41 (m, 3H), 8.27 (s, 1H), 8.07 (d, J ) 8.97 Hz, 1H), 7.71 (d,
J ) 8.24 Hz, 1H), 7.52-7.42 (m, 4H), 7.35 (s, 1H), 7.32 (s, 1H),
4.68-4.62 (m, 1H), 4.53-4.50 (m, 1H), 4.15-4.11 (m, 2H), 2.55
(s, 1H), 2.43 (m, 1H), 2.05-1.95 (m, 1H), 1.84-1.77 (m, 1H),
1.72-1.65 (m, 1H), 1.57-1.50 (m, 1H); HRMS (ES/FTMS)
C26H25ClN4O3 calcd 477.1688 (M + 1), found: 477.1683.
retention time of 4.72 min (18b). Each diastereomer was then
run through a preparative reverse phase HPLC using a TFA
buffered water:acetonitrile gradient to give, after lyophiliza-
tion, solid TFA salts of 2-aza-9-hydroxy-9-fluorenylcarbonyl-
L-prolyl-2-aminomethyl-5-chlorobenzylamide. 18a (40 mg, 18%
yield) HPLC tR ) 2.25 min; >99% purity, LC-MS m/z ) 477;
1H NMR, CD3OD: δ 8.80-8.76 (m, 2H), 8.27 (d, J ) 5.68 Hz,
1H), 8.12-8.09 (m, 1H), 7.67-7.63 (m, 2H), 7.62-7.55 (m, 1H),
7.49 (d, J ) 2.01 Hz, 1H), 7.44-7.38 (m, 2H), 4.56 (d, J ) 15.2
Hz, 1H), 4.46-4.42 (m, 1H), 4.30 (d, J ) 15.39 Hz, 1H), 4.26
(s, 2H), 3.19-3.09 (m, 1H), 2.98-2.95 (m, 1H), 2.10 (m, 1H),
1.82-1.68 (m, 3H). HRMS (ES/FTMS) C26H25ClN4O3 calcd
477.1688 (M + 1), found: 477.1680. 18b (34 mg, 13% yield)
HPLC tR ) 2.28 min; >99% purity, LC-MS m/z ) 477); 1H
NMR, CD3OD: δ 8.76 (d, J ) 5.5 Hz, 1H), 8.64 (s, 1H), 8.16
(d, J ) 5.68 Hz, 1H), 8.08-8.06 (m, 1H), 7.66-7.60 (m, 2H),
7.59-7.57 (m, 1H), 7.50 (d, J ) 1.83 Hz, 1H), 7.45-7.39 (m,
2H), 4.60 (d, J ) 15.2 Hz, 1H), 4.47-4.44 (m, 1H), 4.30-4.26
(m, 3H), 2.72 (s, 2H), 2.08-2.04 (m, 1H), 1.73-1.65 (m, 2H),
1.64-1.56 (m, 1H). HRMS (ES/FTMS) C26H25ClN4O3 calcd
477.1688 (M + 1), found: 477.1658.
2-Aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide (18a,b). Step 1. 4-Phe-
nylpyridine-3-carboxylic acid methyl ester13 (3.8 g, 17.8 mmol)
was hydrolyzed to 4-phenylpyridine-3-carboxylic acid using the
procedure given in Step 2 for 17a,b (4.7 g, theoretical yield )
3.6 g, balance is NaCl). HPLC tR ) 1.80 min; LC-MS m/z )
1
200; H NMR, CD3OD: δ 8.91 (s, 1H), 8.66 (d, J ) 5.22 Hz,
1H), 7.48 (d, J ) 5.22 Hz, 1H), 7.46-7.41 (m, 5H).
Step 2. The compound from the previous step (4.7 g, 23.8
mmol) was converted to 2-aza-9-fluorenone using the procedure
given in Step 3 for 20a,b except that extraction of the product
into CH2Cl2 required prior neutralization of the polyphosphoric
acid with 50% aqueous NaOH to pH 4 (3.1 g, 71% yield). HPLC
tR ) 2.16 min; LC-MS m/z ) 182); 1H NMR, CDCl3: δ 8.87 (s,
1H), 8.75 (d, J ) 4.95 Hz, 1H), 7.75 (d, J ) 7.33 Hz, 1H), 7.65
(d, J ) 7.50 Hz, 1H), 7.62-7.55 (m, 1H), 7.51-7.46 (m, 2H);
HRMS (ES/FTMS) C12H7NO calcd 182.0601 (M + 1), found:
182.0613.
Step 3. The compound from the previous step (3.1 g, 17.2
mmol) was converted to 2-aza-9-hydroxy-9-fluorene carboxylic
acid methyl ester using the procedure given in Step 4 for
20a,b, except that additional amounts of zinc iodide and
trimethylsilyl cyanide and a longer reaction time were required
for complete conversion of the ketone to the trimethylsilyl
cyanohydrin. The product was purified on a silica gel column
using 95:5 CH2Cl2:MeOH as the mobile phase (1.3 g, 31%
yield). HPLC tR ) 2.06 min; LC-MS m/z ) 242); 1H NMR,
CDCl3: δ 8.67-8.65 (m, 2H), 7.75 (d, J ) 6.96 Hz, 1H), 7.57
(d, J ) 5.04 Hz, 1H), 7.51-7.44 (m, 3H), 4.50 (s, 1H), 3.64 (s,
3H); HRMS (ES/FTMS) C14H11NO3 calcd 242.0812 (M + 1),
found: 242.0780.
3-Aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide (19a,b). Step 1. To a solu-
tion of methyl 3-bromopyridine-4-carboxylate14 (5.9 g) in
toluene (140 mL) were added phenylboronic acid (4.3 g, 35.2
mmol), potassium carbonate (4.9 g, 35.3 mmol), and tetrakis-
(triphenylphosphine)palladium (1.4 g, 1.18 mmol). The mixture
was heated to reflux under nitrogen for 4 h, cooled to ambient
temperature, and filtered. The filtrate solvent was removed
in vacuo, and the residue was purified on a silica gel column
using 1:2 EtOAc:hexanes as the mobile phase. Pure fractions
were combined and the solvents were removed in vacuo to yield
3-phenylpyridine-4-carboxylic acid methyl ester (3.6 g, 80%
yield). HPLC tR ) 2.70 min; LC-MS m/z ) 214; 1H NMR,
CDCl3: δ 8.74 (overlapping s and d, 2H); 7.65 (d, J ) 8 Hz,
1H); 7.4-7.5 (m, 3 H), 7.30-7.35 (m, 2H); 3.80 (s, 3H).
Step 2. The product from the previous step (3.6 g, 16.7 mmol)
was converted to 3-phenylpyridine-4-carboxylic acid using the
procedure given in Step 2 for 17a,b (4.5 g, theoretical yield )
3.3 g, balance is NaCl). HPLC tR ) 1.22 min; LC-MS m/z )
200; HRMS (ES/FTMS) C12H9NO2 calcd 200.0706 (M + 1),
found: 200.0695.
Step 3. The product from the previous step (4.5 g, 18.2 mmol
theoretical) was converted to 3-aza-9-fluorenone using the
procedure given in Step 3 for 20a,b except that extraction of
the product into CH2Cl2 required prior neutralization of the
polyphosphoric acid with 50% aqueous NaOH to pH 4 (2.9 g,
95% yield over two steps). HPLC tR ) 2.41 min; LC-MS m/z )
Step 4. The product from the previous step (1.4 g, 5.8 mmol)
was converted to 2-aza-9-hydroxy-9-fluorene carboxylic acid
hydrazide using the same procedure as given in Step 5 for
17a,b (1.4 g, 99% yield). HPLC tR ) 0.8 min; LC-MS m/z )
1
242; H NMR, CD3OD: δ 8.58 (s, 1H), 8.55 (d, 1H), 7.9-7.85
1
182; H NMR, CD3OD: δ 8.94 (s, 1H), 8.65 (d, J ) 4.76 Hz,
(m, 1H), 7.78 (d, 1H), 7.59-7.54 (m, 1H), 7.52-7.47 (m, 2H).
HRMS (ES/FTMS) C13H11N3O2 calcd 242.0924 (M + 1).
Found: 242.0916.
1H), 7.82 (d, J ) 7.33 Hz, 1H), 7.71 (d, J ) 7.32 Hz, 1H), 7.64-
7.62 (m, 1H), 7.56 (d, J ) 4.40 Hz, 1H), 7.46-7.42 (m, 1H);
HRMS (ES/FTMS) C12H7NO calcd 182.0601 (M + 1), found:
182.0592.
Step 5. The product from the previous step (1.4 g, 5.8 mmol)
was converted to 2-aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-
2-(tert-butyloxycarbonylaminomethyl)-5-chlorobenzylamide
using the procedure given in Step 6 for 20a,b (0.27 g, 8% yield).
Step 4. The product from the previous step (2.9 g, 16 mmol)
was converted to 3-aza-9-hydroxy-9-fluorene carboxylic acid
methyl ester using the procedure given in Step 4 for 20a,b
except that neat trimethylsilyl cyanide was used in the
conversion of the ketone to the trimethylsilyl cyanohydrin.
The product was purified on a silica gel column using 1:1
EtOAc:hexanes as the mobile phase (2.2 g, 57% yield). HPLC
1
HPLC tR ) 2.75 min; LC-MS m/z ) 577; H NMR, CDCl3: δ
8.90-8.66 (m, 2H), 7.98-7.90 (m, 2H), 7.68-7.54 (m, 2H), 7.49
(d, J ) 7.15 Hz, 1H), 7.30-7.23 (m, 2H), 7.20 (s, 1H), 5.45 (s,
1H), 4.63-4.54 (m, 2H), 4.31 (s, 2H), 2.46-2.26 (m, 2H), 2.03-
1.87 (m, 2H), 1.81-1.71 (m, 1H), 1.58-1.49 (m, 1H), 1.42 (s,
9H); HRMS (ES/FTMS) C31H33ClN4O5 calcd 577.2212 (M + 1),
found: 577.2185.
1
tR ) 2.00 min; LC-MS m/z ) 242; H NMR, CDCl3: δ 8.93 (s,
1H); 8.56 (d, J ) 5 Hz, 1H); 7.76 (d, J ) 7.5 Hz, 1H); 7.48 (t,
J ) 7 Hz, 2H); 7.38 (t, J ) 7 Hz, 2H); 4.60 (s, 1H), 3.62 (s,
3H); HRMS (ES/FTMS) C14H11NO3 calcd 242.0812 (M + 1),
found: 242.0818.
Step 6. The product from the previous step (0.27 g, 0.47
mmol) was deprotected using the same procedure as given in
Step 7 for 20a,b. The crude product was purified by prepara-
tive reverse phase HPLC using a TFA buffered water:aceto-
nitrile gradient. The diastereomers were then separated on a
Deltapak C18 column (40 × 100 mm, 3 in series) with a mobile
phase gradient of 85% A, 15% B to 60% A, 40% B (A ) 0.1%
NH4OAc in water, B ) acetonitrile) over 1 h, flow rate ) 50
mL/min. On an analytical column using a similar mobile
phase, the first-eluting diastereomer had a retention time of
3.50 min (18a) and the second-eluting diastereomer had a
Step 5. The product from the previous step (2.2 g, 9.1 mmol)
was converted to 3-aza-9-hydroxy-9-fluorene carboxylic acid
hydrazide using the procedure given in Step 5 for 17a,b (2.0
g, 90% yield). HPLC tR ) 0.7 min; LC-MS m/z ) 242; 1H NMR,
DMSO-d6: δ 9.51 (br s, 1H); 9.04 (s, 1H); 8.51 (d, J ) 4 Hz,
1H); 7.89 (d, J ) 7 Hz, 1H); 7.4-7.5 (m, 3H), 7.35 (t, J ) 6 Hz,
1H); 6.78 (s, 1H); HRMS (ES/FTMS) C13H11N3O2 calcd 242.0924
(M + 1), found: 242.0939.

9-Hydroxyazafluorenes
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7 2291
Step 6. The product from the previous step (1.1 g, 4.6 mmol)
was converted to 3-aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-
2-(tert-butyloxycarbonylaminomethyl)-5-chlorobenzylamide
using the procedure given in Step 6 for 20a,b except that the
crude product was first purified on a silica gel column using a
gradient mobile phase of 99:1 to 96:4 CH2Cl2:MeOH, followed
by preparative reverse phase HPLC using a TFA buffered
water:acetonitrile gradient (0.34 g, 13% yield). HPLC tR ) 2.99
min; LC-MS m/z ) 577. The mixture of diastereomers so
obtained could be used directly in the next step as described
below, or they could be separated at this stage on a Chiralcel
OD 5 × 50 cm column (mobile phase 80% A, where A ) 0.1%
diethylamine in hexane, 20% 2-propanol, flow rate ) 80 mL/
min). On an analytical Chiralcel OD column (250 × 4.6 mm,
same mobile phase as above, flow rate ) 2.0 mL/min), the first-
eluting diastereomer had a retention time of 4.89 min and the
second-eluting diastereomer had a retention time of 7.37 min.
Upon deprotection with TFA, the first-eluting diastereomer
produced 19a and the second-eluting diastereomer produced
19b.
Step 7. The product from the previous step (0.34 g, 0.59
mmol) was deprotected using the same procedure as given in
Step 7 for 20a,b. Separation of the closely eluting diastereo-
mers required multiple passes on a preparative reverse phase
HPLC and a TFA buffered water:acetonitrile gradient (99:1
to 50:50 A:B, where A ) 0.1% TFA in H2O, B ) 0.1% TFA in
CH3CN). Lyophilization gave solid TFA salts of 3-aza-9-
hydroxy-9-fluorenylcarbonyl-L-prolyl-2-aminomethyl-5-chlo-
robenzylamide. First-eluting diastereomer 19a (10 mg, 4%
yield) HPLC tR ) 2.20 min; >99% purity, LC-MS m/z ) 477);
1H NMR, CD3OD: δ 9.15 (s, 1H), 8.65 (d, J ) 5.13, 1H), 7.98
(d, J ) 7.6, 1H), 7.74 (d, J ) 5.31, 1H), 7.60-7.56 (m, 1H),
7.51 (s, 3H), 7.46-7.40 (m, 2H), 4.60-4.56 (d, J ) 15.2, 1H),
4.44-4.41 (m, 1H), 4.34-4.31 (d, J ) 15.2, 1H), 4.27 (s, 2H),
2.89-2.82 (m, 2H), 2.09-2.03 (m, 1H), 1.86-1.61 (m, 3H);
HRMS (ES/FTMS) C26H25ClN4O3 calcd 477.1688 (M + 1),
found: 477.1685. Second-eluting diastereomer 19b (6 mg, 2%
yield) HPLC tR ) 2.24 min; >99% purity, LC-MS m/z ) 477;
1H NMR, CD3OD: δ 9.16 (s, 1H), 8.66 (d, J ) 5.22, 1H), 8.00
(d, J ) 7.51, 1H), 7.69 (d, J ) 5.12, 1H), 7.60-7.55 (m, 1H),
7.53-7.47 (m, 3H), 7.45-7.42 (m, 2H), 4.62-4.58 (d, J ) 15.2,
1H), 4.47-4.37 (m, 1H), 4.32 (s, 1H), 4.28 (s, 2H), 2.70 (s, 2H),
2.29-2.02 (m, 1H), 1.90-1.64 (m, 2H), 1.60-1.56 (m, 1H);
HRMS (ES/FTMS) C26H25ClN4O3 calcd 477.1688 (M + 1),
found: 477.1689.
conversion of the active ester to the amide was complete as
determined by HPLC and LC-MS (48 h). The mixture was
warmed to ambient temperature, the solvent was removed in
vacuo, and the residue was purified by preparative reverse-
phase HPLC using a 99:1 to 45:55 A:B mobile phase gradient
over 1 h, A ) 0.1% TFA in water, B ) 0.1% TFA in acetonitrile.
(0.9 g, 44% yield).
Step 4. The product from the previous step (0.9 g, 1.6 mmol)
was deprotected using the same procedure as given in Step 7
for 20a,b. Purification by preparative reverse-phase HPLC
using a gradient of 99:1 to 50:50 A:B (A ) 0.1% TFA in water,
B ) 0.1% TFA in acetonitrile) over 1 h followed by lyophiliza-
tion gave the TFA salt of 19b as a solid (0.6 g, 92% yield).
General Procedure for the Preparation of N-Oxides
21a,b-24a,b. Each of the azafluorene diastereomers 17a,b-
20a,b was converted to its corresponding azafluorene N-oxide
derivative as exemplified with the three-step procedure given
below for the preparation of 24b.
4-Aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide, N-Oxide (24b). Step 1. To
a stirred solution of 20b (0.10 g, 0.21 mmol) in DMF (2 mL)
were added di-tert-butyl-dicarbonate (30 mg, 0.15 mmol) and
DIEA (31 µL, 0.17 mmol). The mixture was stirred at ambient
temperature for 16 h and then evaporated in vacuo. The
residue was partitioned between EtOAc and water. The
organic phase was washed with water and dried (Na2SO4) to
yield 4-aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-(tert-bu-
tyloxycarbonylaminomethyl)-5-chlorobenzylamide (0.11 g, 91%
yield). HPLC tR ) 3.1 min; LC-MS m/z ) 577.
Step 2. The product from the previous step (0.11 g, 0.19
mmol) was dissolved in CH2Cl2 (2 mL) and cooled to 0 °C.
MCPBA (50 mg, 0.3 mmol) was added, and the mixture was
stirred for 16 h while allowing the cooling bath to warm to
ambient temperature. The solvent was evaporated in vacuo,
and the residue was purified by preparative reverse phase
HPLC using a TFA buffered water:acetonitrile gradient to
yield 4-aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-(tert-bu-
tyloxycarbonylaminomethyl)-5-chlorobenzylamide, N-oxide(50
mg, 44% yield). HPLC tR ) 3.1 min; LC-MS m/z ) 593.
Step 3. To a stirred solution of the product from the previous
step (50 mg, 0.087 mmol) in CH2Cl2 (4 mL) was added TFA
acid (2 mL). The mixture was stirred for 1 h and then
evaporated in vacuo. The residue was purified on a preparative
reverse phase HPLC using a TFA buffered water:acetonitrile
gradient. Lyophilization yielded a solid TFA salt of 24b (40
3-Aza-9-hydroxyfluoren-9(R)-ylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide (19b). An alternate, higher
yielding synthesis of 19b was accomplished as follows.
1
mg, 95% yield). HPLC tR ) 2.30 min; LC-MS m/z ) 493; H
NMR, CD3OD: δ 8.64-8.60 (m, 1H), 8.37-8.34 (m, 1H), 7.64-
7.43 (m, 8H), 4.63-4.57 (m, 1H), 4.46-4.42 (m, 1H), 4.32-
4.28 (m, 2H), 3.32 (m, 1H), 2.77-2.73 (m, 2H), 2.10-2.03 (m,
1H), 1.79-1.56 (m,3H); HRMS (ES/FTMS) C26H25ClN4O4 calcd
493.1637 (M + 1), found: 493.1628.
Step 1. 3-Aza-9-hydroxy-9-fluorene carboxylic acid methyl
ester (3.6 g) was prepared as described above. The two
enantiomers were separated on a preparative Chiralpak AD
column (5 × 50 cm) using an 82:18 mixture of A:EtOH (A )
0.1% TFA in hexanes) as the mobile phase. The second-eluting
enantiomer was collected, and the solvents were removed in
vacuo. The residue was partitioned between EtOAc and
saturated aqueous NaHCO3. The aqueous phase was extracted
with EtOAc, and the combined organic extracts were evapo-
rated in vacuo to give 1.0 g of the (-) enantiomer of 3-aza-9-
hydroxy-9-fluorene carboxylic acid methyl ester. On an ana-
lytical column (250 × 4.6 mm Chiralpak AD column, mobile
phase 75:25 A:EtOH, A ) 0.1% TFA in hexanes, 1.5 mL/min
flow rate), the (-) enantiomer had a retention time of 7.6 min,
and the (+) enantiomer had a retention time of 6.3 min.
Step 2. The ester from the previous step (1.0 g, 4.1 mmol)
was converted to 3-aza-9-hydroxy-9-fluorene carboxylic acid
hydrazide (0.81 g, 81% yield) using the procedure given above
for the racemic ester.
Step 3. The product from the previous step (0.73 g, 3.0 mmol)
was converted to 3-aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-
2-(tert-butyloxycarbonylaminomethyl)-5-chlorobenzylamide
using the procedure given in Step 6 for 20a,b except after the
pH was adjusted to 6.5, molecular sieves were added followed
by HOAT (0.4 g, 3.1 mmol). The pH was adjusted to 7 with
DIEA, and the temperature was maintained at -26 ° C until
4-Aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide, N-Oxide (24a). Com-
pound 20a was the starting material for 24a. HPLC tR ) 2.29
min; LC-MS m/z ) 493; 1H NMR, CD3OD: δ 8.60 (d, J ) 8.24
Hz, 1H), 8.33 (d, J ) 6.41 Hz, 1H), 7.62-7.55 (m, 3H), 7.51 (s,
2H), 7.45-7.40 (m, 3H), 4.58-4.54 (d, J ) 15.2 Hz, 1H), 4.43-
4.39 (m, 1H), 4.33-4.30 (d, J ) 15.2 Hz, 1H), 4.26 (s, 2H),
2.99 (s, 1H), 2.86 (s, 1H), 2.07-2.05 (m, 1H), 1.77-1.63 (m,
3H); HRMS (ES/FTMS) C26H25ClN4O4 calcd 493.1637 (M + 1),
found: 493.1629.
1-Aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide, N-Oxide (21a,b). Com-
pound 17a was the starting material for 21a. HPLC tR ) 2.37
min; LC-MS m/z ) 493; 1H NMR, CD3OD: δ 8.02-7.95 (s, 1H),
7.92 (d, J ) 7.33 Hz, 1H), 7.88 (d, J ) 7.51 Hz, 1H), 7.56-
7.51 (m, 5H), 7.49-7.38 (s, 2H), 4.68-4.28 (m, 3H), 4.25-4.23
(m, 2H), 3.33-3.29 (m, 2H), 2.18-2.05 (s, 1H), 1.90 (s, 3H);
HRMS (ES/FTMS) C26H25ClN4O4 calcd 493.1637 (M + 1),
found: 493.1642. Compound 17b was the starting material
for 21b. HPLC tR ) 2.53 min; LC-MS m/z ) 493; 1H NMR,
CD3OD: δ 8.15 (d, J ) 6.50 Hz, 1H), 8.01 (d, J ) 7.69 Hz,
1H), 7.94 (d, J ) 7.60 Hz, 1H), 7.67-7.62 (m, 1H), 7.60 (s,
2H), 7.58-7.39 (m, 4H), 4.58-4.54 (m, 1H), 4.49-4.38 (m, 2H),

2292 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7
Stauffer et al.
4.33-4.24 (m, 2H), 2.53-2.42 (m, 2H), 2.15-2.05 (m, 1H),
1.83-1.78 (m, 1H), 1.76-1.53 (m, 2H); HRMS (ES/FTMS)
C26H25ClN4O4 calcd 493.1637 (M + 1), found: 493.1640.
mg, 19% yield, HPLC tR ) 2.60 min; >99% purity, LC-MS m/z
) 511); 1H NMR, CD3OD: δ 8.59 (d, J ) 6.22 Hz, 1H), 7.95 (d,
J ) 8.06 Hz, 1H), 7.80 (d, J ) 8.97 Hz, 1H), 7.60-7.57 (m,
1H), 7.52-7.46 (m, 2H), 7.44-7.41 (m, 2H), 7.39-7.36 (m, 1H),
4.61 (d, J ) 15.2 Hz, 1H), 4.44-4.41 (m, 1H), 4.30 (s, 1H),
4.27 (s, 2H), 2.71-2.60 (m,1H), 2.59-2.54 (m,1H), 2.11-2.02
(m,1H), 1.72-1.56 (m,3H). HRMS (ES/FTMS) C26H24Cl2N4O3
calcd 511.1298 (M + 1). Found: 511.1293.
2-Aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide, N-Oxide (22a,b). Com-
pound 18a was the starting material for 22a. HPLC tR ) 2.35
min; LC-MS m/z ) 493; 1H NMR, CD3OD: δ 8.38-8.36 (m,
1H), 8.34 (s, 1H), 7.92-7.88 (m, 2H), 7.61-7.55 (m, 1H), 7.54-
7.47 (m, 3H), 7.43-7.36 (m, 2H), 4.61-4.57 (d, J ) 15.02 Hz,
1H), 4.44-4.40 (m, 1H), 4.27-4.17 (m, 3H), 3.31-3.23 (m, 1H),
3.21-3.05 (m, 1H), 2.14-2.10 (m, 1H), 1.84-1.79 (m, 1H),
1.76-1.69 (m, 2H); HRMS (ES/FTMS) C26H25ClN4O4 calcd
493.1637 (M + 1), found: 493.1641. Compound 18b was the
starting material for 22b. HPLC tR ) 2.41 min; LC-MS m/z )
493); 1H NMR, CD3OD: δ 8.39-8.37 (d, J ) 8.14 Hz, 1H), 8.04
(s, 1H), 7.93-7.90 (m, 2H), 7.59-7.54 (m, 2H), 7.52-7.45 (m,
2H), 7.43-7.38 (m, 2H), 4.79-4.74 (m, 1H), 4.57-4.53 (d, J )
15.11 Hz, 1H), 4.50-4.47 (m, 1H), 4.30-4.24 (m, 2H), 2.59-
2.50 (m, 2H), 2.03-1.67 (m, 1H), 1.66-1.53 (m, 2H); HRMS
(ES/FTMS) C26H25ClN4O4 calcd 493.1637 (M + 1), found:
493.1642.
Acknowledgment. The authors would like to thank
Dr. Harri Ramjit, Dr. Art Coddington, and Dr. Charles
Ross for high resolution mass spectroscopy data, Dr.
Sandor Varga and Mrs. Joan Murphy for NMR support,
Ms. Amada Strzelec, Mr. Ken Anderson and Ms. Kristi
Hoffman for log P determinations and Mrs. Bang-Lin
Wan for rat PK determinations. We would also like to
thank Dr. Stephen Brady, Dr. Terry Lyle, Dr. Jules
Shafer, Dr. Joseph Vacca and Dr. Joel Huff for their
guidance, encouragement, and support.
3-Aza-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-ami-
nomethyl-5-chlorobenzylamide, N-Oxide (23a,b). Com-
pound 19a was the starting material for 23a. HPLC tR ) 2.31
min; LC-MS m/z ) 493; HRMS (ES/FTMS) C26H25ClN4O4 calcd
493.1637 (M + 1), found: 493.1653. Compound 19b was the
starting material for 23b. HPLC tR ) 2.35 min; LC-MS m/z )
493; HRMS (ES/FTMS) C26H25ClN4O4 calcd 493.1637 (M + 1),
found: 493.1638.
1-Aza-7-chloro-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-
2-aminomethyl-5-chlorobenzylamide (25a,b). Step 1. 4-Aza-
7-chloro-9-fluorenone15 (4.2 g, 19.6 mmol) was converted to
4-aza-7-chloro-9-hydroxy-9-fluorene carboxylic acid methyl
ester using the procedure as given in Step 4 of 20a,b (4.8 g,
89% yield) HPLC tR ) 2.55 min; LC-MS m/z ) 276.
Step 2. The product from the previous step (4.1 g, 19.2 mmol)
was converted to 4-aza-7-chloro-9-hydroxy-9-fluorene-carbox-
ylic acid hydrazide using the procedure of Step 5 of 17a,b (3.8
g, 72% yield). HPLC tR ) 1.75 min; LC-MS m/z ) 276; HRMS
(ES/FTMS) C13H10ClN3O2 calcd 276.0534 (M + 1), found:
276.0539.
Step 3. The product from the previous step (0.7 g, 2.6 mmol)-
was converted to 4-aza-7-chloro-9-hydroxy-9-fluorenylcarbonyl-
L-prolyl-2-(tert-butyloxycarbonylaminomethyl)-5-chloroben-
zylamide using the procedure given in Step 6 of 20a,b (0.42 g,
27%). HPLC tR ) 3.39 min; LC-MS m/z ) 611.
References
(1) Hauptmann, J. Pharmacokinetics of an Emerging New Class of
Anticoagulant/Antithrombotic Drugs. A Review of Small-Mol-
ecule Thrombin Inhibitors. Eur. J. Clin. Pharm. 2002, 57, 751-
758.
(2) Crowther, M. A.; Weitz, J. I. Ximelagatran: the first oral direct
thrombin inhibitor. Expert Opin. Invest. Drugs 2004, 13, 403-
413.
(3) Brady, S. F.; Stauffer, K. J.; Lumma, W. C.; Smith, G. M.; Ramjit,
H. G.; Lewis, S. D.; Lucas, B. J.; Gardell, S. J.; Lyle, E. A.;
Appleby, S. D.; Cook, J. J.; Holahan, M. A.; Stranieri, M. T.;
Lynch, J. J.; Lin, J. H.; Chen, I. W.; Vastag, K.; Naylor-Olsen,
A. M.; Vacca, J. P. Discovery and Development of L-372, 460, a
Novel Potent Orally Active Small Molecule Inhibitor of Throm-
bin: Co-application of Structure-Based Design and Rapid Mul-
tiple Analogue Synthesis on Solid Support. J. Med. Chem. 1998,
41, 401-406.
(4) Lumma, W. C.; Witherup, K. M.; Tucker, T. J.; Brady, S. F.;
Sisko, J. T.; Naylor-Olsen, A. M.; Lewis, S. D.; Freidinger, R.
M. Design of Novel, Potent, Noncovalent Inhibitors of Thrombin
with Nonbasic P1 Substructures. Rapid Structure-Activity
Studies by Solid-Phase Synthesis. J. Med. Chem. 1998, 41,
3687-3693.
(5) Rittle, K. E.; Barrow, J. C.; Nantermet, P. G.; Newton, C. L.;
Krueger, J. A.; Lewis, S. D.; Lucas, B. J.; Selnick, H. G.
Unexpected Enhancement of Potency With ortho-Aminoalkyl-
benzylamides in the P1 Position. Bioorg. Med. Chem. Lett. 2003
13, 3477-3482.
Step 4. The product from the previous step (0.4 g, 0.7 mmol)
was deprotected using the procedure of Step 7 of 20a,b. The
crude product was purified by preparative reverse phase HPLC
using a TFA buffered water:acetonitrile mobile phase. The
diastereomers were then separated by preparative reverse
phase HPLC using an ammonium bicarbonate buffered mobile
phase (Deltapak C18 column, 40 × 100 mm, 3 columns in
series, flow rate ) 50 mL/min, gradient of 95% A, 5% B to 5%
A, 95% B over 1 h where A ) 0.1% NH4HCO3 in H2O and B )
acetonitrile). Analysis on an analytical Delta Pak C18 column,
150 × 3.5 mm, flow rate ) 1.0 mL/min, using a mobile phase
of 70% A (where A ) 0.1% NH4OAc in water) 30% B (where B
) CH3CN) showed the first peak to elute at 7.52 min. (25a)
and the second at 9.20 min. (25b). The individual diastereo-
mers were then passed though a preparative reverse phase
HPLC using a TFA buffered water:acetonitrile mobile phase
followed by lyophilization to give solid TFA salts of 4-aza-7-
chloro-9-hydroxy-9-fluorenylcarbonyl-L-prolyl-2-aminomethyl-
5-chlorobenzylamide. First-eluting diastereomer 25a (80 mg,
19% yield, HPLC tR ) 2.58 min; >99% purity, LC-MS m/z )
(6) Tucker, T. J.; Lumma, W. C.; Lewis, S. D.; Gardell, S. J.; Lucas,
B. J.; Baskin, E. P.; Woltmann, R.; Lynch, J. J.; Lyle, E. A.;
Appleby, S. D.; Chen, I. W.; Dancheck, K. B.; Vacca, J. P. Potent
Noncovalent Thrombin Inhibitors that Utilize the Unique Amino
Acid D-Dicyclohexylalanine in the P3 Position. Implications on
Oral Bioavailability and Antithrombotic Activity. J. Med. Chem.
1997, 40, 1565-1569.
(7) Nelson, T. D.; LeBlond, C. R.; Frantz, D. E.; Matty, L.; Mitten,
J. V.; Weaver, D. G.; Moore, J. C.; Kim, J.; Boyd, R.; Kim, P.-Y.;
Gbewonyo, K.; Brower, M.; Sturr, M.; McLaughlin, K.; McMas-
ters, D. R.; Kress, M. H.; McNamara, J. M.; Dolling, U. H. The
Stereoselective Synthesis of a Potent Thrombin Inhibitor by a
Novel P2-P3 Lactone Ring Opening J. Org. Chem. 2004, 69,
3620-3627.
(8) Prostakov, N. S.; Soldatenkov, A. T.; Kolyadina, N. M.; Obynoch-
nyi, A. A.; Azafluorenes. Synthesis and Conversions. Russian
Chemical Reviews 1997, 66, 121-138.
(9) Plater, M. J.; Kemp, S.; Lattmann, E.; Heterocyclic Free
Radicals. Part 1. 4,5-Diazafluorene Derivatives of Koelsch’s Free
Radical: An EPR and Metal-Ion Complexation Study. J. Chem.
Soc., Perk. Trans. 1 2000, 971-979.
(10) Evans, D. A.; Truesdale, L. K.; Carroll, G. L. Cyanosilylation of
aldehydes and ketones. A convenient route to cyanohydrin
derivatives. J. Chem. Soc., Chem. Commun. 1973, 55.
(11) Grunewald, G. L.; Brouillette, W. J.; Finney, J. A.; Synthesis of
R-Hydroxyamides Via the Cyanosilylation of Aromatic Ketones.
Tetrahedron Lett. 1980, 21, 1219-1220.
(12) Mazaleyrat, J.-P.; Wakselman, M.; Formaggio, F.; Crisma, M.;
Toniolo, C. Synthesis of terminally protected 9-amino-4,5-
diazafluorene-9-carboxylic acid, the first rigid, transition-metal
receptor, disubstituted glycine. Tetrahedron Lett. 1999, 40,
6245-6248.
1
511); H NMR, CD3OD: δ 8.58 (d, J ) 6.41 Hz, 1H), 7.94 (d,
J ) 8.24 Hz, 1H), 7.83 (m, 1H), 7.56 (m, 1H), 7.50 (s, 2H),
7.42 (m, 2H), 7.38-7.35 (m, 1H), 4.60 (d, J ) 15.01 Hz, 1H),
4.43-4.39 (m, 1H), 4.32 (s, 1H), 4.28 (s, 1H), 4.27 (s, 1H), 2.67-
2.63 (m, 2H), 2.08-1.99 (m, 1H), 1.74-1.61 (m,2H), 1.59-1.54
(m, 1H). HRMS (ES/FTMS) C26H24Cl2N4O3 calcd 511.1298 (M
+ 1). Found: 511.1292. Second-eluting diastereomer 25b (80

9-Hydroxyazafluorenes
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 7 2293
(13) Comins, D. L.; Stroud, E. D.; Herrick, J. J.; Regioselective
Addition of Grignard Reagents to the 1-Phenoxycarbonyl Salts
of Alkyl Nicotinates. Heterocycles 1984, 22, 151.
(14) Epsztajn, J.; Plotka, M. W.; Grabowska, A. Application of
organolithium compounds in organic synthesis. Part 19. Syn-
thetic strategies based on aromatic metalation. A concise re-
giospecific synthesis of 3-halogenated picolinic and isonicotinic
acids. Synth. Commun. 1997, 27, 1075-1086.
(15) DuPriest, M. T.; Schmidt, C. L.; Kuzmich, D.; Williams, S. B.; A
Facile Synthesis of 7-Halo-5H-indeno[1,2-b]pyridines and -py-
ridin-5-ones. J. Org. Chem. 1986, 51, 2021-2023.
(16) Wang, P.; Layfield, R.; Landon, M.; Mayer, R. J.; Ramage, R.;
Transfer active ester condensation: A novel technique for
peptide segment coupling. Tetrahedron Lett. 1998, 39, 8711-
8714.
(17) Lewis, S. D.; Ng, A. S.; Baldwin, J. J.; Fusetani, N.; Naylor, A.
M.; Shafer, J. A. Inhibition of thrombin and other trypsin-like
serine proteases by cyclotheonamide. Thromb. Res. 1993, 70,
173-190.
(18) Lewis, S. D.; Ng, A. S.; Lyle, E. A.; Mellott, M. J.; Appleby, S.
D.; Brady, S. F.; Stauffer, K. J.; Sisko, J. T.; Mao, S. S.; Veber,
D. F.; Nutt, R. F.; Lynch, J. J.; Cook, J. J.; Gardell, S. J.; Shafer,
J. A. Inhibition of thrombin by peptides containing lysyl-alpha-
keto carbonyl derivatives. Thromb. Haemost. 1995, 74, 1107-
1112.
(19) Schumacher, W. A.; Steinbacher, T. E.; Heran, C. L.; Seiler, S.
M.; Michel, I. M.; Ogletree, M. L. Comparison of thrombin active
site and exosite inhibitors and heparain in experimental models
of arterial and venous thrombosis and bleeding. J. Pharmacol.
Exp. Therap. 1993, 267, 1237-1242.
JM049423S