Neutral inhibitors of the serine protease factor Xa

Article abstract of DOI:10.1016/S0960-894X(01)00312-2

A neutral inhibitor of the serine protease factor Xa was identified via a high-throughput screen of a commercial library. The initial lead 1 demonstrated reversible and competitive inhibition kinetics for factor Xa and possessed a high degree of selectivity versus other related serine proteases. Initial modeling efforts and the generation of a series of analogues of 1 are described.

Full text of DOI:10.1016/S0960-894X(01)00312-2

Bioorganic& Medicinal Chemistry Letters 11 (2001) 1801–1804
Neutral Inhibitors of the Serine Protease Factor Xa
William D. Shrader, Wendy B. Young,* Paul A. Sprengeler, Joan C. Sangalang,
Kyle Elrod and Grant Carry
Departments of Medicinal Chemistry, Structural Chemistry and Enzymology, Axys Pharmaceuticals, Inc., 180 Kimball Way,
South San Francisco, CA 94080, USA
Received 26 September 2000; accepted 16 February 2001
Abstract—A neutral inhibitor of the serine protease factor Xa was identified via a high-throughput screen of a commercial library.
The initial lead 1 demonstrated reversible and competitive inhibition kinetics for factor Xa and possessed a high degree of selectivity
versus other related serine proteases. Initial modeling efforts and the generation of a series of analogues of 1 are described. # 2001
Elsevier Science Ltd. All rights reserved.
Factor Xa, a member of the trypsin-like serine protease
class of enzymes, is a pivotal amplification protease in
the blood coagulation cascade. It is the proteolytic
component of the prothrombinase complex that con-
verts prothrombin to thrombin. Thrombin, in turn,
converts fibrinogen to fibrin which polymerizes to form
insoluble fibrin, a major component of blood clots. In
the coagulation cascade, the prothrombinase complex is
the convergent point of the surface-activated intrinsic
pathway and the tissue factor-activated extrinsic path-
way. Therefore, potent and selective inhibitors of factor
Xa are potentially valuable therapeuticagents for the
treatment of thromboembolicdisorders.
physiological pH produced only a 4-fold increase in
factor Xa inhibition. It appears that the HSA binds the
compound in solution and reduces inhibition.
Compound 1 is comprised of a 2-amino-5-methyl-ben-
zoic acid core with a p-fluoroaniline amide occupying
one terminus and a 3-chloro-benzo[b]thiophene-2-car-
boxamide at the other. The structure of 1 has many
characteristics that differentiate it from the majority of
factor Xa inhibitors in the literature,4À6 although there
have been other reports of anthranilicacids being uti-
lized as a core scaffold.5,6 Principally, 1 is neutral and
does not have the highly basicamidine moiety that
defines the majority of factor Xa inhibitors. An amidine
moiety on typical small molecule drug candidates is
often associated with low oral bioavailability, short
half-life and rapid clearance. For this reason, we found
structure 1 a promising lead. Our initial efforts focused
on identifying its binding mode and probing its SAR.
A high-throughput screen of a commercial library iden-
tified compound 11 as an inhibitor of factor Xa. Upon
resynthesis, purification and characterization of 1, we
verified that the activity as observed in the high-
throughput library screen was indeed accurate and due
to the reported structure of compound 1. Enzymatic
studies showed that this compound operated by a com-
petitive mechanism and had good to excellent selectivity
versus related serine proteases. Interestingly, the factor
Xa potency ranged from 0.042 to 2.1 mM depending
upon the assay conditions employed (see Table 1).2,3
For instance, when the assay was performed in a pH 8.2
buffer the Ki0 was 2.1 mM. Lowering the pH of the buf-
fer to 7.4 caused a 50-fold enhancement in potency
while the addition of HSA (human serum albumin) at
Molecular Modeling
To aid the SAR studies, a binding mode hypothesis was
developed. A Monte Carlo conformational search was
performed on 1 employing both the MM2 and Merck
Molecular force fields in the gas phase and with a water
solvation model.7À11 These results suggested that 1 can
adopt the characteristic L-shape conformation that has
been observed for many factor Xa inhibitors that bind
from the S1 to S4 pockets of serine proteases.12 The
optimized solvent conformation of 1 is consistent with
either terminal ring serving as the P1 element (both the
*Corresponding author. Fax: +1-650-829-1019; e-mail: wendy_young@
axyspharm.com
yCurrent address: Elitra Pharmaceuticals, San Diego, CA, USA.
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00312-2

1802
W. D. Shrader et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1801–1804
p-fluoroaniline ring or the 3-chloro-benzo[b]thiophene
ring can be docked into the S1 pocket). At the time this
work was done two facts led to the 3-chloro-ben-
zo[b]thiophene ring being selected as the P1 element.
Primarily, it was realized that the chlorine could occupy
a similar position to that observed for the iodine atom
of 4-iodobenzo[b]thiophene-2-carboxamidine in u-PA
crystal structures.13,14 Also, Lilly Research Laboratories
reported X-ray studies of a 2,3-substituted benzothio-
phene analogue in which the benzothiophene moiety fits
nicely into the S1 pocket of the closely related protease,
thrombin.15
above Tyr228) were retained since 1 has no polar func-
tionality itself to interact with the enzyme at these sites.
A modified version of the flexible docking program
Hammerhead developed at Axys Pharmaceuticals16 was
employed to first dock the starting conformers rigidly
into the enzyme and then optimize the polar and steric
interactions by allowing the ligand to flex. The resulting
bound structure closely resembles the low energy
conformations of 1 in solution.
In our model (Fig. 1), the sulfur of the benzothiophene
ring is directed toward Ser195 of the catalytic triad (3.17
A), while the remainder of the ring makes van der
Waals contact with the S1 residues and water molecules.
The chlorine atom does indeed adopt a position above
Gly218 similar to the iodine in 4-iodobenzo[b]thio-
phene-2-carboxamidine, a u-PA inhibitor.13,14 Both
amide bonds of the core 2-amino-5-methyl-benzamide
are twisted out of plane to allow the turn necessary for
accessing the S4 pocket of factor Xa. The non-planarity
of the two amide dihedral angles (both ca. À35ꢀ)
between the benzothiophene and core rings results in a
total torsional angle of roughly À72ꢀ. This twist is suf-
ficient to direct the p-fluoroaniline into the S4 pocket.
The oxygen of this proximal amide is directed away
from Gly214 towards solvent. Meanwhile, the NH is
directed towards the Gly214 carbonyl, although the
distance is too long for an H-bond (4.94 A). The tor-
sional angles of the remaining distal amide bond are of
opposite sign (ca. 25–30ꢀ) to those on the proximal
amide with the carbonyl directed towards Trp215 NH
(5.01 A). The resulting geometry allows the distal p-
fluoroaniline ring to bind in the ‘hydrophobicbox’ cre-
ated by Trp215, Tyr99, and Phe235 of factor Xa and
directs the fluorine atom toward the back of the S4
pocket (5.26 A from the Ile175 carbonyl).
Thus, a model of a possible binding mode of 1 to factor
Xa was built using the solvated conformers from the
Monte Carlo search which were docked with the 3-
chloro-benzo[b]thiophene ring in the S1 pocket. The
recently reported high resolution (1.95 A) apo factor Xa
crystal structure was used as the model for the
enzyme.13 Two of the S1 pocket water molecules (one
that bridges between Asp189 and Gly218 C¼O and one
Table 1. Ki0 (mM) of 1 screened against selected serine proteases
Protease
Ki0 (mM)
Factor Xaa
Factor Xab
Factor Xac
Thrombina
Factor VIIaa
Plasmina
Trypsina
uPaa
2.1
0.57
0.042
125
>900
>150
>150
>150
>150
>150
Chemistry
tPaa
aPCa
In order to probe the SAR of 1, we produced an initial
set of closely related analogues. The design of this first
set of derivatives was entirely based upon compounds
that could be generated from commercially available
apH 8.2.
bpH 7.4+HAS.
cpH 7.4.
Figure 1. Stereoview of the docked conformation of 1 (ball-and-stick) into the S1 and S4 sites of factor Xa. The catalytic residues His57 and Ser195
as well as Asp189 and the S1 water molecules retained in the docking studies are shown as polytube representations.

W. D. Shrader et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1801–1804
1803
precursors. These analogues, as illustrated in Table 2,
were all prepared by a straight-forward two-step pro-
cedure that progressed via a nucleophilic ring opening
of intermediate benzoxazines. For example, 1 was
synthesized as shown in Scheme 1.17
Table 2. Distal site analogues
Commercially available 3-chloro-benzo[b]thiophene-2-
carboxylic acid was treated with 2-amino-5-methyl-ben-
zoic acid in the presence of carbonyl diimidazole in
DMF. The resulting 2-benzo[b]thiophen-2-yl-6-methyl-
benzo[d][1,3]oxazi-4-one was then treated with p-fluoro-
aniline to afford 1 in 58% overall yield. Analogues 2–15
were generated in an analogous manner utilizing
the requisite benzothiophenes and anthranilicaidc
precursors.
Compound
R1
R2
Factor Xa Ki0 (mM)
pH 7.4+HSA
1
CH3
H
0.57
3.1
2
Based on the modeling indications that the distal
p-fluoroaniline ring was binding in the S4 pocket, we
initiated our search by replacing this group with a range
of functionality. In short, the replacement of the p-
fluoroaniline with larger and more hydrophilicgroups,
that is, 4-morpholino-4-yl-phenylamine (3), N,N-dime-
thyl-benzene-1,4-diamine (4), 4-aminobenzoicaicd ( 5),
and 5-aminobenzoicaicd ( 6) all decreased factor Xa
affinity dramatically. Additionally, more flexible hydro-
philic groups such as a direct fused morpholine (7) and
piperizine (8) also decreased factor Xa potency. These
findings supported our initial binding hypothesis as the
S4 pocket is lipophilic and would discourage generic
hydrophilic groups which do not make distinct interac-
tions with the protein. As suspected, a slightly larger,
more hydrophobic group, such as the chlorine analogue
9, enhanced factor Xa potency roughly 3-fold. Less
dramaticmodifications to the p-fluoroaniline group, as
in analogues 10–14, were well-tolerated by the enzyme,
although 15 had very poor activity. The most reason-
able explanation for this would be that the ortho chloro
group forces this ring to twist and adopt a less favorable
configuration. Not unexpectedly, removal of the methyl
group on the central ring of 1 (as indicated by 2)
decreased factor Xa affinity since this deletion decreases
the overall lipophilicity of the molecule.
3
CH3
CH3
H
436
4
>1200
>150
>150
>150
>1200
0.21
5
6
H
7
H
8
CH3
CH3
H
9
10
11
12
13
14
15
7.8
CH3
CH3
CH3
CH3
CH3
2.1
In an effort to improve the P1 binding element, two
regioisomerichydroxy analogues of 1 were generated as
depicted in Table 3. Both analogues saw a 5- to 10-fold
decrease in fXa potency. The slight loss in potency
would indicate that although the hydroxy groups did
22.1
97
17.7
Table 3. P1 modifications
>1200
Compd
R1
R2
Factor Xa Ki0 (mM)
pH 7.4+HSA
17
18
OH
H
H
OH
3.5
10
Scheme 1. Reagents: (a) 2-amino-5-methylbenzoicacid, CDl, DMF;
(b) p-fluoroaniline.

1804
W. D. Shrader et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1801–1804
not make productive interactions with the protein, there
remains room in the S1 pocket to accommodate moieties
on the benzothiophene which could in fact improve
binding. A more thorough investigation at this location
may prove quite fruitful.
3. Kuzmic, P.; Sideris, S.; Cregar, L. M.; Elrod, K.; Rice,
K. D.; Janc, J. Anal. Biochem. 2000, 281, 62.
4. Zhu, B. Y.; Scarborough, R. M. Curr. Opin. Cardiol., Pulm.
Renal Invest. Drugs 1999, 1, 63.
5. Fevig, J. M.; Wexler, R. R. Annu. Rep. Med. Chem. 1999,
34, 81.
6. Ewing, W. R.; Pauls, H. W.; Spada, A. P. Drugs Future
1999, 24, 771.
7. Goodman, J. M. S.; Still, W. C. J. Comput. Chem. 1991, 12,
1110.
8. Allinger, N. L. J. Am. Chem. Soc. 1977, 130, 8127.
9. Halgren, T. A. J. Comput. Chem. 1996, 17, 490.
10. Qiu, D.; Shenkin, P. S.; Hollinger, F. P.; Still, W. C. J.
Phys. Chem. A 1997, 101, 3005.
11. Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Lis-
kamp, R.; Lipton, M.; Caufield, C.; Chang, G.; Hendrickson,
T.; Still, W. C. J. Comput. Chem. 1990, 11, 440.
12. Rai, R.; Sprengeler, P. A.; Elrod, K. C.; Young, W. B.
Curr. Med. Chem. 2001, 8, 101.
It was realized that incorporation of an amidine group
on the P1 side could have a positive effect upon binding
through the likely interaction with Asp189. Rationally,
we chose not to make this modification as this change
was likely to alter the binding mode and thereby con-
fuse the SAR picture. It has been our experience that
many benzamidine-based compounds bind in the S1
pocket of factor Xa and an inhibitor will adopt an
appropriate binding mode to accommodate such a
strong salt bridge interaction. Additionally, incorpora-
tion of an amidine would have derailed us from the
original attraction of this neutral lead series.
13. Katz, B. A.; Mackman, R. A.; Luong, C. L.; Radika, K.;
Martelli, A.; Sprengeler, P. A.; Wang, J.; Chan, H.; Wong, L.
Chem. Biol. 2000, 7, 299.
Subsequent to this work, reports on the development of
anthranilic acid analogues as factor Xa inhibitors sur-
faced.18,19 In particular, a report by Eli Lilly indicates
that the alternative binding mode (in which the p-fluoro-
aniline is in the S1 pocket) is also highly likely for 1
and should also be considered for the above described
analogues.
14. Sperl, S.; Jacob, U.; Arroyo de Prada, N.; Sturzebecher,
¨
J.; Wilhelm, O. G.; Bode, W.; Magdolen, V.; Huber, R.;
Moroder, L. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 5113.
15. Sall, D. J.; Bastian, J. A.; Briggs, S. L.; Buben, J. A.;
Chirgadze, N. Y.; Clawson, D. K.; Denney, M. L.; Giera,
D. D.; Gifford-Moore, D. S.; Harper, R. W.; Hauser, K. L.;
Klimkowski, V. J.; Kohn, T. J.; Lin, H.-S.; McCowan, J. R.;
Palkowitz, A. D.; Smith, G. F.; Takeuchi, K.; Thrasher, K. J.;
Tinsley, J. M.; Utterback, B. G.; Yan, S.-C. B.; Zhang, M. J.
Med. Chem. 1997, 40, 3489.
Conclusion
16. Welch, W.; Ruppert, J.; Jain, A. N. Chem. Biol. 1996, 3,
449.
The identification of 1, a neutral, uncharged factor Xa
inhibitor, exemplifies the ability of identifying desirable
lead molecules via high-throughput screening of sui-
tably selected libraries. The SAR of the initial series of
analogues generated is consistent with our initial bind-
ing hypotheses.
17. Synthesis of 1: In a 100 mL round-bottom flask under a
nitrogen atmosphere, 3-chloro-benzo[b]thiophene-2-carboxylic
acid (0.030 g, 1.4 mmol), anhydrous DMF (30 mL) and 1,1-
carbonyldiimidazole (0.23 g, 1.4 mmol) were stirred at room
temperature for 20 min. 2-Amino-5-methyl-benzoicacid (0.18
g, 1.2 mmol) was added and the mixture was stirred for
another 12 h. The precipitate was filtered and dried over P2O5
in a vacuum oven to give 295 mg of an off-white solid. This
solid was placed in a 50 mL round-bottom flask and p-fluoro-
aniline was added (0.67 g, 6 mmol). The mixture was heated to
180 ꢀC for 30 min, the crude residue was taken up in CH2Cl2,
and purified by column chromatogrpahy on SiO2 with 20%
EtOAc/CH2Cl2 as the eluent to give 1 as an off-white solid:
0.31 g, 0.7 mmol, 58%. MS (ESI) calcd for C23H16ClFN2O2S:
438.1, found: MH+ 439.0. 1H NMR (300 MHz, DMSO) d
11.42 (s, 1H), 10.62 (s, 1H), 8.29 (dd, 1 H, J=8.4, 1.2 Hz),
8.16–8.11 (m, 1H), 7.97–7.92 (m, 1H), 7.77–7.72 (m, 3H),
7.64–7.60 (m, 2H), 7.45 (dd, 1H, J=8.5, 1.6 Hz), 7.26–7.17 (m,
2H), 2.40 (s, 3H).
18. Yee, Y. K.; Tebbe, A. L.; Linebarger, J. H.; Beight, D. W.;
Craft, T. J.; Gifford-Moore, D.; Goodson, T., Jr.; Herron,
D. K.; Klimkowski, V. J.; Kyle, J. A.; Sawyer, J. S.; Smith,
G. F.; Tinsley, J. M.; Towner, R. D.; Weir, L.; Wiley, M. R. J.
Med. Chem. 2000, 43, 873.
19. Arnaiz, D. O.; Chou, Y.-L.; Karanjawala, R. E.;
Kochanny, M. J.; Lee, W.; Liang, A. M.; Morrissey, M. M.;
Phillips, G. B.; Sacchi, K. L.; Sakata, S. T.; Shaw, K. J.; Sni-
der, R. M.; Wu, S. C.; Ye, B.; Zhao, Z.; Griedel, B. Int. Patent
WO99/32477, published 07/01/99.
References and Notes
1. Compound 1 was purchased from Maybridge Chemicals.
2. Kineticmeasurements were performed in 96-well U-bottom
microtiter plates (Falcon) using
a SpectraMax spectro-
photometric plate reader (Molecular Devices). Factor Xa (7
nM) was combined with inhibitor at varying concentrations in
50 mM Tris (pH 7.4 or 8.2 as indicated), 110 mM NaCl, 0.6
mM MgCl2, 5 mM CaCl2, 0.001% Antifoam A (Sigma) and
10% DMSO for 1 h at room temperature. Measurements in
human serum albumin (HSA) assay conditions were per-
formed with the addition of 0.45 mM HSA. Reactions were
initiated by the addition of substrate (0.5 mM MeOC-Nle-
Gly-Arg-pNA, Boehringer Mannheim) and the rate of sub-
strate hydrolysis was measured by monitoring the change in
absorbance (A405) over 5 min. Ki apparent (Ki0) calculations
were performed by a nonlinear regression fit to the Morrison
equation as described.