Imidazole acetic acid TAFIa inhibitors: SAR studies centered around the basic P′1 group

Article abstract of DOI:10.1016/j.bmcl.2004.02.033

Structural modifications of the aminopyridine P₁ ^′ group of imidazole acetic acid based TAFIa inhibitors led to the discovery of the aminocyclopentyl analog 28, a 1nM TAFIa inhibitor with CLT₅₀ functional activity of 14nM but without selectivity against CPB. While not as active, aminobutyl derivative 27 provided an improved 6.7-fold selectivity for TAFIa versus CPB.

Full text of DOI:10.1016/j.bmcl.2004.02.033

Bioorganic & Medicinal Chemistry Letters 14 (2004) 2141–2145
Imidazole acetic acid TAFIa inhibitors: SAR studies centered
around the basic P01 group
Philippe G. Nantermet,a,* James C. Barrow,a Stacey R. Lindsley,a MaryBeth Young,a
Shi-Shan Mao,b Steven Carroll,b Carolyn Bailey,b Michele Bosserman,b Dennis Colussi,b
Daniel R. McMasters,c Joseph P. Vaccaa and Harold G. Selnicka
aDepartment of Medicinal Chemistry, Merck Research Laboratories, PO Box 4, West Point, PA 19486, USA
bBiological Chemistry, Merck Research Laboratories, PO Box 4, West Point, PA 19486, USA
cMolecular Systems, Merck Research Laboratories, PO Box 4, West Point, PA 19486, USA
Received 16 December 2003; revised 6 February 2004; accepted 9 February 2004
Abstract—Structural modifications of the aminopyridine P10 group of imidazole acetic acid based TAFIa inhibitors led to the
discovery of the aminocyclopentyl analog 28, a 1 nM TAFIa inhibitor with CLT50 functional activity of 14 nM but without
selectivity against CPB. While not as active, aminobutyl derivative 27 provided an improved 6.7-fold selectivity for TAFIa versus
CPB.
Ó 2004 Elsevier Ltd. All rights reserved.
Thrombin-activatable fibrinolysis inhibitor (TAFI) was
Arg217
recently identified1 as an inhibitor of fibrinolysis.2 TA-
FIa is a zinc-containingcarboxypeptidase that removes
basic C-terminal arginine and lysine residues from
peptides and proteins, such as those present on partially
degraded fibrin clots. The presence of these residues on
fibrin accelerates plasmin generation, and therefore
fibrinolysis, by servingas an anchor for the formation of
the plasminogen/t-PA complex.3 As a result TAFIa
serves as a clot stabilizer and its inhibition should
therefore stimulate endogenous fibrinolysis and thereby
exert an antithrombotic effect. The activated form
(TAFIa), also known as carboxypeptidase U (CPU),
carboxypeptidase R (CPR), and plasma carboxypepti-
dase B, is generated in plasma from its zymogen by
limited proteolysis primarily by the thrombin/throm-
bomodulin complex.4
His159
Zn+2
Arg235
P1'
H2N
Glu162
His288
CO2H
Asp348
N
N
N
R
P1
R = H: 1, TAFIa IC50 = 0.08 µM, CPB IC50 = 0.07 µM
R = (CH2)2-i-Pr: 2, TAFIa IC50 = 0.005 µM, CPB IC50 = 0.004 µM
Figure 1. 2-Aminopyridine imidazole acetic acid TAFI inhibitors.
CPM and CPN,6 and greater than 400-fold selective
versus the digestive neutral carboxypeptidase CPA.7
However, no selectivity versus the digestive basic car-
boxypeptidase CPB7;8 was observed, a feature that,
hopefully, could be improved on despite the high amino
acid sequence homology between CPB and TAFIa.
Imidazole acetic acids 1 and 2 represent recent examples5
of potent TAFIa inhibitors, and the ()) enantiomer of 2
(Fig. 1) was indeed proven to be efficacious in an acute
primate model of thrombosis.5a Importantly, they are
inactive against the critical regulatory carboxypeptidases
Based on SAR and on dockingstudies with a homology
model generated9 from existingX-ray crystallographic
structures of CPA and CPB,10 the probable binding
mode of these inhibitors has been determined, and is
presented in Figures 1 and 2. In this binding mode, the
imidazole moiety of the inhibitor is ligated to the cata-
lytic zinc atom; the carboxylate forms a salt bridge with
Keywords: TAFI; Fibrinolysis; Carboxypeptidase.
* Correspondingauthor. Tel.: +1-215-6520945; fax: +1-215-6523971;
e-mail: philippe_nantermet@merck.com
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.02.033

2142
P. G. Nantermet et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2141–2145
Figure 2 shows residues that differ between human TAFIa
and human CPB colored in purple. Specifically high-
lighted are Leu340, Leu342, and Gly347 that line the P10
specificity pocket of TAFIa and exist as Ile, Pro, and Ser,
respectively, in CPB.12 These minor structural dissimilar-
ities between the two carboxypeptidases might be exploit-
ed to confer TAFIa specificity versus CPB to the
imidazole acetic acid series of inhibitors. Structural
modification of the P01 aminopyridine group was therefore
initiated with the hope to improve both TAFIa inhibitory
potency and selectivity versus CPB. Compounds were
designed around the imidazole acetic acid template, pre-
sentinga variety of basic amino termini in P 01.
Imidazole acetic acid derivatives presented in this letter
were evaluated as racemic mixtures for their inhibitory
potency against TAFIa, CPA, CPB, and the regulatory
carboxypeptidases CPM and CPN.13
Figure 2. Homology model of TAFIa with inhibitor 1 docked in the
active site. Purple residues differ between TAFIa and CPB. Leu340Õs
side chain is shown for clarity.
Table 1 illustrates modifications of the 2-aminopyridine
P01 group displayed by inhibitor 1. Methyl substitution is
only tolerated at the 3-position (3–5). Compound 3,
although almost equipotent to 1, is less selective versus
CPB (0.9-fold and 0.2-fold, respectively). Replacement
of the 3-methyl group by electron-withdrawing substit-
uents such as chloro and fluoro (6, 7) or insertion of an
two arginine residues (Arg217 and Arg235); and the
2-aminopyridine moiety fills the S01 specificity pocket of
the enzyme and forms a salt bridge with aspartic acid 348
at the distal end of the pocket. Alkyl substituents (R) on
the imidazole ringoccupy the S region of TAFIa and
1
provide potency enhancement but no improvement in
selectivity regarding TAFIa versus CPB inhibition.5a;11
Table 1. 2-Aminopyridine SAR
CO2H
P1'
N
N
H
Compounds
P10
R
TAFIa IC50 (lM)a
CPB IC50 (lM)a
Selectivity CPB/TAFIa
1
3
4
5
6
7
H
0.08
0.1
1.5
5.8
5.3
6.7
0.07
0.02
––
0.9
0.2
––
––
––
––
H2N
N
N
3-Me
4-Me
6-Me
3-Cl
3-F
––
––
R
––
H2N
8
67
––
––
––
N
MeHN
N
9
13
––
N
N
N
10
3.3
––
––
11
12
12
15
––
––
––
––
H2N
H2N
N
a IC50 values are the average of at least two determinations, standard error of the mean <10%.

P. G. Nantermet et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2141–2145
2143
additional nitrogen atom (pyrimidine 8) are detrimental
to potency demonstratingthe importance of the
Asp348-aminopyridine salt bridge interaction.14;15 Sub-
stitution on either nitrogen atom of the aminopyridine is
detrimental to activity (9, 10) presumably due to steric
interference with the salt bridge. Isomeric aminopyridine
derivatives 11 and 12 are also nearly inactive suggesting
optimal basic group presentation to Asp348 for the
original aminopyridine 1.
HN
CO2H
n
n = 0,1
N
20: m = 0-2 TAFI IC50 > 20 µM
m
p
p = 1, 2
N
H
CO2H
4
CO2H
1
2
H2N
N
H2N
N
3
N
H
N
H
3-NH2: 21, TAFI IC50 = > 20 µM
4-NH2: 22, TAFI IC50 = 7 µM
23, TAFI IC50 = 0.1 µM
The data presented in Table 1 demonstrate that the
aminopyridine P01 group cannot easily be improved on,
possibly due to the restricted available space in the S10
pocket. Aminoalkyl P01 groups therefore emerged as a
logical choice since they are sterically smaller and could
possibly mimic the lysine residue of the natural sub-
strate. Indeed, the 4-carbon tether of an aminobutyl P10
group (14, Table 2) appeared optimal when compared to
aminopropyl and aminopentyl derivatives 13 and 15. As
previously observed, alkylation of the terminal amino
group was detrimental to activity (16 and 17). Surpris-
ingly, aminobutyl derivative 14 displayed better selec-
tivity for TAFIa versus CPB (6.3-fold) than
aminopyridines analogs and it was hoped0that it could
Figure 3. Cyclic aminoalkyl and aminocyclopentyl inhibitors.
addition of a P1 group to the imidazole moiety (iso-
pentyl or 3,3-dimethylbutyl) resulted in potency
enhancement (Table 3). Functional activity (CLT50) was
also assessed for inhibitors 24–28 usingan in vitro clot
lysis assay.5a CLT50 values represent the inhibitor con-
centration providing50% of the maximal acceleration of
clot lysis in pooled human plasma, triggered by throm-
bin, CaCl2, and t-PA, and detected by turbidity changes.
While each inhibitor in Table 3 displays a CLT50 in the
10–200 nM range, analogs 27 and 28 stand out. Amino-
butyl derivative 27, although functionally 2-fold less
active than aminopyridine analog 2, is more selective for
TAFIa versus CPB (6.7-fold and 0.8-fold, respectively).
Aminocyclopentyl derivative 28 was identified as one of
the most intrinsically and especially functionally active
TAFIa inhibitor (IC50 ¼ 1 nM, CLT50 ¼ 14 nM), unfor-
tunately it lacked any selectivity against CPB.
be further enhanced by substitutingthe P alkyl linker
1
vide supra. While potency against TAFIa could be fur-
ther improved in this endeavor (18 and 19), selectivity
deteriorated as steric bulk increased (5.5-fold and 1.6-
fold, respectively), suggesting that the S01 affinity pocket
could be tighter in TAFIa than in CPB.
An alternate strategy to alter potency and selectivity
0
consisted of introducinga ringin the aminoalkyl P side
1
Imidazole acetic acid derivatives presented in this letter
were prepared accordingto Scheme 1. Substituted ami-
nopyridine alkylatingagents of type 30 were synthesized
from halides 29 via carbonylation, reduction, and bro-
mination. Alkylation of malonate derivative 315a pro-
vides intermediates 32. Hydrolysis/decarboxylation with
6 N HCl provides imidazole acetic acid derivatives 1–7
while selective tosyl removal and imidazole alkylation
leads to analogs 2 and 24–26. Inhibitors 8–12 were
synthesized from the correspondingalkylatingagents,
chain, as shown in Figure 3. Incorporation of piperidine,
pyrrolidine, and azetidine rings produced analogs of
general structure 20, none of which showed significant
activity against TAFIa. Introduction of an exocyclic
amino group appeared more promising as illustrated by
aminocyclohexyl-methyl analog 22. Eventually, the
installation of an aminocyclopentyl-methyl P01 group
proved very fruitful. The (1S,2R) isomer 23 was identi-
fied as a 100 nM inhibitor of TAFIa although with a
modest 1.5-fold selectivity against CPB. All other pos-
sible stereoisomers displayed TAFIa IC50 > 4 lM.
16
prepared usingknown procedures. Aminoalkyl deriv-
atives 13–23, 27, and 28 were derived from the corre-
spondingprotected aminoalkyl P 01 iodides and 31, in the
same manner as described above. The necessary iodides
P01 findings were then applied to the substituted imid-
azole scaffold illustrated by inhibitor 2. As expected,5a
Table 2. Aminoalkyl SAR
CO2H
N
P1'
N
H
Compounds
P10
TAFIa IC50 (lM)a
CPB IC50 (lM)a
Selectivity CPB/TAFIa
13
14
15
16
17
18
19
H2N–(CH2)3
H2N–(CH2)4
6.8
0.59
4.7
––
––
3.7
––
6.3
––
H2N–(CH2)5
MeHN–(CH2)4
26
––
––
––
––
Me2N–(CH2)4
H2N–CH2–CHMe–(CH2)2
H2N–CH2–CMe2–(CH2)2
>50
0.27
0.38
1.5
0.6
5.6
1.6
a IC50 values are the average of at least two determinations, standard error of the mean <10%.

2144
P. G. Nantermet et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2141–2145
Table 3. P1–P01 combinations
CO2H
N
P1'
N
P1
Compounds
P10
P1
TAFIa IC50 (lM)a
CPB IC50 (lM)a
Selectivity CPB/TAFIa CLT50 (lM)b
1
H
0.08
0.005
0.002
0.07
0.004
0.004
0.9
0.8
2
0.970
0.120
0.071
H2N
N
N
2
24
i-Pr
t-Bu
3
H
0.1
0.023
0.012
0.02
0.001
0.001
0.2
2.1
0.170
0.159
H2N
25
26
i-Pr
t-Bu
0.04
0.08
14
H
0.59
3.7
6.3
>20
H2N
H2N
27
23
28
i-Pr
H
0.024
0.1
0.160
0.15
6.7
1.5
1
0.210
0.44
H2N
H2N
t-Bu
0.001
0.001
0.014
a IC50 values are the average of at least two determinations, standard error of the mean <10%.
b CLT50 values represent inhibitor concentration providing 50% acceleration of clot lysis in pooled human plasma, triggered by thrombin, CaCl2, and
t-PA, and detected by turbidity changes.5a Values are the average of at least two determinations, standard error of the mean <10%.
CO2Me
N
MeO2C
NH2
N
NHBoc
N
MeO2C CO2Me
N
N
Ts
31
N
a-d
BocHN
R
R
e
N
Ts
R
Br
Br
32
29
30
CO2H
g, h, f
f
H2N
N
H2N
N
R
N
CO2H
R
N
N
N
H
R1
2, 24-26
1-7
Scheme 1. Reagents and conditions: (a) Boc2O, DMAP, 60–95%; (b) CO, Pd0, Et3N, 50–60%; (c) LAH, 60–85%; (d) Ms2O, LiBr, or CBr4, Ph3P,
50–80%; (e) NaH, 31, 30–90%;6 (f) 6 N HCl, 100 °C, 30–95%; (g) cat. MeONa, MeOH, 70–95%; (h) R1(CH2)2OTf, 60–85%.
were obtained from readily available protected
(phthalimide or Boc) aminoacids or aminoalcohols via
reduction and iodination (I2, Ph3P, imidazole). Substi-
tuted aminobutyl analogs 18 and 19 were prepared
accordingto Scheme 2. Iodide 34 was synthesized from
commercially available ketone 33 via Wittigreaction
followed by reduction and iodination. Hydroxyester
3517 was converted to iodide 36 via Mitsunobu reac-
tion with phthalimide followed by reduction and
iodination.
Optimization of the P01 group of imidazole acetic acid
TAFIa inhibitors led to the discovery of the amino-
cyclopentyl analog 28, a 1 nM TAFIa inhibitor with a
CLT50 functional activity of 14 nM. While not as active
as 28, aminobutyl derivative 27 provided an improved
6.7-fold selectivity for TAFIa versus CPB. The data
presented in this letter indicates that the S10 affinity
pockets in TAFIa and in CPB present only subtle
structural differences and that additional studies will be
required to combine exquisite potency and selectivity.

P. G. Nantermet et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2141–2145
2145
O
O
a-d
18
19
N
N
O
I
O
O
33
34
O
e, c-d
HO
CO2Me
N
I
35
36
O
Scheme 2. Reagents and conditions: (a) Ph3PCHCO2Me, 77%; (b) H2, Pd(OH)2/C, 74%; (c) DIBAL-H, 30–40%; (d) I2, Ph3P, imidazole, 60–65%;
(e) phthalimide, Ph3P, DEAD, 24%.
ꢀ
gradient of 1 kcal/mole A. The positions of the Zn, Zn-
bindingresidues (His159, Glu162, and His288), and
Asp348 were altered to those found in the crystal structure
of procarboxypeptidase B (PDB code 1NSA). These
residues, alongwith the residues flankingthem, were
allowed to relax in the context of the entire protein using
the AMBER force field and the GB/SA solvation model as
References and notes
1. (a) Hendriks, D.; Scharpe, S.; van Sande, M.; Lommaert,
M. P. J. Clin. Chem. Clin. Biochem. 1989, 27, 277–285; (b)
Campbell, W.; Okada, H. Biochem. Biophys. Res. Com-
mun. 1989, 162, 933–939; (c) Eaton, D. L.; Malloy, B. E.;
Tsai, S. P.; Henzel, W.; Drayna, D. J. Biol. Chem. 1991,
266, 21833–21838; (d) Bajzar, L.; Manuel, R.; Nesheim,
M. E. J. Biol. Chem. 1995, 270, 14477–14484.
2. For reviews see: Bouma, B. N.; Marx, P. F.; Mosnier, L.
O.; Meijers, J. C. M. Thromb. Res. 2001, 101, 329–354;
Bajzar, L. Arterioscler. Thromb. Vasc. Biol. 2000, 20,
2511–2518.
3. (a) Hoylaerts, M.; Rijken, D. C.; Lijnen, H. R.; Collen, D.
J. Biol. Chem. 1982, 257, 2912–2919; (b) Horrevoets, A. J.
G.; Pannekoek, H.; Nesheim, M. E. J. Biol. Chem. 1997,
272, 2183–2191.
4. Bajzar, L.; Morser, J.; Nesheim, M. E. J. Biol. Chem.
1996, 271, 16603–16608.
5. (a) Barrow, J. C.; Nantermet, P. G.; Stauffer, S. R.; Ngo,
P. L.; Hensel, M. A.; Mao, S.; Carroll, S.; Cooper, C.;
Colussi, D.; Cook, J. J.; Sitko, G.; Tiller, P.; Wong, B.;
McMasters, D. R.; Vacca, J. P.; Selnick, H. S. J. Med.
Chem. 2003, 46, 5294–5297; Also see: (b) Allerton, C. M.
N.; Blagg, J.; Bunnage, M. E. WO02/14285 A1 for a
related series of TAFIa inhibitors.
6. Mathews, K. P.; Pan, P. M.; Gardner, N. J.; Hugli, T. E.
Ann. Int. Med. 1980, 93, 443–445.
7. Skidgel, R. A. Structure and function of mammalian zinc
carboxypeptidases. In Zinc Metalloproteases Health Dis.;
Hooper, N. M., Ed.; Taylor & Francis: London, 1996; pp
241–283.
€
implemented in BatchMin (Schrodinger, Inc., Portland,
OR) with the Zn–ligand distances constrained to those
found in 1NSA.
10. (a) Aviles, F. X.; Vendrell, J.; Guasch, A.; Coll, M.;
Huber, R. Eur. J. Biochem. 1993, 211, 381–389; (b)
Vendrell, J.; Querol, E.; Aviles, F. X. Biochem. Biophys.
Acta 2000, 1477, 284–298; (c) Christianson, D. W.;
Lipscomb, W. N. Acc. Chem. Res. 1989, 22, 62–
69.
11. A full account of our studies regarding SAR studies in P1
is in preparation. Stauffer, S. R. et al.
12. Whereas Leu340 directs its side chain toward the inside of
the S01 specificity pocket, Pro342Õs side chain is directed
away from the active site, and access to Gly347 is hindered
by Asp348 and Asn234. Leu340/Ile340 is therefore
expected to have a more direct influence on compoundsÕ
selectivity.
13. All compounds presented in this letter were found to be
greater than 100-fold selective versus CPA, CPM, and
CPN.
14. Calculated (Chemdraw) pKaÕs for protonated 2-amino-3-
R-5-methyl-pyridines and 2-amino-5-methyl-pyrimidine
correspondingto compounds 1, 3, 6, 7, 8 are 7.0, 7.3,
4.3, 4.1, and 4.3, respectively.
15. Direct replacement of the amino group with hydrogen or
chloro on inhibitor 1 results in a 140-and 300-fold loss in
potency, respectively.
8. The significance of CPB inhibition in vivo remains to be
determined, as CPB is thought to be present only in the
gastrointestinal tract.
9. The homology model of TAFIa was created from the
crystal structure of CPA (PDB code 2CTC) usingMOE
software (Chemical ComputingGroup, Inc., Montreal).
Ten intermediate models were created, and the best-
scoringmodel was minimized to a root-mean-squared
16. (a) For 8 see: Linschoten, M. et al. WO00/66557; (b)
Compound 9 was derived from methylation of the NHBoc
intermediate; (c) For 10 see: Sanderson, P. E.; Naylor-
Olsen, A. M. US6093717; (d) For 11 and 12 see: Askew, B.
C. et al. US5852045.
17. See: Belley, M. L. et al. US5428033.