Novel bicyclic lactam inhibitors of thrombin: Potency and selectivity optimization through P1 residues

Article abstract of DOI:10.1016/S0960-894X(01)00661-8

Peptidomimetic inhibitors of thrombin lacking the important Ser195-carbonyl interaction have been prepared. The binding energy lost after the removal of the activated carbonyl was recaptured through a series of modifications of the P1 residues of the bicyclic lactam inhibitors. Selected substituted compounds displayed useful pharmacological profiles both in vitro and in vivo.

Full text of DOI:10.1016/S0960-894X(01)00661-8

Bioorganic & Medicinal ChemistryLetters 11 (2001) 3161–3164
Novel Bicyclic Lactam Inhibitors of Thrombin: Potency and
Selectivity Optimization through P1 Residues
Sophie Levesque,a,* Yves St-Denis,a Benoit Bachand,a Patrice Preville,a
b
Lorraine Leblond,a Peter D. Winocour,a JeremyJ. Edmunds, J. R. Rubinb
and M. Arshad Siddiquia
a
´
Shire BioChem Inc., 275 Armand-Frappier Blvd., Laval, Quebec, Canada H7V4A7
bPfizer Global Research and Development, Ann Arbor, MI 48105, USA
Received 26 June 2001; accepted 27 September 2001
Abstract—Peptidomimetic inhibitors of thrombin lacking the important Ser195–carbonyl interaction have been prepared. The
binding energylost after the removal of the activated carbonyl was recaptured through a series of modifications of the P1 residues
of the bicyclic lactam inhibitors. Selected substituted compounds displayed useful pharmacological profiles both in vitro and in
vivo. # 2001 Elsevier Science Ltd. All rights reserved.
We have recentlypublished the preparation and biolo-
gical activityof 11 and 2,2 which are examples of bicyclic
lactam inhibitors of thrombin with high potencyand in
vivo activity. Most of the inhibitors prepared in the
earlystudies leading to 1 and 2 were designed to capture
the interactions of the catalytic triad of thrombin with
the inhibitors. However, due to the reactive nature of
the carbonyl function of the P1 residue, an improve-
ment in inhibitor design was envisaged byexcising the
ketothiazole functions of inhibitors 1 and 2.3 As a result
of this removal, we expected a dramatic reduction in
thrombin binding affinities.
The synthesis of the bicyclic template with numerous P3
hydrophobes has been already described.1 The prepara-
tions of the different P1 moieties of the inhibitors are
exemplified in Schemes 1–4. The preparation of the suit-
ablyprotected
trans-4-aminomethyl-cyclohexylamine
(entry 28, Table 1) has been reported elsewhere.4
A common intermediate was used for the preparation of
the different 4-aminomethyl-piperidines based inhibitors
7 and 8, as illustrated in Scheme 1. Thus, the amino
group of isonipecotic acid 3 was first protected as its t-
butyl carbamate. The acid group was then reduced
using BH3THF, to give the protected aminoalcohol 4.
Mesylation of 4, followed bydisplacement with NaN
3
in anhydrous DMF afforded an azide, which was sub-
mitted to hydrogenolysis to give amine 5. This amine
was coupled with the suitable bicyclic template using
conventional peptide coupling reagents (HATU, colli-
dine, and DMF) and the BOC protecting group was
removed bytreatment with 4 N HCl in dioxane. Piper-
idine salts 6 were converted to guanidines 7 using the
appropriate guanylating agent, followed by deprotec-
tion and reverse phase HPLC purification. Inter-
mediates 6 could also be converted to a cyanamide,
which in turn was transformed into N-hydroxy-
guanidine 8.5
We planned to compensate for this reduction in potency
byincorporating additional chemical features in the
inhibitors, primarilythrough introduction of various P1
residues. In this paper, we will describe our efforts in
this direction.
The synthesis of the benzimidazolyl–methylamine is
described in Scheme 2. Commerciallyavailable acid
was first esterified, after which the amine was tritylated.
Ester 10 was then reduced (LiAlH4, THF) and the
9
*Corresponding author. Tel.: +1-450-978-7786; fax: +1-450-978-
7777; e-mail: slevesque@ca.shire.com
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00661-8

´
S. Levesque et al. / Bioorg. Med. Chem. Lett. 11 (2001) 3161–3164
3162
ꢀ
.
Scheme 1. (a) (BOC)2O, CH2Cl2, 100%; (b) BH3 THF, THF, 100%; (c) MsCl, Et3N, CH2Cl2, 0 C; (d) NaN3, DMF, Á, 95% (two steps); (e) H2,
Pd/C 10%, MeOH, 100%; (f) bicyclic-CO2H, HATU, 2,4,6-collidine, 95%; (g) HCl in dioxane, 100%; (h) N,N-(t-butoxycarbonyl)-1H-pyrazole-
.
carboxamidine, iPr2NEt, DMF, 82%; (i) 4 N HCl in dioxane, 70%; (j) BrCN, CH3CO2Na, MeOH, 80%; (k) NH2OH HCl, Na2CO3, MeOH, 87%.
Scheme 2. (a) EDC, DMAP, MeOH, 26%; (b) Tr–Cl, TEA, CH2Cl2,
Scheme 3. (a) Ethylene glycol, pTSA, benzene, Á, 85%; (b) Super-
77%; (c) LiAlH4, THF, 72%; (d) SOCl2, CH2Cl2, 0 ꢀC, 84%; (e)
NaN3, DMF, Á, 77%; (f) H2, Pd/C 10%, MeOH, 100%; (g) Bicyclic-
CO2H, HATU, 2,4,6-collidine, DMF, 65%; (h) TFA/MeOH/CHCl3,
51%.
hydride, THF, ꢁ78!0 ꢀC; (c) MsCl, Et3N, CH2Cl2, 0 ꢀC; (d) NaN3,
DMF, Á; 62% (three steps); (e) H2, Pd/C 10%, MeOH, 22%; (f)
bicyclic-CO2H, HATU, 2,4,6-collidine, DMF, 100%; (g) PPTS, ace-
tone/H2O, 57%; (h) (i) NH2OH, MeOH; (ii) NaBH3CN, AcOH, 26%.
obtained alcohol transformed into chloride 11. Displace-
ment of the chloride with NaN3 in DMF, followed by
hydrogenolysis of the azide afforded amine 12, which was
coupled with the bicyclic template. Removal of the trityl
protecting group gave the benzimidazole compound 13.
Scheme 3 describes the synthesis of the aminomethyl–
cyclohexyl–hydroxylamine. Commercially available
ketone 14 was first protected as an acetal, and the ester
group was reduced to give alcohol 15. The alcohol
was then transformed into amine 16 as in Scheme 1.
Scheme 4. (a) BuLi, DMF, THF, ꢁ100 ꢀC, 86%; (b) NaBH4, EtOH, 96%; (c) MsCl, DIPEA, CH2Cl2; (d) NaN3, DMF, Á, 95% (two steps); (e) (i)
HCl in EtOH, 4 ꢀC; (ii) NH3 in EtOH; (f) H2, Pd/C 10%, MeOH; (g) bicyclic-CO2H, HATU, NMM, DMF, 36% (three steps); (h) (i) HCl in EtOH,
ꢀ
.
4 C; (ii) NH3 in EtOH, 70%; (i) NH2OH HCl, Na2CO3, MeOH, 26%.

´
S. Levesque et al. / Bioorg. Med. Chem. Lett. 11 (2001) 3161–3164
3163
Coupling of amine 16 with the bicyclic template fol-
lowed by hydrolysis of the acetal lead to ketone 17,
which was transformed into hydroxylamine 18 by
reductive amination.
aldehyde produced was then reduced with NaBH4 to
give alcohol 20. Transformation of the cyano group of
21 into the amidine group was done according to a
published procedure.6 Formation of amine 22 and cou-
pling with the bicyclic template was accomplished as in
Scheme 1. Cyanobenzyl amine 24 from the coupling of
the bicyclic template with commercially available 4-
cyanobenzyl amine, was transformed into amidine 25
and amidoxime 26 in modest to good yield.
Finally, the preparation of the phenylamidines is
described in Scheme 4. Bromide 19 was first lithiated,
and the anion formed was formylated with DMF. The
Table 1. In vivo potencyand selectivity
Inhibition of the amidolytic activity of thrombin (Ki)
and in vivo coagulation parameters in the rat arterial
thrombosis model such as the mean occlusion time
(MOT), the activated partial thromboplastin time
(aPTT) and the thrombin time (TT) were measured
Compd
P3
P1
Ki thr (nM)
Selb
—
27a
900
7
according to an alreadypublished procedure.
Diverse range of P1 residues explored are presented in
Table 1. The starting point of the SAR was inhibitor 27
7a
7b
180
78
—
51
(Ki=900 nM), where the ketothiazole moietyof
1
(Ki=0.6 nM) has been removed completely.8 As expec-
ted, the binding affinityof this inhibitor was rather low,
as one of the major interactions with thrombin has been
nullified, namely, the interaction between the electro-
philic carbonyl and Ser195. The inhibitor 7a displayed
ꢂ5-fold improvement in potency. This improved bind-
ing affinityof the inhibitor could be due to the higher
rigidityof the P1 group, and hence better alignment of
the guanidine group, and also to the additional hydro-
phobic interactions introduced bythe piperidine group.
Additional improvement in potencywas attainable by
replacement of the phenylpropanoyl group with the
benzyl sulfonyl group (7b). Replacement of the amidine
fragment of the P1 residue byamidoxime significantly
improved the potency(compare 8 and 7b). In order to
explain this unexpected improvement in potency, com-
pound 8 was co-crystallized with thrombin (Fig. 1).9 We
can clearlysee in Figure 1, that the benzylsulfonamide
group penetrates deeplyin the S3 pocket of thrombin
and also that the Ser195 is no more implicated in the
binding of the inhibitor. The amidine fragment makes
a salt bridge with Asp189 and the presence of the
amidoxime OH group forms an hydrogen bond with
Tyr22B in the S1 pocket. This additional and unprece-
8
14
100
1100
8
296
250
1
28
18
25
23
5
dented interaction mayexplain the ꢂ5- to 6-fold
improved potencyof inhibitor 8 compared to inhibitor
7b.
7
11
26
13
>10,000
—
16
1000
aAll new target compounds were characterized by 1H NMR, reverse-
phase HPLC and mass spectroscopy.
bKi trypsin/Ki thrombin
.
Figure 1.

´
S. Levesque et al. / Bioorg. Med. Chem. Lett. 11 (2001) 3161–3164
3164
Table 2. In vivo parameters
Boudreau for conducting in vivo experiments, and Ms.
Lyne Marcil for her help in the preparation of this
manuscript.
Compd
Rat arterial thrombosis modela
APTTc
MOTb
TTd
8
25
23
48ꢃ13
>60
46ꢃ10
38ꢃ5
231ꢃ29
455ꢃ64
393ꢃ102
References and Notes
58ꢃ11
54ꢃ14
1. St-Denis, Y.; Augelli-Szafran, C. E.; Bachand, B.; Berry-
man, K. A.; DiMaio, J.; Doherty, A. M.; Edmunds, J. J.;
Leblond, L.; Levesque, S.; Narasimhan, L. S.; Penvose-Yi,
J. R.; Rubin, J.; Tarazi, M.; Winocour, P. D.; Siddiqui, M. A.
Bioorg. Med. Chem. Lett. 1998, 8, 3193.
aDose: intravenous bolus dose (0.75 mg/kg) followed byan infusion
(50 mg/kg/min).
bMean occlusion time in min (control: 17/19 min).
cActivated partial thromboplastin time in seconds (control: 20/22 s).
dThrombin time in seconds (control: 40/45 s).
2. Bachand, B.; Tarazi, M.; St-Denis, Y.; Edmunds, J. J.;
Winocour, P. D.; Leblond, L.; Siddiqui, M. A. Bioorg. Med.
Chem. Lett. 2001, 11, 287.
3. For other examples of non-covalent thrombin inhibitors
see: Fevig, J. M.; Wexler, R. R. Annu. Rep. Med. Chem. 1999,
34, 81.
4. Lyle, T. A.; Chen, Z.; Appleby, S. D.; Freidinger, R. M.;
Gardell, S. J.; Lewis, S. D.; Li, Y.; Lyle, E. A.; Lynch, J. J.;
Mulichak, A. M.; Ng, A. S.; Naylor-Olsen, A. M.; Sanders,
W. M. Bioorg. Med. Chem. Lett. 1997, 7, 67.
5. Kim, K. S.; Moquin, R. V.; Qian, L.; Morrison, R. A.;
Seiler, S. M.; Roberts, D. G. M.; Ogletree, M. L.; Youssef, S.;
Chong, S. Med. Chem. Res. 1996, 6, 377. See also: Kim, K. S.;
Kimball, S. D.; Das, J.; Iwanowicz, E. J.; Han, W.-C.
EP95108266.8, 1995.
The cyclohexylamine and piperidylamidine moieties
were comparable P1 residues in terms of thrombin affi-
nities. However, the cyclohexylamine residue displayed
an improved selectivityindex ( 28 vs 7b). The introduc-
tion of phenyl amidines improved the binding affinities
of inhibitors 25 and 23. However, there was a significant
loss of selectivity, presumably due to the reduced lipo-
philicityof the aromatic ring compared to the piperidyl
and cyclohexyl rings. Inhibitors 13, 18, and 26 were all
inactive, suggesting the highlyspecific nature of the P1
subsite.
The three inhibitors having the optimal in vitro poten-
cies were then tested in the rat arterial thrombosis
model (see Table 2, analogues 8, 25, and 23). All of
them displayed greater than 2-fold increase in mean
occlusion time. No occlusion was observed with inhi-
bitor 25 for the whole duration of the experiment
(60 min).
6. Nagahara, T.; Yokoyama, Y.; Inamura, K.; Katakura, S.-
I.; Komoriya, S.; Yamaguchi, H.; Hara, T.; Iwamoto, M. J.
Med. Chem. 1994, 37, 1200.
7. Finkle, C. D.; St-Pierre, A.; Leblond, L.; Deschenes, I.;
DiMaio, J.; Winocour, P. D. Thromb. Haemostasis. 1998, 79,
431.
8. Compound 27 was prepared in a straightforward manner
.
as shown below:
We have demonstrated that the activated carbonyl of
compounds 1 and 2 was not necessaryfor high potency
and selectivity. By modifying P1 residues of these inhi-
bitors, we could capture additional interactions in order
to improve the binding affinityand selectivityfor
thrombin. Selected compounds were administered via
intravenous infusion in the rat arterial thrombosis
model and were shown to have high in vivo efficacy.
Acknowledgements
The authors would like to acknowledge Ms. Brigitte
Grouix and Ms. Quan Yang for providing binding
constants, Ms. Isabelle Deschenes and Ms. Chantal
9. The authors have deposited X-raycrsytallographic data
with the Brookhaven Protein Data Bank. Deposition code
1JWT.