Improved stability of proline-derived direct thrombin inhibitors through hydroxyl to heterocycle replacement

Article abstract of DOI:10.1021/acsmedchemlett.5b00047

Modification of the previously disclosed (S)-N-(2-(aminomethyl)-5-chlorobenzyl)-1-((R)-2-hydroxy-3,3-dimethylbutanoyl)pyrrolidine-2-carboxamide 2 by optimization of the P3 group afforded novel, low molecular weight thrombin inhibitors. Heterocycle replace

Full text of DOI:10.1021/acsmedchemlett.5b00047

Letter
pubs.acs.org/acsmedchemlett
Improved Stability of Proline-Derived Direct Thrombin Inhibitors
through Hydroxyl to Heterocycle Replacement
Harry R. Chobanian,*,† Barbara Pio,† Yan Guo,† Hong Shen,† Mark A. Huffman,‡ Maria Madeira,§
Gino Salituro,§ Jenna L. Terebetski,∥ James Ormes,∥ Nina Jochnowitz,⊥ Lizbeth Hoos,⊥ Yuchen Zhou,⊥
Dale Lewis,⊥ Brian Hawes,⊥ Lyndon Mitnaul,# Kim O’Neill,# Kenneth Ellsworth,# Liangsu Wang,#
Tesfaye Biftu,† and Joseph L. Duffy†
†
‡
§
∥
Departments of Medicinal Chemistry, Process Chemistry, Drug Metabolism and Pharmacokinetics, Preclinical Development,
⊥Pharmacology, and Thrombosis, Merck Research Laboratories, Kenilworth, New Jersey 07033, United States
#
S
* Supporting Information
ABSTRACT: Modification of the previously disclosed (S)-N-
(2-(aminomethyl)-5-chlorobenzyl)-1-((R)-2-hydroxy-3,3-
dimethylbutanoyl)pyrrolidine-2-carboxamide 2 by optimiza-
tion of the P3 group afforded novel, low molecular weight
thrombin inhibitors. Heterocycle replacement of the hydroxyl
functional group helped maintain thrombin in vitro potency
while improving the chemical stability and pharmacokinetic
profile. These modifications led to the identification of
compound 10, which showed excellent selectivity over related
serine proteases as well as in vivo efficacy in the rat
arteriovenous shunt. Compound 10 exhibited significantly
improved chemical stability and pharmacokinetic properties
over 2 and may be utilized as a structurally differentiated preclinical tool comparator to dabigatran etexilate (Pro-1) to
interrogate the on- and off-target effects of oral direct thrombin inhibitors.
KEYWORDS: Thrombin, proline, dabigatran, thrombosis, warfarin, serine protease
hromboembolic diseases remain the leading cause of
1
Tpreventable hospital death in the United States. Venous
thromboembolism is estimated to affect 1−2 million people in
the United States, progressing to pulmonary embolism in
approximately 600,000 of these patients and becoming fatal in
up to 100,000 patients annually.1,2 The conditions amenable to
treatment with anticoagulants are broad, reflecting the
importance of thrombosis in the pathophysiology of multiple
diseases. Indications for anticoagulant therapy include stroke,
myocardial infarction, and cerebral ischemia. Thrombin is a
serine protease that plays a central role in the blood coagulation
cascade by mediating the conversion of fibrinogen to fibrin.
Thrombin also affects arterial thrombosis via activation of the
protease activating receptor PAR1.3
Dabigatran etexilate (Pro-1, Figure 1) remains the only
approved oral direct thrombin inhibitor and has been shown to
reduce the risk of stroke and systemic embolism in nonvalvular
atrial fibrillation, as well as the treatment of deep venous
thrombosis.4 Dabigatran etexilate is dosed as a double prodrug
and is rapidly converted by a serum esterase to dabigatran (1)
in vivo. One shortcoming for dabigatran etexilate is the oral
bioavailability of the double prodrug, which is reported to be
between 3 and 7%.5 In addition, for desired efficacy, dabigatran
etexilate needs to be dosed twice a day (BID), which can be
problematic for compliance in some of the patient population.
Figure 1. Dabigatran etexilate (Pro-1), dabigatran (1), and previously
reported thrombin inhibitor 2.
A reported liability of dabigatran etexilate is an increased risk of
gastrointestinal bleeding as compared to warfarin. It is not
currently known whether this bleeding is unique to dabigatran,
a byproduct of the prodrug metabolism, or general to all oral
direct thrombin inhibitors. Therefore, we sought to discover
structurally distinct direct thrombin inhibitors, devoid of the
need for prodrug derivatization, which would be useful tools to
further understand the pharmacology of this target. Addition-
ally, we sought a compound with a narrow peak to trough ratio
(≤3) in order to minimize bleeding risk and increase the
Received: January 29, 2015
Accepted: March 13, 2015
Published: March 13, 2015
© 2015 American Chemical Society
553
DOI: 10.1021/acsmedchemlett.5b00047
ACS Med. Chem. Lett. 2015, 6, 553−557

ACS Medicinal Chemistry Letters
Letter
possibility for a once daily (QD) dosing paradigm. As a starting
point, we further refined our previously disclosed thrombin
inhibitor 2, which has been profiled extensively.6
Table 1. Direct Thrombin Inhibitors from Scheme 1
On the basis of the pharmacokinetic profile in dogs (t1/2
=
3.9 h, F = 81%) for compound 2, we surmised that BID dosing
would be required to maintain efficacious levels of exposure.7 A
major additional shortcoming recognized with compound 2 is
chemical instability at physiologically relevant pH. Under
slightly basic conditions the secondary alcohol drives intra-
molecular cyclization, giving rise to the bicyclic morpholine
dione 2a while extruding dibenzylamine 2b (Figure 2).8 In this
Figure 2. Cyclization pathway of compound 2 under physiologically
relevant pH ranges.
Letter, we fully characterize the chemical stability of 2 at pH 7.4
and describe our approach to identify more stable derivatives
that maintain the high thrombin potency and selectivity profile
of 2. In addition, we desired to improve upon the
pharmacokinetic profile of 2 to provide a compound
appropriate for a QD dosing regimen.
Our initial goal was to identify a heterocyclic H-bond donor
that maintains the exquisite in vitro potency profile of
compound 2. The synthesis of these analogues commenced
with use of the previously described proline intermediate 4,
prepared according to procedures outlined in Scheme 1.6
a
Scheme 1. Synthesis of Compounds in Table 1
a
Reagents and conditions: (a) Fmoc-L-proline, EDCI, HOBT, Hunig’s
base, DMF; (b) piperidine, DMF; (c) R1COOH, EDCI, HOBT,
Hunig’s base, DMF; (d) 4 N HCl in dioxane.
Standard amino acid coupling of various acyl heterocyclic
moieties at the P-3 region of the molecule was accomplished
followed by BOC deprotection using 4 N HCl in dioxane to
provide the desired final products.
vitro Ki potency of 44 nM. Capping of the pyrrole with a Me
group (9) decreased the potency 20-fold, indicating that the
free N−H of the pyrrole was critical for in vitro potency similar
to the hydroxyl moiety of compound 2. Additional substitution
of a Me group directly on the pyrrole ring led to compounds 10
and 11, with superior potency (9.6 and 17 nM respectively)
and very good trypsin selectivity. Further modification of the 2-
position of the pyrrole with a Cl atom led to compound 12,
which also showed excellent overall potency (11 nM) but
afforded a lower, 100-fold window of selectivity over trypsin as
compared to 10. Further substitution of the pyrrole by means
of a chlorophenyl substituent such as compound 13 or 4-
chlorobenzoyl pyrrole 14 at the 3-position also led to
compounds with excellent thrombin potency and significantly
improved trypsin selectivity. In addition, an indole and two
azaindoles were prepared in order to ascertain the effect of a
fused heterocyclic system. Potencies were quite respectable
with compounds 15 and 17 displaying potency values of 65 and
33 nM, respectively. The regioisomeric azaindole 16 exhibited a
The compounds synthesized were evaluated for inhibition of
thrombin (Table 1). Thrombin itself is a serine protease in the
trypsin family. We regularly counterscreened all compounds
against trypsin, a serine protease present in the gut with similar
substrate prerequisites to thrombin. Inhibition of trypsin-like
enzymes unrelated to the coagulation pathway, but known to
be necessary for physiological functions, could have deleterious
consequences.9 Both dabigatran (1) and compound 2 were
found to exhibit exquisite potency as thrombin inhibitors in our
isolated enzyme assay (Ki = 1.3 and 2.1 nM, respectively).
Among the heterocycle replacements for the hydroxyl
substituent in 2, pyrazole 6 and imidazole 7 were far less
potent with thrombin Ki values of 366 and 193 nM,
respectively, while maintaining good selectivity against trypsin.
Interestingly, a simple pyrrole (compound 8) afforded an in
554
DOI: 10.1021/acsmedchemlett.5b00047
ACS Med. Chem. Lett. 2015, 6, 553−557

ACS Medicinal Chemistry Letters
Letter
10-fold drop in thrombin potency from the azaindole 17 while
maintaining trypsin selectivity. This indicates that the pyridynyl
nitrogen had a substantial effect on the H-bonding capability of
the azaindole.
Additionally, we did examine the selectivity of compounds 10
and 12 against serine proteases besides trypsin. In the case of
compound 10, there was a 30× selectivity window against
Factor Xa and a >1500× window against both Factors VIIa and
XIIa (Table 2). Compound 12 displayed a 200× window over
Factor VIIa, a 40× window over Factor Xa, and a >1500×
window over Factor XIIa.
kinetic profiles for both analogues 10 and 12 were measured in
dogs, we were delighted to find both compounds were orally
bioavailable (42% and 82%, respectively) and possessed very
low clearance values (0.31 and 0.29 mL/min/kg). In addition,
the half-life for both compounds (t1/2) was significantly
improved over 2 and is consistent with QD dosing in humans,
showing an advantage over dabigatran (1). To our delight, over
a 24 h time course, compounds 10 and 12 also maintained a
peak to trough ratio of ≤3, which was a key initial goal for the
program (Figure 3).14
Table 2. In Vitro Coagulation Pathway Selectivity Profile for
Compounds 10 and 12
compd
factor VIIa
factor Xa
factor XIIa
10
12
>15 μM
2.3 μM
295 nM
397 nM
>15 μM
>15 μM
On the basis of promising in vitro data, we sought to further
profile compounds 10 and 12 in an assay that measures the
concentration of a test compound required to double the
activated partial thromboplastin time (2 × APTT) in human
plasma (Table 3).10 Both dabigatran (1) and compound 2 were
Figure 3. Dog IV PK curve for compounds 10 and 12.
Table 3. APTT Data for Thrombin Compounds 1, 2, 10, and
12
We also examined the chemical stability of compounds 10
and 12 versus 2 at a pH that was physiologically relevant. The
true question of whether the acyl pyrrole moiety is as prone to
cyclization was answered by subjecting each compound in a pH
7.4 buffer solution at 25 °C and examining the percentage of
parent species remaining after 24 h (Figure 4). Compound 2
compd
APTT (2×) (μM)
1
0.63
0.23
6.9
2
10
12
14
found to exhibit good functional activity in this coagulation
assay in human plasma (2 × APTT = 0.63 and 0.23 μM,
respectively). Compound 10 possessed anticlotting activity (2
× APTT = 6.9 μM) roughly 10-fold less potent than
dabigatran. Compound 12 was the least effective (2 × APTT
= 14 μM) in the APTT assay for thrombosis.11
Pharmacokinetic data (PK) was generated for compounds 2,
10, and 12 in both rat and dog (Table 4). Compound 2
suffered from poor rat PK with very high clearance and poor
oral absorption at 10 mg/kg.6 The dog half-life for 2 was
measured to be 3.9 h, which was critical to improve for a
potential QD clinical profile.12,13 Rat PK for 10 exhibited poor
oral bioavailability (F = 7%) and very high clearance (Cl = 64
mL/min·kg) overall. Compound 12 exhibited a similarly short
half-life and high clearance; however, the bioavailability was
improved by 3-fold over compound 10. When the pharmaco-
Figure 4. Chemical stability of compounds 2, 10, and 12 at 25 °C, pH
7.4 (50:50 10 mM phosphate buffer/acetonitrile).
degraded over 24 h by 20.5% through conversion to dione 2a.
Interestingly, compounds 10 and 12 did not show any
appreciable degradation over the 24 h time period. The
Table 4. PK Data for Compounds 2, 10, and 12
f
compd
species
dose IV/PK (mg/kg)
AUCN (μM·kg/mg)
t1/2 (h)
Cl (mL/min/kg)
F (%)
a
b
2
rat
2 /10
0.2
2.9
3.9
1.4
19
81
37
81
7
c
c
2
dog
rat
0.75 /1
10.1
3.5
64
d
d
a
10
10
12
12
1 /3
0.96
e
e
dog
rat
1 /2
166
0.9
0.31
52
42
23
82
d
d
1 /3
1.4
15.5
e
e
dog
1 /2
241
0.29
a
b
c
d
Dosing vehicle DMSO. Dosing vehicle 1% methylcellulose. Dosing vehicle 5% dextrose/PEG (60:20). Dosing vehicle DMSO/PEG300/H2O
e
f
(10:50:40), Dosing vehicle 5% DMA/30% PEG400/65% (40% HPCD). Normalized AUC (PO, μM·h·kg/mg).
555
DOI: 10.1021/acsmedchemlett.5b00047
ACS Med. Chem. Lett. 2015, 6, 553−557

ACS Medicinal Chemistry Letters
Letter
improved chemical stability could be a direct result of the
higher measured pKa for the acyl pyrroles 10 and 12 (∼pKa =
16.5) versus compound 2 (pKa =13.6) when subjected to a pH
of 7.4.
We chose to compare the in vivo efficacy of compound 2 to
compound 10 in the rat arteriovenous shunt (AV shunt)
thrombosis model rather than to compound 12 (Figure 5),15
ASSOCIATED CONTENT
* Supporting Information
Synthetic procedures and analytical data of selected thrombin
inhibitors, and conditions for all the biological assays. This
material is available free of charge via the Internet at http://
pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*Tel: +1-908-740-5131. Fax: +1-908-740-6137. E-mail: harry_
chobanian@merck.com.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The author wishes to thank the department of Laboratory
Animal Resources for their assistance in animal dosing and
sampling. The author also wishes to thank Dr. Andrea
Nawrocki for helpful discussions.
REFERENCES
■
Figure 5. Rat AV-shunt study with compounds 1, 2, and 10.
(1) Beckman, M. G.; Hooper, W. C.; Critchley, S. E.; Ortel, T. L.
Venous thromboembolism: A public health concern. Am. J. Prev. Med.
2010, 38, S495−S501.
(2) Deitelzweig, S. B.; Johnson, B. H.; Lin, J.; Schulman, K. L.
Prevalence of clinical venous thromboembolism in the USA: Current
trends and future projections. Am. J. Hematol. 2011, 86, 217−220.
(3) Dockendorff, C.; Aiksu, O.; VerPlank, L.; Dilks, J. R.; Smith, D.
A.; Gunnink, S. F.; Dowal, L.; Negri, J.; Palmer, M.; MacPherson, L.;
Schreiber, S. L.; Flaumenhaft, R. Discovery of 1,3,-diaminobenzenes as
selective inhibitors of platelet activation at the PAR1 receptor. ACS
Med. Chem. Lett. 2012, 3, 232−237.
(4) Greig, S. L.; McKeage, K. Dabigatran etexilate: A review of its use
in the treatment of acute venous thromboembolism and prevention of
venous thromboembolism recurrence. Drugs 2014, 74, 1785−1800.
(5) Blech, S.; Ebner, T.; Ludwig-Schwellinger, E.; Stangier, J.; Roth,
W. The metabolism and disposition of the oral direct thrombin
inhibitor, Dabigatran, in humans. Drug Metab. Dispos. 2008, 386−399.
(6) Morrissette, M. M.; Stauffer, K. J.; Williams, P. D.; Lyle, T. A.;
Vacca, J. P.; Krueger, J. A.; Lewis, S. D.; Lucas, B. J.; Wong, B. K.;
White, R. B.; Miller-Stein, C.; Lyle, E. A.; Wallace, A. A.; Leonard, Y.
M.; Welsh, D. C.; Lynch, J. J.; McMasters, D. R. Low molecular weight
thrombin inhibitors with excellent potency, metabolic stability, and
oral bioavailability. Bioorg. Med. Chem. Lett. 2004, 14, 4161−4164.
(7) Projected human-half-life of compound 2 is 7−8 h based on
allometric scaling of the dog PK.
due to the superior APTT response of 10 over 12 (Table 2). As
a positive control, dabigatran (1) was also investigated in this
assay. Compound 2 inhibited thrombus formation in a dose-
dependent manner from 0.03 to 1.0 mg/kg via IV infusion. At
the 1 mg/kg dose, a 90% reduction in clot weight was observed
with a mean exposure of 0.74 μM. Compound 10 showed good
activity with a 63% inhibition of clot weight at 1.0 mg/kg with
an exposure of 1.04 μM. In comparison, dabigatran (1) showed
a 96% reduction in clot weight at 0.3 mg/kg with an exposure
of drug of 0.189 μM.
The efficacy profile exhibited by all three compounds in the
rat AV shunt study is reasonably reflective of the human 2 ×
APTT concentration from Table 2. Dabigatran exhibited
excellent in vivo efficacy with a steep dose/response curve.
Compound 2 showed superior efficacy at a lower dose over
compound 10, which is likely a manifestation of better overall
intrinsic potency over compound 10 (∼10-fold) (Figure 5).
Additionally, compound 2 possessed a free fraction in rat of
54%, while compound 10 was slightly lower at 40% free drug.12
When examining the positive aspects, however, of improved
chemical stability and blunted peak to trough ratio leading to
likely QD dosing with minimized bleeding risk, it becomes
apparent that compound 10 could be utilized as a valuable
research tool for further pharmacological elucidation of the
thrombin pathway.
(8) Nelson, T. D.; LeBlond, C. R.; Frantz, D. E.; Matty, L.; Mitten, J.
V.; Weaver, D. G.; Moore, J. C.; Kim, J. M.; Boyd, R.; Kim, P.-Y.;
Gbewonyo, K.; Brower, M.; Sturr, M.; McNamara, J. M.; Dolling, U.
H. Stereoselective synthesis of a potent thrombin inhibitor by a novel
P2-P3 lactone ring opening. J. Org. Chem. 2004, 69, 3620−3627.
(9) Rawlings, N. D.; Barrett, A. J. Families of serine peptidases.
Methods Enzymol. 1994, 219−261.
In summary, we have discovered a novel heterocyclic
replacement for our previously disclosed low molecular weight
thrombin inhibitor 2. The acyl pyrroles 10 and 12 improved
significantly upon both the chemical stability and PK profile
parameters set forth to make the compound amenable to a QD
dosing regimen with a blunted peak to trough ratio. These
features, along with structural diversity and lack of prodrug
derivatization as compared to dabigatran−etexilate provided a
differentiated tool compound 10. This tool may aid in the
development of preclinical models to interrogate the
pharmacology of thrombin inhibition as a structurally distinct
comparator to dabigatran.
(10) Korte, W.; Clarke, S.; Lefkowitz, J. B. Short activated partial
thromboplastin times are related to increased thrombin generation and
an increased risk for thromboembolism. Am. J. Clin. Pathol. 2000, 113,
123−127.
(11) Plasma free fraction (% unbound): Compound 2 human (24%)
rat (54%);6 compound 10 human (2.4%), rat (40%).
(12) Hepatocyte stability % remaining at 30 min: Compound 10:
Human (98%), Dog (88%), Rat (62%); Compound 12: Human
(91%), Dog (80%), Rat (48%).
(13) Cmax values for Compound 10 = 8.4 μM (Dog), 0.02 μM (Rat);
Cmax values for Compound 12 = 8.8 μM (Dog), 0.18 μM (Rat).
(14) The projected human half-life of compound 10 is calculated to
be 35 h based on allometric scaling of the dog PK.
556
DOI: 10.1021/acsmedchemlett.5b00047
ACS Med. Chem. Lett. 2015, 6, 553−557

ACS Medicinal Chemistry Letters
Letter
(15) Smith, J. R.; White, A. M. Fibrin, red cell and platelet
interactions in an experimental model of thrombosis. Br. J. Pharmacol.
1982, 77, 29−38.
557
DOI: 10.1021/acsmedchemlett.5b00047
ACS Med. Chem. Lett. 2015, 6, 553−557