Macrocyclic factor XIa inhibitors

Article abstract of DOI:10.1016/j.bmcl.2017.07.048

A series of macrocyclic factor XIa (FXIa) inhibitors was designed based on an analysis of the crystal structures of the acyclic phenylimidazole compounds. Further optimization using structure-based design led to inhibitors with pM affinity for FXIa, excellent selectivity against a panel of relevant serine proteases, and good potency in the activated partial thromboplastin time (aPTT) clotting assay.

Full text of DOI:10.1016/j.bmcl.2017.07.048

Bioorganic & Medicinal Chemistry Letters 27 (2017) 4056–4060
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Macrocyclic factor XIa inhibitors
Cailan Wang a, James R. Corte a, Karen A. Rossi a, Jeffrey M. Bozarth c, Yiming Wu c, Steven Sheriff b,
Joseph E. Myers Jr. b, Joseph M. Luettgen c, Dietmar A. Seiffert b, Ruth R. Wexler a, Mimi L. Quan a,
⇑
a Bristol-Myers Squibb Company, Research and Development, 350 Carter Road, Hopewell, NJ 08540 United States
b Bristol-Myers Squibb Company, Research and Development, US Rt. 206 & Province Line Road, Princeton, NJ 08540 United States
c Bristol-Myers Squibb Company, Research and Development, 311 Pennington-Rocky Hill Road, Pennington, NJ 08543-2130 United States
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of macrocyclic factor XIa (FXIa) inhibitors was designed based on an analysis of the crystal struc-
tures of the acyclic phenylimidazole compounds. Further optimization using structure-based design led
to inhibitors with pM affinity for FXIa, excellent selectivity against a panel of relevant serine proteases,
and good potency in the activated partial thromboplastin time (aPTT) clotting assay.
Ó 2017 Elsevier Ltd. All rights reserved.
Received 28 June 2017
Revised 13 July 2017
Accepted 19 July 2017
Available online 25 July 2017
Keywords:
Factor XIa inhibitors
FXIa
Activated partial thromboplastin time
aPTT
Thrombosis
Thrombosis, the abnormal occlusion of a blood vessel which can
lead to myocardial infarction and ischemic stroke is the leading
cause of mortality and morbidity.1 Approved thrombin and factor
Xa (FXa) inhibitors2 such as dabigatran, rivaroxaban, apixaban,
and edoxaban are effective in the treatment and prevention of
thrombosis. These novel oral anticoagulants (NOACs) have also
addressed many issues related to warfarin.3 However, there
remains an unmet medical need for more efficacious medications
with less bleeding risk.4 Factor XIa is located upstream in the coag-
ulation cascade and plays a major role in the amplification and
propagation of thrombin production.5 Genetic evidence indicates
FXIa could be an antithrombotic target with an improved net clin-
ical benefit.6–8 Preclinical research has demonstrated that reversi-
ble FXIa inhibitors show robust efficacy with minimum bleeding in
animal models.9 This evidence suggests that inhibition of FXIa
could be an effective means for preventing thrombosis with a
reduced bleeding risk as compared to current therapies. Further-
more, a FXI-directed antisense oligonucleotide (ASO) has been
shown to prevent thrombosis and appeared safe with respect to
bleeding in a recent Phase II clinical trial.10
reported one of our macrocyclization approaches by removing
the P1 prime phenyl and connecting the P1 prime to the ortho-
position of the P2 prime phenyl via an alkyl linker (approach A
in Fig. 1).12 Herein, we describe another macrocyclization approach
by linking the P1 prime phenyl directly to the P2 prime phenyl as
shown by approach B in Fig. 1.
The FXIa bound X-ray structures of compound 111c and other
related compounds showed that the binding conformation of the
phenylimidazole-phenylalanine portion of the molecules are
similar in most of the structures (Fig. 2). We envisioned based on
proximity that the two phenyl groups could be linked as shown
in Fig.
1 to form macrocycles that would pre-organize the
inhibitor into its binding conformation and further improve its
binding efficiency.
Various linkers connecting the meta- or para-position of the P1
prime phenyl to the ortho-position of the P2 prime phenyl were
designed guided by modeling using the X-ray structure of 1. The
initial set of compounds with a 3-atom amide linker resulted in
the m-linked compounds 4 and 5 being more potent than the p-
linked analogs 6 and 7 (Table 1). The same trend was observed
for compounds 8–10 with the 4-carbon linker. The p-linked con-
nection altered the trajectory of the linker which contributed to a
loss in FXIa affinity. Similar to the acyclic series, chloro-substitu-
tion at the 5-position of the imidazole improved FXIa affinity
(see m-linked compounds 5 and 9).11a The compounds with the
4-atom alkyl linker were more potent than the compounds with
the 3-atom amide linker. Molecular modeling suggested the
We have previously described a series of phenylimidazole FXIa
inhibitors exemplified by compound 1 (Fig. 1).11 To further
enhance FXIa affinity, we sought to preorganize the bioactive con-
formation of 1 via a macrocyclization strategy. Recently we
⇑
Corresponding author.
E-mail address: mimi.quan@bms.com (M.L. Quan).
http://dx.doi.org/10.1016/j.bmcl.2017.07.048
0960-894X/Ó 2017 Elsevier Ltd. All rights reserved.

C. Wang et al. / Bioorganic & Medicinal Chemistry Letters 27 (2017) 4056–4060
4057
Fig. 1. From acyclic to macrocycle FXIa inhibitors.
Table 1
Initial macrocycles.
Compd
L
Ring size
R
FXIa Kia (nM)
4
5
6
7
8
9
10
m-CONH
m-CONH
p-CONH
p-CONH
m-(CH2)3
m-(CH2)3
p-(CH2)3
13
13
14
14
14
14
15
H
Cl
H
Cl
H
Cl
H
500
92
6400
8100
17
8.2
71
a
Ki values were obtained from purified human enzyme and a synthetic peptides substrate at 25 °C and were averaged from multiple
determinations (n = 2), as described in Ref. 13.
conformation of the macrocycle with the 3-atom amide linker was
strained. The 14-membered ring macrocycle containing the longer
linker alleviated the strain thus leading to improved FXIa affinity.
The human plasma concentration of FXIa inhibitor which
produced an increase of 50% over the baseline in activated partial
thromboplastin time (aPTT1.5x) was used to assess the in vitro
anticoagulant activity. Compound
nanomolar affinity for FXIa with moderate in vitro anticoagulant
activity (aPTT1.5x = 34 M).
9
achieved single digit
l
Modeling suggested that the overall binding mode for macrocy-
cle 9 was similar to acyclic phenylimidazole 1 (Fig. 2) and further
indicated the possibility to form a hydrogen bond with the car-
bonyl of Leu41 with an NH moiety in the linker, similar to our alkyl
linked macrocycles previously reported.12 Therefore the amide lin-
ker was incorporated in compound 11 resulting in an approximate
10-fold improvement in both the FXIa affinity and in vitro antico-
agulant activity (aPTT) compared to 8 (Table 2). Initial molecular
modeling also suggested that incorporation of substitution at the
para-position of the P1 prime phenyl ring could interact with the
Cys42-Cys58 disulfide bridge to further improve binding affinity.
Indeed, fluorine-analog 12 enhanced FXIa affinity by 3-fold and
achieved subnanomolar affinity. Replacement of the P1 prime phe-
nyl ring with a pyridine led to compound 13 with 5-fold improve-
ment in both FXIa affinity and in vitro anticoagulant activity (aPTT)
Fig. 2. Superposition of crystal structure of 1 (in purple) and model of 9 (in cyan).
compared to 11. Further SAR efforts to incorporate both modifica-
tions into one molecule resulted in compound 14 with picomolar
FXIa affinity (FXIa Ki = 0.02 nM). Compound 14 is highly selective

4058
C. Wang et al. / Bioorganic & Medicinal Chemistry Letters 27 (2017) 4056–4060
Table 2
Linker optimization of the 14-macrocycles.
Compd
R/X
L
FXIa Kia (nM)
aPTTb
(mM)
1.5x
8
H/CH
H/CH
F/CH
H/N
CH2CH2
C(O)NH
C(O)NH
C(O)NH
C(O)NH
17
1.5
0.53
0.32
0.02
23
2.1
1.4
0.40
0.27
11
12
13
14
F/N
a
Ki values were obtained from purified human enzyme and a synthetic peptide substrate at 37 °C (except for 8 at 25 °C) and were averaged
from multiple determinations (n = 2), as described in Ref. 13.
b
aPTT (activated partial thromboplastin time) in vitro clotting assay was performed in human plasma.
over a panel of relevant serine proteases (Trypsin Ki = 670 nM,
FVIIa Ki = 1100 nM, FXa Ki = 1400 nM, FIIa Ki > 13,000 nM). The
in vitro anticoagulant activity was also slightly improved com-
pared to 13 (aPTT1.5x = 0.27 mM, Table 2). Importantly, comparing
macrocycle 14 with acyclic phenylimidazole 1, the macrocyclic
analog improved FXIa affinity by 290-fold and aPTT potency by
19-fold.
An X-ray structure of 14 in the FXIa active site was obtained
(Fig. 3A, 2.35 Å, PBD ID = 5W86). The general binding mode is sim-
ilar to 1. A hydrogen bond is observed between the NH of the
amide with the carbonyl of Leu41 as designed. Interestingly, the
fluorine makes van der Waals contact with Tyr58b C(a) instead
of interacting with the Cys42-Cys58 disulfide bridge as initially
anticipated based on the design (Fig. 3B). The difference in potency
between 12 and 14 can be rationalized by examining the ligand
strain. The macrocycle in general is a highly strained system and
almost all atoms have intramolecular van der Waals contact. By
replacing the phenyl in 12 with a pyridine in 14, some of the ligand
strain is reduced, possibly contributing to the improvement in
potency. Conformational analysis on the macrocycles supports this
theory in that twice as many conformations were observed for 14
compared to 12.
Synthesis of the macrocycles with a 3-atom amide linker is
exemplified in Scheme 1. Condensation of bromomethylketone
1512 with Boc-
L-phenylalanine 16 followed by imidazole formation
gave phenylimidazole 17. Compound 17 was treated with zinc (II)
cyanide under microwave radiation to give 18. Saponification of 18
followed by hydrogenation afforded 19. Macrolactamization of 19
under dilute conditions with BOP and DMAP, followed by Boc
deprotection, provided macrocycle 20. Amide coupling of 20 with
N-hydroxysuccinate ester 2111b produced compound 4.
Chlorination of 4 afforded chloroimidazole analog 5.
The synthesis of the macrocycles with a 4-carbon linker is illus-
trated by the preparation of compounds 8 and 9 (Scheme 2). Com-
pound 23, prepared from 15 and 22 as described previously,
reacted with SEMCl to generate 24. Suzuki coupling of 24 with 2-
allyl-4,4,5,5-tetramethyl-1,3,2-dioxaborolane
afforded
25.
Fig. 3. (A) X-ray crystal structure of 14 bound in the active site of FXIa with omit
electron density (gray mesh). The red spheres depict water molecules, the dotted
lines depict hydrogen bonds, and ethylene diol is an artifact of the flash-cooling
Metathesis of 25 using 2nd generation Grubbs catalyst in the pres-
ence of pTsOH and high dilution conditions yielded macrocycle 26
as a mixture of E and Z isomers. Hydrogenation of 26, followed by
procedure. (B) van der Waals contact between Tyr58b C(
pyridine ring.
a) and the fluorine on the

C. Wang et al. / Bioorganic & Medicinal Chemistry Letters 27 (2017) 4056–4060
4059
Scheme 1. Reagents and conditions: (a) KHCO3, DMF, 0 °C to rt; (b) NH4OAc, xylene, reflux (Dean-Stark trap), 3 h, 53% over two steps; (c) Zn(CN)2, Pd(PPh3)4, DMF, 140 °C,
microwave, 65%; (d) LiOH, THF/water, 100%; (e) Raney Ni, H2, water-NH4OH (2:1), 90%; (f) BOP, DMAP, DIEA, DMF/DCM, syringe pump slow addition, 90%; (g) TFA/DCM; (h)
DIEA, DMF, 43% over two steps; (i) NCS, MeCN, TEA, 0–10 °C, 42%.
Scheme 2. Reagents and conditions: (a) KHCO3, DMF, 0 °C to rt; (b) NH4OAc, xylene, reflux (Dean-Stark), 3 h, 45% over two steps; (c) NaH, THF, SEMCl, 78%; (d) 2-allyl-4,4,5,5-
tetramethyl-1,3,2-dioxaborolane, Pd(dppf)Cl2, K2CO3, THF, 80 °C, 16 h, 31%; (e) pTsOH, DCM (1 mM), reflux, 2nd generation Grubbs catalyst, 34%; (f) H2, Pd/C, MeOH, 91%; (g)
5 M HCl (aq), MeOH, 75 °C, 4 h, 90%; (h) DIEA, DMF, 38%; (i) NCS, MeCN, DIEA, 0–5 °C, 67%.
removal of the Boc and SEM groups, afforded amine 27. Amide for-
mation provided compound 8, and chlorination of 8 yielded com-
pound 9.
intermediate, which was protected with Boc anhydride to yield
35. Reaction of 35 with 31 followed by imidazole formation as
described previously resulted in 36. Compound 36 was coupled
with methyl acrylate to give 37. Hydrogenation and saponification,
followed by BOP-mediated macrolactamization under dilute con-
ditions provided macrocycle 38. Boc deprotection and subsequent
amide formation, followed by chiral separation afforded 14.
In summary, we have successfully designed and optimized a
series of macrocycles with high FXIa affinity and excellent selectiv-
ity using structure-based drug design. Compound 14 achieved
picomolar FXIa affinity and potent in vitro anticoagulant activity.
Compound 14 was prepared as outlined in Scheme 3. 4-Iodo-3-
nitroaniline (28) was converted to methylcarbamate 29 with
methylchloroformate. Reaction of 29 with tributyl(1-ethoxyvinyl)
stannane afforded 30. Bromination of 30 with NBS gave bro-
momethylketone 31. 2-Bromo-3-fluoro-6-methylpyridine (32)
was brominated with NBS to give benzyl bromide 33. Treatment
of 33 and diethyl 2-acetamidomalonate with NaH afforded 34.
Hydrolysis followed by decarboxylation afforded an amino acid

4060
C. Wang et al. / Bioorganic & Medicinal Chemistry Letters 27 (2017) 4056–4060
Scheme 3. Reagents and conditions: (a) Methyl chloroformate, pyridine, DCM, 96%; (b) Tributyl(1-ethoxyvinyl) stannane, PdCl2(PPh3)2, PhMe, reflux, 58%; (c) NBS, THF/
water, 99%; (d) NBS/AIBN, CCl4, reflux; (e) diethyl 2-acetamidomalonate, NaH, DMF, 50% over two steps; (f) HCl (6 N aq.), dioxane, reflux; (g) (Boc)2O, NaOH (1 N aq.), dioxane,
88% over two steps; (h) KHCO3, DMF, 0 °C to rt, 77%; (i) NH4OAc, xylene, reflux (Dean-Stark), 3 h, 12%; (j) methyl acrylate, Pd(OAc)2, tri-tolyphosphine, DIEA, MeCN,
microwave, 7–10%; (k) H2, Pd/C, EtOAc/MeOH, 99%; (l) LiOH, THF/water, 95%; (m) BOP, DMAP, DIEA, DMF/DCM, syringe pump slow addition, 72%; (n) TFA/DCM; (o) DIEA,
DMF, 0 °C to rt, 64% over two steps; (p) Chiral HPLC separation: Chiralcel OD-H, 70% (1:1 MeOH/EtOH with 0.1% DEA in heptane).
3. (a) Ansell J, Hirsh J, Hylek E, Jacobson A, Crowther M, Palareti G. Chest.
2008;133:160S–198S;
mization of other parameters to achieve oral bioavailability which
(b) Stein PD, Grandison D, Hua TA. Postgrad Med J. 1994;70:S72–S83;
The potent FXIa affinity of these compounds enables further opti-
will be reported in due course.
(b) Hirsh J, Poller L. Arch Int Med. 1994;154:282–288.
4. (a) Sharma M, Cornelius VR, Patel JP, Davies JG, Molokhia M. Circulation.
2015;132:194–204;
Acknowledgments
(b) Ruff CT, Giugliano RP, Braunwald E, et al. Lancet. 2014;383:955–962.
5. (a) Emsley J, McEwan PA, Gailani D. Blood. 2010;115:2569–2577;
(b) Gailani D, Renne T. Arterioscler Thromb Vasc Biol. 2007;27:2507–2513;
(c) von dem Borne PA, Meijers JCM, Bouma BN. Blood. 1995;86:3035–3042.
6. Bane CE, Neff AT, Gailani D. Factor XI deficiency or hemophilia C. In: Saba HI,
Roberts HR, eds. Hemostasis and Thrombosis: Practical Guidelines in Clinical
Management. Oxford: John Wiley & Sons Ltd; 2014:71–81.
The authors thank the Department of Discovery Synthesis for
the synthesis of intermediates at BMS Biocon Research Center
and the Department of Lead Discovery
in vitro work.
& Optimization for
7. (a) Salomon O, Steinberg DM, Koren-Morag N, Tanne D, Seligsohn U. Blood.
2008;111:4113–4117;
A. Supplementary data
(b) Salomon O, Steinberg DM, Dardik R, et al.
2003;1:658–661;
J Thromb Haemost.
(c) Salomon O, Steinberg DM, Zucker M, Varon D, Zivelin A, Seligsohn U. Thromb
Haemost. 2011;105:269–273.
8. (a) Doggen CJM, Rosendaal FR, Meijers JCM. Blood. 2006;108:4045–4051;
(b) Meijers JCM, Tekelenburg WLH, Bouma BN, Bertina RM, Rosendaal FR. N Engl
J Med. 2000;342:696–701.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2017.07.
048.
9. Wong PC, Quan ML, Watson CA, et al. Thromb Thrombolysis. 2015;40:416–423.
10. Buller HR, Bethune C, Bhanot S, et al. N Engl J Med. 2015;372:232–240.
11. (a) Hangeland JJ, Friends TJ, Rossi KA, et al. J Med Chem. 2014;57:9915;
(b) Pinto DJ, Smallheer JM, Corte JR, et al. Bioorg Med Chem Lett. 2015;25:1635;
(c) Hu Z, Wong PC, Gilligan PJ, et al. ACS Med Chem Lett. 2015;6:590–595.
12. Corte JR, Fang T, Osuna H, et al. J Med Chem. 2017;60:1060–1075.
13. Quan ML, Wong PC, Wang C, et al. J Med Chem. 2014;57:955–969.
References
1. Reddy KS. Lancet. 2016;388:1459–1544.
2. (a) Morais J, De Caterina R. Cardiovasc Drugs Ther. 2016;30:201–214;
(b) Schaefer JK, McBane RD, Wysokinski WE. Ann Hematol. 2016;95:437–449;
(c) Levy JH, Spyropoulos AC, Samama CM, Douketis JJ. Am Coll Cardiol Intv.
2014;7:1333–1351.