Structure-Based Design and Preclinical Characterization of Selective and Orally Bioavailable Factor XIa Inhibitors: Demonstrating the Power of an Integrated S1 Protease Family Approach

Article abstract of DOI:10.1021/acs.jmedchem.0c00279

The serine protease factor XI (FXI) is a prominent drug target as it holds promise to deliver efficacious anticoagulation without an enhanced risk of major bleeds. Several efforts have been described targeting the active form of the enzyme, FXIa. Herein, we disclose our efforts to identify potent, selective, and orally bioavailable inhibitors of FXIa. Compound 1, identified from a diverse library of internal serine protease inhibitors, was originally designed as a complement factor D inhibitor and exhibited submicromolar FXIa activity and an encouraging absorption, distribution, metabolism, and excretion (ADME) profile while being devoid of a peptidomimetic architecture. Optimization of interactions in the S1, S1β, and S1′ pockets of FXIa through a combination of structure-based drug design and traditional medicinal chemistry led to the discovery of compound 23 with subnanomolar potency on FXIa, enhanced selectivity over other coagulation proteases, and a preclinical pharmacokinetics (PK) profile consistent with bid dosing in patients.

Full text of DOI:10.1021/acs.jmedchem.0c00279

Subscriber access provided by ECU Libraries
Article
Structure-based design and pre-clinical characterization of selective
and orally bioavailable Factor XIa inhibitors: Demonstrating
the power of an integrated S1 protease family approach.
Edwige Lorthiois, James Roache, David Barnes-Seeman, Eva Altmann, Ulrich Hassiepen, Gordon
Turner, Rohit Duvadie, Viktor Hornak, Rajeshri G. Karki, Nikolaus Schiering, Wilhelm A. Weihofen,
Francesca Perruccio, Amy Calhoun, Tanzina Fazal, Darija Dedic, Corinne Durand, Solene Dussauge,
Kamal Fettis, Fabien Tritsch, Celine Dentel, Adelaide Druet, Donglei Liu, Louise Kirman, Julie Lachal,
Kenji Namoto, Douglas Bevan, Rose Mo, Gabriela Monnet, Lionel Muller, Richard Zessis, Xueming
Huang, Loren Lindsley, Treeve Currie, Yu-Hsin Chiu, Cary Fridrich, Peter Delgado, Shuangxi Wang,
Micah Hollis-Symynkywicz, Joerg Berghausen, Eric Williams, Hong Liu, Guiqing Liang, Hyungchul Kim,
Peter Hoffmann, Andreas Hein, Paul Ramage, Allan D'Arcy, Stefanie Harlfinger, Martin Renatus, Simon
Ruedisser, David Feldman, Jason Elliott, Richard C Sedrani, Jürgen Maibaum, and Christopher M. Adams
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.0c00279 • Publication Date (Web): 17 Jun 2020
Downloaded from pubs.acs.org on June 17, 2020
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
RESEARCH EUROPE
Namoto, Kenji; Novartis Institutes for BioMedical Research Basel,
RESEARCH EUROPE
Bevan, Douglas; Novartis Pharmaceuticals Corp
Mo, Rose; Novartis Institutes for BioMedical Research Inc, Global
Discovery Chemistry
Monnet, Gabriela; Novartis Institutes for BioMedical Research Basel,
RESEARCH EUROPE
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Muller, Lionel; Novartis Institutes for BioMedical Research Basel,
RESEARCH EUROPE
Zessis, Richard; Novartis Institutes for BioMedical Research Inc
Huang, Xueming; Novartis Institutes for BioMedical Research Inc
Lindsley, Loren; Novartis Institutes for BioMedical Research Inc
Currie, Treeve; Novartis Institutes for BioMedical Research Inc
Chiu, Yu-Hsin; Novartis Institutes for BioMedical Research Inc
Fridrich, Cary; Relay Therapeutics Inc
Delgado, Peter; Novartis Institutes for BioMedical Research Inc, Global
Discovery Chemistry
Wang, Shuangxi; Novartis Institutes for BioMedical Research Inc
Hollis-Symynkywicz, Micah; Novartis Institutes for BioMedical Research
Inc
Berghausen, Joerg; Novartis Pharma AG,
Williams, Eric; Molecular Templates Inc
Liu, Hong; Novartis Institutes for BioMedical Research Inc
Liang, Guiqing; Novartis Inst BioMed Res, Metabolism and
Pharmacokinetics
Kim, Hyungchul; Novartis Institutes for BioMedical Research Inc
Hoffmann, Peter; Novartis Institutes for BioMedical Research Inc
Hein, Andreas; Novartis Institutes for BioMedical Research Basel,
Expertise Protease Platform
Ramage, Paul; Novartis Institutes for BioMedical Research Basel
D'Arcy, Allan ; Novartis Institutes for BioMedical Research Basel,
RESEARCH EUROPE
Harlfinger, Stefanie; AstraZeneca PLC, Oncology R&D, DMPK
Renatus, Martin ; Novartis Institutes for BioMedical Research, RESEARCH
EUROPE
Ruedisser, Simon; ETH Zürich, BioNMR platform D- BIOL
Feldman, David ; National Kidney Foundation
Elliott, Jason; Novartis Institutes for BioMedical Research Inc
Sedrani, Richard; Novartis Pharma AG, Novartis Inst. for BioMedical
Research
Maibaum, Jürgen; Novartis Pharma AG, Novartis Institutes of Biomedical
Research
Adams, Christopher; Novartis Institutes for BioMedical Research Inc,
Global Discovery Chemistry
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 133
1
2
3
4
5
6
Structure-based design and pre-clinical
7
8
9
characterization of selective and orally bioavailable
Factor XIa inhibitors: Demonstrating the power of
an integrated S1 protease family approach.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Edwige Lorthiois‡*, James Roache†*, David Barnes-Seeman†, Eva Altmann‡, Ulrich Hassiepen‡,
Gordon Turner†, Rohit Duvadie†, Viktor Hornak†, Rajeshri G. Karki†, Nikolaus Schiering‡,
Wilhelm A. Weihofen†, Francesca Perruccio‡, Amy Calhoun†, Tanzina Fazal†, Darija Dedic‡,
Corinne Durand‡, Solene Dussauge‡, Kamal Fettis‡,¶, Fabien Tritsch‡, Celine Dentel‡, Adelaide
Druet‡, Donglei Liu†, Louise Kirman†,ʆ, Julie Lachal‡, Kenji Namoto‡, Douglas Bevan†, Rose Mo†,
Gabriela Monnet‡, Lionel Muller‡, Richard Zessis†, Xueming Huang†, Loren Lindsley†, Treeve
Currie†, Yu-Hsin Chiu†, Cary Fridrich†,ʈ, Peter Delgado†, Shuangxi Wang†, Micah Hollis-
Symynkywicz†, Joerg Berghausen‡, Eric Williams†,ʅ, Hong Liu†, Guiqing Liang†, Hyungchul Kim†,
Peter Hoffmann†, Andreas Hein‡, Paul Ramage‡, Allan D'Arcy‡, Stefanie Harlfinger‡,$, Martin
Renatus‡, Simon Ruedisser‡,%, David Feldman&,!, Jason Elliott†, Richard Sedrani‡, Juergen
Maibaum‡, and Christopher M. Adams†
†Novartis Institutes for BioMedical Research, Cambridge, Massachusetts 02139, USA
ACS Paragon Plus Environment
1

Page 3 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
‡Novartis Institutes for BioMedical Research, Novartis Campus, CH-4056 Basel,
Switzerland.
7
8
&Novartis Institutes for BioMedical Research, East Hanover, NJ 07396, USA.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
KEYWORDS. Coagulation, intrinsic pathway, factor XIa inhibitors, aPTT, PT, S1
protease, serine protease, structure-based drug design, Factor XI.
ABSTRACT:
OH
N
O
O
O
OH
OH
H2N
O
NH2
1
23
FXIa, IC50 0.0004
2×aPTT/PT ratio = 21
S1 protease library Hit
FXIa, IC50 0.63
M

M

The serine protease Factor XI (FXI) is a prominent drug target as it holds promise
to deliver efficacious anti-coagulation without an enhanced risk of major bleeds. Several
efforts have been described targeting the active form of the enzyme, FXIa. Herein we
disclose our efforts to identify potent, selective, and orally bioavailable inhibitors of FXIa.
Compound 1, identified from a diverse library of internal serine protease inhibitors, was
originally designed as a complement Factor D inhibitor and exhibited sub-micromolar
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 133
1
2
3
4
5
6
7
8
9
FXIa activity and an encouraging ADME profile while being devoid of peptidomimetic
architecture. Optimization of interactions in the S1, S1β, and S1` pockets of FXIa through
a combination of structure-based drug design and traditional medicinal chemistry led to
the discovery of compound 23 with sub-nanomolar potency on FXIa, enhanced selectivity
over other coagulation proteases, and a pre-clinical PK profile consistent with bid dosing
in patients.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
INTRODUCTION
The discovery of the anti-coagulant properties of dicoumarol in 1940, and
subsequent development of warfarin, transformed the practice of medicine1 offering
patients at risk of thrombosis a prophylactic therapeutic option.2 However, warfarin’s
clinical success is tempered by a narrow therapeutic index due to its mechanism of action
(vitamin K antagonism), which poses a risk of major life threatening bleeding.3 This is
further compounded by drug-drug interactions and dietary interactions which require dose
titration and frequent laboratory monitoring.4 The advent of non-vitamin K anti-coagulants
(NOACs) have offered patients an alternative to warfarin. Molecules selectively inhibiting
the proteases Factor Xa (FXa) (e.g. rivaroxaban5 and apixaban6) or thrombin (Factor II,
ACS Paragon Plus Environment
3

Page 5 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
FII) (e.g. dabigatran7), which play a pivotal role in coagulation homeostasis, have gained
clinical acceptance over the last decade, in part due to simpler dosing paradigms and a
lower risk of bleeding.8,9 However, NOACs are still associated with substantial risk of
major bleeding events limiting their widespread use.10,11
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
More recently, the serine protease Factor XI (FXI) has come to prominence as a
drug target offering the potential of providing efficacious anti-coagulation without an
enhanced risk of major bleeds.12 The activated form, FXIa, has a critical function in the
amplification of thrombin generation in a low tissue-factor environment by directly
activating FIX without intrinsic cascade activation.13 This hypothesis is supported by
human data, as individuals deficient in FXI activity have reduced risk of thromboembolic
events without an increase in major bleeding events.14 Furthermore, a recent clinical trial
with an antisense oligonucleotide targeting FXI proved efficacious in preventing venous
thrombosis in patients undergoing total knee arthroplasty.15 Importantly, in this trial there
was also a trend toward a reduction in bleeding events versus enoxaparin, a heparin
mimetic. In light of this strong clinical rationale, several efforts have been disclosed
targeting FXIa with low molecular weight compounds and biologics.16,17,18,19 These efforts
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 133
1
2
3
4
5
6
7
8
have afforded several linear and macrocyclic scaffolds targeting the enzyme catalytic site
and most recently culminated in the discovery of first clinical candidates.20,21,22,23
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Herein we disclose our efforts to identify potent and selective inhibitors of FXIa by
leveraging the power of a platform drug discovery approach to target serine proteases.24
At the outset of this program, we decided to focus exclusively on demonstrating in vitro
efficacy and selectivity in combination with high in vivo oral bioavailability for candidate
FXIa inhibitors. These criteria included: 1) sufficient potency and pharmacokinetic
properties to maintain >85% inhibition of FXIa, in line with genetic evidence that correlates
≥85% FXI deficiency to thromboprotection;14 and 2) having sufficient selectivity over other
proteases of the coagulation cascade to mitigate the potential risk of bleeding.25 With
regard to assessing FXI inhibition and selectivity, we chose to employ a combination of
high throughput enzymatic assays and clinically translatable human plasma coagulation
assays. The activated partial thromboplastin time (aPTT) and prothrombin time (PT)
assays provided the primary means of assessing the functional potency/efficacy and
functional selectivity, respectively, of our FXIa inhibitors.26 These assays measure the
ACS Paragon Plus Environment
5

Page 7 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
time it takes for plasma to coagulate when the intrinsic (aPTT assay) or extrinsic (PT
assay) pathway is stimulated by an exogenous trigger (Figure 1). We relied exclusively
on the in vitro assesments of efficacy since the translatability of FXIa inhibition from
preclinical models to the clinic has not been established. Herein we show the results of
our lead optimization efforts leading to a selective and orally bioavailable FXIa inhibitor
that prolonged aPTT without a perturbation in the PT assay.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1: Schematic representation of the coagulation cascade. The aPTT assay
assesses inhibition of the intrinsic (blue) and progression (red) components of the
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 133
1
2
3
4
5
6
pathway, while the PT assay assesses the extrinsic (orange) and progression
components.
7
8
9
RESULTS AND DISCUSSION
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Endeavoring for an accelerated hit finding campaign to identify promising starting
points with a high potential for oral bioavailability, we took advantage of a combination of
high throughput screening (HTS), virtual screening, and NMR-mediated fragment based
screening (FBS) approaches. We also embarked on a more focused screen of a
knowledge-based diversity library (~1750 compounds) of internal serine protease
inhibitors and analogs thereof. This compound collection had been built as part of a
protease drug discovery platform at Novartis.24 While the HTS and FBS efforts delivered
several hits that were initially taken forward for SAR exploration, we eventually became
most intrigued by the activity of compound 127 (Figure 2), which demonstrated an IC50 of
0.63 M against human FXIa, and was structurally diverse from known FXIa inhibitors.
This scaffold had originally been discovered as part of our efforts to identify inhibitors of
complement Factor D (FD),27,28,29 resulting from a dedicated structure-based design
approach tackling a unique active-site architecture of an atypical trypsin-like S1 protease.
ACS Paragon Plus Environment
7

Page 9 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Compound 1 appeared to be particularly attractive due to its structural simplicity, low
molecular weight, and a balanced lipophilicity/polarity ratio (logD6.8 = 1.91). These
properties translated into encouraging ADME properties and an excellent mouse
pharmacokinetic profile.27 Furthermore, the zwitterionic character of 1 provided a very
low-affinity profile toward the hERG ion-channel with IC50 values >30 M in both the
dofetilide binding and the functional Q-patch assays. However, the potency of 1 against
FD (IC50: 0.008 µM) and the relatively moderate selectivity against other trypsin-like S1
proteases of the coagulation cascade (e.g. plasma kallikrein (PKL), IC50: 3.1 µM; plasmin,
IC50: 2.2 µM; FVIIa, IC50: 1.4 µM and urokinase, IC50: 0.77 µM) needed to be addressed
in addition to optimizing the potency against FXIa. Hence, the most prominent question
raised was whether the design of a potent and highly selective FXIa inhibitor would be
feasible at all by starting from a low-nanomolar FD inhibitor scaffold.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 133
1
2
3
4
O
O
Cl
5
6
7
N
N
O
O
O
H
H
O
N
HO2C
HO2C
HO2C
HO2C
8
9
H2N
H2N
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
NH2
NH2
HN
O
O
2
5 (rac.)
1
3
4 (dia. mix.)
FXIa, IC50: 0.97 M
PKL, IC50: 1.2 M
FXIa, IC50: 2.8 M
FD, IC50: 0.12 M
FXIa, IC50: 0.20 M
FD, IC50: 0.34 M
PKL, IC50: 1.3 M
FXIa, IC50: 0.63 M
FD, IC50: 0.008 M
FXIa, IC50: 0.40 M
FD, IC50: 0.80 M
Figure 2. Chemical structures of early-stage S1-benzylamine-based FXIa inhibitors.
With the aim of developing a rational design concept toward improved FXIa
selectivity, we resolved the co-crystal structure of a closely related analog, 2 (FXIa IC50:
2.8 M), bearing a carboxamide spacer instead of the ether linker in compound 1. The
overall binding mode of 2 in FXIa is similar to that observed in FD27 (Figure 3) in that it
spans the S1 and S1’ pockets in a U-shape conformation. The basic primary amine is
within H-bonding distance to the Ala190 backbone carbonyl and forms a salt bridge with
the side-chain of Asp189. The biphenyl portion of the molecule is sandwiched between the
Lys192 side-chain and Trp215 backbone and the carboxylic acid moiety binds to the
oxyanion hole making H-bonding interactions with the side-chain of His57 and the
backbone NH of Gly193. The H-bonding interactions with Ser195 are not visible in the X-
ray structure as a FXIa active site mutant (S195A mutant; chymotrypsinogen numbering
ACS Paragon Plus Environment
9

Page 11 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
has been used throughout this paper) was used to obtain this co-crystal structure. The
absence of the OH in the mutant protein does not have an influence on the position of the
carboxylic acid based on comparison to related compounds in Ser195 FXIa-containing
crystals. The phenyl ring fits nicely in the S1’ pocket and forms an edge-to-face interaction
with His57. Several X-ray structures of FXIa and inhibitor complexes, which were not
available at the time of our work, have since been reported with a phenyl ring binding in
a similar fashion in the S1’ pocket of the FXIa active site.16
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Binding mode of compound 2: (a) Crystal structure of 2 (yellow, PDB code 6TS4)
with FXIa S195A mutant (green). The surface of the protein is shown in gray and oxygen
atoms of crystallographic water molecules are shown as red spheres. (b) Comparison of
the crystal structures of compound 2 with FXIa (yellow and green carbons) and FD (pale
blue, PDB code 6QMT).27 Only selected residues in the ligand-binding pocket are shown.
ACS Paragon Plus Environment
10

Journal of Medicinal Chemistry
Page 12 of 133
1
2
3
4
5
6
FXIa and FD residues are labeled in black and in blue, respectively. H-bonding
interactions are shown as black dotted lines.
7
8
9
Most importantly, the comparison between the FXIa16 and the FD30 inhibitor
binding sites revealed several opportunities to concomitantly improve FXIa potency and
FD selectivity by specifically targeting the deep S1 pocket, the S1β pocket and/or the
prime site of FXIa. With regard to the S1 pocket, trypsin-like proteases can be
differentiated based on the nature of the amino-acid 190 at the bottom of the S1 pocket,
which dictates enzyme specificity. FXIa and thrombin display an alanine in position 190
while FD, urokinase and FVIIa bear the larger, more polar, serine residue at this position.
As a consequence, the latter enzymes show substrate-specificity for the less sterically
demanding P1 lysine residue, while FXIa prefers arginine.31 This difference has been
successfully exploited to enhance selectivity of various trypsin-like protease inhibitors.32,33
Besides the S1 pocket, the small hydrophobic S1β pocket located on top of the Cys191-
Cys219 disulfide bridge was also considered to offer another opportunity for improving
FXIa affinity and specificity, since the S1β pocket of FXIa is limited in size due to the side-
chain of Leu146. A literature survey of known inhibitors of serine proteases indicated that
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
11

Page 13 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
binding in this region could contribute to improved ligand binding affinity.29,34,35 Further
extension into the S2’ site, for which significant structural differences are observed
between FXIa and off-target serine proteases, offers another option to enhance
selectivity. In particular, FXIa exhibits a comparatively more open S1’-S2’ binding region
versus FD.31 However, accessing the S2’ pocket would require increasing the size of the
molecule significantly, and therefore this strategy was initially deferred.36
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
A final key aspect defining our medicinal chemistry strategy related to the the nature of
S1-S1’ linker. At the outset of our work, it was argued that the basic amine of 1, positioned in S1,
and the carboxylic acid are engaged in key interactions with Asp189 and the canonical oxyanion
hole (residues Gly193 and Ser195), respectively, and therefore are unlikely to evoke differentiating
binding interactions on their own for FXIa versus FD and other proteases.37 However, the
appropriate spacing and alignment of these two motifs might offer an opportunity to enhance
potency and selectivity and to modulate ADME properties. Of equal importance was the notion
that the zwitterionic motif of this scaffold provided an advantageous in vitro and in vivo ADME
profile, in particular, when retaining the less polar ether linker of compound 1.27
Our initial efforts focused on optimizing the binding interactions to the deep and
solvent-shielded S1 specificity pocket, in particular, by growing toward Ala190 which was
expected to enhance binding to FXIa and at the same time to be less well accommodated
by the smaller S1 pocket of Ser190 S1 family serine proteases including FD. In order to
ACS Paragon Plus Environment
12

Journal of Medicinal Chemistry
Page 14 of 133
1
2
3
4
5
6
7
8
9
probe the impact of the larger FXIa S1 pocket, compound 3 (Figure 2) was prepared, in
which the benzylamine motif was replaced by the rigid bicyclic tetrahydroisoquinoline,
thereby mimicking the FXIa-bound conformation of the flexible aminomethyl group in 2.
Compound 3 displayed similar potency against FXIa as compound 1, and the inhibitory
activity toward FD was reduced by 100-fold (Figure 2). To confirm our design rationale,
the X-ray crystal structure of 3 bound to FXIa with the native catalytic triad, was resolved
(Figure 4). The binding mode spanning S1, S1’ and the oxyanion hole is very similar as
observed for FXIa-bound 2 (Figure 3). The basic NH motif makes H-bonding interactions
with the Ala190 backbone carbonyl and the side-chain of Asp189 similar to that observed in
the 2-FXIa complex.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
13

Page 15 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 4. (a) Crystal structure of compound 3 (yellow) complexed with FXIa (white, PDB
code 6TS7) (b) Overlay of the crystal structure of compound 3 (yellow) complexed with
FXIa (white) on compound 4 (cyan) complexed with PKL (pale blue, PDB code 6T7P).
Only selected residues in the ligand-binding pocket are shown. H-bonding interactions
are shown as black dotted lines.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Encouraged by this result, we undertook a broad search of P1 moieties by close
inspection of published and in-house co-crystal structures of S1-binding fragments and
S1 protease inhibitors. Overlaying the FXIa-2 and -3 crystal structures with compound 4,
derived from our knowledge-based diversity library, bound to PKL,38 a related trypsin-like
Ala190 S1 protease, proved particularly fruitful. The aminodihydrobenzofuran of 4 (Figure
2) overlapped with the S1 binding phenyl and basic amine motifs from both inhibitors
(Figure 4b). Compound 4 had been co-crystallized with PKL as a mixture of (R)- and (S)-
2,3-dihydrobenzofuranyl-3-amine diastereoisomers. The electron density of the bound
ligand was consistent with the absolute (S)-configuration of the 3-
aminodihydrobenzofuran.39 Racemic 5 (Figure 2), which combines the two key
pharmacophores of 4 and 3, was prepared and showed similar inhibitory affinity for both
FXIa and FD as compared to 3. Compound 5 was also found to be a weak inhibitor of
PKL, albeit with a 5-fold lower IC50 value of 1.3 M versus FXIa (IC50 of 0.2 M). During
ACS Paragon Plus Environment
14

Journal of Medicinal Chemistry
Page 16 of 133
1
2
3
4
5
6
7
8
9
our continued optimization efforts, we preferred both the tetrahydroisoquinoline and the
3-aminodihydrobenzofuran S1 binding motifs as more tractable replacements of the
benzylamine moiety of the screening hit 1, as these scaffolds offered the benefit of a lower
risk for semicarbazide-sensitive amine oxidase (SSAO) mediated metabolism40 and
potential toxicity.41
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Next we focused our attention on targeting the S1β pocket in FXIa. SAR
exploration of a representative selection of compound 1 analogs from the FD inhibitor
program27,42 in the FXIa biochemical assay indicated a preference for small and branched
aliphatic substituents. This finding was in line with the limited size of the FXIa S1β pocket
due to the conformational fold of the Leu146 side-chain. Intriguingly, incorporation of an
isopropyl substituent into the meta-position of the middle phenyl ring at the edge of the
S1β site resulted in a significant 65-fold potency increase for compound 6 bearing the P1
(S)-2,3-dihydrobenzofuranyl-3-amine moiety as single stereoisomer.43 Inhibitor 6
displayed single-digit nanomolar potency toward FXIa with an IC50 value of 0.003 M in
the biochemical assay (Table 1), however a 40-fold drop in potency was observed for 6
ACS Paragon Plus Environment
15

Page 17 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
when measured in the presence of 50% human plasma (cd-FXIa IC50: 0.12 M; Table
1).44 The potency of 6 against FD was also increased, leading to a 10-fold selectivity for
FXIa over FD. There was no improvement in PKL selectivity due to the high homology
between FXIa and PKL. In previous reports,17 achieving selectivity toward PKL has
proven to be difficult and, to our knowledge, only the large FXIa inhibitor BMS-962212
lacking oral bioavailability has demonstrated a significant selectivity against PKL (~400-
fold).20 Among S1 proteases PKL is an exceptionally close homolog of FXI, indicating a
recent divergence from a common ancestor.45 Selectivity over PKL was not considered a
necessary attribute as human congenital deficiency in prekallikrein (PK), the inactive
precursor of PKL, or PK-deficient mice do not encounter an increase in clinically relevant
bleeding.46
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1. In vitro FXIa potency (IC50), selectivity data versus FD and PKL and in vitro
profiles.
ACS Paragon Plus Environment
16

Journal of Medicinal Chemistry
Page 18 of 133
1
2
3
4
5
6
O
R1
O
O
OH
OH
7
O
O
8
9
H2N
H2N
H2N
O
O
HN
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
O
11
1
1
12
10
(R = C(CH3)2OH)
6
7
8
9
(R = CH(CH3)2)
(R1= C(CH3)2CN)
(R1= C(CH3)2OH)
(R1= C(CH3)2CONH2)
Compound FXIa IC50
cdFXIa
FD IC50
PKL IC50
[µM]d
Caco-2
HT-eq solubility
s
[µM]a
plasma IC50 [µM]c
[µM]b
Papp(AB)/(BA) x10-6 pH 6.8 (mM)e
(cm/s) (ratio
(BA)/(AB))
6
7
0.003
0.004
0.004
0.009
0.10
0.12
0.05
0.17
0.067
0.43
0.085
0.26
0.04
0.04
0.32
0.32
0.23
0.05
15
0.003
0.002
0.006
0.011
0.33
7/ 17 (2.3)
1.4/ 41 (30)
BLQf/ 19 (-)
BLQf/ 4 (-)
NDg
0.008
0.06
>1
8
9
0.78
NDg
10
11
12
0.003
0.006
0.007
0.004
12/ 12 (1)
NDg
<0.004
NDg
Half-maximal inhibition of apurified human FXIa and b recombinant catalytic domain of
human FXIa as determined in a fluorogenic rhodamine-110 based peptide cleavage
assay. cHalf-maximal inhibition of recombinant human complement FD as determined in
d
a TR-FRET assay. Half-maximal inhibition of PKL as determined in a fluorogenic
rhodamine-110-based peptide cleavage assay. Data represent geometric mean values
of multiple measurements (n≥2). eEquilibrium high throughput solubility in aqueous buffer
f
g
(pH 6.8). BLQ = below limit of quantitation. ND = not determined.
ACS Paragon Plus Environment
17

Page 19 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The improvement in biochemical potency and selectivity against FD demonstrated
by 6 warranted assessment in the aPTT and PT coagulation assays. In our hands, a two-
fold increase in aPTT coagulation time correlated with ~95% inhibition of FXIa activity in
plasma, surpassing our threshold of 85%.47 Hence, we evaluated 6 and subsequent
analogs on the concentration necessary to effect a doubling of coagulation time assessed
by aPTT (2×aPTT). For comparison, we also evaluated the inhibitor concentration
required to double prothrombin time (2×PT) as a measure of selectivity.48 We aimed for
compounds that could achieve 2×aPTT at total plasma concentrations ≤1.5 µM. These
criteria were based on the assumption that identifying a compound with a
pharmacokinetic profile enabling Ctrough concentrations of >1.5 µM would be exceedingly
difficult. For the PT selectivity assessment, we aimed for a ≥20-fold difference between
aPTT and PT doubling concentrations with the assumption that such a margin should
enable functional selectivity at both peak and trough plasma concentrations when paired
with a suitable PK profile. Inhibitor 6 afforded a 2×aPTT value of 2.2 M and a 2×PT value
of 9.9 M resulting in a 2×aPTT/PT ratio of only about 4-fold. Notably, the significant PT
activity of this compound suggested that off-target activity against other coagulation
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
18

Journal of Medicinal Chemistry
Page 20 of 133
1
2
3
4
5
6
7
8
9
cascade proteases was likely influencing the aPTT assessment as well resulting in an
overestimation of FXIa-driven potency in the latter assay (vide infra; Table 3).
Despite the suboptimal functional selectivity observed with 6, we wanted to
investigate how these structural changes relative to compound 127 impacted ADME and
PK properties (Table 2). Gratifyingly, the favorable ADME properties of compound 1 were
retained and translated to low clearance and excellent oral bioavailability in mice.
Table 2. In vivo mouse pharmacokinetic profiles for compounds 1,27 6, 7, 8 and 11.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compounds
CL (mL·min-1·kg-1 )
VSS (L/kg)
1a
14b
0.7b
6a
7a
8a
11a
4 ± 1
1 ± 0
55 ± 4
19 ± 2
0.4 ± 0.1
5.5 ± 0.2
48927b
100b
0.7 ± 0.1
0.3 ± 0.0
171 ± 146
25 ± 21
137 ± 111
0.4 ± 0.1
1.7 ± 0.2
186 ± 26
3 ± 0
1 ± 0.1
t1/2term (h)
1.6b
6.4 ± 3.4
6525 ± 1508
61 ± 14
AUC p.o. d.n. (nM·h)
%F
3284 ± 138
100 ± 17
2591 ± 36
Cmax d.n. (nM)
5263b
99 ± 10
1126 ± 320
aDiscrete PK profiles, male mice (C57BL/6), dose i.v. 1.0 mg·kg-1, dose p.o. 3.0 mg·kg-
1, except for compound 1 dosed po at 10 mg·kg-1; bNo SD calculated as one animal was
excluded from the dosing group due to experimental issues.
We next explored the interactions to the S1β pocket with the additional aim to
improve solubility and to reduce the potency shift observed with compound 6 in the
presence of plasma (Table 1). To that end, we investigated substitution of the P1β
isopropyl group by various polar residues. Introduction of a nitrile (7), hydroxyl (8), or a
ACS Paragon Plus Environment
19

Page 21 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
primary carboxamide (9) retained the high biochemical potency for FXIa and PKL. For
compound 7, a slight improvement in potency in the 50% plasma assay was observed
(Table 1). Interestingly, biochemical selectivity against FD was also substantially
improved for 8 (40-fold) and 9 (80-fold) but not for 7 (Table 1). Introduction of S1-binding
residues to the P1 tetrahydroisoquinoline scaffold, as exemplified by compound 10,
afforded only a minor 4-fold improvement in potency for FXIa, while inhibitory activity
toward FD was retained as compared to compound 3 (Table 1).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The X-ray crystal structure of 7 in complex with FXIa (Figure 5a) revealed a binding
pose very similar to that observed for compound 3 with a perfect overlap of the respective
biphenyl scaffolds, the ether linkers, and the arylacetic acid motifs of both inhibitors. The
isopropyl nitrile moiety was nicely accommodated by the hydrophobic S1β pocket formed
by the side-chains of Leu146, Tyr143, and Lys192. The 2,3-dihydrobenzofuran-3(S)-amine
moiety closely filled the S1 pocket, with the amine H-bonding with the side-chain of Asp189
and the backbone carbonyl of Ala190 and Gly218.
ACS Paragon Plus Environment
20

Journal of Medicinal Chemistry
Page 22 of 133
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. (a) Crystal structure of compound 7 (yellow, PDB code 6TS6) in complex with
FXIa (white). The surface of the protein is shown in gray and oxygen atoms of
crystallographic water molecules are shown as red spheres. (b) Comparison of the co-
crystal structure of compound 7 (yellow) complexed with FXIa (white) and compound 2
(pale blue) complexed with FD (pale blue, PDB code 6QMT).27 Only selected residues in
the ligand-binding pocket are shown. H-bonding interactions are shown as black dotted
lines. FXIa residues are labeled in black and FD residues are labeled in blue.
We also noted that terminal polar P1 residues, as exemplified by compounds 7
to 9, had a beneficial impact on solubility, but they also proved detrimental to cell
permeability and induced efflux as measured by a Caco-2 assay (Table 1). In addition, in
vivo mouse PK profiles for compounds 7 and 8 indicated increased clearance and
reduced oral bioavailability relative to compound 6 (Table 2). This prompted us to explore
the introduction of locally masked or shielded polarity as part of an additional heterocycle
attached to the upper phenyl ring while keeping the beneficial hydrophobic van der Waals
ACS Paragon Plus Environment
21

Page 23 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
interactions to the S1β pocket. To this end, we synthesized the 3,3-dimethyl-
dihydrobenzofuran analog 11 which exhibited similar low-nanomolar potency toward FXIa
in both the biochemical and in the plasma assay, and also showed a ~17-fold selectivity
over FD (Table 1). This compound demonstrated an improved mouse PK profile, as
compared with 7 and 8, characterized by low clearance and good bioavailability (Table
2). However, a 2×PT value of 19.8 µM in relation to a 2×aPTT value of 2.5 µM (aPTT/PT
ratio of ~8) needed further improvement (vide infra, Table 3). We then explored
modifications of the aliphatic portion of the P1 pharmacophore, which is deeply buried at
the bottom of the S1 recognition site upon binding to FXIa. We reasoned that filling more
tightly the unoccupied space near Ala190 could further improve selectivity versus FD and
other Ser190 proteases. As an example, enlargement of the (S)-2,3-dihydrobenzofuranyl-
3-amine of 11 to the 6-membered ring provided the sterically more demanding (R)-
chromanyl amine 12 (Table 1). Relative to compound 11, compound 12 showed similar
potency toward FXIa in the biochemical assay, but with a 3-fold drop in potency in the
FXIa plasma assay. Despite the insufficient FXIa plasma potency, 12 demonstrated
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
22

Journal of Medicinal Chemistry
Page 24 of 133
1
2
3
4
5
6
7
8
exquisite 2,500-fold selectivity against FD, which was attributed to a steric clash with
Ser190 present in FD.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In order to assess the selectivity profile of our emerging lead FXIa inhibitors, we
tested them against a panel of serine proteases of the coagulation cascade and serine
proteases, which play a role in fibrinolysis, i.e. in breaking down blood clots (e.g. tPa,
urokinase, plasmin). Inhibitors 6 and 11 exhibited acceptable selectivity over FVIIa and
plasmin, but also showed sub-micromolar affinity for FIXa, FXa, thrombin, tissue
plasminogen activator (tPa) and urokinase (Table 3). Compound 12 bearing a larger S1-
binding motif demonstrated improved selectivity against serine proteases FVIIa,
urokinase, FXa and thrombin, and also versus FIXa. However, we considered the
moderate selectivity across the off-target protease panel for these 3 compounds still to
be insufficient, as inhibition of either one specific enzyme, or the additive effect of partial
inhibition of several of these proteases, may negatively affect the risk of bleeding, as
evidenced by the poor 2×aPTT/PT ratios for compounds 6 and 11.
Table 3. In vitro S1 protease selectivity data for selected FXIa inhibitors.
Compounds
6
11
12
ACS Paragon Plus Environment
23

Page 25 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
FVIIa IC50 (µM)a
FIXa IC50 (µM)a
FXa IC50 (µM)a
5.10
0.17
7.1
0.21
0.84
0.13
0.10
6.3
75
1.6
0.50
4.3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Thrombin IC50 (µM)a
tPa IC50 (µM)a
0.090
0.090
2.60
0.86
0.19
14
Plasmin IC50 (µM)a
Urokinase
(µM)a
IC50
0.047
0.016
0.95
aHalf-maximal inhibition of selected proteases as determined in biochemical activity
assays using quenched fluorescent peptide substrates. Data represent mean values of
duplicate measurements.
FXIa inhibitor 6 comprising a very lean zwitterionic structure of low molecular
weight and demonstrating a high inhibitory ligand efficiency for FXIa (BEI49 of 20, LLE50
of 5.4, logD7.4 of 3.1) provided a better understanding of the foundational SAR as it relates
to FXIa and FD activity. Compound 6 also demonstrated a low risk of DDI and hERG
inhibition (CYPs 3A4, 2C9, 2D6: all IC50 >20 M; hERG Qpatch IC50: 24 M) and afforded
good selectivity against a panel of 69 enzymes, receptors and ion channels (all IC50’s>30
M except: Cox1 IC50: 3.3 M; PDE4d IC50: 7.2 M; and VMAT2 IC50: 5.2 M).
Furthermore, 6 afforded an excellent mouse pharmacokinetic profile (Table 2). Therefore,
ACS Paragon Plus Environment
24

Journal of Medicinal Chemistry
Page 26 of 133
1
2
3
4
5
6
7
8
we chose compound 6 as a minimal pharmacophore, maintaining the key 3-amino-
dihydrobenzofuran moiety at S1 and the iPr group at S1β for further optimization.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
With our primary pharmacophore set, we turned our attention toward addressing
protease selectivity. To this end, we explored optimizing the spacing between the 3-
amino-dihydrobenzofuran and the carboxylic acid making the critical interaction in the
oxyanion hole. We reasoned that rigidifying the structure to enforce a conformation that
would better position the carboxylic acid in the oxyanion hole would improve both FXIa
affinity and potentially enhance selectivity against other serine proteases. Molecular
modeling suggested that a fused ring system would provide such an opportunity. A fused
3,4-dihydro-2H-benzo[b][1,4]oxazine, that essentially ties the aryl acetic acid back into
the linker as in 13, appeared particularly promising (Figure 6). In addition, the added
polarity of this linker offered the potential to reduce the plasma IC50 shift associated with
compound 6. Synthesized as a mixture of diastereomers at the 2 position of the 3,4-
dihydro-2H-benzo[b][1,4]oxazine and subsequently separated, the more potent (R)-
isomer, 13,51 exhibited exquisite potency that was at the limit of the biochemical assay
ACS Paragon Plus Environment
25

Page 27 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(<0.7 nM). This increase in biochemical potency also translated to a 10-fold shift in
potency relative to compound 6 in the presence of plasma (Table 4). In addition, the
concentration necessary to elicit a doubling in aPTT time was only 0.4 µM (~5-fold more
potent than compound 6). Unfortunately, the 2×aPTT/PT ratio was low (5.1x) indicating
very little selectivity between the intrinsic and extrinsic coagulation pathways and
suggesting that the observed potency in the aPTT assay was not entirely due to FXIa
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
inhibition.
H
N
R
H
N
O
N
O
O
O
O
OH
O
O
OH
OH
OH
O
OH
15 = OMe
17
14
13
16
18 = Me
19 = Cl
H2N
H2N
O
O
Figure 6. Structure of the fused 3,4-dihydro-2H-benzo[b][1,4]oxazine (13) and benzoic
acids (14-19).
With selectivity remaining a significant challenge, we next explored more
substantial changes in the spacing between the 3-amino-dihydrobenzofuran and the
carboxylic acid. Up to this point, we had focused on aryl acetic acids as a means of
spanning the six-atom distance between the S1 moiety and the carboxylic acid of our
ACS Paragon Plus Environment
26

Journal of Medicinal Chemistry
Page 28 of 133
1
2
3
4
5
6
7
8
9
pharmacophore. It was reasoned that pushing the P1’ phenyl ring further toward the
oxyanion hole by converting the aryl acetic acid to a benzoic acid and by extending the
linker by one atom to maintain the requisite spacing might offer new SAR to aid in
addressing the selectivity challenge. To this end, an introductory set of benzoic acid
compounds exemplified by 14, 15, 16, and 17 were synthesized (Figure 6). The inclusion
of indoline 17 was inspired by the exquisite potency offered by the fused ring system of
13. The ether-linked benzoic acid, 14, proved to be 5-fold less potent than 6 against FXIa
and exhibited a larger plasma shift (Table 4). In addition, 14 did not demonstrate a
significant improvement in selectivity over other serine proteases. Surprisingly, addition
of a m-methoxy group to the benzoic acid, 15, demonstrated a substantial increase in
selectivity over most of the coagulation factors and fibrinolysis proteases, although FXIa
potency, especially in the presence of plasma, did not improve compared to compound 6
(Table 4). Exchanging the ether oxygen in the linker for a nitrogen, 16, resulted in a
significant loss of FXIa potency (IC50 2.52 M). However, the indoline 17 provided
biochemical potency on par with compound 6 and, like 15, also afforded an increase in
selectivity against the coagulation proteases (Table 4). In addition, indoline 17
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
27

Page 29 of 133
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
demonstrated an encouraging in vivo mouse PK profile with a long t1/2 (9 h), and promising
oral bioavailability (36%) (Table 5).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 4. In vitro potency and selectivity data: IC50 (µM)a:
Compound
FXIa
6
13
14
15
16
17
18
19
20*
21*
22*
23
0.003 <0.0007 0.012 0.003 2.52
0.006 0.017 0.015 0.04
0.002 <0.0007 0.0004
0.12
0.01
1.19
0.24
0.22
6.62
5.49
8.11
0.17
0.045
0.01
b
cd-FXIa
ND
plasma
FVIIa
5.10
0.17
2.00
0.03
3.27
0.04
28.3
10.2
6.23
3.30
45.1
11.3
20.5
9.60
12.5
15.5
8.75
7.86
16.7
13.4
3.65
1.92
b
b
ND
ND
FIXa
FXa
0.50
0.22
0.24
0.05
6.21
9.09
1.83
3.08
12.5
9.82
11.1
6.22
16.3
7.43
1.71
6.14
11.2
7.21
1.05
2.57
0.001
0.25
1.89
0.17
b
b
b
b
b
b
b
ND
ND
ND
ND
ND
ND
ND
0.090 0.03
0.003 0.001
0.090 0.05
Thrombin
PKL
0.025 0.009
0.058 0.041 0.021 0.16
0.002 0.016
0.09
0.59
0.02
1.32
2.20
30.8
0.40
17.5
0.66
2.56
0.52
4.41
4.13
>100
2.48
>100
3.67
34.0
0.93
34.2
6.68
46.2
1.13
27.2
0.49
9.58
0.25
73.6
3.70
17.4
1.42
8.92
tPA
2.60
0.69
Plasmin
uPA
0.047 0.02
0.04
0.06
FD
b
ND
*
aAll values are geometric means of ≥2 replicates; Diastereomeric mixture. bND = not
determined.
From the introductory set of compounds, 15 and 17 emerged as frontrunners being
equipotent with compound 6, but with improved selectivity in the coagulation pathway.
However, further attempts to expand the SAR around 15 to improve FXIa potency,
ACS Paragon Plus Environment
28