Structure-Guided Design of Novel, Potent, and Selective Macrocyclic Plasma Kallikrein Inhibitors

Article abstract of DOI:10.1021/acsmedchemlett.6b00384

A series of macrocyclic analogues were designed and synthesized based on the cocrystal structure of small molecule plasma kallikrein (pKal) inhibitor, 2, with the pKal protease domain. This led to the discovery of a potent macrocyclic pKal inhibitor 29, w

Full text of DOI:10.1021/acsmedchemlett.6b00384

Subscriber access provided by FLORIDA ATLANTIC UNIV
Letter
Structure-Guided Design of Novel, Potent and
Selective Macrocyclic Plasma Kallikrein Inhibitors
Zhe Li, James Robert Partridge, Abel Silva-Garcia, Peter Rademacher, Andreas Betz, Qing Xu, Hing Leung
Sham, YunJin Hu, Yuqing Shan, Bin Liu, Ying Zhang, Haijuan Shi, Qiong Xu, Xubo Ma, and Li Zhang
ACS Med. Chem. Lett., Just Accepted Manuscript • DOI: 10.1021/acsmedchemlett.6b00384 • Publication Date (Web): 06 Dec 2016
Downloaded from http://pubs.acs.org on December 12, 2016
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 free 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 accessible to all
readers and 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.
ACS Medicinal Chemistry Letters 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 7
ACS Medicinal Chemistry Letters
1
2
3
4
5
6
7
Structure-Guided Design of Novel, Potent and Selective
Macrocyclic Plasma Kallikrein Inhibitors
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
Zhe Li,†,* James Partridge,† Abel Silva-Garcia,† Peter Rademacher,† Andreas Betz,† Qing Xu,† Hing
Sham,† Yunjin Hu,‡ Yuqing Shan,‡ Bin Liu,‡ Ying Zhang,‡ Haijuan Shi,‡ Qiong Xu,‡ Xubo Ma,‡ Li
Zhang‡
† Global Blood Therapeutics, South San Francisco, CA 94080
‡ Pharmaron Xi’an Co., Xi’an, Shaanxi, China 710018
KEYWORDS: Hereditary angioedema, plasma kallikrein, protease inhibitor, structure-based design,
macrocycle.
ABSTRACT: A series of macrocyclic analogs were designed and synthesized based on the co-crystal structure of small molecule
plasma kallikrein (pKal) inhibitor, 2, with the pKal protease domain. This led to the discovery of a potent macrocyclic pKal inhibitor
29, with an IC50 of 2 nM for one olefinic isomer and 42.3 nM for the other olefinic isomer.
The serine protease plasma kallikrein (pKal) is a multifunc-
tional enzyme involved in the early phase of intrinsic blood co-
agulation, kinin formation and fibrinolysis.1 The native protease
consists of a heavy chain, which mediates the interaction of
pKal with other proteins, and the protease domain is located in
the light chain. The two-domain structure of pKal results from
the activation of its single chain precursor kallikrein (preKal)
by the protease factor XIIa (FXIIa). In plasma the activation of
preKal is triggered after autocatalytic conversion of FXII to
FXIIa on surfaces and is accelerated by the cofactor high mo-
lecular weight kininogen (HMWK). The reaction is then ampli-
fied, by reciprocal activation of FXIIa by pKal. Under physio-
logic conditions FXII, HMWK and preKal form a complex with
a variety of receptors and enzymes expressed on endothelial
surfaces, which catalyzes the activation of pKal. After activa-
tion, pKal catalyzes the release of the proinflammatory peptide
bradykinin from HMWK.2 Further pKal appears to be involved
in the generation of plasmin from plasminogen and renin from
prorenin, which links the protease to fibrinolysis and the regu-
lation of blood pressure. Because of the overlapping activities
of pKal in clinically important processes, it appears to be a via-
ble target for therapeutic intervention.3
was also introduced for the treatment of HAE.5 More recently,
a monoclonal antibody against pKal has been evaluated in clin-
ical trials and demonstrated early evidence of efficacy.6
A small molecule oral drug for prophylactic treatment of
HAE is highly desirable. Only two small molecule non-peptidic
pKal inhibitors are known, BCX4161 (1)7 and compound 2
(Figure 1). BCX4161 and its backup compound BCX73538 are
the only small molecule pKal inhibitors that have reached hu-
man clinical trials for HAE. Despite initial success in Phase 1b
studies with HAE patients, BCX4161 failed in recent Phase 2
studies likely due to its poor PK profile and lack of adequate
exposure and efficacy. The second generation compound
BCX7353 (structure not reported) has also entered human trials
recently.8 A class of small molecules reported in patent litera-
ture9 caught our attention. Many examples described in the ap-
plication possess excellent potency against pKal. One com-
pound (compound 2,9 Figure 1) was tested and found to have an
Figure 1. Literature small molecule inhibitors of pKal.
O
N
O
H
N
N
N
O
H
N
N
In plasma the activity of pKal is mainly regulated by macro-
gobulin, antiplasmin and C1 esterase inhibitor (C1-INH). How-
ever, C1-INH appears to be the most important, since an inher-
ited reduced expression or deficiency of the inhibitor results in
excess release of bradykinin and leads to episodic swelling at-
tacks in these patients termed hereditary angioedema (HAE).
The frequency and severity of HAE attacks is highly variable
and unpredictable. Some attacks of swelling that affect the lar-
ynx may block airway and can be fatal. Current treatment op-
tions for HAE patients are limited to biologics. These treat-
ments, targeting to correct C1-INH deficiency, include replace-
ment therapy either by plasma transfusion or infusion of recom-
binant C1-INH.4 An engineered Kunitz-type protease inhibitor
H
N
NH
NH2
NH2
HO
O
O
N
N
1
2
(WO2012017020A1)
BCX4161)
(
enzymatic IC50 of 2.4 nM in our conditions. More importantly,
the compound displayed excellent selectivity over a panel of
eight closely related proteases. Subsequently, the compound
was successfully co-crystallized with the purified protease do-
main of pKal and an X-ray structure was obtained (PDB code
5TZ9).10 The inhibitor demonstrated a surprising mode of inter-
action, which upon binding resulted in a dramatic rearrange-
ACS Paragon Plus Environment

ACS Medicinal Chemistry Letters
Page 2 of 7
ment of amino acids surrounding the active site of pKal. Com- Compound 2 had little oral exposure when it was dosed in
1
2
3
4
5
6
7
8
pound 2 binds to the active site of pKal and forces a rearrange-
ment of Trp598 to expose a deep binding pocket. This allows
the molecule to form critical stacking interactions with the cat-
alytic histidine residue, His434, and two other residues: Trp598,
and Tyr555. The lack of conservation in surrounding residues
explains the >1,000-fold specificity over structurally similar
proteases.
rats by oral gavage. After a 20 mg/kg dose, the Cmax was 25
ng/mL and has a calculated oral bioavailability of ~6%. A pos-
sible explanation for the lack of oral exposure could be at-
tributed to the physiochemical properties, mainly the presence
of seven rotatable bonds which may have a negative impact on
its permeability. Reducing the number of freely rotatable bonds
by conformational restriction is a common medicinal chemistry
approach to improve overall properties of lead compounds,
such as selectivity,11 potency,12 and bioavailability.13
Compound 2 adopts a unique U-shape conformation that
compliments the induced binding pockets of pKal. Several key
interactions are visible and likely important for the binding po-
tency. The middle pyrazole ring of 2 interacts with the imidaz-
ole ring of catalytic His434 via pi-pi stacking with an average
distance of ~ 3.6 Å between them. The phenyl ring of 2 sits in a
pocket surrounded by Ser597, Gly480 and Trp598. In addition
to a pi-pi stacking interaction with Try555, the terminal pyra-
zole ring of 2 also forms water mediated interaction with the
Pro-R hydrogen of CH2 of Gly480 (Figures 2&3).
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 4. Co-crystal structure of 2 with yellow dots indicating link-
ing points for macrocyclic analogs.
Figure 2. Co-crystal structure of pKal with 2. Numbers in paren-
thesis represent the canonical chymotrypsin numbering scheme.
Two water molecules are shown as blue spheres.
We sought to make rigidified analogs of 2 in hope to improve
its oral bioavailability. Closer examination of the co-crystal
structure of 2 with pKal revealed a possible way of modifying
and restraining the molecule, namely, connecting the amino-
pyridine group, residing in the S1 pocket, to the phenyl group,
residing near the S4 pocket, via a straight-chain alkyl linker that
spans over the top of Trp598. As seen in the co-crystal structure,
the space over Trp598 is open to solvents and has room to easily
accommodate a straight-chain linker. Further analysis sug-
gested that the best connecting points may be 2-methyl of the
aminopyridine and 2-position of the phenyl ring (Figure 4).
Figure 5. Modeling of proposed macrocyclic 3 with an 8-atom
trans-double bond linker.
Figure 3. Schematic drawing generated by MOE showing key in-
teractions between 2 and the active site residues of pKal.
Asp
437
H
O
2
His
434
Ser
578
Glu
479
N
N
O
Gly
480
Met
561
HN
H
-
N
+
N
Ser
478
Ser
597
Lys
575
N
H
Cys
574
Tyr
555
Trp
598
Thr
596
NH2
H
O
2
H
O
2
Computer modeling software MOE14 was used to guide the
design of the linker. Specifically, a potential linker alkyl chain
was allowed to grow atom-by-atom from both the amino-
pyridine and the phenyl positions until the two half-chains met
Gly
601
Ala
573
Asp
572
Gly
599
Cys
602
ACS Paragon Plus Environment

Page 3 of 7
ACS Medicinal Chemistry Letters
(15), DMF 50 oC, 16 h, 74%; for 18 K2CO3, 5-bromopent-1-ene
in close proximity (< 2 Å) over the top of residue Trp598. The
(16), DMF 50 oC, 16 h, 100%; (d) LAH, Et2O, reflux, 16 h, 46%
for 9 and 38% for 10.
1
2
3
4
5
6
7
8
two half-chains were then connected with a single- or double-
bond and the resulting macrocyclic structure was minimized
(Amber10 EHT force field) within the binding pocket of pKal.
For synthetic ease and accessibility, oxygen atoms were intro-
duced at both ends of the linker as anchor points to the phenyl
and aminopyridine rings, respectively. A double-bond in the
middle of the linker may be advantageous in reducing the num-
ber of conformers associated a flexible alkyl chain linker. Our
modeling suggested a six-atom straight-chain alkylene or
alkenylene group with oxygen at both ends as the shortest pos-
sible linker. Based on modeling it seems that the -O(CH2)6O- or
-O(CH2)2CH=CH(CH2)2O- linker adopts a favorable staggered
gauche-trans-gauche conformation (Figure 5). Key interactions
observed between 2 and pKal would be preserved in the pro-
posed macrocyclic structure according to the model. These in-
clude the pi-pi stacking of pyrazole with His434; the hydrogen-
bond between C=O of Ser597 and amide-NH of the macrocycle.
To synthesize the acid intermediate 7 (scheme 2), 2-hydroxy-
4-methylbenzoic acid 19 with acetone under acidic conditions
gave acetonide 20. This was followed by benzylic bromination
and displacement of the newly formed bromide 21 with pyra-
zole 22 to deliver 23. Reduction of the ester and selective alkyl-
ation of the phenolic hydroxyl provided alkylated alcohol 25,
which was converted to the chloride 26 by reacting with thionyl
chloride. N-alkylation with ethyl pyrazole-4-carboxylate 27,
followed by hydrolysis of the ethyl ester, provided the key acid
intermediate 7 which was coupled with amine 9 to give bis-ter-
minal alkene 4. Ring-closing metathesis under standard condi-
tions using Grubbs’ II catalyst produced the macrocycle 3.
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
Scheme 2. Synthesis of Macrocycles 3, 29, 30 and 31a
O
O
O
N
N
X
OR3
a
g
d
O
OH
OH
We sought to employ Grubbs ring-closing metathesis
(RCM)15-17 to synthesize the targeted macrocycle 3 (Figure 6).
Macrocycle 3 or others with different ring sizes could be ob-
tained directly from the key bis-alkene intermediates 4-6 via
RCM reactions. These intermediates are derived from the cor-
responding acids (7 and 8) and amines (9 and 10).
OR2
OH)
R1
N
O
H)
Br)
N
O
=
=
=
=
=
=
=
X
20
21
23
24
H;
(R1
(R1
(R2
(R2
e
f
b
19
=
=
25
26
But-3-enyl;
X
OH)
N
N
c
But-3-enyl; X
Cl)
(R
pyrazol-1-yl)
(R
2
1
n
=
=
28
7
1)
Et)
n =
(R3
(R3
h
H,
N
N
O
N
O
O
NH2
NH2
N
NH2
j
H
NH
O
NH
O
i
O
N
Figure 6. Structure-based design of macrocycles.
N
O
N
N
m
N
N
O
m
m
N
O
N
N
N
N
N
N
n
O
m =
m =
m =
N
N
H
=
=
=
m =
m =
m =
4
5
6
1)
1)
2)
n
n
1,
2,
1,
(n
(n
(n
=
=
=
3
29
30
1)
1)
2)
1,
2,
1,
=
(n
m =
1)
1,
(n
(n
(n
31
N
linker
NH2
2
2
3
macrocycle
aReagents and conditions: (a) TFAA,TFA, acetone, rt, 3 d, 46%;
(b) NBS, BPO, CCl4, 75 oC, 1 h, 66%; (c) pyrazole 22, K2CO3, 50
oC, 2 h, 100%; (d) LAH, THF, -78 oC, 1 h, 84%; (e) 15, Na2CO3,
binding
conformation
N
N
o
O
DMF, 80 C, 14 h, 30%; (f) SOCl2, DCM, rt, 30 min, 93%; (g)
O
N
OH
N
ethyl 1H-pyrazole-4-carboxylate 27, K2CO3, DMF, rt, 14 h, 65%;
(h)NaOH, MeOH/H2O, rt, 85%; (i) 9 or 10, HATU, DIPEA, DM,
rt, 14 h, 63% for 4; 59% for 5; (j) Grubbs' catalyst (II), DCE, 70 oC,
12 to 48 h; 4% for 3; 13% for 29.
NH2
H
O
N
m
N
O
N
+
N
O
NH2
N
m
N
H2N
N
N
N
O
1)
2)
2)
n
n
=
=
9
10
1)
2)
=
=
=
=
m =
m =
m =
7
8
4
5
6
1)
(m
(m
1,
2,
1,
(n
(n
(n
(n
(n
=
2)
Two macrocyclic products, presumably cis- and trans-iso-
mers, were isolated from the ring-closing metathesis reaction
after HPLC purification. One analog, 3a, was found to have
moderate pKal inhibitory activity with an IC50 of 1.21 µM; the
other isomer 3b had a similar potency with an IC50 of 1.57 µM.
Due to the complexity of their proton NMR spectra, unambigu-
ous assignment of the cis- and trans-isomers was not possible.
Nevertheless, it seemed that the influence of double-bond ge-
ometry on in-vitro enzymatic potency was minimal as sug-
gested by their IC50 numbers. However, the macrocyclic ana-
logs, either 3a or 3b were much weaker in inhibitory activity
compared to the original small molecule 2. One possible expla-
nation for their poor potency was that the 8-atom linker was too
short for efficient binding of 3a or 3b to pKal. However, it may
also be possible that macrocyclic 3a or 3b was forced into a less
favorable conformation by the relatively rigid double-bond in
their linkers. To rule out this possibility, the double-bond in 3
was hydrogenated to produce the corresponding saturated mac-
rocycle 31 (due to the poor solubility of the free amino com-
pounds, N-Boc-protected intermediates were used to synthesize
31, Supporting Information). The resulting macrocycle 31 with
saturated linker was found to be even less potent than either of
the two isomers 3a or 3b with an IC50 of 8.31 µM (31), suggest-
ing the rigidity of the double-bond did not affect the enzymatic
potency. Together, these results suggested that the linker in
these macrocyclic analogs was likely too short to support strong
The O-alkenyl aminopyridines, 9 and 10, were synthesized
according to synthetic route outlined in Scheme 1. Condensa-
tion of 3-aminocrotononitrile 11 and 2-cyanoacetic acid 12 in
the presence of acetic anhydride provided bis-nitrile 13, which
was treated with sodium ethoxide in ethanol to give 4-hydrox-
ypyrdine intermediate 14 as its sodium salt. This intermediate
was then O-alkylated with ω-alkyenyl bromides 15 and 16 fol-
lowed by reduction with LAH to produce the desired O-alkenyl-
ated aminopyridine derivatives 9 and 10, respectively.
Scheme 1. Synthesis of Intermediates 9 and 10a
O
ONa
NC
O
NC
CN
CN
a
b
+
NC
OH
m
H2N
H2N
H2N
N
11
12
13
14
O
N
O
m
c
CN
d
NH2
Br
H2N
N
m
H2N
=
=
=
=
=
=
15
16
17
18
9
1)
2)
1)
2)
1)
2)
(m
(m
(m
(m
(m
(m
10
aReagents and conditions: (a) Ac2O, dioxane, reflux 2 h, 44%; (b)
EtONa, EtOH, reflux, 89%; (c) for 17 K2CO3, 4-bromobut-1-ene
ACS Paragon Plus Environment

ACS Medicinal Chemistry Letters
Page 4 of 7
and 3b with IC50s of 53.2 µM for isomer 32a and IC50 > 3 mM
for the other isomer 32b, respectively. Clearly, these results
demonstrate the pyridine residue in macrocycle 3 was correctly
installed with its ring nitrogen pointing “down”. This was con-
firmed by a high-resolution, 1.7 Å, co-crystal structure of 2 with
the protease domain of pKal. As a matter of fact, this nitrogen
was involved in a water mediated H-bond with the carboxylate
of Asp572 near the bottom of the S1 pocket (Figures 2&3).
binding with pKal. Before moving on to synthesize and test
1
2
3
4
5
6
7
8
macrocycles with longer linkers, we decided to further investi-
gate the aminopyridine group and its orientation in the S1
pocket. As seen in the co-crystal structure (Figures 2&3), the
aminopyridine group in the S1 pocket is sandwiched between a
group of residues with Cys574 and Cys602 on one side and
Trp598 on the other; its pyridine ring was locked in approxi-
mately parallel fashion to the indole ring of Trp598. However,
due to the poor resolution of the initial X-ray crystal structure
(2.3 Å), the exact orientation of the pyridine ring could not be
determined based solely on the electron density map. Two pos-
sibilities exist: the N-atom was facing “up” to solvents or point-
ing “down” as illustrated in Figure 7. The resolution of the ini-
tial co-crystal structures obtained with compound 2 was 2.3 Å,
not high enough to distinguish between these two poses. Poorer
binding to pKal and negative impact on enzymatic potency
would be certainly expected if the aminopyridine group in mac-
rocyclic 3a and 3b was forced in an unfavorable conformation
by the linker. To address this question, we sought to make com-
pound 32, a macrocyclic analog with the exact same linker as
that in 3 but opposite nitrogen orientation on the pyridine ring
(Figure 7).
Scheme 4. Synthesis of Macrocycle 32a
9
O
O
O
OH
NH2
NH2
N
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
b
H
NH
N
N
H2
N
N
O
N
O
O
N
38
N
O
N
O
N
N
O
a
N
N
N
N
N
N
7
39
32
aReagents and conditions: (a) HATU, DIPEA, DMF, rt, 14 h, 52%;
(b) Grubbs' catalyst (II), DCE, 70 oC, 48 h, 63%.
With proper resolution of the pyridine ring orientation, we
were ready to make macrocycle 29 (Scheme 2), substituting the
8-atom linker with a longer 9-atom linker, aiming to improve
binding efficiency and potency. To access this compound we
applied a modified synthetic sequence to make the key acid 8
(Scheme 5) as that for macrocycle 3 (Scheme 2). Salicylic acid
derivative 19 was transformed, after a series of simple func-
tional groups manipulations, to the phenolic intermediate 46;
the phenol group was O-alkylated with a longer 5-bromopent-
1-ene 16 to produce the O-alkene acid 8, after hydrolysis of the
ester group. Subsequent amide coupling and RCM reaction pro-
duced the corresponding macrocycle 29 with a 9-atom linker
(Scheme 2, steps i and j).
Figure 7. Two possible orientations of the pyridine ring.
2
2
up
nitrogen down
nitrogen
unfavorable ?
NH2
Scheme 5. Synthesis of Acid 8 for Macrocycle 29a
N
O
O
NH2
NH
NH
N
O
O
O
OR4
N
N
N
N
O
O
O
a
d
N
N
X
f
N
N
N
N
N
OR2
OR1
OH
OH
N
OMe
X
O
N
N
3
32
=
=
=
=
=
=
43
44
19
OH)
Cl)
40
R
R
Me; X H)
,
,
(X
(X
(R1
2
b
c
e
OR3
=
41
42
Me; X Br)
,
(R1
2
The required O-alkenyl aminopyridine derivative 38 was
synthesized according to reaction sequence described in
Scheme 3. First, commercial cyanopyridine compound 33 was
hydrogenated in the presence of boc-anhydride to produce bis-
Boc derivative 34. O-alkylation followed by hydrolysis pro-
vided acid 36, which was converted to the corresponding pro-
tected N-Boc amine 37 through Curtius rearrangement via an
acyl azide intermediate. Finally, the aminopyridine 38 was ob-
tained after removal of the Boc group under TFA conditions.
=
=
=
=
45
46
47
R
4
Et
)
Me;
(R3
(R3
4
g
h
i
R
Me;
(R1
2
=
R
H;
Et
)
=
X
pyrazol-1-yl)
=
R
Et
)
(R3 pen-4-enyl;
4
=
=
H
8
R
(R3 pen-4-enyl;
)
4
aReagents and conditions: (a) CH3I, K2CO3, DMF, rt, 16 h, 79%;
(b) NBS, BPO, CCl4, 80 oC, 16 h, 100%; (c) pyrazole 22, K2CO3,
50 oC, 16 h, 21%; (d) LAH, THF, 0 - 25 oC, 1 h, 100%; (e) SOCl2,
o
DCM, 0 - 25 C, 90 min, 100%; (f) ethyl 1H-pyrazole-4-carbox-
ylate 27, K2CO3, DMF, 50 oC, 14 h, 64%; (g) BBr3, DCM, -78 - 0
oC, 2 h, 52%; (h) 5-bromopent-1-ene 16, Cs2CO3, DMF; 90 oC, 2
h, 66%; (i) NaOH, MeOH/H2O, 50 oC, 85%.
Scheme 3. Synthesis of Aminopyridine 38a
N
a
b
N(Boc)2
N(Boc)2
To gain an approximate estimate of its activity, the crude
mixture 29 from initial silica gel column purification, contain-
ing both cis- and trans-isomers, was assayed directly for pKal
activity without separating the two isomers. Consistent with our
SAR analysis for longer linkers, 29 was found to be at least 250-
fold more potent than compound 3a or 3b, having an IC50 of 4.8
nM. The mixture was then purified by reverse phase HPLC to
separate both isomers. The purified analogs displayed IC50 of 2
nM for one olefinic isomer and 42.3 nM for the other, respec-
tively. Once again, unambiguous identification of the cis- and
trans-isomers was not possible due to the complex nature of
their NMR spectra.
EtO
EtO
EtO
N
N
H
O
N
O
O
H
O
O
O
e
33
34
35
c
N(Boc)2
d
NH2
N(Boc)2
HO
N
O
BocHN
N
37
O
H2N
N
O
O
36
38
aReagents and conditions: (a) Raney-Ni, H2, Boc2O, MeOH, rt, 16
o
h, 33%; (b) 4-bromobut-1-ene 15, K2CO3, Ag2CO3, DMF, 80 C,
o
16 h, 50%; (c) NaOH, THF, MeOH, H2O, 50 C, 16 h, 77%; (d)
TEA, DPPA, t-BuOH, toluene, 100 oC, 16 h, 84%; (e) TFA, DCM,
rt, 12 h, 98%.
Similar to previous macrocyclic analogs, two versions of
(32), likely the cis- and trans-isomers, with identical molecular
mass were isolated from the reaction mixture (Scheme 4) after
HPLC purification. The compounds suffered at least 40-fold
potency loss compared to the corresponding macrocycles 3a
Given the promising potency of 29a and 29b, we decided to
explore other macrocycles with an isomeric 9-atom linker to
evaluate possible influence of double-bond position on their
pKal activity (cmpd 30, Scheme 2 and Table 1). The isomeric
ACS Paragon Plus Environment

Page 5 of 7
ACS Medicinal Chemistry Letters
macrocycles were synthesized according to a similar synthetic panel containing eight closely-related proteases to evaluate
1
2
3
4
5
6
7
8
strategy as outlined in Schemes 1&2 for compound 3, substitut-
ing 4-bromobut-1-ene (15) with a longer ω-alkene bromide 5-
bromopent-1-ene (16) when converting 4-hydroxypyridine 14
to its O-alkylated analog 18 (Scheme 1). Compound 30 was ob-
tained after RCM with the bis-alkene intermediate 6 (Scheme
2) (Supporting Information). Compound 30, tested as a mixture
of cis- and trans-isomers, was found to have an IC50 of 153 nM,
suffering approximately 30-fold potency loss compared to the
corresponding isomeric congener (29), suggesting that the dou-
ble bond at this position is less preferred.
their selectivity profiles. All four macrocycles displayed excel-
lent selectivity similar to that seen with the precursor molecule
2 (Table 2). These results were not unexpected since the in-
duced binding conformation, the origin of good selectivity, was
preserved with the macrocyclic analogs. These macrocyclic an-
alogs had little effects (IC50 > 200 µM) on the enzymatic ac-
tivities of several proteases including plasmin, trypsin, TPA,
thrombin, FXa, and FXIIa; and even though these compounds
inhibited the activity of FXIa with moderate IC50s ranging from
single-digit to double-digit micromolar (Table 2), these activi-
ties were at least 1,800-fold less potent compared to the IC50s
for pKal.
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
To determine if the induced binding conformation by com-
pound 2 was preserved with these macrocycles, several analogs,
including 29a, 29b and 30, were selected for co-crystallization
experiments using the protease domain of pKal. Unfortunately,
except for compound 30, these analogs failed to yield any co-
crystals after several attempts. Interestingly, only the trans-iso-
mer was observed in the co-crystal structure, with a final maxi-
mum resolution of 1.4 Å (PDB code 5TJX; Supporting Infor-
mation). The macrocycle 30 sits in the active site of the protease
domain as expected and aligns well when comparing to the
structure of the pKal protease domain co-crystallized with com-
pound 2 (Figure 8). The compounds bind in the same orienta-
tion and the same key interactions are preserved, this includes
hydrogen bonds to the sidechain of Asp572 and carbonyl of
Ser597. There are however subtle differences stemming from
the addition of a large linker to connect the aminopyridine and
the phenyl group. The phenyl group appears shifted out of plane
when compared to compound 2. This likely results in the shift
of the S4 pyrazole and the second pyrazole shown to form a pi-
pi stacking interaction with His434. Compared to the in-silico
modeling, compound 30 aligns well in the S1 pocket but
changes drastically after exiting the S1 pocket (Figure 8). By
aligning the two molecules we calculate an overall RMSD of
approximately 0.763 Å between the in-silico model and crystal-
lographic model, with a majority of differences coming from
arrangement of the linker.
Table 1. Plasma kallikrein potency for 2 and macrocycles.
Cmpd
2
pKal IC50 (nM)
2.4 + 1.1
N (Tests performed)
82
4
3a
(1.21 + 0.040) x103
(1.57 + 0.58) x103
4.8 + 1.8
3b
4
29
6
29a
29b
30
2.0 + 0.0
3
42.3 + 12
4
153 + 14.1
5
31
(8.31 + 2.28) x103
(53.2 + 2.8) x103
inactive
3
32a
32b
3
3
Table 2. Inhibitory activities of 2 and macrocycles against
eight proteases.
Cmpd
2
29
29a
29b
30
pKal
IC50 (µΜ)
2.4
4.8 x10-3
2.0 x10-3
42.3 x10-3
153 x10-3
x10-3
Plasmin
IC50 (µΜ)
> 103
> 103
106
21.4+4.1
> 200
> 200
> 200
> 200
> 200
9.97+ 3.86
> 200
fXa
IC50 (µΜ)
Figure 8. Comparison of 2, macrocycle 30 (co-crystal and in-silico
model generated using MOE). Numbers in parenthesis represent
the canonical chymotrypsin numbering scheme. Two water mole-
cules are shown as blue spheres.
fXIa
IC50 (µΜ)
11.7 +
0.3
3.61+1.79
> 200
40.9+2.8
> 200
12.7+ 0.54
> 200
fXIIa
IC50 (µΜ)
> 103
578
> 200
> 200
> 200
> 200
Trypsin
IC50 (µΜ)
> 200
> 200
> 200
tPA IC50
(µΜ)
> 103
> 103
> 200
> 200
> 200
Thrombin
IC50 (µΜ)
> 200
> 200
> 200
In-vivo rat pharmacokinetic (PK) studies on the most active
compounds in the series were performed to evaluate PK profiles
and assess their potential as possible drug leads. Instead of dos-
ing each isomer, the mixtures 29 and 30 containing both the cis-
and trans-isomers were administered in rat by oral gavage to
gain quick readouts on their oral exposure. Unfortunately, both
compounds displayed little or no exposure. To better under-
stand the PK results, microsomal stability and permeability
were measured for these compounds. All four analogs were
found to have poor permeability with their Papp numbers well
below 1 cm-6/s (Supporting Information). More surprisingly, all
four compounds displayed extremely short half-life (< 5 min;
comparing to 48 min for 2) and were metabolized rapidly by rat
Next, the most active macrocyclic compounds 29a, 29b, 29
and 30, along with compound 2 were profiled in a protease
ACS Paragon Plus Environment

ACS Medicinal Chemistry Letters
liver microsomal isolates in vitro. Due to these poor properties,
Page 6 of 7
1
2
3
4
5
6
7
8
further work on the macrocycles were terminated.
REFERENCES
(1) Bryant, J.W.; Shariat-Madar, Z. Human plasma kallikrein kinin
system: physiological and biochemical parameters. Cardiovasc. Hema-
tol. Agents Med. Chem. 2009, 7, 234-50.
(2) Kaplan, A. P. Bradykinin and the Pathogenesis of Hereditary An-
gioedema. World Allergy Organ J. 2011, 4, 73-75.
(3) Maas, C.; Oschatz, C.; Renne, T. The plasma contact system 2.0.
Semin. Thromb. Hemost. 2011, 37, 375-38.
(4) Levy, J.H.; Freiberger, D.J.; Roback, J. Hereditary angioedema:
current and emerging treatment options. Anesthesia & Analgesia 2010,
110, 1271-1280.
(5) Schneider, L.; Lumry, W.; Vegh, A.; Williams, A.H.; Schmal-
bach, T. Critical role of kallikrein in hereditary angioedema pathogen-
esis: a clinical trial of escallantide, a novel kallikrein inhibitor. J. Al-
lergy Clin. Immunol. 2007, 120, 416-422.
In summary, novel macrocyclic plasma pKal inhibitors were
designed based on the co-crystal structure of pKal with small
molecule pKal inhibitor 2. Guided by structure information and
modeling tool, we discovered a series of potent macrocyclic
pKal inhibitors, typified by 29a, having low nanomolar activity
(2.0 nM), after two rounds of linker optimization. Importantly,
these novel macrocyclic analogs retained the unique binding
conformation and excellent selectivity for a group of closely re-
lated proteases. The work demonstrated the power and utility
of SBDD in aiding the design of novel enzyme inhibitors in
drug discovery project.
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
ASSOCIATED CONTENT.
(6) Chyung, Y.; Vince, B.; Iarrobino, B.; Sexton, D.; Kenniston, J.;
Faucette, R.; TenHoor, C.; Stolz, L.E.; Stevens, C.; Biedenkapp, J.;
Adelman, B. A phase 1 study investigating DX-2930 in healthy sub-
jects. Annals of Allergy, Asthma & Immunology 2014, 113, 460-466.
(7) Aygören-Pürsün, E.; Magerl, M.; Graff, J.; Martinez-Saguer, I.;
Kreuz, W.; Longhurst, H.; Nasr, I.; Bas, M.; Straßen, U.; Fang, L.;
Cornpropst, M.; Dobo, S.; Collis, P.; Sheridan, W.P.; Maurer, M.
Prophylaxis of hereditary angioedema attacks: A randomized trial of
oral plasma kallikrein inhibition with avoralstat. J Allergy Clin Immu-
nol. 2016, pii: S0091-6749, 30270-30276.
Supporting Information. Experimental details and analytical data
for all macrocycles and synthetic intermediates, in vitro and in vivo
biological procedures, as well as modeling and crystallographic
methods. This material is available free of charge via the internet
at http://pubs.acs.org.
AUTHOR INFORMATION.
Corresponding Author
(8) Cornpropst, M.; Dobo, S.; Collier, J.; Rose, A.; Wilson, R.; Babu,
Y.S.; Collis, P.; Zong, J.; Sheridan, W.P. BCX7353, a Potent Inhibitor
of Plasma Kallikrein, Shows Sustained Maximal Enzyme Inhibition
when Dosed Orally Once Daily: Results from a Phase I Trial in Healthy
Subjects. AAAAI Annual Meeting, March 3-8, 2016, Los Angeles, CA.
(9) Brandl, T.; Flohr, S.; Kopec, S.; Lachal, J.; Merkert, C.; Namoto,
K.; Nganga, P.; Pirard, B.; Renatus, M.; Sedrani, R.; Zoller, T. N-((6-
amino-pyridin-3-yl)methyl)-heteroaryl-carboxamides as inhibitors of
plasma kallikrein. WO2012017020A1, February 12, 2012.
(10) Partridge, J.; Choy, R., Li, Z. Manuscript in preparation.
(11) Kobayashi,T.; Suemasa,A.; Igawa, A.; Ide,S.; Fukuda,H.; Abe,
H.; Arisawa, M.; Minami, M.; Shuto, S. Conformationally Restricted
GABA with Bicyclo[3.1.0]hexane Backbone as the First Highly Selec-
tive BGT‑ 1 Inhibitor. ACS Med. Chem. Lett. 2014, 5, 889−893.
(12) Spear, K.L.; Brown, M.S.; Reinhard, M.J.; McMahon, E.G.;
Olins, G.M.; Palomo, M.A.; Patton, D. R. Conformational Restriction
of Angiotensin II: Cyclic Analogues Having High Potency. J. Med.
Chem. 1990, 33, 1935-1940.
(13) Bakali, J.E.; Muccioli, G.G.; Body-Malapel, M.; Djouina, M.;
Klupsch, F.; Ghinet, A.; Barczyk, A.; Renault,N.; Chavatte, P.;
Desreumaux, P.; Lambert, D.M.; Millet, R. Conformational Restriction
Leading to a Selective CB2 Cannabinoid Receptor Agonist Orally Ac-
tive Against Colitis. ACS Med. Chem. Lett. 2015, 6, 198−203.
(14) Molecular Operating Environment (MOE), 2014.09; Chemical
Computing Group Inc., 1010 Sherbooke St. West, Suite #910, Mon-
treal, QC, Canada, H3A 2R7, 2014.
* E-mail: zli@globalbloodtx.com
Present Addresses
Andreas Betz: abetz@allostery.com
Notes. The authors declare the following competing financial in-
terest(s): The authors ZL, JP, AS, PR, AB, QX and HS are, em-
ployees of Global Blood Therapeutics which has a commercial in-
terest in pKal inhibitors.
ACKNOWLEDGMENT. The authors thank Drs. Manuel
Zancanella and Shahul Nilar for reviewing the manuscript. We
thank J. Holton, G. Meigs and ALS staff of beamline 8.3.1 for data
collection and helpful discussions. Beamline 8.3.1 at the Advanced
Light Source is operated by the University of California Office of
the President, Multicampus Research Programs and Initiatives
grant MR-15-328599 and Program for Breakthrough Biomedical
Research, which is partially funded by the Sandler Foundation. Ad-
ditional support comes from National Institutes of Health
(GM105404, GM073210, GM082250, GM094625), National Sci-
ence Foundation (1330685), Plexxikon Inc. and the M.D. Anderson
Cancer Center. The Advanced Light Source (Berkeley, CA), a na-
tional user facility operated by Lawrence Berkeley National Labor-
atory on behalf of the US Department of Energy under contract
number DE-AC02-05CH11231, Office of Basic Energy Sciences,
through the Integrated Diffraction Analysis Technologies program,
supported by the US Department of Energy Office of Biological
and Environmental Research.
(15) Fu, G. C.; Grubbs, R. H. The Application of Catalytic Ring-
Closing Olefin Metathesis to the Synthesis of Unsaturated Oxygen Het-
erocycles. J. Am. Chem. Soc. 1992, 114, 5426-5427.
(16) Fu, G. C.; Grubbs, R. H. Synthesis of Nitrogen Heterocycles
via Catalytic Ring Closing Metathesis of Dienes. J. Am. Chem.
Soc. 1992, 114, 7324-7325.
(17) Scholl, M.; Ding, S.; Lee, C. W.; Grubbs, R. H. Synthesis and
Activity of a New Generation of Ruthenium-Based Olefin Metathesis
Catalysts Coordinated with 1,3-Dimesityl-4,5-dihydroimidazol-2-yli-
dene Ligands. Org. Lett. 1999, 6, 953–956.
ABBREVIATIONS. BPO, benzoyl peroxide; DIPEA, N,N-diiso-
propylethylamine; DPPA, Diphenylphosphoryl azide; fXa, factor
Xa; fXIa, factor XIa; fXIIa, factor XIIa; HAE, hereditary angi-
oedema; HATU, 2-(1H-7-azabenzotriazol-1-yl)-1,1,3,3-tetra-
methyluronium hexafluorophosphate; pKal, plasma kallikrein;
RCM, ring-closing metathesis; tPA, tissue plasminoogen activa-
tor.
ACS Paragon Plus Environment

Page 7 of 7
ACS Medicinal Chemistry Letters
For Table of Contents Use Only
1
2
3
4
5
Structure-Guided Design of Novel, Potent and Selective Macrocyclic Plasma Kallikrein Inhibitors
Zhe Li,†,* James Partridge,† Abel Silva,† Peter Rademacher,† Andreas Betz,† Qing Xu,† Hing Sham,† Yunjin Hu,‡ Yuqing Shan,‡ Bin
Liu,‡ Ying Zhang,‡ Haijuan Shi,‡ Qiong Xu,‡ Xubo Ma,‡ Li Zhang‡
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
N
O
O
NH2
N
NH
N
N
O
H
N
N
N
O
NH2
N
N
2
N
N
=
nM
IC50 2.4
Cmpd
IC
pKal
50 (nM)
29a
29b
2
42.3
-isomer 1
-isomer 2
7
ACS Paragon Plus Environment