Discovery of danoprevir (ITMN-191/R7227), a highly selective and potent inhibitor of hepatitis C virus (HCV) NS3/4A protease

Article abstract of DOI:10.1021/jm400164c

HCV serine protease NS3 represents an attractive drug target because it is not only essential for viral replication but also implicated in the viral evasion of the host immune response pathway through direct cleavage of key proteins in the human innate im

Full text of DOI:10.1021/jm400164c

Subscriber access provided by Cape Breton University | Library
Article
Discovery of Danoprevir (ITMN-191/R7227), a Highly Selective
and Potent Inhibitor of Hepatitis C Virus (HCV) NS3/4A Protease
YUTONG JIANG, Steven W Andrews, Kevin Condroski, Brad Buckman, Vlad Serebryany, Steve
Wenglowsky, April Kennedey, Machender Madduru, Bin Wang, Michael Lyon, George Doherty, Benjamin
Woodard, Christine Lemieux, Mary Geck Do, Hailong Zhang, Joshua A Ballard, Guy Vigers, Barbra
Brandhuber, Peter Stengel, John Josey, Leonid Beigelman, Lawrence Blatt, and Scott D. Seiwert
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/jm400164c • Publication Date (Web): 14 May 2013
Downloaded from http://pubs.acs.org on May 17, 2013
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.
Journal of Medicinal Chemistry 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 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of Danoprevir (ITMNꢀ191/R7227), a Highly
Selective and Potent Inhibitor of Hepatitis C Virus
(HCV) NS3/4A Protease
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
Yutong Jiang,*,† Steven W. Andrews,† Kevin R. Condroski,† Brad Buckman,§ Vlad Serebryany,§ Steve
Wenglowsky,† April L. Kennedy,† Machender R. Madduru,† Bin Wang,† Michael Lyon,† George A.
Doherty,† Benjamin T. Woodard (deceased),† Christine Lemieux,† Mary Geck Do,† Hailong Zhang,†
Joshua Ballard,† Guy Vigers,† Barbra J. Brandhuber,† Peter Stengel,† John A. Josey,† Leonid
Beigelman,§ Lawrence Blatt,§ Scott D. Seiwert§
† Array BioPharma, 3200 Walnut Street, Boulder, Colorado 80301, USA
§ InterMune Inc., 3280 Bayshore Boulevard, Brisbane, California 94005, USA
Abstract: HCV serine protease NS3 represents an attractive drug target because it is not only essential
for viral replication, but also implicated in the viral evasion of the host immune response pathway
through direct cleavage of key proteins in the human innate immune system. Through structureꢀbased
drug design and optimization, macrocyclic peptidomimetic molecules bearing both a lipophilic P2
isoindoline carbamate and a P1/P1’ acylsulfonamide/acylsulfamide carboxylic acid bioisostere were
prepared that possessed subꢀnanomolar potency against the NS3 protease in a subꢀgenomic repliconꢀ
based cellular assay (Huhꢀ7). Danoprevir (compound 49) was selected as the clinical development
candidate for its favorable potency profile across multiple HCV genotypes and key mutant strains, and
for its good in vitro ADME profiles and in vivo target tissue (liver) exposures across multiple animal
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 47
species. Xꢀray crystallographic studies elucidated several key features in the binding of danoprevir to
HCV NS3 protease and proved invaluable to our iterative structureꢀbased design strategy.
1
2
3
4
5
6
7
8
9
F
R
O
R'
O
N
N
IC50 = 0.2ꢀ0.4 nM (1a, 1b, 4, 5 & 6), 1.6 nM (2b), 3.5 nM (3a)
EC50 = 1.6 nM (1b), 20 nM (2 & 3)
MT4ꢀMTT CC50 = 54 ꢀM
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
O
O
H
N
N
O
CYP 1A2, 2C19, 2C9, 2D6 > 25 ꢀM; 3A4 = 12 ꢀM
Timeꢀdependent CYP3A4 inhibition/induction: Negative
hERG %block @ 10 ꢀM: Negative
Selectivity: IC50 > 10 ꢀM in panel of 53 proteases
Oral bioavailability (liver): 15 % (rat); 78 % (monkey)
Liver/plasma ratio: ~10 (rat); ~30 (dog); ~100 (monkey)
O
OH
O
O
O
O
H
N
H
NH
O
N
O
S
N
N
HN
H
O
O
O
IC50 = 200-50,000 nM
ꢀ
EC50 > 10
M
49
2
Introduction
Hepatitis C virus (HCV) infection afflicts approximately 150 million people worldwide, with 3ꢀ4
million new infections every year.1a HCV infections may present few initial symptoms, but over time
can lead to liver fibrosis, cirrhosis, and hepatocellular carcinoma, making HCV the leading cause of
liver transplantation in the United States, and more than 350,000 annual deaths globally.1 Prior to the
arrival of direct acting antivirals (DAAs), the standard of care (SoC) for chronic HCV infection was
pegylated interferon alpha (PEG IFNꢀα) in combination with ribavirin (RBV), which provided sustained
virologic response (SVR) in about 50% of treated patients and was associated with significant side
effects.2 In the past decade, major drug discovery endeavors have focused on combating HCV with
DAAs,3 which target essential enzymes encoded by hepatitis C virus—such as nonꢀstructural proteins
NS3/4A, NS5A4a and NS5B.4b
The NS3/4A is a chymotrypsinꢀlike serine protease that plays an essential role in the HCV viral
replication process. Small molecule DAAs inhibiting the NS3 proteolytic activity can achieve rapid
viral load drop in patients, as was first demonstrated with inhibitor ciluprevir by Boehringer Ingelheim
in 2002.5a,b In addition to its crucial role of processing viral polypeptide, the NS3/4A protease has also
been shown to dampen cellular sensing of viral components, thus reducing endogenous type 1 interferon
production.6 Therefore, an NS3 protease inhibitor would potentially suppress HCV replication by
simultaneously disrupting two processes critical to the survival of the virus.
2
ACS Paragon Plus Environment

Page 3 of 47
Journal of Medicinal Chemistry
In 2011, two first generation DAAs, HCV NS3 protease inhibitors telaprevir7 and boceprevir,8 were
1
2
approved by the FDA. The new treatment regimens significantly improved overall SVR from the
previous 50% range to around 70%, but these inhibitors must still be administered in combination with
the old SoC (PEG IFNꢀα/RBV) in order to prevent the rapid emergence of resistance.7c,8c The side
effects brought on by these triple combination therapies can compromise their clinical utility due to
treatment discontinuation. In addition, to many parts of the world interferon is still prohibitively
expensive and can deter chronic HCV patients from seeking treatments in the first place. Therefore, a
high unmet medical need exists to develop second generation DAAs with milder adverse effects, high
barriers to resistance and action through multiple inhibition mechanisms, eventually leading to a future
SoC that is IFNꢀsparing. Some of the second generation NS3 protease inhibitors currently under
advanced clinical investigations include: simeprevir (TMC435350)5c, faldaprevir (BI201335),5d
vaniprevir (MKꢀ7009)5e and MKꢀ51725f.
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
Our drug discovery campaign aims at maximizing inhibitor potency across HCV genotypes and
clinically relevant drugꢀresistant strains (i.e. A156T and D168A), minimizing offꢀtarget activities, and
optimizing compound preferential tissue distribution to the liver—the primary HCV replication site.
Herein we report the structureꢀbased design, SAR investigation and the ensuing DMPK optimization
that lead to the discovery of a novel HCV NS3/4A protease inhibitor danoprevir (ITMNꢀ191/R7227).
Danoprevir is a macrocyclic nonꢀcovalent reversible NS3 inhibitor with a slowꢀoff rate.9f It displays
unique preclinical characteristics and elicits potent antiviral activity in the clinic.9,10 Currently
danoprevir is in phase II clinical development both in combination with PEG IFNꢀα/RBV and with
NS5B inhibitor mericitabine.
Structure-based Design and Initial Leads
At the outset of this work, the available Xꢀray crystallographic structures of the NS3 protease domain
revealed a relatively shallow and highly solventꢀexposed active site. This binding pocket presented a
challenging task in designing selective and potent small molecule inhibitors against the HCV NS3
protease. To pursue our novel class of NS3 inhibitors, a virtual chemical library was first designed and
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 47
evaluated in docking studies using GOLD.14a This virtual library consisted of tetrapeptides that
possessed a P1/P1’ carboxylic acid head group and probed diverse P2 substituents that occupy the
lipophilic S2 protein surface region (1, Scheme 1). Selection of the most promising chemical targets
was based upon a consensus scoring technique that was subsequently applied to the GOLD docking
results.14b This smaller set was then synthesized as a focused library of P2 carbamates (2). Gratifyingly,
the inhibition results from the NS3 enzymatic assay demonstrated that the virtual screening picks
clustered around the most potent hits, supporting the validity of our proposed inhibitor binding mode.
After several iterations of P2 optimization, we identified a P2 tetrahydroisoquinoline (THIQ) carbamate
as a promising lead (3).
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
Scheme 1. Structureꢀbased library design to lead 5 through depeptization and P2, P1’ꢀ modifications.
P2
P2 Substituent
R
O
R'
O
N
Linker
P3
Virtual screening
Parallel synthesis
P2 optimization
O
O
H
H
N
N
N
O
N
O
OH
OH
H
NH
H
NH
O
O
O
O
N
N
O
O
1
2
P1
Virtual library design
P4
N
N
N
P3 cap
O
O
O
O
O
O
Depeptidization
P1' Modification
O
O
O
O
O
O
O
O
O
H
H
H
N
N
N
N
S
N
N
O
OH
N
HN
HN
OH
H
H
NH
O
O
O
O
O
O
N
P1'
O
3*
IC50 = 210 nM
4
5
IC50 = 177 nM
EC50 = 2352 nM
IC50 = 2.2 nM
EC50 = 17.5 nM
ꢀ
EC50 > 10
M
(* 1:1 diastereomeric mixture at P1)
Although compound 3 was significantly potent in the NS3 enzyme functional assay, it showed no
cellular potency in a repliconꢀbased assay at its top titration level (10 ꢀM). This was attributed to the
4
ACS Paragon Plus Environment

Page 5 of 47
Journal of Medicinal Chemistry
compound’s poor cell permeability as a result of high polar surface area (solvent accessible PSA = 187
Å2), three hydrogenꢀbond donors, and a charged carboxylic acid. To solve this problem, we opted to
first transform the small peptide to a macrolide so that the molecule’s highly exposed polar surface area
is partially shielded, and the labile amide bonds are rigidified—both of which aimed at improving cell
permeability and metabolic stability.11 Aided by computational modeling, we linked the P1 and P3 side
chains of 3 to form a 15ꢀmembered macrocyclic acid 4. This modification indeed increased the
compound’s replicon potency to approximately 2 ꢀM, with enzymatic potency equivalent to its
tetrapeptide predecessor. To further improve cell permeability, we tested the replacement of the
carboxylic acid P1 head group with an acid bioisostere acylsulfonamide, reasoning that while still
maintaining an acidic group, acylsulfonamide could better distribute the anionic charge over additional
atoms, hence increasing cell permeability.12 In addition, our docking studies also suggested that an alkyl
sulfonamide could simultaneously fill the S1’ pocket and form additional hydrogen bonds in the
oxyanion hole, further improving the inhibitor’s binding affinity. The resulting cyclopropyl sulfonamide
5 improved both enzymatic and cellular potency by more than two orders of magnitude. This dramatic
improvement demonstrated that the acylsulfonamide indeed provided superior cell permeability as well
as increase in binding affinity. Although these exciting results validated our drugꢀdesign approach, the
magnitude of improvement was unexpected.
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
As shown in Figure 1, the xꢀray coꢀcrystal structure of a close analog of 5 [in which the P3 tꢀbutyl
carbamate was replaced by a cyclopentyl carbamate (6, IC50 = 1 nM, EC50 = 6 nM)] was obtained bound
in the NS3 protease domain active site. This result provided experimental support for our docking
hypothesis and initial SAR observations: 1) the peptide backbone of the molecule forms tight βꢀsheetꢀ
like interactions with the protein; 2) the acylsulfonamide is well positioned for multiple hydrogenꢀbond
interactions with protein active site residues G137, S138 and S139; 3) the cyclopropyl P1’ group binds
effectively in the S1’ pocket; 4) neither the P2 carbamate linker nor the THIQ group appears to form any
specific interactions with the protein. However, the largely flat and lipophilic THIQ group likely
achieves binding affinity by conforming closely atop the flat, lipophilic “peninsula” on the S2 region of
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 47
the protein surface, which is primarily composed of hydrophilic surface residues. This crystal structure
provided the starting point for the structureꢀbased design of our HCV protease inhibitors with improved
potency and pharmacokinetic profiles.
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 1. Xꢀray crystal structure of inhibitor 6 bound to HCV NS3 protease
Chemistry
The synthesis of the tetrapeptide P2ꢀcarbamate chemical library is shown in Scheme 2. Initially,
racemic P1 vinyl cyclopropyl amino acid (7) was used, which was later replaced by the active
enantiomer (7a, Scheme 3) at the lead optimization stage. Through HATUꢀfacilitated peptide coupling
steps a, c and d, compound 10 was readily obtained as a common intermediate. After surveying various
carbamateꢀforming reactions, we selected the CDI method (step e) for its ready access to our desired
amine inputs and clean reaction profile suitable for parallel syntheses. As the last step, the final
compounds 2 were obtained through mild hydrolysis without any observed deterioration of enantiomeric
purity.
6
ACS Paragon Plus Environment

Page 7 of 47
Journal of Medicinal Chemistry
Scheme 2a
1
2
3
4
OH
OH
5
6
7
8
a
d
c
OEt
N
N
H2N
HCl
HN
R
R
OEt
OEt
N
O
N
H
O
O
H
O
O
O
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
7
8
9
: R = Boc
: R = Boc
b
b
8a
: R = H
9a
: R = H
NR'R"
O
O
OH
O
O
f
H
H
N
N
N
N
O
R
N
H
N
OEt
OR
H
N
H
N
O
O
O
H
O
O
O
2a
: R = Et
10a
: R = Boc
g
b, e
2
: R = H
10
: R = Ac
a
Reaction conditions: (a) (2S,4R)ꢀ1ꢀ(tertꢀbutoxycarbonyl)ꢀ4ꢀhydroxypyrrolidineꢀ2ꢀcarboxylic acid,
HATU, DIEA, DMF, 81%. (b) HCl, dioxane, 99%. (c) (S)ꢀ2ꢀ(tertꢀbutoxycarbonylamino)ꢀ3ꢀ
methylbutanoic acid, HATU, DIEA, DMF. (d) (S)ꢀ2ꢀ(tertꢀbutoxycarbonylamino)ꢀ2ꢀcyclohexylacetic
o
acid, HATU, DIEA, DMF, 59%. (e) AcOH, HATU, DIEA, DMF, 60%. (f) (i) CDI, DCM, 25 C. (ii)
R'R"NH, DCM, 25 oC. (g) LiOHꢀH2O, THF/MeOH/H2O 2:2:1.
The route used to prepare the macrocyclic P1/P1’ acylsulfonamide 5 is shown in Scheme 3. The
enantiomerically enriched (>99% ee) P1 vinyl cyclopropyl amino acid (7a) was prepared from an
enzymatic resolution of the racemic mixture 7.16 The P3 long chain amino acid (11) was initially
obtained from a commercial source and later prepared based on literature precedent.16 Through HATUꢀ
facilitated peptide coupling steps, intermediate 12 was prepared in good yield without the need to
protect the P2 hydroxyl group. Using the same CDIꢀmediated coupling method as in Scheme 2, the P2
carbamate motif was then attached to the macrocyclic hydroxyproline intermediate 13. This method was
amenable to rapid analog synthesis. The ringꢀclosingꢀmetathesis (RCM) step from 12 to 13 was
accomplished in the presence of a variety of catalyst systems with similarly good yields and purity.16
The polymeric impurities generated from the RCM step were easily removed via a silica gel plug,
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 47
Scheme 3a
1
2
3
4
5
6
7
8
9
OH
OH
OH
O
O
O
a
d
c
H
HCl H2N
N
N
N
N
OEt
BocHN
R
OEt
BocHN
OEt
NH
O
O
O
N
H
O
O
O
OEt
8b
R = Boc
: R = H
b
7a
12
8c
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
13
N
N
e
g
O
O
O
O
O
O
O
O
H
H
N
N
N
N
S
OR
N
BocHN
BocHN
H
O
O
O
O
14
R = Et
R = H
f
5
4
:
a
Reaction conditions: (a) (2S,4R)ꢀ1ꢀ(tertꢀbutoxycarbonyl)ꢀ4ꢀhydroxypyrrolidineꢀ2ꢀcarboxylic acid,
HATU, DIEA, DMF, 80%. (b) HCl, dioxane, 99%. (c) (S)ꢀ2ꢀ(tertꢀbutoxycarbonylamino)nonꢀ8ꢀenoic
acid (11), HATU, DIEA, DMF, 87%. (d) 5 mol% RCM catalyst, DCE, 50 oC, 52%. (e) (i) CDI, DCM,
o
o
25 C. (ii) 1,2,3,4ꢀtetrahydroisoquinoline, DCM, 25 C, 90%. (f) LiOHꢀH2O, THF/MeOH/H2O 2:2:1,
87%. (g) (i) CDI, DCE, 50 oC. (ii) cyclopropanesulfonamide (15), DBU, 50 oC, 94%.
followed by crystallization of 13 directly from the filtrate. In the final steps, carboxylic acid 4 was first
activated with CDI, and then coupled to cyclopropanesulfonamide in the presence of DBU to readily
yield the final product 5. As before, no deterioration of enantiomeric purity was observed throughout
this synthetic sequence.
Discussion
The P2 carbamate tetrapeptide inhibitors prepared from the focused library (2, Scheme 1) were tested
for their inhibitory effect on the NS3 protease activity in cleaving a quenchedꢀfluorogenic substrate,
RET S1 (AnaSpec Inc., AcꢀDED(EDANS)EEAbu Ψ[COO]ASK(DABCYL)), in a continuous
fluorescence resonance energy transfer (FRET) assay. IC50 values were determined using a fourꢀ
parameter curve fit equation, and potent compounds were then evaluated in a subꢀgenomic repliconꢀ
8
ACS Paragon Plus Environment

Page 9 of 47
Journal of Medicinal Chemistry
based ELISA cellular assay that quantifies the HCV viral replication level using a selectable marker
protein, Neomycin phosphotransferase II (NPTII). EC50 values, the compound concentrations required
to achieve 50% inhibition of the HCV viral RNA replication in Huhꢀ7 hepatoma cells as compared to a
DMSOꢀtreated control signal, were calculated using a fourꢀparameter curve fit equation as well. This
ELISA quantification assay showed strong correlation to Taqman RNA analysis method. All
biochemical and cellular data reported here are an average of at least three individual measurements (n ≥
3). In all cases, individual measurements were within threeꢀfold for each compound. As mentioned
earlier, none of the inhibitors from library 2 showed detectable inhibitory effect in this cellular assay at
up to 10 ꢀM concentration. Subsequent macrocyclic analogs with much enhanced cell potency (e.g. 5)
were further tested for inhibitory activity in a panel of protease functional assays (e.g. Cathepsin B,
Chymotrypsin, Thrombin and Leukocyte Elastase) to gauge the selectivity profile of this lead series.
Representative inhibitors were also evaluated for their inhibitory effect on hERG channel activity at 1, 5
and 10 ꢀM levels. In vivo plasma and liver drug exposures were examined both intravenously and
orally in rats for selected inhibitors that met our replicon potency and in vitro hepatic stability criteria.
Orally dosed compounds with target tissue (liver) exposures above predicted efficacious levels would
then be advanced into higher species including monkeys and dogs for additional PK studies on liver and
plasma drug exposures.
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
P2 Carbamate Initial Tetrapeptide Leads. A chemical library of about 70 tetrapeptides bearing
carbamates attached to the P2 hydroxyproline was synthesized utilizing a diverse set of primary and
secondary amine inputs (2, Scheme 1). Key results from this series are summarized in Table 1. A clear
trend in the SAR emerged by which an appropriately placed lipophilic group, preferably a fused or
pendent aromatic, provided significant increase in NS3 enzyme inhibition. For example, the simple
phenyl and benzyl carbamates (16 and 18) were micromolar inhibitors, while aryl ring substituents failed
to provide any significant potency improvement (data not shown). However, the naphthyl analogs (17
and 19) provided a 3ꢀ to 4ꢀfold increase in potency and the first submicromolar inhibitors in this series.
The unexpected improvement in potency by incorporating an appropriately positioned aromatic group
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 47
Table 1. Inhibitory Activity against HCV NS3 for P2 Carbamate Tetrapeptide Lead Series.
1
2
3
4
5
6
7
8
9
R
O
O
O
H
N
N
N
H
OH
O
O
N
O
H
O
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
IC50 (nM)
Cmpd
16
R
HN
HN
3,440
17
18
19
20
21
22
23
3
835
2,755
818
HN
HN
47,103
4,769
5,500
4,286
210
N
O
O
N
O
N
N
N
24
25
26
27
14,877
735
N
N
200
N
186
N
10
ACS Paragon Plus Environment

Page 11 of 47
Journal of Medicinal Chemistry
extended to heterocyclic templates as well. The morpholine carbamate 20 was essentially inactive,
while the 2ꢀ and 3ꢀphenylmorpholine carbamates 21 and 22 demonstrated an order of magnitude
improvement in potency. Piperidine carbamate 23 was a weakly potent inhibitor at 4.3 ꢀM, whereas
tetrahydroisoquinoline carbamate 3 was 20ꢀfold more potent at 210 nM. Likewise, pyrrolidine
carbamate 24 was nearly inactive; however, 3ꢀphenyl pyrrolidine, indoline, and isoindoline carbamates
(25, 26, and 27) demonstrated up to an 80ꢀfold increase in potency at 735 nM, 200 nM and 186 nM,
respectively. The surprising SAR from this series clearly suggests that 1) lipophilic interactions in the
S2 pocket provide significant enhancements in potency, and 2) fused bicyclic amines were the optimal
P2 groups for accessing this space through a carbamate linker. From this library, compound 3 was
selected as a tool molecule for further enhancing cellular potency.
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
P2 Carbamate Macrocyclic Leads. The lack of cell potency in the tetrapeptide series was ascribed
to poor cell permeability. Two depeptidization strategies were therefore implemented—truncation of
the P4 amino acid and macrocyclization of the resulting tripeptide. Both strategies aimed at reducing
the solvent exposed polar surface areas of the tetrapeptide lead, while macrocyclization would also
rigidify the peptide backbone and protect the amide bonds from proteolytic cleavage. Truncation and
macrocyclization of inhibitor 3 led to 15ꢀmembered macrocyclic tripeptide mimetic 4. This compound
retained the same level of enzymatic potency (IC50 = 177 nM) as its tetrapeptide predecessor 3 (IC50 =
210 nM), indicating that the newly formed 15ꢀmembered macrocyclic ring system did not disrupt any
key inhibitorꢀenzyme binding site interactions. Thus, the SAR that had been established for the
tetrapeptide lead series should be analogous to the new macrocyclic series. More importantly,
compound 4 marked the program’s first NS3 inhibitor with antiviral activity inside human cells
(replicon EC50 = 2352 nM) and validated the improvement in cell permeability by the above
depeptidization approaches.
To further improve potency and reduce the EC50/IC50 ratio of inhibitor 4 (~ 10), the P1/P1’ carboxylic
acid group was replaced with an acylsulfonamide as a carboxylic acid bioisostere. It was postulated that
the greater delocalization of negative charge around the acylsulfonamide versus the carboxylic acid
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 47
would lead to increased cell permeability. Furthermore, molecular modeling suggested that additional
binding interactions could be gained between the extended P1’ group and the S1’ binding pocket of the
active site. Remarkably, this seemingly simple modification indeed provided a profound potency
enhancement. Inhibitor 5, a P1/P1’ acylsulfonamide analog, gave an IC50 of 2.2 nM—a boost of about
100 times over its carboxylic acid parent 4. Its replicon EC50 of 17.5 nM was over 130 times more
potent than 4, although it is worth noting that the EC50/IC50 ratio remained essentially unchanged as
compared with its carboxylic acid counterpart.
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
After intravenous administration to rats, inhibitor 5 was found to be rapidly cleared from the plasma
compartment (clearance greater than blood flow). However, analysis of the animals’ livers
demonstrated excellent drug exposure and much lower clearance. This drastic tissueꢀspecific clearance
difference was again observed after the drug was orally administered to rats, and the compound’s
liver/plasma exposure ratio was greater than 1000.
Because HCV viral replication occurs predominantly, if not exclusively, in hepatocytes, excellent drug
exposure in the liver—not plasma—should be a key predictor for sustained antiviral efficacy in patients.
However, it was also observed that in vivo liver pharmacokinetics and [liver]/[plasma] ratios could
range widely (>10 times), even among closely related analogs of 5. Therefore, we resorted to
systematically modifying various regions of lead molecule 5 to establish not only the structureꢀactivityꢀ
relationship, but the structureꢀpharmacokineticꢀrelationship as well. With the newly obtained xꢀray
crystal structure (Figure 1) in hand, we quickly focused on the following strategies for further lead
optimization.
P2 Optimizations. As shown in Table 2, most of the substituted THIQ analogs did little to improve
the replicon cell potency of the parent compound 5. Efforts to block the likely metabolic site on the
THIQ ring also failed to improve the compounds in vitro hepatic stability. Compounds 29 and 30
demonstrated a 3ꢀfold improvement in replicon potency compared to 5, but their hepatocyte stability was
worse.
12
ACS Paragon Plus Environment

Page 13 of 47
Journal of Medicinal Chemistry
As described in Table 1, tetrapeptide analog 27 bearing a P2 isoindoline carbamate was equipotent to
3 (IC50 = 186 and 210 nM, respectively), and the macrocyclic analog of 3 maintained its potency (4, IC50
= 177 nM). However, the macrocyclic counterpart of 27 demonstrated an unexpected potency boost (40,
Table 3, IC50 = 11 nM). It is plausible that isoindoline, a flatter ring system than THIQ, may enable
better lipophilic interactions with the flat and lipophilic S2 protein surface. Further attempts to optimize
the isoindoline group in this lead series via substitutions around the ring only yielded modest
improvement (Table 3). Through these optimization efforts isoindoline was identified as a more potent
carbamate moiety than THIQ. Moreover, inhibitor 40 also enjoyed slower in vitro hepatic clearance and
higher in vivo exposures in rats than its THIQ counterpart 4. The lower hepatic clearance may be
attributed to the elimination of the metabolically more vulnerable benzylic protons on the 4ꢀposition of
the THIQ ring.
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
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 47
Table 2. P2 THIQ Carbamate Inhibitory Activity against HCV NS3.
1
2
3
4
6
7
5
6
7
8
5
4
8
1
(THIQ)
O
S
NH
N
3
9
O
O
O
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
HN
N
O
O
O
O
HN
IC50
EC50
(nM)
Cmpd
THIQ
(nM)
5
unsubstituted
4-OH
2.2
2.2
1.0
0.8
0.7
2.4
2.0
2.6
4.1
1.1
2.1
0.8
1.2
18
718
5.7
7.7
12
20
11
35
65
12
43
13
24
28
29
30
31
32
33
34
35
36
37
38
39
4-Me
4,4-dimethyl
5-F
5-NMe2
6-F
6-CF3
7-Cl
8-F
8-Cl
8-CF3
5,8-di-F
14
ACS Paragon Plus Environment

Page 15 of 47
Journal of Medicinal Chemistry
Table 3. P2 Isoindoline Carbamate Inhibitory Activity against HCV NS3.
1
2
3
4
5
6
7
5
6
7
4
1
8
9
N
O
3
O
O
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
OH
HN
N
O
O
O
O
HN
IC50
(nM)
EC50
(nM)
Cmpd
Isoindoline
40
41
42
43
44
45
46
47
unsubstituted
4-Cl
11
13
1550
516
4-F
10
1268
3220
7515
3159
25000
1475
5-F
38
5-CF3
6-Cl
195
110
483
21
5,6-di-Cl
4,7-di-Cl
The superior potency of isoindoline over THIQ P2 carbamate was also clearly demonstrated in the
macrocyclic P1/P1’ acylsulfonamide series (Table 4). For instance, a 9ꢀfold cell potency boost was
observed comparing inhibitor 48 with 5 (EC50 = 2 vs. 18 nM). Various substitutions around the
isoindoline ring did not yield significant cell potency improvements, even though the enzymatic potency
of some of these new analogs was enhanced. Regardless of the steric or electronic nature of the
substituents and their positions, nearly all the acylsulfonamide inhibitors tested in this series resulted in
singleꢀdigit nanomolar enzymatic potency. Compound 55 was the only exception with an IC50 value of
21 nM. Greater discrimination among the inhibitors was observed primarily in their cellular activities.
Overall, singleꢀpoint substitutions of small lipophilic groups—particularly at the 4ꢀposition of the
isoindoline ring—resulted in the most active inhibitors in cells (49, 50, 51, 54, 60 and 61). It is also
worth noting that, because of the close proximity of the P3 cap and P2 parts of the inhibitor when bound
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 47
Table 4. Inhibitory Activity against HCV NS3.
1
2
3
4
5
6
7
5
6
7
1
4
O
8
9
N
O
3
O
O
S
O
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
NH
HN
N
O
O
O
O
HN
IC50
(nM)
EC50
(nM)
IC50
(nM)
EC50
(nM)
Cmpd
Cmpd
Isoindoline
Isoindoline
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
unsubstituted
4-F
1.2
1.0
3
2.0
2.4
2.2
49
6-Cl
2.0
1.5
0.7
1.7
1.0
0.8
9.2
42
5.1
7.3
24
45
41
11
77
6-NH2
4-Cl
4,7-di-F
4-Br
1.1
1.2
0.2
3
4,7-di-Cl
5,6-di-Cl
5-morpholine
4-CF3
4-NH2
5-F
29
4.1
4.0
20
5-(N-Me-piperazine) 2.0
5-CF3
5-OCH3
5-CN
21
2.3
2.5
5-NH(iPr)
4.3
1.8
22
4-morpholine
112
to NS3 active site, an intramolecular interaction between the inhibitor’s P2 and P3 cap moieties may
have contributed to the inhibitor’s overall binding affinity. However, it is not clearly understood why
some of these inhibitors with equal or better enzymatic potency to inhibitor 48 lost much of their activity
when tested in a cellular environment.
P1’ Modifications. As shown in the xꢀray crystal structure of 6 bound to the NS3 protease active site
(Figure 1), the cyclopropyl P1’ group appears to only partially fill the relatively large S1’ pocket.
Inducedꢀfit computational docking suggested that larger groups such as phenyl rings might be better
accommodated within the S1’site. Moreover, it is well known in literature that amino acid residues in
the NS3 S1’ pocket demonstrate considerable conformational flexibility, producing an “induced fit” to
16
ACS Paragon Plus Environment

Page 17 of 47
Journal of Medicinal Chemistry
Table 5. P1’ Inhibitory Activity SAR against HCV NS3.
1
2
3
4
5
6
7
8
9
O
N
O
O
O
R
S
O
NH
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
HN
N
O
O
O
O
HN
IC50
(nM)
EC50
(nM)
IC50
(nM) (nM)
EC50
Cmpd
Cmpd
R
R
48
67
68
69
70
71
72
73
74
75
76
84
85
86
87
N
cyclopropyl
CH3
1.2
6.6
4.8
37
4.8
3.5
3
2.0
156
21
2.5
4.2
2.4
2.6
2.8
32
i-Pr
N
H
tert-Bu
194
166
11
N
CF3
6.5
C6H5
p-Cl-C6H5
p-MeO-C6H5
p-CN-C6H5
p-NMe2-C6H5
(2,5-di-F-C6H5)
S
58
11.8
N
4.1
3
10
125
45
N
N
88
89
11
0.7
4.4
22
52
O
285
509
77
78
2.5
4.6
27
NH
S
Cl
N
90
91
111
17
70
46
N
S
Cl
Cl
79
N(i-Pr)2
NH(C6H5)
0.5
32
139
80
81
82
83
R' = CH3
92
93
94
95
0.5
7.7
3
3.1
13
5
2.2
7.9
11
30
97
R'
CH2OCH3
NH(p-Cl-C6H5)
CH2CH2CH3
CH2Ph
NH(p-MeO-C6H5)
NH(p-CF3-C6H5)
184
70
1.6
25
9.8
accommodate a wide range of P1’ substituent sizes. We investigated this by preparing a variety of
P1/P1’ aryl and alkyl acylsulfonamide analogs in an attempt to find an optimal motif that balances both
potency and pharmacokinetic properties. For instance, switching from alkyl to aryl acylsulfonamides
would not only adjust the P1’ substituent’s steric properties, but would also modulate its acidity, thereby
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 47
potentially affecting the overall pharmacokinetics of these new analogs. A number of these examples
validated this approach, and possessed comparable cell potency and rat pharmacokinetics to parent 48.
As shown in Table 5, small alkyl groups are tolerated in the P1’ position up to the size of tertꢀbutyl,
which drastically diminished potency (69). Phenyl groups, on the other hand, were unexpectedly well
tolerated even with paraꢀsubstitutions, which appeared too large to fit in the S1’ pocket in docking
studies. This can be explained by examining the Xꢀray coꢀcrystal structure of inhibitor 71,which shows
that the trajectory of the phenyl ring does not point toward the bottom of the S1’ pocket as is the case for
inhibitor 48, but is rather directed toward solvent.
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
Acylsulfamides (84–95, Table 5) were introduced to further expand the scope of P1’region and
explore possible DMPK benefits. Compared to its acylsulfonamide counterpart, acylsulfamide (or
sulfonyl urea) was only slightly less acidic (e.g. ACDꢀpredicted pKa = 4.81 for 84, and 4.49 for 48), had
very similar lipophilicity (cLogP = 5.4 for both 48 and 84), exhibited similar trend of steric tolerance for
the P1’ group and same level of potency, but it tended to show higher hepatic clearance rate as measured
by in vitro hepatocyte stability assay (e.g. ER% = 1% and 16% for 48 and 84, respectively, in rat
hepatocytes).
One of the most interesting changes observed in this series was some compounds’ preferential tissue
distributions to liver. For example, after orally dosed in rats, compound 48 had a liver/plasma ratio of
approx. 1:1 throughout the time course (as measured by AUC0ꢀ24h), whereas compound 84 displayed a
liver/plasma ratio of >100:1 (Table 10). It was clear from this initial data that it would be almost
impossible to use plasma concentration to predict whether the target organ (liver) would have adequate
drug exposure, and that even seemingly small changes in drug design could result in drastically altering
compound liver levels. Overall, based on both EC50 and rat liver PK data, the optimal P1’ group for this
series were determined to be small lipophilic groups as in inhibitors 48, 80 and 84.
P3 Cap Modifications. As our crystallographic studies illustrated, the P3 caps in the macrocyclic P2
carbamate inhibitors interact with the S4 protein residues mainly through lipophilic interactions. A
variety of modifications at this position are well tolerated in terms of maintaining their enzymatic
18
ACS Paragon Plus Environment

Page 19 of 47
Journal of Medicinal Chemistry
potency (Table 6). Introduction of solublizing groups into this region maintained enzymatic potency,
but resulted in a loss of cell potency (inhibitors 101ꢀ103, Table 6). Switching the P3 cap from
carbamate to urea improved the inhibitor’s in vitro hepatic stability but lost about 10ꢀ to 20ꢀfold cell
activity (99, 100).
1
2
3
4
5
6
7
8
9
Macrocyclic Linker Modifications. Reduction of the double bond on the macrocycle provided closeꢀ
in analogs with slightly increased lipophilicity which sometimes improved cell potency. Unfortunately,
the fully saturated aliphatic macrocycle linkers also tended to decrease the compounds’ hepatic stability.
Compounds in Table 7 exemplify our efforts to block potential metabolic sites on the macrocyclic
linker. Reducing macrocyclic ring size from 15 to 14 resulted in a loss of cell potency (105 vs. 106 and
104 vs. 49). Replacing methylene groups on the 15ꢀmembered macrocycle with oxygen also
significantly reduced cell potency (49 vs. 107-109). In several examples, saturation of the double bond
on the macrocyclic linker noticeably improved cell potency (104 vs. 105 and 107 vs. 110), but the
alteration often led to deteriorated in vitro metabolic stability. Replacing a methylene with CF2 did yield
analogs with the same level of replicon potency as parent 49 (113 and 118). Unfortunately, none of
these analogs that were designed to reduce metabolic soft spots actually improved in vitro metabolic
stability or in vivo exposures in rats.
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
P2 Linker Modifications. A variety of P2 linkers besides carbamate were prepared based on the xꢀ
ray crystal structure of inhibitor 6 bound to the NS3 active site. The enzymatic and cellular activities of
these analogs are shown in Table 8. Analogs with a sulfonamide linker (e.g. 122) demonstrated a
significant loss of enzymatic potency, whereas analogs with an amide or a urea linker maintained the
same level of enzymatic potency as the carbamateꢀlinked inhibitors. Certain amideꢀlinked inhibitors
such as 126 and 127 showed the least cell potency loss, but were still 5ꢀ10 fold less potent than 49.
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 47
Table 6. Inhibitory Activity SAR against HCV NS3 for P3 Caps.
1
2
3
4
5
6
7
O
8
9
N
O
O
S
O
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
NH
HN
N
O
O
R
HN
IC50
(nM)
EC50
(nM)
Cmpd
R
48
96
(t-Bu)O(CO)-
1.2
1
2
(cyclopentyl)O(CO)-
FCH2CH2O(CO)-
CF3CH2CH2(CO)-
(t-Bu)NH(CO)-
(cyclopentyl)NH(CO)-
O
4
97
1.5
0.6
2.4
0.7
54
48
38
26
98
99
100
O
O
O
O
O
O
101
102
103
2.8
2.2
2.2
107
217
97
O
O
20
ACS Paragon Plus Environment

Page 21 of 47
Journal of Medicinal Chemistry
Table 7. Inhibitory Activity SAR against HCV NS3 for Various Macrocyclic Linkers.a, b
1
2
3
4
5
6
F
O
N
O
7
8
9
O
O
S
NH
O
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
HN
N
O
O
O
O
HN
Q
IC50
(nM)
EC50
(nM)
IC50
(nM)
EC50
(nM)
Cmpd
Q
Cmpd
Q
n=14
n=15
104
105
49
1.5
1.5
1.0
76.6
22.5
2.4
112
1.0
1.5
0.5
9.7
1.8
7.0
0.8
1.3
254
7.2
44
O
n=14
n=15
n=15
n=15
113
F
F
114
F
F
n=15
n=15
106
107
108
0.7
1.8
1.2
5.8
70.2
33
n=15
115
116
117
118
119
1827
24
HN
O
O
n=15
F3C
N
n=15
837
6.8
46
n=15
n=15
n=15
O
109
0.8
155
N
F
n=15
O
O
110
111
1.2
0.7
12.4
19
F
n=15
n=15
F
F
a The n values herein refer to the macrocyclic analog’s overall ring size.
b The syntheses for inhibitors listed in this table can be found in supplementary materials.
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 47
Table 8. Inhibitory Activity SAR against HCV NS3: P2 nonꢀcarbamate Linkers.a
1
2
3
4
5
6
O
O
R
S
O
HN
NH
7
HN
8
9
N
O
O
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
O
HN
IC50
(nM)
EC50
(nM)
IC50
(nM)
EC50
(nM)
Cmpd
R
Cmpd
R
O
S
O
O
N
120
121
122
362
47
2.9
2.3
470
125
126
127
21
9.5
13
1.4
1.4
3.5
S
O
N
Cl
Cl
S
N
O
O O
S
>10,000
S
O
S
N
123
124
365
730
8.2
21
128
126
614
5.2
6.4
N
Cl
H
N
N
O
129
O
N
a The syntheses for inhibitors listed in this table can be found in supplementary materials.
In summary, potent antiviral activity of this novel class of macrocyclic NS3 protease inhibitors was
achieved through optimization of the P1’, P2 and P3 groups (Table 9). Individually, either a flat bicyclic
lipophilic ring at P2 or a small alkyl acylsulfonamide at P1/P1’ brought a similar magnitude of EC50
enhancement—from inactive (130) to ~ 2 ꢀM (4, 40 and 131). Taken together, the two modifications
further improved the EC50 to < 20 nM (5, 48ꢀ50). Attempts to trim the high molecular weights of these
macrocyclic peptidomimetics at the P2 or P4 position generated inhibitors with good enzymatic activity
(IC50 < 20 nM), but with seriously eroded cell potency (133 – 136).
22
ACS Paragon Plus Environment

Page 23 of 47
Journal of Medicinal Chemistry
Table 9. Inhibitory Activity SAR against HCV NS3.
1
2
3
4
R2
5
6
7
8
O
H
N
N
R1
R3HN
O
O
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
R1
R2
R3
IC50 (nM)
EC50 (nM)
Cmpd
OH
OH
OH
OH
>50000
Boc
Boc
Boc
Boc
Boc
Boc
Boc
Boc
Boc
Boc
H
NA
2352
1550
2131
6300
3913
91
130
4
O(CO)(THIQ)
O(CO)(isoindoline)
OH
177
11
75
220
12
3
40
NHSO2(cyclopropyl)
NHSO2(cyclopropyl)
NHSO2(cyclopropyl)
NHSO2(cyclopropyl)
NHSO2(cyclopropyl)
NHSO2(cyclopropyl)
131
132
133
134
5
OMe
OAc
OBut
O(CO)(THIQ)
O(CO)(isoindoline)
2
18
1
2
48
NHSO2(cyclopropyl) O(CO)(4'-F-isoindoline)
NHSO2(cyclopropyl) O(CO)(4'-F-isoindoline)
NHSO2(cyclopropyl) O(CO)(4'-Cl-isoindoline)
NHSO2(cyclopropyl) O(CO)(4'-Cl-isoindoline)
1
2
49
20
3
3050
2
135
50
Boc
H
3
150
136
Preclinical Development
After investigation of the DMPK profiles of this new class of NS3 inhibitors, numerous candidates
emerged as having both the desired replicon potency and excellent in vivo liver exposures in multiple
animal species after oral dosing. The drug distribution between plasma and the target organ liver varied
drastically even among very closeꢀin analogs—from approx. 1:1 to >4000:1 (Table 10). The
significantly different liverꢀpreference displayed by these inhibitors probably have much to do with their
different affinities to active uptake and efflux transporters in addition to their physiochemical
properties.18 Given their excellent antiviral potency and good range of preferential liver distribution,
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 47
inhibitors 49 and 84 were selected as preclinical candidates for animal safety studies. Based on the
overall safety study results, compound 49 was selected as the clinical development candidate. The
preclinical profile of 49 is summarized in Figure 2.17 Detailed discussions of its preclinical
characteristics including pharmacokinetics in rodents and cynos have been disclosed elsewhere.9a
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
F
N
O
O
S
NH
O
O
O
HN
O
N
O
O
O
HN
49 (ITMNꢀ191, R7227)
M.W. = 731.8, cLogP = 5.6, LogD = 2.1
pKa = 5.27 (ACD pred. 4.47)
m.p. = 137.3 oC
IC50 = 0.2ꢀ0.4 nM (1a, 1b, 4, 5 & 6), 1.6 nM (2b),
3.5 nM (3a)
EC50 = 1.6 nM (1b), 20 nM (2 & 3)
MT4ꢀMTT CC50 = 54 M
CYP 1A2, 2C19, 2C9, 2D6 > 25 M; 3A4 = 12 M
Timeꢀdependent CYP3A4 inhibition/induction: Negative
hERG %block @ 10 M: Negative
Selectivity: IC50 > 10 M in panel of 53 proteases
Oral bioavailability (liver): 15 % (rat); 78 % (monkey)
Liver/plasma ratio: ~10 (rat); ~30 (dog); ~100 (monkey)
Figure 2. Danoprevir preclinical data profile.
24
ACS Paragon Plus Environment

Page 25 of 47
Journal of Medicinal Chemistry
Table 10. ADME/PK Profiles of Danoprevir and Close Analogs.
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
Cmpd
R1
5
48
49
84
R2
N
cLogP / LogD / tPSA
IC50 / EC50 (nM)
5.9 / 2.2 / 180
2 / 18
5.4 / 2.4 / 180
1 / 2
5.6 / 2.1 / 180
1 / 2
5.4 / 2.5 / 184
1 / 3
Hepatocytes Stabilitya
ER%
(human/rat/cyno/ dog)
49 / 5 / 53 / 36
65 / 1 / 28 / 33
58 / 6 / 54 / 23
56 / 16 / 56 / 16
0.5 / 165 / >1000
Solubility (ꢀg/mL)b
@pH 1.2 / 6.5 / 7.4
0.5 / 812 / >1000
0.5 / 702 / >1000
0.5 / 548 / >1000
Caco Permeabilityc
Plasma CL (ER%)d
Vss (mL/kg)d
low / efflux
> 100%
5,982
low / efflux
6 %
low / efflux
4 %
low / efflux
> 100%
2,583
89
155
Liver AUC0-24h
276.5
40.3
90.8
136.3
(ꢀg*h/mL) e
Plasma AUC0-24h
0.06
32.9
9.5
1.03
(ꢀg*h/mL) e
AUCliver / AUCplasma
4608
1.2
10
132
a.Test article concentration = 1 ꢀM; b.Tested in aqueous pH buffers; c. Test article’s permeability is
reported as “low” when PeAB < 2 × 10ꢀ6 cm/s, and “efflux” indicates PeBA/PeAB ratio > 2. d. Rat, 1
mg/kg, I.V. ER% = CLobs/(hepatic blood flow rate); Both CLobs and blood flow rate are expressed in
mL/min/kg, and 70 mL/min/kg blood flow rate was used to calculate rat plasma clearance values; e. Rat,
30 mg/kg, P.O. Cohorts of 3 animals were sacrificed for each time point of the liver PK.
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 47
X-ray Crystal Structure
1
2
3
4
5
6
7
8
9
An xꢀray crystal structure of danoprevir bound in HCV NS3 protease was determined inꢀhouse at 2.7ꢀ
angstrom resolution by soaking the inhibitor into an apoꢀstructure of the enzyme that was solved earlier
in the program. This effort revealed a unique, nonꢀcovalent binding mode for danoprevir, distinct from
the binding modes observed for covalent inhibitors telaprevir and boceprevir.7,8 As shown in Figures 3
and 4, the S1/S1’ pocket is occupied by the cyclopropyl acylsulfonamide, with the acyl carbonyl oxygen
projecting deep in the oxyanion hole of the protease active site as it makes hydrogen bonds to Gly137
and Ser138. The sulfonyl oxygens make hydrogen bonds with Gly137 and Ser139, while the nitrogen of
the acyl sulfonamide makes a hydrogen bond with His57. It is postulated that the acyl sulfonamide
exists as an anion when bound, with His57 being protonated when hydrogenꢀbonding to the acyl
nitrogen. The P2 fluoroisoindoline lies in a lipophilic portion of the extended S2 pocket with the
fluorine directed toward the P3 tertꢀbutyl carbamate cap, which in turn occupies the shallow S4 pocket.
The aliphatic chain linking the P1 cyclopropyl to the P3 position lies in a deep hydrophobic cleft
extending from the S1 to S3 sites. The extraordinary inhibitory potency observed for danoprevir arises
from the highly optimized and synergistic combination of its tight hydrogenꢀbonding network in the
S1/S1’ site along with the extensive Van der Waals interactions within the S1ꢀS3 lipophilic cleft and
along the extended S2 pocket.
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
Recently a crystal structure of danoprevir in complex with wildꢀtype NS3 protease (PDB ID 3M5L)
was reported by an independently conducted research.19a It is of higher resolution than our previously
undisclosed inꢀhouse structure, but the inhibitor binding conformation and protein residue coordinates
reported are in good agreement with what we have observed. One difference is that in this recent report
two conformations of danoprevir’s isoindoline P2 moiety were observed (with fluorine either facing or
180 degrees facing away from the P3 cap), whereas we had only observed one conformation (with
fluorine facing the P3 cap). The existence of two danoprevir P2 conformations is not completely
unexpected, considering their being nearly equiꢀenergetic rotamers.
26
ACS Paragon Plus Environment

Page 27 of 47
Journal of Medicinal Chemistry
Danoprevir Still Potent Against A156T Resistance Mutation
1
2
3
4
5
6
7
8
9
In the crystal structure of danoprevir bound to wildꢀtype NS3 protease (Figure 3), the methyl group
side chain of Ala 156 projects upward and is surrounded by the P2 fluoroisoindoline and the tertꢀbutyl
carbamate of the P3 cap. In the presence of the drugꢀresistant A156T mutation, telaprevir loses >100ꢀ
fold potency and becomes essentially inactive, while both ciluprevir and vaniprevir lose >1000ꢀfold of
their potency.15,19b Danoprevir, on the other hand, only experiences a ~2ꢀfold drop in potency (from
<0.25 nM to 0.50 nM).21 Our docking studies predict that danoprevir should be accommodated in the
A156T resistant mutant protease for two reasons. First, unlike the large P2 group present in ciluprevir,
the P2 isoindoline group of danoprevir can undergo significant movement in the presence of T156
without creating additional steric clashes in the extended S2 pocket. Second, relative to telaprevir and
vaniprevir, the inherent rotational flexibility in the P2 and P3 cap carbamate linkage of danoprevir
allows it to undergo movements to avoid steric clashes with T156. These features are illustrated in the
docking of danoprevir into an A156T mutant HCV NS3 protease generated in silico and energyꢀ
minimized for the lowest energy rotamer of residue T156. The relatively flexible P2 and P3 cap
substituents of danoprevir ease apart and twist slightly to accommodate the larger T156 residue (Figure
5).
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
Romano et al. have recently reported a crystal structure of danoprevir in complex with A156T mutant
NS3 protease (PDB ID 3SU2).19b The inhibitor binding conformation and protein residue positions
observed are in good agreement with what we had anticipated from docking studies. One difference is
that we have previously hypothesized the possibility for the P2 fluoroisoindoline to rotate 180 degrees
(with the fluorine facing away from the P3 cap) to further alleviate steric clashes with T156. Both
conformations were indeed reported for danoprevir in complex with wildꢀtype NS3 protease as
aforementioned. However, only a single P2 isoindoline conformation was observed in this mutant
structure (with the fluorine facing the P3 cap). This is likely a result of danoprevir’s flexible P2 group
making compensatory packing against the larger T156 side chain after shifting apart from the P3 cap,
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 47
therefore retaining its potency against the mutant. Other common resistance mutations observed in the
clinic during danoprevir monotherapy (e.g. R155K) have been reported elsewhere.20
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 3. Xꢀray crystal structure of danoprevir (49) bound to NS3 protease active site.
O
N
H
N
O
P2
F
N
H
His 57
P1'
O
N
O
NH
P1
N
O
O
N
H
O
S
P3 cap
P3
O
O
HO
Ser 139
Danoprevir
O
NH
HN
(49)
H
N
O
Gly 137
O
OH
Ser 138
Figure 4. Unique binding mode of danoprevir (49) to NS3 protease active site maximizes potency
28
ACS Paragon Plus Environment

Page 29 of 47
Journal of Medicinal Chemistry
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
Figure 5. Danoprevir docked into A156T mutant HCV NS3 protease (T156 shown as spaceꢀfilling
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
representation).
Conclusions
Through structureꢀbased design and optimization, a new class of macrocyclic peptidomimetic NS3/4A
protease inhibitors has been discovered. A virtual chemical library of approximately 500 molecules was
designed and screened using a unique docking hypothesis that was established on the basis of analyzing
NS3 xꢀray crystallographic structures and carefully studying the resulting binding modes predicted by
docking. Based on these virtual screening studies, tetrapeptides with a P1/P1’ carboxylic acid head
group and P2 carbamates extending from a hydroxyproline were selected as leadꢀgeneration chemical
targets. A facile chemistry route was then identified to couple hydroxyproline with various amine
inputs, which enabled efficient modifications of the P2 substituent. Initial P2 SAR trends gleaned from
this tetrapeptide carbamate series focused our attention on several preferred structures that were all
relatively flat and rigid. Compound 3 (Scheme 1), which features a tetrahydroisoquinoline (THIQ)
carbamate motif, became our first lead. However, the lack of cellular potency of 3 necessitated
29
ACS Paragon Plus Environment